Digital PDFs

AA-N383B-TK

May 1987

592 pages

Original

22MB

Document:	FORTRAN Language Manual Feb87
Order Number:	AA-N383B-TK
Revision:	0
Pages:	592
Original Filename:	AA-N383B-TK_FORTRAN_Language_Manual_Feb87.pdf

OCR Text

TOPS-10/TOPS-20
FORTRAN Language Manual
AA-N383B-TK,AD-N383B-T1

February 1 987
This document describes the language elements of
FORTRAN-10 and FORTRAN-20.
This manual supersedes the TOPS-10jTOPS-20
FORTRAN Language Manual, order number
AA-N383B-TK.

Operating System and Version: TOPS-10 V7.03
TOPS-20 V4.1 (KS)
TOPS-20 V6. 1 (KL)
Software Version:

FORTRAN-10 V 11
FORTRAN-20 V 11
LlNK-10 V5.1
LlNK-20 V6
RMS-20 V3

digital equipment corporation
maynard, massachusetts

First Printing, February 1983
Revised, May 1985
Updated, February 1987

The information in this document is subject to change without notice and should
not be construed as a commitment by Digital Equipment Corporation. Digital
Equipment Corporation assumes no responsibility for any errors that may appear in
this document.
The software described in this document is furnished under a license and may only
be used or copied in accordance with the terms of such license.
No responsibility is assumed for the use or reliability of software on equipment that
is not supplied by Digital Equipment Corporation or its affiliated companies.

The following are trademarks of Digital Equipment Corporation:

DEC
DECmate
DECUS
DECwriter
DIBOL
MASSBUS
MicroVAX
PDP

PjOS
Professional
Q-BUS
Rainbow
RSTS
RSX
RT
UNIBUS

VAX
VAXBI
VAXELN
VMS
VT
Work Processor

~DrnDIl~D

The postpaid READER'S COMMENTS form on the last page of this document
requests the user's critical evaluation to assist us in preparing future documentation.

CONTENTS

PREFACE
CHAPTER 1
1.1
1.2
1.2.1
1.2.2
1.2.3
1.2.4
CHAPTER 2
2.1
2.2
2.2.1
2.2.2
2.2.3
2.2.4
2.3
2.3.1
2.3.2
2.3.3
2.3.4
2.3.5
2.4
CHAPTER 3
3.1
3.2
3.3
3.4
3.4.1
3.5
3.6
3.7
3.8
3.9
3.10
CHAPTER 4
4.1
4.2
4.3
4.3.1
4.3.2
4.3.3
4.4
CHAPTER 5
5.1
5.1.1

INTRODUCTION
OVERVI EW • • • • • • • • • • • • • •
• • • 1-1
MANUAL ORGANIZATION • • • • • • • • •
• 1-1
FORTRAN Language Elements (Part I) • • • • • • • 1-2
FORTRAN Statements (Part II) • • • •
• 1-2
FORTRAN Language Usage (part III)
• • 1-3
APPENDIXES •
• • • • •
1-4
CHARACTERS AND LINES
CHARACTER SET
• • • • • • • • • • • • •
STATEMENT DEFINITION AND FORMAT • • • • • • •
Statement Label Field and Statement Numbers
Line Continuation Field
Statement Field
••••
Remark Field • • • • • •
• •
LINE TYPES • • • • • • • •
• • • •
•
Initial and Continuation Lines
Multi-Statement Lines
•
• • • •
Comment Lines and Remarks
••••
Debug Lines
•••••••
• • • •
Blank Lines
•••••••••••••
LINE-SEQUENCED SOURCE FILES
• • • •

• • 2-1
• • 2-2
2-3
• • 2-3
• • 2-4
• • 2-4
• • 2-4
• 2-4
2-5
• • 2-5
2-6
• 2-7
• • 2-7

CONSTANTS
INTRODUCTION • • • • • • • • •
• • • • • • •
INTEGER CONSTANTS
••••
• • • •
REAL CONSTANTS • • • • • •
•
DOUBLE-PRECISION CONSTANTS • • • • • • • • • • • •
Comparison of Real, D-floating, and G-floating •
COMPLEX CONSTANTS
• • • • •
• •
CHARACTER CONSTANTS
•••••••••••
••
OCTAL AND DOUBLE-OCTAL CONSTANTS •
LOGICAL CONSTANTS
••••
•
HOLLERITH CONSTANTS
••••
••• •
STATEMENT LABEL CONSTANTS
• • • •
• •

3-1
3-1
3-2
3-2
3-3
3-4
3-4
3-5
3-6
3-6
3-7

SYMBOLIC NAMES, VARIABLES, AND ARRAYS
SYMBOLIC NAMES • • • • • •
• • 4-1
•••••••••
4-2
VARIABLES
ARRAYS • • • • • • • • • • • • • •
• 4-3
Array Element Subscripts • • • • • • • • • • • • 4-3
Dimensioning Arrays
• • • •
• 4-4
Order of Stored Array Elements •
• 4-5
CHARACTER SUBSTRINGS • • • • • • • • • • • •
• 4-6
EXPRESSIONS
ARITHMETIC EXPRESSIONS • • • • • • • • • •
Rules for Writing Arithmetic Expressions
iii

• • 5-1
• 5-2

February 1987

5.1.2
5.1.2.1
5.2
5.2.1
5.3
5.3.1
5.4
5.5
5.5.1
5.5.2
5.5.3
5.5.4
5.6
CHAPTER 6
6.1
6.2
6.3
6.4
6.4.1
6.4.2
CHAPTER 7
7.1
7.1.1
7.1.2
7.2
7.2.1
7.2.2
7.3
7.4
7.4.1
7.4.2
7.5
7.5.1
7.6
7.7
7.8
7.9
7.10
CHAPTER 8
8.1
8.2
8.3
8.4
CHAPTER 9
9.1
9.1.1
9.1.2
9.1.3
9.2
9.2.1
9.2.2

Arithmetic Constant Expressions
• • • • • • 5-5
Integer Constant Expression
• 5-5
CHARACTER EXPRESSIONS • • • • • • • • •
• • 5-6
Character Con~tant Expression
• • • • • • 5-7
LOGICAL EXPRESSIONS • • • • • • • • • •
• • • 5-7
Logical Constant Expression
••••
5-10
RELATIONAL EXPRESSIONS • • • • •
5-11
EVALUATION OF EXPRESSIONS
5-13
Parenthetical Subexpressions • •
5-13
Hierarchy of Operators • • • • • •
5-14
Mixed-Mode Expressions • •
•••
5-15
Use of Logical Operands in Mixed-Mode
Expressions
• • ••
••••
5-15
CONSTANT EXPRESSIONS • • • • • • • • • • •
5-15
EXECUTABLE AND NONEXECUTABLE STATEMENTS
EXECUTABLE STATEMENTS
NONEXECUTABLE STATEMENTS • • • • • • • •
ORDERING OF FORTRAN STATEMENTS •
COMPILATION CONTROL STATEMENTS •
PROGRAM Statement
INCLUDE Statement • • • • •

• • 6-1
• • • 6-2
• 6-2
• • • 6-4
• • 6-4
• • 6-4

SPECIFICATION AND DATA STATEMENTS
DIMENSION STATEMENT • • • • • • • • •
• •
Adjustable Dimensions
•••••
• •
Assumed-size Arrays
•••••••••
• •
TYPE SPECIFICATION STATEMENTS
•
Numeric Type Specification Statements
•
Character Type Specification Statements
IMPLICIT STATEMENTS • • • • • • • • • • • •
COMMON STATEMENT • • • • • • • • • • • • • • • •
Dimensioning Arrays 'in COMMON Statements •
Character Data in COMMON • • • • • •
EQUIVALENCE STATEMENT
• • • • • • •
EQUIVALENCE and Extended Addressing
•••••
EXTERNAL STATEMENT • •
INTRINSIC STATEMENT
PARAMETER STATEMENT
DATA STATEMENT •
SAVE STATEMENT • • •

•
•
•
•
•
•
•
•

7-1
7-3
7-4
7-4
7-5
7-6
7-7
7-8
7-10
7-10
7-11
7-17
7-17
7-18
7-20
7-21
7-24

ASSIGNMENT STATEMENTS
ARITHMETIC ASSIGNMENT STATEMENT • • • • • • • • • 8-1
LOGICAL ASSIGNMENT STATEMENTS • • • • • • • •
8-3
ASSIGN (STATEMENT LABEL) ASSIGNMENT STATEMENT
8-3
CHARACTER ASSIGNMENT STATEMENT • • • • • • • • • • 8-4
CONTROL STATEMENTS
GO TO STATEMENTS • • • • • • • • • •
Unconditional GO TO Statements •
Computed GO TO Statements
Assigned GO TO Statements
IF STATEMENTS • • • • • • • • • • • • • •
Arithmetic IF Statements • • • • • • • •
Logical IF Statements

• • 9-1
• 9-2
• • • 9-2
9-3
9-3
9-3
• 9-4

February 1987

9.2.3
9.2.4
9.2.4.1
9.2.4.2
9.2.4.3
9.3
9.3.1
9.3.1.1
9.3.1.2
9.3.2
9.3.3
9.3.4
9.3.5
9.3.6
9.4
9.5
9.6
9.7
9.8
CHAPTER 10

Logical Two-Branch IF Statements •
Block IF Statements • • • • •
Statement Blocks • • • • •
Block IF Examples • • • • • • • • • • •
Nested Block IF Constructs •
•
DO STATEMENT • • • • • •
Indexed DO Statement •
• • • • • • •
Executing an Indexed DO Statement
DO Iteration Control • • • •
DO WHILE Statement • • • • •
The Range of a DO Statement
Nested DO Statements • • • • •
Extended Range • • • • • • • •
Permitted Transfer Operations • • • • •
END DO STATEMENT •
CONTINUE STATEMENT •
STOP STATEMENT • • • • •
PAUSE STATEMENT • • • •
END STATEMENT
• • • •

• • 9-4
• • 9-5
• • 9-8
9-8
• • 9-9
9-10
9-10
9-12
9-12
9-13
9-14
9-14
9-15
9-16
9-17
9-17
9-18
9-19
9-20

DATA TRANSFER STATEMENTS

DATA TRANSFER OPERATIONS
10.1
10.2
DATA ACCESS • • • • • •
Sequential Access
10.2.1
Direct Access • • • •
10.2.2
10.2.3
Keyed Access (TOPS-20 RMS)
10.3
FORMATTED AND UNFORMATTED DATA TRANSFERS • •
10.3.1
Formatted Data Transfers • •
10.3.1.1
Internal Files • • • • • •
• • • •
10.3.2
Unformatted Data Transfers •
•
Unformatted Data Transfer to ASCII Devices
10.3.3
10.4
DATA TRANSFER STATEMENT FORMS
• • • •
10.4.1
Data Transfer Statement Names
• • • • • •
10.4.2
Data Transfer Control-Information List • •
10.4.3
Unit References in Data Transfer Statements •
10.4.3.1
FORTRAN Logical Unit Identifier
••••• •
10.4.3.2
Internal File Identifier • • • • • • • • • •
Record Number References In Data Transfer
10.4.4
Statements • • • • • • • • • • • • • • • • • •
Format References in Data Transfer Statements
10.4.5
FORMAT-Statement Formatting
••••••••
10.4.5.1
List-Directed Formatting • • • •
•
10.4.5.2
NAMELIST-Statement Formatting • • • • • • •
10.4.5.3
10.4.6
Optional End-of-File Transfer of Control
(END=) • • • • • • • • • • • • • • • • •
•
Optional Data Transfer Error Control (ERR=)
•
10.4.7
Optional Error Variable For Error Reporting
10.4.8
(IOSTAT=)
••••••••••••••••••
10.4.9
Key-Field-Value Specifier (TOPS-20 RMS)
•
10.4.9.1
Key Attributes. • • • • • • • • •
• •
10.4.10
Key-of-Reference Specifier. • • • •
• ••
10.4.11
Data Transfer Statement Input/Output Lists • •
10.4.11.1
Simple List Elements. • • • • • •
•
10.4.11.2
Implied DO Lists. • • • • • • •
•
10.5
READ STATEMENT • • • • • • • • • • • •
• •
10.5.1
Formatted READ Transfers. • • • • • •
•
10.5.1.1
Sequential FORMAT-Statement READ • • • • • •
10.5.1.2
Direct-Access FORMAT-Statement READ
••
10.5.1.3
Sequential List-Directed READ
•
10.5.1.4
Sequential NAMELIST-Statement READ.
•
10.5.2
Unformatted READ Transfers. •
• ••
10.5.2.1
sequential Unformatted READ
•
v

10-5
10-5
10-6
10-6
10-6
10-6
10-7
10-8
10-8
10-8
10-9
10-9
10-9
10-11
10-11
10-13
10-13
10-14
10-15
10-16
10-17
10-18
10-19
10-20
10-21
10-23
10-23
10-23
10-24
10-25
10-27
10-29
10-29
10-30
10-30
10-31
10-32
10-32

February 1987

10.5.2.2
Direct-Access Unformatted READ.
• •••
10.5.3
Indexed READ Transfers (TOPS-20)
•••
10.5.3.1
Sequential Indexed READ • • • • • • • • • •
10.5.3.2
Formatted Indexed READ
••••
10.5.3.3
Unformatted Indexed READ
••••
•
10.6
WRITE STATEMENT • • • • • •
• •
10.6.1
Formatted WRITE Transfers
• • • •
•
10.6.1.1
Sequential FORMAT~Statement WRITE
•••
10.6.1.2
Direct-Access FORMAT-Statement WRITE • • • •
10.6.1.3
Sequential List-Directed WRITE • • • • • • •
10.6.1.4
Sequential NAMELIST-Statement WRITE
••
10.6.2
Unformatted WRITE Transfers
••
10.6.2.1
Sequential Unformatted WRITE. • •
• ••
10.6.2.2
Direct-Access Unformatted WRITE
•••
10.6.3
Writing to RMS Files (TOPS-20) • •
• •••
10.7
REREAD STATEMENT. • • • • • • • • •
•
10.7.1
Sequential FORMAT-Statement REREAD.
•
10.7.2
Sequential List-Directed REREAD
•••••
10.8
ACCEPT STATEMENT. • • • • • • • • • •
• ••
10.8.1
Sequential FORMAT-Statement ACCEPT
•
10.8.2
Sequential List-Directed ACCEPT
••
10.9
TYPE STATEMENT. • • • • • • • • • •
•
10.9.1
Sequential FORMAT-Statement TYPE. •
•
10.9.2
Sequential List-Directed TYPE
•
10.10
PRINT STATEMENT • • • • • • • • • • •
•
10.10.1
Sequential FORMAT-Statement PRINT • • • •
10.10.2
Sequential List-Directed PRINT. • •
• ••
10.11
PUNCH STATEMENT • • • • • • • • • • •
10.11.1
Sequential FORMAT-Statement PUNCH • • • • • •
10.11.2
Sequential List-Directed PUNCH. • • • •
10.12
INTERNAL FILES AND ENCODE/DECODE STATEMENTS • •
10.12.1
Internal READ and WRITE Statements
•
10.12.2
ENCODE and DECODE Statements.
• ••••
10.13
DELETE STATEMENT (TOPS-20) • •
• ••••
10.13.1
Current-Record DELETE
••••••••••
10.13.2
Direct-Access DELETE. • • •
• ••••
10.14
REWRITE STATEMENT (TOPS-20)
••••
•
10.14.1
Formatted REWRITE • • • • •
•
10.14.2
Unformatted REWRITE Statement
•••••
10.15
UNLOCK STATEMENT (TOPS-20) • • •
•
CHAPTER 11

10-33
10-33
10-34
10-34
10-35
10-35
10-37
10-37
10-38
10-39
10-40
10-40
10-41
10-41
10-42
10-43
10-43
10-44
10-45
10-45
10-46
10-47
10-48
10-49
10-50
10-50
10-51
10-52
10-53
10-53
10-54
10-55
10-56
10-59
10-60
10-60
10-60
10-61
10-61
10-62

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS

11.1
FILE-CONTROL STATEMENTS
11.2
OPEN STATEMENT
Impl ic i t OPEN
11.2.1
11.2.2
OPEN on a Connected Unit
11.3
OPEN STATEMENT SPECIFIERS
11.3.1
ACCESS Specifier
11.3.2
ASSOCIATEVARIABLE Spec! f ier
11.3.3
BLANK Specifier
11.3.4
BLOCKSIZE Specifier
11.3.5
BUFFERCOUNT Specifier
BYTESIZE Specifier
11.3.6
11.3.7
CARRIAGECONTROL Specifier
11.3.8
DEFAULTFILE Specifier (TOPS-20)
11.3.9
DENSITY Specifier
11.3.10
DEVICE Specifier
11.3.11
DIALOG Specifier
11.3.12
DIALOG= Specifier
11.3.13
DIRECTORY Specifier (TOPS-10)
11.3.14
DIRECTORY Specifier (TOPS-20)
DISPOSE Specifier
11.3.15

. .. · · · · ·

· · · · ··
·····

· · · · 11-1
11-1
·· ·· · ·

· · ·· ·· ·· ·· .. ·· · ·
·
······
· · · ·· ··
·· ··· ··· ··· ··· ··· ··· ··· · · ·· ·· ··
· · · · · · · · ·· · · ··
· ··
········

11-3
11-4
11-4
11-7
11-8
11-9
11-10
11-11
11-12
11-12
11-14
11-15
11-15
11-16
11-16
11-17
11-18
11-19

February 1987

11.3.16
ERR Specifier • • • • • • • • • • • • • • • •
11.3.17
FILE Specifier • • • • • • • • • • • • • • • •
11.3.17.1
Remote File Specifications (TOPS-20) • • • •
11.3.18
FILESIZE (INITIALIZE) Specifier (TOPS-10)
••
11.3.19
FORM Specifier • • • • • • • •
• •
11.3.20
IOSTAT Specifier • • • • • •
•
11.3.21
KEY Specifier (TOPS-20)
• • • • •
• •
11.3.22
LIMIT Specifier • • • • •
•
11.3.23
MAXREC Specifier (TOPS-20) • • • •
•
11.3.24
MODE Specifier • • • •
• ••••
11.3.25
NAME Specifier • • • ~
• • • •
• •
11.3.26
NOSPANBLOCKS Specifier (TOPS-20) • • • •
•
11.3.27
ORGANIZATION Specifier (TOPS-20)
•
11.3.28
PADCHAR Specifier
••••••••••••
11.3.29
PARITY Specifier • • • • • • •
• ••
11.3 .. 30
PROTECTION Specifier (TOPS-10) • • • • • • • •
11.3.31
PROTECTION Specifier (TOPS-20)
11.3.32
READONLY Specifier • • • • •
• ••
11.3.33
RECL (RECORDSIZE) Specifier
•••••••
11.3 .. 34
RECORDTYPE Specifier • • • • • • • • •
•
11.3.35
SHARED Specifier (TOPS-20)
••••
11.3.36
STATUS (TYPE) Specifier
••••••
•
11.3.37
TAPEFORMAT SPECIFIER • • •
• •
11.3.38
UNIT Specifier (Required)
••
• •••••
11.3.39
USEROPENSpecifier (TOPS-20)
••
11.3.40
VERSION Specifier (TOPS-10)
•••••••••
11.4
CLOSE STATEMENT • • • • •
• • • •
11.4.1
Impl ic it CLOSE • '. • • •
• • •
•
11.5
CLOSE STATEMENT SPECIFIERS
• • • • • •
11.5.1
DEVICE, DIRECTORY, FILE, NAME, and PROTECTION
Specifiers • • • • •
• ••••••••••
11.5.2
DIALOG Specifier •
• •••••
11.5.3
DIALOG= Specifier
•••••
11.5.4
DISPOSE Specifier
•
11.5.5
ERR Specifier
••
• ••••••••
11.5.6
IOSTAT Specifier • • • • • •
• •
11.5.7
LIMIT Specifier
•
• ••••
11.5.8
STATUS Specifier • • • • •
• ••••
11.5.9
UNIT Specifier (Required)
••
11.6
OPEN AND CLOSE STATEMENT EXAMPLES
•
11.7
INQUIRE STATEMENT
• • • • •
•
11.7.1
INQUIRE by File
• • • • •
•
11.,7.2
INQUIRE by Unit
••
• • • • •
•
11.7.3
INQUIRE Specifiers
• • • •
• ••
11.7.3.1
ACCESS Specifier •
• •
11.7.3.2
BLANK Specifier.
•• • •
•
11.7.3.3
BYTESIZE Specifier • • • • • • • •
• ••
11.7.3.4
CARRIAGECONTROL Specifier
••••
• •
11.7.3.5
DIRECT Specifier •
• •
11. 7 • 3 ..-6
ERR Specifier
• • • •
•
11.7.3.7
EXIST Specifier
• • • • • •
•
11.7.3.8
FORM Specifier • • • • • • • • • • • •
•
FORMATTED Specifier
•••
11.7.3.9
IOSTAT Specifier
• • • •
• ••
11.7.3.10
11.7.3.11
KEYED Specifier
•
11.7.3.12
NAME Specifier • • • • • • • • • • •
•
NAMED Specifier
• • • •
•
11.7.3.13
NEXTREC Specifier'
•
11.7.3.14
NUMBER Specifier •
•
11.7.3.15
11.7.3.16
OPENED Specifier • • • •
•
11.7.3.17
ORGANIZATION Specifier • •
• ••••
11.,7.3.18
RECL (RECORDSIZE) Specifier
••••••••
RECORDTYPE Specifier
11.7.3.19
SEQUENTIAL Specifier • • • • •
• •
11.7.3.20
vi i

11-20
11-20
11-21
11-21
11-22
11-23
11-24
11-25
11-25
11-26
11-28
11-28
11-29
11-29
11-30
11-30
11-32
11-34
11-34
11-36

I1-j8
11-39
11-41
11-42
11-42
11-43
11-43
11-44
11-44
11-46
11-46
11-47
11-47
11-49
11-49
11-49
11-50
11-51
11-51
11-52
11-52
11-53
11-54
11-54
11-54
11-54
11-54
11-55
11-55
11-55
11-56
11-56
11-56
11-56
11-57
11-58
11-58
11-58
11-58
11-59
11-59
11-59
11-60

February 1987

11.7.3.21
UNFORMATTED Specifier
•••••••
11.8
DEVICE CONTROL STATEMENTS
•••
11.8.1
FIND Statement. • • • • •
• ••
11.8.2
REWIND Statement..
••••••
•
11.8.3
UNLOAD Statement. •
• • • •
11.8.4
BACKSPACE Statement • • • • • • • • • •
11.8.5
ENDFILE Statement
• • • • • • • • •
11.8.6
SKIPRECORD Statement
••••
11.8.7
SKIPFILE Statement.
• ••••••••••
11.8.8
BACKFILE Statement •
• • •
• •
CHAPTER 12

11-60
11-61
11-62
11-63
11-64
11-64
11-65
11-66
11-67
11-67

FORMATTED DATA TRANSFERS

12.1
12.1.1
12.1.2

FORMAT-STATEMENT FORMATTING • • • • • • • • • • 12-2
Specifying a Format List in a FORMAT Statement 12-2
Specifying a Format Specification as a
Character Expression • • • • • • • • • • • • • 12-3
12.1.3
Specifying a Format Specification in a Numeric
Array
• • • • • • • • • • • • • • • • • 12-4
12.1.4
Specifying a FORMAT Statement Using an ASSIGNed
Variable • •
•••••• • • •
• • • • 12-4
12.1.5
The Ordering and Interpretation of Format List
Items
• • • • • • • • •
• • •
12-5
12.2
EDIT DESCRIPTORS • • • • • • • • • •
• • ••
12-6
12.2.1
Repeatable Edit Descriptors
• • • • •
12-7
12.2.2
Nonrepeatab1e Edit Descriptors •
12-8
12.2.3
Carriage-Control Specifiers • •
• • ••
12-9
12.3
INTERACTION OF INPUT/OUTPUT LIST AND FORMAT LIST 12-11
12.3.1
General Description • • • • • •
• 12-11
12.3.2
Formatted Input • • • • • • •
• • • • • 12-14
12.3.3
Formatted Output • • • • • • • • • • • • • • • 12-15
12.3.4
Embedded Format Specifications • • • • •
• 12-15
12.4
FORMAT EDITING • • • • • • •
• • 12-16
12.4.1
Apostrophe (I) Editing • • • • •
• • 12-16
12.4.2
H Editing
••••••••
•• •
12-17
12.4.3
positional Editing • • • • • • •
• 12-18
12.4.3.1
T, TL, and TR Editing
•••••••
• 12-19
12.4.3.2
X Editing
••••••••••••••
• 12-21
12.4.4
$ (Dollar Sign) Editing
• • 12-22
12.4.5
/ ( S 1 ash) Ed i t i ng
• • • • • •
• • 12 - 23
12.4.6
: (Colon) Editing
••
• • • • 12-24
12.4.7
S, SP,and SS Editing • • • • • • •
• • • 12-24
12.4.8
P Editing
••••
• • • • 12-25
12.4.9
BN and BZ Editing
• • • • • • • • • • • 12-28
12.4.10
Q Editing
• • •
12-29
12.4.11
Numeric Editing
• • • •
• • 12-29
12.4.11.1
I Editing
• • • • •
• • • • 12-31
12.4.11.2
F Editing
• • • •
• • • • 12-32
12.4.11.3
E and D Editing
• • • • • • • • ••
12-32
12.4.11.4
G Editing
• • • • • • • • • • • • • • 12-33
12.4.11.5
Complex Editing
• • • • • • • • • • 12-36
12.4.11.6
o (Octal) Editing
• • •
• • • 12-36
12.4.11.7
Z Editing
• • • • 12-37
12.4.12
L Editing
• • • • • • • • • • • •
• 12-38
12.4.13
A Editing
• • • • • • • • • •
• • • • 12-39
12.4.14
R Editing
• • • • • • • • • • •
• • 12-40
12.5
LIST-DIRECTED FORMATTING •
• • • • 12-41
12.6
NAME LIST-STATEMENT FORMATTING
• • 12-44
12.7
NAMELIST STATEMENT • • • •
• • • • • • • • • 12-44
12.7.1
NAMELIST-Contro11ed Data Input Transfer • • • 12-45
12.7.2
NAMELIST-Contro11ed Data Output Transfers
• • 12-46

vi ii

February 1987

CHAPTER 13

FUNCTIONS AND SUBROUTINES

13.1
INTRINSIC FUNCTIONS
• • • • • • • • •
13.1.1
Using an Intrinsic Function
• • • ••
13.1.2
Character Intrinsic Functions • • • • • • • •
13.1.3
Character Comparison Functions • • • • • • • •
13.1.4
Bit Manipulation Functions • •
13.2
STATEMENT FUNCTIONS • • • • • • •
•
13.2.1
Defining a Statement Function • • • • • • • •
Using a Statement Function • •
• • ••
13.2.2
Statement Function Restrictions • • • •
13.2.3
13.3
EXTERNAL FUNCTIONS • • • • • • .. • • •
•
FORTRAN-Supplied External Functions
•
13.3.1
User-Defined External Functions
•
13.3.2
Function Subprogram Restrictions •
•
13.3.3
Using a Function Subprogram
•••••
13.3.4
13.4
SUBROUTINES • • • • • • • • • • • • • • • • • •
FORTRAN-Supplied Subroutines. • • • • •
•
13.4.1
13.4.1.1
ALCCHR Subroutine
• • • •
•
CDABS Function • •
13.4.1.2
13.4.1.3
CDCOS Subroutine • •
• • • • • •
13.4.1.4
CDSXP Subroutine.
• • • • • • •
CDLOG Subroutine •
••• • • • • •
•
13.4.1.5
13.4.1.6
CDSIN Subroutine • • • • • • •
•
13.4.1.7
CDSQRT Subroutine
• • • • • • •
•
CHKDIV Subroutine
• • • •
• •
13.4.1.8
CLRFMT Subroutine
• •
13.4.1.9
DATE Subroutine
• • • •
•
13.4.1.10
DIVERT Subroutine
•
13.4.1.11
13.4.1.12
DTOGA • • • • • • • • • • • •
••• • •
DUMP Subroutine • • • •
• • • •
13.4.1.13
ERRSET Subroutine
• • • • • • • •
13.4.1.14
ERRSNS Subroutine
• • • • • • •
•
13.4.1.15
EXIT Subroutine
•
13.4.1.16
FFUNIT Subroutine
• • • • •
•
13.4.1.17
GTODA Subroutine • •
•
13.4.1.18
ILL Subroutine • •
• • • • • • • • • • •
13.4.1.19
LEGAL Subroutine •
13.4.1.20
MVBITS Subroutine
• • • •
•
13.4.1.21
OVERFL Subroutine
• • • •
•
13.4.1.22
PDUMP Subroutine •
13.4.1.23
QUIETX Subroutine
• • • • •
13.4.1.24
SAVFMT Subroutine • • • • • • •
• • •
13.4.1.25
SAVRAN Subroutine
• • • •
•• • • •
13.4.1.26
SETRAN Subroutine
• • • • • • •
13.4.1.27
SORT Subroutine
• • • • • • ••
•
13.4.1.28
SRTINI Subroutine • • • •
• • • • • • •
13.4.1.29
TIME Subroutine
• • • • •
•
13.4.1.30
TOPMEM Subroutine
•••••
13.4.1.31
TRACE Subroutine • • •
• • • •
•
13.4.1.32
User-Defined Subroutines
••••
•
13.4.2
SUBROUTINE Statement • • • • • • •
•
13.4.2.1
CALL Statement • • • •
13.4.2.2
Execution of a CALL Statement
• • • • •
13.4.2.3
Actual Arguments for a Subroutine
• • •
13.4.2.4
ENTRY Statement
••••••••••
•
13.4.3
RETURN Statement • • • • • • • • • • • •
•
13.4.4
Dummy and Actual Arguments • • • • • • • • • •
13.4.5
Length of Character Dummy and Actual
13.4.5.1
Arguments • • • • • • • • • • • • • • • • •
Character and Hollerith Constants as Actual
13.4.5.2
Arguments
•••••••••••••••••
0

•

13-1
13-2
13-12
13-14
13-14
13-15
13-16
13-16
13-17
13-18
13-18
13-20
13-21
13-21
13-22
13-24
13-24
13-25
13-25
13-26
13-26
13-27
13-28
13-28
13-29
13-29
13-30
13-30
13-30
13-31
13-32
13-33
13-33
13-34
13-34
13-34
13-35
13-35
13-36
13-36
13-37
13-37
13-38
13-38
13-39
13-39
13-40
13-41
13-42
13-43
13-44
13-44
13-45
13-45
13-47
13-49
13-51
13-51

February 1987

CHAPTER 14
14.1
CHAPTER 15

BLOCK DATA SUBPROGRAMS
BLOCK DATA STATEMENT •

. . . . . . . . . . . . . 14-1

WRITING USER PROGRAMS

15.1
15.1.1
15.1.2

GENERAL PROGRAMMING CONSIDERATIONS • • • • • •• 15-1
Accuracy and Range of Double-Precision Nu.mbers 15-1
Writing FORTRAN Programs for Use on Other
Computers • • • • • • • • • •
• • • •
15-2
15.1.2.1
Remote Links • • • • • • • • • • •
15-2
15.1.2.2
Reducing Network Overhead • • • •
15-2
Using Floating-Point DO Loops
••••
15.1.3
15-2
15.1.4
Computation of DO Loop Iteratio~s • • • •
15-2 .. 1
15.1.5
Subroutines - Programming Considerations •
15-3
15.1.6
Reordering of Computations •
• .- •• 15-4
Dimensioning of Dummy Arrays • • • • • • •
15.1.7
15-5
15.2
FORTRAN GLOBAL OPTIMIZATION • • • • • • •
15-6
15.2.1
Optimization Techniques • • • • • • • •
15-6
Elimination of Redundant Computations
15.2.1.1
15-6
15.2.1.2
Reduction of Operator Strength • • •
15-7
Removal of Constant Computation from Loops • 15-7
15.2.1.3
Constant Folding and Propagation •
15.2.1.4
15-8
Removal of Inaccessible Code • • • • • • • • 15-9
15.2.1.5
Global Register Allocation • • •
15.2.1.6
15-9
I/O Optimization • • • • • • • • •
15.2.1.7
15-9
Uninitia1ized Variable Detection
15.2.1.8
15-9
Test Replacement ••
•• • • • • • • • • 15-9
15.2.1.9
programming Techniques for Effective
15.2.2
Optimization . . . . . . . . . . . . . . . . . . 15-10
15.3
FUNCTION SIDE EFFECTS • • • • • • • • • • • • • 15-10
15.4
INTERACTING WITH NON-FORTRAN PROGRAMS AND FILES 15-10
15.4.1
Using The Sharable High-Segment FOROTS • • • • 15~10
15.4.2
Calling Sequences
• • • 15-11
15.4.3
Accumulator Usage • • • •
• • • 15-11
15.4.4
Argument Lists • •
• • • • • • • • • • • 15-12
15.4.5
Argument Types • •
• • • • • • •
• • 15-14
15.4.6
Description of Arguments •
• • 15-14
15.4.7
Interaction with COBOL. • • • • •
• • 15-16
Calling FORTRAN Subprograms from COBOL
15.4.7.1
Programs • • • • • • • • • • • •
• • • 15-16
15.4.7.2
Calling COBOL Subroutines from FORTRAN
Programs • • • • • • • • • • • •
• •• 15-18
15.4.8
Interaction with BLISS-36 • • • • • • • • • • 15-18
15.4.8.1
Calling FORTRAN Subprograms From BLISS-36
Programs • • • • • • • • • • • • • • • • • • 15-18
15.4.8.2
Calling BLISS-36 Routines From FORTRAN • • • 15-19
15.4.9
LINK Overlay Facilities • • • • • • • • • • • 15-19
15.4.10
FOROTS and Overlay Memory Management • • • • • 15-20
15.4.11
Extended Addressing Memory Layout (TOPS-20
only)
••••••••••
• • • 15-22
15.5
USING INDEXED FILES (TOPS-20 ONLY) • • • •
• 15-24
Creating an Indexed File • • • • • • • • • • • 15-24
15.5.1
15.5.2
Writing Indexed Files • • • • • • • • •
• 15-26
15.5.2.1
Duplicate Values in Key Fields • • •
15-26
15.5.2.2
Preventing the Indexing of Alternate Key
Fields • • • • • • • •
15-27
15.5.3
Reading Indexed Files
• • • • • •
15-28
15.5.4
Updating Records • • • • • • • • • ••
•• 15-29
15.5.5
Deleting Records • • • • • • • • • • • •
15-29
15.5.6
Current Record and Next Record Pointers • • • 15-29
15.5.7
Exception Conditions • • • • • • • • • • • • • 15-30

February 1987

CHAPTER 16

USING THE FORTRAN COMPILER

16.1
16.1.1
16.1.2
16.1.3
16.2
16.2.1
16.2.2
16.2.3
16.3
16.4
16.5
16.5.1
16.5.2

USING THE FORTRAN-10 COMPILER
• • • •
TOPS-10 COMPILE-Class Commands • •
RUNNING THE FORTRAN-10 COMPILER
• • ••
TOPS-10 Compiler Command Switches
• • ••
USING THE FORTRAN-20 COMPILER
TOPS-20 COMPILE-Class Commands • •
RUNNING THE FORTRAN-20 COMPILER •
TOPS-20 Compiler Commands Switches
•••••
THE /DEBUG SWITCH • • • • • • • • • •
•
THE /NOWARN SWITCH • • • • • • • • • •
•
THE /EXTEND SWITCH (TOPS-20 ONLY)
•••••••
/EXTEND and Applications with Large Arrays • •
/EXTEND and Applications with Large Executable
Code • • • • • • • • • • • • • • •
• ••
/EXTEND PSECT Placement • • • • •
16.5.2.1
Building Large-Code Applications • •
16.5.2.2
• •
Arguments to /EXTEND • • • • • • • • •
16.5.3
Linking with TWOSEG REL Files
16.5.4
• • •
THE /FLAG (/FLAG-NON-STANDARD) SWITCH • • • • •
16.6
READING A FORTRAN COMPILER LISTING • •
16.7
• • •
Compiler-Generated Variables • • • •
16.7.1
• • •
ERROR REPORTING • • • • • • • • • •
16.8
•
Fatal Errors and Warning Messages
16.8.1
•
Message Summary
• • • •
16.8.2
•
CREATING A SHARABLE HIGH SEGMENT FOR A FORTRAN
16.9
PROGRAM • • • • • • • • • • • •
• • • • •
CHAPTER 17
17.1
17.1.1
17.1.2
17.1.3
17.2
17.3
17.4
17.5
17.6
17.7 I
17.8
17.9
CHAPTER 18

16-1
16-1
16-2
16-2·
16-5
16-5
16-6
16-8
16-11
16-15
16-16
16-17
16-17
16-18
16-19
16-19
16-21
16-21
16-22
16-24
16-33
16-34
16-34
16-35

USING THE FORTRAN INTERACTIVE DEBUGGER (.FORDDT)
17-2
..·········
17-2
·· ·· ·· ·· ·· ··
17-3
17-4
17-4
. · · ··
17-5
· · · ··
17-6
17-7
· · ·· ·· ·· ·· ·· · · · · · · ·· 17-16
17-17
. · · · · ·· ·· ·· · . · · · · ·· 17-17
· 17-17

INPUT FORMAT
Variables and Arrays
Constant Conventions
Statement Labels and Source Line Numbers
FORDDT AND THE FORTRAN /DEBUG SWITCH
LOADING AND STARTING FORDDT
SCOPE OF NAME AND LABEL REFERENCES
FORDDT COMMANDS
ENVIRONMENT CONTROL
FORTRAN /OPTIMIZE SWITCH
CALLING FORDDT
FORDDT AND FORTRAN-20 EXTENDED ADDRESSING

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)

FEATURES OF FOROTS • • •
••• • • • • • ••
ERROR PROCESSING • • • •
••• • • • • • ••
INPUT/OUTPUT FACILITIES
••••••••••
Input/Output Channels Used By FOROTS (TOPS-10
Only)
••••••••••
• ••••
18.3.2
File Access
•••••••
• • • ••
18.3.3
Closing Files After Non-standard Termination.
18.3.3.1
Sequential Access • • • • • •
• • ••
18.3.3.2
Direct (Random) Access Mode • • • • • • • •
18.4
ACCEPTABLE TYPES OF DATA FILES AND THEIR FORMATS
18.4.1
ASCII Data Files. • • • • • • •
18.4.2
FORTRAN Binary Data Files
• • • • • • ••
18.4.2.1
Format of Binary Files. • • •
• • ••
18.4.3
RMS Data Files (TOPS-20 only)
••••
18.5
USING FOROTS • • • • •
•••••
•
18.5.1
FOROTS Entry Points • • • • •
• •
18.1
18.2
18.3
18.3.1

18-1
18-2
18-2.1
18-2.1
18-3
18-3
18-3
18-4
18-4
18-4
18-4
18-4
18-14
18-15
18-15

February 1987

18.S.2
18.S.3
18.S.3.1
18.S.3.2
18.S.3.3

Calling Sequences • • • • • • • • • • •
18-16.1
MACRO Calls for FOROTS Functions • • • • * • • 18-17
Sequential-Access Calling Sequences
• 18-18
Internal File Calling Sequences • • •
• 18-19
NAMELIST I/O, Sequential-Access Calling
Sequences • • • • • • • • • • • • • •
• 18-20
18.S.3.4
Array Offsets and Factoring
• • • • • • • • 18-20
I/O Statements, Direct-Access Calling
18.S.3.S
Sequences • • • • • • • • • • • • • •
• 18-22
Default Devices Statements, Calling
18.S.3.6
Sequences • • • • • • • • • • • • •
18-23
18.5.3.7
Statements to position Files. • • •
• • 18-24
18.5.3.8
List-Directed Input/Output Statements • • • 18-24
18.5.3.9
Input/Output Data Listg
•••••••
• 18-25
18.5.3.10
OPEN and CLOSE Statements, Calling Sequences 18-28
18.S.3.11
Memory Allocation Routines • • • • • • • • • 18-29
18.S.3.12
Channel Allocation and Dea11ocation Routines 18-30
18.6
FUNCTIONS TO FACILITATE OVERLAYS • •
• • • 18-31
18.7
LOGICAL/PHYSICAL DEVICE ASSIGNMENTS
• • 18-36
18.8
FOROTS AND INQUIRE BY FILE STATEME~T • • • • • • 18-36
18.9
USEROPEN PROCEDURES • • • • • • • • • •
• • 18-38
18.9.1
Example of a USEROPEN • • • • • • • • • • • • 18-38
18.9.2
RMS/FOROTS Data and Control Structures.
• 18-40
CHAPTER 19

USING THE FORTRAN REAL-TIME SOFTWARE (TOPS-10 ONLY)

19.1
INTRODUCTION
19.2
USING FORRTF • • • •
19.2.1
Memory • • • • • • • • • •
19.2.2
Modes • • • • • • •
priority-Interrupt Levels
19.2.3
19.2.4
Masks
• • • • • • • •
SUBROUTINES
19.3
19.3.1
LOCK • • • • • • • • • •
19.3.2
RTINIT • • • • • •
19.3.3
CONECT • • • • •
RTSTRT •
19.3.4
19.3.S
BLKRW • • • • • •
•
19.3.6
RTREAD •
• • • • •
19.3.7
RTWRIT
••••
STATO
19.3.8
19.3.9
STATI • • • •
RTSLP
19.3.10
RTWAKE
19.3.11
19.3.12
DISMIS • • • • • • • • •
19.3.13
DISCON
19.3.14
UNLOCK • • • • • • • • • •
Error Messages •
19.3.1S

•
• • • •

• • •
•

19~7

• • • • • • • •

19-7
19-7

ALPHABETICAL DESCRIPTION OF FOROTS MESSAGES

D-10

APPENDIX A

SUMMARY OF FORTRAN STATEMENTS

APPENDIX B

ASCII-1968 CHARACTER CODE SET

APPENDIX C

COMPILER MESSAGES

APPENDIX D

FOROTS ERROR MESSAGES

D.1

19-1
19-2
19-2
19-2
19-2
19-2
19-3
19-3
19-3
19-4
19-4
19-4
19-5
19-5
19-5
19-6
19-6
19-6
19-6

xii

February 1987

APPENDIX E

INTERACTIVE DEBUGGER (FORDDT) ERROR MESSAGES

APPENDIX F

FORTRAN-SUPPLIED PLOTTER SUBROUTINES

F.1
PLOTS SUBROUTINE •
• • F-2
F.2
AXIS SUBROUTINE
• • • •
• • F-2
CAXIS SUBROUTINE • • • • • • • • • •
F-3
F.3
F.4
LINE SUBROUTINE
• • • • •
• F-4
MKTBL SUBROUTINE •
• F-5
F.5
Character Tables
• • F-5
F.5.1
Creating a Character Table
• F-6
F.5.1.1
Creating a Stroke Table
• • • •
• F-6
F.5.1.2
Sample Character Stroke Table • • • • • • • • F-7
F.5.1.3
FORTRAN- and User-Defined Character Sets • • • F-9
F.5.1.4
F.6
NUMBER SUBROUTINE • • • •
• • • • • F-9
F.7
PLOT SUBROUTINE
F-10
F.8
SCALE SUBROUTINE • •
F-10
F.9
SETABL SUBROUTINE
F-11
SYMBOL SUBROUTINE
• • • • •
F-12
F.10
F.11
WHERE SUBROUTINE • • • • • •
F-12
. INDEX
FIGURES
2-1
4-1
6-1
7-1
7-2
7-3
7-4
7-5
9-1
10-1
11-1
11-2
15-1
F-1
F-2
F-3

Fields Within a FORTRAN Line • • • • • • • •
• 2-3
A 3 x 3 x 2 Array • • • • • • • • • • • •
• • 4-4
Ordering of FORTRAN Statements • • • • • •
• • 6-3
Shared Storage using EQUIVALENCE Statement •
7-12
Equivalence of Substrings • • • •
7-13
Equivalence of Character Arrays • • • • •
7-14
Valid Equivalencing
••••••
7-16
Invalid Equivalencing
••••••
• • • • 7-16
Examples of Block IF Constructs • • • • •
• • 9-7
Components of Data Transfer Statements • •
10-9
TOPS-10 File Protection Number • •
• • 11-31
TOPS-20 Protection Number
••••
• • 11-33
Run-time Memory Layout for Section Zero
• • 15-21
Plotter Character Table Entry
• • • F-6
Character Stroke Table Entry • • • • •
• F-6
Sample Character Stroke Table
••••••
• F-8

2-1
3-1

FORTRAN Character Set • • • • • • • • • • • • •
Comparison of Real, D-f1oating, and G-floating
Numbers
•••••••••••••••••••
Use of Symbolic Names
•••••••••••
Arithmetic Operations and Operators • • • •
Type of the Result Obtained from Mixed-Mode
Operations • • • • • • • • • • • • • • • •
•
Permitted Base/Exponent Type Combinations
•
• •
Logical Operators • • • • • • • • • •
Logical Operations Truth Table • • • • • •
Relational Operators and Operations • • • •
Hierarchy of FORTRAN Operators • • • • • • • • •
Rules for Conversion in Mixed-Mode Assignments •
FORTRAN I/O Statement Categories •
• •••
Summary of Data Transfer Statement Forms
FORTRAN Logical Device Assignments
•

TABLES

4-1
5-1
5-2
5-3
5-4
5-5
5-6
5~

8-1
10-1
10-2
10-3

xiii

• 2-2
3-4
• 4-1
• 5-2
•
•
•
•

5-3
5-5
5-8
5-9
5-11
5-14
• 8-2
10-2
10-3
10-12

February 1987

10-4
10-5
10-6
10-7
10-8
10-9
10-10
10-11
10-12
10-13
10-14
11-1
11-2
11-3
11-4
11-5
11-6
12-1
12-2
12-3
12-4
12-5
12-6
13-1
13-2
15-1
15-2
16-1
16-2
16-3
16-4
16-5
16-6
16-7
16-8
16-9
17-1
18-1
18-2
19-1
D-1

Summary of READ Statement Forms
• • • •
Summary of WRITE Statement Forms • •
•
Summary of REREAD Statement Forms
•
Summary of ACCEPT Statement Forms
••••
•
Summary of TYPE Statement Forms
• •
Summary of PRINT Statement Forms •
•
Summary of PUNCH Statement Forms • • • • • • • •
Summary of Internal READ/WRITE and ENCODE/DECODE
Statement Forms • • • • • • • • • •
• •
Summary of DELETE Statement Forms
•
Summary of REWRITE Statement Forms • • • • • • •
Summary of UNLOCK Statement Forms • • • •
•
Summary of OPEN Statement Specifiers and
Arguments
••••••••••
• • • •
DEVICE and MODE Cross-Table • • • •
•
TOPS-10 Protection Code Values • • •
•
TOPS-20 Protection Code Values • • •
•
Summary of CLOSE Statement Specifiers and
Arguments
•••••••••••••
Summary of Device-Control Statements •
•
Repeatable FORTRAN Edit Descriptors
• • ••
Nonrepeatable FORTRAN Edit Descriptors •
Carriage-Control Specifiers • • • • • •
•
Record, Format List, and I/O List Interaction •
Default Field Widths for Numeric Edit
Descriptors
••••••••••••••••••
Effect of Data Magnitude on G-Format Output
Conversion • • • • • • •
•
•
FORTRAN Instrinsic Functions •
• • • •
FORTRAN-Supplied Subroutines.
• • • • • • •
Argument Types and Types Codes
• • • • • • •
Memory Allocations for /EXTEND and /NOEXTEND • •
FORTRAN-10 Compiler Switches •
• • • ••
FORTRAN-20 Compiler Switches •
Arguments to /DEBUG Switch • •
• • • • •
Legal Dummy and Actual Argument Type
Associations • • • • • • • • • • • •
• •••
Legal Dummy and Actual Argument Structure
Associations • • • • • • • • • •
• ••••
Arguments to /NOWARN Switch
•••
/EXTEND Default Memory Layout
•••
Arguments to /EXTEND Switch • • • • • • • • • •
Arguments to /FLAG Switch
••••••
FORDDT Commands • • • • • • • • • • • • • • • •
FOROTS Entry Points
••••••••••••••
Function Numbers and Function Codes
•
Error Messages - Code Format and Full Message
Format
• • • •
FOROTS Error Codes • • • • •
•

xiv

10-28
10-36
10-43
10-45
10-48
10-50
10-52
10-54
10-59
10-61
10-62
11-5
11-27
11-32
11-33
11-45
11-61
12-8
12-9
12-11
12-13
12-30
12-35
13-4
13-23
15-14
15-23
16-3
16-9
16-12
16-14
16-15
16-15
16-16
16-20
16-22
17-1
18-15
18-32
19-8
• D-3

February 1987

PREFACE

The TOPS-10/TOPS-20 FORTRAN Language Manual reflects the software as
of Version 11 of the FORTRAN-10/20 compiler, the FORTRAN-10/20 Object
Time System
(FOROTS),
and the FORTRAN-10/20
debugging
program
(FORDDT) •
This manual describes the FORTRAN language as
implemented
for
the
TOPS-10 operating system (FORTRAN-10) and the TOPS-20 operating system
(FORTRAN-20). Any differences between FORTRAN-10 and FORTRAN-20 are
noted in this manual.
Since this is a reference manual, we assume that you have used FORTRAN
before.
If you haven't, you should read one of the many introductory
FORTRAN texts.
CONVENTIONS
The following conventions are used throughout the manual:
Braces {

indicate that a choice must be made from
of the enclosed lines.

Brackets [ ]

indicate an optional feature.

one

or •

indicate the omission of information from a
programming example or that items in a
command line can be optionally repeated.

Lowercase letters

indicate variable information you supply in a
command string.

UPPERCASE LETTERS

indicate fixed (or literal) information that
you must enter as shown in a command string.

Ellipsis

indicates a blank.
The standard for FORTRAN is the American National Standards Institute
(ANSI)
FORTRAN, X3.9-1978 (also known as FORTRAN-77).
FORTRAN-10/20
extensions and additions to ANSI FORTRAN are in blue print
in this
manual.

FORTRAN-10/20 VII

February 1987

MANUALS REFERENCED
The following manuals are referenced from TOPS-10 publications:

• TOPS-10 Operating System Comma!1 ds Manual
• SOS Reference Manual
• TOPS-10 Monitor Calls Manual

• TOPS-10 Hardware Reference Manual
•

TOPS-10 LINK Reference Manual

• TOPS-10 SORT/MERGE User's Guide
•

TOPS-10 FORTRAN Installation Guide

---

The following manuals are referenced from TOPS-20 publications:

•
•

TOPS-20 Commands Reference Manual
TOPS-20 EDIT Reference Manual

• TOPS-20 User's Guide
•
•
•
•
•

TOPS-20 Monitor Calls Manual
TOPS-20 Link Reference Manual
TOPS-20 SORT/MERGE Use r' s Guide

---

TOPS-20 FORTRAN Installation Guide
TOPS-20 RMS User's Guide

• RMS-20 Pro9rammer's Reference Manual
The following TOPS-10/TOPS-20 manual are referenced:
•

FORTRAN-10/20 and VAX FORTRAN Compatibility Manual

•

TOPS-10/TOPS-20 FORTRAN Pocket Guide

•

TOPS-10/TOPS-20 COBOL-74 Language Manual

•

TOPS-10/20 BLISS Lan9uage Guide

•

TOPS-10/20 Common Math Library Manual

FORTRAN-10/20 VII

xvi

February 1987

CHAPTER 1
INTRODUCTION

1.1

OVERVIEW

The FORTRAN language, as implemented on the TOPS-IO and TOPS-20
operating systems,
is compatible with and encompasses the standard
described in "American National Standard FORTRAN, X3.9-l978" (referred
to as the FORTRAN-77 standard) at the full-language level.
FORTRAN-IO/20 provides many extensions and additions to the FORTRAN-77
standard that greatly enhance the usefulness of FORTRAN and increase
its compatibility with FORTRAN languages implemented by other computer
manufacturers.
The
extensions
and additions to the standard
FORTRAN-77 are printed in this manual in blue print.
A FORTRAN source program consists of a set of statements constructed
using the language elements and the syntax described in this manual.
A given FORTRAN statement performs anyone of the following functions:
1.

It causes operations such as
branching to be carried out.

multiplication,

division,

and

It specifies the type and format of the data being processed.

It specifies the characteristics of the source program.

FORTRAN statements are composed of keywords (words that are recognized
by the compiler) used with elements of the language set: constants,
variables, and expressions. There are two basic types of FORTRAN
statements: executable and nonexecutable.
Executable statements
specify
the
actions
of
the
program;
nonexecutable statements describe the characteristics and arrangement
of data, editing information, statement functions, and the kind of
subprograms that may be included in the program.
The compilation of
executable statements results in the creation of executable code in
the object program.
Nonexecutable statements provide information only
to the compiler; they do not create executable code.

1.2

MANUAL ORGANIZATION

This

manual

is divided
into three parts:
Language
Elements,
and Language Usage.
Sections 1.2.1 through 1.2.3 contain
general descriptions for each of these three parts.
Section 1.2.4
contains a general description of the appendixes.
State~~nts,

1-1

INTRODUCTION
1.2.1

FORTRAN Language Elements (Part I)

Part I of this manual describes the fundamental
programs, including (by chapter):

elements

FORTRAN

Chapter 2, CHARACTERS AND LINES, describes FORTRAN source program
characters and lines.
The FORTRAN compiler interprets your
source program and translates it into machine code
(executable
code) •
Chapter 3, CONSTANTS, describes FORTRAN data types and constants.
FORTRAN enables you to manipulate information (data) in a variety
of ways. This chapter describes the techniques for defining
FORTRAN constants of various data types.
Chapter 4,
SYMBOLIC NAMES,
VARIABLES, AND ARRAYS, describes
symbolic names, variables, and arrays in FORTRAN.
The symbolic
name is used in a variety of ways in a source program;
this
chapter describes the conventions for using symbolic names to
define both variables and arrays.
Chapter 5,
EXPRESSIONS,
introduces
and
describes
FORTRAN
operators.
Expressions in FORTRAN are formed using variables,
constants, and operators.

1.2.2

FORTRAN Statements (Part II)

Part II of this manual describes all the statements in the FORTRAN
language.
The following list describes (by chapter) the information
presented in Part II:
Chapter 6, EXECUTABLE AND NONEXECUTABLE STATEMENTS, defines the
term "FORTRAN Statement", and describes the "Compilation Control
Statements".
Chapter 7, SPECIFICATION AND DATA STATEMENTS, describes all the
statements
within
the
category
"Specification
and Data
Statements".
The specification statements
enable
you
to
explicitly define the data types of variables used within your
program; the DATA statement enables you to create initially
defined constants within your program.
Chapter 8, ASSIGNMENT STATEMENTS, describes all the statements
within the category "Assignment Statements". FORTRAN assignment
statements enable you to assign values to variables, and to
assign statement labels to symbolic names.
Chapter 9, CONTROL STATEMENTS, describes all the statements
within the category "Control Statements".
The default execution
sequence of lines in a FORTRAN program is each line from
left-to-right, and all lines from top-to-bottom.
You use the
FORTRAN control statements to alter the default
execution
sequence,
to stop or pause during program execution, or to mark
the end of an executable program.
Chapter 10, DATA TRANSFER STATEMENTS, describes the data transfer
category of "FORTRAN Input/Output (I/O) Statements".
As the term
implies, a data transfer statement moves data from one place to
another.

1-2

INTRODUCTION
Chapter 11, FILE-CONTROL AND DEVICE-CONTROL STATEMENTS, describes
file-control
and device-control categories of "FORTRAN I/O
Statements". The file-control statements enable you to associate
a unit number with a file. Device-control statements enable you
to position a storage medium (for example, magnetic tape)
on a
connected unit.
Chapter 12, FORMATTED DATA TRANSFERS, describes three types of
data
formatting.
During
certain types of data transfer
operations, you must specify the format of the data being
transferred.
FORTRAN provides three techniques for specifying
the format of data:
FORMAT-Statement,
List-Directed,
and
NAMELIST-Statement formatting.
Chapter 13, FUNCTIONS AND
SUBROUTINES,
describes
FORTRAN
functions and subprograms. Functions and subprograms provide a
technique for producing clear and concise FORTRAN programs.
FORTRAN-lO/20 provides both predefined functions and subprograms,
and the statements for defining your
own
functions
and
subprograms.
Chapter 14, BLOCK DATA SUBPROGRAMS, describes the block-data
subprogram.
This type of subprogram enables you to define
initial values for variables in COMMON.

1.2.3

FORTRAN Language Usage (Part III)

Parts I and II of the manual contain complete descriptions of FORTRAN
elements and statements. Part III of the manual contains explanations
of how you use FORTRAN-lO/20.
The following usage topics are covered
in Part III:
Chapter 15, WRITING USER PROGRAMS, presents
some
general
considerations that you should follow when you are creating
FORTRAN source programs.
In addition, this chapter contains a
description of the FORTRAN optimizer.
Chapter 16, USING THE FORTRAN COMPILER, describes how to use the
FORTRAN compiler and contains descriptions on how to compile,
load, and execute a FORTRAN program.
In addition, this chapter
contains descriptions of how to read a compiler-generated program
listing, and how to create a reentrant FORTRAN program.
This
Chapter also describes how to use FORTRAN-20 extended addressing.
Chapter 17, USING THE FORTRAN INTERACTIVE DEBUGGER
(FORDDT),
describes how to use the FORTRAN interactive debugging program
(FORDDT) to test and debug a running program. This chapter also
contains a brief explanation of how to debug a running FORTRAN
program using DDT, the system debugger.
Chapter 18, USING THE FORTRAN OBJECT-TIME SYSTEM
(FOROTS),
describes the FORTRAN Object-Time System (FOROTS). This chapter
also contains descriptions of how you can use the FOROTS
software.
Chapter 19, USING THE FORTRAN REAL-TIME SOFTWARE (TOPS-lO ONLY),
describes how to use the FORTRAN real-time software. This
chapter is for TOPS-lO installations only.

1-3

INTRODUCTION
1.2.4

APPENDIXES

The appendixes describe various useful
tqpics are covered in the appendixes:

The

information.

following

Appendix A, SUMMARY OF FORTRAN STATEMENTS, summarizes the forms
of all FORTRAN statements and provides a section reference where
each statement is described in detail.
Appendix B, ASCII-1968 C~ARACTER CODE SET, lists the character
code set defined in the X3.4-1968 version of the American
National Standard Code for Information Interchange (ASCII).
Appendix C, COMPILER MESSAGES,
messages.

describes

the

FORTRAN

compiler

Appendix D, FOROTS ERROR MESSAGES,
messages.

describes

the

FOROTS

error

Appendix E, FORDDT ERROR MESSAGES,
messages.

describes

the

FORDDT

error

describes

the

Appendix F, FORTRAN-SUPPLIED PLOTTER SUBROUTINES,
FORTRAN-supplied plotter subroutines.

1-4

CHAPTER 2
CHARACTERS AND LINES

The basic elements of the FORTRAN source program are its characters
and lines.
Characters are used to form statements, expressions, and
comments in FORTRAN source programs. Lines, and fields within lines,
are used to define the context in which characters are interpreted by
the FORTRAN compiler.
program

This chapter describes the relationships among
source
characters, lines, and fields within source program lines.

2.1

CHARACTER SET

Table 2-1 lists the digits,
letters, and symbols recognized by
FORTRAN.
The remainder of the ASCII-1968 character set is acceptable
within character or Hollerith constants or comment text, but these
characters cause fatal errors in other contexts.
NOTE
The complete ASCII character set is defined in the
X3.4-1968 version of the "American National Standard
Code for Information Interchange". A summary of the
standard ASCII set is also contained in Appendix B of
this manual.
NOTE
Lowercase alphabetic
upper-case
outside
character constants.

characters
are
the context of

2-1

treated
Hollerith

as
or

CHARACTERS AND LINES
Table 2-1:

FORTRAN Character Set
Letters

Uppercase:

ABC D E F G H I J K L M N 0 P Q R STU V W X Y Z

Lowercase:

abc d e f g h i j k 1 m n 0 p q r s t u v w x y z
Digits

o 1 234 5 6 789
Symbols
!

"
#
$
&

,
(

)

Exclamation point
Quotation Mark
Number Sign
Dollar Sign
Ampersand
Apostrophe
Left Parenthesis
Right Parenthesis
Asterisk
Plus
Underscore

:
i

<
=

>
...

Comma
Hyphen (Minus)
Period (Decimal Po i nt)
Slant (Slash)
Colon
Semicolon
Less Than
Equal To
Greater Than
Circumflex

Line Termination Characters
Line Feed (LF), Form Feed (FF), Vertical Tab (VT)
Line Formatting Characters
Carriage Return (RET), Horizontal Tab (TAB), Blank
Note that horizontal tabs normally advance the character position
pointer to the next position that is an even mUltiple of 8. An
exception to this is the initial tab, which is defined as a tab that
either includes or starts in character position 6.
(Refer to Section
2.3.1 for a description of initial and continuation line types.)
Tabs
with~n
character specifications count as one character, even though
they may advance the character position as many as eight places.

2.2

STATEMENT DEFINITION AND FORMAT

Source program statements are divided into physical lines. A line is
defined as a string of adjacent character positions, terminated by the
first occurrence of a line termination character, regardless of
context.
As shown in Figure 2-1, each source program line is divided
into four fields.

FORTRAN-10/20 VII

2-2

February 1987

CHARACTERS AND LINES

I. .------------ Line Character Positions ----------11.....1
\1

....

5~.1

Statement Continuation
Label Field
Field

Figure 2-1:
2.2.1

70
y

Statement Field

• J
72 " 73
...
Remarks
MR-S-1 761 -81

Fields Within a FORTRAN Line

Statement Label Field and Statement Numbers

You can place a number ranging from 1 to 99999 in the statement label
field of an initial line to identify the statement. Any source
program statement that is referenced by another statement must have a
statement number.
Leading zeros and all blanks in the label field are
ignored; for example, the numbers 00105 and 105 are both accepted as
statement number 105.
You can assign the statement numbers in a source program in any order;
however, each statement number must be unique with respect to all
other statements in the program or subprogram.
A main program and a subprogram can contain identical statement
numbers.
In this case, references to these numbers are understood to
mean the numbers in the same program unit in which the reference is
made. An example follows:
Assume that main module MAINMD and subprogram SUBI both
contain statement number 105. A GO TO 105 statement, for
instance, in MAINMD will refer to statement 105 in MAINMD,
not to 105 in SUBI.
A GO TO 105 in SUBI will transfer
control to 105 in SUBI.
An initial tab may be used to skip all or part of the label field.
If
an initial tab is encountered during compilation, FORTRAN-IO/20
examines the character immediately following the tab to determine the
type of line being entered.
If this character is one of the digits 1
through 9, FORTRAN-lO/20 considers the line as a continuation line and
uses the second character after the tab as the first character of the
statement field.
If the character following the tab is not one of the digits I
through
9, FORTRAN-IO/20 considers the line to be an initial line and the
character following the tab is considered to be the first character of
the statement field.
The character following the initial tab is
considered to be in character position 6 for a continuation line, and
in character position 7 for an initial line.

2.2.2

Line Continuation Field

Any character of the FORTRAN character set (except a blank, a zero, or
an exclamation point) placed in the line continuation field (position
6) identifies the line as a continuation
line
(see Section 2.3.1).
Whenever you use an initial tab to skip all or part of the label field
of a continuation line, the next character you enter must be one of
the digits 1 through 9 to identify the line as a continuation line.

2-3

CHARACTERS AND LINES
2.2.3

Statement Field

Any FORTRAN statement can appear in the statement field.
Blanks
(spaces)
and tabs do not affect compilation of the statement. Blanks
and tabs may be used freely in this field for appearance purposes,
with the exception of textual data given within either a character or
Hollerith specification, where blanks and tabs are
significant
characters.

2.2.4

Remark Field

In lines consisting of 73 or more character positions, only the first
72 character positions are interpreted by FORTRAN. Note that tabs
generally occupy more than one character position, usually advancing
the cursor to the next character position that is an even multiple of
8. The exception is the tab in a label field, which advances the
cursor either to column 6 or 7, depending on the character following
the tab (see Section 2.2.1).
All other characters after character
remarks and do not affect compilation.

position

are

treated

Note that remarks may also be added to a line in character positions 1
through 72, provided the text of the remark is preceded by the symbol
"!" (see Section 2.3.3).

2.3

LINE TYPES

A line in a FORTRAN source program may be:
1.

An initial line

A continuation line

A multi-statement line

A comment line

A debug line

A blank line

These lines are described in Sections 2.3.1 through 2.3.5.

2.3.1

Initial and Continuation Lines

FORTRAN statement may occupy the statement fields of up to 100
consecutive lines.
The first line in a multi-line statement is
referred to as the initial line; the succeeding lines are referred to
as continuation lines.
Initial lines may be assigned a statement
number and must have either a blank or a zero in character position 6.

An initial tab may be used to skip all or part of the label field.
If
you use an initial tab for this purpose, you must immediately follow
it with a nonnumeric character; that is, the first character of the
statement field must be nonnumeric.

FORTRAN-10/20 VII

2-4

Feb!:'uary 1987

CHARACTERS AND LINES
You cannot assign a statement label to a continuation line.
Instead,
you identify a continuation line by placing a character from the
FORTRAN character set (except blank, zero, or exclamation point)
in
character position 6 of that line.
This position is the line
continuation field.
The label field of a continuation line must be
blank.
Note that blank lines, comments, and debug lines that are treated like
comments
(that is, debug lines that are not compiled with the rest of
the program) are legal continuation lines and do not terminate a
continuation sequence (see Section 2.3.4).
The following is an example of a 3-line FORTRAN FORMAT statement
two continuation lines:
105

FORMAT (lX,'This example shows how continuation lines',
2 'are used to accommodate FORTRAN statements that do not'
3 'entirely fit on a single line.')

In this example the characters 2 and 3 in position
lines as continuation lines.

2.3.2

with

identify

those

Multi-Statement Lines

You may write more than one FORTRAN statement in the statement field
of one line.
The rules for structuring a multi-statement line are:
1.

Successive statements must be separated by a semicolon (;).

Only the first statement in the series can have
number.

The last statement in a line is continued to the next line if
that next line is made a continuation line.

statement

An example of a multi-statement line is:
450

DIST=RATE

* TIME; TIME=TIME+0.05; CALL PRIME(TIME,DIST)
NOTE

If
a
statement
sequence
in
a
multi-statement
line
consists of a
logical IF (see Section 9.2.2)
followed
by any other executable statement, then
the statement following the IF will be
executed in all cases, even if the IF
expression evaluates as false.

2.3.3

Comment Lines and Remarks

Lines that contain descriptive text only are called comment lines.
Comment lines commonly identify and introduce a source program,
describe the purpose of a particular set of statements, and introduce
subprograms.

2-5

CHARACTERS AND LINES
To structure a comment line:
1.

You must place one of the characters C (or c), *, $, I, or
in character position 1 of the line to identify it as a
comment line.

You place the text of the comment in
line.

You may place comment lines anywhere in the
including preceding a continuation line.

You may write a large comment as a sequence of any number of
lines;
however,
each line must carry the
identifying
character (C (or c), *, $, I, or !) in its first character
position.

the

remainder
source

The following is an example of a comment that occupies more
line:
C
C
c
c

the

program,

than

one

SUBROUTINE - A12
This subroutine formats
and stores the results of
the HEAT-TEST program

Comment lines are printed on all listings, but are
by the compiler.

otherwise

ignored

You may add a remark to any statement field, in character positions 7
through 72, provided the symbol!
precedes the text.
For example, in
the line
IF(N.EQ.O)STOP

! Stop if card is blank

the text "Stop if card is blank" is identified as a
remark by the
preceding
!
symbol.
The compiler ignores all characters from the
exclamation point to the end of the line.
The characters following
the exclamation point, however, appear in the source program listing.
To be treated as a remark symbol,
the exclamation point must not
appear in a Hollerith or character constant.
Note that characters appearing in character positions 73 and beyond
are automatically treated as remarks, so that you need not use the
symbol!
(see Section 2.2.4).

2.3.4

Debug Lines

As an aid in program debugging, a D (or d) in character position 1 of
any line causes the line to be interpreted as a comment line; that is,
not compiled with the rest of the program unless the IINCLUDE switch
is present
in the compiler command string.
(See Chapter 16 for a
description of the compiler switches.)
When the IINCLUDE switch is present in the compiler command string,
the 0 (or d) in character position 1 is treated as a blank so that the
remainder of the line is compiled as an ordinary (noncomment) line.
A
debug
line can have a label following the 0 (or d).
Note that if the
debug statement is an initial line, all of its continuation lines must
contain a 0 (or d) in character position 1.

2-6

CHARACTERS AND LINES
2.3.5

Blank Lines

You may insert lines consisting of only blanks, tabs, or no characters
anywhere in a FORTRAN source program.
Blank lines that contain
remarks only, are considered as blank lines. Blank lines are used for
formatting purposes only;
they cause blank lines to appear in their
corresponding positions in source program listings; otherwise, they
are ignored by the compiler.

2.4

LINE-SEQUENCED SOURCE FILES

FORTRAN-IO/20
accepts
line-sequenced
files
as
produced
by
line-oriented text editors
(for example, SOS on TOPS-IO or EDIT on
TOPS-20).
These sequence numbers are used in place of the listing
line numbers normally generated by FORTRAN.
The listing line numbers
are not the same as FORTRAN statement numbers.

2-7

CHAPTER 3
CONSTANTS

3.1

INTRODUCTION

Constants are quantities that do not change value during the execution
of the object program.
The data types you can use for constants in
FORTRAN-10/20 source programs are:
1.

Integer

Real

Double-precision

Complex

Character

Logical

Octal

Double-octal

Hollerith

10.

Statement label

The use and format of each
Sections 3.2 through 3.10.

3.2

these

data

types

are

discussed

INTEGER CONSTANTS

An integer constant is a string of one to eleven digits that
represents a whole decimal number
(a number without a fractional
part).
Integer constants must be within the range of -(2**35-1)
to
(+2**35)-1 (-34359738367 to +34359738367).
Positive integer constants
may optionally be signed; negative integer constants must always be
signed.
You cannot use decimal points, commas, or other symbols in
integer constants (except for a preceding sign, + or -).
Examples of valid integer constants are:
345
+345
-345

3-1

CONSTANTS
Examples of invalid integer constants are:
+345.
3,450
34.5

3.3

(use of decimal point)
(use of comma)
(use of decimal point; not a whole number)

REAL CONSTANTS

A real constant can have any of the following forms:
1.

A basic real constant: a string of decimal digits followed
by
a decimal point,
followed optionally by a decimal
fraction, for example, 1557.42.

A basic real constant followed by a decimal integer exponent
written in E notation
(exponential notation)
form,
for
example, 1559.E2 or 1559.e2.
The number following the E (or
e)
specifies a power of ten by which the basic real constant
will be multiplied.

An integer constant (no decimal point) followed by a decimal
integer exponent written in E notation, for example, 1559E2
or l55ge2.

Real constants may be of any size; however, each will
fit the precision of 27 bits (7 to 9 decimal digits).

rounded

Precision for real constants is maintained to approximately eight
significant digits;
the absolute precision depends upon the numbers
involved.
The exponent field of a real constant written in E notation cannot be
empty
(blank);
it must be either a zero or an integer constant.
The
range of magnitude permitted a real constant is from approximately
1.47 * 10**(-39) to 1.70 * 10**(+38).
The following are examples of valid real constants:
-98.765
7.0E+0
.7E-3
5E+5
50115.
50.E1

(=
(=
(=

7.)
.0007)
500000.)

500.)

The following are examples of invalid real constants:
72.6E512
.375E
500

3.4

(exponent is too large)
(exponent incorrectly written)
(no decimal point given)

DOUBLE-PRECISION CONSTANTS

Double-precision constants are similar to real constants written in
notation form; the differences between these two constants are:
1.

Double-precision constants, depending on their magnitude,
have precision from 16 to 18 places, rather than the 8-digit
precision obtained for real constants.

3-2

CONSTANTS
constant

occupies

two

storage

Each
double-precision
locations.

The letter 0
(or d),
instead
of
E,
is
used
in
double-precision constants to identify a decimal exponent.

On KL model B systems, there are two forms of double-precision number
formats.
If the /GFLOATING compiler switch is specified (see Chapter
16), the double-precision number format is called G-floating.
If the
/OFLOATING compiler switch
(the default) is specified (see Chapter
16), the double-precision number format is called O-floating.
See
Section 3.4.1 for a comparison of the different double-precision
number formats.
On KS systems, only the O-floating double-precision number
provided.

format

You must use both the letter 0 and an exponent
(including zero)
in
writing a double-precision constant.
The range of magnitude permitted
a double-precision constant is from approximately:
1.47 * 10**(-39) to 1.70 * 10**(+38) for O-floating
or
2.78 * 10**(-309) to 8.99 * 10**(+307) for G-floating
The following are examples of valid double-precision constants:
7.9003
7.90+03
7.90-3
79003
7900

(=
(=
(=
(=
(=

7900.)
7900.)
.0079)
79000.)
79.)

The following are examples of invalid double-precision constants:
7.90999
7.9E5

3.4.1

(exponent is too large)
("E" denotes single precision; "0" denotes double
precision)

Comparison of Real, D-floating, and G-floating

For KL model B systems, G-floating double-precision is provided as an
alternative double-precision number format.
You must specify the
/GFLOATING compiler switch (see Chapter 16) to invoke the G-floating
double-precision format.
If you specify the /OFLOATING compiler
switch (the default),
the O-floating format
is used.
Table 3-1
summarizes the comparisons among real, O-floating, and G-floating.

3-3

CONSTANTS
Table 3-1:

Comparison of Real, D-floating, and G-floating Numbers
Bits of
Exponent

Bits of
Mantissa

Real

1.47 * 10**(-39)
to 1.70 * 10**(+38)

8.1

D-floating

1.47 * 10**(-39)
to 1.70 * 10**(+38)

18.7

G-floating

2.78 * 10**(-309)
to 8.99 * 10**(+307)

17.8

3.5

Range

Digits of
Precision

COMPLEX CONSTANTS

You can represent a complex constant by an ordered pair of integer,
real, or octal constants written within parentheses and separated by a
comma. For example,
(.70712,
-.70712)
and
(8.763E3,
2.297)
are
complex constants.
In a complex constant, the first
(leftmost)
constant of the pair
represents the real part of the number; the second constant represents
the imaginary part of the number. Both the real and imaginary parts
of a complex constant can be signed.
The constants that represent the real and imaginary parts of a complex
constant occupy two consecutive storage locations in the object
program.

3.6

CHARACTER CONSTANTS

A character constant is a string of printable ASCII characters
enclosed by apostrophes.
Both delimiting apostrophes must be present,
and the string must be at least one character in length. The compiler
accepts control characters in character constants with the following
exceptions:
Character

Octal Value

- NUL

"0
"12
1113
1114
1115

"@
"J

- LF

"K - VT
"L - FF
"M - CR

NOTE
The CHAR function (see Chapter 13)
can be used to
build variables that contain these control characters.
The value of a character constant is the string of characters between
the delimiting apostrophes. The value does not include the delimiting
apostrophes, but does include all spaces or tabs
within
the
apostrophes.
Within a character constant, the apostrophe character
is represented
by two consecutive apostrophes
(with no space or other character
between them).
3-4

CONSTANTS
The length of the character constant is the number of characters
between the apostrophes, except that two consecutive apostrophes count
as a single apostrophe.
Each character in the string has a character position that is numbered
consecutively starting at one.
The number indicates the sequential
position of a character in a string, from left to right.
There is one
character storage location for each character in the string.
If a character constant appears in a numeric context (for example, as
the
expression on the right side of an arithmetic assignment
statement), it is considered a Hollerith constant (see Section 3.9).
Examples of valid character constants and their lengths are:
Length

Value

'WHAT? '

WHAT?

'TODAYS"S DATE IS:

TODAY'S DATE IS:

'He said, "hello'"

He said, "hello"

,,,,

Examples of invalid character constants are:
'HEADINGS

,,
"Now or Never"

3.7

(no trailing apostrophe)
(a character constant must contain
one character)

least

(quotation marks cannot be used in
apostrophes)

place

OCTAL AND DOUBLE-OCTAL CONSTANTS

You may use octal numbers
(radix 8)
as constants in arithmetic
expressions,
logical expressions, and data statements. Octal numbers
up to 12 digits in length are considered standard octal constants;
they are stored right-justified
in one storage location.
When
necessary, standard octal constants are padded with leading zeros to
fill their storage location.
If you specify more than 12 digits in an octal number,
it is
considered a double-octal constant. Double-octal constants occupy two
storage locations and may contain up to 24 right-justified octal
digits; leading zeros are added to fill any unused digits.
If you assign a single-octal constant to a double-precision or complex
variable,
it is stored right-justified in the high-order word of the
variable. The low-order portion of the variable is set to zero.
If
you assign a double-octal constant to a double-precision or complex
variable, it is stored right-justified in the two words.
All octal constants must:
1.

Be preceded by a double quote (")
octal, for example, "777

Be signed if negative, but optionally signed if positive

Contain one or more of the digits 0 through 7, but not 8 or 9
3-5

to identify the

digits

CONSTANTS
The following are examples of valid octal constants:
"123456700007
+"12345 (optional sign)
-"7777
"-7777
The following are examples of invalid octal constants:
"12368
7777

(contains an 8)
(no identifying double quote)

When you use an octal constant as an operand in an expression,
its
form
(bit pattern) is not converted to accommodate it to the type of
any other operand.
For example, the subexpression (A + "202400000000)
has as its result the sum of A with the floating point number 2.0;
while the subexpression (I + "202400000000) has as its result the sum
of I with a large integer.
You cannot use octal constants as stand-alone arguments for
library
functions that require non-octal arguments.
MINO,
for instance,
requires integer arguments and cannot accept octal arguments.
When you combine a double-octal constant in an expression with
(or
assign it to) either an integer or real variable, only the contents of
the high order location (leftmost) are used.

3.8

LOGICAL CONSTANTS

The Boolean values of truth and falsehood are represented
in FORTRAN
source programs as the logical constants .TRUE. and .FALSE.. Always
write logical constants enclosed by periods, as in the preceding
sentence.
You may use logical quantities in arithmetic and logical statements.
Only the sign of a numeric value used in a logical IF statement is
tested to determine if it is true (negative) or false (nonnegative).

3.9

HOLLERITH CONSTANTS

A Hollerith constant is a string of alphanumeric and/or special
characters preceded by nH (for example, nHstring).
In the prefix nH,
the letter n represents a number that specifies the exact number of
characters (including blanks) that follow the letter H.
NOTE
If a character constant appears in a numeric context
it is considered a Hollerith constant (see Section
3.6) •
The following are examples of Hollerith constants:
2HAB
14HLOAD TEST #124
6H#124-A

3-6

CONSTANTS
NOTE
A tab in a Hollerith constant
character; for example, 3H
AB.

counted

one

You may enter Hollerith constants into DATA statements as a string of:
1.

Up to ten
7-bit
ASCII
characters
double-precision type variables

Up to five
variables

7-bit

ASCII

characters

for
for

all

complex

other

type

The 7-bit ASCII characters that comprise a Hollerith constant are
stored
left-justified
(starting
in
the
first
word
of
a
double-precision constant (the high-order word) or the real part of a
complex constant)
with blanks placed in empty character positions.
Hollerith constants that occupy more than one variable are stored as
successive variables
in the list.
The following example illustrates
how the string of characters is stored in a six-element array called
A:
DIMENSION A(6)
DATA A/27HA string of many characters/
A (1)
A(2)
A (3)
A(4)
A (5)
A (6)

3.10

is
is
is
is
is
is

set
set
set
set
set
set

to
to
to
to
to
to

'A str'
ling 0'
'f man'
'y chat
'racte'
Irs

STATEMENT LABEL CONSTANTS

Statement labels are
statement numbers.

numeric

identifiers

that

represent

program

You write statement label constants as strings of one to five decimal
digits,
that are preceded by either an asterisk (*), a dollar sign
($), or an ampersand (&).
For example, *11992, $11992, and &11992 are
all
valid statement label constants.
You use statement label
constants only in the argument list of CALL statements to identify the
number of the executable statement to return to in a multiple RETURN
statement (see Chapter 13).

3-7

CHAPTER 4
SYMBOLIC NAMES, VARIABLES, AND ARRAYS

4.1

SYMBOLIC NAMES

Symbolic names consist of any alphanumeric combination of one to
31
characters,
the first of which must be a letter.
The special
characters dollar sign ($) and underscore ( ) are allowed.
If you use
more than 31 characters
in a symbolic name, the compiler prints a
warning message and ignores all but the first
31 characters.
The
compiler
interprets lowercase letters in symbolic names as uppercase
letters.
The following are examples of legal symbolic names:
A12345
IAMBIC PENTAMETER
ABLE
C

The following are examples of illegal symbolic names:
.AMBIC
8AB

(first character is not a letter)
(first character is not a letter)

You use symbolic names to identify specific items of a FORTRAN source
program;
Table 4-1 lists these items, together with an example of a
symbolic name and text reference for each.
Table 4-1:

Use of Symbolic Names

Symbolic Names
Can Identify
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.

Variables
Arrays
Array elements
Substrings
Functions
Subroutines
Intrinsic functions
PROGRAM Statement
PARAMETER Statement
COMMON block names
NAMELIST list

FORTRAN-10/20 VII

For Example

For a Detailed
Description
See Section

PI, CONST, LIMIT
TAX
TAX (3,5)
FOO (1: N)
MYFUNC, VAL FUN
CALCSB, SUB2, LOOKUP
SIN, ATAN, COSH
TEST
Vl,C2,K
DATAR, COMDAT
DATA3

4.2
4.3
4.3.1
4.4
13.2
13.4
13.1
6.4.1
7.8
7.4
12.6

4-1

February 1987

SYMBOLIC NAMES, VARIABLES, AND ARRAYS
4.2

VARIABLES

A variable is a data storage location identified by a symbolic name; a
variable is not a constant, an array, or an array element& Variables
specify values that are assigned to them in such ways as assignment
statements
(Chapter 8), DATA statements (Chapter 7), or at run time
through I/O data transfers (Chapter 10). Before you assign a value to
a variable,
its value is undefined; and you should not reference it
except to assign a value to it.
The value you assign to a variable can be either a constant or the
result of a calculation that is performed during the execution of the
object program. For example, the statement IAB=5 assigns the constant
5 to the variable lAB.
In the statement IAB=5+IB, however, the value
assigned lAB depends on the value of variable IB at the time the
statement is executed.
The type of a variable determines the interpretation of its
Variables can be:
1.

Integer

Real

Logical

Double-precision

Complex

Character

contents.

The type of a variabLe is determined either implicitly, by the first
letter of the variable name
(described below), or explicitly, by
declaring the variable type in a type declaration statement
(see
Chapter 7).
FORTRAN uses the following default
types are not explicitly declared:

conventions

for

variables

whose

Variable names that begin with the letters I, J, K, L, M,
N are integer variables.

Variable names that begin with any letter other than I, J, K,
L, M, or N are real variables.
NOTE
These default conventions can be altered
use of the IMPLICIT statement, which
described in Section 7.3.

The following are examples of determining
according to the preceding conventions:

the

type

by
is
a

variable

Beginning Letter

Assumed Data Type

ITEMP
OTEMP
KA123
AABLE

I
0
K

Integer
Real
Integer
Real

4-2

SYMBOLIC NAMES, VARIABLES, AND ARRAYS
4.3

ARRAYS

An array is an ordered set of data identified by an array name. Array
names are symbolic names and must conform to the rules for writing
symbolic names (see Section 4.1).
Arrays are made up of smaller units of data called array elements. As
with variables, you may assign a value to an array element. Before
you assign a value to an array element it has an undefined value. You
should not reference an array element until you have assigned it a
value.
An array element is referenced by using the array name together with
some number of subscripts that describe the position of the element
within the array.

4.3.1

Array Element Subscripts

The general form of an array element name is AN (Sl, S2, ••• Sn), where
AN is the array name and Sl through Sn represent 1 through n subscript
expressions.
You may use any number of subscript quantities in an
element name; however, the number used must always equal the number of
dimensions (see Section 4.3.2) specified for the array.
A subscript can be any constant or expression
example:

(see

Chapter

S),

for

Subscript quantities may contain arithmetic expressions that
involve addition, subtraction, multiplication, division, and
exponentiation.
For
example,
(A+B,C*S,D/2)
and
(A**3,(B/4+C)*E,3) are valid subscripts.

Arithmetic expressions
(see Chapter 5)
used
in
array
subscripts may be of any type, but noninteger expressions
(including complex)
are converted to integer when
the
subscript is evaluated.

A subscript may contain function references (see Chapter 13).
For example, TABLE
(SIN(A)*B,2,3) is a valid array element
identifier.

Subscripts may contain array element identifiers nested to
any level as subscripts.
For example,
in the subscript
(I(J(K(L»),A+B,C) the first subscript expression given is a
nested 3-level array reference.

Some examples of valid array elements are:
1.

IAB(1,S,3)

ABLE (A)

TABLEl(lO/C+K**2,A,B)

MAT(A,AB(2*L),.3*TAB(A,M+l,D) ,55)

4-3

SYMBOLIC NAMES, VARIABLES, AND ARRAYS
4.3.2

Dimensioning Arrays

You must declare the size (number of elements) of an array to enable
FORTRAN to reserve the number of locations needed to store the array.
Arrays are stored as a series of sequential storage locations.
Arrays, however, are visualized and referenced as if they were single
or multi-dimensional,
rectilinear matrices dimensioned on a row,
column, and plane basis.
For example, Figure 4-1 represents a 3-row,
3-column, 2-plane array.

3 ROWS

~
'l-~

____3 COLUMNS

--~v------_J

Figure 4-1:

q,v

MR-S-1755-81

A 3 x 3 x 2 Array

You specify the size of an array by an array declarator written as a
subscripted array name.
In an array declarator each subscript
quantity is a dimension of the array and must be either an integer
expression, an integer variable, or an asterisk (*).
Only the upper bound in the last dimension declarator in a list of
dimension declarators can be an asterisk.
An asterisk marks the
declarators as an assumed-size array declarator (see Section 7.1.2).
NOTE
Variable array dimensions are
only
allowed
in
subprograms.
See adjustable dimension statements,
Section 7.1.1.
For example,
declarators.

TABLE(I,J,K)

and

MATRIX(lO,7,3,4)

are

valid

array

The total number of elements that comprise an array is the product of
the dimension quantities given in its array declarator.
For example,
the array lAB dimensioned as IAB(2,3,4) has 24 elements (2 * 3 * 4
24) •
You dimension arrays only in the specification statements DIMENSION,
COMMON, and type declaration (see Chapter 7).
Subscripted array names
appearing in any of the these statements are array declarators;
subscripted array names appearing in any other statements are always
array element identifiers.

4-4

SYMBOLIC NAMES, VARIABLES, AND ARRAYS
In array declarators, the position of a given subscript quantity
determines the particular dimension of the array (for example, row,
column, or plane) that it represents.
The first
three subscript
positions specify the number of rows,
columns, and planes that
comprise the named array;
each following subscript given
then
specifies a set comprised of n-number (value of the subscript) of the
previously defined sets.
For example:

The Dimension Declarator

Specifies the Array(s)

TAB(2)

TAB(2,2)

TAB(2,2,2)

....

TAB(2,2,2,2 )

......
...... ......

...... ......

......

1,1,1

1,2,1

2,1,1

2,2,1

-"
----

1,2,1,1

2,1,1,1

2,2,1,1

1,2

2,1

2,2

1,1,2

1,2,2

2,1,2

2,2,2 ....

,,-

_......

1,1,1,1

1,1

_...... --

1,1,2,1
I

......

1,2,2,1

-------

2,1,2,1

2,2,2,1

------- .~
I

1,1,1,2

1,2,1,2

2,1,1,2

2,2,1,2

1,1,2,2

1,2,2,2

2,1,2,2

2,2,2,2

_......

--'
--

MR-S-1762-81

NOTE
FORTRAN-10/20 permits up to 127 dimensions in an array
declarator.
(The FORTRAN-77 Standard allows a maximum
of 7 dimensions.)

4.3.3

Order of Stored Array Elements

The elements of an array are stored in ascending order.
The value of
the first (leftmost) subscript varies between its minimum and maximum
values most rapidly.
The value of the last
(rightmost)
subscript
increases to its maximum value least rapidly.
For example, the
elements of the array dimensioned as 1(2,3)
are stored
in the
following order:
1(1,1)

1(2,1)

1(1,2)

1(2,2)

1(1,3)

1(2,3)

In the following list, the elements of the three-dimensional array
(B(3,3,3»
are stored row by row from left to right and from top to
bottom.

4-5

SYMBOLIC NAMES, VARIABLES, AND ARRAYS

8(1,1,1)

8(2,1,1)

8(3,1,1)--1

r------------------------ - 8(1,2,1)
8(2,2,1)
8(3,2,1)-,
,-------------------------8(1,3,1)
8(2,3,1)
8(3,3,1)-,
[:B(1 ~ ~2) - -- B(2~1 ~2) - - -8(3,1 ~)= =,

[------------- ----------- 8(1,2,2)

8(2,2,2)

8(3,2,2)-"'j

1------ - - - - - - - - - - ---------

--8(1,3,2)

8(2,3,2)

8(3,3,2)-"'j

[": B(1,1 ~3) - - - B(2~ ~3) - - - 8(3,1 ~3): J

r- ------------ ----------

=-.:~~,~~3~ ___ ~~~~~ ___~(~,~~3~:J
L - 8(1,3,3)

8(2,3,3)

8(3,3,3)
MR-S-1756-81

Thus B(3,1,1) is stored before B(1,2,1), and so forth.
Character array elements are stored in successive character positions,
and do not necessarily start on a word boundary.
Character array
elements are stored five characters per word
(seven bits
per
character), and the low order bit is never used, for example:
CHARACTER*3 A(4)
The array A will be stored in the following way:

A(1)

A(2)

A(3)

A(4)

"
0

Ix I
28

"
21

Ix I
35

Unused

,
x I x I x Ix I
A

MR·S·2528-83

where:
x

4.4

means bits are not used.

The value in bit 35 is zero.

CHARACTER SUBSTRINGS

A character substring is a contiguous segment of a character variable
or character array element. A character substring is identified by a
substring name and can be assigned values and referenced.
A character substring reference has one of the following forms:
v([el]:[e2])
or
a(s[,s] ••• )

([el]:[e2])

4-6

SYMBOLIC NAMES, VARIABLES, AND ARRAYS
where:
v

is a character variable name.

is a character array name.

is a subscript expression.

is an optional numeric expression that specifies
leftmost character position of the substring.

the

is an optional numeric expression that specifies
rightmost character position of the substring.

the

Character positions within a character variable or array element are
numbered from left to right, beginning at 1. For example, LABEL(2:7)
specifies the substring beginning with the second character position
and ending with the seventh character position of the character
variable LABEL.
If the value of the numeric
integer, FORTRAN converts
fractional part before use.

expression el or e2 is not of type
it to an integer value by truncating any

The values of the numeric expression el and e2 must meet the following
conditions:
1 • LE.

e 1 • LE.

e 2 • LE •

1 en

where:
len is the length of the character variable or array element.
If e1 is omitted, FORTRAN assumes that e1 is 1.
FORTRAN assumes that e2 equals len.

omitted,

For example, NAMES(1,3) (:7) specifies the substring starting with the
first character position and ending with the seventh character
position of the character array element NAMES(1,3).

4-7

CHAPTER 5
EXPRESSIONS

5.1

ARITHMETIC EXPRESSIONS

An arithmetic expression is formed with arithmetic operands and
arithmetic operators.
The evaluation of such an expression produces a
numeric value.
Arithmetic expressions may be either simple or compound.
arithmetic expression consists of an operand that can be:
1.

A numeric constant

A numeric variable

A numeric array element

An arithmetic function reference (see Chapter 13)

An arithmetic
parentheses

logical

expression

written

Operands may be of integer, real, double-precision, complex,
octal, double-octal, or Hollerith type.

simple

within
logical,

The following are examples of valid simple arithmetic expressions:
105
lAB
TABLE(3,4,5)
SIN (X)
(A+B)

(integer constant)
(integer variable)
(array elemen t)
(function reference)
(a parenthetical expression)

A compound arithmetic expression consists of two or more operands
combined by arithmetic operators.
Table 5-1 lists the arithmetic
operations permitted in FORTRAN and the operator recognized for each
operation.

5-1

EXPRESSIONS
Table 5-1:

Arithmetic Operations and Operators

Operation

Binary Operator
+

Addition
Subtraction
Multiplication
Division
Exponentiation
Operation

** or

Unary Operator
+

Identity
Negation

5.1.1

Example
A+B
A-B
A*B
AlB
A**B or A. . B
Example
+A
-B

Rules for Writing Arithmetic Expressions

Observe the following rules in structuring arithmetic expressions:
1.

The operands comprising an arithmetic expression can be of
different types. Tables 5-2 and 5-3 illustrate all permitted
combinations of data types and the type assigned to the
result of each.
NOTE
All combinations of numeric data types except
double-precision with complex are allowed in
FORTRAN.

If you specify two adjacent operators, and the second is a
minus or a plus, the second operator is considered a unary
operator and acts only on the term immediately following
it.
Thus,
in the example (A*X+B)*+C, the subexpression, *+C, is
interpreted as the binary operator * and the unary +.
You cannot, however, have two adjacent binary operators in an
expression.
For
example,
the expression A*/B is not
permitted.

All operators must be included; no operation is implied. For
example, the expression A(B) does not specify multiplication,
although this is implied in standard algebraic notation. The
expression A*(B)
is required to specify a multiplication of
the operands.

5-2

Table 5--2: Type of the Result Obtained from Mixed-Mode Operations

Type of Argument 2
For operators

+,-,\1

Integer

Real

Integer

Real

Complex

Logical

Octal

Double
Octal

Literal

1. Type of operation
used
2. Type associated
with result
3. Conversion on
Argument 1

1. Integer

1. Real

1. Double Precision

1. Complex

1. Integer

2. Integer

2. Real

2. Double Precision

2. Complex

2. Integer

3. None

3. From Integer
to Real

3. From Integer to
Double Precision

3. None

4. Conversion on
Argument 2

4. None

3. From Integer to
Complex. Value
used as Real part
4. None

4. None

4. High order word
is used directly;
low order word
is Ignored

4. High order word
is used directly;
further words are
ignored.

1. Type of operation
used
2. Type associated
with result
3. Conversion on
Argument 1

1. Real

1. Double Precision

1. Complex

1. Real

I. Real

1. Real

2. Real

2. Double Precision

2. Complex

2. Real

3. None

4. None

3. Used directly as
the Real part;
imaginary part
is zero
4. None

3. None

4. From Integer
to Real

3. Used directly as
the high order
word; low order
word is zero.
4. None

4. None

4. High order word
is used directly;
low order word
is ignored.

4. High order word
is used directly;
further words
are ignored.

w
4. Conversion on
Argument 2

Double
Precision

1. Type of operation
used
2. Type associated
with result
3. Conversion on
Argument 1
4. Conversion on
Argument 2

1. Double Precision

2. Double Precision

3. None

4. From Integer to
Double Precision

4. Used directly as
the high order
word; low order
word is zero

4. None

4. Used directly as
the high order
word; low order

4. None

4. First two words
are used directly
further words
are ignored

word is zero

Complex

1. Complex

2. Complex

3. None

4. From Integer to
Complex. Value
used as Real part.

4. Used directly as
the Real part;
imaginary part
is zero.

4. None

1. Type of operation
used
2. Type assocleted
with result
3. Conversion on
Argument 1

1. Integer

1. Real

1. Double Precision

2. Integer

2. Real

2. Double Precision

3. None

4. Conversion on
Argument 2

4. None

3. Used directly as
the high order
word; low order
word Is zero
4. None

3. Used directly as
the Real part;
imaginary part
is zero
4. None

1. Type of operation
used
2. Type associated
with result
3. Conversion on
Argument 1
4. Conversion on
Argument 2

1. Complex

2. Complex

3. None

4. Used directly as
the Real part;
imaginary part
is zero.

4. None

4. First two words
are used directly.
Further words
are ignored.

1. Complex

1. Integer

I. Integer

2. Complex

2. Octal

3. None

4. None

4. High order word
is used directly;
low order word

4. High order word
is used directly;
further words

1. Complex

1 Complex

1. Complex

::J

~
<C

~
Logical

is ignored

Octal

Double
Octal

1. Integer

1. Real

1. Double Precision

1. Complex

1. Integer

2. Integer

2. Real

2. Double Precision

2. Complex

2. Octal

3. None

4. None

3. Used directly as
the Real part;
Imaginary part
is zero.
4. None

3. None

4. Conversion on
Argument 2

3. Used directly as
the high order
word; low order
word is zero.
4. None

4. None

4. High order word
is used directly;
low order word
is ignored

4. High order word
is used directly;
further words
are ignored

1. Type of operation
used
2. Type associated
with result
3. Conversion on
Argument 1

1. Integer

1. Real

1. Double Precision

1. Complex

1. Integer

2. Integer

2. Real

2. Double Precision

2. Complex

2. Octal

1. Octal

3. High order word
is used direclly;
low order word
is Ignored
4. None

3. High order word
is used direclly;
low order word
is ignored
4. None

3. None

4. None

3. High order word
is used directly;
low order word
is ignored
4. None

3. High order word
is used direclly;
low order word
is ignored
4. None

3. High order word
is used directly;
low order word
is ignored
4. High order word
is used directly.
low order word
is ignored

3. High order word
is used directly;
low order words
are ignored
4. High order word
is used directly;
low order words
are ignored

1. Type of operation
used
2. Type associated
with result
3. Conversion on
Argument 1

4. Conversion on
Argumenl2

Literal

are ignored

1. Type of operation
used
2. Type associated
with result
3. Conversion on
Argument 1
4. Conversion on
Argument 2

,. Integer

1. Real

1. Double Precision

1. Complex

1. Integer

2. Integer

2. Real

2. Double Precision

2. Complex

2. Octal

3. High order word
Is used direclly;
further words
are ignored
4. None

3. High order word
Is used directly;
further words
are ignored
4. None

3. First two words
are used directly;
further words
are ignored
4. None

3. High order word
is used directly;
further words
are ignored
4. None

3. High order word
is used directly;
further words
are ignored
4. High order word
is used directly.
low order word
is ignored

3. High order word
is used directly;
further words
are ignored
4. High order word
is used directly;
further words
are Ignored
MR-S·17S1-81

EXPRESSIONS
Table 5-3:

Permitted Base/Exponent Type Combinations

Base Operand

Exponent Operand
DoublePrecision

Complex

Real

DoublePrecision

Complex

Real

DoublePrecision

Complex

DoublePrecision

(Illegal)

Complex

(I llega 1)

Complex

Integer

Real

Integer

Real

5.1.2

Arithmetic Constant Expressions

An arithmetic constant expression is an arithmetic expression in which
each operand is one of the following:
1.

A numeric constant

A symbolic name of a numeric constant

An arithmetic constant expression enclosed in parentheses

A call to the function
ICHAR
(see Chapter
argument is a character constant expression

13)

where

the

The exponentiation operator is not permitted unless the exponent is of
type integer.
Note that variables, array elements, and function
references are not allowed.
Example:
5+6*(ICHAR('Z')-ICHAR('A')+1)*4.1**3

5.1.2.1 Integer Constant Expression - An integer constant expression
is an arithmetic constant expression in which each constant or
symbolic name of a constant is of type integer.
Example:
3+4**6+2

5-5

EXPRESSIONS
5.2

CHARACTER EXPRESSIONS

Character expressions consist of character operands and character
operators.
The evaluation of a character expression yields a single
value of character data type.
A character operand can be anyone of the following:
1.

A character constant

A symbolic name of a character constant

A character variable

A character array element

A character substring

A character expression, optionally enclosed in parentheses

A character function reference

The only character operator is the concatenation operator (//).
A character expression has the form:
character operand [//character operand] •••
The value of a character expression is a character string formed by
successive left-to-right concatenations of the value of the elements
of the character expression.
The length of a character expression is
the sum of the lengths of the character elements.
For example, the
value of the character expression 'AB'//'CDE' is 'ABCDE', which has a
length of 5.
Note that the expression:
A=A//B
has no effect on A, since the concatenation result is truncated to the
length of A.
Parentheses do not affect the value of a character expression.
example, the following character expressions are equivalent:

For

('ABC'//'DE')//'F'
'ABC'//('DE'//'F')
'ABC'//'DE'//'F'
Each of these character expressions has the value 'ABCDEF'.
If a character element in a character expression contains spaces,
spaces are
included
in the value of the character expression.
example, 'ABC '//'D E'//'F' has a value of 'ABC D EF'.

5-6

the
For

EXPRESSIONS
5.2.1

Character Constant Expression

A character constant expression is a character
each operand is one of the following:

expression

A character constant

The symbolic name of a character constant

A character constant expression enclosed in parentheses

A call to the function CHAR
(see Chapter
argument is an integer constant expression

13)

which

where

the

Variables, array elements, substrings, and function references are not
allowed.
Example:
'HELLO'//CHAR(13)//CHAR(lO)//'GOODBYE'

5.3

LOGICAL EXPRESSIONS

Logical expressions can be either simple or compound. Simple logical
expressions consist of a logical operand, which can be one of the
following:
1.

A constant

A variable

An array element

A function reference (see Chapter 13)

An expression written within parentheses

Compound logical expressions consist of two or more logical or numeric
operands combined by logical operators.
The evaluation of a logical
expresslon produces a truth value (type logical, true or false)
as
determined by the resulting bit pattern.
Table 5-4 gives the logical operators permitted
description of the operation each provides.

5-7

FORTRAN

and

EXPRESSIONS
Table 5-4:
Operator

Logical Operators
Description

.AND.

AND operator. Both of the logical operands combined
this operator must be true to produce a true result •

• OR.

Inclusive OR operator.
If either or both of the logical
operands combined by .OR. are true, the result will be
true.

.NEQV.

Exclusive OR operator (also .XOR.).
If either but not
both of the logical operands combined by .NEQV. is
true, the result will be true.

.EQV.

Equivalence operator.
If the logical operands being
combined by .EQV. are both the same (both are true or
both are false), the result will be true.

.NOT.

Complement
operator.
This
operator
specifies
complementation
(inversion)
of the item (operand or
expression) that it modifies.
The original
item,
if
true by itself, becomes false, and vice versa.

Logical expressions are of the general form P .op. Q, where P and Q
are logical operands and .op.
is any logical operator except ".NOT.".
The .NOT. operator complements the value of an operand;
it must
appear immediately before the operand that it modifies, for example,
.NOT.P.
Table 5-5 is a truth table illustrating all possible
logical
combinations of two logical operands (P and Q) and the result of each
combination.

5-8

EXPRESSIONS
Table 5-5:

Logical Operations Truth Table

When P is

And Q is:

True

-----

.NOT. P

False

-----

.NOT. P

True

P .AND. Q

True

False

P .AND. Q

False

True

P .AND. Q

False

P .AND. Q

False

True

P .OR. Q

True

False

P .OR. Q

True

False

True

P .OR. Q

True

False

P .OR. Q

False

True

P .NEQV. Q

False

True

False

P • NEQV. Q

True

False

True

P .NEQV. Q

True

False

P • NEQV. Q

False

True

P .EQV. Q

True

False

P .EQV. Q

False

True

P. EQV. Q

False

P • EQV. Q

True

Then the Expression:

For example, consider the following variables:
Variables
PHETT, RUN
I,J,K
DP,D
L,A,B
CPX,C

Type
Real
Integer
Double-Precision
Logical
Complex

5-9

Is:

EXPRESSIONS
Examples of valid logical
variables are:

expressions

consisting

the

preceding

L.AND.B
(PHETT*I).NEQV. (DP+K)
L.AND.A.OR •• NOT. (I-K)
Logical operations are performed on
the
full
36-bit
binary
representation of the operands involved.
However, when an operand of
a logical expression is double-precision or complex, only the first
word of a double-precision operand (the high-order word) or the real
part of the complex operand is used
in the specified logical
operation.
The result of a logical operation is found by performing the specified
operation simultaneously for each of the corresponding bits in each
operand. For example, consider the expression A=C.OR.D, where C="456
and D="201.
The operation performed by the processor and the result
is:
Word
Bits
0
Operand C 0
Operand D 0
Result A 0

0
0
0

--24
·0
·0
.0

25 26
0 0
0 0
0 0

27
1
0
1

28
0
1
1

29
0
0
0

30
1
0
1

31
0
0
0

32 33
1 I
0 0
1 1

34 35
I
0
0 I
1 1

Table 5-5 also illustrates all possible logical combinations of two
one-bit binary operands
(P and Q)
and glves the result of each
combination. Simply read 1 for true and 0 for false.
If a logical expression is used as an operand in an arithmetic
expression,
its value is not converted to accommodate it to the type
of any other operand.

5.3.1

Logical Constant Expression

A logical constant expression is a logical expression
operand is one of the following:

which

A logical constant

The symbolic name of a logical constant

A relational expression in which each operand is
expression

A logical constant expression enclosed in parentheses

each

constant

Variables, array elements, and function references are not allowed.
Example:
.NOT. (PARML1.NE.PARML2)
where PARMLI and PARML2 are specified in a
Section 7.8).

5-10

PARAMETER

statement

(see

EXPRESSIONS
5.4

RELATIONAL EXPRESSIONS

Relational expressions consist of two arithmetic expressions or
character
expressions
combined by a relational operator.
relational operator allows you to test the relationship between
arithmetic or two character expressions.

two
The
two

The result of a relational expression is always a
false value.

logically

true

You can write relational operators either as a
2-1etter mnemonic
enclosed within periods
(for example,
.GT.) or use the symbolic
equivalent, for example, >, instead of .GT.
Table 5-6 lists the mnemonic and symbolic forms of the FORTRAN-lO/20
relational operators and specifies the type of test performed by each.
Table 5-6:

Relational Operators and Operations
Operators

Relation Tested

Mnemonic

Symbolic

.GT.
.GE.
.LT.
.LE.
.EQ.
.NE.

>
>=

Greater than
Greater than or equal to
Less than
Less than or equal to
Equal to
Not equal to

<
<=

Relational expressions are of the general form A .op. B, where A and B
represent arithmetic or character operands, and .op.
is a relational
operator.
You can mix arithmetic operands of type
double-precision in relational expressions.

integer,

real,

and

A relational expression cannot be used to compare the value of an
arithmetic expression with the value of a character expression.
However, you can compare a numeric expression to a character constant.
In this case, the character constant is considered to be a Hollerith
(see Section 3.9).
You can compare complex operands using only the operators
.EQ.
(==)
and
.NE.
(#) •
Complex quantities are equal if the corresponding
parts of both words are equal.
For example, assume the following variables:
Variables

Type

PHETT, RON
I,J,K
DP,D
L,A,B
CPX,C
CHR,RA

Real
Integer
Double-Precision
Logical
Complex
Character

5-11

EXPRESSIONS
expressions

consisting

the

above

Examples of invalid relational expressions
variables are:

consisting

the

above

.EQ.

operands

and

Examples of valid
variables are:

relational

(PHETT) .GT.IO
[

-- 5

C.EQ.CPX
CHR.LT.RA

(PHETT) .GT 10

(closing period missing from operator)

C.GT.CPX

(complex operands can only
and .NE. operators)

RA.EQ.RON

(you cannot compare
character operands)

compared

arithmetic

Examples of valid expressions that use both logical
operators to combine the preceding variables are:

and

relational

(I.GT. 10) .AND. (J.LE.K)
( (I * RON) • EQ. (I / J) ) • OR • L
(I. AND. K) # ( (PHETT) • OR. (RON) )
C#CPX.OR.RON
rf a logical expression
is used as an operand in an arithmetic
expression,
its value is not converted to accommodate it to the type
of any other operand.

In character relational expressions "less than" means "precedes in the
ASCII collating sequence," and "greater than" means "follows in the
ASCII collating sequence", for example:
, AB ' / / ' ZZZ'

• LT.

'CCCCC'

This expression tests whether 'ABZZZ' is less than
'CCCCC'.
Since
that relationship does exist, the value of the expression is true.
If
the relationship stated does not exist, the value of the expression is
false.
If the two character expressions in a relational expression are not
the same length, the comparison is performed as if the shorter one is
paddad on the right with spaces until the lengths are equal, for
example:
, ABC'

• EQ.

' ABC

, AB '

• LT • ' C '

The first relational expression has a value of true even though the
lengths of the expressions are not equal, and the second has a value
of true even though 'AB' is longer than 'C'.

5-12

EXPRESSIONS
NOTE
The rule that character relationals extend the shorter
operand with spaces to match the length of the longer
operand has an interesting effect when the longer
string ends with characters in the range CHAR(O) to
CHAR(31) (ASCII control characters such as 'bell'
and
line feed).
Since space is CHAR(32) , the trailing spaces supplied
as filler by FORTRAN compare being greater than
trailing control characters. Thus, the string
'FOO'
is .GT. 'FOO . . G' (FOO followed by a bell).

5.5

EVALUATION OF EXPRESSIONS

The following considerations determine the order of computation
FORTRAN expression:
1.

The use of parentheses

An established hierarchy for the
relational, and logical operations

The location of operators within an expression

5.5.1

execution

arithmetic

Parenthetical Subexpressions

In an expression, all subexpressions enclosed within parentheses are
evaluated first.
When parenthetical subexpressions are nested (one
contained within another), the most deeply nested subexpression is
evaluated
first;
the next most deeply nested subexpression is
evaluated second;
and so on, until the value
of
the
final
parenthetical expression is computed.
When more than one operator is contained in
a
parenthetical
subexpression,
the required computations are performed according to
the hierarchy assigned to operators by FORTRAN (see Section 5.5.2).
For example, the separate computations
expression A+B/«A/B)+C)-C are:
1.

Rl=A/B

R2=Rl+C

R3=B/R2

R4=A+R3

R5=R4-C

performed

evaluating

the

results

the

where:
Rl through R5

represent the interim and
computations performed.

5-13

final

EXPRESSIONS
5.5.2

Hierarchy of Operators

The following hierarchy
(order
classes of FORTRAN operators:
first,
second,
third,

execution)

assigned

the

arithmetic operators
relational operators
logical operators

Table 5-7 specifies the precedence
operators of the above classes.

assigned

the

individual

With the exception of exponentiation and
integer division, all
operations on expressions or subexpressions involving operators of
equal precedence are computed
in any order that is algebraically
correct.
A subexpression of a given expression may be computed
in any order.
For example,
in the expression (F(X) + A*B), the function reference
may be computed either before or after A*B.
Table 5-7:

Hierarchy of FORTRAN Operators

Class

Level

EXPONENTIAL

First

** or '"

ARITHMETIC

Second
Third
Fourth

-(negation) and + (identity)
*,/
+,-

RELATIONAL

Fifth

.GT.,.GE.,.LT.,.LE.,.EQ.,.NE.
or

Symbol or Mnemonic

>,>=,<,<=,==,#

LOGICAL

Sixth
Seventh
Eighth
Ninth

.NOT.
.AND.
• OR.
.EQV.,.NEQV.

Operations specifying integer division are evaluated from left to
right.
For example,
the expression I/J*K is evaluated as if it had
been written as (I/J)*K), but this left-to-right evaluation process
can be overridden by parentheses.
I/J*K (evaluated as(I/J) *K) does
not equal I/(J*K).
When a series of exponentiation operations occurs in an expression, it
is evaluated in order from right to left.
For example, the expression
A**2**B is evaluated in the following order:
first Rl = 2**B (intermediate result)
second R2 = A**Rl (final result).
As with other expressions, parentheses alter the evaluation of the
above expression.
The expression (A**2)**B is evaluated in these two
steps:
first Rl = A**2 (intermediate result)
second R2 = Rl**B (final result)

5-14

EXPRESSIONS
5.5.3

Mixed-Mode Expressions

Mixed-mode
expressions
are
evaluated
on
a
basis
of
subexpression-by-subexpression, with the type of the results obtained
converted and combined with other results or terms according to the
conversion procedures described in Table 5-2.
For example, assume the following variables and data types:
Variables

Type
Double-Precision
Real
Integer

I,J

The mixed-mode expression D+X*(I/J)
manner:

evaluated

the

following

I/J

Rl is integer

X*Rl Rl is converted to type real and is multiplied by X
to produce R2

D+R2 R2 is converted to type
added to D to produce R3

double-precision

and

where:
Rl and R2,
and R3 represent the
interim
respectively of the computations performed.

5.5.4

and

final

results

Use of Logical Operands in Mixed-Mode Expressions

When you use logical operands in mixed-mode expressions, the value
the logical operand is not converted in any way to accommodate it
the type of the other operands in the expression.
For example,
L*R, where L is type logical and R is type real, the expression
evaluated without converting L to type real.

5.6

of
to
in
is

CONSTANT EXPRESSIONS

A constant expression is an arithmetic constant expression
(see
Section 5.1.2), a character constant expression (see Section 5.2.1),
or a logical constant expression (see Section 5.3.1).

5-15

CHAPTER 6
EXECUTABLE AND NONEXECUTABLE STATEMENTS

Each statement is classified as
executable
or
nonexecutable.
Executable statements specify actions and form an execution sequence
in a program.
Nonexecutable statements do the following:
1.

Specify characteristics, arrangement, and initial
data

Contain editing information

Specify statement functions

Classify program units

Specify entry points within subprograms

values

Nonexecutable statements are not part of the execution sequence.
Nonexecutable statements may be labeled, but such statement labels
must not be used to control the execution sequence.

6.1

EXECUTABLE STATEMENTS

The following statements are classified as executable:
1.

Arithmetic, logical, statement label (ASSIGN), and
assignment statements

Unconditional GO TO, assigned
statements

Arithmetic IF, logical IF statements, and two-branch logical
IF statements, IF THEN, ELSE, and ELSE IF THEN statements

CONTINUE statement

STOP and PAUSE statements

DO and DO WHILE statements

READ, REREAD, WRITE,

DELETE and UNLOCK statements

OPEN and CLOSE statements

10.

REWIND, BACKSPACE, ENDFILE, BACKFILE,
FIND and UNLOAD statements

FORTRAN-10/20 Vll

TO,

and

character

computed

REWRITE, and PRINT statements

6-1

SKIPRECORD,

SKIPFILE,

February 1987

EXECUTABLE AND NONEXECUTABLE STATEMENTS

6.2

11.

CALL and RETURN statements

12.

END, END IF, and END DO statements

13.

DECODE and ENCODE statements

14.

ACCEPT, PUNCH, and TYPE statements

15.

INQUIRE statement

NONEXECUTABLE STATEMENTS

The following statements are classified as nonexecutable:

6.3

SUBROUTINE,

ENTRY,

and

PROGRAM,
FUNCTION,
statements

DIMENSION,
COMMON,
EQUIVALENCE,
IMPLICIT,
INTRINSIC, EXTERNAL, and SAVE statements

INTEGER,
REAL,
DOUBLE PRECISION,
COMPLEX,
CHARACTER type-specification statements

DATA statement

FORMAT statement

Statement function statement

INCLUDE statement

NAMELIST statement

BLOCK

DATA

PARAMETER,
LOGICAL,

and

ORDERING OF FORTRAN STATEMENTS

The order in which you place FORTRAN statements in a program unit is
important.
Certain types of statements must be processed before
others to guarantee that compilation takes place as you expect.
Figure 6-1 shows the required order of statements and comment lines
within a program unit.
Horizontal lines indicate (from the top of the
diagram to the bottom) the order in which statements and comment lines
must appear in a program. Tor example, a PROGRAM statement must occur
before FORMAT statements.
FORMAT statements,
in turn, must occur
before an END statement.
vertical lines
in the diagram
indicate how comment lines
and
statements may be interspersed in the program.
For example, PARAMETER
statements must be placed after all PROGRAM, FUNCTION, or SUBROUTINE
statements,
and
before
all statement function and executable
statements.
PARAMETER statements can be placed before, after, or
between all
IMPLICIT and other specification statements.
Comment
lines may be interspersed anywhere in a program.
Generally if FORTRAN encounters statements that are out of place,
it
prints warning messages and continues compilation.
In some cases,
however, out-of-place statements cause the compiler to terminate
compilation or generate unexpected results.

6-2

February 1987

EXECUTABLE AND NONEXECUTABLE STATEMENTS

PROGRAM, FUNCTION, SUBROUTINE, or
BLOCK DATA l Statements

IMPLICIT
Statements
PARAMETER
Statements
Comment
Lines
and
INCLUDE3
Statements

Other
Specification
Statements
FORMAT
and
Ent ry 2
Statements

NAMELIST
and
DATA
Statements

Statement
Function
Definitions

Executable
Statements

END Statement

BLOCK DATA subroutines cannot contain any executable statements, statement
functions, FORMAT statements, EXTERNAL statements, INTRINSIC statements,
or NAMELIST statements (See Section 14.1).

The ENTRY statement is allowed only in functions or subroutines. All executable
statements which reference any dummy parameters must physically follow the
ENTRY statement unless the references appear in the FUNCTION statement, the
SUBROUTINE statement, or in a preceding ENTRY statement.

The placement of an INCLUDE statement is dictated by the types of statements to
be included.
MR-S-3822-85

Figure 6-1:

Ordering of FORTRAN Statements

NOTE
In FORTRAN-IO/20, a DATA statement can precede a
PARAMETER
statement
or
another
specification
statement.

6-3

EXECUTABLE AND NONEXECUTABLE STATEMENTS
6.4

COMPILATION CONTROL STATEMENTS

You use compilation control statements to
identify FORTRAN programs
and to specify their termination.
Statements of this type do not
affect either the operations performed by the object program,
or the
manner in which the object program is executed.
The three compilation
control statements are:
1.

PROGRAM statement

INCLUDE statement

END statement

The PROGRAM statement and the INCLUDE statement are described
in the
following sections.
The END statement is described in Section 9.8.

6.4.1

PROGRAM Statement

This statement allows you to give the main program a name other than
the compiler-assumed name "MAIN." The general form of a PROGRAM
statement is:
PROGRAM name
where:
name

is a symbolic name that begins with an alphabetic
character and contains a maximum of six characters.
(See Section 4.1 for a description of symbolic names.)

The PROGRAM statement must be the first statement in a program unit.
(see Section 6.3 for a discussion of the ordering of FORTRAN
statements.)

6.4.2

INCLUDE Statement

This statement allows you to
include code segments or external
declarations in a program unit without having them in the same file as
the primary program unit. When an INCLUDE statement is encountered
during compilation,
the compiler replaces the INCLUDE statement with
the contents of the specified file.
The general form of the
INCLUDE
statement is:
INCLUDE 'filespec [/switch]'
where:
filespec

is a TOPS-I0 or TOPS-20 file specification
(refer
to
the
TOPS-IO or TOPS-20 Operating System
Commands manual).
The only restrlction is that
unde~'fOPS-IO
you cannot specify subdirectory
information.

switch

is one of the following optional switches:
/CREF

indicates the
included statements
are to be used to augment the
cross-reference listing (default).

6-4

EXECUTABLE AND NONEXECUTABLE STATEMENTS
/LIST

indicates that the statement in the
specified file
is to be listed in
the compilation source listing.
A
number
indicating
the depth of
nesting of include files
precedes
each statement listed (default).

/NOLIST

indicates
that
the
included
statements are not to be printed in
the compilation listing.

/NOCREF

indicates
statements
augment
listing.

that
the
included
are not
to be used to
the
cross-reference

The following rules govern the use of the INCLUDE statement:
1.

The INCLUDEd file can contain any legal FORTRAN statement
except a statement that terminates the current program unit,
such as the END, PROGRAM, FUNCTION, SUBROUTINE, or BLOCK DATA
statements.
The INCLUDEd file can contain other INCLUDE statements.
This
is called nesting INCLUDE statements.
The number of nested
levels is restricted to 12.

The proper placement of the
INCLUDE statement within a
program unit depends upon
the types of statements to be
INCLUDEd.
(See Section 6.3 for information on the ordering
of FORTRAN statements.)

The file to be INCLUDEd must be on disk.

Note that an asterisk (*) is appended
to the line numbers of the
INCLUDEd statements on the compilation listing.
The level of nesting
is indicated following the asterisk.

6-5

CHAPTER 7
SPECIFICATION AND DATA STATEMENTS

Specification statements are used
to specify
storage allocation,
and data arrangement.
specification statements:
1.

DIMENSION

Statements that explicitly specify type,
including INTEGER,
REAL, DOUBLE PRECISION, COMPLEX, LOGICAL, or CHARACTER

IMPLICIT

COMMON

EQUIVALENCE

EXTERNAL

INTRINSIC

PARAMETER

DATA

10.

SAVE

Specification statements are nonexecutable and
ordering guidelines described in Section 6.3.

7.1

type characteristics,
There are ten types of

must

conform

the

DIMENSION STATEMENT

The DIMENSION statement provides FORTRAN with information needed to
identify and allocate the space required for arrays.
You may specify
any number of subscripted array names as array declarators
in a
DIMENSION statement.
The general form of a DIMENSION statement is:
DIMENSION a(d)

[,a(d) ••• ]

where:
each a(d)

is an array declarator.
An array declarator
provides the name and dimension(s) of an array.
An array declarator is written in the following
form:
a(d [,d ••• ])

7-1

SPECIFICATION AND DATA STATEMENTS
where:
a

is the symbolic name of the array.

is the array dimension declarator.
declarator is as follows:

The form of a

dimension

dl: ] d2
where:
dl

is an optional integer expression or integer
variable specifying the lower dimension bound.
The lower dimension bound is the first element in
that dimension of the array.
If dl is not
specified the default is 1.

is an integer expression or integer variable that
specifies the upper dimension bound. The upper
dimension bound is the greatest element in that
dimension of the array. You must have at least
one d2 specification in each array declaration.
If both dl and d2 are specified, dl cannot have a
value greater than d2.
The values dl and d2 can,
however, be the same.
An asterisk (*) can also occur as an upper bound,
but only as the last dimension. An asterisk marks
the declarator as an assumed-size array declarator
(see Section 7.1.2).
NOTE

A dimension bound expression must
function or array element reference.

not

contain

If the array is a dummy argument to a subprogram, then dl and d2 can
be integer dummy arguments and d2 can be an asterisk; otherwise, they
must be constants.
If the symbolic name of a constant or variable that appears in a
dimension bound expression is not of implicit default integer type
(see Section 4.2),
it must be specified integer by an IMPLICIT
statement or a type-statement.
Examples:
DIMENSION EDGE (-1:1,4:8), NET (5,10,4), TABLE (567)
DIMENSION TABLE (IAB:J,K,M,10:20)
where:
lAB, J, K, and M are of type integer.

7-2

SPECIFICATION AND DATA STATEMENTS
7.1.1

Adjustable Dimensions

When used within a subprogram, a declarator for an array that is a
dummy argument can use integer dummy arguments as dimension bounds.
The following rules govern the use of adjustable dimensions in a
subprogram:
1.

The array name must be a dummy argument. Each variable that
is used as a dimension bound must be either a dummy argument
or in COMMON (see Section 7.4).

For multiple entry subprograms, if any variables that specify
dimension bounds are dummy arguments which do not occur in
the formal argument list of the entry point used,
the value
of the variables as passed for a previous call are used.
However, this is only permitted if the subprogram has not
changed those dummy arguments. Futhermore, when overlays are
used, a SAVE statement that preserves the local variables of
the subprogram is needed.

If the value of an array dimension variable is altered within
the program, the dimensionality of the array is not affected.

The size of an array within a subprogram cannot
size of the original array, as defined in
program.

exceed the
the calling

Example 1:

SUBROUTINE SBR (ARRAY,Ml,M2,M3,M4)
DIMENSION ARRAY{Ml:M2,M3:M4)
DO 27 L=M3,M4
DO 27 K=Ml,M2
ARRAY (K,L)=VALUE
CONTINUE
END

In the example above, the dimensions of ARRAY
each time the subroutine SBR is entered.

will

re-specified

Example 2:
SUBROUTINE SBI (ARRl,M,N)
DIMENSION ARRl{M,N)
ARRl{M,N)=VALUE
ENTRY SB2{ARRl,M)
ENTRY SB3{ARRl,N)
ENTRY SB4{ARRl)

END
In the example above, the first call made to the subroutine must be
made to S81 so that all of the dimension bounds have defined values.
If a call is made to SBI with the values M=ll and N=13,
a succeeding
call to SB2 will use the value N=13, but will give M a new value.
If
a succeeding call is made to SB4,
the last values passed through
entries SBl, SB2, or SB3 are used for M and N.

7-3

SPECIFICATION AND DATA STATEMENTS
7.1.2

Assumed-size Arrays

An assumed-size array is a dummy array for which the upper bound
the last dimension is specified as an asterisk(*), for example:

SUBROUTINE SUB(A,N)
DIMENSION A(1:N,1:*)
Since storage for array A is allocated in the calling routine,
the
upper bound of the rightmost dimension of A does not affect the
subscript calculations or storage allocation for A.
Therefore, subroutine SUB can be written to handle arguments with any
rightmost dimension
(the last subscript is never range checked for
being too large, even when the /DEBUG:BOUNDS compiler switch is
specified). Such a subroutine can declare assumed-size arrays.
The size of an assumed-size array, and the number of elements that can
be referenced, are determined as follows:
1.

If the actual argument corresponding to the dummy array is a
noncharacter array name, the size of the dummy array is the
size of the actual-argument array.

If the actual argument corresponding to the dummi argument is
a noncharacter array element name, with a subscript value of
s in an array of size a, the size of the dummy array is:
a+l-s.

If the actual
argument
is
a
character-array
name,
character-array element name,
or character-array element
substring name, and begins at character storage unit b of an
array with n character storage units, the size of the dummy
array is INT«n+l-b)/y). Where y is the length of an element
of the dummy array.

Because the actual size of an assumed-size
assumed-size array name cannot be used as:

7.2

array

not

known,

An array name in the list of an I/O statement

A unit identifier for an internal file in an I/O statement

A format specifier in an I/O statement

A NAMELIST statement element

TYPE SPECIFICATION STATEMENTS

Type specification statements explicitly declare the data
type of
variables, arrays, or function names.
You can give an array name in a
type specification statement, either alone (unsubscripted) to declare
the type of all its elements, or with dimension bounds, to specify
both its type and dimensions.
There are two forms of type specification statements:
numeric type
specification
(see Section 7.2.1)
and character type specification
(see Sec t i on 7. 2. 2) •

7-4

SPECIFICATION AND DATA STATEMENTS
7.2.1

Numeric Type Specification Statements

The general form of numeric type specification statements is:
type v [ ,v ••• ]
where:
type

can be anyone of the following declarators:
1.

INTEGER

REAL

DOUBLE PRECISION

COMPLEX

LOGICAL

is a variable, array, or function name to be declared
the specified type.
The names listed must be separated
by commas and can appear in only one type statement
within a program unit.

Examples:
INTEGER A, B, TABLE, FUNC
REAL R, M, ARRAY(5:l0,lO:20,5)
If a name that is the same as an intrinsic FORTRAN function name
appears in a conflicting type statement, it is assumed that the name
refers to a user-defined routine, variable, or array of the given
type.
If you place a generic FORTRAN function name in an explicit
type statement, it loses its generic properties.
NOTE
The data type size modifier, *n,
is accepted by
FORTRAN-lO/20
to
be
compatible
with the type
statements used by other manufacturers.
You may
append this size modifier to the declarators, causing
some to elicit messages warning users of the form of
the variable specified by FORTRAN-IO/20:
Declarator
INTEGER*2
INTEGER*4
LOGICAL*l
LOGICAL*2
LOGICAL*4
REAL*4
REAL*8
COMPLEX*8
COMPLEX*16
REAL*16
COMPLEX*32

Form of Variable Specified
Full word integer with warning message
Full word integer
Full word logical with warning message
Full word logical with warning message
Full word logical
Full word real
Double-precision real
Complex
Complex with warning message
Double-precision real with warning message
Complex with warning message

7-5

SPECIFICATION AND DATA STATEMENTS
In addition, you can append the data
type
size
modifier
to
individual
variables, arrays, or function names.
Its effect is to override, for
the
particular element,
the size modifier
(explicit or implicit)
of the primary
type.
For example,
REAL*4 A, B*8, C*8(lO), 0
A and 0 are single-precision (one word)
real, and Band C are double-precision
(two words for each element) real.

7.2.2

Character Type Specification Statements

The form of the character type specification statement is:
CHARACTER [*len [,]] v [*len]

[,v [*len] ] •••

where:
v

len

is one of the following:
•

The name of a symbolic constant, variable,
function subprogram

•

An array declarator

array,

is the length of the character data item and is one of the
following:
•

An unsigned, nonzero integer constant

•

An integer constant expression enclosed in parentheses
and with a positive value

•

An asterisk enclosed in parentheses

If you specify CHARACTER*len, len is the default length specification
for that list.
If an item in that list does not have a length
specification, the item's length is len. But if an item does have a
length specification,
it overrides the default length specified in
CHARACTER*len.
A length specification of asterisk
(for example, CHARACTER*(*»
specifies that a dummy argument or function name assumes the length
specification of the corresponding actual argument or
function
reference
(see Chapter 13). A length specification of asterisk for
the symbolic name of a constant specifies that the symbolic constant
assumes the actual length of the constant that it represents.
If you do not specify a length, a length of one is assumed.
Note that
a length specification of zero is invalid.
You can use a character
type declaration statement to define arrays by including array
declarators
(see Section 4.3.2) in the list.
If you specify both an
array declarator and a length, the array bounds precede the length,
the form is:
a [(d)] [*len]

7-6

SPECIFICATION AND DATA STATEMENTS
where:
a

is an array name, and a(d)

is an array declarator.

Examples of character type specification statements follow:
CHARACTER*32 SOCSEC(100)*9, NAMES (100)
The above statement specifies an array SOCSEC comprising one hundred
9-character elements, and an array NAMES comprising one hundred
32-character elements.
PARAMETER (LENGTH=4)
CHARACTER*(4+LENGTH) LAST, FIRST
The above statements specify two 8-character variables, LAST
FIRST.
(The PARAMETER statement is described in Section 7.8.)

and

SUBROUTINE Sl(BUBBLE)
CHARACTER LETTER(26), BUBBLE*(*)
The above statements specify an array LETTER comprising twenty-six
I-character elements and a dummy argument, BUBBLE, which has a length
defined by the calling program.
CHARACTER*16 QUEST*(5*INT(A»
The above statement is invalid.
an integer constant expression.

7.3

The length specifier for QUEST is not

IMPLICIT STATEMENTS

IMPLICIT statements declare the data type of variables and functions
according to the first letter of each symbolic name.
The IMPLICIT
statement has two forms:
IMPLICIT type (a[,a] ••• )[,type (a[,a] ••• )] •••
IMPLICIT NONE
As shown in the statement above, an IMPLICIT statement consists of one
or more type declarators separated by commas.
Each type declarator
has the form:
type (a [ , a] ••• )
where:
type

can be anyone of the following declarators:
1.

INTEGER

REAL

DOUBLE PRECISION

COMPLEX

LOGICAL

CHARACTER [*len]

7-7

SPECIFICATION AND DATA STATEMENTS

is an alphabetic specification in either of the general
forms:
c or cl-c2, where c, cl, or c2 is an alphabetic
character.
The latter form specifies a range of
letters,
from cl through c2, which must occur in
alphabetical order.

When you specify type as CHARACTER*len, len specifies the length for
character data type.
Len is an unsigned, nonzero integer constant or
an integer constant expression enclosed in parentheses and with a
positive value.
If you do not specify a length, a length of one is
assumed.
Each letter in a type declarator list specifies that each symbolic
name
(not declared
in an explicit type specification statement)
starting with that letter is assigned the data type named
in the
declarator.
For example, the statement:
IMPLICIT REAL (R,M,N,O)

declares that all names that begin with the letters R, M, N, or
are
type REAL names, unless declared otherwise in an explicit type
statement.
NOTE
Type declarations
given
in
an
explicit
type
specification override those also given in an IMPLICIT
statement.
IMPLICIT declarations do
not
affect
intrinsic functions.
The length is also overridden
when a particular name appears in a CHARACTER or
CHARACTER FUNCTION statement (see Chapter 13).
You may specify a range of letters within the alphabet by writing the
first and last letters of the desired range separated by a dash, for
example, A-E for A,B,C,D,E.
Thus, the statement:
IMPLICIT INTEGER (I,L-P)
declares that all symbolic names that
I,L,M,N,O, and P are of type INTEGER.

begin

with

the

letters

You may use more than one IMPLICIT statement, but they must appear
before any other declaration statement in the program unit.
(See
Section 6.3 for a discussion on ordering FORTRAN statements.)
The same letter must not appear as a single letter, or be included in
a
range of letters, more than once in all of the IMPLICIT statements
in a program unit.
You can use an IMPLICIT NONE statement to provide warning messages for
variables not explicitly declared,
including variables implicitly
declared by other IMPLICIT statements.
If you specify IMPLICIT NONE,
no other IMPLICIT statement should be included in the program unit.

7.4

COMMON STATEMENT

The COMMON statement enables you to establish storage that may be
shared by two or more programs and/or subprograms, and to name the
variables and arrays that are to occupy the common storage.
The use
of common storage conserves storage and provides a means to reference
the same data in different subprograms without passing arguments.
7-8

SPECIFICATION AND DATA STATEMENTS
COMMON statements have the following form:
COMMON [/cb/] nlist[[,]/[cb]/nlist] •••
where:
(See Section 4.1 for

is an optional common block name.
the rules for symbolic names.)

nlist

is a list of variable names, array names, and array
declarators that are
to occupy the common area.
The
items specified for a common area (block)
are ordered
within storage as .they are listed
in the COMMON
statement.

type or all
The items in nlist must be either all numeric data
character data type.
A common block cannot contain both numeric and
character data.
A symbolic name can be used to identify each block.
However, you can
omit the symbolic name for one block
in a
program unit.
This
unlabeled block is called the blank common block.
The elements of a named common block can be assigned initial values by
DATA statements appearing in the BLOCK DATA subprograms (see Chapter
14).
In standard-conforming programs,
the elements of the blank
common
block
may
not
be assigned
initial values.
However,
FORTRAN-I0/20 allows any common block elements to be defined in a DATA
statement in any program unit.
A given common block name may appear more than once in the same COMMON
statement, and
in more than one COMMON statement within the same
program or subprogram.
When extended addressing is in effect, COMMON blocks reside
in
the
large
data
area
by default.
However,
the /EXTEND:COMMON or
/EXTEND:NOCOMMON switches can be used to explicitly allocate COMMON
blocks in the large data area or small data area (see Section 16.5).
During compilation of a source program, FORTRAN strings together all
items listed for each common block in the order in which they appear
in the source program.
For example:
COMMON X,Y,Z/ST1/A,B
COMMON/ST1/TST(3,4)/ST2/TAB(2,2)
COMMON/ST2/C,D,E//P,Q
COMMON W
has the same effect as the single statement:
COMMON X,Y,Z,P,Q,W/ST1/A,B,TST(3,4)/ST2/TAB(2,2),C,D,E
All elements specified for a common block are placed
into one area.
Common block names must be unique with respect to all subroutine,
function, and entry point names.

7-9

SPECIFICATION AND DATA STATEMENTS
NOTE
If you use overlays, you can use the SAVE statement to
retain the value of variables in a named common across
overlays (see Section 7.10).
(Blank common is always
saved.)
For example:
Main Program

Subprogram

COMMON DELTA, LENGTH
COMMON /COMl/KILOS,PRICE

SUBROUTINE CALC
COMMON/COMl/N,A
COMMON Z,KOUNT

CALL CALC

The COMMON statements in the main program put DELTA and LENGTH into
the blank common block, and put KILOS and PRICE into a common block
named COMl.
The COMMON statements in the subroutine make Z correspond to DELTA in
the main program, KOUNT correspond to LENGTH, N correspond to KILOS,
and A correspond to PRICE.
To prevent conflict with other COMMON blocks, LINK requires
largest definition for each common block be loaded first.

7.4.1

that

the

Dimensioning Arrays in COMMON Statements

Array names with dimension bounds can be given in COMMON statements.
However, variables cannot be used as dimension bounds in a declarator
appearing in a COMMON statement; adjustable dimensioning
is not
permitted in COMMON.
Each array name given in a COMMON statement must be dimensioned either
by the COMMON statement or by another dimensioning statement within
the program or subprogram that contains the COMMON statement, but not
both.
For examples,
COMMON /A/B(lOO), C(lO,lO)
COMMON X(5,15),Y(5)

7.4.2

Character Data in COMMON

Each character variable in a COMMON block is allocated to start at the
first available character position.
For example,
CHARACTER B*3,C*3,D(3)*2
COMMON B,C,D

7-10

SPECIFICATION AND DATA STATEMENTS
The COMMON block will be allocated in the following way:

Ito

"
14

0(1)

0(2)

\~~

Ix I
28

0(3)

Unused

Ix I I
0

x I x I x
7

Ix I
35

MR-S-2527-83

where x means the bits are not used.

7.5

EQUIVALENCE STATEMENT

The EQUIVALENCE statement associates two or more
same storage location.

variables

with

the

The format of the EQUIVALENCE statement is:
EQUIVALENCE (nlist)

[,(nlist ••• )]

where:
nlist

is a list of variable names, array elements, array
names, and character substring references separated by
commas and enclosed in parentheses. You must specify
two or more of these items in each list.

In an EQUIVALENCE statement, each expression in a subscript
substring reference must be an integer constant expression.
The EQUIVALENCE statement allocates all of the items
in
parenthesized list beginning at the same storage location.
example, the statements:

a
one
For

EQUIVALENCE (A,B,C)
EQUIVALENCE (LOC,SHARE(3»
specify that the variables A, B, and C are to share the same storage
location, and that the variable LOC and the array element SHARE(3) are
to share the same location.
The relationship of equivalence is transitive.
following statements have the same effect:

For

example,

the

EQUIVALENCE (A,B), (B,C)
EQUIVALENCE (A,B,C)
The following EQUIVALENCE statement makes the first character of the
character variables KEY and STAR share the same storage location.
The
character variable STAR is equivalent to the substring KEY (1:10):
CHARACTER KEY*16, STAR*lO
EQUIVALENCE (KEY,STAR)
You can equivalence variables of different numeric data types.
Character variables must not be equivalenced to numeric variables.
For example, you can equivalence a real variable equivalent to a
complex variable.
In this case, since ea~h complex variable occupies

7-11

SPECIFICATION AND DATA STATEMENTS
two words of storage, and each single-precision variable occupies one
word of storage, if you equivalence a real and a complex variable, the
real variable shares storage with the real part of the complex
variable.
Figure 7-1 depicts the shared storage when a complex
variable is equivalenced with a real variable.
Source Program Statements~
COMPLEX A
REAL B
EQUIVALENCE (A,B)

Memory Location
A or B

Stores: Real Part of Complex A
or Entire Real B

Second Part of
Memory Location A

Stores: Imaginary Part of
Complex A

1----36-8il WO,d---!
MR-S-1764-81

Figure 7-1:

Shared Storage using EQUIVALENCE Statement

The EQUIVALENCE statement does not imply
(or perform)
any type
conversions.
If you equivalence a real variable and an integer
variable, the data within the equivalenced location will be treated as
a real or integer number, depending on whether it is referenced by the
real or integer variable.
If you equivalence a real variable with a double-precision variable,
the data in the high-order word of the double-precision variable will
be
used
by
the
real
variable.
For
positive
D-floating
double-precision numbers (see Section 3.4), the high-order word is in
the same format as a single-precision number.
For G-floating double-precision numbers (KL model B only - see Section
3.4),
the high-order word is not in the format of a single-precision
number.
Thus, equivalencing a real variable and a
G-floating
double-precision variable will produce incorrect results.
Equivalencing a negative D-floating number and a real variable may not
produce correct results either, for example the number:

577000000000
000000000001
777777777777, almost 1.0)

(the

negative

200777777777

does not have a valid single-precision number in its high-order word.
If you equiva~ence an array and a variable, the array does not assume
any of the properties of the variable, and the variable does not
assume any of the properties of the array.
When you use an array element in EQUIVALENCE statements, it must have
either as many subscripts as dimensions of the array, or only one
subscript.
In either case, the subscripts must be integer constants.
Note
that
the
single subscript case treats the array as a
one-dimensional array of the given type.

7-12

SPECIFICATION AND DATA STATEMENTS
The following example shows
the
effect
I-dimensional and a 2-dimensional array:

equivalencing

DIMENSION A(3,2),B(6)
EQUIVALENCE (A(l,l) ,B(l»
or
EQUIVALENCE (A(l) ,B(l»
In this example, each array element of array A shares the same storage
area with an element of array B:
A(l,l)
A(2,1)
A(3,1)
A(1,2)
A(2,2)
A(3,2}

B(l)
B(2)
B(3)
B(4)
B(5)
B(6}

Specifying an array name without subscripts in an
EQUIVALENCE
statement is the same as specifying the first element of the array.
When you make one character substring equivalent to another character
substring, the EQUIVALENCE statement also sets equivalences between
the other corresponding characters in the character strings, for
example,
CHARACTER NAME*16, 10*9
EQUIVALENCE (NAME(lO:13), ID(2:5})
As a result of these statements, the character variables NAME
share space as illustrated in Figure 7-2.

NAME
Character
Position

1
2
3

4
5
6
10
Character
Position

7
8
9

1
2

10
11
12
13
14
15
16

3
4
5
6
7
8
9
MR·S·2523·83

Figure 7-2:

Equivalence of Substrings

7-13

and

SPECIFICATION AND DATA STATEMENTS
The following statement also aligns the variables as shown
7-2:

Figure

EQUIVALENCE (NAME(9:9) ,ID(1:1»
If the character substring references are array elements,
the
EQUIVALENCE
statement
sets
equivalences
between
the
other
corresponding characters in the complete arrays.
Character elements of arrays can overlap at
for example:
CHARACTER FIELDS(6)*4, STAR(5)*5
EQUIVALENCE (FIELDS (1) (2:4), STAR (2)

any

character

position,

(3: 5»

As a result of these statements, the character arrays FIELDS and
share storage space as shown in Figure 7-3.
STAR

Character
Position

1
2

Character
Position

1
2
3
4
1
2
3
4
1
2
3
4
1

5
1
2
3
4
5
1
2
3

5
1
2
3
4

2
3
4
1
2
3
4
1
2

5
1
2
3
4
5

3
4
MR-S-2524-83

Figure 7-3:

3
4

FIELDS

Subscript

Equivalence of Character Arrays

7-14

STAR

SPECIFICATION AND DATA STATEMENTS
General EQUIVALENCE Restrictions:
1.

You cannot cause two different elements of an array to become
equivalenced to each other.
Thus, the following statement
sequence is prohibited because it specifies the same storage
location (B) for A(l) and A(2):
DIMENSION A(2)
EQUIVALENCE (A(l) ,B), (~(2) ,B)

An EQUIVALENCE statement must
not
specify
that
two
consecutive locations are nonconsecutive.
For example, the
following statement sequence is prohibited because B(l) takes
two storage locations,
the second of which would make A(2)
nonconsecutive to A(l):
INTEGER A(2)
DOUBLE PRECISION B(2)
EQUIVALENCE (A(l) ,B(l»,

(A(2) ,B(2»

An EQUIVALENCE statement in a SUBROUTINE
or
FUNCTION
subprogram must not refer to an argument of the subprogram.
For example, the following statement sequence is prohibited:
SUBROUTINE A(B,C)
EQUIVALENCE (B,X)

You cannot cause two different substrings of the same
character variables or arrays to become equivalenced to each
other. For example,
the following statement sequence is
prohibited because it specifies the same substring B(1:3) for
A ( 1 : 3) and A ( 2 : 4) :
CHARACTER A(3)*4,B*4
EQUIVALENCE (A(l) (1:3) ,B(1:3»,

(A(l) (2:4) ,B(1:3»

You also cannot use the EQUIVALENCE statement to assign
memory locations in a way that is inconsistent with the
normal linear storage of character variables and arrays.
For
example,
the following statement sequence is prohibited
because it would require the character variable B(2:2) to be
equivalent to both A(l) (2:2) and A(l) (1:1):
CHARACTER A(2)*lO,B*lO
EQUIVALENCE (A(l) (1:3) ,B(1:3», (A(l) (4:6) ,B(5:7»

Restrictions on EQUIVALENCE and COMMON:
1.

You cannot use the EQUIVALENCE statement to equivalence two
elements in different common blocks.
Thus, the following
statement sequence is prohibited:
COMMON /BLOCKl/A,B,F/BLOCK2/C,D,E
EQUIVALENCE (A,C)

You cannot set two quantities declared in a COMMON block to
be equivalent to one another.
Thus, the following statement
sequence is prohibited:
COMMON A,B,C
EQUIVALENCE (A,C)

7-15

SPECIFICATION AND DATA STATEMENTS
3.

Quantities placed in a common area by means of an EQUIVALENCE
statement are permitted to extend the end of the common area
forward.
For example, the statements:
COMMON/R/X,y,Z
DIMENSION A(4)
EQUIVALENCE (A,Y)
cause the common block R to extend from Z to A(4) arranged as
shown in Figure 7-4.

Location X

Location Y and A(1)

}

1----------1

Shared Locations

Location Z and A(2)

Location A(3)
Location A(4)
MR-S-1746-S1

Figure 7-4:

Valid Equivalencing

You cannot use EQUIVALENCE statements that cause the start of
a common block to be extended backwards.
For example, the
invalid sequence:
COMMON/R/X,Y,Z
DIMENSION A(4)
EQUIVALENCE(X,A(3»
would require A(l) and A(2) to extend the starting location
of block R in a backwards direction as illustrated in Figure
7-5.

Location A(1)
Location A(2)

Location X and A(3)
t-----------I

}
Causes COMMON R to Extend Backward

Location Y and A(4)

Location Z
MR-S-'1747-S1

Figure 7-5:

Invalid Equivalencing

7-16

SPECIFICATION AND DATA STATEMENTS
7.5.1

EQUIVALENCE and Extended Addressing

When extended addressing is in effect, and an EQUIVALENCE statement
causes a variable to be in COMMON, that variable resides in the same
psect as the rest of the COMMON block.
For variables not in COMMON,
if you equivalence a
large variable
(default 10,000 or more words)
with other variables
(including
scalars), all these variables will reside in the large data psect.
For example,
REAL A(20000),X
EQUIVALENCE (A{l),X)
causes variable X to be placed in the large data psect,
equivalenced with a large array.

since

Conversely, if each equivalence variable is small (default less than
10,000 words),
they will reside in the small data psect, even if the
total size of the equivalence class is over the small variable limit.
For example,
REAL A(8000) ,B(8000)
EQUIVALENCE (A(8000), B(l»
The arrays will reside in the small data psect, because
smaller than 10,000 words.

each

one

See Sections
addressing.

extended

7.6

15.4.11

and

16.5

for

information

EXTERNAL STATEMENT

Any user subprogram name to be used as an argument to another
subprogram must appear
in an EXTERNAL statement in the calling
subprogram. The EXTERNAL statement declares names to be subprogram
names to distinguish them from other variable or array names.
The subprograms mentioned in the EXTERNAL statement cannot be FORTRAN
intrinsic
functions;
they can be only user-supplied functions,
subroutines, or block data subprograms.
(The INTRINSIC statement
discussed in Section 7.7 allows intrinsic function names to be used as
arguments.) The EXTERNAL statement has the following form:
EXTERNAL proc[,proc •.• ]
where:
proc

is the symbolic name of a user-supplied subprogram, the
name of a dummy argument associated with the name of a
subprogram, or a block data subprogram.

The EXTERNAL statement declares each symbolic name included in it to
be the name of an external procedure, even if a name is the same as
that of an intrinsic function.
For example, if SIN is specified in an
EXTERNAL statement
(EXTERNAL SIN), all subsequent references to SIN
are to a
user-supplied function name SIN,
not to the
intrinsic
function of the same name.
The name specified in an EXTERNAL statement can be used as an actual
argument to a subprogram, which can then use the corresponding dummy
argument in a function reference or a CALL statement.
7-17

SPECIFICATION AND DATA STATEMENTS
NOTE
Note that a complete function reference used as an
argument,
for
instance,
FUNC (B)
in CALL SUBR
(A(FUNC(B) ,C», represents a value, not a subpro.gram.
therefore,
A complete function reference is not,
defined in an EXTERNAL statement.
The interpretation of the EXTERNAL statement described above is
different from that of earlier versions of FORTRAN-10/20.
If the
/NOF77 compiler switch is specified (see Sections 16.1.3 and 16.2.3),
the subprogram names can be intrinsic functions.
For compatibility with previous versions of FORTRAN-10/20,
the names
of external subprograms can be preceded by an asterisk (*) or an
ampersand (&) within an EXTERNAL statement. For example,
EXTERNAL *SIN, &COS
declares SIN and COS to be user subprograms.
(If a prefixed
not an intrinsic function, then the prefix is ignored.)

name

Note that specifying an intrinsic function in an EXTERNAL statement
without a prefix (that is, EXTERNAL SIN) has no effect upon the usage
of the function name outside of actual argument lists.
If the name
has generic properties, they are retained outside the actual argument
list.
(The name has no generic properties within an argument list.)
The names declared in a program EXTERNAL statement are reserved
throughout the compilation of the program, and cannot be used in any
declarator statement other than a type statement.

7.7

INTRINSIC STATEMENT

The INTRINSIC statement allows you to use intrinsic function names as
arguments to subprograms.
See Section 13.1 for further information on
intrinsic functions.
The INTRINSIC statement has the form:
INTRINSIC fun[,funJ •••
where:
fun is the symbolic name of an intrinsic function.
The INTRINSIC statement declares each symbolic name included in it to
be the name of an intrinsic procedure. This name can then be used as
an actual argument to a subprogram, which can use the corresponding
dummy argument in a function reference or a CALL statement.
The appearance of a generic function name in an INTRINSIC
does not cause that name to lose its generic property.

7-18

statement

SPECIFICATION AND DATA STATEMENTS
NOTE
You cannot use a generic-only name in an INTRINSIC
statement.
The generic name must be the same as an
instrinic function name. For example,
INTRINSIC LOG
is illegal because there is no function named
'LOG'.
LOG is the generic name that selects between functions
such as ALOG, DLOG, or CLOG.
Only one appearance of a symbolic name is permitted in all of the
INTRINSIC statements of a program unit.
Also, a symbolic name must
not appear in both an EXTERNAL and an INTRINSIC statement in a program
unit.
An example of
follows:

the

use

the

EXTERNAL

and

INTRINSIC

statements

Main Program

EXTERNAL CTN
INTRINSIC SIN, COS

CALL TRIG(ANGLE,SIN,SINE)

CALL TRIG(ANGLE,COS,COSINE)

CALL TRIG(ANGLE,CTN,COTANT)

Subprograms
SUBROUTINE TRIG(X,F,Y)
Y = F(X)
RETURN
END
FUNCTION CTN(X)
CTN = COS(X)/SIN(X)
RETURN
END
In this example, when TRIG is called with a second argument of SIN or
COS, the function reference F(X) references the math library functions
SIN and COS; but when TRIG is called with a second argument of CTN,
F(X) references the user function CTN.

7-19

SPECIFICATION AND DATA STATEMENTS
7.8

PARAMETER STATEMENT

The PARAMETER statement allows you to
during compilation.

define

constants

symbolically

The general form of the PARAMETER Statement is:
PARAMETER

(p= c [ , p= c J ••• )

where:
p

is a symbolic name.

is a constant expression (except for expressions involving
multiplication,
division, or exponentiation of complex
numbers).
(See Chapter 3 for a description of FORTRAN
constants.)

The constant acquires the same data type as the symbolic name.
If the
symbolic name is of type integer, real, double precision, or complex,
the corresponding expression
(c)
must be an arithmetic constant
expression
(see Section 5.1.2).
If the symbolic name is of type
character or logical, the corresponding expression
(c)
must be a
character constant expression
(see Section 5.2.1)
or a logical
constant expression (see Section 5.3.1), respectively.
The data type of a symbolic name defined to be a constant is specified
by a
type-statement or
IMPLICIT statement preceding the defining
PARAMETER statement.
Also, if the length specified for
the symbolic
name of a character constant is not the default length of one, its
length must be specified by a type-statement or
IMPLICIT statement
preceding the symbolic name of the constant.
NOTE
The form and the interpretation to the PARAMETER
statement described above are different from those of
the PARAMETER statement provided in earlier versions
of FORTRAN-IO/20.
The earlier version is described
below. This form and
interpretation can still be
used, however a warning message will be issued.
This
form of the PARAMETER statement is:
PARAMETER p=c[,p=c ••• ]
The symbolic name acquires the same data type
constant.

the

During compilation,
the symbolic names are replaced
by
their
associated constants, provided the following rules are observed:
1.

Symbolic names may appear only where
acceptable.

Symbolic name references
statement definition.

Symbolic names must be unique with respect to all ather names
in the program unit.

must

7-20

appear

FORTRAN
after

constants
the

are

PARAMETER

SPECIFICATION AND DATA STATEMENTS
4.

Symbolic names must not be redefined in subsequent
statements.

Symbolic names must not be used as part of another constant,
such as within a character constant or as the count for a
Hollerith constant.

Symbolic names
specification.

part

DATA statements are used to supply the initial values
arrays, array elements, substrings, and COMMON areas.

variables,

7.9

must

not

used

PARAMETER

format

DATA STATEMENT

The form of the data statement is:
DATA nlist/clist/ [[,]nlist/clist/] •••
where:
nlist

identifies a set of items to be initialized.

clist

contains the set of values to be assigned the items
nlist.

For example, the statement:
DATA IA/5/,IB/10/,IC/15/
initializes variable IA to the value 5, variable IB to the value 10,
and variable IC to the value 15. The number of storage locations you
specify in the list of variables must be equal to the number of
storage locations you specify in its associated list of values.
If
not, a warning message is output.
When the value list specifies more storage locations than the variable
list, the excess values are ignored. When the value list specifies
fewer storage locations than the variable list, the excess variables
are not initialized.
The nlist portion of each nlist/clist/ set can contain the names of
one
or more variables, array names, array elements, character
substring names, or labeled COMMON variables.
You may specify an
entire array (unsubscripted array name) or a portion of an array in a
DATA statement nlist as an implied DO loop construct.
(See Section
10.4.9.2 for a description of implied DO loops.)
The form of an implied-DO list in a DATA statement is:
(dlist,i=~1,n2[,n3])

where:
dlist

is a list of array element names,
names, or implied-DO lists.

is the name of an integer
index variable.

variable,

called

nl,n2,n3

are integer expressions that can
constants and loop index variables.

contain

7-21

character

substring
the

loop

integer

SPECIFICATION AND DATA STATEMENTS
For example, the statement:
DATA (NARY(I),I=1,5)/1,2,3,4,5/
initializes the first five elements of array
NARY(2)=2, NARY(3)=3, NARY(4)=4, and NARY(5)=5.

NARY

NARY(l)=l,

When you use an implied DO loop in a DATA statement, the loop index
variable must be of type INTEGER, and the Initial, Terminal, and
Increment parameters of the loop must be of type INTEGER.
In a DATA statement, references to an array element or substring must
be integer expressions in which all terms are either integer constants
or indices of embracing implied DO loops.
These types of integer
expressions can include the exponentiation operator.
The clist portion of each nlist/clist/ set can contain one or more
numeric,
logical,
Hollerith,
octal, hexadecimal, or character
constants. You may specify literal data as either a Hollerith
specification,
for example, 5HABCDE, or a string enclosed in single
quotes,
for example,
'Abcde'.
Each ASCII data item is stored
left-justified and is padded with blanks if necessary.
When you assign the same value to more than one item in nlist, a
repeat specification may be used.
The repeat specification has the
form:
n*d
where:
n

is an integer that specifies how many times the value d is
to be used.
For example, a clist specification of /3*20/
specifies that the value 20 is to be assigned to the first
three items named in the preceding list.
The statement:
DATA M,N,L/3*20/
assigns the value 20 to the variables M, N, and L.

When the specified data type is not the same as that of the variable
to which it is assigned, FORTRAN converts the data item to the type of
the variable.
The type conversion is performed using the rules given
for type conversion in arithmetic assignments.
(See Table 8-1.)
Octal, logical, Hollerith, hexadecimal, and character constants are
not converted.
Sample Statement

Result

DATA PRINT,I,O/'TEST' ,30,"77/,(TAB(J) ,J=1,30)/30*5/

The first 30
elements of
array TAB are
initialized to
5.0.

DATA«A(I,J),I=1,5),J=1,6)/30*l.0/

No conversion
required.

DATA«A(I,J),I=5,10),J=6,15)/60*2.0/

No conversion
required.

7-22

SPECIFICATION AND DATA STATEMENTS

When character variables are initialized, length conversions are made.
The conversion is based on the following rules:
1.

If the constant contains fewer characters than the length of
the element in nlist, the rightmost character positions of
the element are initialized with spaces.

If the constant contains more characters than the length of
the element in nlist, the character constant is truncated to
the right.

Character constants and Hollerith constants can also be used to
initialize numeric variables.
The character string is stored left
justified in the numeric variable.
When the character
string
specified is longer than one numeric variable can hold, the string is
stored left justified across as many variables as are needed to hold
it.
If necessary, the last variable used is padded with blanks up to
its right boundary.
For character variables, each variable or array
exactly one character constant in the data list.
For example, assuming that X,
statement:

and

are

element

must

have

single-precision,

the

DATA X,Y,Z/'abcdefghijkl'/
causes:
X to be initialized to 'abcde'
Y to be initialized to 'fghij'
Z to be initialized to 'klbbb'
When a character string is to be stored in double-precision and/or
complex variables, and the specified string is only one word long, the
second word of the variable is padded with blanks.
For example, assuming that the variable C is complex, the statement:
DATA C/'ABCDE','FGHIJ'/
causes the first word of C to be initialized to 'ABCDE' and its second
word to be initialized to 'bbbbb'.
The string 'FGHIJ' is ignored.
The first word contains the real part of the complex variable; the
second word contains the imaginary part.
In addition, the following two forms of bit data constants are allowed
in DATA statements:
O'di ••• dn'
Z'hi ••• hn'
where di are octal digits and hi are hexadecimal digits with A-F
representing the decimal equivalent of 10-15. These constants are
right-justified.
Note that you can also use the double quote (") form
of octal constants as described in Section 3.7.

7-23

SPECIFICATION AND DATA STATEMENTS
7.10

SAVE STATEMENT

The SAVE statement retains the values stored in a variable, array, or
common block after execution of a RETURN or END statement in a
subprogram.
The SAVE statement has the following form:
SAVE [a[,a] ••• ]
where:
a

is a named common block name (preceded and
slash), a variable name, or an array name.

followed

NOTE
Ordinarily, the values of a~l variables are retained
after
execution
of a
RETURN or END statement.
However, when overlays are used,
the SAVE statement
must be used to ensure retention of values.
An entity specified by a SAVE statement within a program unit does not
become undefined upon execution of a RETURN or END statement in that
unit.
If the entity is in a common block, however,
it may be
redefined in another program unit that references that common.
Procedure names, the names of variables and arrays in a common
and dummy argument names cannot be used in a SAVE statement.

block,

A SAVE statement that does not explicitly contain a list is treated as
though it contained a list of all allowable items in the program unit
that contains the SAVE statement.
If a particular common block name is specified by a SAVE statement in
a subprogram of an executable program, it must be specified by a SAVE
statement in every subprogram in which that common block appears.
NOTE
It is not necessary to use the SAVE statement to
retain
the value of a blank common block;
the
definition status of blank common is automatically
retained after a RETURN or END statement.
Also, when the SAVE statement is used,
it is not
necessary to specify the LINK switch /OVERLAY:WRITABLE
when loading a program.

7-24

CHAPTER 8
ASSIGNMENT STATEMENTS

Assignment statements assign values to variables, array elements, or
character substrings.
There are four kinds of assignment statements:

8.1

Arithmetic assignment statements (see Section 8.1)

Logical assignment statements (see Section 8.2)

Statement Label assignment (ASSIGN) statements
8.3)

Character assignment statements (see Section 8.4)

(see

Section

ARITHMETIC ASSIGNMENT STATEMENT

You use statements of this type to assign numeric values to numeric
variables or array elements. Arithmetic assignment statements have
the form:
v=e
where:
v

is the name of the numeric variable or array element that is
to receive the specified value.

is an arithmetic expression.

In assignment statements, the equal symbol (=) does not imply equality
as it would in algebraic expressions; it implies replacement.
For
example, the expression v=e is interpreted as "the contents of the
location identified as v are to be replaced by the value of expression
e; the previous contents of v are lost."
When the type of the specified variable or array element name differs
from that of its assigned value, FORTRAN converts the value to the
type of its assigned variable or array element. Table 8-1 describes
the type conversion operations performed by FORTRAN for each possible
combination of variable and value types.

8-1

ASSIGNMENT STATEMENTS
Table 8-1:

Rules for Conversion in Mixed-Mode Assignments

Expression
Type (e)

Variable Type (v)

REAL

INTEGER

COMPLEX

DOUBLEPRECISION

LOGICAL

CHARACTER

Real

R,I

H,L

Integer

R,F,I

Complex

C,R

R,L

Doubleprecision

H,I

Logical

R,I

H,L

Octal

R,I

H,L

Hollerith

Character

DoubleOctal*

Legend
D = Direct replacement
C = Conversion from real to integer with truncation, overflow is
possible
F = Conversion from integer to real with rounding
R = Real part only
I = Set imaginary part to 0
H = High-order only
L = Set low-order part to 0
M = Convert with no truncation and no rounding
N = Convert with rounding; truncation can occur and overflow is
possible
0 = Round to one word of precision, overflow is possible
X = Not allowed
Notes

Octal numbers with 13 to 24 digits are termed double-octal.
Double-octals require two storage locations.
They are stored
right-justified and are padded with zeros to
fill
the
locations.

8-2

ASSIGNMENT STATEMENTS
&

Use the first two words of the Hollerith constant.
If the
Hollerith constant is only one word long, the second word is
padded with blanks.

% Use the first word of the Hollerith constant.
#

8.2

To convert double-octal numbers to complex, the low-order octal
digits are assumed to be the imaginary part, and the high-order
digits are assumed to be the real part of the complex value.

LOGICAL ASSIGNMENT STATEMENTS

Statements of this type are used to assign values to variables and
array elements of type logical.
Logical assignment statements have
the following form:
v=e
where:
v

is the name of a variable or array element

is a logical expression

For example, assuming that the variables L, F, M, and G
logical, the following statements are valid:

type

replaced

Results

Sample Statement
L=.TRUE.

The contents of L are
logical truth.

F=.NOT.G

The contents of F are replaced by
the complement of the contents of
G.

M=A.GT.T
or
M=A)T

L = ( (I • GT • H) • AND. (J <= K) )

8.3

are

If A is greater than
T,
the
contents
of M are replaced by
logical truth; if A is less than or
equal to T, the contents of Mare
replaced by logical false.
This
can also be read:
If A is greater
than T, then·M is tru~, otherwise,
M is false.
The contents of L are replaced by
either the true or false resultant
of the expression.

ASSIGN (STATEMENT LABEL) ASSIGNMENT STATEMENT

The ASSIGN statement is used to assign a statement label constant (a 1
to 5 digit statement number)
to a variable name. The form of the
ASSIGN statement is:
ASSIGN s TO i

8-3

ASSIGNMENT STATEMENTS
where:
s

is a statement number in the current program unit.

is a variable name.

For example, the statement:
ASSIGN 2000 TO LABEL
specifies that the variable
2000.

LABEL

references

the

statement

number

With the exception of complex, double-precision, or character, you can
use any type of variable in an ASSIGN statement.
Use the ASSIGN statement in conjunction with assigned GO TO control
statements
(see Chapter 9), or as a format identifier in an I/O
statement (See Chapter 10). The ASSIGN statement sets up statement
label variables that are then referenced in subsequent GO TO control
statements, or in format specifiers in I/O statements. The following
sequence illustrates the use of the ASSIGN statement:
555 TAX=(A+B+C)*.05

ASSIGN 555 TO LABEL

GO TO LABEL

8.4

CHARACTER ASSIGNMENT STATEMENT

The character assignment statement assigns the value of the character
expression on the right of the equal sign to the character variable,
array element, or substring on the left of the equal sign.
The form of the character assignment statement is:
v=e
where:
v

is a character variable, array element, or substring.

is a character expression.

If the length of the expression on the right side of the assignment is
greater than the length of the variable on the left side, the
character expression is truncated on the right.
If the length of the expression on the right side of the assignment is
less than the length of the variable on the left side, the character
expression is filled on the right with blanks.

8-4

ASSIGNMENT STATEMENTS
FORTRAN-10/20 allows overlap between the character expression and the
character variable, array element, or substring.
(That is, the
character positions defined in the character variable, array element,
or substring can be referenced in the character expression.) For
example, the following assignments are allowed:
CHARACTER *4 A,B
DATA A/'abcd'/,B/'efgh'/
A(1:3)
A(2:4)
B(2:4) = B(1:3)
After the above assignment statements, A is 'bcdd', and B is 'eefg'.
The expression must be of character data type.
You cannot assign
numeric value to a character variable, array element, or substring.

Note that assigning a value to a character substring does not affect
character positions in the character variable or array element not
included in the substring.
If a character position outside of the
substring has a value previously assigned, it remains unchanged.
If
the character position is undefined, it remains undefined.
Examples of valid and invalid character assignment statements follow.
All variables and arrays in the examples are assumed to be of
character data type.
Valid

= 'PROG2'
REVOL(l} = 'MAR'//'CIA'
LOCA(3:8) = 'PLANT5'

FILE

TEXT(I,J+l) (2:N-l)

NAME//X

Invalid
'ABC'

CHARS

(the left side must be a character variable,
array element, or substring reference)

CHARS

(expression on the right must be
da ta type)

8-5

character

CHAPTER 9
CONTROL ST1\TE.MENTS

FORTRAN object programs normally execute statement by statement in the
order in which they were presented to the compiler. The following
control statements, however, enable you to alter the normal sequence
of statement execution:
1.

CALL (Section 13.4.2.2)

CONTINUE (Section 9.5)

3•

DO (Sec t ion 9. 3)

DO WHILE (Section 9.3.2)

ELSE (Section 9.2.4)

ELSE IF THEN (Section 9.2.4)

END (Section 9.8)

END DO (Section 9.4)

END IF (Section 9.2.4)

10 •

GO TO ( Sec t ion 9. 1 )

11.

IF (Section 9.2)

12.

IF THEN (Section 9.2.4)

13.

STOP (Section 9.6)

14.

PAUSE (Section 9.7)

15.

RETURN (Section 13.4.4)

The CALL and RETURN statements are described in Sections 13.4.2.2 and
13.4.4, respectively.
The remaining statements are described in this
chapter.

9.1

GO TO STATEMENTS

A GO TO statement causes the statement that it identifies
executed next, regardless of its position within the program.

There are three kinds of GO TO statements: Unconditional (see Section
9.1.1), Computed
(see Section 9.1.2), and Assigned
(see Section
9.1.3).
9-1

CONTROL STATEMENTS
9.1.1

Unconditional GO TO Statements

An unconditional GO TO statement
specified statement label.

transfers

program

control

the

You can position an unconditional GO TO statement anywhere in
source program, except as the terminating statement of a DO loop.

the

The form of the unconditional GO TO statement is:
GO TO s
where:
s

is a statement label of an executable statement.

For example:
GO TO 300

9.1.2

Computed GO TO Statements

The form of a computed GO TO statement is:
GO TO ( s

[, s J ••• ) [ , J e

where:
(s[,s] ••• )

is a list of statement labels.

is an integer expression.

You may include any number of statement labels in the list of a
computed GO TO statement.
However, each statement label must appear
within the program unit containing the GO TO statement.
The same
statement label can appear more than once in the list.
The value of the expression must be an integer value
(it will be
truncated if necessary) that is greater than 0 and less than or equal
to the number of statement labels given in the list.
If the value of
the expression is not within this range, the next sequential statement
is executed.
When a computed GO TO statement is executed, the value of the
expression is computed first.
The value of the expression specifies
the position of the label (within the given list of statement labels)
that identifies the statement to be executed next.
For example, in
the statement sequence:
GO TO (20, 10, 5)K
CALL XRANGE(K)
the variable K acts as a switch, causing a transfer to statement 20 if
K=l, to statement 10 if K=2, or to statement 5 if K=3.
The subprogram
XRANGE is called if K is less than 1 or greater than 3.

9-2

CONTROL STATEMENTS
9.1.3

Assigned GO TO Statements

The form of an assigned GO TO statement is:
GO TO i

[[,]

(s [, s] ••• ) ]

where:
i

is a variable name and the optional parenthesized list is a
list of statement labels.
The statement labels specified
must appear within the program unit containing the GO TO
statement.

Assigned GO TO statements must be logically preceded by an ASSIGN
statement
(see Section 8.3) that assigns a statement label value to
the variable i.
The assigned GO TO statement transfers program
control to the label that has been ASSIGNed.
The statement label value assigned must appear within the same program
unit as the GO TO statement that uses that value.
In statements with
a specified list, if i is not assigned one of the statement label
values given in the list, the next sequential statement is executed.
Examples:
ASSIGN 300 TO STAT1
GO TO STATl
GO TO STAT1, (177,300,777)

9.2

IF STATEMENTS

There are four kinds of IF statements:
arithmetic
(see Section
9.2.1), logical
(see Section 9.2.2), logical two-branch (see Section
9.2.3), and block IF (see Section 9.2.4).

9.2.1

Arithmetic IF Statements

The form of the arithmetic IF statement is:
IF (e) sl, s2, s3
where:
e

is an expression enclosed within parentheses and sl, s2, and
s3 are statement labels of three executable statements
appearing within the program unit containing
the
IF
statement.
The expressIon e must not be of type complex.
The same statement label can appear more than once in the IF
statement.

This type of IF statement transfers control of the program to one of
the given statements according to the computed value of the given
expression.
If the value of the expression is:
1.

Less than 0, control
identified by label sl.

transferred

the

statement

Equal to 0, control is
identified by label s2.

transferred

the

statement

9-3

CONTROL STATEMENTS
3.

Greater than 0, control
identified by label s3.

transferred

the

statement

Examples:
IF(ETA)4, 7, 12

Transfers control to statement 4 if
ETA is negative, to statement 7 if
ETA is 0, and to statement 12 if
ETA is greater than O.

IF(KAPPA-L(10»20, 14, 14

Transfers control to statement 20
if KAPPA is less than the 10th
element of array L and to statement
14 if KAPPA is greater than or
equal to the 10th element of array
L.

9.2.2

Logical IF Statements

The form of the logical IF statement is:
IF (e)

where:
e

is any expression.
complex.

The

expression

is an executable statement.

must

not

type

If the value of the expression is true
(negative), control is
transferred to the executable statement within the IF statement.
If
the value of the expression is false
(nonnegative), control is
transferred
to
the next sequential executable statement.
The
statement you give in a logical IF statement may be any executable
statement except a DO statement, an END statement, or another logical
IF statement.
Examples:
IF(T.OR.S) X=Y+l

Performs
an
arithmetic
assignment operation if the
result of the IF is true.

IF(Z.GT.X(K»

Performs a subroutine call if
the result of the IF is true.

CALL SWITCH(S,Y)

IF(K.EQ.INDEX) GO TO 15

9.2.3

Performs
an
unconditional
transfer
if the result of the
IF is true.

Logical Two-Branch IF Statements

The format of a logical two-branch IF statement is:
[F

(e)

sl, s2

9-4

CONTROL STATEMENTS
where:
e

is any expression, and sl and s2 are statement labels
appearing
within
the program unit containing the IF
statement. The expression must not be of type complex.

Logical two-branch IF statements transfer control to either statement
sl or s2, depending on the computed value of the given expression.
If
the value of the given logical expression is true (negative), control
is transferred to statement sl.
If the value of the expression is
false (nonnegative), control is transferred to statement s2.
Examples:
IF (LOGl) 10,20

Transfers control to statement 10
if
LOGl
is
true
(negative) ;
otherwise transfers
control
to
statement 20.

IF (A.LT.B.AND.A.LT.C) 31,32

Transfers control to statement 31
if A is less than both Band Ci
transfers control to statement 32
if A is greater than or equal to
either B or C.

9.2.4

Block IF Statements

Block IF statements conditionally execute blocks
statements.
The four block IF statements are:
•

IF THEN

•

ELSE IF THEN

•

ELSE

•

END IF

(or

groups)

The block IF
These statements are used in block IF constructs.
where the ELSE IF THEN and ELSE
construct has the following form,
statements are optional:
IF (e) THEN
block
ELSE IF (e) THEN
block

ELSE
block
END IF
where:
e

is a logical expression.

block

is a sequence of zero or more complete FORTRAN statements.
This sequence is called a statement block.
9-5

CONTROL STATEMENTS
Each block
IF statement, except the END IF statement, has an
associated statement block.
The statement block consists of all the
statements following the block IF statement up to (but not
including)
the next block IF statement in the block IF construct.
The statement
block is conditionally executed based on the values of logical
expressions
in
the preceding block IF statements.
A statement block
can be empty.
The IF THEN statement begins a
block IF construct.
The block
following it is executed if the value of the logical expression in the
IF THEN statement is true.
The first statement of the block cannot
directly follow the THEN on the same line.
For example, the following
is illegal:
IF (T.LT.X) THEN T

The correct form is:
IF (T.LT.X) THEN
T = X
The ELSE statement specifies a statement block to be executed
if no
preceding statement block in the block IF construct was executed.
The
ELSE statement is optional.
The ELSE IF THEN statement is similar to the ELSE statement, except it
requires an additional condition for execution.
The ELSE IF THEN
statement specifies a statement block to be executed if both the value
of the logical expression in the ELSE IF THEN statement is true, and
no preceding statement block in the block IF construct was executed.
A block
IF construct can contain any number of ELSE
IF THEN
statements.
The ELSE IF THEN statement is optional.
The END IF statement terminates the block IF construct.
Figure 9-1 describes the flow of control for four examples of block IF
constructs.

9-6

CONTROL STATEMENTS

CONTROL STATEMENTS
Construct

Flow of Control

IF (e) THEN
block
END IF
Execute
block

IF (e) THEN
block1
ELSE
block2
END IF

IF (e 1) THEN
block1
ELSE IF (e2) THEN
block2
END IF

IF (e1) THEN
block1
ELSE IF (e2) THEN
block2
ELSE IF (e3) THEN
block3
ELSE
block4
END IF

False

Execute
block1

Execute
block2

Execute
block 1

Execute
block2

False

Execute
block1

Execute
block2

Execute
block3

Execute
block4

MR-S-2525-83

Figure 9-1:

Examples of Block IF Constructs

After the last statement in a statement block is executed, control
passes
to the next executable statement following the END IF
statement. Consequently, at most one statement block in a block IF
construct is executed each time the IF THEN statement is executed.

9-7

CONTROL STATEMENTS
ELSE IF THEN and ELSE statements can have statement labels, but these
labels cannot be referenced.
The END IF statement can have a
statement label to which control can be transferred, but only from
within the block IF construct.
Section 9.2.4.1 describes restrictions on statements in a statement
block.
Section 9.2.4.2 describes examples of block IF constructs.
Section 9.2.4.3 describes nested block IF constructs.

can
contain
any
9.2.4.1 Statement Blocks - A statement block
executable FORTRAN statement except an END statement (see Section
9.8).
You can transfer control out of a statement block, but you
cannot transfer control back into the block.
Note that you cannot
transfer control from one statement block into another.
DO loops cannot overlap statement blocks.
When a statement block
contains a DO statement (see Section 9.3), it must also contain the DO
loop's terminal statement or END DO statement. Conversely, if a block
IF construct appears within the range of a DO loop, the corresponding
END IF statement must also appear within the range of that DO loop.

9.2.4.2 Block IF Examples - The simplest block IF construct consists
of the IF THEN and END IF statements; this construct conditionally
executes one statement block.
Form

Example

IF (e) THEN
block
END IF

IF (LOWER.LE.UPPER) THEN
MIDDLE=(LOWER+UPPER)/2
END IF

The statement block consists of all the statements between the IF THEN
and END IF statements.
The IF THEN statement first evaluates the logical expression
(e),
(LOWER.LE.UPPER).
If the value of e is true, the statement block is
executed.
If the value of e is false, control transfers to the next
executable statement after the END IF statement; the block is not
executed.
The following example contains a block IF construct with
THEN sta teme'nt:
Form

Example

IF (el) THEN
blockl
ELSE IF (e2) THEN
block2
END IF

IF (ITEM.LT.A(MIDDLE»
THEN
UPPER=MIDDLE-l
ELSE IF (ITEM.GT.A(MIDDLE»
THEN
LOWER=MIDDLE+l
END IF

ELSE

Blockl consists of all statements between the IF THEN and the ELSE IF
THEN statements;
block2 consists of all the statements between the
ELSE IF THEN and the END IF statements.
If ITEM is less than A(MIDDLE) , blockl is executed.
If ITEM is not less than A(MIDDLE) ,
A(MIDDLE) , block2 is executed.

9-8

but

ITEM

greater

than

CONTROL STATEMENTS
If ITEM is not less than A(MIDDLE)
and ITEM is not greater than
A(MIDDLE), neither blockl nor block2 is executed; control transfers
directly to the next executable statement after the END IF statement.
The following example contains a
statement:

block

construct

Form

Example

IF (e) THEN
blockl
ELSE
block2

IF (ITEM.GT.A(MIDDLE»
LOWER=MIDDLE+I
ELSE
SEARCH=MIDDLE
RETURN
END IF

END IF

with

ELSE

THEN

Blockl consists of all the statements between the IF THEN and the ELSE
statements; block2 consists of all the statements between the ELSE and
the END IF statements.
If ITEM is greater than A(MIDDLE), blockl is executed.
If ITEM is not greater than A(MIDDLE), block2 is executed.

9.2.4.3 Nested Block IF Constructs - A block IF construct can be
included in a statement block of another block IF construct. But the
nested block IF construct must be completely contained within a
statement block; it must not overlap statement blocks.
The following example contains a nested block IF construct.
Form

Example
FUNCTION SEARCH(A,N,ITEM)
CHARACTER*(*) A(N),ITEM
INTEGER SEARCH,N,LOWER,MIDDLE,UPPER
LOWER=l
UPPER=N
IF(LOWER.LE.UPPER) THEN
MIDDLE=(LOWER+UPPER)/2

IF (e) THEN

blockl

IF (e) THEN
10
blocka
ELSE IF (e) THEN
blockb
ELSE
blockc

IF (ITEM.LT.A(MIDDLE»
THEN
UPPER=MIDDLE-1
ELSE IF (ITEM.GT.A(MIDDLE»
THEN
LOWER=MIDDLE+l
ELSE
SEARCH=MIDDLE
RETURN
END IF

END IF

GOT010
END IF

END IF
20 SEARCH=O
RETURN
END

9-9

CONTROL STATEMENTS
If LOWER is less than or equal to UPPER, blockl is executed.
Blockl
contains a nested block IF construct.
If ITEM is less than A(MIDDLE) ,
blocka is executed.
If ITEM is greater than A(MIDDLE)
blockb is
executed.
If ITEM is equal to A(MIDDLE) , blockc is executed.
If LOWER is greater than UPPER, control is transferred to the first
executable statement after the last END IF statement. The nested IF
construct is not executed.

9.3

DO STATEMENT

The two types of DO statements are:
1.

Indexed DO (DO statement)

Pretested indefinite DO (DO WHILE statement)

The indexed DO statement is described in Section
WHILE statement is described in Section 9.3.2.

9.3.1

9.3.1,

and

the

Indexed DO Statement

DO statements simplify the coding of iterative procedures;
that is,
the statements in the DO statement range are executed repeatedly a
specified number of times.
The form of an indexed DO statement is:

/[,

Indexing Parameters

----=--

TERMINAL
STATEMENT
LABEL
INDEX
VARIABLE

I = el,

\~~IN~C~R~EM~EN~T~
\
(OPTIONAL)
TERMINAL
PARAMETER
PARAMETER

INITIAL
PARAMETER

MR-S-1760-81

where:
s

Terminal statement label s identifies the last statement of
the DO statement range.
The statement must follow the DO
statement in the same program unit.
If s
is omitted,
then
the loop must be terminated by an END DO statement (see
Section 9.4).
The terminal statement can be any executable statement
than one of the following:
•

Unconditional or assigned GO TO statement

•

Arithmetic IF or logical two-branch IF statement

•

Block IF, ELSE IF, ELSE, or END IF statement

9-10

other

CONTROL STATEMENTS
•

RETURN statement

•

STOP statement

•

END statement

•

DO statement

If the terminal statement is a logical IF, it can contain any
executable statement except one of the following:

•

DO statement

•

Block IF, ELSE IF, ELSE, or END IF statement

•

END statement

•

Another logical IF statement

Index variable i is an unsubscripted numeric variable whose
value is defined at the start of the DO statement operations.
The index variable must not be of type complex.
The index variable is available for use
execution of the range of the DO statement,
value within the DO loop does not change the
the DO loop will execute.
The DO loop index
available for use in the program when:

throughout each
but altering its
number of times
variable is also

Control is transferred outside the range of the DO loop
by a GO TO, IF, or RETURN statement located within the DO
range

Control is transferred outside the range of the DO loop
by an I/O statelnent with either or both the options END=
or ERR= (see Chapter 10)

A subprogram is executed from within the DO statement
range having the
index variable as an argument or in
COMMON

Initial parameter e1 assigns the index variable i its initial
value.
This parameter can be any expression, but cannot be
of type complex.

Terminal parameter e2 provides the value used to determine
how many repetitions of the DO statement range are performed.
This parameter can be any expression, but cannot be of type
complex.

Increment parameter e3 specifies the value to be added to the
initial parameter (e1) on completion of each cycle of the DO
loop. The increment parameter is optional.
If e3 and
its
preceding comma are omitted, e3 is assumed to be equal to 1.
This parameter can be any expression, but cannot be of type
complex.

9-11

CONTROL STATEMENTS
9.3.1.1 Executing an Indexed DO Statement - The indexing parameters
e1, e2 or e3 can be any expressions.
Their values are calculated only
once, at the start of each DO loop operation, to determine the values
for
the
initial,
terminal, and increment parameters.
If necessary,
the initial, terminal, and increment parameters are converted, before
use, to the data type of the index variable.
The number of times that a DO loop will execute, called the
count, is specified by the formula:

iteration

MAX (INT «e2-e1+e3)/e3) ,0)
If the iteration count is less than or equal to zero, the body of the
loop is not executed.
The index variable retains its assigned value
(e1) •
NOTE
The interpretation of the iteration count described
above is different from that of earlier versions of
FORTRAN-IO/20.
If the /NOF77 compiler switch is
specified
(see Sections 16.1.3 or 16.2.3), and the
iteration count is less than or equal
to
zero,
the
body of the loop is executed once.
Since the iteration count is computed at the start of a DO loop
operation, changing the value of the loop index variable within the
loop cannot affect the number of times that the loop is executed.
At the start of a DO loop operation, the index value
is set to the
value of the initial parameter
(el);
and the iteration count is
established.

9.3.1.2 DO Iteration Control - At the end of each DO loop cycle,
following steps are executed:
is

(e3)

added

the
the

The value of the increment parameter
index variable.

The iteration count is decremented.

If the
iteration count is greater than zero,
control
transfers to the first executable statement after the DO
statement for another iteration of the loop.

If the iteration count is less than
execution of the DO loop terminates.

equal

zero,

Exit from a DO loop upon completion of the number of
iterations
specified by the loop count is referred to as a normal exit.
If no
other DO loop shares the
terminal
statement, or
if this DO loop
statement is outermost,
control passes to the first executable
statement after the terminal statement of the DO loop.
The final value of the index variable is the value determined by
1.

9-12

step

CONTROL STATEMENTS
NOTE
The interpretation of the index variable described
above is different from that of earlier versions of
FORTRAN-IO/20.
If the /NOF77 compiler switch is
specified
(see Sections 16.1.3 or 16.2.3), the final
value of the index variable of the DO statement is
undefined after a normal loop exit.
Exit from a DO loop may also be accomplished by a transfer of control
by a statement within the DO loop range to a statement outside the
range of the DO statement.
This is called an extended range DO loop
(see Section 9.3.5).
When execution of a DO loop terminates, and other DO loops share its
terminal statement, control transfers outward to the next most
enclosing DO loop in the DO nesting structure (see Section 9.3.4).
Examples of DO Iteration Control:
100

DO 100 I = 1,10
J=I

After execution of these statements, 1=11 and J=lO.
switch is specified, I is undefined and J=lO).

200

L=O
DO 200 K
L=K

the

/NOF77

(If

the

/NOF77

5,1

After execution of these statements, K=5 and
switch is specified, K is undefined and L=5).

9.3.2

(If

L=O.

DO WHILE Statement

The DO WHILE statement is similar to the DO statement described in
Section 9.3.1.
Instead of executing a fixed number of iterations, the
DO WHILE statement executes for as long as a logical expression
contained in the statement continues to be true.
The form of the DO WHILE statement is:
DO [s[,]] WHILE (e)
where:
s

is the label of an executable statement that must physically
follow in the same program unit.

is a logical expression.

The DO WHILE statement tests the logical expression at the beginning
of each execution of the loop, including the first.
If the value of
the expression is true, the statements in the body of the loop are
executed;
if the expression is false,
control transfers to the
statement following the loop.
If no label appears in a DO WHILE statement, the DO WHILE loop must be
terminated with an END DO statement (see Section 9.4).

9-13

CONTROL STATEMENTS
The following example demonstrates the use of the DO WHILE statement:
CHARACTER*132 LINE
1=1

LINE (132:) = 'X,
DO WHILE (LINE (I: I)
I = I + 1
END DO

9.3.3

.EQ. '

The Range of a DO Statement

The range of a DO statement is defined as the series of statements
that follows the DO statement, up to and including the specified
terminal statement or END DO statement.
If another DO statement appears within the range of a DO statement,
the range of that statement must be entirely contained within the
range of the first DO statement. More than one DO statement may have
the
same labeled terminal statement but not unlabeled END DO
statement.
(See Section 9.3.4, Nested DO Statements.)
If a DO statement appears within an IF block, ELSE IF block, or ELSE
block
(see Section 9.2.4), the range of the DO statement must be
contained entirely within that block.
If a block IF statement appears within the range of a DO statement,
the corresponding END IF statement must also appear within the range
of the DO statement.

9.3.4

Nested DO Statements

One or more DO statements can be contained within the range of another
DO statement. This is called nesting.
The following rules govern the
nesting of DO statements:
1.

The number of nested levels
restricted to 79 DO loops.

(DO

loop

The range of each nested DO statement must be entirely within
the range of the containing DO statement (such as, they
cannot overlap).
For example:
Valid

Invalid

DO 1

DO 2

C
The ranges of
loop DO 2 and
DO 3 overlap

DO 3

c:::=

MR-S-17SB-B 1

9-14

within

loop)

CONTROL STATEMENTS
3.

More than one DO loop within a nest of labeled DO loops can
end on the same statement. When this occurs, the terminal
statement is considered ·to belong to the
innermost DO
statement that ends on that statement. Only a statement that
occurs within the range of the innermost DO statement can use
the statement label of the shared terminal statement for
transfer of control.
For example:
r°..;;;0_4.;....._ _ _ -411- Outermost DO Loop
004
DO 4

pO 4

____ Innermost DO Loop
Terminal Statement

l-..I.--'--.l-_ _ _ - - - -

MR·S·1759·81

Although all four DO loops share the same terminal statement,
the terminal statement "belongs" to the innermost DO loop.
4.

Nested loops cannot share an unlabeled END DO statement.
Each unlabeled END DO terminates exactly one DO loop.
For example:
Correctly Nested
DO Loops

Incorrectly Nested
DO Loops

DO 10 1=1,20

DO 10 1=1,5

DO J=1,5

DO J= 1,10

DO K= 1,10

CONTINUE

END DO

[

END DO
10

9.3.5

CONTINUE

MR·S·2526·83

Extended Range

By following certain rules, it is possible to transfer out of a DO
loop,
perform a
series of statements elsewhere in the program, and
then transfer back into the DO loop.
The statements that are executed
after a transfer out of a DO loop and before a transfer back into the
same DO loop are collectively known as the "extended range." A DO loop
that permits transfer
in and out of its range is called an extended
range DO loop.

9-15

CONTROL STATEMENTS
NOTE
This feature makes the flow of a program difficult
to
follow,
does not conform to the FORTRAN-77 standard,
and is therefore discouraged.
The following rules govern the use of extended range DO loops:
1.

The statement that causes the transfer out of the DO loop
must be contained within the most deeply nested DO (innermost
loop having the same terminal
statement).
This
loop must
also contain
the statement
to which the extended range
returns.

A transfer into the range of a DO statement is permitted only
if the transfer
is made from the extended range of that DO
statement.

The extended range of a DO statement must not contain another
DO statement.

The extended range of a DO statement cannot change the
variable or indexing parameters of the DO statement.

You can call a subprogram within an extended range.

The following exampLe illustrates the use
loop:

extended

index

range

200

DIMENSION TABLE(10,5), VALUE (10)
LOGICAL LOGARR(10)
DO 1000 I = 1, 10
An extended range DO loop
IF (LOGARR(I»
GOTO 500
Test logical array item
I=K
CALL SUBROT(K)
Invoke subroutine using
current index value
no 200 J= 1, 5
Nonextended range loop
'rABLE (I ,J)
0
CONTINUE

500

GOTO 2000
VALUE(I)

1000

CONTINUE

2000
2100

Extended range invocation
Invoke function GETVAL with
current index
Terminal statement for outer
loop

GETVAL(K)

STOP
TYPE 2100, I
FORMA T (' I = " I 2 )
LOGARR(I) = .TRUE.
GOTO 500

Extended range starts
Extended range ends and
returns

END

9.3.6

Permitted Transfer Operations

The following rules govern the transfer of program control from within
a DO statement range or the ranges of nested DO statements:
1.

A transfer out of the range of any DO loop is permitted at
any time.
When such a
transfer occurs, the value of the
controlling DO loop's
index variable
is defined as the
current value.
9-16

CONTROL STATEMENTS
2.

A transfer into the range of a DO statement is permitted
it is made from the extended range of the DO statement.

You can call a subprogram from within the range of any:
a.
b.
c.

DO loop
nested DO loop
extended range loop (in which you leave the loop through
a GO TO, execute statements in the extended range, and
return to the original loop)

The following examples illustrate the transfer
from within the ranges of nested DO statements:

Valid
Transfers

Invalid
Transfers

operations

permitted

•
extendedtrange

MR-S-1757-Bl

9.4

END DO STATEMENT

The END DO statement terminates the range of a DO or DO WHILE
statement.
The END DO statement must be used to terminate a DO block
if the DO or DO WHILE statement defining the block does not contain a
terminal-statement label.
The END DO statement may also be used as a
labeled terminal statement if the DO or DO WHILE statement does
contain a terminal-statement label.
The form of the END DO statement is:
END DO

9.5

CONTINUE STATEMENT

The form of the CONTINUE statement is:
CONTINUE
Execution of the CONTINUE statement has no effect.
the terminating statement of a DO loop.

9-17

It may be used

CONTROL STATEMENTS
In the following example, the labeled CONTINUE statement
legal termination for the range of the DO loop.

provides

DIMENSION STOCK(100)
DO 20 1=1,100
STOCK(I)=O
CALL UPDATE (STOCK(I»
IF(STOCK (I) .EQ. 0) GO TO 30
20 CONTINUE
STOP
30 TYPE 35
35 FORMAT (' UPDATE ERROR')
END

9.6

STOP STATEMENT

Execution of the STOP statement causes program execution to be
terminated.
A descriptive message may optionally be included in the
STOP statement to be output to your terminal
immediately before
program execution is terminated.
The form of the STOP statement is:
STOP [n]
where:
n

is an optional decimal integer constant of up to 6 digits,
or a character constant.
The constant is printed at the
terminal when the STOP statement is executed.
You can have any number of characters in the character
constant.
You can use continuation lines to accommodate
large character strings.
The constant is printed without
leading zeroes, unless they are specified in the statement.
NOTE
The word STOP is not printed when the STOP statement
is executed unless the word STOP is included in the
statement as a character constant.

The following examples show the results of executing STOP statements
that contain a 6-digit decimal string and a character constant.
10

PROGRAM TEST
STOP 123456
END

EXECUTE STOP1.FOR
FORTRAN: STOPI
TEST
LINK:
Loading
[LNKXCT TEST execution]
123456
Elapsed time 0.3
CPU time 0.1

9-18

CONTROL STATEMENTS

PROGRAM TEST
STOP 'The program has stopped'
END

EXECUTE STOP2.FOR
FORTRAN: STOP2
TEST
LINK:
Loading
[LNKXCT TEST execution]
The program has stopped
CPU time 0.1
Elapsed time 0.3

9.7

PAUSE STATEMENT

Execution of a PAUSE statement suspends the execution of the object
program and gives you the option of continuing execution of the
program, exiting from the program, or beginning a TRACE operation.
The form of the PAUSE statement is:
PAUSE [n]
where:
n

is an optional integer constant of up to 6 digits, or a
character constant.
The constant is printed at the terminal
when the PAUSE statement is executed.
You can have any number of characters in the character
constant.
You can use continuation lines to accommodate
large character strings.
The constant is printed without
leading zeros, unless they are specified in the statement.

If execution of the program is resumed after a PAUSE, program control
continues as if a CONTINUE had been executed. Execution of the PAUSE
statement causes the word PAUSE, the optionally specified constant,
and the following prompt to be printed at the terminal:
Type G to Continue, X to Exit, T to Trace
The responses to this prompt are:
G

continues program execution at
following the PAUSE statement.

causes program termination.

produces a trace back list at the terminal.
This list
consists of invoked routine names and locations, plus the
location and module names of the callers of those routines.
Using this information you can track the active path of
execution from the main program to the PAUSE trace routine.
(See Section 13.4.1.32 for a detailed description of this
feature.)

9-19

the

statement

immediately

CONTROL STATEMENTS
PROGRAM PTEST
PAUSE
PAUSE 234
PAUSE 'Character String'
END
EXECUTE PTEST.FOR
FORTRAN: PTEST
PTEST
LINK: Loading
[LNKXCT PTEST execution]
PAUSE
Type G to Continue, X to Exit, T to Trace.
G

PAUSE
234
Type G to Continue, X to Exit, T to Trace.
G
PAUSE
Character String
Type G to Continue, X to Exit, T to Trace.
X
CPU time 0.3

9.8

Elapsed time 18.8

END STATEMENT

This statement signals FORTRAN that the physical end of a program unit
has been reached.
END is an executable statement.
The general form
of an END statement is:
END
The following rules govern the use of the END statement:
1.

This statement must be the last physical statement
source program unit (main program or subprogram).

When executed in a main program, the END statement has the
effect of a STOP statement;
in a subprogram, END has the
effect of a RETURN statement.

An END statement may be labeled, but it must not be continued
(that is, it must appear only on an initial line).

9-20

CHAPTER 10
DATA TRANSFER STATEMENTS

FORTRAN I/O statements are divided into three categories by
as follows:

function,

Data Transfer Statements
transfer data between memory and
files.
The "files" can be devices such as TTY: or MTA:.
Internal files and
ENCODE/DECODE statements are used
for
memory-to-memory data transfers.

File Control Statements
associate and disassociate files
and
FORTRAN
logical
unit
numbers,
and can specify
characteristics of such an association.

Device Control Statements
position files.
For example,
using the device control statements you can position magnetic
tape to a particular file or record.

This chapter describes data transfer statements.
file-control and device-control statements.

Chapter 11 describes

Table 10-1 lists the three categories of I/O statements,
the
statements within each category, and the sections in which each I/O
statement is further described.

10-1

February 1987

DATA TRANSFER S.TATEMENTS
Table 10-1

FORTRAN I/O Statement Categories

directory

•

filename

•

filetype

•

file generation number

•

network attributes

default

file

following

file

node

name,

RMS

When yo u spec i fy any of the above components
in
the
FILE or NAME
specifier, they override those values specified in DEFAULTFILE.
Since
FOROTS processes DEFAULTFILE,
FILE,
and NAME in that order,
the
components specified
in NAME supercede all previous corresponding
components.

FORTRAN-10/20 VII

11-14

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS

DENSITY
Specifier

11.3.9

DENSITY Specifier

The DENSITY specifier is ignored except when used with magnetic
tape;
it permits you to specify the tape density.
If you do not specify a
tape density, FORTRAN assumes that you have set the density at monitor
level or that you are satisfied with the system default for the
device.
The form of the DENSITY specifier is:
DENSITY = dens
where:
dens

is a character expression having a value equal to
the following:

one

for

the

'200'
'556'
'800'
'1600'
'6250'
'SYSTEM'
SYSTEM specifies that the density
magnetic tape device being used.

the

default

density

DEVICE
Specifier

11.3.10

DEVICE Specifier

The DEVICE specifier enables you to specify the name of the device
involved in the data transfer.
A logical name always takes precedence
over a physical name.
The DEVICE arguments can specify I/O devices
located at remote stations, as well as logical devices.
The form of the DEVICE specifier is:
DEVICE= Character expression
If you omit this option, the logical name un (where un is the decimal
unit number}
is tried.
If this
is not successful, the standard
(default) device associated with the unit is used (see Table 10-3).

11-15

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS

DIALOG
Specifier

11.3.11

DIALOG Specifier

The DIALOG specifier enables you to type in additional OPEN specifiers
when the OPEN statement is actually executed.
If the DIALOG specifier is found in an OPEN list, then each time
the
OPEN statement
is executed FOROTS suspends execution and prompts the
terminal with an asterisk.
The form of the DIALOG specifier is:
DIALOG
You can respond
to
the asterisk prompt by
entering
a
file
spec if ica t ion,
DIALOG swi tches
(see below), or a file spec if ica t ion
followed by DIALOG switches.
The file specification may be a
full
file specification including the device, directory name, and so on.
NOTE
A DIALOG switch is any OPEN specifier (except DIALOG,
DIALOG=, UNIT, NAME, FILE, IOSTAT, or ERR) preceded by
a slash (/).
For example, when FOROTS enters DIALOG mode, you
such as:

can

type

string

*DSK:FOO.BAR/MODE:BINARY/ACCESS:DIRECT

DIALOG =
Specifier

11.3.12

DIALOG= Specifier

The DIALOG= specifier enables you to include all or a portion of the
OPEN specifiers
in a character expression.
The contents of the
character expression are interpreted as if you had given the DIALOG
specifier
(see above);
and, when the asterisk prompt was given, you
had typed in the same string as
is contained
in the character
expression, followed by a carriage return.
The form of the DIALOG= specifier is:
DIALOG= Character expression

11-16

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
Example:
OPEN(UNIT=l,DIALOG=' DSK:FOOeBAR/ACCESS:SEQOUT/MODE:ASC II')
I=22
WRITE(1,100)I
10 0 FORMAT ( I5 )
END

DIRECTORY (TOP5-10)
Specifier

11.3.13

DIRECTORY Specifier (TOPS-10)

On TOPS-10, the DIRECTORY specifier is ignored except for disk files.
It specifies the location of the User File Directory (UFD) and,
optionally, the Sub File Directory (SFD), either of which can contain
the file specified in the OPEN statement.
The form of the DIRECTORY specifier is:
DIRECTORY=

Character expression

The UFD is the directory in which a user's files are stored;
the SFD
exists within the UFD.
An SFD is often used to group files into
separate subdirectories.
The following is a sample of the UFD and SFD specification:
10,7,SFDA,SFDB
In the sample specification, 10,7 is the project-programmer number.
This is an adequate directory specification if the file is in the UFD
specified by 10,7.
The SFDA and SFDB specify two levels of subfile
directories.
The complete directory specification indicates that the
file is located in subfile directory SFDB.
As indicated, the path to
SFDB is through the UFD 10,7 and through the SFD SFDA.
NOTE
Refer to the TOPS-10 Monitor Calls
complete description of directories
directory structures.

Manual
for
a
and multilevel

The following is an example of a character expression specification:
DIRECTORY='10,7,SFD1,SFD2,SFD3'

---.--

project
SubFile
Programmer Directory
Number
Path

11-17

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
The following is an example of how to assemble
individual elements:

specification

from

CHARACTER*10 PROJ,PROG,PATHl,PATH2
CHARACTER*l COMM
PROJ= '10'
PROG= '7'
PATHl= 'SFDA'
PATH2= 'SFDB'
COMM= ' , '
OPEN(UNIT=I,DIRECTORY=PROJ//COMM//PROG//COMM//PATHI//COMM//PATH2)
The above specification
expression:

equivalent

the

following

character

'10,7,SFDA,SFDB'

DIRECTORY (TOPS-20)
Specifier

11.3.14

DIRECTORY Specifier

(TOPS-2~)

On TOPS-20, the DIRECTORY specifier is ignored except for disk files.
The DIRECTORY specifier enables you to define the path through the
directory structure to a file specified in the OPEN statement.
The form of the DIRECTORY specifier is:
DIRECTORY=

Character expression

The argument to the DIRECTORY specifier
is a character expression
whose elements comprise the directory path specification, for example:
OPEN(UNIT=22,DIRECTORY= 'GUEST')
!Looks for DSK:<GUEST)FOR22.DAT
or
CHARACTER*12 ID
ID= 'GUEST.CLASS3'
OPEN(UNIT=22,DIRECTORY=ID)
!Looks for DSK:(GUEST.CLASS3>FOR22.DAT

FORTRAN-10/20 VII

11-18

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS

DISPOSE
Specifier

11.3.15

DISPOSE Specifier

The DISPOSE specifier enables you to specify an action to
the file is closed.

occur

when

The form of the DISPOSE specifier is:
DISPOSE = dis
where:
dis

is a character expression having a value equal to
the following:

one

'KEEP'
'SAVE'
'DELETE'
'EXPUNGE'
'PRINT'
, LIST'
'PUNCH'
The DISPOSE specifier must have one of the following values:
KEEP

Specifies that the file is to be left where the OPEN
statement specifies.
DISPOSE='KEEP' is the default.

-SAVE

Same as KEEP.

DELETE

Specifies on TOPS-10 that, if the device
is either a
DECtape or disk, delete the file; otherwise, take no
action.
On TOPS-20, if the device involved is
the file; otherwise, take no action.

disk,

delete

EXPUNGE

On TOPS-10, same as DELETE.
On TOPS-20, if the device
involved
is a disk, expunge the file; otherwise, take
no action.

Specifies that the file will be printed and kept.
file must be on disk.

LIST

Specifies that the file will be
The file must be on disk.

PUNCH

Specifies that the file will be punched on the
tape punch and kept. -The file must be on disk.

11-19

printed

and

The

deleted.
paper

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS

ERR
Specifier

11.3.16

ERR Specifier

The ERR specifier enables you to designate a statement number of an
executable statement,
in the current program unit, to which control
passes if an error occurs during the execution of an I/O statement.
If an error occurs and no ERR specifier or
IOSTAT specifier
(see
Section 11.3.20) is supplied, the program types an error message.
If
the program is running under batch, it is aborted.
If the program is not running under batch, it enters DIALOG mode after
processing all of the other specifiers, as if you had supplied the
DIALOG specifier (see Section 11.3.11). This is true regardless of
whether or not the OPEN statement was explicitly executed or implied
by the execution of the first data transfer statement for a unit.
The form of the ERR specifier is:
ERR= s
where:
s

is the number of an executable statement to which program
control passes if an error occurs during the execution of
the statement that includes the ERR specifier.

The subroutine ERRSNS can be called to pinpoint the
Appendix D for FOROTS error values returned by ERRSNS.

error.

See

The FILE specifier enables you to name the file involved in the
transfer operation. You can specify a full file specification.

data

FILE
Specifier

11.3.17

FILE Specifier

The form of the FILE specifier is:
FILE= Character expression
The value of the character expression Is any legal TOPS-10 or TOPS-20
file specification.
(See the TOPS-10 Operating System Commands Manual
or the TOPS-20 User's Guide.)

11-20

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
If you omit the period and extension, the extension .DAT is assumed.
If just the extension is omitted, a null extension is assumed.
Thus,
if you want a filename without an extension,
remember to
use the
per iod.
If a filename is not specified, a default name is generated
the fo rm:

that

has

FORxx.DAT
where:
xx

is the FORTRAN logical unit number (decimal) or the logical
unit name for the default statements ACCEPT, PRINT, PUNCH,
READ, WRITE, or TYPE.

11.3.17.1 Remote File Specifications
(TOPS-20) - A
remote
file
specification is a
file specification that contains a nodename and
optionally, network access information in the following format:
node::dev:<dir>file[.ext.gen;USERID:id;PASSWORD:pwdiACCOUNT:act]
USERID, PASSWORD, and ACCOUNT attributes may be optional depending on
the default access privileges of the remote directory or file.
DEVICE
may be optional depending on the defaults
in effect at the
remote
host.
When the file specification is remote, RMS is used to access
the file.
For all remote access, RMS uses the default protection of
the remote directory or system.
NOTE
FORTRAN supports remote
file access only to other
TOPS-20
and VMS systems.
Note that logical names
defined at the local host will
not be expanded or
interpreted by FORTRAN/RMS before transmission to the
remote host.

FILESIZE
(INITIALIZE)
Specifier
(TOPS-10 only)

11.3.18

FILESIZE (INITIALIZE) Specifier (TOPS-10)

The FILESIZE (or INITIALIZE) specifier is used
for disk operations
only.
It enables you to estimate the number of words that an output
file is going to contain.
The form of the FILESIZE specifier is:
FILESIZE= Integer expression
The value assigned as a FILESIZE argument can be a integer expression,
and is rounded up to the next higher block boundary (multiple of 128).

FORTRAN-10/20 VII

11-21

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
The value specified by FILESIZE= is used as an estimate only.
The
effect of FILESIZE=
is
to help the monitor try to choose the best
place on the disk to put the file.

FORM
Specifier

11.3.19

FORM Specifier

The FORM specifier designates whether the records in a data
transfer
operation are formatted or unformatted.
You should not mix formatted
(character) and unformatted (binary) records in the same file.
The form of the FORM specifier is:
FORM

where:
ft

is a character expression having
'FORMATTED' or 'UNFORMATTED'.

value

equal

The arguments to the FORM specifier are:
FORMATTED

specifies that the records being
contain character (formatted) data.

transferred

UNFORMATTED

specifies that the records being
contain binary (unformatted) data.

transferred

, LIN ED',
the de fa u 1 t
If FORM is not specified and MODE is'ASCII' or
value
for
FORM is
'FORMATTED'.
Otherwise,
if MODE is 'BINARY',
FORM is
'IMAGE', or 'DUMP' (TOPS-10 only),
the default value for
'UNFORMATTED'.
If both FORM and MODE are specified and they are
incompatible,
then
DIALOG
mode
is
entered,
and
you are asked
to correct the
incompatibility.
In
the
following
example,
MODE='BINARY'
and
FORM='FORMATTED'
are specified in the same OPEN statement.
As shown
below, when the program
is executed,
interactive DIALOG mode
is
automatically
entered
to
enable
the
user
to
correct the
incompatibility.

101

PROGRAM TRIMP
OPEN(UNIT=l,MODE='BINARY' ,FORM='FORMATTED')
WRITE(UNIT=1,FMT=101)
FORMAT(lX,'This is a test.')
END

EXECUTE TRIMP
LINK:
Loading
[LNKXCT TRIMP execution]
?OPEN unit 1 DSK:FOR01.DAT at MAIN.+4 in TRIMP (PC 165)
?Incompatible attributes /MODE:BINARY /FORM:FORMATTED

11-22

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
[Enter correct file specs]
*/MODE:ASCII
CPU time 0.3
Elapsed time 11.4
TYPE FOR01.DAT
This is a test.
If neither FORM nor MODE is specified
the default value for
FORM
depends on the access.
If the access is SEQUENTIAL (or is defaulted),
the default for FORM is FORMATTED.
If the access is DIRECT or RANDOM,
the default for FORM is UNFORMATTED.
NOTE
For ASCII devices (line printer,
plotter,
terminal,
industry magnetic
tape),
the FORM= specifier has no
meaning and is ignored; both formatted and unformatted
data transfers are legal (see Section 10.3.3).

IOSTAT
Specifier

11.3.20

IOSTAT Specifier

The IOSTAT specifier identifies an integer variabie that
is
used
store the I/O status code during the execution of a statement.

The form of ,the IOSTAT specifier is:
IOSTAT= Integer variable or integer array element
If no error occurs during the execution of the statement, the
variable is set to zero.

defined

If an error does occur during the execution of the statement,
the
defined variable is assigned a positive integer value that corresponds
to the number of the FOROTS error that occurred (see Appendix D for
FOROTS error codes).
When an error occurs, no error message is typed; instead, the program
either continues at the ERR= statement number (if the ERR specifier is
included), or continues at the statement immediately following
the
OPEN statement (if no ERR specifier is included).

11-23

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS

KEY (TOPS-20)
Specifier

11.3.21

KEY Specifier (TOPS-20)

The KEY specifier defines the keys used by RMS to access records in an
i n de xed f i 1 e •
For new RMS
indexed
files
the KEY specifier
is mandatory.
The
presence
of
KEY=
causes
RMS to be
invoked,
whether or
not
ORGANIZATION= is specified.
If the KEY specifier is present without
the
ORGANIZATION=
being
specified,
ORGANIZATION= defaults
to
'INDEXED'.
If
ORGANIZATION=
and
KEY=
are
both
specified,
ORGANIZATION= must be 'INDEXED', or an error occurs or the ERR= branch
is taken.
For existing RMS indexed files, the KEY specifier
is optional;
if
given, it is ignored, although RMS does update the KEY XAB chain given
by the user with the key information from the file.
(See
Chapter
18
for more information on RMS data structures.)
The form of the KEY specifier is:
KEY = (kspec[,kspec] ••• )
where:
kspec

has the form:
e1:e2[:dt]

where:
el

is an integer expression specifying the first byte
position of the k~y.

is an integer expression specifying the last
position of the key.

is the optional data
type
INTEGER or CHARACTER.

the

key:

byte
either

The default data type of a key is CHARACTER.
The position of a
key
specification
in the list determines a key's key-of-reference number.
This number is used in any subsequent I/O statement to
specify the
same
key.
The primary key is key-of-reference number 0, the first
alternate key is key-of-reference number 1, and so forth.
The key fields and key-of-reference numbers are permanent attributes
of an indexed file and are established when the file is created. When
an existing file
is opened,
key definitions and
key-of-reference
numbers are obtained from the file itself.

FORTRAN-10/20 VII

11-24

February 1987

FI LE-CONTROL AND DEVICE-CONTROL STATEM.ENTS
The KEY specifier defines the access key for
records
in an
indexed
file.
The key starts at position el in the record and has a length of
e2-el+l.
The values of el and e2 must be such that:
1 .LE.
1 .LE.

(el) .LE.
(e2-el+l)

(e2) .LE. record-length
.LE. 255

You must define at least one key for a new indexed
file.
This
mandatory key is called the primary key of the file and usually has a
unique value of each record (this is the default condition).
You can
also define other keys, called alternate keys.
RMS allows up to 254
alternate keys.
When using keys of type INTEGER, RMS requires that key length e2-el+l
equals 4.
INTEGER key lengths other than 4 generate an error.

LIMIT
Specifier

11.3.22

LIMIT Specifier

The LIMIT specifier designates the number of output units
(such as
pages)
for
spooled
print or punch requests, which result from using
DISPOSE='PRINT',
DISPOSE='PUNCH', or DISPOSE='LIST'
(see
Section
11.3.15) •
The form of the LIMIT specifier is:
LIMIT= Integer expression

MAXREC (TOPS-20)
Specifier

11.3.23

MAXREC Specifier (TOPS-20)

The MAXREC specifier defines the maximum number of
records permitted
in a direct access RMS file.
If MAXREC is omitted, the maximum number
is unlimited.
If MAXREC is specified, RMS returns an error when a
record number greater than the value specified is written to the file.
This keyword is ignored for non-RMS files.
The form of the MAXREC specifier is:
MAXREC = Integer expression

FORTRAN-10/20 VII

11-25

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS

MODE

~____s_p_e_C_if_ie_r____
11.3.24

MODE Specifier

The MODE specifier defines the
record.

data

mode

external

file

is a character expression having a value equal to
the following:

one

After a MODE has been assigned (either explicitly or by default),
cannot be changed until the file is closed and then reopened.

The form of the MODE specifier is:
MODE

mod

where:
mod

'ASCII'
'LINED'
'BINARY'
, IMAGE'
'DUMP'

The default value of MODE depends on the values of FORM and ACCESS.
If FORM is FORMATTED, then the default MODE is ASCII.
If FORM is
UNFORMATTED, then the default MODE is BINARY.
If ACCESS is SEQUENTIAL
and no FORM= is specified, then the default MODE is ASCII; if ACCESS
is DIRECT or RANDOM, and no FORM= is specified, then the default MODE
is BINARY.
(See Section 11.3.19 for details on FORM, and Section
11.3.1 for details on ACCESS.)
Character data is supported in formatted BINARY and IMAGE mode
it is not supported in DUMP mode files.

files;

The possible values of MODE are:
ASCII

Specifies the data to be 7-bit ASCII characters.
Records are terminated with a line feed, form feed, or
vert i c'al tab.

LINED

Takes effect on input only.
For output, this mode
defaults to ASCII.
LINED specifies the data to be
7-bit ASCII characters with optional line sequence
numbers. FORTRAN removes the line sequence numbers, if
(The
present, before supplying the data to the user.
1 ine sequence number can be obtained by using the
function LSNGET, see Section 13.3.1.)
Note that a page mark in a file containing line
sequence numbers is a separate record. FORTRAN removes
the blank sequence number, and the carriage return/form
feed sequence is read as a blank record.
This value is ignored for RMS files.

FORTRAN-10/20 VII

11-26

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS

BINARY

specifies that data is formatted as a
FORTRAN binary
data
file.
A BINARY file is composed of 36-bit words
(see Section 10.3.2).
The first word of each record is
written by FOROTS and specifies the beginning of the
binary record; this 36-bit value is called a
type 1
Logical Segment Control' Word (LSCW).
Each binary record can contain one or more
type
2
LSCWs.
The
type
2 LSCW,
written by FOROTS under
certain conditions, is used to mark a record that spans
internal buffer boundaries.
A third FOROTS-written word, the type 3 LSCW, is always
written as the last 36-bit value in each BINARY record.

IMAGE

All data in a "BINARY" transfer remains equivalent
its external form with its internal representation.

For RMS files,
specifying BINARY is equivalent
specifying IMAGE.
RMS files do not contain LSCWs.

Specifies an unformatted binary mode.
Like the
BINARY
form of
unformatted
transfers,
IMAGE specifies that
data is transferred as 36-bit values, with the internal
and external
representation of the data remaining the
same.
Unlike BINARY files, IMAGE files do not contain
record
information
(LSCWs);
they contain only the data
transferred.
IMAGE files can be backspaced if a record
size is specified.

DUMP
TOPS-10

Corresponds to TOPS-10 DUMP mode I/O.
(See the TOPS-10
Monitor Calls Manual.)
Record
size
is
ignored.
Character data cannot be written into or read from DUMP
mode
files.
Note that
there
is
little
or
no
performance advantage
to
using
DUMP, as FOROTS uses
DUMP mode internally for all disk files.
NOTE

For ASCII devices (line printer,
plotter,
terminal,
industry magnetic
tape),
the MODE= specifier has no
meaning and is ignored; both formatted and unformatted
data transfers are legal (see Section 10.3.3).
Table 11-2 summarizes the different MODE arguments that are
on different devices.

supported

Table 11-2: DEVICE and MODE Cross-Table
MODE =
Device
Disk (sequential)
Disk (direct)
DECtape
Terminal
Magtape
Line Printer
Card Reader
Card Punch
Paper Tape Reader
Paper Tape Punch

'ASCII'

'LINED'

'BINARY'

'IMAGE'

X
X
X
X
X
X
X
X
X
X

X
X

X
X
X

X
X
X
X
X
X

.x
X
X
X

(TOPS-10)
'DUMP'

~--

FORTRAN-10/20 VII

11-27

._--,..,--

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS

NAME
Specifier

11.3.25

NAME Specifier

The NAME specifier is used to specify a full file specification.
You
can use
this specifier instead of the DEVICE, FILE, and/or DIRECTORY
specifiers.
The form of the NAME specifier is:
NAME= Character expression
Examples of the NAME specifier are:
(TOPS-H~):

NAME='DSK:FOO.BAR[10,34]'

(TOPS-20):

NAME='PS:<SMITH)BILLING.DAT'

The NAME specifier can not be given in DIALOG mode.
Also,
the OPEN
statement cannot have both a DIALOG= specifier and a NAME specifier.

NOSPANBLOCKS (TOPS-20)
Specifier

11.3.26

NOSPANBLOCKS Specifier (TOPS-20)

The NOSPANBLOCKS specifier enables you to specify that RMS records are
not to cross disk block (page) boundaries.
If any record exceeds the
size of the physical block, an error occurs.
The form of the NOSPANBLOCKS specifier is:
NOSPANBLOCKS
This specifier has no effect on non-RMS files or on
than new relative or sequential files.

FORTRAN-10/20 VII

11-28

RMS

files

other

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS

ORGANIZATION (TOPS-20]
Specifier

11.3.27

ORGANIZATION Specifier (TOPS-20)

The ORGANIZATION specifier determines the internal organization of an
RMS file.
This specifier invokes RMS (unless 'UNKNOWN' is specified).
The form of the ORGANIZATION specifier is:
ORGANIZATION

org

where:
org

is a character expression having a value equal
of the following:

one

'SEQUENTIAL'
'RELATIVE'
'INDEXED'
'UNKNOWN'
For new files, the default file organization is SEQUENTIAL.
For
existing
files, the value specified must be the same value as that of
the existing file, or an error message is displayed or the ERR= branch
is taken.

PADCHAR
Specifier

11.3.28

PADCHAR Specifier

The PADCHAR specifier is used only
operations.
PADCHAR enables you
used to pad fixed-length formatted
specified
record length (see the
11.3.33) •

with formatted output data transfer
to specify a character that will be
records, on output only,
to their
RECL(RECORDSIZE) specifier, Section

The form for the PADCHAR specifier is:
PADCHAR= A character expression in which the first
used

FORTRAN-10/20 VII

11-29

character

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS

if
ignored
The default pad character is space.
The pad character is
is,
if the
fixed-length
records
are
not
being
used
(that
I/O is not
RECL(RECORDSIZE) specifier is absent), or
if
formatted
being done.
NOTE
To specify a null character for the pad character, you
must use the function CHAR (see Chapter 13), since the
compiler does not allow null character constants,
for
example:
OPEN (UNIT=1,PADCHAR=CHAR(0))

PARITY
Specifier

11.3.29

PARITY Specifier

The PARITY specifier is only used for magnetic
tape operations.
It
permits you to specify parity to be observed (odd or even) during the
transfer of data.
The form of the PARITY specifier is:
PARITY = par
where:
par

is a character expression having a value equal to 'ODD' or
'EVEN' •

PROTECTION (TOP5-10)
Specifier

11.3.30

PROTECTION Specifier (TOPS-10)

This option specifies a protection code to be assigned
to
the data
file
being transferred.
The protection code is a 3-digit octal value
indicating the level of access to the file.
Each of the three numbers
in the
protection code has a
specific
meaning.
The
number
in
the
leftmost position designates the file
owner's protection; the middle number designates the project member's
protection; and the rightmost number designates the protection for all
others
users on the system.
The system default
for
the
file
protection is assigned if a protection is not specified.

11-30

February 1987

FILE~CONTROLAND

DEVICE-CONTROL STATEMENTS

On TOPS-10, the form of the PROTECTION specifier is:
PROTECTION= Integer expression
Figure 11-1 illustrates the
code.

TOPS-10

3-digit

octal

file

protection

NOTE
When setting the protection code for a file open for
output, be sure not to set the protection such that
the file is protected against writing by the program;
if you should do this, the OPEN statement will fail.
To protect the file against writing by the owner,
you
should give the PROTECTION specifier
in the CLOSE
statement.

tr---- File Owner
PROTECTION = nnn....-Other Users
+I----prOject Members

File owner

An octal digit in the first position specifies the file access for the file owner. The file owner
is the user whose programmer number matches the directory in which the file is
contained.

Project Members

An octal digit in the second position specifies the file access for the project members.
Project members are users whose project number matches the directory in which the file
is contained.

Other Users

An octal value in the third position specifies the file access for all users other than the file
owner or a project member.
MR-S-1748-B1

Figure 11-1:

TOPS-10 File Protection Number

Table 11-3 lists all
the possible values for each field
of the
protection code.
Refer to the TOPS-10 Monitor Calls Manual for more
information.
NOTE
The File Daemon referred
to
in Table
11-3 provides extended file protection.
The File Daemon allows any user to
specify who can and who cannot access
their files (if applicable).
(Refer to
the TOPS-10 Operating System Commands
Manual.)

11-31

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS

Table 11-3: TOPS-lO Protection Code Values
,-

Octal
Value

Meaning in Owner Field

1---

The file owner can execute, read, append to, update, write, rename, or change the protection of
the file. The File Daemon is not called on a protection failure.

The file owner can execute, read, append to, update, write, or rename the file. The File Daemon
is not called on a protection failure.

The file owner can execute, read, append to, update, or write the file. The File Daemon is not
called on a protection failure.

The file owner cannot access the file. However, the owner can use the RENAME monitor call to
change the file protection. The File Daemon is not called on a protection failure.

The file owner can execute, read, append to, update, write, rename, or change the protection of
the file. The monitor calls the File Daemon on a protection failure.

The file owner can execute, read, append to, update, or write the file. The monitor calls the File
Daemon on a protection failure.

The file owner can execute or read the file. The monitor calls the File Daemon on a protection
failure.

The file owner cannot access the file. However, the owner can use the RENAME monitor call to
change the file protection. The monitor calls the File Daemon on a protection failure.

Octal
Value

Meaning in Project Member and Other User Fields

The project member (or other) can execute, read, append to, update, write, rename, or change the
protection of the file.

The project member (or other) can execute, read, append to, update, write, or rename the file.

The project member (or other) can execute, read, append to, update, or write the file.

The project member (or other) can execute, read, append to, or update the file.

The project member (or other) can execute, read, or append to the file.

The project member (or other) can execute or read the file.
The project member (or other) can only execute the file.
The project member (or other) has no access to the file.
-------------------

On TOPS-10, PROTECTION can be an integer expression.
If the argument
is assigned a zero value or is not specified, the default protection
code established for the TOPS-10 installation is used.

PROTECTION (TOPS-20)
Specifier

11.3.31

PROTECTION Specifier (TOPS-29)

The PROTECTION specifier enables you to designate an octal protection
code for the file.
The protection code is a 6-digit octal value that
designates the access to the file of the owner, group members, and
other system users.

11-32

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
On TOPS-20, the form of the PROTECTION specifier is:
PROTECTION= Integer expression
Figure 11-2 illustrates a TOPS-20 protection code.
File Owner
r:==Grou p Member
Other Users

++
[fr==

PROTECTION =nnnnnn

File Owner

The leftmost two digits designate the file access for the file owner. The file owner is the
user who is connected to the directory in which the file resides.

Group Member

The middle two digits designate the file access for group members. Group members are
established by the system administrator. A group membership enables a user to share
files among other users in the same group.

Other Users

The rightmost two digits designate the file access for all users on the system, other than
the file owner or a group member.
MA-S-1749-81

Figure 11-2:

TOPS-20 Protection Number

NOTE
This specifier is ignored for remote files.
Instead,
the default protection of the remote directory or
system is used.
Table 11-4 lists the possible protection values and their meanings
the TOPS-20 file protection code.

Table 11-4: TOPS-20 Protection Code Values
Octal
Value

Meaning in Owner, Group Member, or Other User Fields

Permits full access to the file.

Permits read-only access to the file.

Permits write and delete access to the file.

Permits execute-only access to the file.

Permits append-only access to the file.

Permits listing of the file specification using the DIRECTORY command.

Permits no access to the file.

FORTRAN-10/20 VII

11-33

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
On TOPS-20, PROTECTION specifies a protection code to be assigned to
the data
file being transferred.
The protection code determines the
level of access that three classes of users have to the file.
PROTECTION takes an integer expression as an argument.
If PROTECTION
is assigned a zero value or is not specified, the default protection
code established for your connected directory is used.

READONlY
Specifier

11.3.32

READONLY Specifier

For local non-RMS files, the READONLY specifier
is used
to specify
that the program will only read from the file.
The accessing process
can read but not write to the file.
Output to the file is illegal and
will cause an error at execution time.
For RMS or remote files, if the first
process to access the
file
specifies READONLY,
the file
is open for read-only access by that
process, as for non-RMS files.
Any subsequent accessing process must
specify READONLY to obtain any a~cess to the file.
If READONLY is not
specified by the subsequent accessor, or
if SHARED
(see Section
11.3.35)
is specified, access to the file is denied with a "?File is
locked" error.
Thus, the access-sharing capabilities of the file are
determined by the first process to access the file.
The form of the READONLY specifier is:
READONLY

RECl
(RECORDSIZE)
Specifier

11.3.33

RECL

(RECORDSIZE) Specifier

For non-RMS files, the RECL (or RECORDSIZE) specifier enables you to
specify the number of characters or words in each record transferred.
RECL lS required
for
all
files
opened
for
direct
access
(ACCESS='DIRECT', 'RANDOM', or 'RANDIN').
(See Section 11.3.1.)
For RMS files, RECL specifies the Maximum Record Size (MRS).
The form of the RECL specifier is:
RECL= Integer expression

FORTRAN-10/20 VII

11-34

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
For non-RMS files:
•

In an ASCII transfer (MODE='ASCII'
or
'LINED'),
the value
assigned
to RECL specifies the number of characters in each
record.
For output to disk or CORE-DUMP tape files,
in addition to
these characters, FOROTS adds a carriage return/line feed to
each record, followed by enough null characters to
fill
the
current word,
so
that records are word-aligned.
RECL is
enforced on output by padding short records with the padding
character for formatted records.
Long records are truncated.

•

FORTRAN enforces a specified RECL for all
input operations.
If the RECL specified in an OPEN statement is different from
the actual size of the records, FOROTS reads the number of
characters specified by RECL.

•

For input to disk or CORE-DUMP tape files,
specifying a
record
size
directs FOROTS to
read
records that are
word-aligned.
The calculation of the actual
recordsize
is
the size specified,
plus two for the carriage return/line
feed, plus the number of nulls necessary to word-align the
record.

•

For INDUSTRY tapes, with RECORDTYPE='FIXED',
RECL specifies
the exact number of characters in each record; there are no
terminators or padding characters.
For INDUSTRY tapes with
RECORDTYPE='VARIABLE', RECL specifies the maximum record size
in the file, excluding the RCW.
When the record is read, regardless of the contents of the
record,
it
is
interpreted as specified by the rules above;
there are RECL characters of data, and the rest are
ignored.
No
interpretation is done of the characters in the data part
of the record.
These characters appear in the
FOROTS line
buffer exactly as they appear in the file, including nulls
and control characters.

•

In the case of MODE='LINED', the value of RECL excludes
five characters and tab in each line-sequence number.

the

•

In a binary transfer (MODE='BINARY', or 'IMAGE'),
the value
assigned to RECL specifies the number of 36-bit words in each
record.
For MODE='BINARY', the value in RECL excludes the
LSCWs written by FOROTS.
NOTE
If MODE=' DUMP'

is specified, RECL is ignored.

For RMS files:
•

If the file contains fixed-length records, RECL specifies the
size of each record.

•

If the file is a sequential file and contains variable-length
records,
RECL
is optional and specifies the maximum length
for each record.
If no RECL has been specified, RMS does not
perform maximum-length checking.

FORTRAN-10/20 V11

11-35

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
•

As with non-RMS files, if the records are formatted,
length is in bytes; if unformatted, in words.

the

•

If RECL is not specified, the MRS value
file was created is used.

the

•

If RECL is specified for an existing file with fixed-length
records or relative organization, and the value is different
from the actual length of the records in the file, an error
message is displayed, or the ERR= branch is taken.

•

If a program attempts to write a variable-length record that
is longer than the non-zero MRS value specified when the file
was created, an error is displayed or the ERR= branch is
taken.
Writing
fixed-length records that exceed RECL
produces a run-time warning, and the record is truncated to
the specified size.

•

RECL is required when creating a new relative or indexed
file, or when the RECORDTYPE is (or defaults to) FIXED.

specified

when

For local RMS files, RMS allows the following maximum record sizes for
file/record formats:
Organization

Record format

Maximum

Sequential
Sequential
Relative
Relative
Indexed
Indexed

Fixed-length
Variable-length
Fixed-length
Variable-length
Fi x ed-leng th
Variable-length

262,143 bytes
262,143 bytes
262,143 bytes
262,143 bytes
3,579 words
3,578 words

For remote files, the maximum size for an RMS record transfer cannot
exceed RMS's DAP buffer size. If the record exceeds this limit, a
"?Record too big" error is displayed or the ERR= branch is taken.

RECORDTVPE
Specifier

11.3.34

RECORDTYPE Specifier

The RECORDTYPE specifier defines the format of the records in a file.
The form of the RECORDTYPE specifier is:
RECORDTYPE = rtype
where:
rtype

is a character expression having a value equal
of the following:

one

'FIXED'
'STREAM'
'VARIABLE'
'UNKNOWN'
FORTRAN-10/20 VII

11-36

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
The possible values for RECORDTYPE are:
FIXED

is not defined for 36-bit format
(CORE-DUMP).
For
INDUSTRY magnetic
tape,
this is the standard ANSI
"F" tape record format.
Thus, the record data
is
written with no
terminators or carriage control
characters, one record directly after another,
to
fill
exactly fixed-length blocks.
For this record
format, a RECORDSIZE must be specified in the OPEN
statement.
With this record format, all physical blocks on the
tape are the same size except for the last block,
which may be a short block.

STREAM

For all tape formats, this record
format specifies
that for
formatted files, a standard stream record
terminator (CRLF) is
placed at the end of each
record, and
that a standard stream terminator is
expected on
input
to
delimit
records.
For
unformatted files, the behavior is identical to that
of MODE='IMAGE'.
With this record format, all physical blocks on the
tape are the same size except for the last block,
which may be a short block.
For 36-bit mode (CORE-DUMP), the bytes in the last
36-bit word after the last character of data will be
nulls.
For all tape formats,
if a RECORDSIZE is
specified
for
a
formatted file, the actual record
read or written consists of the data, CRLF, and the
number of nulls necessary to word-align the record.

VARIABLE

is not defined for 36-bit tape format
(CORE-DUMP).
For
INDUSTRY magnetic tape,
this is the standard
ANSI "D" tape record
format.
Thus,
records are
variable
in length, and the record data is written
with no terminators or carriage-control characters,
and preceded by a 4-byte record size (referred to as
the Record Control Word or RCW).
Blocks are variable length, not to exceed the block
length specified
in the OPEN statement.
If a
RECORDSIZE is specified in the OPEN statement,
this
is the maximum number· of bytes specified in the
4-byte RCW that precedes each record.
The
record
actually read or written
is four characters less
than specified in the RCW.

UNKNOWN

If the record type cannot be determined.

If no RECORDTYPE specifier is used, its
default record types are the following:

value

UNKNOWN,

File Type

Default Record Type

Non-RMS files

'STREAM'

RMS relative and indexed files

'FIXED'

All other RMS files

'VARIABLE'

FORTRAN-10/20 VII

11-37

and

the

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
NOTE
VMS FORTRAN may write
special
'segmented'
RMS
sequential files with embedded record control bytes as
part of the data for each RMS record in the file
(VMS
FORTRAN RECORDTYPE='SEGMENTED').
These files can be
read correctly only from VMS
FORTRAN.
Attempts
to read these files from a remote
TOPS-20 system using FORTRAN-10/20 VII will return two
bytes of control information as part of the data.

SHARED (TOPS-20)
Specifier

11.3.35

SHARED Specifier (TOPS-20)

The SHARED specifier, for RMS and remote non-RMS files, specifies that
the file is opened for write sharing:
at least one program is writing
a file and at least one other program is either reading or writing the
same
file.
For write sharing to occur, all programs that access the
file must specify SHARED.
Attempts by a subsequent accessor to write
a
record that has already been locked results in a "?Record locked"
error, or the ERR= branch is taken.
The form of the SHARED specifier is:
SHARED
If neither READONLY nor SHARED is specified,
the
file
is open for
exclusive access; that is, the initial accessor can read and write to
the file, but subsequent accessors are denied any access.
This specifier is ignored for local non-RMS files.

FORTRAN-10/20 VII

11-38

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS

STATUS
(TYPE)
Specifier

11.3.36

STATUS

(TYPE) Specifier

The STATUS (or TYPE) specifier lets you specify whether or not the
file being opened must exist, or what to do with the opened file when
it is closed.
The form of the STATUS specifier is:
STATUS = sta
where:
sta

is a character expression whose value is equal to
the following:

one

'EXPUNGE'
'OLD'
'NEW'
'SCRATCH'
'UNKNOWN'
'KEEP'
'DELETE'
The arguments to STATUS are:
EXPUNGE

On TOPS-10, this specifies that the file
is deleted
when it is closed.
On TOPS-20, this specifies that the
file is deleted and expunged when it is closed.
NOTE
On TOPS-10, any delete also expunges a file
from storage.
On TOPS-20, a DELETE operation
marks the file as deleted; an EXPUNGE operation
immediately erases the file from storage.

OLD

Specifies that the file being opened must already
exist.
If the file does not exist, an error results.

NEW

On TOPS-10, STATUS='NEW' specifies that the file must
not exist.
If the file does exist, an error results.
An error also occurs if you specify STATUS='NEW'
with
ACCESS='SEQIN' ,'SEQUENTIAL' ,'SEQINOUT'
(to a read-only
device), or 'RANDIN' (see Section 11.3.1).
On
TOPS-20,
the
STATUS=' NEW'
specifier
acts
differently,
depending
on what you have
in the
directory before STATUS=' NEW' is executed.

11-39

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
Also, the way you specify the file in the OPEN
statement which contains the STATUS='NEW' specifier
influences the way the STATUS=' NEW' specifier operates.
The
following list describes the ways that this
specifier can operate when used on TOPS-20.

SCRATCH

If the file specified in the OPEN statement does
not currently exist in the directory, and no
generation
number
is
specified,
then
the
STATUS='NEW' specifier creates the specified file
and gives it a generation number of 1.

If the file specified in the OPEN
statement
contains a name, extension, and generation number
that does not exist, the specified file is used.

If the file specified in the OPEN or
CLOSE
statement
contains
a
name,
extension,
and
generation number that is exactly the same as an
existing file in your directory, then STATUS='NEW'
causes an error, and no file is created.

If you did not specify a generation number, but the
file specified has the same name and extension as
an existing file in your directory, then the file
with the same name and extension and the next
highest generation will be created.

Specifies that the file will be automatically deleted
when the file is closed.
STATUS='SCRATCH' impl ies
DISPOSE='EXPUNGE' •
A SCRATCH file is always given a name by FOROTS.
The
name of the file is inaccessible to the FORTRAN
program.
If STATUS=' SCRATCH' is used, you must not specify FILE,
DIRECTORY, PROTECTION, or VERSION. If your program is
writing a file with STATUS='SCRATCH', and the file is
being written to disk, you can retain it by executing a
CLOSE statement that renames the file to a specified
name.
If a file is opened with STATUS='SCRATCH', the access
must
be
ACCESS='SEQUENTIAL'
(or
'SEQINOUT')
or
ACCESS='DIRECT' (or 'RANDOM') (see Section 11.3.1).

UNKNOWN

Specifies that a file opened for an input operation
must exist.
When a file is opened for output with
STATUS='UNKNOWN', if the file exists, it is superseded;
if it does not exist, it is created.
UNKNOWN is the default for STATUS. This value is used
unless you specify STATUS or unless the value of STATUS
is otherwise determined by the ACCESS specifier.

KEEP

Specifies that the file is not deleted.
STATUS='KEEP'
is
equivalent
to
DISPOSE='SAVE' and STATUS='UNKNOWN'.

11-40

Specifying
specifying

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
DELETE

On TOPS-10, specifies that the file will be erased when
the file is closed.
On TOPS-20, specifies that the
file will
be deleted
when the
file
is closed.
The file is erased when a
TOPS-20 EXPUNGE command
is given.
To
undelete a
deleted file, use the TOPS-20 UNDELETE command.

TAPEFORMAT

Specifier

11.3.37

]

TAPEFORMAT SPECIFIER

The TAPEFORMAT specifier defines the physical format of
tape.

the

magnetic

The form of the TAPEFORMAT specifier is:
TAPEFORMAT = tmode
where:
tmode

is a character
expression
'CORE-DUMP' or 'INDUSTRY'.

having

value

equal

The values for the TAPEFORMAT specifier are:
CORE-DUMP specifies the "DIGITAL-compatible" tape
format,
which
is 36-bits. stored
in
five frames on a 9-track tape.
The SET TAPE RECORDSIZE
(TOPS-20)
or SET BLOCKSIZE
(TOPS-10)
command
is
interpreted to be the number of
36-bit words in the magnetic tape blocks on the tape,
and
is used if no BLOCKSIZE specifier is given in the
OPEN statement.
If a BLOCKSIZE specifier is given
in
the OPEN statement, it is interpreted to be the number
of 36-bit words
for both formatted
and
unformatted
files.
INDUSTRY

specifies characters are read or written
in standard
industry tape format, one character per tape frame.

11-41

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS

UNIT
Specifier
(required)

11.3.38

UNIT Specifier (Required)

The UNIT specifier defines the FORTRAN logical unit number to be used.
FORTRAN devices are identified by assigned decimal numbers within the
range 0-99 (see Table l0-3).
UNIT must be an integer expression.
The form of the UNIT specifier is:
UNIT= Integer expression
If the unit specifier
is the
first
specifier given
in
the OPEN
statement,
the
keyword UNIT= is optional.
For example the following
statement opens a file on unit 20:
OPEN (20,FILE='MYFILE')
NOTE'
The FORTRAN standard
logical
unit assignments are
described
in Section 10.4.3.1. Although the range of
logical unit numbers shown in Table 10-3 is 0-99,
the
range of
UNIT numbers
is an
installation-defined
parameter.

USER OPEN (TOPS-20)
Specifier

11.3.39

USEROPEN Specifier (TOPS-20)

The USEROPEN specifier is used to specify the name of an optional
user-supplied
external
function.
This function can set various RMS
parameters that FORTRAN otherwise defaults,
and must perform the
actual
OPEN
in place of FOROTS.
This function is intended for use
with RMS files only; a USEROPEN function used with a
remote
non-RMS
file can produce indeterminate results.
The form of the USEROPEN specifier is:
USEROPEN

proc ed ur e- name

where:
proced ure-name

FORTRAN-10/20 VII

is the symbolic name of the USEROPEN function.
The function name must be declared EXTERNAL.

11-42

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
The function must perform both $OPEN and $CONNECT RMS service calls,
returning
AC0 non-zero on any non-recoverable failure.
If AC0 is
returned zero, FOROTS assumes the function has succeeded, and
program
execution continues.
If AC0 is returned non-zero, FOROTS assumes the
USEROPEN function has failed, and issues an error (or takes the ERR=
branch) based on the STS and STV values returned by RMS to FOROTS.
See Section 18.9 for more information on USEROPEN.

VERSION (TOP5-10)
Specifier

11.3.40

VERSION Specifier (TOPS-10)

Use the VERSION specifier for disk operations only; it enables you
assign a 12-digit octal version number to an output file.

The form of the VERSION specifier is:
VERSION=Integer expression

11.4

CLOSE STATEMENT

The CLOSE statement disassociates an active file from a logical
unit
and releases the memory occupied by I/O buffers and other unit-related
data.
The CLOSE statement can also change some of the
file
characteristics that were assigned during the OPEN, such as the name,
protection, directory, and disposition of the file.
Once a CLOSE statement has been executed, you must use another OPEN
statement (or implicit OPEN) to regain access to the closed file.
The form of the CLOSE statement is:
CLOSE (closelist)
where:
closelist

FORTRAN-10/20 VII

is a list of CLOSE statement specifiers.
This
list must contain the UNIT specifier (see Section
11.5.9)
and
can
optionally
contain
other
specifiers.

11-43

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
11.4.1

Implicit CLOSE

An implicit CLOSE occurs when FOROTS automatically closes a
logical
unit without execution of a CLOSE statement.
This can happen when a
program terminates, or when you execute an OPEN for a unit that is
already connected to another file.

11.5

CLOSE STATEMENT SPECIFIERS

All of the CLOSE statement specifiers are optional, except the UNIT
specifier which is required.
Some CLOSE statement specifiers have the
same formats as the corresponding specifiers in the OPEN statement.
Table 11-5 summarizes the specifiers in the CLOSE statement, and the
type of value required by each. A section number is provided to refer
to detailed descriptions of each specifier.

11-44

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
Table 11-5: Summary of CLOSE Statement Specifiers and Arguments
Argument
DEVICE =

Possible Value
Character expression

DIALOG

Section
11.5.1
11.5.2

DIALOG =

Character expression

11.5.3

DIRECTORY =

Character expression

11.5.1

DISPOSE =

Character expression with one of the following values:
'SAVE', 'DELETE', 'PRINT', 'KEEP', 'LIST', 'PUNCH',
'EXPUNGE'

11.5.4

ERR =

Statement number

11.5.5

FILE =

Character expression

11.5.1

IOSTAT=

Integer variable or integer array element

11.5.6

LIMIT =

Integer expression

11.5.7

NAME =

Character expression

11.5.1

PROTECTION =

Integer expression

11.5.1

STATUS =
TYPE =

Character expression with one of the following values:
'KEEP', 'DELETE', 'EXPUNGE'

11.5.8

UNIT =

Integer expression

11.5.9

NOTE
For
compatibility
with
previous
versions
of
FORTRAN-10/20, you can specify a numeric array name as
the value of each of the following specifiers:
DIALOG=
DIRECTORY
NAME
When a numeric array name is used, FOROTS assumes that
it contains a
string of characters terminated by a
null character.
In addition, you can specify a numeric variable as the
value of the
DEVICE and
FILE specifiers.
If the
variable is single precision, FOROTS assumes
that
it
contains 5 characters;
if
it
is double precision,
FOROTS assumes that is contains 10 characters.
numeric variables
The use of numeric array names and
is a nonstandard
in place of character variables
feature.

11-45

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS

DEVICE, DIRECTORY,
FILE, NAME, and
PROTECTION
Specifiers

11.5.1

DEVICE, DIRECTORY, FILE, NAME, and PROTECTION Specifiers

The CLOSE statement file identification specifiers can be
used when
you want to rename the output file when it is closed.
Their formats
are the same as the corresponding specifiers in the OPEN statement.
If any of these specifiers are given in the CLOSE statement, the
file
is renamed when
it
is closed.
If some, but not all of the file
identification parameters are specified on a CLOSE statement, only the
specified parameters are changed when the file is renamed.
Example:
OPEN(20,ACCESS='SEQOUT' ,FILE='FOO.BAR')

CLOSE(20,FILE='NEWFIL')
The above sequence renames the output file to DKSB:NEWFIL.BAR.
Refer to the following sections under the OPEN statement:
FILE - see Section 11.3.17
NAME - see Section 11.3.25
DEVICE - see Section 11.3.10
DIRECTORY (TOPS-10) - see Section 11.3.13
DIRECTORY (TOPS-20) - see Section 11.3.14
PROTECTION (TOPS-10) - see Section 11.3.30
PROTECTION (TOPS-20) - see Section 11.3.31

DIALOG
Specifier

11.~.2

DIALOG Specifier

The DIALOG specifier enables you to
type
in additional
specifiers when the CLOSE statement is actually executed.

If the DIALOG specifier is found in the CLOSE list, then each time the
CLOSE statement is executed, FOROTS suspends execution and prompts the
terminal with an asterisk.
The form of the DIALOG specifier is:
DIALOG

FORTRAN-10/20 Vll

11-46

February 1987

FILE-CONTROL AND

DEV~CE-CONTROL

STATEMENTS

You can respond
to the asterisk prompt by
entering
a
file
specification,
DIALOG switches' (see below), or a file specification
followed by DIALOG switches.
The file specification can be a
full
file specification including the device,.directory name, and so on.
If you enter a file specification that
is different from the
file
specification currently assigned to the file, FOROTS RENAMEs the file
after it is closed to the new name.
NOTE
A DIALOG switch is any CLOSE specifier (except DIALOG,
DIALOG=, UNIT, NAME, FILE, IOSTAT, or ERR) preceded by
a slash (/).

DIALOG
Specifier

11.5.3

DIALOG= Specifier

The DIALOG= specifier enables you to include all or a portion of the
CLOSE specifiers
in a character expression.
The contents of the
character expression are interpreted as if you had given the DIALOG
specifier
(see above);
and, when the asterisk prompt was given, you
had typed in the same string as
is contained
in the character
expression, followed by a carriage return.
The form of the DIALOG= specifier is:
DIALOG= Character expression
Example:
CLOSE (UNIT=20,ERR=10,DIALOG='/DISPOSE:DELETE')
When DIALOG= is given in the CLOSE list, it
other specifiers except DIALOG.

processed

after

all

DISPOSE
Specifier

11.5.4

DISPOSE Specifier

The DISPOSE specifier enables you to specify an action to
the file is closed.

occur

when

The form of the DISPOSE specifier is:
DISPOSE = dis

FORTRAN-10/20 VII

11-47

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
where:
dis

is a character expression having a value equal to
the following:

one

'KEEP'
, SAVE'
'DELETE'
'EXPUNGE'
'PRINT'
'LIST'
'PUNCH'
The DISPOSE specifier must have one of the following values:
KEEP

Specifies that the file is to be left on the connected
unit.
DISPOSE='KEEP'
is the default.
You can not
specify DISPOSE='KEEP' if
in the corresponding
OPEN
statement you specified STATUS='SCRATCH' (see Section
11.3.36) •

SAVE

Same as KEEP.

DELETE

Specifies on TOPS-10 that, if the device
is either
a
DECtape or disk, delete the file; otherwise, take no
action.
On TOPS-20, if the device involved is
the file; otherwise, take no action.

EXPUNGE

disk,

delete

On TOPS-10, same as DELETE.
On TOPS-20, if the device
involved
is a disk, expunge the file; otherwise, take
no action.
EXPUNGE deletes but does not expunge a local RMS file.
Depending
on
the capabilities of the remote system,
EXPUNGE may also expunge a remote RMS or non-RMS file.

Specifies that the file will be printed and
kept
(the
file
will
not be
kept
if you specify the
CLOSE
statement STATUS=' DELETE' or 'EXPUNGE').
The file must
be on disk.
This value is not supported for TOPS-20
RMS files; however, it is supported for VMS RMS files.

LIST

Specifies that the file will be printed, deleted,
and
expunged
(the
file will not be deleted if you specify
the CLOSE statement STATUS='KEEP').
The file must be
on disk.
This value is not supported for TOPS-20 RMS
files; however, it is supported for VMS RMS files.

PUNCH

Specifies that the file will be punched
on the
paper
tape punch and kept.
The file must be on disk.
This
value is not supported for RMS or remote non-RMS files.
NOTE

The value of the CLOSE statement DISPOSE specifier
supersedes the value of the OPEN statement DISPOSE
specifier and the OPEN statement STATUS specifier
if
STATUS='EXPUNGE', 'KEEP', and 'DELETE'.

FORTRAN-10/20 VII

11-48

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS

ERR
Specifier

11.5.5

ERR Specifier

The ERR specifier enables you to designate a statement label of an
executable statement,
in the current program unit, to which control
passes if an error occurs during the execution of an I/O statement.
The form of the ERR specifier is:
ERR= s
where:
s

is the number of an executable statement to which program
control passes if an error occurs during the execution of
the statement in which the ERR specifier is included.

The ERR specifier works the same way when it is given in a CLOSE as it
does when given in an OPEN statement (see Section 11.3.16).

IOSTAT
Specifier

11.5.6

IOSTAT Specifier

The IOSTAT specifier identifies an integer variable that is used
store the I/O status code during the execution of a statement.

The form of the IOSTAT specifier is:
IOSTAT= Integer variable or integer array element

LIMIT
Specifier

11.5.7

LIMIT Specifier

The LIMIT specifier designates the number of output units
(such as
pages)
for spooled print or punch requests, which result from using
DISPOSE=' PRINT',
DISPOSE=' PUNCH', or DISPOSE=' LIST'
(see
Section
11.5.4) •

11-49

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
The form of the LIMIT specifier is:
LIMIT= Integer expression

STATUS
Specifier

11.5.8

STATUS Specifier

it is
The STATUS specifier tells FOROTS what to do with a file when
closed.
In the CLOSE statement, STATUS values are a subset of the
DISPOSE specifier (see Section 11.5.4) values.
NOTE
The ANSI-77 standard does not have DISPOSE
allows STATUS='KEEP' or STATUS='DELETE'.

and

only

The form of the STATUS specifier is:
STATUS

= sta

where:
sta

is a character expression having a value equal to
the following:

one

'KEEP'
'DELETE'
'EXPUNGE'
The arguments to STATUS are:
KEEP

Specifies that the file is not deleted.

DELETE

On TOPS-10, specifies that the file is deleted.
On TOPS-20, specifies that the file
is marked for
deletion when the file is closed.
The file is erased
when a TOPS-20 EXPUNGE command is given. To undelete a
deleted file, use the TOPS-20 UNDELETE command.

EXPUNGE

On TOPS-10, the same as delete.
On TOPS-20,
specifies that the file is deleted and expunged.

this

EXPUNGE deletes but does not expunge a local RMS file.
Depending on the capabilities of the remote system,
EXPUNGE may also expunge a remote RMS or non-RMS file.
NOTE
The value of the CLOSE statement STATUS specifier
supersedes the value of the OPEN statement DISPOSE
specifier and OPEN statement STATUS specifier
if
STATUS=' EXPUNGE' , 'KEEP', and 'DELETE'.

FORTRAN-10/20 VII

11-50

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS

UNIT
Specifier
(Required)

11.5.9

UNIT Specifier (Required)

The UNIT specifier tells FOROTS which logical unit number
is to be
closed.
This specifier corresponds to the UNIT specifier in the OPEN
statement (see Section 11.3.38) and to the UNIT specifiers in the data
transfer statements (see Section 10.4.3).
The form of the UNIT specifier is:
UNIT= Integer expression
If the unit specifier is the
first
specifier given
in the CLOSE
statement,
the keyword
UNIT=
is optional.
For example, to close a
file on unit 20, you can use the following command:
CLOSE (20)

11.6

OPEN AND CLOSE STATEMENT EXAMPLES

The following are examples of OPEN and CLOSE statements:
OPEN (UNIT=l,DEVICE='DSK' ,ACCESS='SEQIN' ,MODE='BINARY')
causes a disk file named
FOR01.DAT
(since no
FILE= option was
specified) to be opened on unit 1 for sequential input in binary mode.
OPEN (UNIT=3,DEVICE='DSK' ,FILE='PAYROL.DAT',
1
ACCESS='RANDOM' ,MODE='ASCII' ,RECORDSIZE= 80,
2
ASSOCIATEVARIABLE=I,ERR=240)
Causes a disk file named PAYROL.DAT to be opened on unit 3 for
random
I/O operations in ASCII mode.
The records
in PAYROL.DAT are 80
characters long; the ASSOCIATE VARIABLE for this file
is
I.
If an
error occurs during
the execution of this OPEN statement, the OPEN
terminates, and control is transferred to the statement labeled 240.
OPEN
1
1

(UNIT=I, FILE='IDX.DAT',
ORGANIZATION='INDEXED' ,STATUS='NEW',
KEY=(1:15:CHARACTER) ,RECORDSIZE=80)

Invokes RMS and causes an RMS indexed disk file named
IDX.DAT to be
created on unit 1. A primary index of datatype CHARACTER, specifying
character positions 1 through 15 of each record has been defined
for
the file.
CLOSE (UNIT=3,DISPOSE='DELETE')
The above statement causes the file
closed and deleted.

FORTRAN-10/20 VII

11-51

associated

with

unit

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
11.7

INQUIRE STATEMENT

The INQUIRE statement inquires about specific properties of a
file
name or of a logical unit number on which a file might be opened.
The
INQUIRE statement has two forms:
one inquires by file, and the other
inquires by unit.

11.7.1

INQUIRE by File

An INQUIRE by file is an INQUIRE statement containing the following:
•

A FILE= keyword

•

An associated file specification

•

No UNIT= keyword

It is used to obtain information about a file based on the file name.
INQUIRE by file can be used to get information on the following files:
•

Files that are "connected"; meaning files for which an OPEN
statement has been executed or
for which a data transfer
statement has been executed.

•

Files that are not "connected."

INQUIRE by file has the following form:
INQUIRE (FILE=fi[DEFAULTFILE=dfi] ,flist)
where:
fi

is a character expression whose value
name of the file to be inquired about.

flist

is a list that can contain at most one of each
inquiry specifiers (see Section 11.7.3).

dfi

is a character expression specifying
name specification string.

specifies

the

default

file

DEFAULTFILE=dfi can be used in addition to or in place of FILE=fi when
used
in connection with an inquiry about a file.
If both FILE and
DEFAULTFILE are used, then the DEFAULTFILE specification
is merged
with
FILE
to
obtain the
full
fIle
specification, with
file
specification fields
in
FILE superceding corresponding
fields
in
DEFAULTFILE.
NOTE
If the file specification is remote, RMS
access the file.

used

INQUIRE by file may be used
any time during
the execution of a
program.
It can be used before a file is opened to find out about the
existence of the file, or after the file is opened to find
out other
attributes of the file.
It can also be used to find the unit number
on which the file is opened.
If the same file is opened on more
than
one unit, the smallest number on which the file is opened is returned.

FORTRAN-10/20 VII

11-52

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
The determination of whether a file specified in an INQUIRE
is opened on a unit is the following:

statement

The file specification given in the INQUIRE statement is used
to lookup the file.

If the file exists, the file specification, expanded with the
physical device name and generation
(TOPS-20 only),
is
compared with the file specification for each open unit,
in
ascending order, until there is an exact string match.

If the file does not exist, the specification given
in the
INQUIRE statement (with a default of DSK: added if necessary
for the device name), is compared with the file specification
for each open unit,
in ascending order, until there is an
exact string match.
Note
that this match will only be
successful for 'deferred' OPEN files, since non-deferred OPEN
files are always established
in the specified directory
immediately.
Therefore, the file exists (see item 2 above) •
NOTE
If a file exists, INQUIRE by file will not generally
match the file with a unit for which a 'deferred' OPEN
has been done, since the file specification for
the
unit has not been expanded.
For example, the file's
logical device name has not been replaced by a
physical device name.

(See Section 18.8 for information on FOROTS and INQUIRE by file.)

11.7.2

INQUIRE by Unit

INQUIRE by unit is an INQUIRE statement containing a UNIT= keyword and
no FILE=
keyword.
It is used to find out information about the file
that may be "connected" to the specified unit.
INQUIRE by unit has the following form:
INQUIRE ([UNIT=]u,ulist)
where:
u

is the number of the logical unit to be inquired about.
The unit need not exist, nor need it be connected to a
file.
If the unit is connected to a
file,
the
inquiry
encompasses both the connection and the file.

ulist is a list that can contain at most one of
inquiry specifiers (see Section 11.7.3).

each

the

If the optional UNIT= keyword if omitted, u must be the first item
the list.

INQUIRE by unit can be used at any time during
the execution of a
program.
It can be used before a file is opened to find out if there
is another file open on the unit, or after the file is opened to find
out other attributes of the file.

11-53

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
11.7.3

INQUIRE Specifiers

The specifiers described in the following
either farm of the INQUIRE statement.

11.7.3.1
form:

sections

ACCESS Specifier - The ACCESS specifier

ACCESS

may

has

the

used

following

acc

where:
acc

11.7.3.2
form:

is a character variable, array element,
or substring
reference.
It
is assigned the value 'SEQUENTIAL' if the
file is connected for sequential access, 'DIRECT'
if the
file
is connected
for direct access, or 'KEYED' if the
file is connected
for
keyed access.
If there
is no
connection, acc is 'UNKNOWN'.

BLANK Specifier - The

BLANK

specifier

has

the

following

blk

where:
blk

is a character variable, array element,
or substring
reference.
It is assigned the value 'NULL' if the file
was last opened with BLANK='NULL', and
is assigned
the
value' ZERO' if the file was opened with BLANK='ZERO'.
If
the file is not open, blk is 'UNKNOWN'.

11.7.3.3 BYTESIZE
follow i ng form:
BYTESIZE

Specifier - The

BYTESIZE

specifier

has

the

is the integer bytesize of the file (or 0 if
is not open).

the

file

byte

where:
byte

11.7.3.4 CARRIAGECONTROL
has the following form:

Specifier - The

CARRIAGECONTROL

specifier

CARRIAGECONTROL = cc
where:
cc

is a character variable, array element, or substring
reference.
It is assigned the following values:
1.

'FORTRAN' if the file has the FORTRAN carriage-control
attribute

FORTRAN-10/20 VII

11-54

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS

11.7.3.5
form:

'LIST' if the file has
attribute

the

implied

'NONE'

'TRANSLATED' if the file has FORTRAN carriage-control
characters being translated directly into vertical
motion characters.

'UNKNOWN'
if the CARRIAGECONTROL value
determined, or the file is not open.

if the file has no carriage-control attribute

DIRECT Specifier - The DIRECT specifier

DIRECT

carriage-control

has

the

cannot

following

dir

where:
dir

11.7.3.6

is a character variable, array element, or substring
reference.
It is assigned the following values:
1.

'YES'
file

if DIRECT is an allowed access

method

for

the

'NO' if DIRECT is not an allowed access method for the
file

'UNKNOWN' if the processor
is unable to
whether DIRECT is an allowed access method

determine

ERR Specifier - The ERR specifier has the following form:

ERR = s
where:
s

11.7.3.7
form:
EXIST

is the label of an executable statement.
ERR is a control
specifier;
if an error occurs during execution of the
INQUIRE statement, control is transferred to the statement
whose label is s.

EXIST Specifier - The

EXIST

specifier

has

the

following

where:
ex

is a logical variable or logical array element.
It is
assigned the value
.TRUE. if the specified file or unit
exists, and the value .FALSE.
if the specified
file or
unit does not exist.

11-55

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
11.7.3.8

FORM Specifier - The FORM specifier has the following form:

FORM = fm
where:
fm

is a character variable, array element, or substring
reference.
It
is assigned the value 'FORMATTED' if the
file is connected for formatted I/O, and 'UNFORMATTED'
if
the file is connected for unformatted I/O.
If there is no
connection, fm is 'UNKNOWN'.

11.7.3.9 FORMATTED
follow i ng form:
FORMATTED

Specifier - The

FORMATTED

specifier

has

the

fmd

where:
fmd

11.7.3.10
form:
IOSTAT

is a character variable, array element, or substring
reference.
It is assigned the value 'YES' if formatted is
an allowed form for the file.
It is assigned
the value
'NO'
if formatted is not an allowed form of the file, and
the value 'UNKNOWN' if the form cannot be determined.

IOSTAT

Speci~ier

- The IOSTAT specifier has

the

following

ios

where:
ios

11.7.3.11
form:
KEYED

is an integer variable or integer array element.
It
is
assigned a
processor-dependent positive integer value if
an error occurs during execution of the INQUIRE statement,
or assigned the value zero if there is no error condition.

KEYED Specifier - The KEYED

specifier

has

the

following

kyd

where:
kyd

is assigned the value 'YES'
if KEYED is an allowed
access method
for
the file
(that is,
the file is
indexed), 'NO'
if KEYED is not an allowed access
method, and 'UNKNOWN' if the file is not open.

FORTRAN-10/20 VII

11-56

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
11.7.3.12

NAME Specifier - The NAME specifier has the following form:

NAME = nme
where:
nme

is a character
reference.
It
inquired about.

variable, array element, or substring
is assigned the name of the file being

The value assigned to nme is not necessarily identical
to
the value specified with FILE=.
For example, the value
that the processor returns may contain a directory name or
generation number (TOPS-20 only).
However, the value that
is assigned is always valid for use with FILE= in an OPEN
statement.
NOTE
FILE and NAME are synonyms when used with the OPEN
statement, but not when used with the INQUIRE
statement.
For INQUIRE by unit, FOROTS returns the
full,
expanded
file specification if any of the following is true:
•

If there is a file open on the specified
STATUS other than UNKNOWN or SCRATCH.

unit

with

•

If there is a file open on the
ACCESS other than SEQUENTIAL.

unit

with

•

An I/O transfer statement has been executed using
specified unit and the unit has not been closed.

the

specified

FOROTS returns the string given in the OPEN for NAME= with
defaults applied if both of the following are true:
•

If the file is open on the specified unit as a
result
of an OPEN statement
in which STATUS='UNKNOWN' and
ACCESS='SEQUENTIAL' are specified or implied.

•

No I/O transfer statement has been executed using
specified unit.

the

If STATUS='SCRATCH', FOROTS returns blanks for NAME=.
If there has been no OPEN statement, and no I/O transfer
statement has been executed
using the specified unit,
FOROTS returns for NAME= the default file specification
for that unit.
For INQUIRE by file, FOROTS returns the full,
expanded
file specification if the file exists in the specified
directory.
If the file does not yet exist on the
specified directory, but has been opened by a 'deferred'
OPEN, FOROTS returns the string given
in the
INQUIRE
statement, with defaults applied
for
the device and
generation number (TOPS-20 only).
Otherwise, blanks are
returned.
NOTE
INQUIRE on a remote file that returns a full
file
specification, returns the network attributes.
FORTRAN-10/20 VII

11-57

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
11.7.3.13
form:

NAMED Spec~fier - The NAMED

specifier

has

the

following

NAMED = nmd
where:
nmd

is a logical variable or logical array element.
It is
assigned the value .TRUE. if the specified file has a
name, and the value .FALSE. if the file does not have a
name.

11.7.3.14
form:

NEXTREC Specifier - The NEXTREC specifier has the following

NEXTREC

where:
nr

11.7.3.15
form:

is an integer variable or integer array element.
It is
assigned an integer value that is one more than the last
record number read or written on the specified direct
access file. If no records have been read or written, the
value of nr is one. If the file is not connected for
direct access, or if the position is indeterminate because
of an error condition, nr is zero.

NUMBER Specifier - The NUMBER specifier has

the

following

NUMBER = num
where:
num

11.7.3.16
form:

is an integer variable or integer array element.
It is
assigned the number of a logical unit currently connected
to the specified file.
If there is no logical unit
connected to the file, num is not defined. If more than
one unit is connected to the file, the smallest unit
number is returned.

OPENED Specifier - The OPENED specifier has

the

following

OPENED = od
where:
od

is a logical variable or logical array element.
It is
assigned the value .TRUE. if the specified file is opened
on a unit or if the specified unit is opened; it is
assigned the value .FALSE. if the file or unit is not
open.

11-58

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
11.7.3.17 ORGANIZATION Specifier - The ORGANIZATION specifier has the
following form:
ORGANIZATION

org

where:
org

is a character variable, array element, or substring
reference.
It
is assigned the value 'SEQUENTIAL' if
the file is a sequential file, 'RELATIVE' if the
file
is a relative file, 'INDEXED' if the file is an indexed
file, or 'UNKNOWN' if the file is not open or the file
is not an RMS file.

11.7.3.18 RECL
(RECORDSIZE)
Specifier specifier has the following form:

The

RECL

(RECORDSIZE)

RECL = rcl
where:
rcl

is an integer variable or integer array element.
For non-RMS files, if the file (or unit) is opened, rcl is
the
record
length for fixed-length record files.
In all
other cases, rcl is zero.
If the file
is opened
for
formatted
I/O,
rcl
is expressed
in characters, and in
words if the file is unformatted.
For RMS files, if the file
is open,
RECL returns the
maximum record size value associated with the file in
either bytes (for formatted I/O) or words (for unformatted
I/O).
If the file is not open, RECL returns zero.

11.7.3.19 RECORDTYPE Specifier - The
following form:
RECORDTYPE

RECORDTYPE

specifier

has

the

rtype

where:
rtype is a character variable, array element, or substring
reference.
It
is assigned the value 'FIXED' if the file
has fixed-length records, assigned the value 'VARIABLE' if
the file
has variable-length records, and assigned the
value 'STREAM' if the file is a stream file
(default
for
disk and magnetic tape).
If the file is not open, rtype
is assigned the value 'UNKNOWN'.

FORTRAN-10/20 VII

11-59

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
11.7.3.20 SEQUENTIAL Specifier - The
following form:
SEQUENTIAL

SEQUENTIAL

specifier

has

the

seq

where:
seq

is a character variable, array element,
or substring
reference.
It is assigned the following values:
1.

'YES' if SEQUENTIAL is an allowed
the specified file

access

method

'NO'

'UNKNOWN' if the processor cannot determine
SEQUENTIAL is an allowed access method

if SEQUENTIAL is not an allowed access method

11.7.3.21 UNFORMATTED Specifier - The UNFORMATTED specifier
following form:
UNFORMATTED

for

whether

has

the

unf

where:
unf

is a character variable, array element,
or substring
reference.
It is assigned the value 'YES' if unformatted
is an allowed form for the file.
It is assigned the value
'NO'
if unformatted
is not an allowed form of the file,
and the value 'UNKNOWN' if the form cannot be determined.

11-60

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
Table 11-6 summarizes the form and use of the FORTRAN
statements.

device

control

Table 11-6: Summary of FORTRAN Device Control Statements
Statement Form

Section

FIND (UNIT = un,REC = rnf,ERR = s][,IOSTAT = ios])
FIND (un'rn[,ERR=sl[,IOSTAT=iosD

11.8.1

REWIND un
REWIND (UNIT = un[,ERR = s][,IOSTAT = ios])
REWIND (un[,ERR=s][,IOSTAT=ios])

11.8.2

UNLOAD un
UNLOAD (UNIT = un[,ERR = s][,IOSTAT = ios])
UNLOAD (un[,ERR = s If,IOSTAT = iosD

11.8.3

BACKSPACE un
BACKSPACE (UNIT = un[,ERR = s][,IOSTAT = ios])
BACKSPACE (un[,ERR = s][,IOSTAT = ios])

11.8.4

ENDFILE un
END FILE (UNIT = un[,ERR = s][,IOSTAT = ios])
ENDFILE (un[,ERR=s][,IOSTAT=ios])

11.8.5

SKIPRECORD un
SKIPRECORD (UNIT = un[ ,END = s][,ERR = s][,IOSTAT = ios 1)
SKIPRECORD (un[,END= s][,ERR=s][,IOSTAT=ios])

11.8.6

SKIPFILE un
SKIPFILE (UNIT = un[ ,ERR = s][,IOSTAT = ios])
SKIPFILE (un[,ERR=s][,IOSTAT=iosD

11.8.7

BACKFILE un
BACKFILE (UNIT = un[,ERR = s][,IOSTAT = ios 1)
BACKFILE (unLERR= sl[,IOSTAT= ios])

11.8.8

11.8

DEVICE CONTROL STATEMENTS

Device control statements enable you to position external devices.
For example, when performing data transfers with magnetic tape, you
use device control statements to position the tape.
The device
control statements may be used for both formatted and unformatted
files.
The following list contains all of the device control statements,
the section in which each statement is described.
1.

FIND (Section 11.8.1)

REWIND (Section 11.8.2)

UNLOAD (Section 11.8.3)

BACKSPACE (Section 11.8.4)

ENDFILE (Section 11.8.5)

SKIPRECORD (Section 11.8.6)

SKIPFILE (Section 11.8.7)

BACKFILE (Section 11.8.8)

11-61

and

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
NOTE
The results of the BACKSPACE and SKIPRECORD statements
are
unpredictable when used on list-directed and
NAMELIST formatted data.
The general form of every device control statement is:
keyword un
or
keyword

(UNIT=un[ ,END=s] [,ERR=s] [,IOSTAT=ios])

or
keyword

(un[ ,END=s] [,ERR=S] [,IOSTAT=ios])

where:
keyword

is the statement name.

is the FORTRAN logical unit number.
Table 10-3
lists the default logical unit numbers.
If you
enclose the
logical
unit
specification
in
parentheses,
you can include the keyword portion
(UNIT=) of the logical
unit specification.
The
keyword
form of the
unit specifier makes the
specification posi.tional1y independent
in
the
parenthetical list.

END=s

is the optional end-of-file specifier (see Section
10.4.6) •

ERR=s

is the optional
10.4.7) •

IOSTAT=ios

is the optional I/O error
Section 10.4.8).

error

specifier

(see

Section

status

specifier

(see

The operations performed by the device control statements are
usually
used only for magnetic tape devices (MTA).
In FORTRAN, however, the
device control operations are simulated for disk devices.

FIND
Statement

11.8.1

FIND Statement

In earlier versions of FORTRAN-10/20, the FIND statement could be used
during
direct-access
READ/WRITE operations to
reduce
the
time
necessary to do data
transfers and
to
locate
records
in
a
direct-access
file.
For FORTRAN-10/20 Version 11, the FIND statement
has no effect on non-RMS files, except for setting the current record
number and, therefore, the associate variable.

FORTRAN-10/20 VII

11-62

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
For an RMS relative file, the FIND statement positions the file at the
record
number
specified.
This establishes the record as the current
record, sets th€ associate variable, and returns.
The
file on the
specified unit must be a relative file opened for direct access.
The first form of the FIND statement is:
FIND (UNIT=un,REC=rn[ ,ERR=s] [,IOSTAT=ios])
The second form of the FIND statement operates in the same way as the
first
form.
The difference between the two forms is in the way that
the unit number and record number are specified.
The second form of this statement is:
FIND(un' rn[ ,ERR=s] [, IOSTAT=ios])
In this form, the unit number and record number do not contain
the
keywords UNIT= and REC=.
Instead, the unit number is specified first,
followed by a single quote ('), and finally the record number.
The following example demonstrates how the FIND statement is used:
FIND(UNIT=1,REC=100)
This statement positions the file so
reads record 100.

that

the

READ

statement

REWIND
Statement

11.8.2

REWIND Statement

The REWIND statement, used for disk files, causes a specified file
to
be positioned at its initial point.
For magnetic tape, execution of
the REWIND statement positions the magnetic tape at its initial point.
The REWIND
sequential
files.

statement can be
used with
access RMS relative files.

RMS sequential
files or
It
is illegal for indexed

The forms of the REWIND statement are:
REWIND un
or
REWIND (UNIT=un[,ERR=s] [,IOSTAT=ios])
or
REWIND (un [ , ERR=S] [ ,IOSTAT= ios] )

FORTRAN-10/20 VII

11-63

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
where:
un

is the logical unit number of the device on which the REWIND
is being
performed.
Table 10-3 lists the default logical
unit numbers.

REWIND is illegal for files opened with ACCESS='APPEND'.

Statement

11.8.3

UNLOAD Statement

UNLOAD]

' - - - - -

The UNLOAD statement rewinds and unloads the magnetic tape
with the specified unit.

associated

The forms of the UNLOAD statement are:
UNLOAD un
or
UNLOAD (UNIT=un[,ERR=s] [,IOSTAT=ios])
or
UNLOAD (un[ ,ERR=s] [,IOSTAT=ios])
where:
un

is the logical unit number of the device on which the UNLOAD
is being performed.
Table 10-3 lists the default logical
unit numbers.

BACKSPACE
Statement

11.8.4

BACKSPACE Statement

Execution of a BACKSPACE statement causes the file connected
to
the
specified unit to be positioned before the preceding record.
If there
is no preceding record, the position of the file is not changed.
If
the preceding
record
is an ENDFILE record (see Section 11.8.5), the
file is positioned before the ENDFILE record.
For RMS files, this statement repositions a
file
with sequential
organization to
the beginning of the previous record. When the next
I/O statement is executed, this
preceding
record
is available for
processing.

FORTRAN-10/20 VII

11-64

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
The BACKSPACE statement cannot be used
for direct-access
files,
append-access files, or files that are formatted with list-directed or
NAMELIST-statement formatting.
For RMS files, BACKSPACE cannot be used for files open ACCESS='APPEND'
or with ORGANIZATION='RELATIVE' or 'INDEXED'.
The forms of the BACKSPACE statement are:
BACKSPACE un
or
BACKSPACE (UNIT=un[,ERR=s] [,IOSTAT=ios])
or
BACKSPACE (un[,ERR=s] [,IOSTAT=ios])
where:
un

is the logical unit number of the device on which the
Table 10-3 lists the default
BACKSPACE is being performed.
logical unit numbers.

ENDFILE
Statement

11.8.5

ENDFILE Statement

The ENDFILE statement closes the file on the specified
unit.
On
magnetic
tape,
an 'ENDFILE record' is written and is then positioned
after the end-of-file mark.
For RMS files, ENDFILE is supported only
organization.

for

files

with

sequential

For disk, the file is closed and then positioned at the end of the
file, and an end-of-file status is set.
This status is the equivalent
of the file being positioned after an 'ENDFILE record'.

FORTRAN-10/20 VII

11-65

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
Thus, for both disk and magnetic tape, a
BACKSPACE operation given
after
an end
file operation positions the file after the last data
record (that is, before the physical (or virtual) 'ENDFILE record').
The ENDFILE statement can be used only with sequential access files.
The forms of the ENDFILE statement are:
ENDFILE un
or
ENDFILE (UNIT=un[ ,ERR=s] [,IOSTAT=ios])
or
ENDFILE (un[ ,ERR=s]J ,IOSTAT=ios])
where:
un

is the logical unit number of the device on which the
ENDFILE is being
performed.
Table 10-3 lists the default
logical unit numbers.

SKIPRECORD
Statement

11.8.6

SKIPRECORD Statement

The SKIPRECORD statement skips the record
immediately following
the
current
(last accessed)
record.
If the SKIPRECORD statement is
executed prior to accessing any records, then the first record in the
file
is skipped.
You cannot use SKIPRECORD on direct-access files.
You can use SKIPRECORD on RMS indexed files; it does a sequential read
on the current index as established by a previous indexed READ, or the
primary index if no current index has been set.
The forms of the SKIPRECORD statement are:
SKIPRECORD un
or
SKIPRECORD(UNIT=un[ ,END=s] [,ERR=s] [,IOSTAT=ios])
or
SKIPRECORD (un [ ,END=s] [ ,ERR=S] [ ,IOSTAT=ios] )
where:
un

is the logical unit number of the device on which the
SKIPRECORD is being performed.
Table 10-3 lists the default
logical unit numbers.

FORTRAN-10/20 VII

11-66

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS

SKIPFILE
Statement

11.8.7

SKIPFILE Statement

This statement is used only for magnetic tape operations.
Unless an
end-of-file has been encountered, the SKIPFILE statement advances to
the beginning of the next
file.
If an end-of-file has
been
encountered,
SKIPFILE skips the next file.
If the number of SKIPFILE
executions exceeds the number of files available to be skipped,
an
error occurs.
The forms of the SKIPFILE statement are:
SKIPFILE un
or
SKIPFILE (UNIT=un[ ,ERR=s] [,IOSTAT=ios])
or
SKIPFILE (un[ ,ERR=s] [,IOSTAT=ios])
where:
un

is the logical unit number of the device on which the
SKIPFILE is being performed.
Table 10-3 lists the default
logical unit numbers.

BACKFILE
Statement

11.8.8

BACKFILE Statement

This statement is used only for magnetic
tape operations.
If an
end-of-file has been encountered, the BACKFILE statement positions to
the start of the file whose end-of-file was detected.
Otherwise,
the
BACKFILE statement positions to the start of the file that precedes
the current (last accessed) file.
The forms of the BACKFILE statement are:
BACKFILE un
or
BACKFI LE (UNIT=un [ ,ERR=S] [ ,IOSTAT= ios] )
or
BACKFILE (un [ ,ERR=s] [ ,IOSTAT=s] )
FORTRAN-10/20 VII

11-67

February 1987

FILE-CONTROL AND DEVICE-CONTROL STATEMENTS
where:
un

is the logical unit number of the device on which the
BACKFILE is being performed. Table 10-3 lists the default
logical unit numbers.
NOTE
On a magnetic tape with multiple files,
the position
of the tape after an ENDFILE record of one file is
equivalent to the position at the beginning of the
next file.

11-68

February 1987

CHAPTER 12
FORMATTED DATA TRANSFERS

Data transfers can be either
formatted or
unformatted.
When the
internal
(memory)
representation of the data
is translated to a
different external (peripheral storage) representation during a data
transfer, that data transfer is considered formatted.
Conversely, when the internal and external representations of the data
are the same, that data transfer is considered unformatted.
A formatted data transfer
involves editing of data as
it
is
transferred
to and from memory.
FORTRAN provides you with three ways
for specifying how the data
is formatted during a
formatted data
transfer.
These are:
1.

FORMAT-Statement Formatting

List-Directed Formatting

NAMELIST-Statement Formatting

Of the three types, FORMAT-statement formatting provides you with the
most control over how the data is formatted.
Section 12.1 describes
FORMAT-statement formatting.
List-directed formatting means that the formatting
is controlled by
the data
types of the
I/O list elements.
Section 12.5 describes
list-directed formatting.
NAMELIST-statement formatting is the third method for
formatting
the
data;
the
formatting is controlled by the data types of the namelist
elements.
In this form,
the
I/O list
is defined
in a NAMELIST
statement and referenced by the data transfer statement.
Section 12.6
describes NAMELIST-statement formatting.
NOTE
FOROTS/RMS supports transparent data transfers to and
from VMS systems for all
formatted data, and for
unformatted character data.
Numeric data
transfers
are not supported.
DIL
(Data interchange Library)
routines or DIU (Data Interchange Utility)
should be
used
to correctly exchange numeric data between
TOPS-20 and VMS systems.

FORTRAN-10/20 VII

12-1

February 1987

FORMATTED DATA TRANSFERS

FORMAT-Statement
Formatting

12.1

FORMAT-STATEMENT FORMATTING

A FORMAT statement directs the editing of data during its transfer
between internal and external storage.
Every formatted
(FORMAT
statement) data transfer statement contains a reference to one of the
following:
1.

A line containing a FORMAT statement with a format list

A numeric array containing a format list

A character expression containing a format list

An integer, real, or logical variable that has been
a FORMAT statement number with an ASSIGN statement

assigned

The format list is made up of format specifiers.
During execution of a formatted data transfer statement, items in the
I/O list are associated with specifiers in the referenced format list.
The specifiers dictate how the various data items are formatted.
Section 12.1.1 describes how to create a format list in a FORMAT
statement;
Section
12.1.2
describes
how to create a format
specification as a character expression. Section 12.1.3 describes how
to create a numeric array that contains a format list. Section 12.1.4
describes how to specify a FORMAT statement using an ASSIGNed
variable.

12.1.1

Specifying a Format List in a FORMAT Statement

The general form of a FORMAT statement is:
n

FORMAT fs

where:
n

is the required statement number.
This number,
referenced in the control-information list of an
I/O statement (see Section 10.4.2) provides the
association between the data transfer statement
and the FORMAT statement.

is a format specification.
specification is:
([format list])

12-2

The form of

format

FORMATTED DATA TRANSFERS
where:
format list

is a list of items
following forms:

which

may

take

any

the

[r] ed
or
ned
or
[r]

(fl)

where:
r

is a nonzero, unsigned, integer constant called
repeat specification.

is a repeatable
12.2.1) •

edit

descriptor

(see

Section

ned

is a non repeatable edit
12.2.2) •

descriptor

(see

Section

is a nonempty format list.

The only placement restrictions for FORMAT statements are that they
follow PROGRAM, FUNCTION, SUBPROGRAM, or BLOCK DATA statements, and
that they precede the END statement.
The following example illustrates FORMAT-statement formatting.
The
FMT specifier in the WRITE statement references the label of FORMAT
statement 101.
This FORMAT statement contains a list of edit
descriptors
(X,
I, and F) that dictate the formatting of the data in
I/O list (variables J, Y, and Z).

101

12.1.2

J=2
Y=3.0
Z=5.1
WRITE{UNIT=5,FMT=101)J,Y,Z
FORMAT(1X,I,F,F)

Specifying a Format Specification as a Character Expression

You can store format specifications in character variables, character
arrays, character array elements, character substrings, or character
expressions.
A character format specification must be of the form described in
Section 12.1.1. Note that the form begins with a left parenthesis and
ends with a right parenthesis. Character data may follow the right
parenthesis that ends the format specification, with no effect on the
format specification.
Blank characters may precede the
format
specification.
If the format identifier is a character array name,
the format
specification may be contained in more than the first element of the
array.
{A character array format specification is considered to be a
concatenation of all the array elements of the array in the order
given by array element ordering (see Section 4.3.3).)
12-3

FORMATTED DATA TRANSFERS
However, if a character array element name is specified as a
format
identifier, the length of the format specification must not exceed the
length of the array element.
The following example shows the same format specification used in the
examples
in
Section
12.1.1.
This time,
however,
instead of
referencing
the format
specification by
statement
number,
or
referencing
the name of a numeric array, the data transfer statement
references the name of the character variable
in which the
format
specification is contained.
J=2
Y=3.0
Z=5.1
CHARACTER FORNAM*lO
FORNAM = I (1 X, I , F , F) I
WRITE(UNIT=5,FMT=FORNAM)J,Y,Z

12.1.3

Specifying a Format Specification in a Numeric Array

An alternative to using FORMAT statements
specification in a numeric array.

store

the

format

The
format specifications are associated with a data
transfer
statement
by referencing
the array name containing
the
format
specification, instead of a statement label of a FORMAT statement.
The following example shows the same format specification used in the
example in Section 12.1.1.
This time, however, instead of referencing
the format
specification by statement number,
the data
transfer
statement references the name of the numeric array in which the format
specification is contained.
INTEGER FORNAM(2)
FORNAM(l)=' (lX,I'
FORNAM(2)=' ,F,F)'
,J'=2
Y==3.0

2=5.1
WRITE(UNIT=5,FMT=FORNAM)J,Y,Z
END
In the above example the format specification is stored in both words
of array FORNAM.
This is because the format contains ten characters:
the first five are in FORNAM(l); and the last five are in FORNAM(2).
NOTE
When storing a
format specification
in an array,
always
include
the outer most parentheses enclosing
the format specifiers.
Note
that
the word
FORMAT
should not be included in the string.

12.1.4

Specifying a FORMAT Statement Using an ASSIGNed Variable

Integer, real, or logical variables that have been ASSIGNed
FORMAT
statement numbers can be used as format specifiers.
(See Section 8.3
for information on the ASSIGN (statement label) assignment statement.)
12-4

FORMATTED DATA TRANSFERS
The variable is assigned a statement number by an ASSIGN statement.
The format specifier references the variable that refers to the
statement number it has been assigned.
The following example shows the same format specification used in the
examples in Sections 12.1.1, 12.1.2, and 12.1.3. This time, however,
instead of referencing the format specification by statement number,
the data transfer statement references a variable that has been
assigned a statement number by an ASSIGN statement.

101

12.1.5

ASSIGN 101 TO IFORMT
WRITE(UNIT=5,FMT=IFORMT}J,Y,Z
FORMAT(lX,I,F,F}

The Ordering and Interpretation of Format List Items

For standard conforming programs, all items within the
format
list
should be separated by commas, with the exception of the following
cases:
1.

Between a P edit descriptor and an immediately following
E, D, or G edit descriptor (See Section l2.4.ll)

Before or after a slash edit descriptor (See Section 12.4.5)

Before or after a colon edit descriptor (See Section l2.4.6)

In FORTRAN-10/20, the use of commas to delimit format edit descriptors
within a format list is optional as long as no ambiguity exists.
For
example,
FORMAT (3X,A2)
can be written as
FORMAT (3XA2)
But the specification
FORMAT (I22I5)
is ambiguous, since it can represent
FORMAT (I22,I5)

FORMAT (I,22I5)

and requires the comma to eliminate ambiguity.

12-5

FORMATTED DATA TRANSFERS

FORMAT-Statement
Edit Descriptors

12.2

EDIT DESCRIPTORS

Edit descriptors within the format list describe the manner of editing
performed on the data being transferred.
For example, when you transfer integers from a file to memory, you use
an I
edit descriptor. When the data transfer statement is executed,
an item in the I/O list is associated with the I
edit descriptor
in
the format list, and the following results:
1.

Before being stored in memory, the data is converted to an
internal
integer
format by the I
edit descriptor in the
format list.

The memory location in which the data is stored is identified
by the I/O list element.

The following sample program demonstrates how an integer is read
from
the terminal (external device) into the memory location identified in
the I/O list of the ACCEPT statement.
PROGRAM FORMAT

101
102

TYPE *,'Please enter a two digit numbera'
ACCEPT 101,K
FORMAT(I2)
TYPE 102,K
FORMAT(lX,I5)
END

The sample output below shows what happens when the user executes the
above program.
The user enters 78
in
response to the ACCEPT
statement.
This causes the integer value to be stored in the variable
K according
to the I edit descriptor in FORMAT statement 101. Then
the type statement causes the value of variable K to be printed at the
terminal according to the I edit descriptor in FORMAT statement 102.
EXECUTE TEST
LINK: Loading
(LNKXCT FORMAT execution]
Please enter a two digit number:
78
78
CPU Time 0.1
Elapsed Time 8.7
The I edit descriptor is an example of a repeatable edit descriptor.
FORTRAN has two
types of edit descriptors:
repeatable (Section
12.2.1) and nonrepeatable (Section 12.2.2). The third type of
item
that appears
in a
format list
is
the carriage-control specifier
(Se c t ion 12.2.3).

12-6

FORMATTED DATA TRANSFERS

Repeatable Edit
Descriptors

12.2.1

Repeatable Edit Descriptors

A repeatable edit descriptor may be preceded by an optional, unsigned,
nonzero, integer constant that specifies a repeat count.
This integer
is called a repeat specification.
Using a repeat specification in an edit descriptor gives you a
shorthand way to specify multiple fields with a single specification.
For example, without using the repeat specification, if you wanted to
specify four
fields,
each of which contain an integer value that is
six characters long, you might construct the following
FORMAT
statement:
101

FORMAT (lX,I6,I6,I6,I6)

If you use the repeat specification, however, you need only specify
the edit descriptor and field width a single time, as follows:
101

FORMAT (lX,4I6)

These two FORMAT statements are equivalent.
Table 12-1 lists the repeatable edit descriptors.
Each descriptor
listed in the table
is shown in its complete form.
The key at the
bottom of Table 12-1 describes all the optional elements in each edit
descriptor.
The right-most column of Table 12-1 references the
section in which each edit descriptor is discussed.

12-7

FORMATTED DATA TRANSFERS
Table 12-1: Repeatable FORTRAN Edit Descriptors
Edit Descriptor

Descriptor Type

Refer to:

[r]lfw[.m]l

Integer

Section 12.4.11.1

[r]Flw.dl

Floating Point

Section 12.4.11.2

[r]E[ w.d[Ee]l

Scientific Notation

Section 12.4.11.3

[r]D! w .d[Ee1]

Scientific Notation

Section 12.4.11.3

[r ]G[ w.d[Ee]]

General Conversion

Section 12.4.11.4

F,E,D,G
(Two successive)

Complex

Section 12.4.11.5

Ir JO[ w[ .m]]

Octal

Section 12.4.11.6

!rlZ[w[.mll

Hexadecimal

Section 12.4.11.7

[rlUwl

Logical

Section 12.4.12

[r]A[wl

Character or Hollerith

Section 12.4.13

lrlR[w]

Hollerith

Section 12.4.14

Key:
r

is a nonzero, unsigned, integer constant called a repeat specification.

is a nonzero, unsigned, integer constant which is equal to the total number of characters in the numeric field
being described. The numeric edit descriptors are described in Section 12.4.11.

.m is an unsigned, integer constant which specifies the minimum number of digits to be output to the field
being described. If necessary, leading zeros are output. The value of m must not exceed the value of w.
If m is zero and the value of the internal data item is zero, the output field consists of only blank characters,
regardless of the sign control in effect.

is a nonzero, unsigned, integer constant which specifies the total number of digits to the right of the decimal
point in the numeric field being described. If.d is specified, w must also be specified. The maximum value is
63 digits.

is a nonzero, unsigned, integer constant which is equal to the total number of digits in the exponent field of
the numeric field being described. The maximum value is 15 digits.

Nonrepeatable Edit
Descriptors

12.2.2

Nonrepeatable Edit Descriptors

A nonrepeatable edit descriptor can not be preceded by a
repeat
specification.
The nonrepeatable edit descriptors provide a variety
of editing possibilities,
such as positioning within a
record,
including character constants
in a FORMAT statement, and delimiting
records within a single format descriptor.
Table 12-2 lists the nonrepeatable edit descriptors.
The
format,
function,
and section number where each descriptor is discussed are
listed in the table.

12-8

FORMATTED DATA TRANSFERS
Table 12-2:

Nonrepeatable FORTRAN Edit Descriptors
Function

Edit Descriptor

Refer to:

'hl. .. hn'

Character Data

Section 12.4.1

nHh

Hollerith Data

Section 12.4.2

Tc
TLc
TRc

In-Record Positioning

Section 12.4.3.1

InlX

In-Record Positioning

Section 12.4.3.2

(Dollar sign) Prevents record from terminating with END
OF LINE

Section 12.4.4

(Slash) Record Delimiter

Section 12.4.5

(Colon) Format-Control Termination

Section 12.4.6

S
SP
SS

Plus sign control for output of positive numeric fields

Section 12.4.7

Scaling Factor for Numeric Fields

Section 12.4.8

BN
BZ

Specifies the handling of blanks during the input of
Numeric Fields

Section 12.4.9

Input Only Descriptor - returns the number of characters left in the current record.

Section 12.4.10

Key:
n

is a nonzero, unsigned, integer constant which is equal to a number of spaces (X descriptor) or the total
number of characters (H descriptor).

is a character capable ofrepresentation by the processor. This type of character is described in Appendix B.

is a nonzero, unsigned, integer constant which is equal to a number of characters within a record relative to
the current position.

is an optionally signed integer constant which declares the scaling factor for the field being described.

Carriage-Control
Specifiers

12.2.3

Carriage-Control Specifiers

In a data output transfer, the first character of each reco~d can be
used
for carriage control.
A carriage-control specifier dictates the
action of the printing mechanism on output devices.
For example,
carriage-control
specifiers
determine
the vertical
spacing
for
line-printer output.

12-9

FORMATTED DATA TRANSFERS
NOTE
The CARRIAGECONTROL specifier of
the
OPEN statement
enables you to decide how the first character of each
record is treated.
Depending on
the value of
the
CARRIAGECONTROL specifier, the first character can be:
1.

Replaced with
character (s) •

the

appropriate

printer-control

Disregarded as a carriage-control
character and,
instead, be transferred as part of the record.

For more information on the CARRIAGECONTROL specifier,
see Section 11.3.6

The carriage-control specifier may be written as a character constant.
The
following example shows the blank carriage-control character in a
FORMAT statement:
WRITE(5,101)
101

FORMAT('

','This is a string')

END
When this example is executed,
the string
printed on unit 5, the terminal, as follows:

the

format

list

This is a string
If you omit the carriage-control specifier from a data output transfer
format list, FOROTS interprets the first character to be output to the
record as the carriage-control character.
Using the example above, if
we omit the blank specifier, FOROTS assumes that the first character
encountered (in this case, the "T" in "This") is the carriage-control
character.
Executing
this
example,
after
removing
the
carriage-control specifier, causes the first character to be stripped
from the character constant.
Thus, the output at the terminal is:
his is a string
The carriage-control characters are summarized in Table 12-3.
The $
(dollar sign)
output edit descriptor modifies the action of the
carriage-control specifier (see Section 12.4.4).

12-10

FORMATTED DATA TRANSFERS
Table 12-3: Carriage-Control Specifiers
Specifier

Format List Form

Printer Character

Octal Value

Effect on Carriage Control

012

Skip to next line (form feed after
60 lines on printer).

blank

' ,

plus

'+'

zero

'0'

LF,LF

012,012

Skip a line.

one*

'1'

014

Form feed to top of next page.

two*

'2'

DLE

020

Space to next half page.

three*

'3'

013

Space to next one-third of a page.

minus

'-'

LF,LF,LF

012,012,012

Skip two lines.

asterisk*

,*,

DC3

023

Skip to next line; suppress form
feed. (Continous print)

period*

DC2

022

Triple space, with a form feed
after every 20 lines printed.

comma*

' ,,

DC1

021

Double space, with a form feed
after every 30 lines printed.

slash*

'1'

DC4

024

Space to next one-sixth of a page.

Suppress line feed; overprint the
line.

Indicates carriage-control specifiers for which the effect on carriage control is device dependent. The effect
described is for a line printer with a standard form setup.

Note - This table assumes a standard form setup for your line printer (or other output device).

I/O List & FORMAT List
Interaction

12.3

INTERACTION OF INPUT/OUTPUT LIST AND FORMAT LIST

This section describes how the I/O list and the format
during a data transfer.

12.3.1

list

interact

General Description

Format control is initiated by execution of a formatted data transfer
statement.
The actions performed by format control depend on the
interaction of the edit descriptors in the
format specification and
the I/O list elements in the data transfer statement.
The following example shows how the
I/O list elements
in a data
transfer statement
interact with the edit descriptors in the format
list in a simple data transfer.

100

READ (5,100) N,X,Y
FORMAT (I5,F12.0,FlO.0)

12-11

FORMATTED DATA TRANSFERS
In this example, the I/O list is
N,X,Y
and the format specification is
(IS,F12.0,FI0.0)
The variables in the I/O list and the specifiers
matched up as follows:
N
X

the

format

are

IS
F12.0
FI0.0

A formatted data transfer statement matches elements of the
I/O list
and specifiers
in the format specification.
The matching proceeds
from left to right, one I/O list element to one
repeatable edit
descriptor.
In the above example, there are three elements in the
I/O list and
three format specifiers.
However,
the
interactions can be more
complicated than those in the example.
A format specifier can be
preceded by a repeat count, in which case it corresponds to more than
one element in the I/O list. Also, an element of the I/O list can be
an array name, in which case it can correspond to more than one format
specifier.
The number of elements in the I/O list and in the format
specification do not have to be the same.
Table 12-4 details what happens in these more complex cases.

12-12

FORMATTED DATA TRANSFERS
Table 12-4: Record, Format List, and 110 List Interaction
1.

Record ends
Format specification continues
1/0 list continues

Action:
The transfer continues as if the record was extended with blanks.
Example:
10

READ (S dO) A ,6 ,C
FORMAT (3Fl0.0)

Record contains:
40

Resulting values:
A = 40.0
B
10.0
C = 0.0

Note that this situation is not applicable to an output transfer.
2. 110 list ends
Format specification continues
Action:
The format scan continues until it encounters a repeatable edit descriptor, a colon, or until the rightmost right
parenthesis of the format is reached.
Nonrepeatable edit descriptors up to and including the first colon, if present, are processed if they are encountered during the scan.
Example:
A = 12
B = 123

C = 1234
D = 12345
20

WRITE (S ,20) A ,6 ,C
FORMAT (' A=' ,F3.0, ',6=' ,FLI.O, ',C=' ,FS.O, "
1
D=' ,F8.0)

Resulting output:
A = 12., B = 123., C = 1234., D =
Note that the D = descriptor was processed. The format scan does not terminate until it encounters a descriptor which requires an I/O list element, and there is no I/O list element to supply. However, the colon edit
descriptor will cause the format scan to stop if there is nothing left in the I/O list (see Section 12.4.6).
I,

3. Format specification ends
1/0 list continues
Action:
A new record is started. The format scan continues, starting at the beginning of the last complete parenthesized
group within the format specification. If there is no parenthesized group within the format, the format is
restarted from the beginning (see Section 12.3.4).
Example:
A=123.
6=78SLI.12S
C=1.812S

WRITE (SilO) A,6,C
FORMAT (F1S.8)

12-13

FORMATTED DATA TRANSFERS
Table 12-4: Record, Format List, and 1/0 List Interaction (Cont.)
Resulting output:
123.000000
7654.125000
1.612500
4. 1/0 list ends

Format specification ends
Action:
On input, if there are any characters remaining in the record, they are ignored. On output, the record is simply
terminated without any extra characters added.
Example:
10

READ (5 t1 0) GAMMA
FORMAT (F5. 2)

Record contains:
12.34 value of GAMMA
Resulting values:
GAMMA = 12.34
The extra data (the comment) in the input record is ignored.

The execution of a formatted I/O statement proceeds by matching I/O
list elements and FORMAT edit descriptors. The edit descriptors I, 0,
Z, R, F, E, 0, G, L, A, and Q each correspond to one element of the
I/O list. No I/O list element corresponds to H, X, P, T, :, $, S, SP,
SS, BN, BZ, or apostrophe edit descriptors.
If one of these
descriptors is encountered, it is executed and the format scan
continues.

12.3.2

Formatted Input

A formatted input statement begins by reading a record from the
specified unit.
The format is scanned from left to right. X, P, T,
BN, BZ, and / edit descriptors are executed as they are encountered.
If an I, 0, Z, R, F, E, 0, G, L, A, or Q descriptor is encountered,
data is read into the corresponding I/O list element as specified by
the edit descriptor.
If the I/O list contains no more elements, execution of the READ
statement ends.
Additional records will be read from the specified
unit when a slash occurs in the format, or when the last right
parenthesis of the format is reached, and I/O list elements remain to
be filled.
When an input record is terminated by a slash or by the end of the
format, any data left in the input record is discarded. If the input
record is exhausted before the data transfers are completed, the
remainder of the transfer is completed as if the record were extended
with blanks.

12-14

FORMATTED DATA TRANSFERS
12.3.3

Formatted Output

A formatted output statement begins by scanning the format.
The H, X,
P, T, BN, BZ, :, /, $, S, SS, SP, and apostrophe edit descriptors are
executed as they are encountered.
If an I, 0, Z, R, F, E, D, G, L, or
A descriptor is encountered, data is translated from the corresponding
I/O list element, as specified by the edit descriptor, and placed in
the output record.
If the I/O list contains no more elements, the output record
is
written to the specified unit, and execution of the WRITE statement
ends. Additional records will be written to the specified unit when a
slash occurs in the format,
or when the last right parenthesis is
reached, and I/O list elements remain to be transferred.

12.3.4

Embedded Format Specifications

Format specifications may contain embedded format specifications with
optional repeat specifications.
If a repeat specification is used,
the entire format specification that it precedes is scanned the
specified number of times during the I/O transfer.
In the example:
100

WR I TE (1, 100) A, B, C , D, E , F , G , H, I , J
FORMAT (FlO.2,4 (I5,lX,I3) ,18)

the variable A is matched with the format
item F10.2.
Then,
the
variable B is matched with 15, variable C with 13, variable D with IS,
and so on for four iterations of the embedded format specification.
Finally, the variable J is matched with the format item 18.
If no repeat specification is used preceding
specification, a repeat count of 1 is implied.

embedded

format

When the last right parenthesis of the format is reached, and more I/O
list elements remain to be transferred, a new record is started and
format scanning continues. The scanning continues at the beginning of
the format specification whose right parenthesis is the next to last
right parenthesis in the format.
If there are no embedded format
specifications,
format scanning continues at the beginning of the
format.
For example:
DIMENSION A(lOO)
INTEGER CASE
CASE=33
100

WRITE (1,100)CASE, (A(I) ,1=1,100)
FORMAT ('THIS IS CASE' ,I6,//,4(lX,FlO.5»

After A(l) through A(4) are written, a new line is started and format
scanning
continues
at
the
beginning of the embedded format
specification. Thus, A(5) through A(8) are written; a new record
is
started, and so forth.

12-15

FORMATTED DATA TRANSFERS
The output file would appear as follows:
THIS IS CASE 33
A(l)
A(5)
A(9)

A (2)

A(3)

A(4)

A(6)
A(lO)

A(7)
A (11)

A (12)

A(8)

Example:
DIMENSION A(5) ,DATl(lOO) ,B(4,100) ,DAT2(100)
WRITE (1,100) CASE, (A(K) ,K=1,5), (DATI (J), (B(I,J) ,I=1,4),
1 DAT2 (J) ,J=l, 100)
100 FORMAT ('CASE' ,I5,11,5(lX,FlO.3) ,1,(FI0.3,4(3X,F15.5)
1 ,lX,FlO.3»
In this example, after A(l)
through A(5),
DATl(l), B(l,l)
through
B(4,1), and DAT2(1) are written, a new record is started, and format
scanning begins. The format scanning begins at the embedded format
specification
following the 'I'
(the specification whose right
parenthesis is the next to last right parenthesis).
The output file would appear as follows:
CASE
A (1)
DATI (1)
DATl(2)
DAT1(3)

12.4

33
A(2)
B(l,l)
B(1,2)
B(1,3)

A(3)
B(2,1)
B(2,2)
B(2,3)

A(4)
B(3,1)
B (3 ,2)
B(3,3)

A (5)
B(4,1)
B(4,2)
B(4,3)

DAT2(1)
DAT2(2)
DAT2(3)

FORMAT EDITING

Tables 12-1 and 12-2 describe forms of all the
FORMAT
edit
descriptors.
The edit descriptors enable you to specify the form of a
record and to specify the editing of the data as it is transferred.
The edit descriptors are described according to the character used to
accomplish a particular modification to the data or record in which
the data are stored.

APOSTROPHE (')
Editing

12.4.1

Apostrophe (I) Editing

The apostrophe (')
edit descriptor
(single-quote)
include a character constant in a format list.

12-16

enables

you

FORMATTED DATA TRANSFERS
The form of the apostrophe edit descriptor is:
'hl ••• hn'
where:
'hl ••. hn'

is a character constant.

To include an apostrophe as part of the character constant,
use two successive apostrophes.
This descriptor is only used for output; the
the apostrophes are written.

characters

you

must

enclosed

Example:
10

TYPE 10
FORMAT (' That' 's the way!')

will output
That's the way!

H
Editing

12.4.2

H Editing

The H edit descriptor (also called the Hollerith descriptor)
you to include character strings in a format list.

enables

The form of the H edit descriptor is:
nHhl ••• hn
where:
n

is a nonzero, unsigned, integer constant that indicates
the total number of ASCII characters included in the
string.

hl ••• hn

is a string of ASCII characters.
set is described in Appendix B.)

(The ASCII

character

You may transmit alphanumeric data directly from the FORMAT
using either the H or apostrophe specifiers.

statement

This descriptor is only used for output; the n characters that follow
the H are written.
For example, you can use the following statement
sequence to print the words PROGRAM COMPLETE on the printer:
PRINT 101
101 FORMAT (17HiPROGRAM)iCOMPLETE)

12-17

FORMATTED DATA TRANSFERS
The result of apostrophe editing is the same
For example, you may use the descriptors:

Hollerith

editing.

101 FORMAT (17H~PROGRAM~COMPLETE)
and
101 FORMAT ('~PROGRAM~COMPLETE')
in the same manner.
Apostrophes can appear anywhere within a Hollerith edit descriptor
without having to be represented by two apostrophes.
However, if the
H edit descriptor occurs within a character constant, the apostrophe
is written as two apostrophes, which are counted as one character.

POSITIONAL
Editing

12.4.3

Positional Editing

The positional edit descriptors specify the position at which the next
character will be transmitted to or from the record.
The positional
edit descriptors are:
T, TL, TR, and X.
The T edit descriptor specifies the character position within a record
where the next character will be transmitted (see Section 12.4.3.1).
The TL and TR descriptors specify the number of character positions to
the left or right,
respectively, of the current position for the
character position of the next character (see Section 12.4.3.1).
The X descriptor specifies the number of character positions to the
right of the current position for the character position of the next
character (see Section 12.4.3.2).
NOTE
On output, a record is initially filled with blanks.
Therefore,
fields skipped by the positional editing
descriptors will be blank-filled.
However, the output
record length is determined by actual output. Merely
specifying a positional editing descriptor with no
output will not change the record size. Thus, the
record written with:
FORMAT (I6,SOX,TIO,I3)
will have a record length of 13 characters,
output was done after the SOX.
Examples:
The statement sequence:
PRINT 2
2 FORMAT (TSO,'BLACK' ,T30,'WHITE')
12-18

since

FORMATTED DATA TRANSFERS
causes the following line to be printed:
fLACK

rHITE
(print position 30)

(print position 50)

The statement sequence:
1 FORMAT (T35,'MONTH')
READ (2,1)
causes the first 34 characters of the input data associated with
logical unit 2 to be skipped, and the next five characters to replace
the characters M, 0, N, T, and H in storage.
If an input record containing:

ABC}J)i~XYZ
is read with the format specification:
10 FORMAT (T7,A3,T1,A3)
then the characters XYZ and ABC are read in that order.
You can use the field descriptor nX to introduce blanks into output
records or to skip characters of input records.
The letter X
specifies the operation, and n is "a positive integer that specifies
the number of character positions to be either made blanks (output) or
skipped (input).
The statement:
FORMAT (5H~STEP,I5,10X,2Hy~,F7.3)
may be used to print the line:

T, TL, and TR
Editing

12.4.3.1 T, TL, and TR Editing - The T edit descriptor specifies that
the transmission of the next character to or from a record is to occur
at the specified character position.
The form of the T descriptor is:
Tc
where:
c

is a positive, unsigned, integer constant that indicates the
character position to or from which the next character will
be transferred.

For example:
FORMAT (T20,I5,T8,I2)
12-19

FORMATTED DATA TRANSFERS
specifies that the characters read or written for
the 15 descriptor
will
start
in character position 20,
and
the characters read or
written for the 12 descriptor will start in character position 8.
For
output to carriage-control devices
(line printer and terminal), Tn
specifies that n-l will be
the next character written,
since the
character position 1 is the carriage-control character position.
The TL edit descriptor specifies that the
transmission of the next
character
to or from the record is to occur at a position which is a
specified number of positions backward from the current position.
The form of the TL edit descriptor is:
TLc
where:
c

is a positive, unsigned, integer constant that indicates the
character position,
c
positions backward from the current
position, to or
from which the next character will
be
transferred.
If c
would cause transmission to start at a
position before the beginning of the current
record,
transmission will start instead at position one.

For example:
FORMAT (I5,T13,A5,TLlO,I2)
specifies that the characters read or written for
the A5 descriptor
will
start at character position 13, and that the characters read or
written for the 12 descriptor at will start at character position 8.
The TR edit descriptor specifies that the
transmission of the next
character
to or
from a
record will occur at a position that is a
specified number of positions forward from the current position.
The
function of this
form is identical to that of the X descriptor (see
Section 12.4.3.2).
The form of the TR edit descriptor is:
TRc
where:
c

is a positive, unsigned, integer constant that indicates the
character position,
c
positions forward from the current
position, from which the next character will be transferred.

Example using all three types of T descriptor:
10

TYPE 10
FORMAT(' 234567890123456789012345678901234567890')

TYPE 20
FORMAT(T29,'BLACK' ,TlO,'WHITE')

TYPE 30
FORMAT(TlO,'90l2' ,TR5,'890l')
TYPE 10
TYPE 40
FORMAT(T20,'(SECOND)' ,TLI0,')FIRST(')
END
12-20

FORMATTED DATA TRANSFERS
EXECUTE TEST2.FOR
FORTRAN: TEST2
MAIN.
LINK:
Loading
[LNKXCT TEST2 execution]
234567890123456789012345678901234567890
WHITE
BLACK
9012
8901
234567890123456789012345678901234567890
)FIRST<ND)
CPU time 0.1

Elapsed time 0.5

In FORMAT 20, 'BLACK' is written, then 'WHITE' is written to the left
of it.
In FORMAT 30,
five positions are skipped between the two
character strings being printed.
In FORMAT 40, '<SECOND)' is written,
the format goes back ten positions and writes ')FIRST<' over the
previously written character string.

X
Editing

12.4.3.2 X Editing - The X edit descriptor specifies that
the
transmission of characters to or from a record will occur a specified
number of characters forward from the current position.
The form of the X edit descriptor is:
[n] X

where:
n

is an optional, unsigned, positive,
integer constant that
indicates the number of characters forward from the current
position, at which the next character will be transmitted.
The default value is 1.

Example:
10

TYPE 10
FORMAT(' 1234567890123456789012345678901234567890')

TYPE 20
FORMAT(' A WORD OR TWO',10X,'OR THREE')
END

EXECUTE TEST3.FOR
FORTRAN: TEST3
MAIN.
LINK: Loading
[LNKXCT TEST3 execution]
1234567890123456789012345678901234567890
A WORD OR TWO
OR THREE
CPU time 0.2

Elapsed time 2.1
12-21

FORMATTED DATA TRANSFERS
In this example, ten positions are skipped between the printing of
'A WORD OR TWO' and 'OR THREE'.

-c:- '--'--;-":J
(DOLLAR SIGN)
Editing

12.4.4

(Dollar Sign) Editing

The $ (dollar sign) output edit descriptor suppresses all carriage
control at the end of the current record (for CARRIAGECONTROL='LIST')
or at the beginning of the next record (for CARRIAGECONTROL='FORTRAN'
or 'TRANSLATED').
This descriptor is used for interactive I/O; it leaves the terminal
position at the end of the text so that a response will follow the
output on the same line.
Example:
WRITE (5,10)
READ (5,*) N
FORMAT (' Number of samples:'
WRITE (5,20)
READ (5,*) X
FORMAT (I Mean value:
END

10
20

If the user enters 100 for Nand 1.23 for
will produce the typescript:

executing

the

program

the

several

Number of samples: 100
Mean value:
1.23
The $ edit descriptor can be used to append
statements into a single line.
For example:
DO 10 I = 1,10
WR I TE ( 5 , 20 ) I
FORMAT (lX,I3,$)
WRITE (5,20)
END

10
20

will produce one line of output:
1

The $ edit descriptor is ignored for input.

12-22

output

FORMATTED DATA TRANSFERS

/ (SLASH)
Editing

12.4.5

(Slash) Editing

The / (slash) edit descriptor indicates the end of data transfer for a
record.
Two consecutive slashes indicate the transmission of an empty
record.
On input to a file connected for sequential access,
the remaining
portion of the current record is skipped, and the file is positioned
at the beginning of the next record.
This new record becomes the
current record.
On output to a file connected for sequential access,
the current record is terminated, and a new record is created, which
becomes the current and last record of the file.
A record that contains no characters may be written.
record may be skipped on input.

Also, an

entire

If the file is connected for direct access,
the record number
is
increased by one, and the file is positioned at the beginning of the
record that has that record number. This record becomes the current
record.
The following statements will write a record with no characters:
100

WRITE (1,100)
FORMAT(/)

To handle a group of I/O records where different records have
different field descriptors,
use a slash to indicate a new record.
For example, the statement
FORMAT (308/15,2F8.4)
is equivalent to
FORMAT (308)
for the first record, and
FORMAT (I5,2F8.4)
for the second record.
You may omit separating commas when you use a slash. When n slashes
appear at the beginning or end of a format, n blank records will be
in the
written on output or skipped on input. When n slashes appear
format,
n-l blank records are written on output or n-l
middle of a
records are skipped on input.

12-23

FORMATTED DATA TRANSFERS

: (COLON)
Editing

12.4.6

(Colon) Editing

The:
(colon) edit descriptor terminates format control if there are
no more
items
in the
I/O list of the corresponding data transfer
statement.
The colon edit descriptor has no effect if there are any
items left
in the I/O list of the corresponding data transfer
statement.
Example:
Xl
X2
X3
10
20

100
200
300

TYPE 10, Xl,X2
FORMAT(F6.1,F6.1,:,'THIS SHOULD NOT PRINT' ,F6.1)
TYPE 20,X1,X2,X3
FORMAT(F6.1,F6.1,:,' BUT THIS SHOULD' ,F6.1)
END

EXECUTE TEST4.FOR
FORTRAN: TEST4
MAIN.
LINK:
Loading
[LNKXCT TEST4 execution]
100.0 200.0
100.0 200.0 BUT THIS SHOULD 300.0
CPU time 0.1
Elapsed time 1.5

S, SP, and 5S
Editing

12.4.7

S, SP,and SS Editing

The S, SP , and SS edit descriptors control the output of the optional
plus sign
in numeric output fields.
These descriptors are used as
follows:
S

indicates that the system-defined action is taken.
system-defined specification for Digital FORTRAN is SSe

indicates that the plus sign
numeric output fields.

indicates that the plus sign is not printed
in positive
numer ic output fields.
This descriptor is the default.
12-24

printed

all

The

positive

FORMATTED DATA TRANSFERS
The S, SP, and SS edit descriptors affect only I,
F,
E,
D, and G
editing during the execution of a data transfer output statement;
these edit descriptors have no effect during input transfers.
These
descriptors stay in effect until the end of the I/O transfer or until
another S, SP, or SS is encountered.
Example:
Xl
X2
X3
X4
10

100
200
300
400

TYPE 10, XI,X2,X3,X4
FORMAT("
,S,F7.l, SP,F7.1, SS,F7.l, SP,F7.l)
END

EXECUTE TEST5.FOR
FORTRAN: TEST5
MAIN.
LINK:
Loading
[LNKXCT TEST5 execution]
100.0 +200.0 300.0 +400.0
CPU time 0.1
Elapsed time 0.2
In this example,
X2 and X4 have plus signs because of the
descriptors in front of the F descriptors that correspond to them.

p
Editing

12.4.8

P Editing

The P edit descriptor is used to specify a scale factor for F,
and G edit descriptors.

representing

the

The form of the P edit descriptor is:
kP

where:
k

is an optionally signed integer
scale factor.

constant

If a scale factor is not specified in a format list, a scale factor of
zero
is assumed.
Once a scale factor is specified in a format list,
that scale factor remains in effect for
all
F,
E,
D,
and G edit
descriptors until a new scale factor is specified, or to the end of
the execution of the current I/O statement.
Scale factors have no
effect on I, Z, and 0 edit descriptors.

12-25

FORMATTED DATA TRANSFERS
The scale factor affects the F, E, D, and G data transfers as follows:
On input:
If there is an exponent field, the scale factor k has no effect.
If
there
is no exponent field on the number read in, the number is
multiplied by 10**(-k) before being assigned to the input variable.
On output:
The basic real constant part of the quantity, in E and D editing only,
is multiplied by 10**k and the exponent is reduced by k.
For G
editing, the scale factor has no effect unless the magnitude of the
data item to be edited is outside of the range that permits the use of
F editing.
If G editing is specified, and the magnitude of the data
item to be edited is such that E editing is required, the scale factor
has the same effect as with E output editing (see Section 12.4.11.3).
The comma is optional between a
P edit descriptor
following F, E, D, or G edit descriptors.

and

For example, assume the data involved is the real number
edit descriptor

immediately
26.451;

the

F8.3
produces the external field

The addition of the scale factor of -lP, as in
FORMAT (-lP,F8.3)
produces the external field

When you add a scale factor to D, E,
and G (external
field not a
decimal fixed-point) edit descriptors, the scale factor multiplies the
number by the specified power of ten,
and
the exponent
is changed
accordingly.
In input operations, type F (and type G,
if the external
field
is
decimal
fixed-point)
conversions are the only ones affected by scale
factors.
When you add a scale factor to a D or E edit descriptor, it specifies
a power of 10 so that the external form of the number has its mantissa
multiplied by the specified power of 10;
its exponent
is adjusted
accordingly.
For example, assume the data involved is the real
edit descriptor
Ell.3
produces the external field
1jO.125E+02
The addition of the scale factor 2P, as in
FORMAT (2P,Ell.3)
12-26

number

12.49;

the

FORMATTED DATA TRANSFERS
produces the external field
.J2fj512.49E+00
with a scale factor of zero, the number of significant digits
by a format of the form:

printed

Ew.d
or
Dw.d
is the number of digits to the right of the decimal point.
For a negative scale factor nP, for -d<n<O, there will be ABS(n)
leading zeros and d-ABS(n) significant digits after the decimal point
(for a total of d digits after the decimal point).
If n<-d, there
will be d insignificant digits (zeros) to the right of the decimal
point.
If the scale factor nP is positive, for 0<n<d+2 there will be n
significant digits to the left of the decimal point and d-n+l
significant digits to the right of the decimal point (for a total of
d+l significant digits).
If n>d+2, there will be d+l significant
digits and n-d-l insignificant trailing zeros on the left of the
decimal point.
If the data to be printed is 12.493, these formats produce results
follows:
FORMAT

OUTPUT

SIGNIFICANT
DIGITS

REASON

E15.3
lPE15.3
-lPE15.3
2PE15.3
-3PE15.3
4PE15.3
6PE15.3

}6p5J6J6J6~0.125E+02

3
4

n=O
n<d+2
-d<n
n<d+2
n<-d
n<d+2
n>d+2

J'6t6J6)6,lSjSl.249E+Ol

~J'6t6J6}6,kS0.012E+03

2
4

t6t6J6J6J6J6l2.49E+00
t6.kS.kSJ6.kS)60.000E+05
J6J6)6,sJ'6jS1249.E-02
Jz$JzSJDjlS124900.E-04

o
4
4

Example:
10
20
30

TYPE 10
FORMAT(' Type in a real number')
ACCEPT 20,Xl
FORMAT(2P,F)
TYPE 30,Xl
FORMAT(' Number read with P=2 =' ,F,
1 I,' (Number read)*10**(-2) ')
TYPE 40,Xl
FORMAT(/,' The above number written with P=2' ,I,
1 ' i s =' , 2 P , F , I " ( Nurn b era b 0 v e) * 1 0 * * (2) , )
END

12-27

FORMATTED DATA TRANSFERS
EXECUTE TESTP.FOR
FORTRAN: TESTP
MAIN.
LINK:
Loading
[LNKXCT TESTP execution]
Type in a real number
5.
Number read with P=2 =
(Number read)*10**(-2)

0.0500000

The above number written with P=2
is =
5.0000000
(Number above)*10**(2)
CPU time 0.2
Elapsed time 5.1
The number the program receives is (5.)*(10**(-2»
out is (.05) * (10** (2» •

and the value typed

BN and BZ
Editing

12.4.9

BN and BZ Editing

The BN and BZ edit descriptors specify how blanks other than leading
blanks are interpreted only for numeric input fields where a width has
been specified. These edit descriptors have no effect on numeric
output fields.
The BZ descriptor specifies that blanks will be read as zeroes.
The
BN descriptor specifies that blanks will not be read as zeroes.
The
use of the BN or BZ edit descriptors in a format overrides the BLANK=
specifier
in the OPEN statement for the duration of the use of that
format.
(The BLANK= specifier is described in Section 11.3.3.)
For example:
101

ACCEPT (FMT=101)A,B,C,D
FORMAT (BN,I5,FIO.2,BZ,FIO.2,F8.5)

reads the first two numbers of data, ignoring blanks embedded
in the
numbers.
Then the program reads the second two numbers, substituting
zeroes for blanks embedded in the numbers.

12-28

FORMATTED DATA TRANSFERS

Q
Editing
..

12.4.10

. ------J

_---------

Q Editing

The Q edit descriptor sets a corresponding integer variable in the I/O
list to the number of characters left in the record being transferred.
This descriptor is for use with input transfers only.
You can
use
multiple Q descriptors in the same format list.
The Q edit descriptor
is useful when you need to know the number of characters remaining
in
a record.
For example:

100

TYPE *,'Enter text:'
ACCEPT 100,L,Jl
FORMAT (AS,Q)

when used to read the data
Enter text:
HELLO THIS IS A TEST
would yield
the value
IS
for
variable Jl,
characters in the data, and AS reads S of them.

since

there

are

Numeric
Editing

12.4.11

Numeric Editing

The I, F, E, D, G, Z, and a edit descriptors are used to
specify the
input
and
output of
integer,
real,
complex,
double-precision,
hexadecimal, and octal data.
The numeric edit descriptors are repeatable, and can be
used
without
specifying
size.
For output,
if you use a numeric edit descriptor
without specifying a field width, the defaults shown in Table l2-S are
used.
For input, the data is scanned until
a
blank,
comma,
or
character
illegal for the specified edit descriptor is encountered, except for A
format, which uses the defaults shown in Table 12-S.

12-29

FORMATTED DATA TRANSFERS
Table 12-5:

Default Field Widths for Numeric Edit Descriptors

r - - - - - - - - - - - -.. -.- .-.--..-.-..---....... ---. - . - - - - - - - - - - - - - - - - - - - - - - - - - - - - - ,

Edit Descriptor

Default Field Width

----.----- --.- .---. .-----.- . .-.----. -. --...-. -- _..__._.__._...._._-_._--_._._._-------------1
115
-_._-----_._-_ .. _._+ .. _......_._... .

F (single prec.)

*1<'15.7

F (double prec.)

*F25.18

E (single prec.)

E15.7

. -.---.----.- - .-.- ._._ ..._-... _-_._._--_._------------------1

-...- - . - . - - - - - - - - - - - - - - - - . - - . - - - - - - - - . - - . - - - - - - - - - - - - -- - - - - I

E (double prec.)

E25.18

- - - - - - -. --.--- --.--.-----.---- .--.----. ----.--. - - - - - - - - - - - . - - - - - - - - - - - - - - - - 1
D15.7
D (single prec.l

. .---- ._-_._._._--------_._----------------1
D (double prec.l

D25.l8

r----------.---+-.----.-.--.--.-.-.------.-------.---------------------/

G (single prec.)
1-----_._--------

GI5.7
.-.--------..---.--.------... -----

G (double prec.)

G25.18

o (single prec.)
o (double prec.)

015

....-..---.-.--.-.-..-...---.-- ... - - - - - - - - - - - - - - - - - - - - - - 1

----.-.. --.-------.-.-.. - - - - - - - - - - - - - - - 1

--.------..... ------.------. -. -----_._---_._----------------1
025

------_._._--_._---_.__. _ - - - - - - - - - - - - - - - - - - - - - - - - - 1

LI5

Z (double prec.)

Z25

A (single prec.l

A (double prec.)

Ala

R (single prec.l

r - - - - - - - - - - --+-------------.--- - . - - - -..--.------.-------.----------1
Z (single prec.l
Z15
-----------.--.---------------------1
-.. --.---.. -- .---.---.-.------------------1
.-.-- ..--..-.-.-----.-.--.---.-.-.---.-.--.. - . - - - - - - - - - . - - . - - . - - - - - f

1------------_._-

R (double prec.)

---.-.. - - - - - - - . - - - - - - - - - - - - - - - - - - - - - - - - - f

.-. ----.----- ..--.-------------.--------------------1

RIO

If the default field width for F tormat is too small for the data, the field width expands to fit the data.

The following conventions apply to all I/O transfers using the numeric
edit descriptors:
1.

The interpretation of blanks is determined by a combination
of any BLANK= specifier in the corresponding OPEN statement
(see Section 11.3.3), and any BN or BZ edit descriptor
(see
Section 12.4.9)
that is currently in effect in the format
list. A field of all blanks is always equal to zero.

On input transfers, with F, E, D, and G editing, a decimal
point appearing in the input field overrides the portion of
the edit descriptor that specifies the location of the
decimal point.

On output transfers, the representation of a positive or zero
value in the field may be prefixed with a plus, as controlled
by the S, SP, and SS edit descriptors (see Section 12.4.7).
The representation of a negative value in the field is
prefixed with a minus sign.

On output transfers, the representation of the transferred
datum is right-justified in the specified field.
If the
number of characters produced by the editing is smaller than
the specified field width, leading blanks are inserted in the
field.

12-30

FORMATTED DATA TRANSFERS
5.

On output transfers, if the number of characters produced
exceeds the field width, or if an exponent exceeds its length
(as specified in the Ew.dEe or Gw.dEe forms),
the entire
field width, represented by w, is filled with asterisks.

The next sections describe the individual numeric edit descriptors.

I
Editing

12.4.11.1
editing.

Editing - The

edit

descriptor

specifies

integer

The form of the I edit descriptor is:
[r]I[w[.m]]
where:
r

is an optional,
nonzero,
unsigned,
integer
constant
indicating how many fields of I are being specified.
The
default is one field.

is an optional, nonzero, unsigned, integer constant that is
equal
to the
total number of digits in the integer field
being described.
If w is not specified,
for output,
the
value
is 15
(the default); for input, the data is scanned
until a blank, comma, or character illegal for
the
I
edit
descriptor is encountered •

•m

is an optional, unsigned, integer constant (separated from w
by a period) that indicates the minimum number of digits to
be output to the integer field being described.
The default
is one digit
(I15.l).
If necessary, leading zeroes are
output.
The value of m must not exceed the value of w.
If m is zero
and the value of the internal data item is zero, the output
field consists of only blank characters, regardless of the
sign control in effect.

On input, the Iw.m and the Iw forms
treated the same.

the

edit

descriptor

are

Example:
10

FORMAT(I,I8,219.5)

The first data item is output as a
1- to
IS-digit right-justified
integer
in the first 15 columns.
The second item is a 1- to 8-digit
integer occupying the next 8 columns.
The third and fourth items are
5- to 9-digit integers occupying 9 columns each, with leading zeroes
appended to the data to make them 5 digits if necessary.

12-31

FORMATTED DATA TRANSFERS

F
Editing

12.4.11.2 F Editing - The
(floating-point) editing.

edit

descriptor

specifies

real

The form of the F edit descriptor is:
[r] F[w.d]
where:
r

is an optional,
nonzero,
unsigned,
integer
constant
indicating the number of fields of F being specified.
The
default is one field.

is an optional, nonzero, unsigned, integer constant equal to
the total number of digits in the F field being described.
This total includes the digits to the right and left of the
decimal
point,
the decimal point itself, and (if included)
the sign.
On input, if the decimal point is omitted,
the
rightmost d digits of the string, with leading zeros assumed
if necessary, are interpreted as the fractional part of the
value represented.
If w is not specified,
for output,
the
value
for
single-precision is 15
(the default),
and the value for
double-precision is 25 (the default).
For input,
the data
is scanned
until a blank, comma, or character illegal for
the F edit descriptor is encountered •

•d

is an optional, nonzero"
unsigned,
integer constant that
specifies the total number of fractional digits in the field
of width w.
The default for single-precision is 7 digits;
the default for double-precision is 18. The maximum is 63
digits.
NOTE
If the default field width for F format (with no width
specified)
is too small for the data, the field width
expands to fit the data.

E and 0
Editing

12.4.11.3 E and D Editing - The E
editing of real data.

and

The form of the E edit descriptor is:
[r]E[w.d[Ee]]
12-32

edit

descriptors

specify

FORMATTED DATA TRANSFERS
The form of the D edit descriptor is:
[r]D[w.d[Ee]]
where:
r

is an optional, nonzero, unsigned, integer constant that is
equal
to
the number of E or D fields being described.
The
defaults are one single-precision
E
field
and
one
double-precision D field.

is an optional, nonzero, unsigned, integer constant equal to
the total number of digits
in the E or D field being
described.
The total for both types of fields is equal to:
1.

The total number of digits on both sides of the
point

The decimal point itself

The sign for the number (if included)

The exponent character (if included)

The digits in the exponent

decimal

If w is not specified, for output,
the value
for
single
precision is 15
(the default),
and the value for double
precision is 25 (the default).
For
input,
the data
is
scanned until a blank, comma, or character illegal for the E
or D edit descriptor is encountered.
d

is an optional, nonzero, unsigned integer constant equal to
the
total number of fractional digits in the field being
described (unless a scale factor greater than one is in
effect).
The defaults for both the E and D edit descriptors
are 7 digits
if single precision,
and 18
if
double
precision. The maximum is 63 digits.

is an optional, nonzero, unsigned, integer constant equal to
the total
number of digits
in the E or D field being
described.
The default for both types of edit descriptors
is two digits.
The maximum is 15 digits.
For KL model B systems, if the program is compiled with the
/GFLOATING switch
(see Section 16.1.3 or 16.2.3), you may
want to specify three digits to accommodate the exponent
field of double-precision numbers.

G
Edmng

12.4.11.4 G Editing - The G edit descriptor allows editing
of
integer,
real, double-precision, logical, complex, or character data.
with the exception of real, double-precision, and complex data,
the
type of conversion performed by the G edit descriptor depends on the
type of the corresponding variable in the I/O list.
12-33

FORMATTED DATA TRANSFERS
The form of the G edit descriptor is:
[r]G [w.d [Ee]]
where:

is an optional, nonzero, unsigned, integer constant that
is
equal
to
the
number of G fields
being described.
The
default is one.

is an optional, nonzero, unsigned, integer constant equal to
the
total
number of digits in the G field being described.
The total for both types of fields is equal to:
1.

The total number of digits on both sides of the
point

The decimal point itself

The sign for the number (if included)

The exponent character

The digits in the exponent

decimal

(if included)

If w is not specified, for
output,
the value
for
single
precision
is 15
(the default),
and the value for double
precision is 25 (the default).
For
input,
the data
is
scanned until a blank, comma, or character illegal for the G
edit descriptor is encountered.
d

is an optional, nonzero, unsigned integer constant equal
to
the
total
number of fractional digits in the field being
described (unless a scale factor greater
than one
is
in
effect).
The defaults are 7 digits if single precision, and
18 if double precision.
The maximum is 63 digits.

is an optional, nonzero, unsigned, integer constant equal to
the
total
number of digits in the G field being described.
The default is two digits.
The maximum is 15 digits.
For KL model B systems, if the program is compiled with
the
/GFLOATING switch
(see Section 16.1.3 or 16.2.3), you may
want to specify three digits
to accommodate
the exponent
field of double-precision numbers.

For input, in the case of real, double-precision,
and complex data,
the G-format conversion is the same as for E-format conversion.
For
output, however, the type of conversion performed depends on the
magnitude of the data items.
Table 12-6 illustrates the conversion
performed for various ranges of real,
double-precision,
and complex
data.

12-34

FORMATTED DATA TRANSFERS
Table 12-6: Effect of Data Magnitude on G·Format Output
Conversions
Data Magnitude (m)

Effective Conversion

m .LT. 0.1.
0.1 .LE. m .LT. 1.0
1.0 .LE. m .LT. 10.0

Ew.d
F(w - n).d,n(x)
F(w - n).(d-1),n(x)

10**d-2 .LE. m .LT. 10**d-1
10**d-1 .LE. m .LT. 10**d
m .GE. 10**d

F(w - n).l,n(x)
F(w - n).O,n(x)
Ew.d

where:
x

is a blank

n is 4 for Gw.d and e + 2 for Gw.dEe

where:
x

is a blank.

is 4 for Gw.d and e+2 for Gw.dEe
NOTE

In all numeric field conversions, the field width
(w)
you specify should be large enough to include the
decimal point, sign, and, where applicable,
the
exponent character
(E), the exponent sign, plus the
exponent digits. This is in addition to the number of
digits in the number to be represented.
If the specified width is t~o small to accommodate the
converted number, the field will be filled with
asterisks (*). If the number converted occupies fewer
character positions than specified by w, it will be
right-justified in the field, and leading blanks will
be used to fill the field.
If the numeric data representation cannot fit into the
field width F{w-n), the n spaces (n{x»
are removed
from the right, and the numeric data representation is
again processed into the field width Fw.
Examples of G output conversions (where the ~ signifies a blank) are:
Format

Internal Value

External Representation

G13.6
G13.6
G13.6
G13.6
G13.6
G13.6
G13.6
G13.6
G13.6

0.01234567
-1.12345678
1.23456789
12.34567890
123.45678901
-1234.56789012
12345.67890123
123456.78901234
-1234567.89012345

9I0.123457E-01
-0. 123457J6Jb;5»
Jb)51. 23457t6~)5J6

12-35

y.sp512.3457)5J6t5JzS:
J6t6123.457j2S}6.J6~
t6-1234.57J6J6J6~
,t6,t612 345. 7p5J6,l6iz{
J2Sj2S123457.J6J2S}6'P{
-0.123457E+07

FORMATTED DATA TRANSFERS
For comparison, consider the following example of the same
output under the control of an equivalent F field descriptor.

values

Format

Internal Value

External Representation

F13.6
F13.6
F13.6
F13.6
F13.6
F13.6
F13.6
F13.6
F13.6

0.01234567
-0.12345678
1.23456789
12.34567890
123.45678901
-1234.56789012
12345.67890123
123456.78901234
-1234567.89012345

~J6)6)5;S0 • 012346
,JzS)6)S)S-0.123457
Jzl)6JlSJ6J61.234568
J6Jl5JlSJzH2.345679
~Jijp5123.456789

J2l-1234.567890
~12345.678901

123456.789012

*************

Complex
Editing

12.4.11.5 Complex Editing - A complex number consists of a pair of
separate real numbers.
The first number of the pair is the real part
of the complex number; the second number is the imaginary part.
The editing of a complex number involves specifying two successive F,
E,
D, or G edit descriptors. The edit descriptors need not be the
same.
A sample format list description for a complex number is:
101

FORMAT (FI0.2,EI0.2)

In this sample, the FI0.2 edit descriptor provides editing for
the
real part of the complex number; the E10.2 edit descriptor provides
editing for the imaginary part.
You may include any nonrepeatab1e edit descriptors between
and imaginary edit descriptors for a complex number.

(O~al)

the

real

]

Editing
'----------"~-.-.--.-----'"-

12.4.11.6 a (Octal) Editing - The 0 (octal) edit descriptor specifies
octal editing.
The form of the

a edit descriptor is:

[r]O[w[.m]]

12-36

FORMATTED DATA TRANSFERS
where:
r

is an optional,
nonzero,
unsigned,
integer
constant
specifying the number of successive octal
fields being
described.
The default is one octal field.

is an optional,
nonzero,
unsigned,
integer
constant
specifying the
total
number of digits in the octal field
being described.
If w is not specified,
for
output,
the value
for
a
single-precision octal
field
is 15 (the default), and the
value
for
a double-precision octal
field
is 25
(the
default).
For
input,
the data is scanned until a blank,
comma, or character illegal for the 0 edit descriptor
is
encountered.

is an optional, unsigned, integer constant specifying
the
minimum number of digits to be output to the field.
The
defaults are 12 for single-precision octal values and 24 for
double-precision octal values.

~--"----

Z
(Hexadecimal)
Editing
--~.-,-,---,-

------_.__ ...

12.4.11.7 Z Editing - The
Z edit descriptor specifies input and
output of hexadecimal values.
Hexadecimal is a base 16 number system
where the characters 0-9 and A-F (or a-f) represent the numbers 0-9
and 10-15, respectively.
(On output, A-F only.)
The form of the Z edit descriptor is:
[r] Z [w[ .m]]

where:
r

is an optional,
unsigned,
nonzero,
integer
constant
specifying the number of consecutive hexadecimal fields
being specified.
The default is one hexadecimal field.

is an optional,
unsigned,
nonzero,
integer
constant
specifying the
total
number of digits in the hexadecimal
field being described.
If w is not specified,
for
output,
the value
for
a
single-precision hexadecimal field is 15 (the default), and
the value for a double-precision hexadecimal
field
is 25
(the default).
For
input,
the data
is scanned until a
blank, comma, or character illegal for the Z edit descriptor
is encountered.

is an optional, unsigned, integer constant specitying
the
minimum number of digits to be output in the field.
The
default for single-precision hexadecimal fields is 9 digits;
the default for double-precision fields is 18 digits.

12-37

FORMATTED DATA TRANSFERS

L
Editing

12.4.12

L Editing

The L edit descriptor provides editing of logical data.
The form of the L edit descriptor is:
[r] L[w]
where:
r

is an optional,
unsigned,
nonzero,
integer
constant
specifying the number of consecutive logical fields being
described.
The default is one logical field.

is an optional,
unsigned,
nonzero,
integer
constant
specifying the
total number of characters in the logical
field being described.
If w is not specified, for output,
the value
is 15
(the
default)
For
input,
the data
is scanned until a blank,
comma, or character illegal for the L edit descriptor
is
encountered.

You may transfer logical data under format control in a manner similar
to numeric data transfer by use of the field descriptor
Lw
where:
L

is the control character and w is an integer specifying the
field width.
The data
is transmitted as the value of a
corresponding logical variable in the associated I/O list.

The input field consists of optional blanks, optionally followed by a
period,
followed
by a T for true or F for false, optionally followed
by any series of characters (such as, .TRUE.
for true or .FALSE.
for
false) •
If the enti re input data field is blank or empty, a value of
false is stored.
On output, w minus 1 blanks followed by T or F will be output
value of the logical variable is true or false, respectively.

12-38

the

FORMATTED DATA TRANSFERS

A
Editing

12.4.13

A Editing

The A edit descriptor specifies the editing of character or Hollerith
data.
The data are stored left-justified in a word and padded with
blanks to the right.
NOTE
The R edit descriptor performs the same
function
for
Hollerith
data,
only
it
stores
the
data
right-justified in a word with leading nulls.
The R
edit descriptor
is not supported for character data.
For a description of the R edit descriptor,
see
Section 12.4.14.
The form of the A edit descriptor is:
[r]A[w]
where:
r

is an optional, unsigned, integer constant specifying the
number of consecutive A fields being defined.
The default
is one A field.

is an optional, unsigned, integer constant specifying the
total
number of characters in the field being defined.
The
default for single-precision values
is 5 characters;
the
default
for double-precision and complex values is 10
characters.

Depending on the
I/O operation,
the A edit descriptor
transfers
character or Hollerith data into or from a variable in an I/O list.
A
list variable may be of any type.
For example,
READ (6,5) V
5 FORMAT (A4)
causes four character or Hollerith characters to be read from
and stored in the variable v.

unit

The A descriptor deals with variables containing left-justified,
blank-filled characters.
The following list summarizes the result of
character or Hollerith data
transfer
(both internal and external
representations)
using
the A descriptor.
These explanations assume
that w represents the field width and m represents the total number of
characters possible
in the variable.
Double-precision and complex
variables contain 10 characters (m=lO);
integer,
real,
and logical
variables contain 5 (m=5).

12-39

FORMATTED DATA TRANSFERS
A Descriptor
1.

INPUT, where w > m -- The rightmost m characters of the field
are read in and stored in the corresponding variable.

INPUT, where w < m -- All w characters are read in and stored
left-justified
and
blank-filled
in
the corresponding
var iable.

OUTPUT,
where w > m
m characters
are
output
and
right-justified
in the field.
The remainder of the field is
blank-filled.

OUTPUT, where w < m -- The
leftmost
corresponding variable are output.

characters

the

R
Editing

12.4.14

R Editing

The R edit descriptor specifies the editing of Hollerith data.
The
Hollerith data are stored right-justified with leading nulls.
The R
edit descriptor is not supported for character data.
NOTE
The A edit descriptor, described in Section 12.4.13,
performs the same function as the R descriptor except
that
it left-justifies the data
in storage with
trailing spaces.
The form of the R edit descriptor is:
[r]R[w]
where:
r

is an optional,
unsigned,
nonzero,
integer
constant
specifying the number of consecutive R fields being defined.
The default is one R field.

is an optional, unsigned,
nonzero,
integer constant that
specifies the total number of characters in the R field.
If w is not specified,
for
output,
the value
for
a
single-precision field is 5 (the default), and the value for
a double-precision or complex field is 10.
For
input,
the
data
is scanned until a blank, comma, or character illegal
for the R edit descriptor is encountered.

12-40

FORMATTED DATA TRANSFERS
The R descriptor deals with variables containing
right-justified,
zero-filled characters.
The following list summarizes the result of
Hollerith data transfer (both internal and external
representations)
using
the R descriptor.
These explanations assume that w represents
the field width and m represents the
total number of characters
possible
in
the variable.
Double-precision and complex variables
contain 10 characters (m=lO)i integer,
real,
and logical variables
contain 5 (m=5).
NOTE
When more than five characters are stored, bit zero of
the
low-order
word
is
skipped.
Thus,
a
double-precision or complex variable
filled by an
R-format data transfer is of the form:
, 0

da ta

I I
'0

da ta

R Descriptor
1.

INPUT, where w > m -- The rightmost m characters of the field
are read in and stored in the corresponding variable.

INPUT, where w < m -- All w characters are read in and stored
right-justified
and
zero-filled
in
the corresponding
variable.

m characters
are
output
and
OUTPUT, where w > m
in the field.
The remainder of the field is
right-justified
zero-filled.

OUTPUT, where w < m -- The
rightmost
corresponding variable are output.

characters

the

List-Directed
Formatting

12.5

LIST-DIRECTED FORMATTING

The use of an asterisk in a data transfer statement in place of a
FORMAT statement label specifies list-directed formatting.
For this
type of formatting,
the
type of each transferred data
item
is
specified by the types of respective elements in the I/O list.
List-directed input data transfers are performed without
regard
for
column,
card, or line boundaries.
List-directed output transfers
produce records with a maximum length of 72 characters
(the default)
or the length specified by the RECL specifier (see Section 11.3.27 for
devices other than the terminal).
Otherwise, the maximum length of
the current terminal width is used.
The following is a sample list-directed data transfer statement:
READ (5,*)I,IAB,M,L

12-41

FORMATTED DATA TRANSFERS
You may use list-directed transfers to read data from any acceptable
input
device,
including
a
terminal.
However,
do not use
device-positioning commands in conjunction with list-directed data
transfers.
If you do, the results are unpredictable.
Data for list-directed transfers should consist of alternate constants
and
delimiters.
The constants used should have the following
characteristics:
1.

Input constants must be of a form acceptable to FORTRAN.

Character constants must be enclosed within single quotes,
for example, 'ABLE'.
Each apostrophe in a character constant
must be represented by two apostrophes.

The end of a record is equivalent to a blank except when
it
occurs in a character constant.
In this case, the end of the
record is ignored and the character constant
is continued
with the next record.
The first character of the continued
record must be blank, which is ignored.

If the string of a character constant exceeds the length of
the data
item,
the string is truncated.
If the string is
shorter than the data item, the string is left-justified and
remaining character positions are blank filled.

Blanks are used as delimiters
in list-directed
input.
not permitted
in any
Embedded
blanks
are,
therefore,
list-directed data item,
with the exception of character
constants.

Decimal points may be omitted from real constants that do not
have a fractional part.
In this case, it is assumed that the
decimal point follows the rightmost digit of a real constant.

Complex constants must be enclosed within parentheses.

Octal constants must be preceded with a double quote (").

A numeric data
item can correspond only to a numeric
constant,
and a character data item can correspond only to a
character constant.

A delimiter in a list-directed list of data items separates one data
item from another.
Delimiters in data for list-directed input must
comply with the following:
1.

Delimiters may be commas, blanks, or slashes.

Delimiters may be either preceded by or
followed by any
number of blanks, carriage return/line feed characters, tabs,
or line terminators; any such combination is treated as a
single delimiter.

A null
item
(the complete absence of a data
item)
is
represented
by
two
consecutive
commas that have no
intervening constant(s).
You may place any number of blanks,
tabs,
or carriage return/line
feed characters between the
commas of a null item.
Each time you specify a null item in
the
input data,
its corresponding list element is skipped
(unchanged) •

12-42

FORMATTED DATA TRANSFERS
The following illustrates the effect of the input of
item:
I/O List

A,B,IAB,N

Data input

101,'A' ,,20

null

Resulting contents of I/O list items:

A
B
lAB
N

101.
'A'
unchanged
20

Slashes (/) cause the current input operation to
terminate
even if all
the items of the I/O list are not filled.
The
contents of items of the I/O list that either are skipped (by
null
items)
or have not received an input data item before
the transfer is terminated remain unchanged.
Once the
I/O
list of the data transfer statement is satisfied, the use of
the / delimiter is optional.

Once the I/O list has been satisfied
(values have been
transferred to each item of the list), any items remaining in
the input record are skipped.

Constants or null
items
in data
for
list-directed
assigned a repeat count so that an item is repeated.

input

may

A constant with a repeat count is written as:
r*K
where:
is an integer constant that specifies the number of
times the constant is repeated, the asterisk delimits
the repeat count from the constant,
and K represents
the constant.

A null item with a repeat count is written as an
specifies the repeat count, followed by an asterisk.

integer,

The following are examples of constants and null items:

10*5
3*'ABLE'
3*

represents 5,5,5,5,5,5,5,5,5,5
represents 'ABLE' ,'ABLE' ,'ABLE'
represents null,null,null
NOTE

The asterisk form representing nulls must be delimited
by a comma or slash; in this case spaces are ignored
and not treated as delimiters.

12-43

which

FORMATTED DATA TRANSFERS

t2.6

NAMELIST-STATEMENT FORMATTING

The data transfer statements described in Chapter 10 usually include
~n
I/O list,
which
is a list of variable, array, or array element
names that identify the names of the data being transferred.
An alternative way of creating
I/O lists
is to
use the NAMELIST
statement.
Using
this method,
you can specify the I/O list in a
NAMELIST statement and
then
reference
the list by name
in the
appropriate data transfer statement.
When
you
use
NAMELIST-statement
formatting,
as
opposed
to
fORMAT-statement or list-directed formatting, you need only reference
an I/O list by NAMELIST name in a data transfer statement.

[

"---------------"-""]
NAMELIST
Statement

----"-.------.-~.~-

12.7

..".,- ..

,~-,--

-.---~~-

NAMELIST STATEMENT

The form of the NAMELIST statement is:
NAMELI ST /name/ list [/name/l i st] •••
where:
name

is the name of the NAMELIST I/O list.
This is the name
referenced
in data transfer statements.
Each NAMELIST
name must be enclosed in slashes.

1 i st

is the list of items comprising the NAMELIST I/O list.
[terns within
the
list may be variable names or array
names.
Separate multiple list items with commas.
Each list of a NAMELIST statement
is
identified and
referenced by the name immediately preceding the list.

The following is an example of creating two NAMELIST I/O lists
the names TABLE and SUMS.

having

DIMENSION C(2,4) ,TOTAL(lO)
NAMELIST/TABLE/A,B,C/SUMS/TOTAL
In this example, the name TABLE identifies the list consisting of the
scalars A and B and the array C, and the name SUMS identifies the list
consisting of the array TOTAL.

12-44

FORMATTED DATA TRANSFERS
Once a list has been defined in a NAMELIST statement, one or more
statements may reference its name.

I/O

The rules for structuring a NAMELIST statement are:
1.

You may use a maximum of six characters for a NAMELIST name.

You must begin the list name with an alphabetic character.

You must enclose the NAMELIST name in slashes.

You should use NAMELIST names
program.

You may define a NAMELIST name only once, and you must define
it by a NAMELIST statement.
Once defined, you may use the
name only in I/O transfer statements.

You must define the NAMELIST name before
statements in which it is used.

You must define any dimensioned variable contained
in a
NAMELIST
statement
in
an array declaration statement
preceding the NAMELIST statement.

12.7.1

that

are

unique

the

within

data

the

transfer

NAMELIST-Controlled Data Input Transfer

During data input transfers
in which a
NAMELIST-defined name
is
referenced,
records are read until a record is found that begins with
a blank, then $ (dollar sign), and then the desired NAMELIST name.
The dollar sign must be the second character in the record; the first
character in the record must be a blank.
NOTE
You may use "&" instead of "$" in
input.

NAMELIST-controlled

Data items of records to be input (read) using NAMELIST-defined
must be separated by commas and may be of the following form:

lists

V=Kl,K2, ••• ,Kn
where:

may be a variable, array, or array element name.

Kl, ••• ,Kn

are constants.
A series of
identical constants
may be represented as a single constant preceded
by a repetition count (5*5 represents
5,5,5,5,5).
You can specify more than one constant only if V
is an array.
If V is a scalar, then you may have
only Kl.

The input data is always converted to the type of the
list variable
when there is a conflict of types.
A character constant is truncated
from the right, or extended on the right with blanks, if necessary, to
yield a constant of the same length as the variable, array, or
substring.

12-45

FORMATTED DATA TRANSFERS
The input operation continues until another $ symbol is detected.
If
variables appear
in the NAMELIST record that do not appear in the
NAMELIST list, an error condition will occur.
A character constant must have delimiting apostrophes.
If
an
apostrophe
is part of a character constant, it must be represented by
two consecutive apostrophes, which must be contained in the same
record
(one apostrophe cannot end a record, and the other apostrophe
start a record).
For example, assume:

A is a 2-dimensional real array

B is a I-dimensional integer array

C is an integer variable

D is a character variable of length 5.

The program contains the NAMELIST declaration:
NAMELIST /FRED/ A,B,C,D

The input data is as follows:
~$FRED

A{7,2)=4, B=3,6*2.8, C=3.32, D='RON'$

A READ statement referring to
result in the following:

the

NAMELIST-defined

name

FRED

The integer 4 will be converted to floating point and
in A{7 ,2) •

2•

Th e in t e g e r 3 will be pIa c ed in B ( 1) •

The integer 2 (after
B(2) ,B(3) , ••• ,B(7).

The floating point number 3.32
integer 3 and placed in C.

The character string 'RON~~' will be placed in D.

12.7.2

being

truncated)
will

will
be

will
placed

placed

converted

in
the

NAMELIST-Controlled Data Output Transfers

When a WRITE statement refers to a NAMELIST-defined name, all
variables and arrays and their values belonging to the named list are
written out, each according to
its type.
Character constants are
written with delimiting apostrophes.
Arrays are written out by
columns.
Output data is written so that:
1.

The fields for the data will be large enough to
the significant digits.

The output can be read by an input
NAMELIST-defined list.

12-46

statement

contain

all

referencing

FORMATTED DATA TRANSFERS
For example, if ARRAY is a 2 X 3 real array, Al is a real variable, Kl
is an integer variable, and D is a character variable containing the
five characters AB'CD, the statements:
REAL ARRAY(2,3)
CHARACTER D*5
DATA ARRAY,Al,KI,D/-6.75, 0.234E-04, 680.0, -17.8,0.0,00,
1
73 • 1, 3, 'AB' I CD' /
NAMELIST/NAMl/ARRAY,Al,Kl,D
WRITE (u,NAM1)
generate the following form of output:

I01umn 1
9f$NAMl
ARRAY=
Al=
$END

-6.750000,

0.2340000E-04,

680.0000,

-17.80000,

2*0.0000000,

73.10000, Kl= 3, D='AB' 'CD I
NOTE
Do not use device-positioning commands
such
as
BACKSPACE or SKIPRECORD with NAMELIST-controlled I/O
operations.
If you do, the results are unpredictable.

12-47

CHAPTER 13
FUNCTIONS AND SUBROUTINES

Procedures you use repeatedly in a program can be written once and
then referenced each time you need the procedure.
Procedures that may
be referenced are either contained within the program in which they
are referenced, or self-contained executable procedures that can be
compiled separately. The kinds of procedures that can be referenced
are:
1.

Intrinsic functions (FORTRAN-defined functions)

Statement functions

External functions

Subroutines

The first three of these categories are referred to collectively as
functions or function procedures; procedures of the last category are
referred to as subroutines or subroutine procedures.
Intrinsic functions perform a predefined computation with a specific
number and type of arguments. These functions are provided by FORTRAN
(see Section 13.1).
Statement functions are user-defined, single statement procedures that
resemble
assignment statements.
The appearance of a statement
function reference in an
expression
causes
the
user-defined
computation to be performed (see Section 13.2).
External functions are separate program units that generally compute a
single value using one or more parameters. There are two types of
external functions available:
user-defined and FORTRAN-supplied.
A
user-defined external function is defined with a FUNCTION statement.
Both types of external functions are invoked by including a function
reference in an expression (see Section 13.3).
Subroutines are external program units that are used to perform
multiple computations or alter variables.
There are two types of
subroutines
available:
user-defined
and
FORTRAN-supplied.
User-defined subroutines are defined with a SUBROUTINE statement (see
Section 13.4.2.1). Both types of subroutines are invoked with a CALL
statement (see Section 13.4.2.2).

13.1

INTRINSIC FUNCTIONS

Intrinsic functions are supplied with the FORTRAN software.
Each
intrinsic function performs a predefined computation. There are two
types of intrinsic functions:
specific and generic.
13-1

FUNCTIONS AND SUBROUTINES
Specific functions have an implicitly defined data type.
Each
specific function requires arguments of a particular type and returns
results of a predefined type.
The IMPLICIT statement cannot be used
to change the type of a specific intrinsic function.
The data type of the return value of a generic function is determined
by the data type of its arguments. The FORTRAN generic functions are:
ABS
ACOS
AINT
ALOG
ALOGIO
AMAXI
AMINI
ANINT
ASIN
ATAN
ATAN2
CMPLX
COS
COSH
DBLE

DIM
EXP
INT
LOG
LOGIO
MAX
MIN
MOD
NINT
REAL
SIGN
SIN
SINH
SQRT
TAN
TANH
NOTE

Table 13-1 lists all the specific
and
generic
intrinsic functions.
For ease of identification, each
generic function name in the table is indicated by an
asterisk.

13.1.1

Using an Intrinsic Function

An intrinsic function is used in a FORTRAN expression by referencing
the name of the function in an expression.
For example, the following
program contains two intrinsic functions: ABS (returns the absolute
value of the argument)
and SQRT (returns the square root of the
argument) •
PROGRAM TEST

Y = -64
A = ABS (Y)
TYPE ,A
B = SQRT CA)
TYPE ,B
END
When the preceding program is executed, variable Y is assigned the
value -64.
The ABS function in the second expression calculates the
absolute value of -64.
Next,
in the third expression,
the SQRT
function calculates the square root of the absolute value of Y, which
is A. The square root of A is assigned to B in the third expression.
Executing the program yields the following results:

13-2

FUNCTIONS AND SUBROUTINES
EXECUTE TEST. FOR
FORTRAN: TEST
TEST
LINK:
Loading
[LNKXCT TEST execution]
64.00000
8.000000
CPU time 0.1 Elapsed time 4.0
The following example contains specific and generic functions.
In the
example,
the generic function SQRT is used to find the square root of
the double-precision value 64.0. Next, the specific function DSQRT is
used to find the square root of the double-precision value 64.0.
If
the argument supplied to DSQRT was not a double-precision number, a
fatal compilation error would result.
PROGRAM TESFUN
DOUBLE PRECISION A,B,AR,BR
REAL C,CR
A
B
C
C
C

64.00DO
64.00DO
64.00

GENERIC SQRT RETURNS DP
SQRT BECAUSE ARG TYPE IS DP
AR = SQRT(A)

C
C
C

SPECIFIC DSQRT PERFORMS THE
SAME FUNCTION WHEN GIVEN A DP
ARGUMENT
BR = DSQRT(A)

C
C

SPECIFIC SQRT RETURNS A REAL
VALUE RESULT
CR = SQRT(C)
TYPE , AR,BR,CR
END

Executing the program above yields the following results:
EXE TESFUN
LINK:
Loading
[LNKXCT TESFUN Execution]
8.0000000000000000, 8.0000000000000000, 8.000000
CPU time 0.1
Elapsed time 0.7
Table 13-1 lists the FORTRAN intrinsic functions.
This table gives
function definitions, argument and function types, and ranges of
acceptable values. Each function contains a description of the range
for valid arguments(s) and the range within which the function returns
valid results.
If function arguments do not fall within the specified
range, the result of the function is undefined.
For more information on the precision and accuracy of the FORTRAN
intrinsic functions, refer to the TOPS-IO/TOPS-20 Common Math Library
Manual.

13-3

FUNCTIONS AND SUBROUTINES
Table 13-1: FORTRAN Intrinsic Functions
Name

Definition

Argument
Type

Argument Restrictions

Function
Type

Result
Range

Exponential
EXP*

y = e**x

Real

-89.415 .LE. x
.LE. 88.029

Real

y .GT. 0

DEXP**

y= e**x

Double

D-floating:
-89.415 .LE. x
.LE. 88.029

Double

D-floating:
y .GT. 0

G--floating:
-710.475 .LE. x
.LE. 709.089
CEXP

w = e**z

Complex

-89.415 .LE. REALCz)
.LE.88.029
IAIMAG(z)1 .LE. 36394.429

G-floating:
y .GT. 0
Complex

All COMPLEX Numbers

Logarithm (LOG, LOG10=Generic Functions)
ALOG*

y = log(x) [base e)

Real

x .GT. 0

Real

-89.415 .LE. y .LE. 88.029

DLOG**

y = log(x) [base e)

Double

D-floating:
x .GT. 0

Double

D-floating:
-89.415 .LE. y .LE. 88.029

G--floating:
x .GT. 0
CLOG

w = log(z) [base e)

Complex

z .NE. (0,0)

G-floating:
-710.475 .LE. y .LE. 709.089
Complex

-89.415 .LE. REAL(w) .LE. 88.029
-PI .LT. AIMAG(w) .LE. PI

ALOGlO*

y = log(x) [base 10]

DLOGIO** y = log(x) [base 10]

Real

x .GT. 0

Real

-38.832 .LE. y .LE. 38.230

Double

D-floating:
x .GT. 0

Double

D-floating:
-38.832 .LE. y .LE. 38.320

G-floating:
x .GT. 0

G-floating:
-308.555 .LE. y .LE. 307.953

Square Root
SQRT*

y = SQRT(x) = x**112

Real

x .GE. 0

Real

y .GE. 0

DSQRT**

y = SQRT(x) = x**112

Double

D-floating:
x .GE. 0

Double

D-floating:
y .GE. 0

G-floating:
x .GE. 0
CSQRT

w = SQRT(z) = z**112

Complex

Any COMPLEX Number

13-4

G-floating:
y .GE. 0
Complex

All COMPLEX Numbers
REAL(w) .GE. 0

FUNCTIONS AND SUBROUTINES

Table 13-1: FORTRAN Intrinsic Functions (Cont.)
Name

Definition

Argument
Type

Argument Restrictions

Function
Type

Result
Range

Trigonometric
SIN*

y = sin(x)

Real

Ixl .LE. 210828714

Real

-1 .LE. y .LE. 1

SIND

y = sin(x) (degrees)

Real

Ixl .LE. 47185919

Real

--1 .LE. y .LE. 1

DSIN**

y = sin(x)

Double

D-floating:
Ixl .LE. 6746518852

Double

D-floating:
-1 .LE. y .LE. 1
G-floating:
-1 .LE. y .LE. 1

G-floating:
Ixl .LE. 1686629713
CSIN

w = sin(z)

Complex

IREAL(z)1 .LE. 210828714
IAIMAG(z)1 .LE. 88.895

Complex

All COMPLEX Numbers

COS*

y = cos(x)

Real

Ixl .LT. 210828714

Real

-1 .LT. y .LE. 1

COSD

y = cos(x) (degrees)

Real

Ixl .LT. 47185919

Real

-1 .LT. y .LE. 1

DCOS**

y = cos(x)

Double

D-floating:
Ixl .LT. 6746518852

Double

D-floating:
-1 .LE. y .LE. 1
G-floating:
-1 .LE. y .LE. 1

G-floating:
Ixl .LT. 1686629713
CCOS

w = cos(z)

Complex

IREAL(z)1 .LE. 210828714
IAIMAG(z)1 .LE. 88.895

Complex

All COMPLEX Numbers

TAN*

y = tan(x)

Real

Ixl .LE. 36396

Real

All REAL Numbers

DTAN**

y = tan(x)

Real

D-floating:
Ixl .LE. 3373259426

Double

D-floating:
All D-FLOATING Numbers
G-floating:
All G-FLOATING Numbers

G-floating:
Ixl .LE. 843314856
COTAN

y = cot(x)

DCOTAN** y = cot(x)

Real

Ixl .LE. 36396

Real

All REAL Numbers

Double

D--floating:
Ixl .LE. 3373259426

Double

D-floating:
All D-FLOATING Numbers

G-floating:
Ixl .LE. 843314856

13-5

G-floating:
All G-FLOATING Numbers

FUNCTIONS AND SUBROUTINES
Table 13-1: FORTRAN Intrinsic Functions (Cont.)
Name

Definition

Argument
Type

Argument Restrictions

Function
Type

Result
Range

Inverse
Trigonometric
ASIN*

y = arcsin(x)

Real

-1 .LE. x .LE. 1

Real

-PI/2 .LE. y .LE. PII2

DASIN**

y = arcsin(x)

Double

D-floating:
-1 .LE. x .LE. 1

Double

D-floating:
-PI/2 .LE. y .LE. PI/2

G-floating:
-1 .LE. x .LE. 1

G-floating:
-PI/2 .LE. y .LE. PII2

ACOS*

y = arccos(x)

Real

-1 .LE. x .LE. 1

Real

DACOS**

y = arccos(x)

Double

D-floating:
-1 .LE. x .LE. 1

Double

y = arctan(x)

DATAN** y = arctan(x)

G-floating:

o .LE. y .LE. PI

Real

x = any
REAL Numbers

Real

-P1/2 .LE. y .LE. PII2

Double

D-floating:
x = any D-FLOATING
Numbers

Double

D-floating:
-PI/2 .LE. y .LE. PII2

G-floating:
x = any G-FLOATING
Numbers
ATAN2*

y = arctan(argllarg2)

DATAN2** y = arctan(argllarg2)

D-floating:

o .LE. y .LE. PI

G-floating:
-1 .LE. x .LE. 1
ATAN*

o .LE. y .LE. PI

G-floating:
-PII2 .LE. y .LE. PII2

Real

arg1,arg2 = any
REAL Numbers

Real

-PI .LE. y .LE. PI

Double

D-floating:
argl,arg2 = any
D-FLOATING Numbers

Double

D-floating:
-PI .LE. y .LE. PI

G-floating:
argl,arg2 = any
G-FLOATING Numbers

G-floating:
-PI .LE. y .LE. PI

Hyperbolic
SINH*

y = sinh(x)

Real

Ixl .LE. 88.722

Real

All REAL Numbers

DSINH**

y = sinh(x)

Double

D-floating:
Ixl .LE. 88.722

Double

D-floating:
All D-FLOATING
Numbers

G-floating:
Ixl .LE. 709.782

G-floating:
All G-FLOATING
Numbers

COSH*

y = cosh(x)

Real

Ixl .LE. 88.722

Double

y .GE. 1

DCOSH**

y = cosh(x)

Double

D-floating:
Ixl .LE. 88.722

Double

D-floating:
y .GE. 1

G-floating:
Ixl .LE. 709.782
TANH*

y = tanh(x)

DTANH** y = tanh(x)

G-floating:
y.GE.l

Real

Any REAL Numbers

Real

-1 .LE. y .LE. 1

Double

D-floating:
Any D-FLOATING
Numbers

Double

D-floating:
-1 .LE. y .LE. 1

G-floating:
Any G-FLOATING
Numbers

13-6

G-floating:
-1 .LE. y .LE. 1

FUNCTIONS AND SUBROUTINES
Table 13-1: FORTRAN Intrinsic Functions (Cont.)
Name

Definition

Argument
Type

Argument Restrictions

Result
Range

Function
Type

Absolute Value
ABS*

y = Ixl

Real

Any REAL Numbers

Real

y .GE. 0

lABS

y = Iii

Integer

Any INTEGER Numbers

Integer

y .GE. 0

DABS**

y = Ixl

Double

D-floating:
Any D-FLOATING
Numbers

Double

D-floating:
y .GE. 0

G-floating:
Any G-FLOATING
Numbers
CABS

y = Izl

Complex

Any COMPLEX Numbers

G-floating:
y .GE. 0
Real

y .GE. 0

Truncation
AINT*

Sign of arg *
largest integer
.LT.largl

Real

Any REAL Numbers

Real

All REAL Numbers

INT*

Sign of arg *
largest integer
.LT·largl

Real

Any REAL Numbers

Integer

All INTEGER Numbers

IDINT

Sign of arg *
largest integer
.LT·largl

Double

Any DOUBLE
PRECISION Numbers

Integer

All INTEGER Numbers

DlNT**

Sign of arg*
largest integer
.LT·largl

Double

D-floating:
Any D-FLOATING
Numbers

Double

D-floating:
All D-FLOATING
Numbers

G-floating:
Any G-FLOATING
Numbers

G-floating:
All G-FLOATING
Numbers

Nearest Whole Number
ANINT*

y = int(x + .5) if
x .GE. 0 else
y = int(x -.5)

Real

Any REAL Numbers

Real

All REAL Numbers

DNINT**

y = int(x + .5) if
x .GE. 0 else
y = int(x -.5)

Double

D-floating:
Any D-FLOATING
Numbers

Double

D-floating:
All D-FLOATING
Numbers

G-floating:
Any G-FLOATING
Numbers

G-floating:
All G-FLOATING
Numbers

Nearest Integer
NINT*

y = int(x + .5) if
x .GE. 0 else
y = int(x -.5)

Real

x .LE. (2**35)-1
x .GE. -(2**35)

Integer

All INTEGER Numbers

IDNINT

y = int(x + .5) if
x .GE. 0 else
y = int(x -.5)

Double

x .LE. (2**35)-1
x .GE. -(2**35)

Integer

All INTEGER Numbers

13-7

FUNCTIONS AND SUBROUTINES

Table 13-1: FORTRAN Intrinsic Functions (Cont.)
Argument
Type

Argument Restrictions

Remainder when
argi is divided
byarg2

Real

arg2 .NE. 0

Real

o .LE. y .LT. arg2

MOD*

Remainder when
argi is divided
byarg2

Integer

arg2 .NE. 0

Integer

o .LE. y .LT. arg2

DMOD**

Remainder when
argi is divided
byarg2

Double

D-floating:
arg2 .NE. 0

Double

Name

Definition

AMOD

Function
Type

Result
Range

Remaindering

G-floating:
arg2 .NE. 0

D-floating:

o .LE. y .LT. arg2
G-floating

o .LE. y .LT. arg2

Maximum Value (MAX = Generic Function)
AMAXO

Argument with
greatest value

Integer

Any INTEGER Numbers

Real

All REAL Numbers

AMAXI*

Argument with
greatest value

Real

Any REAL Numbers

Real

All REAL Numbers

MAXO

Argument with
greatest value

Integer

Any INTEGER Numbers

Integer

All INTEGER Numbers

MAXI

Argument with
greatest value

Real

Any REAL Numbers

Integer

All INTEGER Numbers

DMAXl**

Argument with
greatest value

Double

D-floating:
Any D-FLOATING
Numbers

Double

D-floating:
All D-FLOATING:
Numbers

G-floating:
Any G-FLOATING
Numbers

G-floating:
All G-FLOATING
Numbers

Minimum Value (MIN = Generic Function)
AMINO

Argument with
least value

Integer

Any INTEGER Numbers

Real

All REAL Numbers

AMINI*

Argument with
least value

Real

Any REAL Numbers

Real

All REAL Numbers

MINO

Argument with
least value

Integer

Any INTEGER Numbers

Integer

All INTEGER Numbers

MINI

Argument with
least value

Real

Any REAL Numbers

Integer

All INTEGER Numbers

DMINl**

Argument with
least value

Double

D-floating:
Any D-FLOATING
Numbers

Double

D-floating:
All D-FLOATING
Numbers

G-floating:
Any G-FLOATING
Numbers

13-8

G-floating:
All G-FLOATING
Numbers

FUNCTIONS AND SUBROUTINES
Table 13-1: FORTRAN Intrinsic Functions (Cont.)
Name

Definition

Argument
Type

Argument Restrictions

Function
Type

Result
Range

Transfer of Sign
SIGN*

If arg2 .GE. 0
then largll
else -Iargll

Real

Any REAL Numbers

Real

All REAL Numbers

ISIGN

If arg2 .GE. 0
then larg11
else -Iargll

Integer

Any INTEGER Numbers

Real

All INTEGER Numbers

DSIGN**

If arg2 .GE. 0

Double

D-floating:
Any D-FLOATING
Numbers

Double

D-floating:
All D-FLOATING
Numbers

then larg11
else -larg11

G-floating:
Any G-FLOATING
Numbers

G-floating:
All G-FLOATING
Numbers

Positive Difference
DIM*

If argl .GT. arg2
then arg1 - arg2
else 0

Real

Any REAL Numbers

Real

y .GE. 0

IDIM

If argl .GT. arg2
then arg1 - arg2
else 0

Integer

Any INTEGER Numbers

Integer

y .GE. 0

DDIM**

If arg1 .GT. arg2
then arg1 - arg2
else 0

Double

D-floating:
Any D-FLOATING
Numbers

Double

D-floating:
y .GE. 0

G-floating:
Any G-FLOATING
Numbers

G-floating:
y .GE. 0

Double Precision Product
DPROD

arg1*arg2

Real

Any REAL Numbers

CONJG

arg = x + i*y,
CONJG = x - i*y

Complex

REAL*

arg = x + i*y
returns x

AIMAG

Double

ALL DOUBLE PRECISION
Numbers

Any COMPLEX Numbers

Complex

All COMPLEX Numbers

Complex

Any COMPLEX Numbers

Real

All REAL Numbers

arg = x + i*y
returns y

Complex

Any COMPLEX Numbers

Real

All REAL Numbers

CMPLX*

Returns
arg1 + i*arg2

Real

Any REAL Numbers

Complex

All COMPLEX Numbers

DFLOAT

Integer to double
precision

Integer

Any INTEGER Numbers

Double

Iyl .LT. 2**35

DBLE*

Real to double
precision

Real

Any REAL Numbers

Double

All DOUBLE PRECISION
Numbers

SNGL

Double precision
to real

Double

Any DOUBLE
PRECISION Numbers

Real

All REAL Numbers

FLOAT

Integer to real

Integer

Any INTEGER Numbers

Real

IYI .LT. 2**35

IFIX

Real to integer

Real

Ixl .LT. 2**35

Integer

All INTEGER Numbers

ICHAR

Character to
Integer

Character

First character of
character value

Integer

o .LE. y .LE. 127

CHAR

Integer to
Character

Integer

o .LE. y .LE. 127

Character

All Single Character

Conversion Routines

13-9

FUNCTIONS AND SUBROUTINES
Table 13-1: FORTRAN Intrinsic Functions (Cont.)

Name

Definition

Argument
Type

Argument Restrictions

Function
Type

Result
Range

Length
LEN

Length of
character entity

Character

Any CHARACTER Value

Integer

y .GE. 1

Integer

y .GE. 0

Index of a Substring
INDEX

Return location of
arg2 within argl
if not found
return 0

Character

Size of character
string argl must
be larger than
or equal to arg2
Character Comparisons

LGE

argl .GE. arg2

Character

Any CHARACTER Value

Logical

.TRUE. or .FALSE.

LGT

argl .GT. arg2

Character

Any CHARACTER Value

Logical

.TRUE. or .FALSE.

LLE

argl .LE. arg2

Character

Any CHARACTER Value

Logical

.TRUE. or .FALSE.

LLT

argl .LT. arg2

Character

Any CHARACTER Value

Logical

.TRUE. or .FALSE.

Bit Manipulation
lAND

Performs a logical
AND on corresponding
bits of argl and arg2

Integer

argl,arg2 = any
Integer Numbers

Integer

All Integer Numbers

lOR

Performs an inclusive
OR on corresponding
bits of argl and arg2

Integer

argl,arg2 = any
Integer Numbers

Integer

All Integer Numbers

lEaR

Performs an exclusive
OR on corresponding
bits of argl and arg2

Integer

arg1,arg2 = any
Integer Numbers

Integer

All Integer Numbers

NOT

Complements each bit
of argument

Integer

Any Integer Number

Integer

All Integer Numbers

ISHFT

Logically shifts argl
left arg2 bits if arg2
is positive; argl is
logically shifted right
arg2 bits if arg2 is
negative

Integer

argl = any
Integer Number
-36 .LE. arg2 .LE. 36

Integer

All Integer Numbers

ISH FTC

Circularly shifts (rotates) Integer
rightmost arg3 bits of
arg1 by arg2 places.
If arg2 is positive,
the rotation is to the
left; if arg2 is negative,
the rotation is to the
right

argl = any
Integer Number
-36 .LE. arg2 .LE. 36
1 .LE. arg3 .LE. 36

Integer

All Integer Numbers

13-10

FUNCTIONS AND SUBROUTINES
Table 13-1: FORTRAN Intrinsic Functions (Cont.)
Argument
Type

Argument Restrictions

Function
Type

Result
Range

Name

Definition

IBITS

Extracts bits arg2
through arg2 + arg3-1
from argl .........

Integer

argl = any
Integer Number
o .LE. arg2 .LE. 35
arg2 + arg3 .LE. 36

Integer

All Integer Numbers

IBSET

Returns the value of
argl with bit arg2 of
argl set to 1

Integer

argl = any
Integer Number
o .LE. arg2 .LE. 35

Integer

All Integer Numbers

IBCLR

Returns the value of
argl with bit arg2 of
,argl set to 0

Integer

argl = any
Integer Number
o .LE. arg2 .LE. 35

Integer

All Integer Numbers

BTEST

Returns .TRUE. if bit
arg2 of argl equals 1;
returns .FALSE. if
bit arg2 of argl is 0

Integer

argl = any
Integer Number
o .LE. arg2 .LE. 35

Logical

.TRUE. or .FALSE.

Bit Manipulation

Notes:
* = Generic function
•• = G-floating double-precision functions (KL model B only) -

are used if IGFLOATING compiler switch is

specified (see Section 16.1.3 or 16.2.3).

"'.* = See also the MVBITS Subroutine, Section 13.4.1.21
(2**35)-1 = 34359738367
-(2**35) = -34359738368

13-11

FUNCTIONS AND SUBROUTINES
13.1.2

Character Intrinsic Functions

Character
intrinsic functions are functions
that take character
arguments or return character values.
Character comparison intrinsic
functions are functions
that take character arguments and
return
logical values.
FORTRAN provides four character intrinsic functions:
1.

LEN
The LEN function
returns the
length
of
a
character
expression.
The LEN function has the following form:
LEN(arg)
where:
arg

is a character
indicates how
expression.

expression.
The value
returned
many characters there are in the

The following example illustrates the LEN function:
C
C

This subroutine reverses an entire character
string.
SUBROUTINE REVERS(S)
CHARACTER T, S*(*)

J = LEN(S)
DO 10 I=1,J/2
T = S(I:I)
S(I:I) = S(J:J)
S(J:J) = T
J = J - 1
CONTINUE
RETURN
END

INDEX
The INDEX function searches for
a
substring
(arg2)
in a
specified character string
(argl),
and,
if it finds the
substring, returns the substring's starting position.
If
arg2 occurs more than once in argl, the starting position of
the first (leftmost) occurrence is returned.
If arg2 does
not occur
in argl, the value zero is returned.
The INDEX
function has the following form:
INDEX(argl,arg2)
where:
arg1

is a character expression specifying the string
to be searched
for the substring specified by
arg2.

arg2

is a character
expression
specifying
substring that is searched for.

13-12

the

FUNCTIONS AND SUBROUTINES
The following example illustrates the INDEX function:
C
C
C
C

This subroutine places the symbol i into the
variable MARKS at places corresponding to the
beginning of all occurrences of the substring
SUB within the string S.
SUBROUTINE FINSTR(SUB,S)
CHARACTER*(*) SUB, S
CHARACTER*132 MARKS
INTEGER I,J

I = 1
MARKS
10

91
3.

, ,

J = INDEX(S(I:), SUB)
IF (J • NE. 0) THEN
I = 1 + J
MARKS(I-l:I-l) = 'i'
IF (I .LE. LEN(S)) GO TO 10
END IF
WRITE (5,91) S, MARKS
FORMAT (2(/lX,A))
END

ICHAR
The ICHAR function converts a character expression to
equivalent ASCII code and
returns the ASCII value.
ICHAR function has the following form:

its
The

ICHAR(arg)
where:
arg

is the character to be converted
to an ASCII
code.
If arg is longer than one character, only
the value of the first
character
is
returned;
the remaining characters are ignored.

CHAR
The CHAR function returns the single character whose ASCII
code
is the
integer or octal argument.
The CHAR function
has the following form:
CHAR(arg)
where:
arg

is an integer expression.

The following example illustrates the CHAR and ICHAR functions:
CHARACTER C*l
INTEGER I
C
C

Convert number between 0 and 9 in I to a character
digit
C = CHAR(I+ICHAR('O'))
END
13-13

FUNCTIONS AND SUBROUTINES
13.1.3

Character Comparison Functions

The four character comparison functions provided with FORTRAN are:
LLT, where LLT(argl,arg2) is equivalent to (arg1.LT.arg2)
LLE, where LLE(argl,arg2) is equivalent to (argl.LE.arg2)
LGT, where LGT(argl,arg2) is equivalent to (argl.GT.arg2)
LGE, where LGE(argl,arg2) is equivalent to (argl.GE.arg2)
The comparison functions have the following form:
func(argl,arg2)
where:
arg is a character expression.
The character comparison functions defined by the FORTRAN-77 standard
are guaranteed to make comparisons according to the ASCII collating
sequence, even on non-ASCII processors. On TOPS-10/20 systems, the
character comparison functions are identical to the corresponding
character relationals.
An example of the use of a character comparison function follows:
CHARACTER*lO CH2
IF (LGT(CH2,'SMITH'»

STOP

The IF statement in this example is equivalent to:
IF (CH2.GT.'SMITH') STOP

13.1.4

Bit Manipulation Functions

Intrinsic bit manipulation functions are used for manipulation of
the
bits
in
the
binary
patterns that represent integers.
Integer data
types are represented internally in binary two's complement notation.
Bit
positions
in
the
binary representation are numbered from right
(least significant bit) to left (most significant bit); the
rightmost
bit
position is numbered 0, and the leftmost bit position is numbered
35. A bit in a binary pattern has a value of 0 or 1.
The intrinsic functions lAND,
lOR,
lEaR,
and
NOT perform bitwise
operations on all the bits of their arguments.
Bit
of the result is
the result of applying the specified logical operation to bit
of the
arguments.
Bit
I
of
the
result
is
the
result of applying the
specified logical operation to bit 1 of the arguments, and so on
for
all the bits of the result.

The shift
functions
ISHFT and
ISHFTC shift binary patterns.
A
positive
shift
count
indicates a left shift, while a negative shift
count indicates a right shift.
A shift count of zero means no
shift.
ISHFT specifies a logical shift; bits shifted out of one ~nd are lost
and zeros are shifted in at the other end.
ISHFTC perform~ a circular
shift;
bits
shifted
out at one end are shifted back in at the other
end.

13-14

FUNCTIONS AND SUBROUTINES
The function IBITS and the subroutine MVBITS (see Section 13.4.1.21)
operate on bit fields.
A bit field is a contiguous group of bits
within a binary pattern.
Bit fields are specified by a
starting bit
position and a length. A bit field must be entirely contained in its
source operand.
For example, the integer 79 is represented
pattern:

the

following

binary

0 ••• 0101111
n ••• 6543210 (bit position)
where:
n

is 35 (the number of bit positions in an integer).

You can refer to the bit field contained
in bits 3 through 6 by
specifying a starting position of 3 and a length of 4.
In the above
example, the selected bit pattern would be the following:
0 ••• 000101
Negative integers are represented in two's complement notation.
integer -79 is represented by the following binary pattern:

The

1 ••• 1010001
n ••• 6543210 (bit position)
where:
n

is 35 (the number of bit positions in an integer).
NOTE
The value of bit position 35
is 1 for
a negative
all the
number and 0 for
a non-negative.
Also,
high-order bits of the pattern to
the left of the
value up to bit 35 are the same as bit 35.

IBITS and MVBITS operate on bit fields.
Both the starting position of
a bit field and its length are arguments to these intrinsics.
IBSET,
IBCLR, and BTEST operate only on one bit.
Thus, they do not require a
length argument.

13.2

STATEMENT FUNCTIONS

A statement function is a procedure specified by a
single statement
that
is similar
in form to an arithmetic, character, or logical
assignment statement. The statement function enables you to define a
single-line computation once in your program, give it a name, and have
that calculation performed each time you reference the statement
function
in the program.
A statement function is classified as a
nonexecutable statement.

13-15

FUNCTIONS AND SUBROUTINES
13.2.1

Defining a Statement Function

Statement functions have the following form:
fun ( [d [, d] ••• ] ) = e
where:
fun

is the symbolic name for
the statement function.
The
function name
follows the rules for forming symbolic names
in FORTRAN (see Chapter 4).

is an optional dummy argument.
Separate multiple dummy
arguments with commas.
(Dummy arguments are described in
Section 13.4.5.) The parentheses are still
required
if no
dummy arguments are specified.

is any type of FORTRAN expression.
The expression part of
the statement function
(to the right of the equal sign)
defines the computation performed using the dummy arguments.

The relationship between fun and e must conform to
the assignment
rules
in Sections 8.1,
8.2,
and 8.4.
Note that the type of the
expression may be different from the type of the statement function
name.
The following
functions:

rules

govern

the

formation

and

use

statement

The dummy argument list in the statement function serves only
to
indicate the order, number, and type of arguments for the
statement function.

The dummy arguments used in a statement function are local to
that statement function.
It is valid to use the same names
in multiple statement functions in the same program unit.
A
dummy argument name may also be used elsewhere in the program
unit to identify a variable of the same type,
including
its
appearance as a dummy argument in a FUNCTION, SUBROUTINE, or
ENTRY statement.

Each dummy argument in a statement function dummy argument
list must be unique; the same dummy argument cannot appear
twice in a single list.

Each variable reference in the function can be either a reference to a
dummy argument of the statement function, or a reference to a variable
that appears within the same program unit as the statement function
statement.
If a statement function dummy argument name is the same as the name of
another entity,
the appearance of that name in the expression of a
statement function statement is a reference to the statement function
dummy argument.

13.2.2

Using a Statement Function

Statement functions are used in FORTRAN programs by referencing
the
name of the statement function in an expression that is in the same
program unit as the statement function definition.
If a
character
function
is
referenced
in a
program unit,
the function
length
specified in the program unit must be an integer constant expression.
13-16

FUNCTIONS AND SUBROUTINES
For example, the following program uses a statement function
(called
PROFIT)
to determine the profit for
a product.
In the statement
function definition, PROFIT is defined as the difference between
wholesale and retail prices minus .05 sales tax.
PROGRAM STAFUN
PROFIT (A,B) = ((A - B) 100

(A * .05»

WRITE(UNIT=5,FMT=100)
FORMAT(lX,'Enter Wholesale Price and Retail Price')
ACCEPT*,WHOSAL,RETAIL

150

C = PROFIT (RETAIL,WHOSAL)

101

WRITE(UNIT=5,FMT=101) C
FORMAT(lX,'The Profit (minus sales tax)

is:

',F8.2)

END
When the program is executed, the retail
and wholesale values are
entered at the terminal.
Next, the expression at statement number 150
uses the values of RETAIL and WHOSAL to calculate the profit as
defined
in the PROFIT statement function.
A sample execution of this
program yields the following results:
EXECUTE STATE
LINK:
Loading
[LINKXCT STAFUN execution]
Enter Wholesale Price and Retail Price
31.67
45.95
The profit (minus sales tax) is:
11.98
CPU time 0.2
Elapsed time 18.5
When a FORTRAN expression that contains a statement function reference
is executed, the following happens:
1.

The actual arguments contained
reference are evaluated.

The actual arguments in the statement function are associated
with
the
dummy
arguments
in the statement function
definition.

The expression portion of the statement function is evaluated
using the actual arguments.

If necessary, the value of the expression is converted to the
type of the statement function.
Finally, the value resulting
from the expression evaluation is substituted
in
the
expression containing the statement function reference.

13.2.3

statement

function

Statement Function Restrictions

The following rules and restrictions must be
statement functions:
1.

the

adhered

when

using

The actual arguments in a function reference must agree
in
type and number with the corresponding dummy arguments in the
statement function dummy argument list.

13-17

FUNCTIONS AND SUBROUTINES
2.

An actual argument in a statement function reference can
itself be an expression; all actual arguments must be defined
when a statement function reference is evaluated.

A statement function can only be referenced
in the
unit that contains the statement function statement.

A statement function must not contain a reference to another
statement function that appears later in the program unit,
but can contain a reference to another statement function
that appears earlier in the program.

The symbolic name used to identify a statement function must
not appear as a symbolic name in any specification statement
except a type statement (to specify the type of the function)
or as the name of a common block in the same program unit.

An external function reference
(see Section 1383)
in the
expression part of a statement function statement must not
cause a dummy argument of the statement function to become
undefined or redefined.

The symbolic name of a statement function may not be an
actual argument.
It must not appear in an EXTERNAL statement
(see Section 7.6).

A statement function statement in a function subprogram (see
Section 13.3.4.) must not contain a function reference to the
name of the function subprogram or an entry name
in the
function subprogram.

An actual argument in a statement function reference can be
any expression,
including a character expression involving
concatenation of an operand whose length specification is an
asterisk in parentheses.

10.

The length specification of a character
must be an integer constant expression.

13.3

statement

program

function

EXTERNAL FUNCTIONS

An external function is a procedure that is defined externally to the
program unit that references it.
FORTRAN offers two types of external
functions:
FORTRAN-supplied and user-defined.
The FORTRAN-supplied
external functions are described in Section 13.3.1; the user-defined
functions are described in Sections 13.3.2 through 13.3.4.

13.3.1

FORTRAN-Supplied External Functions

FORTRAN-supplied external functions
are
similar
to
intrinsic
functions.
To use a FORTRAN-supplied function, you reference its name
in an expression.

13-18

FUNCTIONS AND SUBROUTINES
The following are the FORTRAN-supplied external functions:
x=DTOG (y)

returns a G-floating double-precision number
in the
range
1.47 x 10**-39
to
1.70 x
10**+38.
The argument y is aD-floating
double-precision
number.
(Also,
see the
DTOGA subroutine, Section 13.4.1.12.)

x=GTOD (y)

returns a D-floating double-precision number
in the
range 1.47 x 10**-39 to 1.70 x
10**+38.
The argument y is a G-floating
double-precision
number.
(Also,
see the
GTODA subroutine, Section 13.4.1.18.)

x=LSNGET(unit)

returns the last line number read in a
line
sequenced
file.
LSNGET returns a positive
integer if the last line has a valid line
number;
returns
zero
if the last line is a
page mark; or returns -1
if the last line
number
is invalid (such as, AAAA with bit 35
set) •
It also
returns -1
if the
file
contains no line number, or was opened with a
mode other than LINED (see Section 11.3.20).

x=RAN(O)

returns a pseudo random floating-point number
in the range of O.LT.x.LT.l.
The argument is
a dummy (not used) and may be any number.
Refer
to the related subroutines SETRAN (see
Section 13.4.1.27) and SAVRAN
(see Section
13.4.1.26) •

x=RANS(O)

returns a pseudo random floating-point number
in the
range of O.LT.x.LT.l.
RANS
is a
prime-modulus random number generator with
shuffling
capability.
It calls RAN to
generate
its.
initial
table
of
random
deviates.
Refer to related subroutins SETRAN
(see Section 12.4.1.27)
and SAVRAN
(see
Section 12.4.1.26).

y=SECNDS(x)

returns the system time
in seconds as a
single-precision, floating-point value, minus
the
value
of
its
single-precision,
floating-point argument.
The argument y is
set equal
to
the
time
in seconds since
midnight, minus the user-supplied value of x.

x=TIM2GO(O)

returns the number of seconds remaining
in
the
job's run-time limit.
The time l~mit is
set by the /TIME switch when submitting
the
batch job.
The argument
is a dummy (not
used) and may be any number.
You may also specify a
time limit for
an
interactive
job by using the SET TIME-LIMIT
command on TOPS-20, or the SET TIME command
on TOPS-lO.

FORTRAN-supplied external functions are treated in the same manner as
user-defined functions.
Implicit or explicit type declarations affect
these functions,
and no argument checking
(type or number)
is
performed at compile time.

13-19

FUNCTIONS AND SUBROUTINES
13.3.2

User-Defined External Functions

An external user-defined function is a procedure that is external
to
the program unit that references it.
The function subprogram enables
you to define a multiline function.
By referencing the name of that
function in an expression, the lines of the function are automatically
executed.
The FUNCTION statement is always the first
statement
subprogram.
The form of the FUNCTION statement is:
[type]

FUNCTION fun

([argl

function

[,arg2] ••• ])

where:
type

is an optional
type specification for
the external
function.
This may be INTEGER, REAL, DOUBLE PRECISION,
COMPLEX, LOGICAL, or CHARACTER (plus the optional size
modifier *len).
For CHARACTER, len is the length specification of the
result of the character function.
If you specify
CHARACTER*(*), the function assumes the length declared
for
it in the program unit that invokes it.
If len is
an integer constant, the value of len must agree with
the length of the function specified in the program
unit that invokes the function.
If a
length is not
specified
in a CHARACTER FUNCTION statement, a length
of one is assumed.
If you do not specify a type, the type of the
function
subprogram
name determines the data type of the
external function.

fun

is the symbolic name of
the
external
function
subprogram.
Unless the optional data type is specified
in the FUNCTION statement, the
type of the
function
name
determines
the
data type of the function
subprogram.

arg

is an optional dummy argument. Arg may be a variable
name,
array name,
or dummy procedure name.
Separate
multiple dummy arguments with commas. The parentheses
are optional if no dummy arguments are specified.

You must define the symbolic name assigned a function subprogram as a
variable name in the function.
During each execution of the function,
this variable can be redefined.
The value of the variable at the time
of execution of any RETURN statement is the value of the function.
NOTE
The RETURN statement returns control to the statement
that referenced
the function subprogram (see Section
13.4.4). Additionally,
you may desire to have a
function start executing at a statement other than the
first executable statement in the function subprogram.
The ENTRY statement (see Section 13.4.3) enables you
to define an alternate entry point
in the function
subprogram.

13-20

FUNCTIONS AND SUBROUTINES
13.3.3

Function Subprogram Restrictions

The following rules govern the structuring of a function subprogram:
1.

You can not use the symbolic name of a function subprogram in
any nonexecutable statement in the subprogram except in the
initial FUNCTION statement or a type statement.

Dummy argument names cannot appear
in any EQUIVALENCE,
COMMON, or DATA statement used within the subprogram.

The function subprogram can define or redefine one or more of
its arguments so as to return results in addition to the
value of the function.

The function subprogram can contain any FORTRAN statement
except BLOCK DATA, SUBROUTINE, PROGRAM, or another FUNCTION
statement.

The function subprogram should contain at least one RETURN
statement and must be terminated by an END statement.
The
RETURN statement signifies a logical conclusion of the
computation made by the subprogram, and returns the computed
function value and control
to the calling program.
A
subprogram can have more than one RETURN statement.
The END statement specifies the
subprogram and implies a return.

physical

end

the

If the type of a
function
is specified
in a FUNCTION
statement, the function name must not appear
in a type
statement.
Note that a name must not have
its
type
explicitly specified more than once in a program unit.

A function specified in a subprogram may be referenced within
any other procedure subprogram or the main program of the
executable program. A function subprogram must not reference
itself, either directly or indirectly.

If the name of a function subprogram is of type character,
each entry name
in the function subprogram must be of type
character.
If the name of the
function subprogram or any
entry in the subprogram has a length of (*) declared, all
such items must have a length of (*) declared; otherwise, all
such items must have a length specification of the same
integer value.

13.3.4

Using a Function Subprogram

After defining a function subprogram, you use it by referencing the
name
of
the
function subprogram in an expression.
Function
subprograms are referenced in expressions using the following form:
fun ([argl,arg2, ••• argn])
where:
fun

is the function subprogram name. This is the same name
that appears in the corresponding FUNCTION statement.
The length of the character function
in a character
function reference must be the same as the length of
the character function in the referenced function.
13-21

FUNCTIONS AND SUBROUTINES
arg

is an optional list of actual arguments.
These
arguments must agree in type and number with the dummy
argument list of the corresponding FUNCTION statement.
If the actual and dummy arguments do not agree, no type
conversion is done; and the results are unpredictable.
The
parentheses
are required even if no actual
arguments are specified.

When an expression that contains a function
executed, the following happens:

subprogram

reference

Evaluation of actual arguments that are expressions

Association of actual arguments with the corresponding
arguments

Execution of the referenced function subprogram

dummy

Section 13.3.3 describes all the restrictions that must be adhered
when using function subprograms.

13.4

SUBROUTINES

A subroutine
units that
user-defined
described in
described in

subprogram is a procedure that is external to the program
reference it.
FORTRAN offers two types of subroutines:
and FORTRAN-supplied.
FORTRAN-supplied subroutines are
Sections 13.4.1 - 13.4.1.32; user-defined subroutines are
Sections 13.4.2 through 13.4.2.4.
NOTE

The FORTRAN-supplied subroutines are described in this
manual
in two sections.
Section 13.4.1 describes the
general FORTRAN subroutines; Appendix F describes the
subroutines that enable you to use a plotter.
Program units reference subroutines with the CALL statement
(see
Section 13.4.2.2).
The subroutine reference in the CALL statement
contains the unique name of the subroutine, as defined in a SUBROUTINE
statement
(see Section 13.4.2.1). The SUBROUTINE statement is always
the first statement in a subroutine.
When a CALL statement reference is made to a subroutine subprogram,
program execution transfers from that CALL statement to the referenced
subroutine subprogram.
By including the ENTRY statement (see Section 13.4.3) within the body
of a subroutine subprogram, you can enter the subroutine at a point
other than the first statement in the subroutine.
In this case,
the
CALL statement used to reference an entry point in a subroutine
contains a reference to an entry point name, as opposed to the
subroutine name.
Return of program control from the subroutine to the calling program
unit occurs when the RETURN statement is executed
(see Section
13.4.4). The RETURN statement is always the last statement executed
in a subroutine subprogram.

13-22

FUNCTIONS AND SUBROUTINES
Table 13-2:

FORTRAN-Supplied Subroutines
Form

Function

Section

CALL ALCCHR(size)

Allocates space for dynamic character concatenation operations.

13.4.1.1

dpres = CDABS(dparg)

Returns the double-precision absolute value of
the specified double-precision complex number.

13.4.1.2

CALL CDCOS(dparg,dpres)

Finds the complex cosine of the specified
double-precision complex number.

13.4.1.3

CALL CDEXP(dparg,dpres)

Finds the complex exponential of the specified
double-precision complex number.

13.4.1.4

CALL CDLOG(dparg,dpres)

Returns the complex logarithm of a specified
double-precision complex number.

13.4.1.5

CALL CDSIN(dparg,dpres)

Returns the complex sine of the doubleprecision complex number specified.

13.4.1.6

CALL CDSQRT(dparg,dpres)

Returns the complex square root of the double
precision complex number specified.

13.4.1.7

CALL CHKDIV(unitvar)

Returns the number of the unit to which error
messages are being written.

13.4.1.8

CALL CLRFMT(arrayname)

Discards the FORMAT statement saved by the
execution of the SAVFMT subroutine.

13.4.1.9

CALL DATE(name)

Places the current data, left-justified, in a character variable.

13.4.1.10

CALL DIVERT(un)

Enables you to redirect error messages from
the current device to an open file on a specified
device.

13.4.1.11

CALL DTOGA(sname,dname,n)

Converts elements of double-precision arrays
from D--floating double-precision format to
G-floating double-precision format.

13.4.1.12

CALL DUMP(LB1,UB1,formatl

Causes specified portions of memory to be
dumped to the line printer (LPT:).

13.4.1.13

CALL ERRSET(n)
CALL ERRSET(n,1)
CALL ERRSET(n,1,subr)

Controls the output of arithmetic error messages during program execution.

13.4.1.14

CALL ERRSNS(l)
CALL ERRSNSO,J)
CALL ERRSNS(l,J,MSG)

Determines the reason for an error trapped by
ERR = on an OPEN, CLOSE, or data transfer
operation.

13.4.1.15

CALL EXIT

Terminates the program and returns control to
the monitor.

13.4.1.16

CALL FFUNIT

Returns the number of the first available
FORTRAN logical unit.

13.4.1.17

CALL GTODA(sname,dname,n)

Converts elements of double-precision arrays
from G-floating double-precision format to
D-floating double-precision format.

13.4.1.18

CALL ILL

Sets the ILLEG flag.

13.4.1.19

CALL LEGAL

Clears the ILLEG flag.

13.4.1.20

CALL MVBITS(m,l,len,n,j)

Transfers a bit field from one storage location
to a second storage location.

13.4.1.21

CALL OVERFUIANS)

Returns information about overflow, underflow, and divide check.

13.4.1.22

CALL PDUMP(LB1,UB1,formatl

Functions exactly like DUMP subroutine except that control returns to the calling program
after the dump has been executed.

13.4.1.23

1... ,LBn,UBn,formatn1)

QUIETX

Suppresses all summary type out when the pro-

13.4.1.24

I ... ,LBn, UBn,formatn 1)

gram
L-_ _ _ _ _ _ _ _ _ _ _L -_
_terminates.
_ _ _ _ _ _ _ _ _ _ _ _ _--JL..-_ _ _.___ .. _.

13-23

FUNCTIONS AND SUBROUTINES
Table 13-2: FORTRAN-Supplied Subroutines (Cont.)
.-------'--------,,-------------,._------------'-.--------..
Section
Function
Form
CALL SA VFMT(namel ,arraysize j)

Directs FOROTS to encode format specifications contained in the specified character variable or array.

13.4.1.25

CALL SAVRAN(n)

Saves the last internal integer seed value generated by the RAN function.

13.4.1.26

CALL SETRAN(n)

Specifies the internal integer seed value for the
RAN function,

13.4.1.27

CALL SORT('sort string')

Sorts one or more files using the SORT
program,

13.4.1.28

CALL SRTINHn)

Directs FOROTS to start allocating memory
from top downward to account for large overlay
programs and preallocates pages 600:677 for
SORT.

13.4.1.29

CALL TIME(x}
CALL TIME(x,y)

Returns the current time of day in left-justified
ASCII.

13.4.1.30

CALL TOPMEM(n)

Directs FOROTS to start allocating memory
from top downward to account for large overlay
programs.

13.4.1.31

CALL TRACE

Generates a list of active subprograms on the
terminal.

13.4.1.32

13.4.1

FORTRAN-Supplied Subroutines

The FORTRAN software includes a set of predefined subroutines.
This
section describes the general FORTRAN subroutines (and a function that
is similar)
in alphabetical order.
(See Appendix
F
for
the
FORTRAN-supplied plotter subroutines.)
NOTE
sections
13.4.1.2
through
13.4.1.7
describe
subroutines
(and a
function)
that are used
for
calculations on double-precision complex numbers.
You
must
supply. your own subroutines
for
performing
addition, subtraction, multiplication, and division of
double-precision complex numbers.
In
addition,
FORTRAN
does
not
support
the
double-precision
complex
data
type
(called
COMPLEX*16).
These numbers are
kept as arrays and
cannot be used in expressions or as the arguments of
generic routines.

ALCCHR
Subroutine

13.4.1.1 ALCCHR Subroutine - The ALCCHR subroutine allocates space
for dynamic character concatenation operations.
You do not normally
need to allocate space for this purpose unless you are doing very
large character concatenation operations.
13-24

FUNCTIONS AND SUBROUTINES
The form of the ALCCHR subroutine is:
CALL ALCCHR(size)
where:
size

is the integer size in characters for
or expanding the character stack.

either

creating

COABS
Function

Function - The
COABS
function
returns
the
13.4.1.2 COABS
double-precision absolute value of the specified double-precision
(Although COABS is a
function and
is not
a
complex number.
subroutine,
it
is included here because
it
is similar to some
subroutines.)
The form of the CDABS function is:
dpres

= COABS(dparg)

where:
dparg

is a 2-element double-precision array containing the
complex value whose absolute value you want calculated.
The first element of dparg contains the real part of
the double-precision complex number; the second element
contains the imaginary part.

dpres

is a double-precision variable that
absolute value of the complex number.

set

the

COCOS
Subroutine

13.4.1.3 COCOS Subroutine - The COCOS subroutine
finds
cosine of the specified double-precision complex number.

the

complex

The form of the COCOS subroutine is:
CALL COCOS(dparg,dpres)
where:
dparg

is a 2-element double-precision array containing the
complex value whose cosine you want calculated.
The
first element of dparg contains the real part of the
double-precision complex number;
the second element
contains the imaginary part.
13-25

FUNCTIONS AND SUBROUTINES
dpres

is a 2-element double-precision array in which the
subroutine returns the result of the calculation. The
first element of dpres contains the real part of the
double-precision complex number;
the second element
contains the imaginary part.

Example:
DOUBLE PRECISION dparg(2) ,dpres(2)
dparg(l) = IDa
!arg is (1,-1)
dparg(2) = -IDa
CALL CDCOS(dparg,dpres)

COEXP
Subroutine

13.4.1.4 CDEXP Subroutine - The CDEXP subroutine finds the
exponential of the specified double-precision complex number.

complex

The form of the COEXP subroutine is:
CALL CDEXP(dparg,dpres)
where:
dparg

is a 2-element double-precision array that contains the
complex argument to the subroutine.
The first element
of dparg contains the real part of the double-precision
complex
number;
the second element contains the
imaginary part.

dpres

is a 2-element double-precision array that stores the
result of the calculation.
The first element of dpres
stores the real part of the result; the second element
stores the imaginary part.

Example:
DOUBLE PRECISION dparg(2) ,dpres(2)
dparg(l) = 000
dparg(2) = 100
!arg is (0,1)
CALL CDEXP(dparg,dpres)

COLOG
Subroutine

13.4.1.5 COLOG Subroutine - The COLOG subroutine returns the
logarithm of a specified double-precision complex number.

13-26

complex

FUNCTIONS AND SUBROUTINES

The form of the CDLOG subroutine is:
CALL CDLOG(dparg,dpres)
where:
dparg

is a 2-element double-precision array that contains the
double-precision complex number whose logarithm you
want calculated.
The first element of dparg contains
the real part of the complex number; the second element
contains the imaginary part.

dpres

is a 2-element double-precision array that stores the
result returned by CDLOG.
The first element of dpres
contains the real part of the double-precision complex
number; the second element contains the imaginary part.

Example:
DOUBLE PRECISION dparg(2) ,dpres(2)
dparg(l) = 100
!arg is (1,0)
dparg(2) = 000
CALL CDLOG(dparg,dpres)

CDSIN~

Subroutine
1 - -_ _ _ _ __

13.4.1.6 CDSIN Subroutine - The CDSIN subroutine returns the
sine of the double-precision complex number specified.

complex

The form of the CDSIN subroutine is:
CALL CDSIN(dparg,dpres)
where:
dparg

is a 2-element double-precision array that contains the
number whose sine you want calculated.
The first
element of dparg contains the real part of
the
double-precision complex number;
the second element
contains the imaginary part.

dpres

is a 2-element double-precision array in which the
result of the calculation
is returned.
The first
element of dpres contains the real part of the result;
the second element contains the imaginary part.

Example:
DOUBLE PRECISION dparg(2) ,dpres(2)
dparg(l) = -lDO
!arg is (-1,01)
dparg(2) = IDa
CALL CDSIN(dparg,dpres)

13-27

FUNCTIONS AND SUBROUTINES

CDSQRT
Subroutine

13.4.1.7 CDSQRT Subroutine - The CDSQRT subroutine
returns
the
complex square root of the specified double-precision complex number.
The form of the CDSQRT subroutine is:
CALL CDSQRT(dparg,dpres)
where:
dparg

is a 2-element double-precision array that contains the
double-precision complex number whose square root you
want calculated.
The first element of dparg contains
the real part of the double-precision complex number;
the second element contains the imaginary part.

dpres

is a 2-element double-precision array that contains the
result of the calculation.
The first element of dpres
contains the real part of the complex square root;
the
second element contains the imaginary part.

Example:
DOUBLE PRECISION dparg(2) ,dpres(2)
dparg(l) = 1000
dpres(2) = -1000
!arg is (10,-10)
CALL CDSQRT(dparg,dpres)

CHKDIV
Subroutine

13.4.1.8 CHKDIV Subroutine - The CHKDIV subroutine returns the number
of the
unit to which error messages are being written.
This
subroutine returns the value -1 if the messages are being sent to the
terminal.
The form of the CHKDIV subroutine is:
CALL CHKDIV(unitvar)
where:
unitvar

is the variable in which the unit number is stored.

13-28

FUNCTIONS AND SUBROUTINES

13.4.1.9 CLRFMT Subroutine - The CLRFMT subroutine discards
the
encoded
form of the FORMAT statement saved by the execution of the
SAVFMT subroutine (see Section 13.4.1.25).
The form of the CLRFMT subroutine is:
CALL CLRFMT(arrayname)
where:
is the name of the array that contains the encoded
form of the FORMAT specifications saved by the
SAVFMT subroutine.

arrayname

13.4.1.10 DATE Subroutine - The DATE subroutine stores the current
date as a
left-justified ASCII string in a character variable.
The
date is in the form:
dd-mmm-yyb
where:
dd

is a 2-digit day (if the first digit is 0, it
is converted
to a blank), mmm is a 3-letter month abbreviation (such as,
Jan,Feb) , yy is a 2-letter year, and b is a blank.
The data
is stored in ASCII, left-justified, and blank filled.

The form of the DATE subroutine is:
CALL DATE(name)
where:
name

is the name of a character variable. The date returned
by the subroutine is stored in this variable.
NOTE

For
compatibility
with
previous
versions
of
FORTRAN-IO/20, you can specify a numeric array name as
the argument for the DATE subroutine.
The current
date
is stored as a left-justified ASCII string in a
2-word array or data
item
(double-precision
or
complex).
The array must have at least two elements.

13-29

FUNCTIONS AND SUBROUTINES

DIVERT
Subroutine

]

'-----_.

13.4.1.11 DIVERT Subroutine - The DIVERT subroutine enables you to
redirect error messages from the current device to an open file on a
specified device.
The form of the DIVERT subroutine is:
CALL DIVERT(un)
where:
un

is the logical device number of the file on which
file resides.

DTOGA
Subroutine

the

open

' - - - - - - -

13.4.1.12 DTOGA - The
DTOGA
subroutine
converts
elements
of
double-precision arrays fromD-floating double-precision format to
G-floating double-precision format.
The form of the DTOGA subroutine is:
CALL DTOGA (sname,dname,n)
where:
sname

is the name of the source array.

dname

is the name of the destination array.

is the number of elements to convert.

(See Section 13.4.1.18 for the GTODA subroutine.)

DUMP
Subroutine

13.4.1.13 DUMP Subroutine - The DUMP subroutine causes
portions of memory to be dumped to the line printer (LPT:).

13-30

specified

FUNCTIONS AND SUBROUTINES
The form of the DUMP subroutine is:
CALL DUMP(LBl,UBl,formatl[ ••• ,LBn,UBn,formatn])
where:
LBl,UBl

are the integer values of the upper
addresses to be dumped.

and

lower

formatl

is an integer that
indicates
possible specifications are:

dump

format.

a
1
2
3

the

memory
The

octal
real
integer
ASCII

If no arguments are supplied, all of user memory is dumped
in octal.
If only the bounds arguments are specified, or if the format value is
out of range, the dump format is octal.
If only the
first
bound
argument
is specified, all locations from that address to the end of
memory are dumped.
The dump is terminated by a call to EXIT.

ERRSET
Subroutine

13.4.1.14 ERRSET Subroutine - The ERRSET subroutine
output of warning messages during program execution.

controls

the

The ERRSET subroutine has three forms:
1.

CALL ERRSET(n)

CALL ERRSET(n,i)

CALL ERRSET(n,i,subr)

where:
n

is the maximum number of error messages to type.

is the code to which error the call applies; one of the
following:
-1

1
2
3
4
5
6
7
8

any of the following
integer overflow
integer divide
input integer overflow
input floating overflow
floating overflow
floating divide check
floating underflow
input floating underflow
library routine error

13-31

FUNCTIONS AND SUBROUTINES
9
21
22
23
if i
subr

output field width too small
FORLIB warnings
nonstandard usage warnings
Bounds check warnings

is not specified, -1 is assumed

is the name of the user-defined error handling
routine
to
be invoked each time any of the above errors occur.
The effect is as if
CALL SUBR (I,IPC,N2,ITYPE,UNFIXD,FIXED)
were placed in the program just after
causing the trap.
I

the

instruction

= error number of trap, same as above

IPC
PC of
trap
instruction
(if
trapped, IPC = PC of the IOLST. call

code

N2 = second error number (reserved for Digital)

= data type of value
UNFIXD = value returned by the hardware
ITYPE

FIXED = value after fixup
val ue)

(SUBR

can

change

If SUBR is not specified, no routine is called
the APR trap.

this
on

ERRSNS
Subroutine

13.4.1.15
ERRSNS Subroutine - The ERRSNS subroutine
returns
integer
values
that describe the status (success or failure) of the last I/O
operation (see Appendix D).
This subroutine can be used to determine
the
reason
for
an error trapped by ERR= on an OPEN, CLOSE, or data
transfer operation.
Both return values are always cleared
after a
successful data transfer operation.
The forms of the ERRSNS subroutine are:
CALL ERRSNS

(I)

or
CALL ERRSNS (I,J)
or
CALL ERRSNS

FORTRAN-10/20 VII

(I,J,MSG,K)

13-32

February 1987

FUNCTIONS AND SUBROUTINES
where:
I

returns a FORTRAN-supplied number
class of failure that occurred.

that

describes

the

optionally returns
further describes
error.

MSG

If present, is a character variable used to return the
ASCII
text of the last error message.
If the variable
for MSG
is less than 80 characters,
the text
is
truncated;
if
the
variable
is greater
than 80
characters, the text
is padded
to
the right with
blanks.

a processor-specific number that
or qualifies the type of the last

NOTE
For compatibility with previous versions of
FORTRAN-10/20 ,
you can specify a numeric array
as the MSG argument.
The numeric array is used
as a l6-word array to return the ASCII text of
the last error message.
K

optionally returns the RMS STV code that
reason for the last RMS error.

explains

the

EXIT
Subroutine

13.4.1.16 EXIT Subroutine - The EXIT subroutine
terminates
the
program and returns control to the monitor.
The EXIT subroutine takes
no arguments.
The form of the EXIT subroutine is:
CALL EXIT

FFUNIT
Subroutine

13.4.1.17 FFUNIT Subroutine - The
FFUNIT subroutine returns
lowest available FORTRAN logical unit number (see Table 10-3).

the

The form of the FFUNIT subroutine is:
CALL FFUNIT (n)

FORTRAN-10/20 VII

13-33

February 1987

FUNCTIONS AND SUBROUTINES

GTODA
Subroutine

13.4.1.18 GTODA Subroutine - The GTODA subroutine converts elements
of double-precision arrays from G-f1oating double-precision format to
D-f1oating double-precision format.
The form of the GTODA subroutine is:
CALL GTODA(sname,dname,n)
where:
sname

is the source array name.

dname

is the destination array name.

is the number of array elements to convert.

(See Section 13.4.1.12 for the DTOGA subroutine.)

ILL
Subroutine

13.4.1.19 ILL Subroutine - The ILL subroutine sets the
ILLEG flag.
If this flag
is set and an
illegal character is encountered in
floating-point, double-precision input, the corr~sponding value is set
to
zero and no error message is issued.
The ILL subroutine takes no
arguments.
The ILLEG flag is not set initially.
The form of the ILL subroutine is:
CALL ILL

LEGAL
Subroutine'

ILLEG
13.4.1.20 LEGAL Subroutine - The LEGAL subroutine clears the
flag set by the
ILL subroutine.
The LEGAL subroutine takes no
arguments.
The form of the LEGAL subroutine is:
CALL LEGAL
13-34

FUNCTIONS AND SUBROUTINES

MVBITS
Subroutine

13.4.1.21 MVBITS Subroutine - The MVBITS subroutine transfers a bit
field
from one storage location
(source)
to a field in a second
storage location (destination).
The form of the MVBITS subroutine is:
CALL MVBITS(m,i,len,n,j)
where:
m

is an integer expression that represents the source location
from which a bit field is transferred.

is an integer expression that identifies the
position in the source field transferred from m.

first

len is an integer expression that identifies the length
field transferred from m.

bit
the

is an integer variable or array element that is the
destination location to which a bit field is transferred.

is an integer expression that identifies the
first
position in the destination field transferred from m.

bit

The MVBITS subroutine transfers len bits from position i
through
i+1en-l of the source location (m) to positions j through j+1en-l of
the destination location (n). Other bits of the destination location
and all of the bits of the source location remain unchanged.
The
values of i+1en and j+1en must not be greater than 36.
(See Section 13.1.4 for information on bit manipulation functions.)

~_____O_V_E_R_F_L________]
~

Subroutine

13.4.1.22 OVERFL
Subroutine - The
OVERFL
subroutine
information about overflow, underflow, and divide check.
The form of the OVERFL subroutine is:
CALL OVERFL(IANS)

13-35

returns

FUNCTIONS AND SUBROUTINES
where:
fANS

is an integer variable whose value specifies whether an
overflow, underflow, or divide check has occurred since
the last call to OVERFL.
The values returned are:
1 = at least one overflow,
check occurred.
2

underflow,

divide

none occurred.

13.4.1.23 PDUMP Subroutine - The PDUMP subroutine functions exactly
like
the DUMP subroutine (see Section 13.4.1.13) except that control
returns to the calling program after the dump has been executed.
The form of the PDUMP subroutine is:
CALL PDUMP(LBl,UBl,formatl[ ••. ,LBn,UBn,formatn)
where:
LBl,UBl

are the integer values of the upper
addresses to be dumped.

and

lower

formatl

is an integer that
indicates
possible specifications are:

dump

format.

o
1

2
3

the

memory
The

octal
real
integer
ASCII

aUIETX
Subroutine

13.4.1.24 QUIETX Subroutine - The QUIETX subroutine suppresses all
summary typeout when the program terminates, including library error
summaries and CPU times.
The QUIETX subroutine takes no arguments.
The form of the QUIETX subroutine is:
CALL QUIETX

13-36

FUNCTIONS AND SUBROUTINES

13.4.1.25 SAVFMT Subroutine - The SAVFMT subroutine directs FOROTS to
encode the FORMAT specifications contained in the specified character
variable or array, and to save the encoded form.
This action improves
the performance of any future
I/O statements which that character
variable or array is the format identifier.
FOROTS saves the encoded form of the format until a call
to CLRFMT
(see Section 13.4.1.9)
is executed
for that variable or array, or
until another call to SAVFMT is executed for that variable or array.
NOTE
After a call to SAVFMT, you must not change the value
of the variable or array.
If the value is changed,
the new value may be ignored.
A call to SAVFMT with a
variable or array whose address is identical to that
in a previous call, does an implied call to CLRFMT.
The form of the SAVFMT subroutine is:
CALL SAVFMT(name[,arraysize])
where:
name

is the name of the character variable or array
that contains the FORMAT descriptors that you want
encoded.

arraysize

is the number of array elements
specified.

array

NOTE
For
compatibility
with
previous
versions
of
FORTRAN-IO/20, you can specify a numeric array name as
the argument for the SAVFMT subroutine.

-----.'"--.----

_-_ - ...
...

SAVRAN
Subroutine

13.4.1.26
internal
function,
time
it
to SETRAN

SAVRAN Subroutine - The SAVRAN subroutine saves the last
integer seed value generated by the RAN function.
The RAN
described in Section 13.3.1, returns a
random number
each
is called.
This value can be used at a later time in a call
to reestablish the same random number sequence.

13-37

FUNCTIONS AND SUBROUTINES
The form of the SAVRAN subroutine is:
CALL SAVRAN(n)
where:
n

is an integer variable
into which the SAVRAN subroutine
stores the last internal integer seed value generated.

SETRAN
Subroutine

13.4.1.27 SETRAN Subroutine - The SETRAN subroutine specifies the
internal
integer seed value
for
the RAN function.
If the SETRAN
argument is zero, RAN uses its own default starting value.
The form of the SETRAN subroutine is:
CALL SETRAN(n)
where:
n

is a nonnegative integer constant
2**31) •

variable

(less

than

SORT
Subroutine

13.4.1.28 SORT Subroutine - The SORT subroutine sorts one or more
files
using
the SORT program.
You can successfully use this
subroutine only if the SORT software has been
installed
(see
the
FORTRAN Installation Guide).
The SORT software is sold as a separate
pr-od-uc-E and may not be available at your installation.
The form of the SORT subroutine is:
CALL SORT('sort string')
where:
sort string

is a command line containing file specifications
and SORT switches.
For specific information about
the SORT command line, see the SORT/MERGE User's
Guide.

13-38

FUNCTIONS AND SUBROUTINES
NOTE
The sort string must be compatible with
the current version of the standalone
SORT.
Therefore, the string is not the
same for TOPS-IO and TOPS-20
(see the
SORT/MERGE User's Guide) •

SRTINI

·-----l

Subroutine

13.4.1.29 SRTINI Subroutine - The SRTINI subroutine directs FOROTS to
start allocating memory from top downward to account for large overlay
programs and preallocates pages ~OO:677 (octal) for SORT.
The form of the SRTINI routine is:
CALL SRTINI(n)
where:
n

is top page number to use when allocating memory.

Note that FOROTS will not allocate pages (including DDT pages)
have been marked as unavailable at memory initialization.

that

TIME
Subroutine

13.4.1.30 TIME Subroutine - The TIME subroutine returns
time of day in left-justified ASCII.

the

current

The TIME subroutine has two forms:
CALL TIME(x)
or
CALL TIME(x,y)
where:
x

is a character variable.
In the one argument form,
TIME
returns the time in x in the form:
hh:mm, where hh is the
hour (24-hour time) and mm is the minutes.

is a character variable. When the two argument form of the
TIME subroutine is used, the forms of the time returned in x
is the same as the one argument form, and the value returned
in y has the form:
bss.t, where b is a blank, S5 is the
current seconds, and t is the current tenths of seconds.
13-39

FUNCTIONS AND SUBROUTINES
NOTE
For
compatibility
with
previous
versions
of
FORTRAN-IO/20,
you can specify a numeric variable or
array element as an argument of the TIME subroutine.
The following example demonstrates using the one and the two
forms of the TIME subroutine in a program.

argument

PROGRAM TIMTST
CHARACTER*lO X,Y
CALL TIME(X,Y)
WRITE(UNIT=5,FMT=101)X,Y
CALL TIME(X)
L02

WRITE (UNIT=5, FMT=102)X
FORMAT(lX,'THE ONE ARGUMENT TIME RETURNS:

',A)

END
EXECUTE TIMTST
LINK:
Loading
[LNKXCT TIMTST execution]
THE TWO ARGUMENT TIME RETURNS: 09:00 20.9
THE ONE ARGUMENT TIME RETURNS: 09:00
CPU time 0.1
Elapsed time 0.2

--]

TOPMEM
Subroutine

13.4.1.31 TOPMEM Subroutine - The TOPMEM subroutine directs FOROTS to
start allocating memory from a
specified page number downward to
account for large overlay programs.
The form of the TOPMEM subroutine is:
CALL TOPMEM(n)
where:
n

is the top page number to use in allocating memory_

Note that FOROTS will not allocate pages (including DDT pages)
have been marked as unavailable at memory initialization.

13-40

that

FUNCTIONS AND SUBROUTINES

TRACE
Subroutine
------_..__._------- j

13.4.1.32 TRACE Subroutine-- The TRACE subroutine generates a list of
active subprograms on the terminal.
An active subprogram is one that
has been called but has not yet returned.
The main program is always
active.
The trace listing starts at the currently active routine (the
one containing the call to TRACE)
and proceeds back to
the main
program.
The form of the TRACE subroutine is:
CALL TRACE
The information produced by the TRACE routine consists
subprogram:

of,

The name of the routine

The address of the routine (in octal)

The address of the subprogram call
+ offset)

The address of the subprogram call
offset with the calling routine)

The number of arguments passed to the called routine

The types of the arguments passed including:

for

each

(expressed as routine-name
(expressed

label

C D F G I K L -

Character string descriptor
D-floating double precision
Real
G-floating double precision
Integer or double integer
Literal string
Logical
a - Octal or double octal
S - Statement label
U - Unknown argument type
X - Complex

If there are too many arguments to display, the
contain ' ••• ' .

'types'

column

will

If local
symbols are
loaded with the program,
the
label+offset
information will be much more informative.
A label of the form 12345P
refers to FORTRAN statement number 12345; a label of the
form 56789L
refers to
line number 56789
in the compiler listing.
Line number
labels only appear if the program was compiled with /DEBUG:LABELS (see
Chapter 16).
The traceback listing is sent to
the error-message unit,
which is
normally the
terminal.
You can use the DIVERT subroutine (Section
13.4.1.11) to change where the listing is sent.

13-41

FUNCTIONS AND SUBROUTINES
Example:

10
20

PROGRAM MAIN
TYPE 10
FORMAT (I' Calling 8UB1:')
CALL 5UB1
TYPE 20
FORMAT (I' Calling 5UB2:')
CALL 5UB2 (A,B)
END
SUBROUTINE SUB1
Y = F(X)
END
SUBROUTINE 5UB2 (G,H)
CALL 5UB3 (G,H,I)
END
SUBROUTINE 5UB3
CALL TRACE
END

(Al,A2,A3)

FUNCTION F(Y)
CALL 5UB2 (Y,O)
F=2.
END
EXECUTE TRC.FOR
Load ing
LINK:
(LNKXCT TRC execut ion]
Calling 5UB1:
Name
TRACE.
SUB3
5UB2
F
8UB1

(Loc)
(426373)
(256 )
(232)
(307)
(214)

«--«--«--«--«--«---

Caller
5UB3+2
5UB2+6
F+20
5UB1+2
MAIN.+7

(Loc)
(5UB3+2)
(5U82+6)
(F+20)
(5U8l+2)
(MAIN.+7)

Args
0
3
2
1
0

Types

«--«--«--«---

Caller
5UB3+2
5UB2+6
MAIN.+14

(Loc)
(5U83+2)
(8U82+6)
(MAIN.+14)

Args

Types

FFI
FF

FFI
FI
F

Calling 8UB2:
Name
TRACE.
5UB3
SUB2

( Loc)
(426373)
(256 )
(232)

CPU time 0.4

13.4.2

o
2

Elapsed time 3.1

User-Defined Subroutines

A subroutine subprogram is a separate ptogram unit.
The FORTRAN CALL
statement
is used in a program unit to call a subroutine subprogram.
The CALL statement contains the name of the subroutine to which
control passes when the CALL statement is executed.
The CALL
statement can also optionally contain actual arguments that are passed
to the called subroutine.
The CALL statement is described in Section
13.4.2.2.

13-42

FUNCTIONS AND SUBROUTINES
The SUBROUTINE statement is always the first statement in a subroutine
subprogram.
The
SUBROUTINE
statement
defines the name and,
optionally, any dummy arguments used by the
subroutine.
The
SUBROUTINE statement is described in Section 13.4.2.1.
The ENTRY statement enables you to enter a subroutine subprogram at a
statement other than the first statement of the subroutine. The ENTRY
statement is described in Section 13.4.3.
The last logical statement of a subroutine subprogram is always the
RETURN statement.
By default, the RETURN statement always returns
control to the first executable statement in the calling program that
immediately follows the CALL statement. Optionally, you may use the
alternate return form of the RETURN statement to return control to a
statement other than the default. Both forms of the RETURN statement
are described in Section 13.4.4.

13.4.2.1 SUBROUTINE Statement - The SUBROUTINE statement is always
the first statement in a subroutine subprogram. It is used to define
the name of the subroutine and, optionally, to define dummy arguments
that are used by the subroutine.
The form of the SUBROUTINE statement is:
SUBROUTINE sub [([dl[,d2] ••• ])]
where:
sub

is the symbolic name of the subroutine or dummy procedure.

is an optional dummy argument for the subroutine subprogram.
This argument can be a variable name, an array name, a dummy
procedure name, or any combination of these separated by
commas.
The parentheses are. optional if no dummy arguments
are specified.

The following
subprogram:

rules

control

the

structuring

subroutine

You can not use the symbolic name of the subprogram in any
statement within the defined subprogram except the SUBROUTINE
statement itself. The symbolic name of a subroutine is a
global name and must not be the same as any other global name
or any local name in the program unit.

The symbolic name of a dummy argument is local to the program
unit and must not appear in an EQUIVALENCE, PARAMETER, SAVE,
INTRINSIC, COMMON, or DATA statement except as a common block
name.

The subroutine subprogram may define or redefine one or
of its dummy arguments so as to return results.

If the actual argument is a constant or expression,
subroutine must not change the value of that argument.

the

The subroutine subprogram may contain any FORTRAN statement
except BLOCK DATA, FUNCTION, PROGRAM, another SUBROUTINE
statement, or any statement that
either
directly
or
indirectly references the subroutine being defined or any of
the subprograms in the chain of subprogram references leading
to this subroutine.
13-43

FUNCTIONS AND SUBROUTINES
6.

Dummy arguments that represent statement labels may be either
an *, $, 0 r &.

The subprogram should contain at least one RETURN statement
and must be
terminated by an END statement.
The RETURN
statement indicates a logical end of the
routine;
the END
statement signifies the physical end of the subroutine.
If
no RETURN precedes the END statement,
then the RETURN
statement is implicit.

Subroutine subprograms can have as many entry points as
desired
(see description of ENTRY statement given in Section
13.4.3) •

A character dummy argument whose length specification is an
asterisk
in parentheses may appear as an operand
for
concatenation.

13.4.2.2 CALL Statement - The CALL statement is used
in a program
unit to
reference a subroutine.
Execution of the CALL statement
causes a transfer of program control to the subroutine referenced
in
the CALL statement.
The CALL statement can also contain a list of arguments that
by the computation performed in the referenced subroutine.

used

The form of the CALL statement is:
CAL L sub [( [a 1 [ , a 2] ••• ] ) ]
where:
sub

is the symbolic name of a subroutine or dummy procedure.

is an optional actual argument that is used by
the
subroutine.
The actual arguments in the CALL statement must
agree in position and type with the dummy arguments in the
referenced
SUBROUTINE
statement.
The parentheses are
optional if no actual arguments are specified.

13.4.2.3 Execution of a CALL Statement - When
executed, the following results occur:

CALL

statement

Any actual arguments in the CALL statement argument list that
are expressions are evaluated.

The actual arguments are then associated with
arguments in the referenced SUBROUTINE statement.

Control passes to the subroutine and it is executed.

Return of control
calling
program
statement.

the

dummy

from the referenced subroutine to the
completes
the execution of the CALL

A subroutine can be referenced within any other procedure or the main
program of the executable program.
A subprogram cannot, however,
reference itself, either directly or indirectly.

13-44

FUNCTIONS AND SUBROUTINES
13.4.2.4 Actual Arguments for
a Subroutine - Actual
arguments can
appear
in the CALL statement argument list.
Actual arguments must
agree
in order, number,
and type with the corresponding dummy
arguments
in the dummy argument list of the referenced subroutine.
The exception to the agreement rules between actual and dummy argument
lists is the use of a subroutine name or an alternate return specifier
as an actual argument.
Actual arguments in CALL statements can be any of the following:
1.

Any expression, including a character expression whose length
specification is an asterisk in parentheses.

An array name

An intrinsic function name

An external procedure name

A dummy procedure name

An alternate return label

An actual argument in a subroutine reference may be a dummy argument
name that appears
in a dummy argument list within the subprogram
containing the reference.
An asterisk dummy argument must not be used
as an actual argument in a subprogram reference.

13.4.3

ENTRY Statement

The ENTRY statement provides you with a method for entering a function
or subroutine subprogram at any executable statement.
The ENTRY
statement can appear anywhere within a function subprogram after
the
FUNCTION statement, or within a subroutine subprogram after the
SUBROUTINE statement. An ENTRY statement cannot appear between a DO
statement and the terminal
statement of
its DO-loop or inside a
block-IF statement.
A subprogram may optionally have one or more ENTRY statements.
ENTRY statement is classified as a nonexecutable statement.

The form of the ENTRY statement is:
ENTRY en [(dl [,d2 ••• ])]
where:
en

is the symbolic name of an entry in a function or subroutine
subprogram.
This symbolic name is called an entry name.
If
the entry name appears in a subroutine subprogram,
it
is
referred
to as a subroutine name; if the entry name appears
in a function subprogram, it is called an external
function
name.

is a variable name, array name, dummy procedure name, or the
symbols:
*, $, or & (these symbols can represent a dummy
argument which is an alternate return label).
The symbol
references
(asterisk,. ampersand,
and dollar sign)
are
permitted only when the
ENTRY statement appears in a
subroutine subprogram.

13-45

FUNCTIONS AND SUBROUTINES
If you do not specify any parentheses after the entry name,
you need
not specify any dummy arguments.
If,
however,
you include the
parentheses, you must specify at least one dummy argument.
The rules
for the use of an ENTRY statement follow:
1.

The ENTRY statement allows entry into a subprogram at a place
other than that defined by the SUBROUTINE or FUNCTION
statement.
You may include more than one ENTRY statement in
an external subprogram.

Execution begins at the first executable statement
the ENTRY statement.

Appearance of an ENTRY statement in a subprogram does not
negate the rule that statement functions in subprograms must
precede the first executable statement.

ENTRY statements are nonexecutable
execution flow of a subprogram.

affect

the

You can not use an ENTRY statement in a main program or
a subprogram reference itself through its entry points.

have

You can not use an ENTRY statement
statement construction.

The dummy arguments in the ENTRY statement need not agree
in
order, number, or type with the dummy arguments in SUBROUTINE
or FUNCTION statements or any other ENTRY statement in the
subprogram.
However, the arguments for each call or function
reference must agree with the dummy arguments in
the
SUBROUTINE, FUNCTION, or ENTRY statement that is referenced.

Entry into a subprogram initializes only the dummy
of the referenced ENTRY statement.

You can not reference a dummy argument unless it appears in
the
dummy list of the ENTRY,
SUBROUTINE, or FUNCTION
statement by which the subprogram is entered.

10.

The source subprogram must be ordered such that references to
dummy
arguments
in
executable
statements follow the
appearance of the dummy argument
in the dummy list of a
SUBROUTINE, FUNCTION, or ENTRY statement.

11.

Dummy arguments that were defined for a subprogram by some
previous reference to the subprogram are undefined for
subsequent entry into the subprogram.

12.

The value of a function
current entry name.

13.

If an entry name
in a
function subprogram is of type
character, each entry name and the name of the function
subprogram must be of type character.
If the name of the
function subprogram or any entry in the subprogram has a
length of (*) declared, all such items must have a
length
specification of the same integer value.

must

13-46

and

the

returned

not

following

range

use

arguments

the

FUNCTIONS AND SUBROUTINES
13.4.4

RETURN Statement

The RETURN statement returns control to the referencing
and may appear only in a function subprogram or
subprogram.

prog ram un i t
a subroutine

The RETURN statement has two possible forms, depending on whether it
is the last statement in a function subprogram or a subroutine
subprogram.
The form of the RETURN statement in a function subprogram is:
RETURN
The form of the RETURN statement in a subroutine subprogram is:
RETURN [e]
where:
e

is an integer constant, variable, or expression. This form
of the RETURN statement is called an alternate return. The
alternate return form enables you to select any labeled
statement in the calling program unit as a return point
after execution of the program unit in which the alternate
RETURN statement appears.
The value of e should be a positive integer that is equal to
or less than the number of statement labels given in the
argument list of the calling statement. If e is less than 1
or is larger than the number of available statement labels,
a standard return is performed.
(A
standard
return
transfers control back to the first executable statement
immediately following the calling statement in the calling
prog ram un it) •
NOTE
A dummy argument for a statement label must be either
an asterisk (*), a dollar sign ($), or an ampersand
(&) •

You may use more than one RETURN (standard return)
subprogram.
The use of the alternate return
statement is restricted to subroutine subprograms.

13-47

statement in any
form of the RETURN

FUNCTIONS AND SUBROUTINES
For example,
program:

assume

the

following

statement

sequence

main

Each occurrence of RETURN returns control to the statement GO
in the calling program.

101

CALL EXAMP(I,*10,K,*15,M,*20)
GO TO 101

alternate return #1

alternate return #2

alternate return #3

END
SUBROUTINE EXAMP (L, *,M, *,N,*)

RETURN

RETURN(I+J)

END

If, on the execution of the RETURN(I+J)
is:

statement, the value of

Value

The following is performed:

less than one

a standard
return to
the GO TO
statement is made
the return is made to statement 10
the return is made to statement 15
the return is made to statement 20
a stahdard
return to
the GO TO
statement is made.

1
2
3
Greater than 3

13-48

(I+J)

101

FUNCTIONS AND SUBROUTINES
13.4.5

Dummy and Actual Arguments

Since you may reference subprograms at more than one point throughout
a program, many of the values used by the subprogram may be changed
each time it is used. Dummy arguments in subprograms represent the
actual values to be used, which are passed to the subprogram when it
is called.
For example, shown belaw is a subroutine (TEST) being called from the
maln program by a CALL statement.
In this example, the variables in
the CALL statement, A, B, and C(2), represent actual values in the
main program.
They are therefore called actual arguments.
On the other hand, the variables in the SUBROUTINE statement, R,
X,
and Z, do not represent any values until they have values passed to
them from the CALL statement.
They are therefore called dummy
arguments.
(The CALL, SUBROUTINE, and RETURN statements are described in Sections
13.4.2.2, 13.4.2.1, and 13.4.4, respectively.)
CALL TEST (A,B,C(2»

END
SUBROUTINE TEST (R,X,Z)

RETURN
END
Functions and subroutines use dummy arguments to indicate the type of
the actual arguments they represent and whether the actual arguments
are variables, arrays, subroutine names, or the names of external
functions.
Each dummy argument must be used within a function or
subroutine as if it were a variable, array, subroutine, or external
function identifier.
Dummy arguments are given in an argument list associated with the
identifier assigned to the subprogram; actual arguments are normally
given in an argument list associated with a call made to the desired
subprogram.
The position, number, and type of each dummy argument in a subprogram
list must agree with the position, number, and type of each argument
list of the subprogram reference.
NOTE
If the /DEBUG:ARGUMENTS compiler switch is specified
(see
Section
16.3) ,
optional type checking
is
performed at load time on dummy and actual arguments.

13-49

FUNCTIONS AND SUBROUTINES
Dummy arguments may be:
1.

variables

array names

subroutine identifiers

function identifiers

the symbols *, $, or & that are used to denote
of alternate return labels

the

position

When you reference a subprogram, its dummy arguments are
replaced by
the corresponding actual arguments supplied in the reference.
All
appearances of a dummy argument within a function or subroutine are
related
to the given actual arguments.
Except for
subroutine
identifiers and character constants, a valid association between dummy
and actual arguments occurs only if both are of the same type;
otherwise,
the
results of the subprogram computations will
be
unpredictable.
Argument association may be carried through more than one level of
subprogram reference if a valid association is maintained through each
level.
The dummy/actual argument associations established when a
subprogram is referenced are terminated when the desired subprogram
operations are completed.
The following rules govern the use and form of dummy arguments:
1.

The number and type of the dummy arguments of a
procedure
must be the same as the number and
type of the actual
arguments given each time the procedure is referenced.

Dummy argument names may not appear in EQUIVALENCE, DATA,
COMMON statements.

A variable dummy argument should have a variable,
an array
element
identifier,
an expression, or a constant as its
corresponding argument.

An array dummy argument should have either an array name or
an array element
identifier as
its corresponding actual
argument.
If the actual argument is an array, the length of
the dummy array should be less than or equal to that of the
actual array.
Each element of a dummy array is associated
directly with the corresponding elements of the actual array.

A dummy argument representing a subroutine identifier
have a subroutine name as its actual argument.

should

A dummy argument representing an external function must
an external function as its actual argument.

have

A dummy argument may be defined or redefined in a
referenced
subprogram only if its corresponding actual argument is a
variable.
If dummy arguments are array names, then elements
of the array may be redefined.

Additional information regarding the use of dummy and actual arguments
is given
in the description of how subprograms are defined and
referenced.

13-50

FUNCTIONS AND SUBROUTINES
13.4.5.1 Length of Character Dummy and Actual Arguments - The length
of a dummy argument of type character must not be greater than the
length of its associated actual argument. Note that if the character
dummy argument's length is specified as *(*), the length used is
exactly the length of the associated actual argument. This is known
as a passed length character argument.
The length of the dummy argument is determined each time control
transfers to the function.
The length of the actual argument can be
the length of a character variable, array element, substring, or
expression.
A character array dummy argument can have passed length.
of each element in the dummy argument is the length of the
the actual argument.
The passed length and
the array
together determine the size of the passed length character

The length
elements in
declaractor
array.

The following example of a function subprogram uses a
passed length
character argument.
The function finds the position of the character
with the highest ASCII code value; it uses the length of the passed
length character argument to control the iteration.
(Note that the
intrinsic function LEN is used
to determine the length of the
argument.
See Table 13-1 for a description of the LEN function.)

INTEGER FUNCTION ICMAX(CVAR)
CHARACTER*(*) CVAR
ICMAX = 1
DO 10 I = 2, LEN (CVAR)
IF (CVAR(I:I) .GT. CVAR(ICMAX:ICMAX»
RETURN
END

ICMAX =1

Each of the following function references specifies a different length
for the dummy argument:
CHARACTER VAR*lO, CARRAY(3,5)*20

II
12
13
14
15

ICMAX (VAR)
ICMAX(CARRAY(2,2»
ICMAX(VAR(3:8»
ICMAX(CARRAY (1,3) (5:15»
ICMAX (VAR(3:4)//CARRAY(3,5»

13.4.5.2 Character and Hollerith Constants as Actual ArgumentsActual arguments and their corresponding dummy arguments must agree in
data type.
If the actual argument is a
Hollerith constant
(for
example, 4HABCD) , the dummy argument must be of numeric data type.
In FORTRAN-IO/20, if an actual argument is a character constant
(for
example
'ABCD'),
the corresponding dummy argument can have either
numeric or character data type.
If the dummy argument has a numeric
data type, the character constant 'ABCO' is, in effect, converted to a
Hollerith constant by the FORTRAN compiler and the loader.
An exception to this occurs when the function or subroutine name
is
itself a dummy argument.
It is not possible to determine at compile
time or load time whether a character constant or Hollerith constant
is required.
In this case, a character constant actual argument can
correspond only to a character dummy argument.
13-51

CHAPTER 14
BLOCK DATA SUBPROGRAMS

Block data subprograms provide initial values for variables and
elements in named common blocks.

array

A block data subprogram must start with the BLOCK DATA statement. The
only
valid statements within a block data subprogram are the
specification and DATA statements
(COMMON, DIMENSION,
EQUIVALENCE,
IMPLICIT, PARAMETER, SAVE, type statements, and DATA statements).
The
last statement of a block data subprogram must be an END statement.
You can enter initial values into more than one labeled
in a single subprogram of this type.

common

block

An executable program can contain more than one block data subprogram.

14.1

BLOCK DATA STATEMENT

The form of the BLOCK DATA statement is:
BLOCK DATA [sub]
where:
sub

is the optional symbolic name of a block data subprogram
which the BLOCK DATA statement appears.

The name sub is a global name and must therefore be a unique
symbolic name within the executable program.
The following is an example of a block data subprogram:
BLOCK DATA TEST
COMMON /SQUARE/ CIRCLE,RECTAN,PI
DATA CIRCLE,RECTAN,PI/l.,2.,3.14159/
END
This example initializes the COMMON variables CIRCLE, RECTAN,
to 1., 2., and 3.14159 respectively.

14-1

and

CHAPTER 15
WRITING USER PROGRAMS

This chapter is a guide for writing effective programs with FORTRAN.
It contains techniques for optimization, interaction with non-FORTRAN
programs, and other useful programming hints.

15.1

GENERAL PROGRAMMING CONSIDERATIONS

The following paragraphs describe programming considerations you
should observe when preparing a FORTRAN program to be compiled by
FORTRAN.

15.1.1

Accuracy and Range of Double-Precision Numbers

Floating-point and real numbers may consist of up to 16 digits
in a
double-precision mode.
Their range is specified in Chapter 3, Section
3.2. You must be careful when testing the value of a number within
the specified range,
since,
although numbers up to 10**38 may be
represented, FORTRAN can only test numbers of up to eight significant
digits (REAL precision) and 16 significant digits (DOUBLE precision).
NOTE
For KL model B systems,
if the /GFLOATING compiler
switch is specified
(see Section 16.1.3 or 16.2.3),
double-precision numbers up to
10**307
can
be
represented.
You must also be careful when testing floating-point computations for
a
result of O.
In most cases the anticipated result, that is, 0 will
be obtained; however, in some cases the result may be a very small
number that approximates O.
Such an approximation of 0 will cause
tests for equality to 0 to fail.
To increase the accuracy of its cO~Rile-time arithmetic, the compiler
does all
floating
point arithmetic in double precision (it converts
back to single precision when necessary).
For KL model B systems,
if
the /GFLOATING compiler switch is specified (see Section 16.1.3 or
16.2.3), the compiler does its compile-time arithmetic in G-floating
double-precision.
If the /DFLOATING compiler switch is specified (the
default), the compiler does its compile-time arithmetic in D-floating
double-precision.
Compile-time arithmetic done in G-floating double-precision may not
overflow or underflow as
it might with D-floating double-precision
arithmetic. G-floating double-precision has a greater
range than
D-floating double-precision (see Section 3.4).
15-1

WRITING USER PROGRAMS
15.1.2

Writing FORTRAN Programs for Use on Other Computers

If you prepare a program to run on both TOPS-10 or TOPS-20 and another
manufacturer's computer, you should:
1.

Avoid using
any blue print language extensions
in this
manual.
(All information in blue print represents aspects of
FORTRAN that are extensions to the FORTRAN-77 Standard.)

Consider the accuracy and size of the numbers
that
manufacturer's computer is capable of handling.

another

You can use the /FLAG compiler switch
to
invoke the compatibility
flagger.
This feature provides warning messages for language elements
that are extensions
to
the
FORTRAN-77 standard
or
that
are
incompatible with VAX FORTRAN (see Section 16.6).

15.1.2.1 Remote Links - A remote network link
is established each
time an OPEN statement or
INQUIRE-by-file
for a
remote file is
performed.
The link remains open until the
file
is closed or the
INQUIRE statement completes.
The maximum number of simultaneous open
links available to a
user process
is a site-dependent monitor
parameter.

15.1.2.2 Reducing Network Overhead - A substantial part of
the
overhead associated with 'remote file access consists in establishing
the remote link.
Since a READ, READ, WRITE sequence for a file opened
for
read access causes FOROTS to close and
reopen the file for
read/write access, network overhead can be reduced by ensuring
that
remotes
files
to be written to are opened with an ACCESS specifying
both read and write access.
Likewise, file
specifications
in CLOSE
statements for
remote
files, which cause FOROTS to establish a link
for a network rename operation, should not be routinely used unless a
full remote rename is required.

15.1.3

Using Floating-Point DO Loops

FORTRAN permits you to use noninteger,
single- or double- precision
numbers as
the parameter variables in a DO statement.
This lets you
generate a wider range of values for
the DO loop index variables,
which may, in turn, be used inside the loop for computations.
WARNING
If a
noninteger
index
is used,
accumulation
of
rounding errors may lead to unexpected values for the
loop variable.

FORTRAN-10/20 VII

15-2

February 1987

WRITING USER PROGRAMS
15.1.4

Computation of DO Loop Iterations

The number of times through a DO loop is computed outside the loop and
is not affected by any changes to the DO index parameters within the
loop. The formula for the number of times a DO loop is executed is:
DO 10 I=Ml,M2,M3
Number of cycles=MAX (INT((M2-Ml+M3)/M3) ,O)
The values of the parameters Ml, M2, M3 can be of any type except
complex.
If the
iteration count is less than or equal to zero, the
body of the loop is not executed.
The
index variable retains its
assigned value (Ml).
(See Section 9.3.)
NOTE
The interpretation of the iteration count and
the
index variable described above is different from that
of earlier versions of FORTRAN-10/20.
If the /NOF77
compiler switch is specified (see Section 16.1.3 or
16.2.3), and the iteration count is less than or equal
to
zero,
the body of the loop is executed once.
In
addition, the final value of the index variable of the
DO statement is undefined after a normal exit.

15-2.1

February 1987

WRITING USER PROGRAMS

This page intentionally left blank.

15-2.2

February 1987

WRITING USER PROGRAMS
15.1.5

Subroutines - Programming Considerations

Consider the following items when preparing and executing subroutines:
1.

During execution, no check is made to see if the proper
number of arguments
is passed (unless the /DEBUG:ARGUMENTS
compiler switch. is specified, see Section 16.3).

If the number of actual arguments passed to a subroutine is
less than the number of dummy arguments specified, the values
of the unspecified arguments are undefined
(unless the
/DEBUG:ARGUMENTS compiler switch
is specified, see Section
16.3) •

If the number of actual arguments passed to a subroutine is
greater than the number of dummy arguments given, the excess
arguments are ignored (unless the /DEBUG:ARGUMENTS compiler
switch is specified, see Section 16.3).

If an actual argument is a constant and
its corresponding
dummy argument
is set to another value (an illegal usage),
all references made to the constant in the calling program
may be changed to the new value of the dummy argument.

No check is made to see if the arguments passed are of the
same type as the dummy arguments (unless the /DEBUG:ARGUMENTS
compiler switch is specified, see Section 16.3).
NOTE
An exception is that a check is always made
for G-floating and D-floating type mismatches
regardless of the /DEBUG:ARGUMENTS switch
being specified.
In addition, when you pass character data to
a subroutine or function that is compiled
with the /EXTEND switch, the calling program
must
also be compiled with the /EXTEND
switch.
If an actual parameter is a constant and the corresponding
dummy is of type real, be sure to include the decimal point
In the constant.
If the dummy is double-precision,
be sure
to specify the constant with a "D".
NOTE
You are given no warning
if any of the situations
described
In items 1,2,3,4,
or 5 occur (unless the
/DEBUG:ARGUMENTS compiler switch
is specified,
see
Section 16.3).

Examples:
If a function F has a single dummy argument of
function is called with:

F(2)

15-3

type

real,

and

that

WRITING USER PROGRAMS
F interprets the integer 2 as an unnormalized
In this instance, F(A) should be called with:

floating-point

number.

F(2.0)
Similarly, if the function Fl(D)

is called with:

Fl(2.S)
and D is double-precision, Fl assumes that its parameters have been
specified with two words of precision and picks up whatever follows
the constant 2.S in memory.
The proper method is to use:
Fl(2.SDO)

lS.1.6

Reordering of Computations

Computations that are not enclosed within parentheses may be reordered
by the compiler.
Sometimes it is necessary to use parentheses to
ensure proper results from a specific computation.
For example, assuming that:
1.

RLI represents a large number, such that RLl*RL2
an overflow condition, and

will

cause

RSI is a very small number, that is, less than 1, the program
sequence:

A=RSl*RLl*RL2
B=RS2*RL2*RLl

will not produce an overflow when evaluated left to
right,
since the first computation in each expression (that is,
RS1*RLl and RS2*RL2) will produce an interim result that
is
smaller than either large number (RLI or RL2).
However, the compiler may recognize RLl*RL2 as a common sUbexpression
(see Section lS.2.1.l) and generate the following sequence:
temp
A
B

RLl*RL2
RSl*temp
RS2*temp

The computation of temp will cause an overflow.
You should write the program as follows to
results are obtained:

ensure

that

the

desired

A=(RSl*RLl)*RL2
B=(RS2*RL2)*RLI
Computations may be reordered even when
selected.
lS-4

global

optimization

not

WRITING USER PROGRAMS
15.1.7

Dimensioning of Dummy Arrays

When you specify an array as a dummy argument to a subprogram unit,
you must
indicate to the compiler that the parameter is an array by
dimensioning the array in a specification statement.
This is the only
way the compiler is able to distinguish a reference to such an array
from a function reference.
A dummy array can be dimensioned the
following ways:
1.

Assumed size

Adjustable dimensioned

Fixed'dimension bound

Dimensioning the array with a size of 1
dangerous,
practice.
The alternative to
assumed-size arrays (see Section 7.1.2).

is a common,
although
this practice is to use

Example:
SUBROUTINE SUBl(A,B)
DIMENSION A(l)
There are disadvantages to using the above technique because it may
prevent the compiler from diagnosing illegal programs, specifically:
1.

Reading or writing the array by name
DIMENSION ARRAY (10)
READ (1) ARRAY
The above is a binary read that
ARRAY.

will

read

ten

words

into

SUBROUTINE SUBl(A)
DIMENSION A (1)
READ(l)A
This binary read will cause one word to be read into A.

Using the array as a format
SUBROUTINE SUB2(FMT)
DIMENSION FMT(l)
READ (l,FMT)
Only the first word of the format specification contained
FMT is used.

Using the /DEBUG:BOUNDS compilation switch
(see Section
16.3), the dimension information used
is that which is
specified in the array declaration
SUBROUTINE SUB3(A)
DIMENSION A(l)
A(2)=0
The reference to A(2) will cause
message to be generated.

15-5

the

out-of-bounds

warning

WRITING USER PROGRAMS
15.2

FORTRAN GLOBAL OPTIMIZATION

You have the option of
invoking the global
optimizer
during
compilation.
The optimizer treats groups of statements in the source
program as a single entity.
The purpose of the global optimizer is to
prepare a more efficient object program that produces the same results
as the original unoptimized program,
but takes significantly less
execution time.
The output of the lexical and syntactic analysis phase of the compiler
is developed into an optimized source program equivalent (in results)
to the original.
The optimized program is then processed by the
standard compiler code generation phase.

15.2.1

Optimization Techniques

15.2.1.1 Elimination of Redundant Computations - Often the
same
sUbexpression will
appear
in more than one computation throughout a
program.
If the values of the operands of such a common expression
are not changed between computations, the sUbexpression may be written
as a separate arithmetic expression.
Also, the variable
representing
its resultant may then be substituted where the sUbexpression appears.
This eliminates unnecessary recomputation of the subexpression.
For
example, the instruction sequence:
A=B*C+E*F

H=A+G-B*C

IF«B*C)-H) 10,20,30
contains the sUbexpression B*C three times when it really needs to
computed only once.
Rewriting the preceding sequence as:

T=B*C
A=T+E*F
H=A+G-T
IF(T-H) 10,20,30
eliminates two computations of the subexpression B*C from the
sequence.

overall

Decreasing the number of arithmetic operations performed in a source
program by the elimination of common subexpressions shortens the
execution time of the resulting object program.

15-6

WRITING USER PROGRAMS
15.2.1.2 Reduction of Operator Strength - The time required
to
execute arithmetic operations varies according to the operator(s)
involved. The hierarchy of arithmetic operations according to the
amount of execution time required is:
MOST TIME

OPERATOR
**

/
LEAST TIME

+,-

During program optimization,
the global optimizer replaces, where
possible
some arithmetic operations that require the most time
with operations that require less time.
For example,
consider the
following DO loop that is used to create a table for the conversion of
from 1 to 20 miles to their equivalents in feet:
10

DO 10 MILES=1,20
IFEET(MILES)=5280*MILES

The execution time of the loop wo~ld be shorter if the time-consuming
multiply operation, that is, 5280*MILES, could be replaced by a faster
operation. Since you increment MILES on each pass,
you can replace
the multiply operation by an add and total operation.
In its optimized form, the
equivalent to:

loop

would

replaced

sequence

K=5280
DO 10 MILES=1,20
IFEET(MILES)=K
K=K+5280

In the optimized form of the loop, the value of K is set to 5280 for
the first
iteration of the loop, and is increased by 5280 for each
succeeding iteration of the loop.
This situation occurs frequently
implicitly contain multiplications.

subscript

calculations

that

15.2.1.3 Removal of Constant Computation from Loops - The speed with
which
a given algorithm may be executed can be
increased
if
instructions and/or computations are moved out of frequently traversed
program sequences into less frequently traversed program sequences.
Movement of code is possible only if none of the arguments
in the
items to be moved are redefined within the code sequences from which
they are to be taken.
Computations within a
loop consisting of
variables or constants that are not changed in value within the loop
may be moved outside the loop.
Decreasing the number of computations
made within a loop greatly decreases the execution time required by
the loop.
For example, in the sequence:
10

DO 10 I=1,100
F=2.0*Q*A(I)+F

INumerical analysis considerations severely limit the
cases where this is possible.
15-7

number

WRITING USER PROGRAMS
the value of the computation 2.0*Q,
once calculated on the first
iteration, will remain unchanged during the remaining 99 iterations of
the loop.
Reforming the preceding sequence to:

QQ=2.0*Q
DO 10 1=1,100
F=QQ*A(I)+F

moves the calculation 2.0*Q outside the scope of
movement of code eliminates 99 multiply operations.

the

loop.

This

In addition, it is possible to remove entire assignment statements
from loops.
This action can be easily detected from the macro
expanded listings.
The internal sequence number
remains with the
statement and appears out of order in the leftmost column of the macro
expanded listing (LINE).

15.2.1.4 Constant Folding and Propagation - In this
method
of
optimization,
expressions containing determinate constant values are
detected and the constants are replaced, at compile time,
by their
defined or calculated value.
For example, assume that the constant PI
is defined and used in the following manner:

PI=3.14159

X=2*PI*Y

At compile time, the optimizer will have used the defined value of PI
to calculate the value of the sUbexpression 2*PI.
The optimized
sequence would then be:

PI=3.14159

X=6.28318*Y

thereby eliminating a multiply operation from the object code program.
The evaluation of constant expressions at compile time
is called
"folding";
the replacement of variables with their constant values is
called "constant propagation".
NOTE
For KL model B systems, use of the /GFLOATING compiler
switch
(see Section 16.1.3 or 16.2.3)
may affect
compile-time arithmetic.

15-8

WRITING USER PROGRAMS
15.2.1.5 Removal of Inaccessible Code - The optimizer detects and
eliminates any code within the source program that cannot be accessed.
In general, this will not happen since programmers do not normally
include such code in their programs; however, inaccessible code may
appear in a program during the debugging process.
The removal of
inaccessible code by the optimizer reduces the size of the object
program.
A warning message is generated for each inaccessible line removed.

15.2.1.6 Global Register Allocation - During the compilation of a
source program, the optimizer controls the allocation of registers to
minimize computation time
in the optimized object program.
The
allocation process is designed to minimize the number of MOVE and
MOVEM machine instructions that will appear
in the most frequently
executed portions of the code.

15.2.1.7 I/O Optimization - Every effort is made to minimize the
number of required calls to the FOROTS system.
This is done primarily
through extensive analysis of implied DO loop constructs on I/O data
transfer
statements.
The formats of these special blocks are
described in Chapter 18. These optimizations reduce the size of the
program
(argument code plus argument block size is reduced) and
greatly improve the performance of programs that use implied DO loop
I/O statements.

15.2.1.8 Uninitialized Variable Detection - A warning message may be
generated when a scalar variable is referenced before it has received
a value (only when optimizing).

15.2.1.9 Test Replacement - If the only use of a DO loop index is to
reduce operator strength (see Section 15.2.1.2) and the loop does not
contain exits (GO TOs out of the loop),
the DO loop index is not
needed and can be replaced by the reduced variable.
For example:

DO 10 1=1,10
K=K+7*I
CONTINUE

Reduction of operator strength and test replacement together transform
this loop into:

DO 10 1=7,70,7
K=K+I
CONTINUE

This situation occurs frequently in
execution of these statements, 1=11.

15-9

subscript

computation.

After

WRITING USER PROGRAMS
15.2.2

Programming Techniques for Effective Optimization

Observe the following recommendations during the coding of a FORTRAN
source program. They will improve the effectiveness of the optimizer:
1.

Do not use DO loops with an extended range.

Specify label lists when using assigned GO TOs.

Nest loops so that the innermost index is the
largest range of values.

Avoid the use of associated I/O variables.

Avoid unnecessary use of COMMON and EQUIVALENCE.

15.3

one

with

the

calls

FUNCTION SIDE EFFECTS

Unpredictable results can occur if a statement includes
different functions that modify the same variables.
Consider the following example:
COMMON A
A=5.
P=F(1.)+Q(2.)
END
FUNCTION F(X)
COMMON A
A=O.
F=X+l
END
FUNCTION Q(Y)
COMMON A
Q=A
END

In the preceding sequence, if P is evaluated by calling F before Q,
the value of P will be 2.
If P is evaluated by calling Q before F,
the value of P will be 7.

15.4

INTERACTING WITH NON-FORTRAN PROGRAMS AND FILES

15.4.1

Using The Sharable High-Segment FOROTS

If your program does not contain a FORTRAN main program module,
and
you wish to have the sharable FOROTS GETSEGed at run time, you must do
the following:

Force the loading of the FOROTS initialization routine RESET.
by declaring it as an external.

Define the symbol FOROT% as a global with a positive, nonzero
value before FORLIB.REL is searched.

15-10

WRITING USER PROGRAMS
3.

Initialize FOROTS by the appropriate initialization call:
JSP 16, RESET.
EXP 0

15.4.2

Calling Sequences

The following paragraphs describe the standard procedures for
subroutine calls.
1.

writing

Procedure
a.

The calling program must load the accumulator
(AC)
16
with the address of the first argument in the argument
list.

The subroutine is then
using AC 17.

The return will be made to the
after the PUSHJ 17 instruction.

The FOROTS trace facility requires the
to be:

called

PUSHJ

instruction

immediately

calling

sequence

XMOVEI 16,AP
PUSHJ 17,F
where AP is the pointer to the argument list and F is the
subprogram name.
The word preceding the first word of an
entry point should have its name in SIXBIT.
2.

15.4.3

Restrictions
a.

Skip returns are not permitted.

The contents of the pushdown stack located before the
address specified by AC 17 belong to the calling program;
they cannot be read by the called subprogram.

FOROTS assumes that
it has control of the
stack;
therefore,
you must not create your own stack.
The
FOROTS stack is initialized by the call to RESET.
(See
Section 15.4.1).

Accumulator Usage

The specific functions performed by accumulators (AC) 17,16,0,
are:
1.

and

Pushdown Pointer - AC 17 is always maintained as a pushdown
pointer.
In section zero, its right half points to the last
location in use on the stack, and its left half contains the
negative of the number of words allocated to the unused
remainder of the stack.
In non-zero sections,
the pushdown pointer contains
global address of the last location in use on the stack.

15-11

the

WRITING USER PROGRAMS
2.

Argument List Pointer - AC 16 is used as the argument
pointer. The called subprogram does not need to preserve its
contents.
The calling program cannot depend on getting back
the address of the argument list passed to the called
subprogram. AC 16 cannot point to the ACs or to the stack.

Temporary and Value Return Registers - AC 0 and 1 are used as
temporary registers and for
returning values.
The called
subprogram does not need to preserve the contents of AC 0 or
1
(even if not returning a value).
The calling program must
never depend on getting back the original contents of the
data passed to the called subprogram.

Returning Values - A subroutine
results by modifying arguments.

subprogram

may

pass

back

A numeric function subprogram always returns the value of the
function in AC 0 (or ACs 0-1 if the value is double precision
or complex).
A function subprogram may also pass back
additional
results by modifying the arguments.
(See Section
15.4.4 for a description of character functions.)
5.

Preserved ACs - FORTRAN function subprograms preserve
through 15; subroutine subprograms do not.

ACs

The design of the called subprogram cannot depend on the contents of
any of the ACs being set up by the calling subprogram, except for ACs
16 and 17.
Passing
information must be done explicitly by the
argument list mechanism.
Otherwise, the called subprograms cannot be
written in either FORTRAN or COBOL.

15.4.4

Argument Lists

Since the FORTRAN compiler uses the indirect bits on argument lists
(note that this permits shared, pure code argument lists), it is
essential for code that accesses parameters to take this into account.
Specifically, sequences that obtained the values of parameters through
use of operations such as:
HRRZ R,1(16)
to pick up the address of the second argument should be changed to
XMOVEI R,@1(16)
The latter operation will work when interfacing with FORTRAN.
The format of the argument list is as follows:
Arg count word
Arg list addr.---First arg entry
Second arg entry

Last arg entry
The format of the arg count word is:
bits 0-17
bits 18-35

These contain -n, where n is the number of arg
entries.
Reserved for future DIGITAL development, and must be

15-12

WRITING USER PROGRAMS
The format of an arg entry is as
word) :
bit

bits
bits
bit
bits
bits

1-8
9-12
13
14-17
18-35

follows

(each

entry

single

IFIW (Instruction Format Indirect Word) flag, must be
1.
Reserved for future DIGITAL development, must be O.
Arg type code.
Indirect bit if desired.
Index field, must be O.
Address of the argument.

For character functions, the first argument points to the return
value, which
is a character string descriptor (see Section 15.4.6).
The actual arguments to the function start in the second argument
entry.
The following restrictions should be observed:
1.

Neither the argument list nor the arguments themselves can be
on the stack.
The same restriction applies to any indirect
argument pointers.

The called program may not modify the argument list
itself.
The argument list may be in a write-protected segment.
Note that the arg count word is at position -1 with respect
to the contents of AC 16. This word is always required even
if the subroutine does not handle a variable number of
arguments.
A subroutine that has no arguments must still
provide an argument list consisting of two words,
that
is,
the argument count word with a 0 in it and a zero argument
word.

Example:
XMOVEI 16,AP
PUSHJ 17,SUB

AP:

iSET UP ARG POINTER

iCALL SUBROUTINE
iRETURN HERE

iARGUMENT LIST
-3,,0
IFIW 4,A
IFIW 4,B
IFIW 4,C

iSUBROUTINE TO SET THIRD ARG TO SUM OF FIRST TWO ARGS
SUB:

MOVE
ADD
MOVEM
POPJ

T,@0(16)
T,@1(16}
T,@2(16}
17,

15-13

iGET FIRST ARG
iADD SECOND ARG

iSET THIRD ARG
iRETURN TO CALLER

WRITING USER PROGRAMS
15.4.5

Argument Types

Table 15-1:

Argument Types and Types Codes

Type Code
(Octal)

Description
FORTRAN Use

o
1
2

3
4
5
6
7

10
11
12
13
14
15
16
17

Unspecified
FORTRAN Logical
Integer
Reserved
Real
Reserved
Octal
Label
Double real
(D-floating)
Not applicable
Double octal
Double real
(G-floating)
Complex
Character
Reserved
Hollerith

COBOL Use
Unspecified
Not applicable
I-word CaMP
Reserved
COMP-l
Reserved
Reserved
Procedure address
Not applicable
2-word CaMP
Not applicable
Not applicable
Not applicable
Byte string descriptor
Reserved
Not applicable

Literal arguments are permitted, but they must reside
in a writable
segment.
This
is because the FORTRAN compiler makes a local copy of
all non array elements and may copy dummy arguments back to the actual
arguments.
All
unused type codes are reserved for future DIGITAL
development.

15.4.6

Description of Arguments

The types of the arguments that may be passed are:
1.

Type 0 - Unspecified
The calling program has not specified the type.
The called
subprograms should assume that the argument is of the correct
type if it is checking types.
If several types are possible,
the called subprogram should assume a default as part of its
specification.
If none of the above conditions is true,
the
called subprogram should handle the argument as an integer
(type 2).

Type 1 -

FORTRAN logical

A 36-bit binary value containing 0 or positive
.FALSE.
and negative to specify .TRUE ••
3.

Type 2 -

Integer and l-word-COMP

A 36-bit 2's complement signed binary integer.

15-14

specify

WRITING USER PROGRAMS
4.

Type 4 - Real 'and COMP-l
A 36-bit floating-point number.
bit 0
bits 1-8
bits 9-35

sign
excess 128 exponent
mantissa

Type 6 - Octal
A 36-bit unsigned binary value.

Type 7 - Label and procedure address
The address of the parameter is the address of
return label or a subprogram.

represented

Type 11 - 2-word COMP
A 2-word (72-bit)
word
word
word
word

alternate

Type 10 - Double real (D-floating)
A double-precision floating-point number
D-floating form.
(See Section 3.4.)

Type 12

1,
1,
2,
2,

2's complement signed binary integer.

bit 0
bits 1-35
bit 0
bits 1-35

sign
high order
same as word 1, bit 0
low order

- Double octal

A 72-bit unsigned binary value.

10.

Type 13 - Double real

(G-floating)

(KL model B only)

A double-precision floating-point number
G-floating form.
(See Section 3.4.)
11.

represented

Type 14 - Complex
A complex number represented as an ordered pair of 36-bit
floating-point numbers.
The first represents the real part,
and the second represents the imaginary part.

12.

Type 15 - Character string descriptor
The format of the character string descriptor is:
word 1:
word 2:

ILDB-type pointer,
that is,
aimed at
the
character preceding the first character of the
string
EXP character count

The character descriptor may not be modified by the called
program.
The character string
itself must consist of a
string of contiguous 7-bit ASCII characters.

15-15

WRITING USER PROGRAMS
13.

Type 17 - Hollerith
A string of contiguous 7-bit ASCII characters left
justified
on the word
boundary of the first word and terminated by a
null character in the last word.
The FORTRAN compiler emits constants that are padded with
spaces to a word boundary, followed by a full-word containing
zero.

15.4.7

Interaction with COBOL

FORTRAN programs
conversely,
the
subprograms.

can call COBOL programs as subprograms,
and,
COBOL
programs can call
FORTRAN programs as

Note that I/O operations can be performed only in subprograms that are
written
in the same language as the main program.
Also note that APR
trap handling will be done in a manner consistent with the
language
used in the main program.

15.4.7.1 Calling
FORTRAN Subprograms from COBOL Programs - COBOL
programmers may write subprograms in FORTRAN to use the conveniences
and facilities provided by this language.
The COBOL verb ENTER is
used to call FORTRAN subroutines.
The form of ENTER is as follows:
ENTER FORTRAN program

name~SING

identifier-l
}
literal-l
{ procedure-name-l

[

{

i~entifier-2}J

Ilteral-2 •••
procedure-2

The USING clause names the data within the COBOL program that is to be
passed to the called FORTRAN subprogram.
The passed data must be in a
form acceptable to FORTRAN (see Table 14-1).
NOTE
G-floating double-precision does not exist as
type in COBOL.

data

The calling sequence used by COBOL in calling a FORTRAN subprogram is:
MOVEI 16, address of first entry in argument list
PUSHJ 17, subprogram address
If the USING clause appears
in the ENTER statement,
the compiler
creates an argument list that contains ·an entry for each identifier or
literal in the order of appearance
in the
USING clause.
It
is
preceded by a word containing, in its left half, the negative number
of the number of entries in the list.
If no USING clause is present,
the 'argument list contains an empty word, and the preceding word is
set to 0.
Each entry in the list is one 36-bit word of the form:
0-8

9-12

type

13-35
Effective address (E)

where:
type

is a
4-bit code that
indicates the USAGE of the
argument,
and bits 13-35 contain the effective address
(E) of the first word of the argument.

FORTRAN-10/20 VII

15-16

February 1987

WRITING USER PROGRAMS
If the passed argument is a I-word COMP item, the type code is 2 and E
is the location of the argument.
If the passed argument is a 2-word COMP item,
the type code is 11
(octal)
and E is the location of the first word of the argument; the
second word of the argument is at E+l.
If the passed argument is a COMP-l item, the code is 4 and
location of the argument.

the

if the passed argument is a COMP-2 item, the code is 10 (octal) and E
is the location of the first word of the argument; the second word of
the argument is at E+l.
If the passed argument is a DISPLAY-6, DISPLAY-7, DISPLAY-9, or COMP-3
item,
the code is 15
(octal)
and E is the location of a 2-word
descriptor for the argument.
The first word of the descriptor
is a
byte pointer word pointing to the argument.
Its byte-size is 6 for
DISPLAY-6, 7 for DISPLAY-7, and 9 for DISPLAY-9 or COMP-3.
The second word of the descriptor is of the form:
Bit 0

Reserved

Bit 1-4

Type code:
1= DISPLAY-6
2= DISPLAY-7
3= DISPLAY-9
4= COMP-3

Bit 5

Item is a literal

Bit 6

Item is a figurative constant (such as SPACES)

Bit 7

Item is numeric
if bit 7

Bit 8-11

Reserved

Bit 12-35

Size of item in bytes

If bit 7

Bit 8

Item is signed

Bit 9

Item is scaled (such as PICTURE contains "PiS" to
left of implied decimal point, for example,
999PP)

Bit 10

Item is numeric edited

If bit 9 = 0;
Bits 11-25

Reserved

Bits 26-30

Number of decimal places

Bits 31-35

Size of item in bytes

If bit 9
Bits 11-25

FORTRAN-10/20 VII

1;
Reserved

15-17

February 1987

WRITING USER PROGRAMS
Bits 26-30

Scale factor (such as, 2 if picture was 999PP)

Bits 31-35

Size of item in bytes (such as, 3 if picture was
999PP)

The return from a subprogram (through POPJ 17,)
after the call.

15.4.7.2 Calling COBOL Subroutines from FORTRAN
COBOL subprograms use the CALL statement:

the

statement

Programs -

call

CALL sub (args ••• )
where sub is a COBOL subprogram.
NOTE
COBOL l2C subroutines called
from FORTRAN programs
must be compiled with the /R COBOL switch to prevent
an undefined global error at LINK time.

15.4.8

Interaction with BLISS-36

FORTRAN programs can call
BLISS-36
routines as subprograms, and,
conversely, BLISS-36 programs can call subprograms written in FORTRAN.
BLISS routines called by FORTRAN programs must be able to coexist
compatibly with FOROTS.
For instance, they must use FUNCT.
functions
(see Section 18.6) for dynamic memory management within the section
that FOROTS
is
in, since FOROTS assumes that it has control of that
section.
One problem that the BLISS routines can encounter is stack
The size of the program stack as set up by FOROTS may be too
BLISS routines which have several STACKLOCAL variables.
supported way to
increase the size of the stack is to use
FORLIB that has been assembled with a larger stack.

overflow.
small for
The only
a copy of

15.4.8.1 Calling FORTRAN Subprograms From BLTSS-36 Programs - To call
a
FORTRAN subprogram from a BLISS-36 program, the FORTRAN subprogram
must be declared an EXTERNAL ROUTINE in any module that contains a
call
to the subprogram.
In addition, if the FORTRAN subprogram is a
subroutine, it must 'be declared with a linkage type of FORTRAN SUB.
If the FORTRAN subprogram is a function, then it must be declared-with
a linkage of FORTRAN_FUNC.
For example:
EXTERNAL ROUTINE FOO:
BAR:

FORTRAN SUB,
FORTRAN=FUNC;

declares FOO to be the name of a
FORTRAN subroutine which will be
called
in this module, and declares BAR to be the name of a FORTRAN
function.
After
the
FORTRAN
subprogram
has
been
declared
appropriately,
it can be called
just like any function written in
BLISS.

FORTRAN-10/20 VII

15-18

February 1987

WRITING USER PROGRAMS
15.4.8.2 Calling BLISS-36 Routines From FORTRAN - A BLISS-36
routine
that
is to be called
from a FORTRAN program must have either the
FORTRAN SUB linkage attribute (if the routine
is to be used as a
subroutTne)
or the FORTRAN FUNC linkage attribute (if the routine is
to be used as a function).
The programmer that wishes to write a BLISS-36 routine to be called
from
FORTRAN must be familiar with the calling sequence used by
FORTRAN to call subprograms (see Section 15.4.2), FORTRAN argument
lists
(see Section 15.4.4), and
FORTRAN argument descriptors (see
Sections 15.4.5 and 15.4.6).
This knowledge is necessary because the
values of the formal arguments of the BLISS-36 routine are the FORTRAN
argument list entries that correspond to actual arguments of the BLISS
routine.
In
general,
the
BLISS
routines
must
be
compiled
with
ADDRESSING MODE (INDIRECT) ,
or MACHOP calls must be used to generate
FORTRAN arguments
any instruction that references formals since all
This must be done
must be accessed
through
indirect addressing.
indirect bit in
because the FORTRAN compiler
frequently sets the
argument lists (see Section 15.4.4).
See the BLISS-36 Language Guide for more information.
NOTE
ADDRESSING_MODE (INDIRECT)
can
have
far
reaching
effects on your program, which you should understand
fully before using.

15.4.9

LINK Overlay Facilities

LINK provides several routines that are accessible directly from a
FORTRAN program.
These routines are presented here briefly, together
with the FORTRAN specification of their parameters.
In general,
LINK
performs these functions automatically.
These routines are available
only for your convenience.
Full details of the use of the overlay
facilities can be found in the LINK Reference Manual.
NOTE
Overlays are not allowed when
TOPS-20
extended
addressing
is used.
In addition, the overlay handler
does not support programs using long symbols
(symbols
longer than six characters).
The following terms are
overlay routines.

LINK

File spec

A
character
expression
consisting
'dev:file.ext[directory]'
(TOPS-10),
'dev:<directory>file.typ.gen' (TOPS-20).

of
or

Name

A quoted string giving a
link name, or an
integer constant or variable giving a link
number.

List of link names

A sequence of name items separated by commas.

FORTRAN-10/20 VII

used

describe

15-19

the

parameters

February 1987

WRITING USER PROGRAMS
The routines available are:
CLROVL

Specifies a non-writable overlay.

GETOVL(List of link names)

Used
to
change
structure in core.

INIOVL(File spec)

Used to specify the overlay file to
be
found
if
the
load
time
specification is to be overridden.

LOGOVL(File spec)

Used to specify where the log file
is to be written. If no arguments
are given, the log file is closed.

REMOVL(List of link names)

Removes the
core.

specified

links

from

RUNOVL (Name)

Loads the
specified
transfers to that LINK.

link

and

SAOVL

Specifies a writable overlay.

the

For a full description of these routines, refer to the LINK
Manual.

overlay

Reference

NOTE
The SAVE statement retains the values stored in a
variable, array, or common block after execution of a
RETURN or END statement in a subprogram.
When
overlays are used, the SAVE statement must be used to
ensure retention of values. When the SAVE statement
is used, it is not necessary to specify the LINK
switch /OVERLAY:WRITABLE when loading a program (see
Section 7.10).

15.4.10

FOROTS and Overlay Memory Management

For sharable FOROTS, the FOROTS static data area is several pages
located at the top of FOROTS. FOROTS dynamic memory is allocated at
runtime below FOROTS and in a downward direction (toward the user's
program) •
For nonsharable FOROTS (FOROTS loaded from FORLIB), the FOROTS data
area is located in the low segment, so that it will be linked with
variables used by the user's program.
FOROTS dynamic memory is
allocated at runtime starting at the page designated by the symbol
STARTP in FORPRM.MAC, downward toward the user
program.
The
distributed value for this page number is 577. If the user's program
has two segments, FOROTS allocates memory down to the user's high
segment, skips over the high segment, and begins allocating memory
below the user's high segment toward the user's low segment.
For both sharable and nonsharable FOROTS, when FOROTS can no longer
allocate memory downward toward the user's low segment, it allocates
memory starting at the top of memory downward.
When FOROTS can no
longer allocate any memory, it reports:
?Memory full
and returns to the monitor after attempting to close all files.
15-20

WRITING USER PROGRAMS
Figure 15-1 illustrates the run-time memory layout.

Page

777
Reserved for SORT, DBMS, and DDT

600
FOROTS

FOROTS Static Data

500

-1- - - - - -FOROTS Dynamic Data

-1- -

---

Used for Dynamic
Character Concatenations

User Program

o
MR-S-3878-85

Figure 15-1:

Run-time Memory Layout for Section Zero

FOROTS has a separate memory manager specifically designed for use by
OVRLAY.
This memory manager allocates memory at the top of the
users's low segment.
In general,
user programs that use overlays
should not use the FUNCT. calls GAD, COR, and RAD.
OVRLAY expects to
be able to use memory beginning at the top of the user's low segment,
allowing for a special FOROTS scratch space allocated by the /SPACE
switch.
Under certain circumstances, notably when concatenating character
expressions
whose
length
is not known until
runtime,
FOROTS
dynamically allocates a special scratch area using the FUNCT. COR
function
(see Section 18.6). This area is allocated the first time a
dynamic concatenation is performed.
The /SPACE switch to LINK determines the maximum size for such
"dynamic concatenations"
if overlays are used.
The default value
(See the LINK Reference
given by LINK for this value is 4000 (octal).
Manual.)

15-21

WRITING USER PROGRAMS
You can allocate space for dynamic character concatenation
ALCCHR subroutine (see Section 13.4.1.1).

15.4.11

using

the

Extended Addressing Memory Layout (TOPS-20 only)

The FORTRAN compiler must determine which psect every word of data or
code should reside in.
For non-extended addressing compilations, the
data and code reside in either the low segment or the high segment.
For
extended addressing compilations,
there are three segments
(psects) in which the data and code can be allocated:
1.

The code psect corresponds to the non-extended high
The default name is .CODE ••

segment.

The data psect corresponds to the non-extended
The default name is .DATA ••

low

segment.

The .LARG. psect is where large data resides.
name of this psect cannot be changed.)

(Note that the

A data item can be placed
following ways:

the

.LARG.

psect

either

the

The data item is an array or any character scalar whose size
is greater or equal to the value of the /EXTEND:DATA switch
( de fa u 1 t 10, aaa wo r d s).
( See Sec t ion 16. 5) •

The data item is placed in a COMMON
group that is in the .LARG. psect.

block

Table 15-2 describes the various memory allocations for
non-extended compilations.

15-22

EQUIVALENCE
extended

and

WRITING USER PROGRAMS
Table 15-2:

Memory Allocations for /EXTEND and /NOEXTEND

Item

/NOEXTEND

/EXTEND

User subprogram

Hiseg

Code

FORLIB

Hiseg

.CODE.

Argument blocks

Hiseg

Code

Compile-time constant
character descriptors

Hiseg

Code

Array dimension information

Hiseg

Code

EFIWS

N/A

Code

Symbol tables (from LINK)

Lowseg or Hiseg

.DATA.

FORMAT statements

Lowseg

Data

Constants

Lowseg

Data

Small arrays and scalars

Lowseg

Data

Large arrays

Lowseg

.LARG.

COMMON variables

Lowseg

.LARG. (by default)

Variables EQUIVALENCED to
large arrays

Lowseg

.LARG.

PDV

N/A

.DATA.

(by default)

NOTE
When the sharable FOROTS is used,
LINK loads the
/NOEXTEND high-segment into the low-segment. This
is done so that the sharable FOROTS can be used as
the high-segment (see Section 16.9)

15-23

WRITING USER PROGRAMS
lS.S

USING INDEXED FILES

(TOPS-2~

ONLY)

Traditionally, sequential and direct access have been the only file
access modes available to FORTRAN programs. To overcome some of the
limitations of these access modes', FORTRAN-20 supports a third access
mode, called keyed access (see Section 10.2.3). Keyed access allows
you to retrieve records, at random or in sequence, based on key fields
that are established when you create a file with indexed organization.
You can access files with indexed organization using either sequential
or keyed access, or a combination of both.
1.

Keyed access retrieves records randomly bas~d
on
particular key fields and key values that you specify.

Sequential access retrieves records in an ascending sequence
based on the values within the particular key field that you
specify.

the

The combination of keyed and sequential access is commonly referred to
as the Indexed Sequential Access Method (ISAM). Once you have read a
record by means of an indexed read request, you can then use a
sequential read request to retrieve records with ascending key field
values, beginning with the key field value in the record retrieved by
the initial read request.
Indexed organization is especially suitable for maintaining complex
files in which you want to select records based on one of several
criteria. For example, a mail-order firm could use an indexed
organization file to store its customer list. Key fields could be a
unique customer order number, the customer's zip code, and the item
ordered.
Reading sequentially based on the zip code key would enable
you to produce a mailing list sorted by zip code. A similar operation
based on customer order number or item number key would enable you to
list the records in customer order number or item number sequence.
The remainder of this section provides information
topics:

the

following

•

Creating an indexed file (Section IS.S.l)

•

Writing records to an indexed file (Section IS.S.2)

•

Reading records from an indexed file (Section IS.S.3)

•

Updating records in an indexed file (Section IS.S.4)

•

Deleting records from an indexed file (Section lS.S.5)

Information is also ~rovided about the effects of read and write
operations on positioning your program to records within an indexed
file (Section 15.5.6) and about how to build logic into your program
to handle exception conditions that commonly occur (Section 15.5.7).

lS.5.l

Creating an Indexed File

You can create a file with an indexed organization by using either the
FORTRAN OPEN statement or the RMSUTL utility.

FORTRAN-10/20 VII

15-24

February 1987

WRITING USER PROGRAMS
•

Use the OPEN statement to specify the file options
by FORTRAN.

supported

•

Use the RMSUTL utility
supported by FORTRAN.

directly

select

features

not

When you create an indexed file, you define certain fields within each
record
key fields.
One of these key fields, called the primary key,
is identified as key number zero and must be present in every record.
Additional
keys, called alternate keys, can also be defined; they are
numbered from 1 through a maximum of 254.
An indexed file can have as
many
as
255
key fields defined.
In practice,
however,
few
applications require more than 3 or 4 key fields.
The data
types
CHARACTER.

used

for

key

fields

must

either

INTEGER

In designing an indexed file, you must decide the byte positions of
the key fields.
For example, in creating an indexed file for use by a
mail-order firm, you might define a file
record to consist of
the
following fields:
INTEGER ORDER NUMBER
CHARACTER*20 NAME
CHARACTER*20 ADDRESS
CHARACTER*19 CITY
CHARACTER*2 STATE
CHARACTER*9 ZIPCODE
INTEGER ITEM NUMBER

positions
positions
positions
positions
positions
positions
positions

Given this
record definition,
you
statement to create an indexed file:

1:4, key 0
5:24
25:44
45:63
64:65
66:74, key 1
75:76, key 2

could

use

the

following

OPEN

(UNIT~l,FILE='CUSTOMER.DAT',STATUS='NEW' ,

ORGANIZATION=' INDEXED' ,ACCESS='KEYED',
RECORDTYPE='VARIABLE' ,FORM='UNFORMATTED',
RECL=19,
KEY=(1:4:INTEGER, 66:74:CHARACTER, 75:76:INTEGER) ,
IOSTAT=IOS, ERR=9999)

2
3
4

OPEN

This OPEN statement establishes the attributes of the file,
including
the definition of a primary key and two alternate keys.
The data type
size is determined by the number of character positions allotted to
the key fields,
which in this case are 4 and 2 character positions,
respectively.
If you specify the KEY= specifier (see Section 11.3.21)
when opening
an existing
file,
the key specification that you give is ignored.
However, RMS does update the key XAB chain with key information from
the
file.
(See Chapter 18
for more
information on RMS data
structures.)
FORTRAN uses RMS default key attributes when creating an indexed file.
These default are:
•

The values in primary key fields cannot be changed
record is rewritten and cannot have duplicates.

when

•

The values in alternate key fields can
have duplicates.

and

FORTRAN-10/20 VII

15-25

changed

a
can

February 1987

WRITING USER PROGRAMS
You can use the RMSUTL utility or a USEROPEN routine to override these
defaults and to specify other values not supported by FORTRAN-20, such
as null key field values, null key names, and key data types other
than integer and character.
See Section 18.9 for information on the use of the USEROPEN specifier
in OPEN statements. The RMS-20 programmer's Reference Manual has more
information on indexed file options. Use of the RMSUTL utility is
explaind in the TOPS-20 RMS User's Guide.

15.5.2

Writing Indexed Files

You can write records to an indexed file with either formatted or
unformatted indexed WRITE statements (see Section 10.6.3). Each write
operation inserts a new record into the file and updates the key
index(es) so that the new record can be retrieved in a sequential
order based on the values in the respective key fields.
For example, you could add a new record to the file for the mail-order
firm, described in Section 15.5.1, with the following ~tatement:
WRITE (UNIT=1,IOSTAT=IOS,ERR=9999)
The following two sections
write operations:

describe

considerations

that

relate

values

•

The effect of writing records with duplicate
fields (Section 15.5.2.1)

key

•

The method by which you can prevent an alternate key field in
a record from being indexed during a write operation (Section
15.5.2.2)

15.5.2.1 Duplicate Values in Key Fields - It is possible to write two
or more records with the same value in a single key field. The
attributes specified for the file when it is created determine whether
this duplication is allowed. By default, FORTRAN cr~ates files that
allow duplicate alternate key field values and prohibit duplicate
primary key field values. If duplicate key field values are present
in a file, the records with equal values are retrieved on a
first-in/first-out basis.
For example, assume that five records are written to an
in this order (for clarity, only key fields are shown):
ORDER NUMBER

ZIP CODE

1023
942
903
1348
1263

70856
02163
14853
44901
33032

FORTRAN-10/20 VII

indexed

file

ITEM NUMBER
375
2736
375
1047
690

15-26

February 1987

WRITING USER PROGRAMS
If the file is later opened and
read sequentially by primary key
(ORDER NUMBER),
the order
in which the records are retrieved is not
affected by the duplicated value (375) in the ITEM NUMBER key field.
In this case, the records would be retrieved in the following order:
ORDER NUMBER

ZIP CODE

903
942
1023
1263
1348

14853
02163
70856
33032
44901

ITEM NUMBER
375
2736
375
6~0

1047

However, if the read operation is based on the second alternate key
(ITEM_NUMBER),
the order
is which the records are retrieved is
affected by the duplicate key field value.
In this case, the
records
would be retrieved in the following order:
ORDER NUMBER

ZIP CODE

1023
903
1263
1348
942

70856
14853
33032
44901
02163

ITEM NUMBER
375
375
690
1047
2736

Notice that the records containing the same key field value (375)
retrieved in the order in which they were written to the file.

are

15.5.2.2 preventing the
Indexing of Alternate Key Fields - When
writing to an indexed file that contains variable-length records, you
can prevent entries from being added to the key indexes for any
alternate key fields.
This is done by omitting the names of the
alternate key fields from the WRITE statement.
The ommitted alternate
key fields must be at the end of the record; another key field cannot
be specified after the omitted key field.
For example, assume that the last record (ORDER NUMBER 1263)
mail-order example is written with the following statement:

the

WRITE (UNIT=1,IOSTAT=IOS,ERR=9999) ORDER_NUMBER,
1
NAME, ADDRESS, CITY,
1
STATE, ZIP_CODE
Because the field
name ITEM NUMBER is omitted from the
WRITE
statement,
an entry for that-key field is not created in the index.
As a result, an attempt to read
the file using
the alternate key
ITEM NUMBER would not retrieve the last record and would produce the
following listing:
ORDER NUMBER

ZIP CODE

1023
903
1348
942

70856
14853
44901
02163

ITEM NUMBER
375
375
1047
2736

You can omit only trailing alternate keys from a record;
key must always be present.

FORTRAN-10/20 VII

15-27

the

primary

February 1987

WRITING USER PROGRAMS
15.5.3

Reading Indexed Files

You can read records in an indexed file with either sequential or
indexed
READ statements
(formatted or unformatted) under the keyed
mode of access (see Section 10.5.3).
By specifying ACCESS=' KEYED'
in
the OPEN statement
(see Section 11.3.1), you enable both sequential
and keyed access to the indexed file.
Indexed READ statements position the file
pointers
(see Section
15.5.6) at a particular record, determined by the key field value, the
key-of-reference, and the match criterion.
Once you retrieve a
particular record by an
indexed READ statement, you can then use
sequential access READ statements to retrieve records with
increasing
key field values.
The form of the external record's key field must match the form of the
value you specify in the KEY= specifier (see Section 11.3.21). Thus,
if the key field contains character data, you should specify the KEY=
specifier value as a CHARACTER data type.
If the key field contains
binary data, then the KEY= specifier value should be of INTEGER data
type.
Note that if you write a record to an indexed file with formatted I/O,
the data type
is converted
from its internal representation to an
external representation.
As a result, the key value must be specified
in the external form when you read the data back with an indexed read.
Otherwise, a match will occur when you do not expect it.
The following FORTRAN program segment prints ths order number and
zip
code of each record where the first five characters of the zip code
are greater than or equal to '10000' but less than '50000':
C
C

Read first record with ZIP CODE key greater than or
equal to '10000'.
READ (UNIT=1,KEYGE='10000' ,KEYID=1,IOSTAT=IOS,ERR=9999)
1
ORDER NUMBER, NAME, ADDRESS,
1
CITY,STATE,ZIP_CODE,ITEM_NUMBER

C
C

While the zip code previously read is within range, print
the order number and zip code, then the next record.
DO WHILE (ZIP CODE .LT. '50000')
PRINT *, "Order number' ,ORDER_NUMBER, 'has zip code',
1
ZIP CODE
READ (UNI~=1,IOSTAT=IOS,END=200,ERR=9999)
1
ORDER NUMBER, NAME, ADDRESS
1
. CITY,-STATE, ZIP_CODE, ITEM_NUMBER

C
C
200

END= branch will be taken if there are no more records
in the file.
END DO
CONTINUE

The error branch on the keyed READ in this example
is taken
if no
record
is
found with a zip code greater that or equal to '10000'; An
attempt to access a nonexistent record is an error.
If the sequential
READ has accessed all
records in the file, however, an end-of-file
status occurs, just as with other file organizations.
If you wish to detect a failure of the keyed READ, you can examine the
I/O status variable (lOS) for the appropriate error number (see Table
D-l for a list of the returned error codes).
FORTRAN-10/20 VII

15-28

February 1987

WRITING USER PROGRAMS
15.5.4

Updating Records

The REWRITE statement (see Section 10.14) updates existing records in
an indexed
file.
You cannot replace an existing record simply by
writing it again; a WRITE statement would attempt to add a new record.
An update operation is accomplished in two steps.
First, you must
read the record
in order to make it the current record.
Next, you
execute the REWRITE statement.
For example,
to
update the record
containing ORDER NUMBER 903
(see prior examples) so that the NAME
field becomes 'Theodore Zinck', you might use the following
FORTRAN
code segment:
READ (UNIT=1,KEY=903,KEYID=0,IOSTAT=IOS,ERR=9999)
1
ORDER NUMBER, NAME, ADDRESS, CITY, STATE, ZIP_CODE,
1
ITEM NUMBER
NAME = 'Theodore Zinck'
REWRITE (UNIT=1,IOSTAT=IOS,ERR=9999)
When you rewrite a record, key fields may change.
The attributes
specified
for
the file when it was created determines whether this
type of change is permitted.

15.5.5

Deleting Records

To delete records from an indexed file, you use the DELETE statement
(see Section 10.13).
The DELETE and REWRITE statements are similar; a
record must first be locked by a READ statement before it can be
operated on.
The following FORTRAN code segment deletes the second
file with ITEM_NUMBER 375 (see previous examples):

record

the

READ (UNIT=1,KEY=375,KEYID=2,IOSTAT=IOS,ERR=9999)
READ (UNIT=1,IOSTAT=IOS,ERR=9999) (IOLIST)
IF (ITEM NUMBER .EQ. 375) THEN
DELETE (UNIT=l, IOSTAT=IOS, ERR= 9999)
ELSE
PRINT *, 'There is no second record,'
END IF
Deletion removes a record from all defined indexes in the file.

15.5.6

Current Record and Next Record Pointers

The RMS file system maintains two pointers into an open indexed
the "next record" pointer and the "current record" pointer.
•

file:

The next record pointer indicates the record to be retrieved
by a sequential read.
When you open an indexed file, the
next record pointer indicates the record with the lowest
primary
key
field
value.
Subsequent sequential read
operations cause the next record pointer to be the one with
the next higher value
in the same key field.
In case of
duplicate key field values,
records are retrieved
in the
order in which they were written.

FORTRAN-10/20 VII

15-29

February 1987

WRITING USER PROGRAMS
•

15.5.7

The current record pointer indicates the record most recently
retrieved by a READ operation; it is the record that is
locked from access by other programs sharing the file.
The
current record is the one operated on by the REWRITE
statement and the DELETE statement. The current record is
undefined until a read operation is performed on the file.
Any file operation other than a read causes the current
record pointer to become undefined. Also, an erior results
if a rewrite or delete operation is performed when the
current record pointer is undefined.

Exception Conditions

You can expect to encounter certain exception conditions when using
indexed files. The two most common of these conditions involve valid
attempts to read locked records and invalid attempts to create
duplicate keys.
Provisions for handling both of these situations
should be included in a well-written program.
When an indexed file is shared by several users, any ,read operation
may result in a "record locked" error. One way to recover from this
error condition is to ask if the user would like to reattempt the
read.
If the user's response is positive, then the program can go
back to the READ statement.
You should avoid simply looping back to the READ statement without
first providing some type of delay (caused by a request to try again,
or to discontinue). If your program reads a record but does not
intend to modify the record, you should place an UNLOCK statement (see
Section 10.15) immediately after the READ statement.
This technique
reduces the time that a record is locked and permits other programs to
access the record.
The second exception condition, creation of duplicate keys, occurs
when your program tries to create a record with a key field value that
is already in use. When duplicate key field values are not desirable,
you might have your program prompt for a new key field value whenever
an attempt is make to create a duplicate.

FORTRAN-10/20 VII

15-30

February 1987

CHAPTER 16
USING THE FORTRAN COMPILER

This chapter explains how to use the FORTRAN compiler.
Section 16.1
describes using the FORTRAN-IO compiler and Section 16.2 describes
using the FORTRAN-20 compiler.

16.1

USING THE FORTRAN-IO COMPILER

This section describes how use the FORTRAN-IO compiler. You should be
familiar with the TOPS-IO operating system.
The TOPS-IO Operating
System provides commands that enable you to compile, execute, and
debug FORTRAN programs.
These commands are known as the COMPILE-Class
commands.

16.1.1

TOPS-IO COMPILE-Class Commands

You can invoke the FORTRAN-IO compiler by using TOPS-IO COMPILE-Class
commands.
These commands enable you to compile, execute, and debug a
program by giving the commands at TOPS-IO command level.
The COMPILE-Class commands are:
COMPILE
LOAD
EXECUTE
DEBUG
Example:
.EXECUTE ROTOR. FOR
The compiler switches OPTIMIZE, CREF, and DEBUG may be specified
directly in COMPILE-Class commands and may be used globally or
locally.
(See Section 16.1.3 for a description of
FORTRAN-l 0
switches.)
Example:
.EXECUTE/CREF Pl.FOR,P2.FOR/DEBUG
The other compiler switches must be passed
specific source file.

16-1

parentheses

for

each

USING THE FORTRAN COMPILER
Example:
.EXECUTE Pl.FOR(INCLUDE)
Refer to the TOPS-IO Operating System Commands
information about the COMPILE-Class commands.

16.1.2

Manual

for

further

RUNNING THE FORTRAN-I0 COMPILER

On TOPS-lO, the command to run the FORTRAN compiler directly is:
.R FORTRA
The compiler responds with an asterisk
(*), and is then ready to
accept a command string.
The form of the FORTRAN compiler command
string is:
object filespec, listing filespec=source filespec(s)
You are given the following options:
1.

File specifications consist of an optional device name, a one
to six character filename, an optional one to three character
file extension, and an optional directory path specification.
The path may include SFDs.

You may specify more than one source file in the compilation
command string.
These files will be logically concatenated
by the compiler and treated as one source file.

More than one program unit
source file.

A program unit may consist of more than one source file.

If no object file is specified, no relocatable binary file is
generated.

If no listing file is specified, no listing is generated.

If no extension is given, the defaults are the following
the respective files:

may

contained

single

for

.LST (listing)
.REL (relocatable binary)
.FOR (source)
.CRF (cross reference)
if the
specified (see Section 16.1.3)

16.1.3

/CROSSREF

switch

TOPS-IO Compiler Command Switches

in the
Switches to the FORTRAN-IO compiler are accepted anywhere
command string.
They are totally position and file independent.
Table 16-1 lists the switches.

16-2

USING THE FORTRAN COMPILER
Table 16-1:
Switch

FORTRAN-10 Compiler Switches
Meaning

Defaul ts

/CROSSREF

Generates a file with extension .CRF
that can be input to the CREF program.

OFF

/DEBUG

Includes debugging information in your
program (see Section 16.3).

NONE

/DFLOATING

Indicates that double-precision
numbers are stored in D-floating
format. (See Section 3.4.)

/EXPAND

Includes the octal-formatted version
of the object file in the listing.

OFF

/F66

The FORTRAN-66 standard rules apply for
DO loops and EXTERNAL statements. (Same
function as the /NOF77 switch.)

OFF

/F77

The FORTRAN-77 standard rules apply for
DO loops and EXTERNAL statements.

/FLAG

Invokes the compatibility
Section 16.6).

OFF

/GFLOATING

Indicates that double-precision
numbers are stored in G-floating
format. (See Section 3.4.)

OFF

/INCLUDE

Compiles a D in card column 1 as
space.

OFF

/LNMAP

Produces a line number/octal location
map in the listing only if /MACROCODE
was not specified.

OFF

/MACROCODE

Adds the mnemonic translation of the
object code to the listing file.

OFF

/NOFLAG

Indicates that no compatibility
flagging will be done (see Section
16.6) •

/NOF77

The FORTRAN-66 standard rules apply for
DO loops and EXTERNAL statements. (Same
function as the /F66 switch.)

OFF

/NOERRORS

Does not print error messages on the
terminal.

OFF

/NOWARN

Suppresses warning messages (see
Section 16.4).

NONE

/OPTIMIZE

Performs global optimization.

OFF

/SYNTAX

Performs syntax check only.

OFF

16-3

flagger

(see

USING THE FORTRAN COMPILER
Each switch must be preceded by a slash (/).
Switch names need only
contain
those letters that are
required
to make the switch name
unique.
You are encouraged to use at least three letters
to
prevent
conflict with switches in future implementations.
Example:
.R FORTRA
*OFILE,LFILE=SFILE/MAC,S2FILE
The /MAC switch will cause the MACRO
S2FILE to appear in LFILE.LST.

code

generated

for

SFILE

All switches, used or implied, are printed at the top of each
page.

16-4

and

listing

USING THE FORTRAN COMPILER
16.2

USING THE FORTRAN-20 COMPILER

This section describes how to use the FORTRAN-20 compiler. You should
be familiar with the TOPS-20 Operating System.
The TOPS-20 Operating
System provides commands that enable you to compile, execute, and
debug
FORTRAN program.
These commands are known as the COMPILE-Class
commands.

16.2.1

TOPS-20 COMPILE-Class Commands

The TOPS-20 COMPILE-Class commands enable you to initiate compilation,
execution,
and debugging of FORTRAN programs from TOPS-20 command
level.
The TOPS-20 COMPILE-Class commands are:
COMPILE
LOAD
EXECUTE
DEBUG
Example:
@EXECUTE ROTOR. FOR
The following FORTRAN compiler switches (see Section
specified directly in a COMPILE-Class command:

16.2.3)

can

/ABORT
/BINARY
/CROSS-REFERENCE
/DEBUG
/LIST
/MACHINE-CODE
/NOWARNINGS
/OPTIMIZE
/WARNINGS
NOTE
When you specify the switches /BINARY,
/DEBUG,
/LIST,
and
/NOWARNINGS
directly
in
a
COMPILE-Class command, the switches cannot have
arguments as they can when running the compiler.
All other switches must be specified by using
as shown in the following example:

Language-switches,

@COMPILE TEST.FOR/OPT/LANG:"/NOWARNINGS"
Refer to the TOPS-20 Commands Reference Manual
information about the COMPILE-Class commands.

for

NOTE
You cannot use long TOPS-20
COMPILE-Class commands.

16-5

filenames

with

the

USING THE FORTRAN COMPILER
16.2.2

RUNNING THE FORTRAN-20 COMPILER

On TOPS-20, the command to run the FORTRAN compiler directly is:
@FOR~RA

The compiler responds with the following prompt:
FORTRAN>
and is then ready to accept a command string.
You can use the question mark to list the commands beginning with
a specific letter or letters.
Type
the letter or letters
followed by a question mark.
(Refer to the TOPS-20 User's
Guide.)
You can type commands to the system by using either full
input,
recognition input,
abbreviated input, or a combination of these
methods.
To give a command using full input,
type the complete command
name,
arguments, or switches (if any), using a space to separate
the fields.
To give a command using recognition input, type a portion of the
switch or filename and press the ESC key.
You must type enough
of the switch or filename to make it unique.
Continue typing and
pressing the ESC key until the switch or filename is complete.
Recognition input requires less typing than full
input, so you
are less likely to make a mistake.
You can use recognition in
typing switches, switches arguments, and file specifications.
When typing file specifications,
you can also use CTRL/F to
complete the rest of a partial file specification.
To give a command using abbreviated input, type only enough of
the switch or filename to distinguish one switch or filename from
another.
(Refer to the TOPS-20 User's Guide.)
You should enter a command string in one of the following forms:
1.

[COMPILE]<source-file-spec> [switches]
If no switches are specified, the compiler produces a
.REL
file,
with the same filename as the source file.
The user
must use a /LISTING switch to get a listing file (see Section
16.2.3 for
a description of FORTRAN-20 compiler switches).
COMPILE is optional if the command begins with a switch or
begins with a file specification that cannot be confused with
the words EXIT, HELP, TAKE, or COMPILE.

16-6

USING THE FORTRAN COMPILER
2.

[COMPILE]<source-file-spec>+<source-file-spec>+ •••
[swi tches]
The source files are treated as
if they were
together prior to the beginning of compilation.

concatenated

If no switches are specified, the compiler produces a
.REL
file,
with the same filename as the last source file in the
list.
The user must use a /LISTING switch to get a
listing
(see Section 16.2.3 for a description of FORTRAN-20 compiler
switches).
COMPILE is optional if the command begins with a
switch or begins with a file specification that cannot be
confused with the words EXIT, HELP, TAKE, or COMPILE.
3.

TAKE <file-spec> [/ECHO]
The compiler reads the file specified as the command
input
stream.
The TAKE command is legal within 'take' files.
The
maximum nesting depth is 10 'take' files.
The
on
can
the

/ECHO switch optionally causes commands to be displayed
TTY:
as they are executed.
The optional /NOECHO switch
be used on a nested take command to cancel the affect of
/ECHO switch while processing that nested command file.

RUN <file-spec> [/OFFSET:<integer>]
This command runs another program (for example,
LINK).
It
causes an exit from the FORTRAN compiler and the start of
execution of the program indicated by the file specification,
with the additional option of starting that program at an
OFFSET relative to the normal starting address.

HELP
This command prints information on the user's terminal
how to use the FORTRAN compiler.

about

EXIT
This command exits from FORTRA.

You are given the following options:

Filename specifications consist of the following:
An optional device name (the default device is DSK: )
An optional directory name
An up to 39 alphanumeric character filename
An optional up to 39 alphanumeric character file type
An optional generation
version of the file

number

that

identifies

An optional
file attribute
to specify
characteristics of a file specification
(Refer to the TOPS-20 User's Guide)

16-7

the

distinctive

USING THE FORTRAN COMPILER
NOTE
LINK is restricted to
3-character extensions.

6-character

filenames

and

You may specify more than one source file in the compilation
command string.
These files will be concatenated by the
compiler and treated as one source file.
The name of the
last source file is used as the default name of the object
and listing files.
If the last source file does not have a
name
(such as, TTY:), FORTRAN-OUTPUT is used as the default
filename.

More than one program unit may
source file.

A program unit may consist of more than one file.

If no /LISTING switch is specified (see Section 16.2.3),
listing is generated.

If no extension is given, the defaults are the following for
the respective files:
.LST (listing)

contained

single

if the /CROSSREF switch is not specified

.CRF (cross reference)
if the
specified (see Section 16.2.3)

/CROSSREF

switch

.REL (relocatable binary)
.FOR (source)

16.2.3

TOPS-20 Compiler Commands Switches

Switches to the FORTRAN-20 compiler are accepted anywhere in the
command string.
They are totally position and file independent.
Table 16-2 lists the switches.

16-8

USING THE FORTRAN COMPILER
Table 16-2:

FORTRAN-20 Compiler Switches

Switch

Meaning

Defaul ts

/ABORT

Causes the compiler to exit at the
end of a compilation that contains
errors.

OFF

/BINARY[:relfile]

Indicates that a relocatable binary
file is generated. You can
optionally specify the file
specification.

/CROSS-REFERENCE

Generates a file with extension
.CRF that can be input to the
CREF program.

OFF

/DEBUG [keys: ]

Includes debugging information in
your program (see Section 16.3).

NONE

/DFLOATING

Indicates that double-precision
numbers are stored in D-floating
format. (See Section 3.4.)

/ECHO-OPTION

Echo switches selected from the
SWITCH.INI file.

OFF

/EXPAND

Includes the octal-formatted version
of the object file in the listing.

OFF

/EXTEND[keys:]

Indicates extended addressing.
Programs can have up to 30
sections of code and data (see
Section 16.5).

OFF

/F66

The FORTRAN-66 standard rules apply
for DO loops and EXTERNAL statements. (Same function as the /NOF77
switch.)

OFF

/F77

The FORTRAN-77 standard rules apply
for DO loops and EXTERNAL
statements.

/FLAG-NON-STANDARD

Invokes the compatibility flagger
(see Section 16.6).

OFF

/GFLOATING

Indicates that double-precision
numbers are stored in G-floating
format. (See Section 3.4.)

OFF

/INCLUDE

Compiles a D in card column 1 as
space.

OFF

/LISTING[:listfile]

Indicates a list file will be
generated. You can optionally
specify the file specification.

OFF

16-9

USING THE FORTRAN COMPILER
Table 16-2:

FORTRAN-20 Compiler Switches (Cont.)

switch

Meaning

Defaults

/LNMAP

Produces a line number/octal
location map in the listing only
if /MACHINE-CODE was not specified.

OFF

/MACHINE-CODE

Adds the mnemonic translation of
the object code to the listing
file. This command will cause
a default /LISTING.

OFF

/NOBINARY

Indicates that no relocatable
binary file is generated.

OFF

/NOF77

The FORTRAN-66 standard rules apply
for DO loops and EXTERNAL
statements. (Same function as
the /F66 switch.)

OFF

/NOFLAG-NON-STANDARD

Indicates that no compatibility
flagging will be done (see Section
16.6) •

/NOERRORS

Does not print error messages
on the terminal.

OFF

/NOEXTEND

Indicates extended addressing
is not in effect (see Section 16.5).

/NOWARN

Suppresses warning messages (see
Section 16.4).

NONE

/OPTIMIZE

Performs global optimization.

OFF

/OPTION [:option]

Only read lines from the SWITCH.INI
file that start with FORTRA:option.

OFF

/SYNTAX

Performs syntax check only.

OFF

Each switch must be preceded by a slash (/).
Switch names need only
contain those letters that are required to make the switch name
unique.
You are encouraged to use at least three letters to prevent
conflict with switches in future implementations.
NOTE
When using switches
in control
files,
you
encouraged to type the full name of the switch.

16-10

are

USING THE FORTRAN COMPILER
Example:
@FORTRA
FORTRAN>SFILE+S2FILE/MAC/LIST:LFILE
The /MAC switch will cause the MACRO code generated for SFILE and
S2FILE to appear
in LFILE.LST. An relocatable binary file will be
created with the name S2FILE.REL.
All switches, used or implied, are printed at the top of each
page.

16.3

listing

THE /DEBUG SWITCH

The /DEBUG switch tells FORTRAN to compile a series of debugging
features
into
your
program.
Several of these features are
specifically designed to be used with the FORTRAN debugging program
(FORDDT).
Refer to Chapter 17 for more information.
By using the
DEBUG switch arguments listed in Table 16-3, you can include specific
debugging features.
The form of the /DEBUG switch is:
/DEBUG:arg
or
/DEBUG: (argl,arg2, ••• )

16-11

USING THE FORTRAN COMPILER
Table 16-3:

Arguments to /DEBUG Switch

Arguments

Meaning

DIMENSIONS

Includes dimension information in
FORDDT.

.REL

file

for

TRACE

Generates references to FORDDT required for its
trace features (automatically activates LABELS).

LABELS

Generates a label for each statement of the form
<line-number>L.
(This option can be used without
FORDDT. )

INDEX

Forces DO LOOP indexes to be stored at the
beginning of each iteration rather than held in a
register for the duration of the loop.
In addition, this switch forces all function
values to be stored in memory prior to return from
the function. If this switch is specified, you
can set a FORDDT pause on the RETURN statement
( see Sec t i on 13. 4 • 4) and t h en examine the value to
be returned.

BOUNDS

Generates the bounds checking code for all array
references
and
substring references.
Bounds
violations will produce run-time error messages.
Note that the technique of specifying dimensions
of 1 for subroutine arrays will cause bounds check
errors.
(You may use this option without FORDDT.)

ARGUMENTS

Generates type checking information at load time
for actual argument types and associated dummy
argument types.
Type violations will produce
non-fatal load-time error messages. This switch
also performs type checking at compile-time for
statement functions.

NONE

Do not include any debug features.

ALL

Enable all debugging aids.

NOTE

1
1

1
1
1
1

If the /DEBUG switch is specified without arguments,
then the defaults are used for all the arguments. The
/DEBUG switch used without arguments is the same as
/DEBUG:(ARGUMENTS,BOUNDS,DIMENSIONS,INDEX,LABELS,
TRACE) •

1
1

1
1
1
"I
1
1

You can invoke a single aspect of debugging without
affecting the other features controlled by the switch.
For example, specifying /DEBUG:ARGUMENTS does not
cause BOUNDS checking to also be invoked. However,
you can change multiple features of the /DEBUG switch
in
the
same
command
line.
For
example,
/DEBUG:ARGUMENTS /DEBUG:BOUNDS would
invoke
both
argument checking and bounds checking.

FORTRAN-10/20 VII

16-12

February 1987

USING THE FORTRAN COMPILER
Options available with the /DEBUG arguments are:
1.

No debug features - Either "do not specify the
or include /DEBUG:NONE.
.

All debug features -

Either/DEBUG or /DEBUG:ALL.

Selected features that is:

Either a

series

/DEBUG

modified

switch

switches,

/DEBUG:BOU/DEBUG:LAB
or a list of modifiers
/DEBUG:(BOU,LAB, ••• )

Exclusion of features - If you wish all but one or
modifiers and do not wish to list them all, you can use
prefix "NO" before the switch you wish to exclude.
exclusion of one or more features implicitly includes all
others,
that
is,
/DEBUG:NOBOU
is
the
same
/DEBUG: (DIM,TRA,LAB,IND,ARG).

If you include more than one statement on
first
statement will
receive a
label
reference (/DEBUG:TRACE).

two
the
The
the
as

a single line, only the
(/DEBUG:LABELS)
or FORDDT

NOTE
If a source file contains line sequence numbers that
occur more than once
in the same subprogram, the
/DEBUG option cannot be used. Also, the /DEBUG option
and
the /OPTIMIZE option cannot be used at the same
time.
The following formulas may be used
to determine the
increases
in
program size that will occur as a result of the addition of various
/DEBUG options.
DIMENSIONS

For each array, 3+3*N words where N is the number
of dimensions, and up to three constants for each
dimension.

TRACE

One instruction per executable statement.

LABELS

No increase.

INDEX

One
instruction
per
inner
loop
plus
one
instruction for some of the references to the
index of the loop.
Also one instruction per
subprogram.

BOUNDS

For each array,
DIMENSIONS.

the

formula

the

same

For each reference to an array element, 5+N words
additional words are generated, where N is the
number of dimensions in the array.
If you do not
specify BOUNDS, approximately 1+3*(N-l) words will
be used.
For each reference to a substring, add 5
words.
ARGUMENTS

No increase.

16-13

February 1987

USING THE FORTRAN COMPILER
If the /DEBUG:ARGUMENTS switch argument is specified, type checking is
performed at LINK time for calls to external programs and at compile
time for calls to statement functions." Non-fatal error messages are
issued at LINK time for the following cases:
1.

If the number of arguments in the called subprogram
calling program unit are not equal.

If the length of an array or character scalar actual argument
is less than that of the corresponding dummy argument.
(This
is checked only if the length of the actual is known at
compile time.)

If the associations of actual argument data types with dummy
argument data types are other than those indicated as legal
in Table 16-4.

If a non-routine name is passed where a routine name is
expected, or a routine name is passed where a non-routine
name is expected (see Table 16-5).

Non-fatal error messages are issued at compile time for the
cases (only for statement functions):

and

the

following

If a non-routine name is passed where a routine name is
expected, or a routine name is passed where a non-routine
name is expected.

If the length of the actual character expression being passed
to
a
statement
function
is less than that of the
corresponding character expression dummy argument.

Table 16-4: Legal Dummy and Actual Argument Associations
Actual Argument Type
Alternate
Double
Return
Label
Logical Integer Real D-floating G-floating Complex Character Octal Hollerith Octal
Alternate
Return
Label
Logical

Integer
Real
D-floating
G-floating

X
X
X

Complex

X
X

Character

X indicates legal associations. All others will cause a warning to be issued if IDEBUG:ARGUMENTS is specified.

FORTRAN-10/20 VII

16-14

February 1987

USING THE FORTRAN
Table 16-5:

COMP~LER

Legal Dummy and Actual Argument Structure Associations
Actual Argument Structure
Singleton

Singleton

Array

Routine

legal associations.
All others will cause
warning to be issued if /DEBUG:ARGUMENTS is specified.

X indicates

16.4

THE /NOWARN SWITCH

The /NOWARN switch is used to suppress compiler warning messages.
If
this switch is used with no arguments, all warning messages are
suppressed. The /NOWARN switch may also be used with arguments as
shown in Table 16-6.
Table 16-6:

Arguments to /NOWARN Switch

Arguments

Meaning

ALL

Suppress all warning messages.

NONE

Do not suppress warning messages.

xxx

Where xxx is the three character error mnemonic
the error message to be suppressed. This is
for
for example
the three letters that follow %FTN,
%FTNABD.

For example,
/NOWARN:LID
will suppress all warnings of
characters.

identifier

A list of arguments is also allowed.

having

than

six

For example,

/NOWARN:(LID,DIM)
would suppress both LID and DIM types of warning messages.
(See Appendix C for a list of all compiler error mnemonics.)

FORTRAN-10/20 VII

16-15

February 1987

USING THE FORTRAN COMPILER
THE /EXTEND SWITCH (TOPS-20 ONLY)

16.5

Support for extended addressing is almost completely transparent; a
program compiled with the /EXTENb switch will use extended addressing
without requiring changes to the FORTRAN source program.
When /EXTEND is specified, arrays ~nd COMMON blocks can extend across
multiple sections.
Executable code can also reside in multiple
sections, with the restriction that a single subprogram must not cross
a section boundary.
.
The /EXTEND switch can be specified without arguments to use the
default extended address space layout.
This is suitable for most
applications in which the executable code fits within a single
section, but which may employ data structures that require more than a
section of memory. When such an application is compiled /EXTEND
without further arguments, a default memory layout is used that
depends on the default size settings for arrays and strings:
Table 16-7:

/EXTEND Default Memory Layout

Section

Pages

Contents

O-477

Executable code
Argument blocks
Literals
Non-COMMON scalars
Non-COMMON arrays smaller than
10,000 words
Non-COMMON strings smaller than
10,000 words

50O-577

FOROTS

60O-777

Reserved

2-31

O-777

COMMON blocks
Arrays larger than 10,000 words
Strings larger than 10,000 words

Additional arguments to /EXTEND (COMMON:, DATA:, PSECT:, and CODE), in
conjunction with the /SET switch to LINK, can be supplied to override
these defaults and direct specific placement of COMMON blocks,
non-COMMON arrays and scalars, an~ executable code. For example, a
decimal argument can be specified for the DATA: keyword, which
overrides the 10,000 word default minimum size for large arrays and
character scalars. /EXTEND:COMMON.can be used to direct specific
named COMMON blocks to non-default sections of extended memory.
(Section 16.5.3 describes the /EXTEND arguments.)

16-16

February 1987

USING THE FORTRAN COMPILER
NOTE
If the /EXTEND switch is specified without arguments,
then the default-s are used for all the arguments.
The
/EXTEND switch used without arguments is the same as
/EXTEND:(DATA:10000,COMMON,NOCODE,PSECT:.DATA.:.CODE.)
(See Section 16.5.3.)
You can invoke a single aspect 9f extended addressing
without affecting the other features controlled by the
switch.
For example, specifying /EXTEND:DATA does not
cause COMMON blocks to be placed into the large psect,
which
is done with the /EXTEND:COMMON
argument.
However,
you can change multiple features of the
/EXTEND switch in the same command line.
For example,
/EXTEND:CODE:100/EXTEND:COMMON
would
place
all
non-common data items not less than 100 words long and
all COMMON blocks into the large psect.

16.5.1

/EXTEND and Applications with Large Arrays

The simplest usage of extended addressing is for applications in which
the executable code fits in one section.
These applications may have
very large arrays or common blocks.
In many cases, these applications
can be compiled with the /EXTEND switch with no arguments, using the
default memory layout in Table 16-7.
In some cases, you may need to use the DATA or COMMON arguments to
/EXTEND to
redirect the placement of variables in areas when default
placements cannot be used.
Specifying a smaller /EXTEND:DATA size may
be necessary if the total size of non-COMMON scalars and arrays causes
them to overlap FOROTS pages.
If such overlap occurs,
program
execution will terminate with the error messages:
"?Can't get FOROll.EXE"
and
"?Can't overlay existing pages"
A smaller DATA size will force these d-ata structures into a non-FOROTS
section.

16.5.2

/EXTEND and Applications with _Large Executable Code

A more complex use of extended addressing
is one
where
the
application's executable code is larger than a section.
In order to
build an application that has more than one section of executable
code,
you have to specify which program units will be linked together
in each section.

FORTRAN-10/20 VII

16-17

February 1987

USING THE FORTRAN COMPILER
When /EXTEND is specified, the extended code will be in three PSECTS,
one analogous to the present hiseg, one analogous to the present
lowseg, and one containing the larg~ variables. If the user does not
specify the extended addressing switch, FORTRAN will gene.rate a TWOSEG
REL file as it always has. The three PSECTS generated under /EXTEND
are:
1.

The "large data area" psect (.LARG.) has a maximu~ size of 30
sections.
It will consist of user-specified COMMON blocks,
arrays and character scalars that are larger
than
a
user-specified size (or default).
There is no restriction on the size of an individual array or
COMMON block, beyond the restriction on the total size of the
large data area.

The "small data area" psect (default name .DATA.) of a
program consists of user-specified COMMON blocks, and scalars
and arrays that are smaller than a user-specified size (or
default).
If FORDDT is used, it will reside in the .DATA.
psect.

The "executable code and sharable data area" psect (default
name .CODE.) of a program consists of all executable code,
argument blocks and literals.
Library
functions
and
subroutines used by the program are placed in the .CODE.
psect. This does not include space used by SORT, which
occupies its own section.

For each "executable code area"
(code psect)
there will be a
corresponding "small data area" (data psect). The combined size of
these two areas must not exceed 256K words. The default psect names
.DATA. and .CODE. can be changed at compile time by the /EXTEND:PSECT
command switch (see Section 16.5.3).

16.5.2.1 /EXTEND PSECT Placement - The three psects will
according to the table below:

set

Psect

Default
Origin

Attributes

data

1001000

code

1300000

.LARG.

2000000

Single section, Non-zero section,
Concatenated, Writable
Single section, Non-zero section,
Concatenated, Read-only
Non-zero section, Concatenated, Writable

You can alter the default psect origins when loading programs by
glvlng a /SET switch to LINK (see the LINK Reference Manual). This
may be necessary if a program has too much code or local data to fit
in the regions alLocated for them by the default psect origins. This
would cause the psects to overlap, and would be indicated by a LINK
warning message (%LNKPOV).
Altering the defaults is also useful if
the user wants to reserve one or more sections for some purpose (such
as telling the monitor where dynamic libraries should be loaded).
The first page (locations 000-777) of any section that contains code
is reserved for use by FORDDT.
The LINK /SET switch should not
specify a psect origin less than 1000 for any such section.

16-18

February 1987

USING THE FORTRAN COMPILER
16.5.2.2 Building Large-Code Applications - When compiling large-code
applications, the following considerations apply:
1.

The code and small data psects for a single program unit must
always reside together in the same section, since the PC will
no advance across section boundaries, and the small data area
is assumed to be in the same section.

You will need to use the /EXTEND:CODE keyword.
This
specifies that the· object. code emitted by the compiler
assumes that all subprograms that it calls may be
in a
separate section.

You must also use the PSECT argument to /EXTEND to specify
the psect names for code and small data.
If you wish to link
the program units in several different compilations
(source
files)
together in the same section, you should use the same
psect names for those program units.

At link-time, you must specify the starting address for
using the /SET switch, according to the following rules:

each

The code and small data psects for a program
linked in the same section with each other.

If a section contains any executable code,
section is reserved for FORDDT and FOROTS.

You must always allocate space for
the
.CODE.
and
.DATA.
psects, since FORLIB routines will be linked in these psects.

Pages 500-577 of the section that contains .CODE. and
are reserved for FOROTS.

Pages 600-777 are reserved for FOROTS I/O buffers and DDT.

You must always allocate space for the .LARG. psect.

16.5.3

unit

psect

page

must

that

.DATA.

Arguments to /EXTEND

By using th~ /EXTEND switch arguments listed in Table
include specific extended addressing features.

16-8,

you

can

The form of the /EXTEND switch is:
/EXTEND: a rg
or
/EXTEND: (argl,arg2, ••• )

16-19

February 1987

USING THE FORTRAN COMPILER
Table 16-8:

Arguments to /EXTEND Switch

Arguments

Meaning

CODE

Specifies that the object code produced by the
compiler has to assume that any subprogram that
it calls could be in a separate section.
NOCODE
is the default.

COMMON [ : name]
or
COMMON: (name, ••• )

Without a common block name specified, causes
all common blocks
that
have not already been
allocated by /EXTEND:[NO]COMMON to be allocated
in the
.LARG.
psect.
This is the default.
Individual
·common
blocks
can
be
placed
explicitly in .LARG. by putting their names in a
list after COMMON:.
When you explicitly place
an individual common block in .LARG., any common
blocks that have not already been allocated by
/EXTEND:[NO]COMMON are placed in the small data
psect.

DATA[:decimal number] Specifies a decimal argument that is the minimum
size
(in words)
for
non-common arrays and
character scalars, which will be allocated to
the .LARG. psect.
The default is 10,000.
NOCODE

Allows the compiler to assume that all of the
code will
be
in the same section. A program
compiled with the NOCODE argument cannot call
any subprograms compiled with the CODE argument.
This is the default.

NOCOMMON [ : name]
or
NOCOMMON:(name, ••• )

Without a common block name specified, causes
all common blocks to be
allocated in the data
psect.
Individual common blocks can be placed
explicitly in the data psect by putting their
names in a list after COMMON:.
COMMON is the
default.

NODATA

Specifies that all
non-common variables will
reside
in
• DATA.
This
is equivalent to
DATA:1073741823, which excludes all variables
from • LARG.

PSECT[:[data psect]
[ : [code psect]]]

Allows users to set the code and data psect
names explicitly (the large data psect is always
called
.LARG.)
If PSECT is specified with one
argument, that argument becomes the name of the
small data area psect.
Any second argument
becomes the name of the code psect.
This allows
separate program units
to be put in separate
psects,
then the psect can be
placed
in
different sections with the /SET switch at LINK
time (see the LINK Reference Manual).
NOTE

When using the PSECT argument, the small data psect
and code psect for
any given program unit must be
loaded into the same memory section.

16-20

February 1987

USING THE FORTRAN COMPILER
16.5.4

Linking with TWOSEG REL Files

If a main program unit compiled with /EXTEND is linked with subprogram
units
(FORTRAN or not) that were compiled for non-extended use, then
LINK will automatically place the LOWSEG of non-extended units in the
.DATA. psect, and the HISEG of such units into the .CODE. psect.
A program compiled with /EXTEND can call a subprogram that is not
compiled with /EXTEND; howeve~, it is illegal for a subprogram that is
not compiled with /EXTEND to call a subprogram that is.
Programs that were compiled by old versions of FORTRAN-10/20 (prior to
Version 7) will not work if loaded in a non-zero section.
Most MACRO routines written for non-extended use will
require
conversion to run in non-zero sections.
Data structures accessed with
18-bit address fields, indexed and indirect words, stack pointers and
some monitor calls may need modification to perform correctly in
extended sections. See the TOPS-20 Monitor Call User's Guide for more

information concerning extended MACRO progra~

16.6

-----

THE /FLAG (/FLAG-NON-STANDARD) SWITCH

The /FLAG switch invokes the compatibility flagger.
NOTE
For TOPS-20 systems,
the full
switch names
are
/FLAG-NON-STANDARD and /NOFLAG-NON-STANDARD; however
/FLAG and /NOFLAG work.
This feature provides warning messages for language elements used that
are the following:
•

Extensions to the ANSI FORTRAN-77 standard

•

Features not found in VAX FORTRAN

•

Features that could cause logically
used on the VAX FORTRAN system

different

results

when

NOTE
VAX FORTRAN is used
operating systems.

the

VAX/VMS

and

ULTRIX

This allows the flagging of any element that could cause conversion
problems for
programs written on the TOPS-10/20 system that might be
compiled and executed on a VAX/VMS system or an ANSI-compatible
system.
This
includes problems that could occur at object time, as
well as compilation incompatibilities.

FORTRAN-10/20 VII

16-21

February 1987

USING THE FORTRAN COMPILER
By using the FLAG switch arguments
specify which features to flag.

listed

Table

16-9,

you

can

The form of the /FLAG switch is:
/FLAG [ : arg]
or
/FLAG [: (argl ,arg2, ••• )]
Table 16-9:

Arguments to /FLAG Switch

Arguments

Meaning

ALL

Gives warning
messages
for
language
elements
incompatible with both FORTRAN-77 and VAX FORTRAN.

ANSI

Gives warning messages whenever a language element is
an extension to the FORTRAN-77 standard.

NOANSI

Does not flag FORTRAN-77 extensions.

NONE

Does not flag.

NOVMS

Does not flag VAX incompatibilities.

VMS

Gives warning messages whenever a language element is
incompatible with VAX FORTRAN.

If no /FLAG switch is specified, no flagging is done. If no arguments
are given with the /FLAG switch, then flagging is done for both
FORTRAN-77 and VAX incompatibilities.
The /NOFLAG switch indicates that no flagging will be done.

16.7

READING A FORTRAN COMPILER LISTING

When you request a listing from the FORTRAN compiler, it may contain
the
following
information, depending on the switches used at
compilation time:
1.

A printout of the source program plus an internal sequence
number assigned to each line by the compiler. This internal
sequence number is referenced in any error or warning
messages generated during the compilation. If the input file
is line-sequenced, the number from the file is used. If code
is added by means of the INCLUDE statement, all INCLUDEd
lines will have an asterisk
(*)
appended
to
their
line-sequence number.

A summary of the names and relative program locations (in
octal)
of
scalars
and arrays (including unreferenced
character scalars and arrays~ in the source program plus
compiler-generated variables.

16-22

February 1987

USING THE FORTRAN COMPILER
3.

All COMMON blocks and the relative locations
the variables in each COMMON block.

(in

A listing of all equivalenced variables or arrays and
their
relative locations.
Note that all equivalenced variables
that are also in COMMON are listed only as being in COMMON.

A listing of the subprqgrams referenced
and FORTRAN-defined li~rary functions).

A summary of temporary locations generated by the compiler.

A heading on each page of the listing containing the program
unit name
(MAIN.,
.BLOCK, program, subroutine or function,
principal entry), the input filename, the list of compiler
switches, and the date and time of compilation.

If you used the /MACRO switch, a mnemonic printout of the
generated code (in a format similar to MACRO) is appended to
the listing.
This section has four fields:

(both

octal)

user-defined

LINE:
This column contains the internal sequence number
of the line corresponding
to
the mnemonic code.
It
appears on the first
line of the
code
sequence
associated with that
internal sequence number.
An
asterisk indicates a compiler inserted line.
LOC:
The relative location in the object program of the
instruction.
LABEL:
Any program or compiler
generated
label.
Program labels have the letter UP" appended.
Labels
generated by the compiler are followed
by the letter
"M".
Labels generated by the compiler and associated
with the /DEBUG:LABELS switch consist of the
internal
sequence number followed by an "L".
GENERATED CODE:

The MACRO mnemonic code.

If you use the /LNMAP switch and do NOT use the /MACRO
switch,
a line number/octal location map is appended to the
listing.
This section lists the line numbers
ln increments
of 10 on subsequent lines and each number from 0 through 9
for each line in adjacent columns.
The numbers appearing
inside the matrix are the relative octal locations of the
statements in the FORTRAN program unit.
For example, to find the relative octal
location of line
number 001043,
find the row marked 001040 and then column 3
on that line.
The number
in that place
is the desired
relative location.
This listing can be very large and sparse
for line-numbered files with large increments, such as those
produced by the editor SOS on TOPS-10 (or the editor EDIT on
TOPS-20) •
NOTE
A single FORTRAN line can produce multiple
machine
instructions.
In this case the line
number map lists only the first location.

16-23

February 1987

USING THE FORTRAN COMPILER
9.

A list of all argument blocks generated by the compiler.
A
zero argument appears first followed by argument blocks for
subroutine calls and function references (in order of their
appearance in the program) '.
Argument blocks for all I/O
operations follow this.

10.

FORMAT statement listings.

11.

A summary of errors ,detected
during compilations.

16.7.1

warning

messages

issued

Compiler-Generated Variables

In certain situations the compiler generates internal variables.
Knowing what these variables represent can help you read the macro
expansion. The variables are of the form:
.letter digit digit digit digit
The function of these variables can be determined by the first
of the variable name as described below:
Letter

letter

Function of Variable

Compile-time constant character descriptor

Arithmetic statement function formal parameters.

Result of a DO LOOP initial value expression
parameter of an adjustable dimensioned array.

Result of
a
computat ion.

Temporary storage for expression values.

Result of reduced operator strength expression.

Result of the DO LOOP step size expression of
iteration count for a loop.

common

subexpression

constant

computed

For example:

You may find these variables on the listing under SCALARS and ARRAYS.
The following examples show listings where all of these
pointed out.

16-24

features

are

February 1987

USING THE FORTRAN COMPILER

Example 1

, ,

Program
Name

Source
Name

MAIN.

TIM1.FOR

00001
00002
00003
00004
00005
00006
00007
00008
00009
00010
00011
00012
00013
00014
00015
00016
00017
00018
00019
00020
00021

100

Compiler Version

FORTRAN V.11(4543)/F77/M

2-DEC-86

10:26

MACRO code listing included
IMPLICIT INTEGER (A-Z)
DIMENSION A(100,200)
COMMON B(100,200)
OPEN(UNIT=22,FILE='TIM1.DAT')
INTEGER_SUM1=O
INTEGER_SUM2=0
DO 100 J=1,200
DO 100 1=1,100
K1=I*J
IF (K1 .LT. 500 .OR. K1 .GT. 1500) K1=O
A(I,J)=K1
K2=I+J
IF (K2 .EQ. 100 .OR. K2 .EQ. 200 .OR. K2 .EQ. 300) K2=K2+1
B(I,J)=K2
INTEGER_SUM1=INTEGER_SUM1+K1
INTEGER_SUM2=INTEGER_SUM2+K2
CONTINUE

WRITE(22,10)INTEGER_SUM1,INTEGER_SUM2
FORMAT('INTEGER_SUM1=' ,19,'
INTEGER_SUM2=' ,19)
END

COMMON BLOCKS
/ . COMM. / (+47040) ..._ _ _ Relative addresses of each variable _ _ _ _ _ _ _ _ _.....
B

"*" N0 1EXPLICIT DEFINITION

*K1
*INTEGER_SUM2

1
47045

*J l::2
*1
47046

A
*K2

-J"%,, NOT REFERENCED]
3
47047

.S0001
*INTEGER_SUM1

Relative address of each character data descriptor

CHARACTER DATA
NAME
'TIM1.DAT'

"*" NO EXPLICIT DEFINITION]
DESCRIPTOR ADDRESS
START OF DATA
ADDR(POSITION)
. HSCHD+O

47043
47050

.SOOOO

47044

t
Compiler generated
variable

LENGTH

I 8 - # of characters
t Relative address of first character of string

47b54(1)

Character position of first character of string

FORTRAN-10/20 VII

16-25

February 1987

USING THE FORTRAN COMPILER

Internal sequence number on first instruction
for this source line

LINE

4
5
6
7

Octal displacement of instruction

LOC

LABEL

0
1
2
3
4
5
6
7
10

JFCL
JSP

3M:

4M:

5M:

14
15
16
17
20
21
22

7M:

6M: •

24
25
26
27
30
31
32
33
34
35
36
37

10M:

9M:

8M:

15
16
17

42
43
44
45
46
47
50

12
13

GENERATED CODE

XMOVEI
PUSHJ
SETZB
MOVEM
MOVE
HLREM

0,0
16,RESET.
0,0
16,2M
17,OPEN.
2,INTEGER_SUMl
2,INTEGER_SUM2
2, [777470000001]
2, .SOOOO

HRRZM

2,J

MOVE

2, [777634000001]

MOVE
IMULI
MOVEM
CAlL
CAlLE
JRST
JRST

3,J
3,0(2)
3,Kl
3,764
3,2734
O,7M
O,6M

SETZB

4,Kl

MOVEI
·IMUL
ADD I
MOVE
MOVEM
MOVE
ADD I
MOVEM
MOVE
CAIE
CAIN
JRST

3,144
3,J
3,0(2)
4,Kl
4,A-145(3)
3,J
3,0(2)
3,K2
5,K2
5,144
5,310
O,9M

CAIN

5,454

ADS

3,K2

MOVEI
IMUL
ADD I
MOVE
MOVEM
ADDM
ADDM

3,144
3,J
3,0(2)
5,K2
5,B-145(3)
4,INTEGER_SUMl
5,INTEGER_SUM2

AOBJN
HRRZM
ADS
AOSGE
JRST
XMOVEI
PUSHJ
XMOVEI
PUSHJ
XMOVEI
PUSHJ

2,5M
2,1
2,J
0, .SOOOO
,O,4M
16,l1M
17,OUT.
16,12M
17,IDLST.
16,lM
17 ,EXIT.

Compiler generated label

lOOP: •

51
52
53
54
55
56
57
60
61
62

FORTRAN-10/20 VII

Program label

16-26

February 1987

USING THE FORTRAN COMPILER

ARGUMENT BLOCKS: ...- - - - Function, subroutine, and FOROTS argument blocks
63
64
65
66
67
70
71
72
73
74
75
76

0,,0
0,,0
777776, ,0
436000,,26
406640" . HSCHD+O
777776, ,0
401000, ,26
402340, ,lOP
777775, ,0
401100, ,INTEGER_SUMl
401100, ,INTEGER_SUM2
4000, ,0

1M:
2M:
11M:
12M:

FORMAT STATEMENTS (IN LOW SEGMENT):
20
MAIN.

47056
47057

lOP:

(' INTEGER_SUM1= ',19,'

[No errors detected ] _

INTEGER_SUM2=' ,19)

Summary of errors

Example 2
MAIN.
00001
00002
00003
00004
00005
00006
00007
00008
00009
00010
00011
00012
00013
00014
00015
00016
00017
00018
00019
00020
00021

TIM1.FOR

100

FORTRAN V.ll(4543)/F77/LN

2-DEC-86

10:28

IMPLICIT INTEGER (A-Z)
DIMENSION A(100,200)
COMMON B(100,200)
OPEN(UNIT=22,FILE='TIM1.DAT')
SUM1=0
SUM2=0
DO 100 J=l,200
DO 100 1=1,100
Kl=I*J
IF (Kl .LT. 500 .OR. Kl .GT. 1500) Kl=O
A(I,J)=Kl
K2=I+J
IF (K2 .EQ. 100 .OR. K2 .EQ. 200 .OR. K2 .EQ. 300) K2=K2+1
BO, J)=K2
SUM1=SUM1+Kl
SUM2=SUM2+K2
CONTINUE

WRITE(22,10)SUM1,SUM2
FORMAT(' SUM= ',19,'
END

SUM2=' ,19)

COMMON BLOCKS
/.COMM./(+47040)
B
+0
SCALARS AND ARRAYS [ "*" NO EXPLICIT DEFIIhTION - "%" NOT REFERENCED]
*Kl
*SUM2

1
47045

47046

*K2

47047

.SOOOl
*SUMl

47043
47050

.SOOOO

47044

LINE NUMBER/OCTAL LOCATION MAP ...----Requested with /LNMAP
: 0

00000
00010
00020
MAIN.

23
30
33
61
[ No errors detected ]

13
55

Line #11 starts at octal offset 23 (from the previous example, note that line 11 uses locations 23 through 27,
but only the first location~is shown here).

FORTRAN-10/20 VII

16-27

February 1987

USING THE FORTRAN COMPILER
Example 3
MAIN.
00001
00002
00003
00004
00005
00006
00007
00008
00009
00010
00011
00012
00013
00014
00015
00016
00017
00018
00019
00020
00021

TIM1.FOR

100

FORTRAN V.11(4543)/F77/0PT/M

2-DEC-86

10:30

IMPLICIT INTEGER (A-Z)
DIMENSION A(100.200)
COMMON B(100.200)
OPENCUNIT=22.FILE='TIM1.DAT')
SUM1=O
SUM2=0
DO 100 J=1.200
DO 100 1=1.100
K1=hJ
IF (K1 .LT. 500 .OR. K1 .GT. 1500) K1=O
ACI.J)=K1
K2=I+J
IF (K2 .EQ. 100 .OR. K2 .EQ. 200 .OR. K2 .EQ. 300) K2=K2+1
B(I.J)=K2
SUM1=SUM1+K1
SUM2=SUM2+K2
CONTINUE

WRITE(22.10)SUM1.SUM2 '
FORMAT ( , SUM1= '.19.'
END

SUM2= '.19)

COMMON BLOCKS
/.COMM./(+47040)
B
+0
SCALARS AND ARRAYS [ n"n NO EXPLICIT DEFINITION - n%n NOT REFERENCED]
•

Optimizer,created variables

*K1
1
.S0001 47045
*K2
47052

. R0001
.S0001
*SUM1

CHARACTER DATA
NAME

[ n*n NO EXPLICIT DEFINITION]
DESCRIPTOR ADDRESS
START OF DATA
ADDR(POSITION)

LENGTH

'TIM1.DAT'

. HSCHD+O

FORTRAN-10/20 VII

2
47046
47053

.ROODO
*SUM2

3
47047

*J
*1

47060(1)

16-28

4
47050

A
.00001

5
47051

February 1987

USING THE FORTRAN COMPILER
LINE

4
6

*
7

*
8

LOC
0
1
2
3
4
5
6
7
10
11
12
13
14
15

LABEL

JFCL
JSP

8M:

7M:

26
27
30
31
32
33

11M:

10M:

9M:

15
16

36
37
40
41

*17
*
*

*
*
*
*
19
*
21

42
43
44
45
46
47
50
51
52
53
54
55
56
57
60
61

MOVE
MOVE

6,7
2, [777634000001]

MOVE I
ADD
MOVE
CAlL
CAlLE
JRST
JRST

4,0(2)
4, . ROOOl
5,6
5,764
5,2734
O,8M
O,7M

MOVEI

5,0

MOVEM
MOVE
ADDI
CAIE
CAIN
JRST

5,A-145(4)
3,7
3,0(2)
3,144
3,310
O,10M

CAIN

3,454

ADDI

3,1

MOVEM
ADD
ADD
ADD

3,B-145(4)
11,5
10,3
6,7

AOBJN
HRRZM
MOVEI
ADDM

2,6M
12,144
12,144
12, . ROOOl

ADDI
AOSGE
JRST
MOVEM
MOVEM
MOVEM
MOVEM
MOVEM
XMOVEI
PUSHJ
XMOVEI
PUSHJ

7,1
0, .SOOOO
O,5M
7,J
11, SUMl
10,SUM2
5,Kl
3,K2
16,12M
17,OUT.
16,13M
17,IDLST.

XMOVEI
PUSHJ

16,3M
17,EXIT.

Asterisks indicate optimizer
generated statements

6M:

XMOVEI
PUSHJ
SETZB
MOVEI
MOVEM
MOVNI
MOVEI
MOVEM

0.0
16,RESET.
0,0
16,4M
17,OPEN.
10,11
12,144
12, . ReOOl
12,310
7,1 12, . SOOOO

5M:

16
17
20
21
22
23
24

9
10

GENERATED CODE

lOOP:

1M:

2M:

FORTRAN-10/20 VII

16-29

February 1987

USING THE FORTRAN COMPILER

ARGUMENT BLOCKS:
63
64
65
66
67
70
71
72
73
74
75
76

3M:
4M:
12M:
13M:

0, ,0
0,,0

777776, ,0
436000, ,26
406640, , . HSCHD+O
777776, ,0
401000, ,26
402340, ,lOP
777775, ,0
401100, ,SUMl
401100, ,SUM2
4000, ,0

FORMAT STATEMENTS (IN LOW SEGMENT):
20
MAIN.

47062
6
47063
lOP:
('SUM1=' ,19,'
[No errors detected ]

FORTRAN-10/20 VII

SUM2=' ,19)

16-30

February 1987

USING THE FORTRAN COMPILER
Example 4
MAIN.
00001
00002
00003
00004
00005
00006
00007
00008
00009
00010
00011
00012
00013
00014
00015
00016
00017
00018
00019
00020
00021

TIM1.FOR

100

FORTRAN V.ll(4543)/F77/M/EXT

2-DEC-86

16:27

WRITE(22,10)SUM1,SUM2
FORMAT ( I SUM1= 1,19, I
END

COMMON BLOCKS

SUM2= 1,19)

["!" STORED IN .LARG.

/.COMM./(+47040!)
+0
.L._____ Large common block

SCALARS AND ARRAYS
*Kl
1
.SOOOl 3
*1
6
CHARACTER DATA
NAME

"*" NO EXPLICIT DEFINITION - "%" NOT REFERENCED ]
"!" VARIABLE STORED IN .LARG. ]
*J
2
! J...----O-- Large variable
.SOOOO 4
*K2
7

'TIM1.DAT'

. HSCHD+O
LABEL

14(1)

XMOVEI
PUSHJ
SETZB
MOVEM
MOVE
HLREM

0,0
16,RESET.
0,0
16,2M
17,OPEN.
2,SUMl
2,SUM2
2, [777470000001]
2, .SOOOO

HRRZM

2,J

MOVE

2, [777634000001]

MOVE
lMULI
MOVEM
CAlL
CAlLE
JRST
JRST

3,J
3,0(2)
3,Kl
3,764
3,2734
O,7M
O,6M

SETZB

4,Kl

4
5
6
7

5
6

10
11

3M:

4M:

5M:

14
15
16
17
20
21
22

GENERATED CODE
JFCL
JSP

5
10

["*" NO EXPLICIT DEFINITION - "!" VARIABLE STORED IN .LARG. ]
DESCRIPTOR ADDRESS
START OF DATA LENGTH
ADDR(POSITION)

LINE

LOC

*SUM2
*SUMl

7M:

FORTRAN-10/20 VII

16-31

February 1987

USING THE FORTRAN COMPILER
LINE

LDC

LABEL

6M:

24
25
26
27
30
31
32
33
34
35
36
37

10M:

9M:

8M:

15
16
17

42
43
44
45
46
47
50

12
13

GENERATED CODE
MOVE I
lMUL
ADDI
MOVE
MOVEM
MOVE
ADDI
MOVEM
MOVE
CAIE
CAIN
JRST

3,144
3,J
3,0(2)
4,Kl
4,~[ .EFIWA-145(3)
3,J
3,0(2)
3,K2
5,K2
5,144
5,310
O,9M

CAIN

5,454

ADS

3,K2

MoVEI
lMUL
ADDI
MOVE
MoVEM
ADDM
ADDM

3,144
3,J
3,0(2)
5,K2
5,~[ .EFIW B-145(3)
4,SUMl
5,SUM2

AoBJN
HRRZM
ADS
AoSGE
JRST
XMOVEI
PUSHJ
XMoVEI
PUSHJ
XMoVEI
PUSHJ

2,5M
2,1
2,J
0, .SOOOO
O,4M
16,11M
17,oUT.
16,12M
17,IDLST.
16,lM
17,EXIT.

lOOP:

51
52
53
54
55
56
57
60
61
62

ARGUMENT BLOCKS:
63
64
65
66
67
70
71
72
73
74
75
76

0, ,0
1M:
2M:
11M:
12M:

0, ,0

777776, ,0
436100" [000000000026]
406640" . HSCHD+O
777776, ,0
401000, ,26
402340, , lOP
777775, ,0
401100"SUMl
401100, ,SUM2
4000, ,0

FORMAT STATEMENTS (IN LOW SEGMENT):
20
MAIN.

17
20

lOP:

('SUM1= ',19,'

SUM2=' ,19)

[No errors detected]

FORTRAN-10/20 VII

16-32

February 1987

USING THE FORTRAN COMPILER
NOTE
Note that in the scalars and arrays list, 'NO EXPLICIT
DEFINITION'
indicates that the variable was never
explicitly defined,
as
in a TYPE
or
DIMENSION
statement.
Also, 'NOT REFERENCED' indicates that the
variable was declared, but never used,
and
therefore
was never allocated any storage in the program.
Character variables that are declared,
but
not
referenced,
appear under the scalars and arrays
section of the listing.
No storage is allocated
for
either the character descriptor or the character data.

16.8

ERROR REPORTING

If an error occurs during the initial pass of the compiler (while the
actual
source code is being read and processed), an error message is
printed on the listing immediately following the line
in which the
error occurred. When pertinent and possible, the error references the
internal sequence number of the incorrect line.
The error messages
along wi th "the statement in error are output to the user terminal.
Example:
TYPE DAY.FOR
01000
01100
01200
01300
01400
01500
100
01600
C
01700
01800
200
01900

I=10
IMPLICIT INTEGER (X)
J=I**4
Kl
X=I+J+Kl
CONTNUE
TYPE 200,X
FORMAT (IX, I8)
END

COMPILE DAY.FOR
FORTRAN: DAY
IMPLICIT INTEGER (X)
01100
IMPLICIT statement out of order
%FTNSOD LINE:01100
Kl
01300
?FTNNRC LINE:01300 Statement not recognized
01500
100
CONTNUE
?FTNMSP LINE:01500 Statement name misspelled
?
01600
?FTNICL LINE:01600 Illegal character C in label field
?FTNFTL
If errors
appear in
output to
reference

MAIN.

3 fatal errors and 1 warning

are detected after the initial pass of the compiler,
they
the list file after the end of the source listing.
They are
your terminal without the statement in error, but they may
its internal sequence number.

16-33

February 1987

USING THE FORTRAN COMPILER
16.8.1

Fatal Errors and Warning Messages

There are two levels of messages, warning and fatal error.
Warning
messages are preceded by "%" and indicate a possible problem. The
compilation will continue, and the object program may be correct.
Fatal errors are preceded by a "?". If a fatal error is encountered
in any pass of the compiler, the remaining passes will not be called,
and no relocatable binary file will be generated.
Additional errors that would be detected in later compiler passes may
not become apparent until the first errors are corrected. It is not
possible to generate a correct object program for a source program
containing a fatal error.
The format of messages is:
?FTNxxx Line:n text
or
%FTNxxx Line:n text
where:
?

%
FTN
xxx
Line:n
text

indicates a fatal message
indicates a warning message
is the FORTRAN mnemonic
is the 3-letter mnemonic for the error message
is the optional line number where error occurred
is the explanation of error

The printing of fatal errors and warning messages on your terminal can
be suppressed by the use of the /NOERRORS switch; however, messages
will still appear on the listing. The /NOWARN switch will suppress
warning messages on both the user terminal and in the listing.
Specific warnings can be suppressed by using options to the /NOWARN
switch (see Section 16.4).

16.8.2

Message Summary

At the end of the listing file and on the terminal, a message summary
is printed after each program unit is compiled. This message has two
forms:
1.

When one or more messages were issued
?FTNFTL
%FTNWRN

name 1 fatal error and no warnings
name 2 fatal errors and no warnings
name no fatal errors and 1 warning
name no fatal errors and 2 warnings

or
2.

When no messages were issued
name [No errors detected]

where name is the program or subprogram name. Appendix C
complete list of fatal errors and warning messages.

16-34

contains

February 1987

USING THE FORTRAN COMPILER
16.9

CREATING A SHARABLE HIGH SEGMENT FOR A FORTRAN PROGRAM

For non-extended addressing programs,
the FORTRAN compiler always
generates two segment code for a program unit.
However, by default,
LINK loads all FORTRAN code into the low segment to allow the sharable
run-time system to be bound to the program at run-time rather than at
load-time.
This default action of LINK can be overridden by using the LINK switch
/OTSEGMENT:NONSHARABLE when loading the program.
If this switch is
given, LINK loads the impure code (the data areas) in the low segment,
loads the pure code (the machine instructions) in the high segment,
and binds a private copy of the run-time system to the program at
load-time.
On TOPS-10, a program loaded with the /OTSEGMENT:NONSHARABLE switch
can be saved
to produce an executable file with a sharable high
segment using the TOPS-10 SSAVE command or the TOPS-20 EXEC SAVE
command.
(The LINK switch /SSAVE can also be used to produce the
sharable executable file.) This is an advantage if a large application
program is to be run by several users simultaneously.
Both the code
unique to the program and the copy of the FORTRAN run-time system that
is private to the program are shared between all the program's users.
On TOPS-20, all pure and
impure pages of all programs begin as
sharable pages, however,
as
impure pages are modified by stores,
TOPS-20 creates non-sharable copies of the pages. The default action
of LINK in programs that do not use extended addressing is to load all
FORTRAN code into the low segment.
This results
in the
impure and
pure code of each program unit being loaded as one contiguous unit.
Since the presence of impure code on a page will ultimately result in
the page becoming non-sharable, the effective number of pages that are
sharable in the program is less that the maximum that could be
achieved.
However, if the /OTSEGMENT:NONSHARABLE switch is used, all
of the pure code of the program is loaded with a private copy of
FOROTS into the high segment.
Since the pure code in this case is not
mixed with the impure code, a greater number of pages remain sharable
in the program (also, the number of page faults may be reduced).
Note that extended addressing programs on TOPS-20 by default have
impure code and pure code loaded
into disjoint areas so that
/OTSEGMENT~NONSHARABLE does not increase the sharability
or decrease
the number of page faults of these programs.
In addition, these
programs use the system-wide sharable FOROTS, and thus make the best
use of system resources.
On both TOPS-10 and TOPS-20, the possible benefits gained by the users
of the application program sharing the high segment of their program
containing both user and FOROTS code must be weighed against the loss
of not sharing the common copy of FOROTS with all other users.
(See the LINK Reference Manual for more information on the
switch.)

FORTRAN-10/20 VII

16-35

/OTSEGMENT

February 1987

CHAPTER 17
USING THE FORTRAN INTERACTIVE DEBUGGER (PORDDT)

FORDDT is an interactive program used to debug FORTRAN programs and
control their execution. By using the symbols created by the FORTRAN
compiler, FORDDT allows you to examine and modify the data and FORMAT
statements
in your program, set breakpoints at any executable
statement or routine, trace your program statement-by-statement, and
make use of many other debugging techniques described in this chapter.
Table 17-1 lists all the commands available to the user of FORDDT.
Table 17-1:

FORDDT Commands

Command

Purpose

Control Commands

CONTINUE

Continues
execution
breakpoint.

after

DDT

Enters DDT.

GOTO

Transfers control to some program statement
within the open program unit.

Traces execution of the program.

START

Begins execution of the FORTRAN program.

STOP

Terminates the
monitor mode.

program

and

FOR DDT

returns

Data Access Commands
ACCEPT

Modifies variables or FORMAT statements.

TYPE

Displays variables.

CHARACTER

Defines dimensions of character arrays for
FORDDT
references.
(This
command is
unnecessary if /DEBUG is specified
at
compile time.
See Table 16-3.)

17-1

Declarative Commands

USING THE FORTRAN INTERACTIVE DEBUGGER (FORODT)
Table 17-1:

FORDDT Commands (Cont.)

Command

Purpose

DIMENSION

Defines dimensions of real
and
integer
arrays
for
FORDDT
references.
(This
command
is unnecessary
if
/DEBUG
is
specified
at compile time.
See Table
16-3.)

DOUBLE

Defines dimensions of double-precision and
complex
arrays
for
FORDDT references.
(This command is unnecessary if /DEBUG
is
specified
at compile time.
See Table
16-3.)

GROUP

Defines indirect lists for TYPE statements.

MODE

Specifies format of typeout.

OPEN

Accesses program unit symbol table.

PAUSE

Sets FORDDT breakpoints.

PAUSE ON ERROR

Sets FORDDT breakpoints (for errors such as
arithmetic overflows).

REMOVE

Clears FORDDT breakpoints.

REMOVE ON ERROR

Clears PAUSE ON ERROR breakpoints.

Other Commands
LOCATE

Lists program unit names in which
symbol is defined.

STRACE

Displays routine
program status.

WHAT

Displays current DIMENSION, GROUP,
FORDDT breakpoint information.

traceback

given
current
and

The FORDDT commands are described in detail in Section 17.5.

17.1

INPUT FORMAT

FORDDT commands consist of alphabetic
FORTRAN-like
identifiers and
need consist of only those characters required to make the command
unique.
If you wish to specify parameters, a
space
is
required
following the command name.
Comments may be appended to command lines
by preceding the comment with an exclamation point (!).

17.1.1

Variables and Arrays

FORDDT allows you to access and modify the data in your program using
standard FORTRAN symbolic names.
Variables are specified simply by
name.
For example:
name
17-2

USING THE FORTRAN INTERACTIVE DEBUGGER (FORDDT)
where:
name is a variable name.
Array elements are specified in the following formats:
name
name (sl, ••• ,sn)
where:
name

is the name of the array

(sl, ••• ,sn)

are the subscripts of a particular array.
The
subscripts must be integer constants or variables.

You may reference an entire array simply by typing the array name
(without subscripts).
You may specify a range of array elements by
typing the first and last element in the chosen range, separated by a
dash (-).
The following examples show the various ways of
and arrays to FORDDT:

specifying

variables

ALPHA
ALPHA(7)
ALPHA(PI)
ALPHA(2)-ALPHA(5)

17.1.2

Constant Conventions

FORDDT accepts optionally signed numeric data in the standard
input formats:

FORTRAN

INTEGER - A string of decimal digits.

REAL - A string of decimal digits optionally including a
decimal point.
Standard engineering and double-precision
exponent formats are also accepted.

OCTAL - A string of octal digits
double quote (").

COMPLEX - An ordered pair of integer or real
separated by a comma and enclosed in parentheses.

LOGICAL - A Boolean argument, either .TRUE.

CHARACTER - A string of printable ASCII
by apostrophes.

HOLLERITH
A string of
alphanumeric
and/or
special
characters
delimited
by
any
alphanumeric or special
character, excluding the space character, which does not
occur with the string itself.
Such as, # l2AB#, where # is
the delimiting character.

17-3

optionally

preceded

constants

or .FALSE.

characters

enclosed

USING THE FORTRAN INTERACTIVE DEBUGGER (FORDDT)
17.1.3

Statement Labels and Source Line Numbers

FORTRAN statement labels are input and output by straightforward
numeric reference, such as 1234. However, source line numbers must be
input to FORDDT with a number sign (#) preceding them. This mandatory
sign distinguishes statement labels from source line numbers.

17.2

PAUSE #3

IThis causes a pause at source line number 3.

PAUSE 3

IThis causes a pause at the statement labeled 3.

FORDDT AND THE FORTRAN /DEBUG SWITCH

Most facilities of FOR DDT are available without the FORTRAN /DEBUG
features. However, if you do not use the /DEBUG switch when compiling
a FORTRAN program, the trace features (NEXT command) will not be
available, and several of the other commands will be restricted. In
addition, the /DEBUG switch must be used if long symbols are to be
made available to FORDDT. If long symbols are used in your FORTRAN
program, and you do not use the /DEBUG switch when compiling, only the
first six characters of each symbol is available to FORDDT.
Using the /DEBUG switch tells FORTRAN to compile extra information for
debugging.
(See Chapter 16 for more information.) These features are:
1.

/DEBUG:DIMENSIONS, which generates dimension information in
the .REL file for all arrays dimensioned in the subprogram.
The dimension information is automatically available to
FORDDT if you wish to reference an array in a TYPE or ACCEPT
command.
This feature eliminates the need to
specify
dimension information for FORDDT by using the DIMENSION
command.

/DEBUG:LABELS, which generates labels for every executable
source line in the form <line-number>L. If these labels are
generated, they may be 'used as arguments with the FORDDT
commands PAUSE and GOTO.
This switch also generates labels at the last location
allocated for a FORMAT statement so that FORDDT can detect
the end of the statement.
These labels have the form
<format-label>F.
If they are generated, you can display and
modify FORMAT statements with the TYPE and ACCEPT commands.
Note that the :LABELS switch is automatically activated with
the :TRACE switch, since labels are needed to accomplish the
trace features.

/DEBUG:TRACE, which generates a reference to FORDDT before
each executable statement. This switch is required for the
trace command NEXT to function.
Note that if more than one FORTRAN statement is placed on a
single input line, only the first statement has a FORDDT
reference and line-number label associated with it.
This
also applies to the :LABELS switch.

/DEBUG:INDEX, which forces the compiler to store in its
respective data location, as well as a register, the index
variable of all DO loops at the beginning of each loop
iteration.
You will then be abl~ to examine DO loops by
using FORDDT. If you modify a DO loop index using FORDDT, it
will not affect the number of loop iterations because a
separate loop count is used.
FORTRAN-10/20 Vll
17-4
February 1987

USING THE FORTRAN INTERACTIVE DEBUGGER (FORDDT)
In addition, this switch forces all
function values to be
stored
in memory prior to return from the function.
If this
switch is specified, you can set a FORDDT pause on the RETURN
statement
(see Section 13.4.4) and then examine the value to
be returned.

17.3

/DEBUG:BOUNDS, which generates the bounds checking code for
all array references.
Bounds violations produce run-time
error messages.
Note that the technique of specifying
dimensions of 1 for subroutine arrays causes bounds check
errors.
(You can use this option without FORDDT.)

/DEBUG:ARGUMENTS, which performs type checking at load time
for actual argument types and associated dummy argument
types.
Type violations produce non-fatal,
load-time error
messages.
This switch also performs type checking at
compile-time for statement functions.

LOADING AND STARTING FORDDT
1.

On TOPS-IO, the simplest method of debugging with FORDDT is:
.DEBUG filespec (DEBUG)
On TOPS-20, the corresponding command is:
@DEBUG filespec /DEBUG
On both systems, FORDDT responds with:
STARTING FORTRAN DDT
Program name:
When FORDDT prompts you for a program name, type the same
name specified in the PROGRAM statement of the program being
debugged.
If the PROGRAM statement is not used
ln the
program being debugged,
FORDDT uses MAIN., and will not
prompt for a program name.
FORDDT next prints its command prompt:

»
The angle brackets indicate that FORDDT is ready to receive a
command.
2.

If you are on TOPS-20, you can type a question mark to
prompt to get a list of all FORDDT commands, as follows:
»7 One of the following:
CHARACTER
CONTINUE
ACCEPT
GROUP
HELP
GOTO
PAUSE
REMOVE
OPEN
WHAT
TYPE

DDT
LOCATE
START

Also on TOPS-20, you can use the ESCape key
of FORDDT commands.
For example:
»CON<ESC>TINUE

17-5

the

DIMENSION DOUBLE
MODE
NEXT
STOP
STRACE
for

recognition

USING THE FORTRAN INTERACTIVE DEBUGGER (FORDDT)
On both systems, you need only type the
of a specific FOR DDT command.
3.

unique

abbreviation

FORDDT
You may wish
to
load your compiled program and
directly with
the linking
loader.
(Loading with LINK is
The
accomplished implicitly in the DEBUG command string.)
command sequence is as follows:
On TOPS-IO, to start LINK, type:
.R LINK
On TOPS-20, type:
@LINK
On both systems, when LINK prompts you with an asterisk,
can type a command string in any of the following forms:
*filespec /DEB/G

(loads DDT)

*filespec /DEB:{FORDDT}/G
{FORTRA}

(loads FORDDT)

*filespec /DEB:(DDT,{FORDDT})/G
{FORTRA}

(loads DDT
and FORDDT)

*filespec /DEB:({FORDDT},DDT)/G
{FORTRA}

(loads FORDDT
and DDT)

you

In the last two command forms shown,
the
first debugging
program specified
(FORDDT or DDT) in the command string is
the one you communicate 'with after the LINK command string
is executed.
See Section 17.9 for
addressing programs.

17.4

information

extended

SCOPE OF NAME AND LABEL REFERENCES

Each program unit has its own symbol table. When you initially enter
FORDDT,
you automatically open the symbol table of the main program.
All references to names or labels through FORDDT must be made with
respect to the currently open symbol table.
If you have given the main program a name other than MAIN.
by using
the PROGRAM statement (see Section 6.4.1), FORDDT asks for the defined
program name.
After you enter the program name,
FORDDT opens the
appropriate symbol
table.
At this point, symbol tables in programs
other than the main program can be opened by using the OPEN command.
References to statement labels,
line numbers,
FORMAT statements,
variables,
and arrays must have labels that are defined in the
currently open symbol table.
However, FORDDT will accept variable and
array references outside the currently open symbol table, providing
the name is unique with respect to all program units in the given load
module.

17-6

USING THE FORTRAN INTERACTIVE DEBUGGER (FORDDT)
17.5

FORDDT COMMANDS

This section gives a detailed description of all commands
The commands are given in alphabetical order:
ACCEPT

FORDDT.

Allows you to change the contents of a
FORTRAN
variable, array, array element, array element range, or
FORMAT statement. The command format is:
ACCEPT name[/mode] value
where:
name

is the variable, substring, array, array
element,
array element range, or FORMAT
statement to be modified.

mode

is the format of the data value to be
entered.
The mode keyword must be preceded
by a slash (I) and
immediately follows the
name.
Intervening blanks are not allowed.
(Note that Imode does not apply to FORMAT
modification.)

value

is the new value to be assigned. The format
of the input value must correspond to the
specified mode.
DATA LOCATION MODIFICATION
Data Modes

The following data modes are accepted:
Mode
A
C

D
F
I

L
R

S
X

Meaning

Example

ASCII (left-justified)
CHARACTER
DOUBLE-PRECISION
REAL
INTEGER
OCTAL
LOGICAL
RASCII (right-justified)
SYMBOLIC
COMPLEX

IFOOI
'ABC'
123.4567890
123.45678
1234567890
7654321
.TRUE. or .FALSE.
\BAR\
PSI(2,4)
(1.25,-78.E+9)

If not specified, the default mode is REAL (F).
1.

Two-Word Values
For the data modes ASCII
(A), OCTAL
(O),
RASCII
(R), and SYMBOLIC (S), FORDDT will accept a "/BIG"
modifier on the mode switch.
This
modifier
indicates that the variable and the value are to be
interpreted as two words long.

FORTRAN-10/20 VII

17-7

February 1987

USING THE FORTRAN INTERACTIVE DEBUGGER (FORDDT)
Example:
ACCEPT VAR/RASCII/BIG '1234567890'
assumes that VAR is two words long and
given l0-character literal into it.

stores

the

The /BIG modifier can also be used to display more
than the first
256 characters of long character
strings.
2.

Character Variables
A character variable can be initialized by using an
ACCEPT command of the following form:
ACCEPT VAR/C 'string'
Note that the length of the variable is that which
is specified in the source program.
If the string
is longer than the
variable,
the
rightmost
characters are truncated.
If the string is shorter
than the variable, it is stored left-justified and
padded on the right with blanks.

Initialization of Arrays
If the name field of an ACCEPT contains
an
unsubscripted array name or a
range of array
elements, all elements of the array
or
the
specified range are set to the given value.
Example:
ACCEPT ARRAY/F 1.0
or
ACCEPT ARRAY(5)-ARRAY(10)/F 1.0
Note that this applies only
ASCII and RASCII.

modes

other

than

For character arrays,
if the value has fewer
characters than the length of the array element,
the rightmost character positions of the element
are initialized with spaces.
If the value has more
characters than the length of the array element,
the value is truncated to the right.
4.

Long Literals
When the value field
of an ACCEPT contains an
unsubscripted
array
name
or range of array
elements, and the specified data mode is ASCII, the
value field
is expected to contain a long literal
string. ACCEPT stores the string linearly into the
array or array range.
If the array is not filled,
the remainder of the array or range is filled
with
zeroes.
If the literal is too long, the remaining
characters are ignored.

FORTRAN-10/20 VII

17-8

February 1987

USING THE FORTRAN INTERACTIVE DEBUGGER (FORDDT)
Example:
ACCEPT ARRAY/ASCII
'ABCDEFGHIJKLMNOPQRSTUVWXYZ'
5.

FORMAT Statement Modification
When the name field of an ACCEPT contains a
label,
FORDDT expects this label to be a FORMAT statement
label and that the value
field
contains a new
format specification.
Example:
ACCEPT 10 (lHO,FIO.2,3(I2))
The new specification cannot be longer than the
space originally allocated
to the FORMAT by the
compiler.
The remainder of the area is cleared
if
the new specification is shorter.
Note that FOROTS performs some encoding of FORMAT
statements when
it processes them for the first
time.
If any I/O statement referencing
the given
FORMAT has been executed, the FORTRAN program has
to be restarted (re-initializing FOROTS).

CHARACTER

Defines the dimensions of a character array.
The
result of this command is the same as for the DIMENSION
command except that the array so dimensioned
is
understood by FORDDT to be a character array.
The
command format is:
CHARACTER arrayname ([Ll:]Ul[,[L2:]U2, ••• ])
NOTE
This declarator cannot be used
to specify
element length.
The length specified in the
user program will be used.

CONTINUE

Allows the program to resume execution after a
FORDDT
pause.
After a CONTINUE
is executed,
the program
either runs to completion or
until another
pause
is
encountered.
The command format is:
CONTINUE [n]
where the n is optional and, if omitted, is assumed
to
be one.
If a value is provided, it can be a numeric
constant or program variable, but it is treated as an
integer.
When the value n is specified, the program
continues execution until the nth occurrence of this
pause.
For example,
CONTINUE 20
continues execution until the 20th occurrence of
pause, or until a different pause is encountered.

. DDT

the

Transfers control of the program to DDT,
the standard
system debugging
program.
Any files currently opened
by FOROTS are unaffected, and a
return to FORDDT is
possible so that program execution may be resumed.
17-9

USING THE FORTRAN INTERACTIVE DEBUGGER (FORDDT)
%FDDT is the global symbol used to
FORDDT.
The command format is:

return

control

%FDDT<ESC)G
Your program will be in the same condition as before
unless you have modified your core image with DDT.
DIMENSION

Sets, displays, or removes the user-defined dimensions
of
an
array for FORDDT access purposes.
These
dimensions need not agree with those declared
to the
compiler
in the source code.
FORDDT allows you to
redimension an array to have a larger scope than that
of the source program.
If this is done, a warning is
given.
NOTE
The DIMENSION command cannot be used to declare
double-precision, . complex, or character arrays
(see the CHARACTER and DOUBLE commands) •
The command format is:
DIMENSION name ([Ll:] Ul [, [L2:] U2, .•• ])
where:
name

is the name of the array

([Ll:]Ul ••• )

specifies the bounds of the
array, where L is the lower
bound and U is the
upper
bound.
The default value of L
is
1.
The bounds must be
integer
constants
or
variables.

For example:
DIMENSION ALPHA(7,5:6,10)
FORDDT remembers the dimensions of the array until they
are redefined or removed.
The command:
DIMENSION
gives a full list of all
for all arrays.

the

user-defined

dimensions

DIMENSION ALPHA
displays the current information for
only.

the

array

ALPHA

DIMENSION ALPHA/REMOVE
removes any user-defined
array ALPHA.

17-10

array

information

for

the

USING THE FORTRAN INTERACTIVE DEBUGGER (FORDDT)
DOUBLE

Defines the dimensions of a double-precision or complex
array.
The result of this command is the same as for
the DIMENSION command except that the
array
so
dimensioned is understood by FORDDT to be an array with
two-word elements.
The command format is:
DOUBLE arrayname ([Ll:]Ul[,[L2:]U2, ••• ])

GOTO

Allows you to continue your program from a point other
than the one at which it last paused.
The GOTO allows
you to continue at a statement label or source line
number provided that the IDEBUG:LABELS switch has been
used or the contents of a
symbol
previously ASSIGNed
during the program execution has been used.
Note that the program must be STARTed before this
command can be used, and also note that a GOTO is not
allowed after the <CTRL/C>,
REENTER sequence
(see
Sec t ion 17.6).
The command format is:
GOTO n

GROUP

Sets up a string of text for input to a
TYPE command.
You can store TYPE statements as a list of variables
identified by the numbers 1 through 8.
This feature
eliminates
the need to retype the same list of
variables each time you wish to examine the same group.
Refer to the TYPE command for the proper format of the
list.
The command format is:
GROUP [n list]
where:
n

is the group number 1-8

list

is a list of group numbers preceded by a
slash
(I)
andlor
variable
names,
substrings, array elements, and array
elemerit ranges to be typed when you give
the command:
TYPE In, where n
is the
group number.
The validity of the list
is not checked.

GROUP
with no arguments causes FORDDT to type out the current
contents of all the groups.
GROUP n
types out
requested.

the

contents

the

particular

group

Note that one group may refer to another.
For example:

FORTRAN-10/20 V11

GROUP 2 VAR2,VAR3
GROUP 3 11,/2
17-11

February 1987

USING THE FORTRAN INTERACTIVE DEBUGGER (FORDDT)
LOCATE

Lists the program unit names in which a given symbol is
defined.
This is useful when the variable you wish to
locate is not in the currently open program unit and is
defined in more than one program unit. The command
forma t is:
LOCATE n
where n may be any FORTRAN variable, array, label, line
number, or FORMAT statement number.

MODE

Defines the display format for succeeding FOR DDT TYPE
commands.
You need type only the first character of
the mode to identify it to FOR DDT
because
all
characters after the first are ignored. The modes are:
Mode
C
F
D
X
I

A
R
L

Meaning
CHARACTER
REAL
DOUBLE-PRECISION
COMPLEX
INTEGER
OCTAL
ASCII (left-justified)
RASCII (right-justified)
LOGICAL

Unless the MODE command is given, the
mode is the REAL (F) fo rma t •

default

typeout

The command format is:
MODE list
where list contains one or more of the mode identifiers
separated by commas.
The current setting can be
changed by issuing another MODE command. If more than
one mode is given, the values are typed out in the
order: C, F, D, X, I, 0, A, R, L.
A typical command string might be:
MODE A,I,OCTAL
NEXT

Allows you to cause FOR DDT to trac~ source lines,
statement
labels,
and
entry point names during
execution of your program. This command only provides
trace facilities if the program is compiled with the
FORTRAN /DEBUG switch. If this switch is not used, the
NEXT command acts as a CONTINUE command. The command
format is:
NEXT [n] [/sw]
where:
n

is a program variable or integer numeric
value

is one of the following switches
/S= statement label
/L= source line
/E= entry point
17-12

USING THE FORTRAN INTERACTIVE DEBUGGER (FORDDT)
The default starting value of n is 1,
statement trace. T~e default switch is IL.

single

The command
NEXT 20/L
traces the execution of the next 20 source line numbers
or until another pause is encountered.
Note that if no argument
argument given is used.

specified,

the

last

For example:
NEXT /E
changes the tracing mode to trace only subprogram
entries using the numeric argument previously supplied.
OPEN

Allows you to open a particular program unit of the
loaded program so that the variables are accessible to
FORDDT.
Any previously opened program unit
is closed
automatically when a new one is opened.
Only global
symbols, symbols in the currently open unit, and unique
locals are available at anyone time.
Note that
starting FOR DDT automatically opens the main program.
The command format is:
OPEN name
where name
is the subprogram name.
arguments reopens the main program.

PAUSE

OPEN

with

Allows you to place a FORDDT breakpoint at a statement
number,
source line number, or subroutine entry point.
Up to ten breakpoints may be set at anyone time. When
a breakpoint is encountered, execution is suspended at
that point and control
is returned to FORDDT.
The
symbol
table of that subprogram is also automatically
opened.
The command formats are:
PAUSE
PAUSE
PAUSE
PAUSE
PAUSE
PAUSE
PAUSE

P
P AFTER n
p IF condition
p TYPING Ig
P AFTER n TYPING /g
P IF condition TYPING /g

where:

P
n

is the point where the breakpoint
inserted
is an
integer constant, variable,
array element
is a group number

PAUSE 100

17-13

is
or

USING THE FORTRAN INTERACTIVE DEBUGGER (FORDDT)
sets a breakpoint at statement label 100, causes
execution to be suspended, and causes FORDDT to be
entered on reaching 100 in the program.
PAUSE #245 AFTER MAX(5)
causes a break to occur at source line number 245 after
encountering this point the number of times specified
by MAX(5).
Note that AFTER can not be abbreviated.
PAUSE DELTA IF LIMIT(3,1) .GT.2.5E-3
causes a
FOR DDT break to occur
if the
variable
LIMIT(3,1)
is greater than the value 2.5E-3.
The IF
can not be abbreviated, and the
following
FORTRAN
logical connectives are allowed:
.GT.,

.GE.,

.LT., .LE.,

.EQ., .NE.

Double-precision comparisons and arithmetic operations
are not allowed.
However, comparisons can be made
between variables, constants,
and logical constants
(such as .TRUE. and .FALSE.).
PAUSE 505 TYPING /5
sets a
FORDDT breakpoint at label 505,
and
the
variables
in group 5 are displayed.
The TYPING
specification can not be abbreviated.
PAUSE #24 AFTER 16 TYPING /3
causes a break at source line number 24 after 16 times
through; however, the contents of group 3 are displayed
every time.
Whe~

the TYPING option is used with the PAUSE command,
control can be transferred
to FORDDT at the next
typeout by typing any character on the terminal.

Note that pause requests remain after a
<CTRL/C>
REENTER sequence, a START command, or a <CTRL/C> START
sequence.
PAUSE ON ERROR Causes the program to enter FORDDT whenever an error
occurs
(such as an arithmetic overflow).
It has the
same command format as the PAUSE command.
REMOVE

Removes the previously
command format is:

set

FORDDT

from

the

breakpoints.

The

REMOVE [p]
For example,
REMOVE L#123
removes a breakpoint
number 123.
REMOVE ALPHA

17-14

program

source

line

USING THE FORTRAN INTERACTIVE DEBUGGER (FORDDT)
removes a
ALPHA.

breakpoint

from

the

subroutine

entry

REMOVE with no arguments removes all your FORDDT
breakpoints, and, in this case, no abbreviation of
REMOVE is allowed.
REMOVE ON ERROR Removes a PAUSE ON ERROR breakpoint. It has the same
command format as-the REMOVE command.
START

Starts your program at the normal FORTRAN main
entry point. The command format is:

program

START
STOP

Terminates the program, closes all files opened by
FOROTS, and causes an exit to the monitor. The usual
command format is:
STOP
STOP /RETURN
allows a return to
devices or closing
issued.

STRACE

monitor mode without releasing
files so that a CONTINUE can be

Displays a subprogram level traceback of the
state of the program. The command format is:

current

STRACE
TYPE

Displays FORTRAN defined variables, substrings, arrays,
or array elements on your terminal. The command format
is:
TYPE list
where list may be one or more variables, substrings, or
array
references
and/or
group
numbers.
These
specifications must be separated by commas, and group
numbers must be preceded by a slash (/). The command
with no arguments uses the last argument list submitted
to FORDDT.
An array element range
example:

can

also

specified.

For

TYPE PI(5)-PI(13)
displays the values from PI(5) to PI(13) inclusive. If
an unsubscripted array name is specified, the entire
array is typed.
There are several methods of choosing the
typeout in conjunction with the MODE command:
1.

If you do not specify
real.

You can specify a format through the
described in this chapter.

FORTRAN-10/20 VII

17-15

format,

the

form

default

MODE

command

February 1987

USING THE FORTRAN INTERACTIVE DEBUGGER (FORDDT)
3.

You can change the format(s) previously designated
by the MODE command by including print modifiers in
the TYPE or GROUP string. The print modifiers are:
A
B
C
D
F
I
L

R
X

ASCII(left-justified)
LONG
CHARACTER
DOUBLE-PRECISION
REAL
INTEGER
LOGICAL
OCTAL
RASCII(right-justified)
COMPLEX

If you type a variable in mode CHARACTER (C), the
number of characters printed is equal to the length
declared in the FORTRAN source program, up to a maximum
of 256 characters.
The IB switch can be used to
override the 256 character maximum.
The B switch may be used in conjunction with the A, 0,
and R switches.
This modifier indicates that the
variable is to be interpreted as two words long. The B
switch can also be used with the C switch to display
more than the first 256 characters of long character
strings. The B switch can not be used alone.
The first print modifier specified in a string of
variables determines the mode for the entire string
unless another mode is placed directly to the right of
a particular variable. For example, in:
TYPE II K,L/O,M,N/A,/2
the typeout mode is integer
specified. Therefore,

until

another

mode

K, M are integer - the default mode for group 2 is
integer
L is OCTAL
N is ASCII
WHAT

Displays on your terminal the name of the currently
open program unit, any currently active breakpoints,
any group specifications, and any user-set
array
dimensions. The command format is:
WHAT

17.6

ENVIRONM·ENT CONTROL

If a program enters an infinite loop, you can recover by typing a
<CTRL/C> (twice) REENTER sequence.
This action causes FORDDT to
simulate a pause at the point of reentry and allows you to control
your run-away program.

17-16

USING THE FORTRAN INTERACTIVE DEBUGGER (FORDDT)
Most commands can be used once the program has been reentered;
however, GOTO,
STRACE, TYPE, and ACCEPT cause transfer of control to
routines external to FORDDT.
No guarantee can be made to ensure that
any of these commands following a <CTRL/C> REENTER sequence will not
destroy the program integrity.
The program must be returned to a
stable state before any of these four commands can be issued.
In
order to restore program integrity, you should set a pause at the next
label and then CONTINUE to it.
If the /DEBUG:TRACE switch is used, a
NEXT 1 command can be issued to restore program integrity.

17.7

FORTRAN /OPTIMIZE SWITCH

You should never attempt to use FOR DDT with a program that has been
compiled with the /OPTIMIZE switch.
The global optimizer causes
variables to be kept in ACs.
For this reason, attempts to examine or
modify variables in optimized programs will not work.

17.8

CALLING FORDDT

FORDDT can be called directly
appropriate statement is:

from

user

FORTRAN

program.

The

CALL FORDDT
where no argument is required.
FORDDT must be loaded and
initialized
before a CALL to FORDDT is made.
This is done by starting the program
in debug mode prior to the first call (see Section 17.3, item 1). All
FORDDT commands are allowed. A CONTINUE will resume normal execution
of the user program (similar to a RETURN from a subroutine).
NOTE
Since FORDDT is defined as a global symbol,
users
should be careful
if they decide to use FOR DDT as a
program, subroutine, or function name.

17.9

FORDDT AND FORTRAN-20 EXTENDED ADDRESSING

FORDDT VI0 has been modified to be able to run
in any section and
access data and code in all sections.
The user interface to FORDDT is
the same regardless of whether or not a program uses extended
addressing.
FORDDT VI0 is section independent. The same FORDDT.REL will
either section 0 or a non-zero section.

work

If a program is loaded with the /DEBUG:FORDDT option,
LINK loads
FORDDT.REL with the program.
FORDDT.REL is a single-segment module
(it has only low segment code); therefore, when loaded with a
FORTRAN
object program that was compiled with the /EXTEND switch (see Section
16.5), FORDDT, by default, is redirected to the .DATA. psect.

17-17

USING THE FORTRAN INTERACTIVE DEBUGGER (FORDDT)
FORDDT Version 10 will not
versions of FORTRAN-IO/20.

guaranteed

work

with

FORDDT and FORLIB must be in the same section.
Since they would by
default go into the .DATA./.CODE. section, the user normally would not
need to be concerned about this.
However, you should be cautious when
you use the LINK /REDIRECT switch.
NOTE
The first page of any section that
reserved for FORDDT and FOROTS.

17-18

contains

code

CHAPTER 18
USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)

This chapter describes the facilities that the FORTRAN Object Time
System
(FOROTS) provides for the FORTRAN user.
FOROTS implements all
standard FORTRAN I/O operations as set forth
in the FORTRAN~77
standard.
In addition,
it provides the user with capabilities and
programming features beyond those defined in the ANSI standard.
The primary function of FOROTS is to act as a direct interface between
user-object programs and the TOPS-10 or TOPS-20 monitor during input
and output operations.
Other capabilities include:
1.

Job initialization

Channel and memory management

Error handling and reporting

File management

Formatt ing of data

Mathematical library

User library (nonmathematical)

Specialized applications packages

Overlay facilities

10.

RMS-20 interface (TOPS-20 only)

11.

Network file access (TOPS-20 only)

FOROTS runs on any TOPS-10 or TOPS-20 system.
all TOPS-l~ or TOPS-20 peripheral devices.

18.1

FOROTS interfaces

with

FEATURES OF FOROTS

The following list briefly describes many specific features of FOROTS;
more detailed
information concerning
the
implementation of these
features is given later in this chapter.
1.

Your program can run in either batch or timesharing mode
without requiring a program change. All differences between
batch-mode and timesharing-mode operations are resolved by
FOROTS.

FORTRAN-10/20 VII

18-1

February 1987

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
2.

Your programs can access both
devices in the same manner.

FOROTS helps provide complete data file compatibility between
all system devices.

FOROTS treats devices located at remote stations in the
way it treats local devices.

directory

and

nondirectory

same

Programs

written for magnetic
tape operations will
run
on disk under FOROTS supervision.
FOROTS simulates
the commands needed for magnetic tape operations.

c~rrectly

18.2

You may change or specify object program device
specifications with a FOROTS interactive dialogue.

and

Non-FORTRAN binary data files may be read in
FOROTS.

mode

FOROTS
provides
interactive
program/operating
system
error-processing
routines.
These routines permit you to
route
the
execution
of
the
program
to
specific
error-processing routines whenever designated types of errors
are detected.

An error traceback facility for
fatal
errors provides the
active execution path (by subroutine calls) between the main
program and the subroutine where the fatal error occurred.

10.

FOROTS provides a
trap-handling system
for
arithmetic
functions, including default values and error reports.

11.

FOROTS permits your program to switch from READ to WRITE
the same I/O device without loss of data or buffering.

12.

Although primarily designed for use with the FORTRAN-10/20
object programs,
you can also use FOROTS as an independent
I/O system, and as an I/O system for MACRO object programs.

13.

FOROTS provides an
interface to RMS-20
(TOPS-20 Record
Management Services) for I/O to RMS-20 sequential, relative,
and indexed file organizations.

14.

FOROTS allows transparent access
from TOPS-20 systems to
remote RMS and non-RMS files residing on other TOPS-20 and
VMS systems.

IMAGE

file
by

ERROR PROCESSING

Whenever a run-time error is detected,
the FOROTS error-processing
system takes control of program execution.
This system determines the
class of the error and either outputs an appropriate message at the
controlling
terminal or branches the program to a predesignated
processing routine.

FORTRAN-10/20 VII

18-2

February 1987

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
18.3

INPUT/OUTPUT FACILITIES

On TOPS-10, FOROTS uses monitor-buffered I/O for SEQUENTIAL, SEQINOUT,
SEQIN,
and
SEQOUT files access, and uses dump mode I/O for
DIRECT (RANDOM) , RANDIN, and dump mode files access.
On TOPS-20, FOROTS uses PMAP monitor calls for disk files access other
than APPEND,
and uses monitor-buffered I/O for all other file
accesses.
On TOPS-20, FOROTS uses the RMS-20 subsystem for disk I/O if the
ORGANIZATION keyword specifies an RMS file organization, or if the
file is remote.
The following sections describe I/O data channel and access modes.

18.3.1

Input/Output Channels Used By FOROTS (TOPS-10 Only)

FOROTS uses extended channels starting
operations.
User programs can request
through the ALCHN. and FUNCT. routines.

FORTRAN-10/20 VII

18-2.1

at channel 20 for
I/O
I/O channels 0 through 17

February 1987

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)

This page intentionally left blank.

18-2.2

February 1987

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
When a request is made for an I/O channel, a table is scanned until a
free channel
is found.
The first free channel is assigned to the
requesting program.
On completion of the assigned transfer,
control
of the I/O channel is returned to FOROTS by using the DECHN. routine.

18.3.2

File Access

Data can be transferred between processor storage and
devices using either sequential or direct (random) access.

18.3.3

peripheral

Closing Files After Non-standard Termination

When a FORTRAN program is aborted by <CTRL/C) or an error, open files
cannot be closed with the monitor command CLOSE.
The following
command should be used:
REENTER
FOROTS then asks if you want the files to be closed.
YES, then, the files are closed.

you

answer

18.3.3.1 Sequential
Access - In
a
sequential-access
transfer
operation,
the records involved are transferred in the same order as
they appear in the source file.
Each I/O statement executed transfers
the record immediately following the last record transferred from the
accessed source file.
A type of the sequential access
is available for output
(write)
operations.
This type of access is called APPEND and is specified by
the OPEN statement specifier ACCESS='APPEND'
(see Section 11.3.1).
APPEND lets you write a record immediately after the last logical
record of the accessed file.
During APPEND transfer,
the records
already in the accessed file
remain unchanged; the only function
performed is the appending of the transferred records to the end of
the file.
You must specify transfer types (other than SEQINOUT) by setting the
ACCESS option of a FORTRAN OPEN statement to one of several 'possible
arguments.
For the sequential access, the arguments are:
ACCESS='SEQIN'
ACCESS='SEQOUT'
ACCESS='SEQUENTIAL'
ACCESS='SEQINOUT'
ACCESS='APPEND'

(file is opened for read-only access)
(file is opened for output)
(file is opened for input or output)
(same as SEQUENTIAL)
(sequential append access)
NOTE

A common way to append data to a file opened with
SEQUENTIAL access is to read past the end of file, and
then begin writing.
The FORTRAN-77 standard requires
that a BACKSPACE operation be done to back over the
'end file record' preceding the WRITE.

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
l8.~.3~2
Direct (Random) Access' Mode - Direct access permits records
to be accessed and transferred from a source file in any desired
order. Direct access can only be used with disk files that have been
set up for direct access.
Direct-access files must contain a
specified number of identically sized records that may be individually
accessed by a record number.

Direct-access transfers may be done in either a read/write direction
or a special read-only direction. You must specify random transfer
direction by setting the ACCESS option of an OPEN statement (see
Section 11.3.1) to one of several possible arguments.
ACCESS='DIRECT' (direct read/write access)
ACCESS='RANDOM' (same as DIRECT)
ACCESS='RANDIN' (direct read-only access)
ACCESS='KEYED'
(direct indexed-file read/write access)

18.4

ACCEPTABLE TYPES OF DATA FILES AND THEIR FORMATS

The following sections describe the
acceptable to FOROTS.

18.4.1

types

data

files

that

are

ASCII Data Files

Each record within an ASCII data file consists of a set of contiguous
7-bit characters. A vertical paper-motion character (that is, a form
feed, a vertical tab, or a line feed) terminates each set.
Logical
records may be split across physical blocks. There is no maximum
length for logical records.

18.4.2

FORTRAN Binary Data Files

Each logical record in a FORTRAN binary data file contains data that
the executing program can reference with either a READ or a WRITE
statement. A logical record is preceded and ended by a control word
and may have one or more control words embedded within it. In FORTRAN
binary data files, there is no relationship between logical records
and physical device block sizes. There is no implied maximum length
for logical records.

18.4.2.1 Format of Binary Files - A FOROTS binary file can contain
three forms of Logical Segment Control Words (LSCW). These LSCWs give
FOROTS the ability to distinguish ASCII files from binary files.
The
value in the high-order 9 bits of an LSCW tells what kind of LSCW it
is: START, CONTINUE, or END.
START
CONTINUE
END

FORTRAN-10/20 VII

LSCW
001+ the number of words in the segment including
the START LSCW word (exclusive of the END
LSCW)
002+ the number of words in the segment including
the CONTINUE LSCW
003+ the number of words in the whole record
incl ud ing all LSCWs

18-4

"February 1987

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
If the access you specify for a
file
through the OPEN statement
ACCESS= argument
is
'SEQIN', 'SEQOUT', 'SEQUENTIAL', or 'SEQINOUT',
all three LSCWs can appear in a record.
If you specify a record size,
all records are of the same length, and there are no CONTINUE LSCWs.
The following examples illustrate the LSCW.
file contains only 001 and 003 LSCWs.
C
C

The direct-access

LOOK AT A BINARY FILE AND SEE THE LOGICAL SEGMENT
CONTROL WORDS.
OPEN(UNIT=l,ACCESS='DIRECT' ,MODE='BINARY',
1
RECORDSIZE=lOO)
1=5
WR1TE(I'I)
J=7
WR1TE(l'2)
END

(I, J=l,lO)
(J,K=I,lO)

18-5

binary

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
0/
1/
2/
3/
4/
5/
6/
7/
10/
11/
12/
13/
14/
15/
16/
17/
20/
21/
22/
23/
24/
25/
26/
27/
30/
31/
32/
33/
34/
35/
36/
37/
40/
41/
42/
43/
44/
45/
46/
47/
50/
51/
52/
53/
54/
55/
.56/
57/
60/
61/
62/
63/
64/
65/
66/
67/
70/
71/
72/
73/
74/
75/
76/
77/

001000,000145~Number

000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000

of
words in
record
counting
START LSCW

100/
101/
102/
103/
104/
105/
106/
107/
110/
111/
112/
113/
114/
115/
116/
117/
120/
121/
122/
123/
124/
125/
126/
127/
130/
131/
132/
133/
134/
135/
136/
137/
140/
141/
142/
143/
144/
145/
146/
147/
150/
151/
152/
153/
154/
155/
156/
157/
160/
161/
162/
163/
164/
165/
166/
167/
170/
171/
172/
173/
174/
175/
176/
177/

18-6

000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
003000,000146~END

001000,000145
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000
000000,000000

LSCW
containing
the number
of words in
the record
including
all LSCWs.

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)

200/
201/
202/
203/
204/
205/
206/
207/
210/
211/
212/
213/
214/
215/
216/
217/
220/
221/
222/
223/
224/
225/
226/
227/
230/
231/
232/
233/
234/
235/
236/
237/
240/
241/
242/
243/
244/
245/

246/
247/
250/
251/
252/
253/
254/
255/
256/
257/
260/

L61/
262/
263/
264/
265/
266/
267/
270/
271/
272/
273/
274/
275/
276/
277/
300/
301/
302/
303/
304/
305/
306/
307/
310/
311/
312/
313/

18-7

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
On TOPS-10, in the sequential-access binary file,
the second record
crosses the disk block boundary and contains an 002 (CONTINUE) LSCW.
On TOPS-20, the CONTINUE LSCW occurs on buffer boundaries, whose size
is determined by the BUFFERCOUNT keyword in the OPEN statement (see
Section 11.3.5) (default is four pages, 4000 octal words).
C
C

LOOK AT A BINARY FILE AND SEE THE LOGICAL SEGMENT
CONTROL WORDS.
OPEN(UNIT=l,MODE='BINARY')
1=5
WRI~E(l)

(I, J=l,lOO)

J=7
WRITE(l)
END

(J,K=l,lOO)

18-8

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
0/
1/
2/
3/
4/
5/
6/
7/
10/
11/
12/
13/
14/
15/
16/
17/
20/
21/
22/
23/
24/
25/
26/
27/
30/
31/
32/
33/
34/
35/
36/
37/
40/
41/
42/
43/
44/
45/
46/
47/
50/
51/
52/
53/
54/
55/
56/
57/
60/
61/
62/
63/
64/
65/
66/
67/
70/
71/
72/
73/
74/
75/
76/
77/

001000,000145
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005

001000,000032~Number

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
200/
201/
202/
203/
204/
205/
206/
207/
210/
211/
212/
213/
214/
215/
216/
217/
220/
221/
222/
223/
224/
225/
226/
227/
230/
231/
232/
233/
234/
235/
236/
237/
240/
241/
242/
243/
244/
245/

002000,000114~Continue

000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007

LSCW

246/
247/
250/
251/
252/
253/
254/
255/
256/
257/
260/
261/
262/
263/
264/
265/
266/
267/
270/
271/
272/
273/
274/
275/
276/
277/
300/
301/
302/
303/
304/
305/
306/
307/
310/
311/
312/
313/
313/

18-10

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
Image files contain no LSCWs.
You can only backspace
that is created with a record size.
C
C

IMAGE

LOOK AT AN IMAGE MODE FILE AND SEE NO LOGICAL SEGMENT
CONTROL WORDS.
OPEN(UNIT=l,MODE='IMAGE')
I=5
WRITE(l)
J=7
WRITE(l)
END

(I, J=l,lOO)
(J,K=l,lOO)

18-11

file

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
0/
1/
2/
3/
4/
5/
6/
7/
10/
11/
12/
13/
14/
15/
16/
17/
20/
21/
22/
23/
24/
25/
26/
27/
30/
31/
32/
33/
34/
35/
36/
37/
40/
41/
42/
43/
44/
45/
46/
47/
50/
51/
52/
53/
54/
55/
56/
57/
60/
61/
62/
63/
64/
65/
66/
67/
70/
71/
72/
73/
74/
75/
76/
77/

000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005

100/
101/
102/
103/
104/
105/
106/
107/
110/
111/
112/
113/
114/
115/
116/
117/
120/
121/
122/
123/
124/
125/
126/
127/
130/
131/
132/
. 133/
134/
135/
136/
137/
140/
141/
142/
143/
144/
145/
146/
147/
150/
151/
152/
153/
154/
155/
156/
157/
160/
161/
162/
163/
164/
165/
166/
167/
170/
171/
172/
173/
174/
175/
176/
177/
18-12

000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000005
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007

244/
245/
246/
247/
250/
251/
252/
253/
254/
255/
256/
257/
260/
261/
262/
263/
264/
265/
266/
267/
270/
271/
272/
273/
274/
275/
276/
277/
300/
301/
302/
303/
304/
305/
306/
307/

18-13

000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,0000'07
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007
000000,000007

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
The following example illustrates the
binary files.
C
C

LSCWs

for

character

data

LOOK AT BINARY MODE FILE WITH CHARACTER DATA AND SEE THE
LOGICAL SEGMENT CONTROL WORDS
OPEN (UNIT=l,MODE='BINARY')
WRITE (I) 3, 'ABCDEF' ,4,'GHIJKL'
WRITE (1)
END

0/
1/
2/
3/
4/
5/
6/
7/
10/
11/
12/
13/
14/

'MNOPQR' ,'STUVWX'

001000,000007
000000,000003
406050,342212
430000,000000
000000,000004
436211,145226
460000,000000
003000,000010
001000,0000e4
466351,750242
512472,452654
536600,000000
003000,000005

The following example illustrates the format
image files.
Image 'files contain no LSCWs.
C
C

character

data

LOOK AT IMAGE MODE FILE WITH CHARACTER DATA AND SEE
NO LOGICAL SEGMENT CONTROL WORDS
OPEN (UNIT=l,MODE='IMAGE')
WRITE (1) 3, 'ABCDEF' ,4,'GHIJKL'
WRITE (1)
END

0/
1/
2/
3/
4/
5/
6/
7/

18.4.3

'MNOPQR' ,'STUVWX'

000000,000003
406050,342212
430000,000000
000000,000004
436211,145226
462331,647640
506452,352252
532573,000000

RMS Data Files (TOPS-20 only)

The FOROTS/RMS-20 interface can be used to access RMS-20
(and
remote
VMS RMS-32) files with sequential, relative, or indexed organizations.
,An RMS record file consists of a prologue section and a data section.
The prologue section is the repository of the file's attributes.
The
data section contains data records (and indexes, if applicable).
See the TOPS-20 RMS User's Guide for
formats.
FORTRAN-10/20 VII

18-14

information

RMS

file

February 1987

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
18.5

USING FOROTS

FOROTS has been designed to lenq itself for use as an I/O system for
programs written
in languages other than FORTRAN.
Currently, MACRO
programmers may employ FOROTS as a general
I/O system by
writing
simple MACRO calls that simulate the calls made to FOROTS by a FORTRAN
compiler.
The calls made to FOROTS are to routines that implement
FORTRAN I/O statements such as READ, WRITE, OPEN, or CLOSE.
FOROTS will provide automatic memory allocation, data conversion,
buffering, and device interface operations to the MACRO user.

18.5.1

I/O

FOROTS Entry Points

FOROTS provides the following entry points for
FORTRAN compiler or a non-FORTRAN program.
contained in FORLIB.REL.
Table 18-1:

calls from either a
These entry points are

FOROTS Entry Points

Entry Point

Function

ALCHN.

Allocates an I/O channel
for
use
subroutine (see Section 18.5.3.12)

ALCOR.

Allocates memory (see Section 18.5.3.11)

CLOSE.

Closes a file.
In a FORTRAN program, this call
is
made when the CLOSE statement is executed (see
Section 18.5.3.10)

DBMS.

DBMS interface.

DEC.

DECODE routine.
This call, coupled with an
call, handles decoding.

DECHN.

Deallocates an I/O channel that was
ALCHN (see Section 18.5.3.12)

DECOR.

Deallocates memory that was allocated by
call (see Section 18.5.3.11)

DELTR.

Deletes an RMS relative or indexed file record.
In
a FORTRAN program, this call is made when a DELETE
statement is executed (see Section 10.13).

ENC.

ENCODE routine

EXIT.

Closes all
files,
clears interrupt system, and
terminates
program
execution.
In a
FORTRAN
program, this call is made when an END statement is
executed in the main program.

EXITI.

Writes out buffers, closes and unmaps all files

FIN.

I/O list termination routine (see Section 18.5.3.9)

FORTRAN-10/20 VII

18-15

MACRO

IOLIST

obtained
an

from
ALCOR

February 1987

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
Table 18-1:

FOROTS Entry Points (Cont'd)

Entry Point

Function

FIND.

FIND statement

FORER.

Error processor

FOROP.

Miscellaneous FOROTS utilities

FUNCT.

OTS-independent
interface
to
provide
common
functions
(like memory and I/O channel management)
for programs such as overlay handler and DBMS.
FUNCT. is an interface that works the same way with
FORTRAN, COBOL, and ALGOL run-time systems
(see
Section 18.6).

IFI.

Internal file input (see Section 18.5.3.2)

IFO.

Internal file output (see Section 18.5.3.2)

IN.

Formatted input routine
(see Sections
18.5.3.5, 18.5.3.6, and 18.5.3.8)

IOLST.

I/O list routine (see Section 18.5.3.9)

MTOP.

REWIND,
BACKSPACE,
Section 18.5.3.7).

NLI.

NAMELIST input routine (see Section 18.5.3.3)

NLO.

NAMELIST output routine (see Section 18.5.3.3)

OPEN.

Opens a file.
Connects FORTRAN Logical Unit Number
to a file for I/O (see Section 18.5.3.10)

OUT.

Formatted output routine
(see Sections
18.5.3.5, 18.5.3.6, and·18.5.3.8)

RESET.

Job initialization entry

REWRF.

Formatted REWRITE output routine.
In a FORTRAN
program, this call is made when a formatted REWRITE
statement is executed for RMS files
(see Section
10.14.1) •

REWRU.

Unformatted REWRITE output routine.
In a FORTRAN
program,
this call
is made when an unformatted
REWRITE statement is executed for RMS files
(see
Sec t i on 1 0 • 14 • 2) •

RTB.

Binary input routine
18.5.3.5)

TRACE.

Traces subroutine calls

UNLOC.

UNLOCK record routine for
RMS files.
UNLOCK statement, Section 10.15.)

WTB.

Binary output routine (see
18.5.3.5)

FORTRAN-10/20 VII

18-16

and

(see

ENDFILE

18.5.3.1,

statements

Sections

(see

18.5.3.1,

18.5.3.1

and

(See

the

18.5.3.1

and

February 1987

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
18.5.2

Calling Sequences

You must use the following general form for all calls made to FOROTS:
XMOVEI
PUSHJ

16,ARGBLK
17,Entry Point
(control is returned here)

where:
ARGBLK

is the address of a
specifically
formatted
argument block that contains information needed by
FOROTS to accomplish the desired operation.

Entry Point

is an entry point identifier (see Table 18-1) that
specifies the entry point of the desired FOROTS
routine.

With three exceptions, all returns from FOROTS will be made to the
program instruction immediately following the call (PUSHJ 17, entry
point instruction).
The exceptions are:
1.

An error return to a specified statement number,
that
READ or WRITE statement ERR=option (see Section 10.4.7),

is,

An end-of-file return to a statement number, that is, READ or
WRITE statement END=option (see Section 10.4.6),

A fatal error that returns to the monitor.

Sections 18.5.3.1 through 18.5.3.12 give the MACRO calls and
required
argument block formats needed to
initialize FOROTS and FOROTS I/O
operations.

18-16.1

February 1987

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)

This page intentionally left blank.

18-16.2

February 1987

USING THE FORTRAN
18.5.3

OBJ~CT

TIME SYSTEM (FOROTS)

MACRO Calls for FOROTS Functions

The following sections describe the forms of the MACRO calls to FOROTS
that are made by the FORTRAN compiler.
The calls described are
identified by the language statement that they implement.
The
following
terms and abbreviations are used in the description of the
argument block (ARGBLK) of each call:
pointer to the second word in the argument block
(This
is the address pointed to by the argument ARGBLK in the
calling sequence.)

FORTRAN logical unit number

count of ASCII characters

FORMAT statement address

list

an Input/Output list

name

a NAMELIST name

a variable specifying the
random access mode

list-directed I/O (the FORMAT statement is not used)

type

type specification of a variable or constant

logical

record

number

for

The argument block for all I/O statements is a sequence of keyword
specifiers.
Bits 2-8 of each argument specify which argument is being
supplied, as follows:
1
2
3
4
5
6
7
10
11
12
13
14
15
16

UNIT
FMT address
FMT size (in words)
END= address
ERR= address
IOSTAT= address
REC=
NAMELIST table address
File-positioning function code
ENCODE/DECODE array address
Internal record length
Relational type for indexed READ
Key number of key for indexed READ
Key relational value

FORTRAN-10/20 VII

18-17

February 1987

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
The format of ARGBLK is:
0-1

2-S

9-12

14-17

(-count - negative of number of words in
block not including this one)
~

1 a

ARGBLK:

··
·· ··
1 a

wnere:
I =
y =
kwd

lS-35

kwd

type

kwd

type

indirection bit
IS-bit address or data
keyword number
NOTE

Future versions of FOROTS will not support argument
blocks with index registers specified either in the
arguments or in memory locations referenced indirectly
by these arguments. Arguments must not reside in the
ACs. In addition, so-called 'immediate' arguments
(those
with a type code of zero) will not be
supported.

18.5.3.1 Sequential-Access Calling Sequences - The READ and WRITE
statements for formatted sequential data transfer operations and their
calling sequences are:
READ(u,f) list
XMOVEI 16, ARGBLK
PUSHJ 17, IN.
and
WR I TE ( u , f) 1 i s t
XMOVEI 16, ARGBLK
PUSHJ 17, OUT.
The following arguments must be specified in ARGBLK:
1
2
3

UNIT
FMT address
FMT size

The following may also appear:
4

ERR
END
IOSTAT

18-18

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
The READ and WRITE statements for unformatted sequential data transfer
operations and their calling sequences are:
READ(u) list
XMOVEI 16, ARGBLK
PUSHJ 17, RTB.
and
WRITE(u)

list

XMOVEI 16, ARGBLK
PUSHJ 17, WTB.
The following arguments must be specified in ARGBLK:
1

UNIT

The following may also appear:
4
5
6

END
ERR
IOSTAT

18.5.3.2 Internal File Calling
Sequences - The READ and
WRITE
statements for formatted sequential data transfer operations using
internal files and their calling sequences are:
READ ( u , f)

1i s t

XMOVEI 16, ARGBLK
PUSHJ 17, IFI.
and
WR I T E (u , f)

1i st

XMOVEI 16, ARGBLK
PUSHJ 17, IFO.
The following arguments must be specified in ARGBLK:
1

2
3

UNIT (must be a character scalar or array)
FMT address
FMT size

The following may also appear:
4
5

ERR
END
IOSTAT

18-19

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
lS.5.3.3 NAMELIST I/O, Sequential-Access Calling Sequences - The READ
and WRITE statements for NAMELIST-directed sequential data transfer
operations and their calling sequences are:
READ (u, name)
XMOVEI 16, ARGBLK
PUSHJ 17, NLI.
and
WRITE (u, name)
XMOVEI 16, ARGBLK
PUSHJ 17, NLO.
The following arguments must be specified in ARGBLK:
1
10

UNIT
NAMELIST address

The following may also appear:
4
5
6

END
ERR
IOSTAT

The NAMELIST table is generated from the FORTRAN NAMELIST statement.
The first word of the table is the NAMELIST name; following that are a
number of 2 word entries for scalar variables, and a number of
(N+4)
word entries for array variables, where N is the dimensionality of the
array.
The names you specify in the NAMELIST statement are stored, in SIXBIT
format,
first
in the table.
Each name
is followed by a list of
arguments associated with the name.
The NAMELIST table is terminated
by a zero entry.
The name argument list can be in either a scalar or
an array form.

18.5.3.4 Array Offsets and Factoring - Address calculations used to
reference a given array element
involve factors and offsets.
For
example:
Array A is dimensioned
DIMENSION A (Ll:Ul,L2:U2,L3:U3, ••• Ln:Un)
The size of each dimension is represented by:
Sl
S2

UI-Ll+l
U2-L2+1

In order to calculate the address of an element referenced by:
A (Il,I2,I3, ••• In)
the following formula is used:
A+(I1-L1)+(I2-L2)*Sl+(I3-L3)*S2*Sl+ ••• +(In-Ln)*S[n-1]* ••• *S2*Sl
lS-20

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
The terms are factored out depending on the dimensions of
not on the element referenced, to arrive at the formula:

the

array,

A+(-Ll-L2*Sl-L3*S2*Sl ••• )+Il+I2*Sl+I3*S2*Sl •••
The parenthesized part of this
formula
is the offset
for
single-precision array; it is referred to as the Array Offset.

For each dimension of a giverr array, there is a corresponding
factor
by which a subscript in that position will be multiplied.
From the
last expression, one can determine the factor for dimension n to be:
S[n-l]*S[n-2]* ••• *S2*Sl
For double-precision and complex arrays, the expression becomes:
A+2*(Il-Ll)+2*(I2-L2)*Sl+2*(I3-L3)*S2+Sl+ •••
Therefore, the array offset for a double-precision array is:
2*(-LI-L2*Sl-L3*S2*Sl ••• )
and the factor for the nth dimension is:
2*S[n-l]*S[n-2]* ••• *S2*Sl
The factor for the first dimension of a double-precision array is
always 2.
The factor for the first dimension of a single-precision
array is always 1.
For character arrays, the offset is calculated
in bytes
instead of
words.
The byte offset from the start of a character array whose
elements are of length X is:
X*«II-Ll)+(I2-L2)*S2+ ••• )
This offset is X times the offset of a single-precision numeric array.
NAMELIST Table
0

35
NAMELIST name in SIXBIT
NAMELIST entry 1
NAMELIST entry 2

·
·
·
NAMELIST entry n
4000,,0 (FOROTS FIN. word)

18-21

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
SCALAR ENTRY in a NAMELIST Table

o 1

9-12

18-35

SIXBIT/SCALAR NAME/

1 0

Scalar addr

ARRAY ENTRY in a NAMELIST Table
0-1 1 2-8

9- 12 1 13

1 14-17

18-35

SIXBIT/ARRAY NAME/
1 0

I IDIMS

1 type 1 I

1
ARRAY SIZE

BASE ADDR

OFFSET
Factor 1
Factor 2

·
· n
Factor
·

18.5.3.5 I/O Statements, Direct-Access Calling Sequences - The READ
and WRITE statements for formatted direct-access data transfers and
their calling sequences are:
READ (u I r , f)

1 is t

XMOVEI 16, ARGBLK
PUSHJ 17, IN.
and
WRITE

CUi r ,f)

list

XMOVEI 16, ARGBLK
PUSHJ 17, OUT.
The following arguments must be specified in ARGBLK:
1
2
3

UNIT
FMT address
FMT size
REC

18-22

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
The following may also appear:
4
5
6

END
ERR
IOSTAT

The READ and WRITE statements for unformatted direct-access
and their calling sequences are:

transfers

READ (u I r)
XMOVEI l6,ARGBLK
PUSHJ P,RTB.
and
WRITE (u I r)
XMOVEI l6,ARGBLK
PUSHJ P,WTB.
The following argument must be specified in ARGBLK:
1
7

UNIT
REC

The following may also appear.:
4
5
6

END
ERR
IOSTAT

18.5.3.6 Default Devices Statements, Calling Sequences - The FORTRAN
statements that require the use of a reserved system default device
and their calling sequences are:
Default Device
ACCEPT f, list
READ f, 1 i st
REREAD f, list

UNIT=-4
UNIT=-5
UNIT=-6

(TTY)
(CDR)
(REREAD)

UNIT=-3
UNIT=-2
UNIT=-l

(LPT)
(PTP)
(TTY)

XMOVEI 16, ARGBLK
PUSHJ 17, IN.
and
PRINT f, list
PUNCH f, list
TYPE f, list
XMOVEI 16, ARGBLK
PUSHJ 17, OUT.
The arguments for these calls are the same as
formatted sequential READ and WRITE statements.

18-23

for

the

standard

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)

18.5.3.7 Statements to Position Files - The formatted and unformatted
FORTRAN statements that can be used to control the positioning of
files and their calling sequences are:
Function
(FORTRAN Statement)
SKIPFILE (u)
BACKFILE (u)
BACKSPACE (u)
ENDFILE (u)
REWIND (u)
SKIPRECORD (u)
UNLOAD (u)

FOROTS Code
7
3
2

o
5
1

CALL:
XMOVEI 16, ARGBLK
PUSHJ 17, MTOP.
The following arguments must be specified in ARGBLK:
1
11

UNIT
FOROTS code

The following may also appear:
4
5
6

END
ERR
IOSTAT
NOTE
For disk files, UNLOAD is the same as REWIND; BACKFILE
and SKIPFILE are ignored.

18.5.3.8 List-Directed Input/Output Statements - You may write any
form of a sequential
I/O statem~nt as a list-directed statement by
replacing the referenced FORMAT statement number with an asterisk (*).
The list-directed forms of the READ and
calLing sequences are:
READ (u, *)

WRITE

statements

and

their

1i s t

XMOVEI 16, ARGBLK
PUSHJ 17, IN.
and
WR I T E ( u, *)

1i s t

XMOVEI 16, ARGBLK
PUSHJ 17, OUT.
The arguments for these calls are the same as for
the standard
formatted sequential READ and WRITE statements, except that the FORMAT
statement address and FORMAT statement size must be specified as zero.

18-24

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
18.5.3.9 Input/Output Data Lists - The compiler generates a calling
sequence to the run-time system for the I/O list in a READ or WRITE
statement. The argument block associated with the calling sequence
contains the addresses of the variables and arrays to be transferred
to or from an I/O buffer.
The general form of an I/O list calling sequence is:
XMOVEI 16, ARGBLK
PUSHJ 17, IOLST.
Any number of elements may be included in the ARGBLK.
The end of
argument block is specified by a zero entry or a FIN entry.
Mnemonic Name

FOROTS Value

DATA
SLIST
ELIST
FIN
F77 SLIST
F77 ELIST

the

3
4

5
6

The elements of an I/O list are:
1.

DATA
The DATA element converts one single- or double-precision or
complex item from external
to
internal
form for a READ
statement and from internal to external
form for
a WRITE
statement.
Each DATA element has the following format:
0-1
1

2-8

9-12

14-17

DATA

type

18-35
SCALAR AD DR

SLIST
The SLIST argument, converts an entire array from internal
to
external
form or vice versa, depending on the type of
statement (that
is,
READ or WRITE)
involved.
An SLIST
consist of a table of arguments that has the following form:
0-1

2-8

9-12

14-17

SLIST

type

#ELEMENTS

a
1 a

type

INCREMENT

type

BASE ADDRI

1
1

18-25

18-35

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
For example, the sequence:
DIMENSION A(lOO) ,B(lOO)
READ(-,-)A
or
READ(-,-) (A(I) ,I=l,lOO)

!only when the /OPT switch is used

develops an SLIST argument of the form:
0-1

2-8

9-12

14-17

18-35

1
1
1
0

2
0
0
4

2
2
2
0

0
0
0
0

[144]
[1 ]
A
0

0
0
0
0

More than one base address may appear in a SLIST as
the increment is the same. The sequence:
DIMENSION A(lOO), B(lOO)
WRITE (-,-) (A(I) ,B(I) ,I=lOO)

long

!on1y when the /OPT
switch is used

develops an SLIST argument of the form:

3.'

0-1

2-8

9-12

14-17

18-35

1
1
1
1
0

2
0
0
0
4

2
2
2
2
0

0
0
0
0
0

[144]
[1 ]
A
B
0

0
0
0
0
0

ELIST
The SLIST format permits only a single increment to be
specified for a number of arrays, while the ELIST permits
different increments to be specified for different arrays.
An ELIST consists of a table of arguments that has the
following form:
The format of the ELIST is:
0-1

2-8

9-12

14-17

1 0

ELIST

type

No. Elements to
transfer
increment 1

1 0

type

Base ADDR 1
increment 2

1 0

type

Base AD DR 2
increment N

1 0

type

Base ADDR N

18-26

18-35

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
For example, the FORTRAN sequence:
DIMENSION IC(6,100), IB(lOO)
WR I T E ( - , - ) ( I B ( I) , I C ( 1 , I) , I = 1 , 1 0 0 )
produces the ELIST:
0-1

2-8

9-12

14-17

18-35

1
1
1
1
1
0

3
0
0
0
0
4

2
2
2
2
2
0

0
0
0
0
0
0

[144]
[1 ]
IB
[6]
IC
0

This could

0
0
0
0
0
0

The increment may be zero.

produced

the

s~quence:

WRITE(-,-) (K,I=I,lOO)

!only when the /OPT switch is used

Produces the ELIST:

0-1

2-8

9-12

14-17

18-35

1 0

[144]

1 0

[0]

1 0

0 0

FIN
The end of an I/O list is indicated by a FIN
the I/O processor interprets this element, it
to FIN to terminate the I/O. This call must
each I/O initialization call, including calls
list.

element.
When
performs a call
be made after
with a null I/O

The FIN routine may be entered by an explicit call or by an
argument in this I/O list argument block. Both calls can not
be used. The FIN element has the following format:
9-12

14-17

18-35

EXPLICIT CALL:
PUSHJ 17, FIN.

18-27

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
5.

F77 SLIST
This is the same as SLIST except that if the number
elements is less than or equal to zero, no I/O is done.

F77 ELIST
This is the same as ELIST except that is the number
elements is less than or equal to zero, no I/O is done.

18.5.3.10 OPEN and CLOSE Statements, Calling Sequences - The form and
calling sequences for the OPEN and CLOSE FORTRAN statements are:
OPEN statement call:
XMOVEI 16, ARGBLK
PUSHJ 17, OPEN.
CLOSE statement call:
XMOVEI 16, ARGBLK
PUSHJ 17, CLOSE.
where ARGBLK is:
0-1

2-8

9-12

14-17

Negative of
the number
of words in
block not
including
this one.

1 0
1 0
1 0

G
G
G

type I
type I
type I

X
X
X

·· ··
1· 0·

·
·
·
·
G

· ·
· ·

··

18-35

···
X

·· ··

type I

H
H
H

··
·
·

The G field (bits 2 through 8)
contains a value that defines the
argument name;
the H field (bits 18 through 35) contains an address
that points to the value of the argument.
Note that the G field
values for OPEN and CLOSE statements are not the same as those for
other I/O statements.

18-28

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
The numer ic codes that may appear in the G field are:
G Field

Open Argument

01
02
03
04
05
06
07
10
11
12
13
14
15
16
17
20
21
22
23
24
25
26
27
30

G Field

DIALOG
ACCESS=
DEVICE=
BUFFERCOUNT=
BLOCKSIZE=
FILE=
PROTECTION=
DIRECTORY=
LIMIT=
MODE=
FILESIZE=
RECORDSIZE=
DISPOSE=
VERSION=
ORGANIZATION=
SHARED
IOSTAT=
ASSOCIATEVARIABLE=
PARITY=
DENSITY=
BLANK=
CARRIAGECONTROL=
FORM=
BYTESIZE=

31
32
33
34
35
36
37
40
41
42
43
44
45
46
47
50
51
52
53
54
55
56
57

Open Argument
PADCHAR=
RECORDTYPE=
STATUS=
TAPEFORMAT=
READONLY=
UNIT=
ERR=
EXIST=
FORMATTED=
NAMED=
NEXTREC=
NUMBER=
OPENED=
SEQUENTIAL=
UNFORMATTED=
NAME=
KEY=
USEROPEN=
DIALOG=
DEFAULTFILE=
KEYED=
NOSPANBLOCKS
MAXREC=

18.5.3.11 MemQry Allocation Routines - The memory management module
is called
to allocate or deallocate memory blocks.
There are two
entry points (ALCOR. and DECOR.) that control memory allocation and
deallocation.
When TOPS-20 extended addressing is in effect, ALCOR. and DECOR.
can
be used; however, memory will be allocated in FOROT's section instead
of
in the
user's section.
You
can
use
the
LINK
switch
/OTSEGMENT:NONSHARABLE to put FOROTS in the user's section.
Use the ALCOR.
entry to allocate the number of words specified in the
argument block variable.
Upon return, AC 0 will contain either the
address of the allocated memory block or -1,
which
indicates that
memory is not available.
The calling sequence for an ALCOR. call is:
XMOVEI 16, ARGBLK
PUSHJ 17, ALCOR.
where ARGBLK is:
0-1

2-8

9-12

14-17

-1
1 0

Reserved

FORTRAN-10/20 VII

18-35
0

type

18-29

Address of
number of words
February 1987

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
Use the DECOR. entry to deallocate a previously allocated block of
memory;
the argument variable must be loaded with the address of the
memory block to be returned.
If the number of desired words is N, ALCOR. actually removes N+2 words
from free storage.
The pointer returned points to the third word
(word 2 as opposed to word 0) removed from free storage.
The first
two words are used by FOROTS to maintain linked lists of allocated
(using ALCOR.) and free storage, and must not be modified.
The calling sequence for a DECOR. call is:
XMOVEI 16, ARGBLK
PUSHJ 17, DECOR.
where ARGBLK is:
0-1

2-8

9-12

14-17

-1

1 0

18-35
0

Reserved type

Pointer to word
containing
address of block
to be returned

18.5.3.12 Channel Allocation and Deallocation Routines - You may
allocate software channels in MACRO programs by means of calls to the
ALCHN. routine and deallocate them by calls to the DECHN.
routine.
Values are returned in AC O.
Use the ALCHN. entry to allocate a particular channel or the next
available channel.
The channel to be allocated is passed as an
argument to ALCHN.
Zero is passed as an argument to allocate the next
available channel.
Allowed channels are 1 through 17 (octal).
If the
channel requested is not available, or all channels are in use, ALCHN.
returns with a -1
in AC O.
In normal returns, AC 0 contains the
assigned number.
The calling sequence of an ALCHN. routine is:
XMOVEI 16, ARGBLK
PUSHJ 17, ALCHN.
where ARGBLK is:
0-1

2-8

9-12

14-17

-1

-..

1 0

Reserved type

18-35
0

18-30

Pointer to a word
containing
the channel #
or zero

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
Use the DECHN. entry to deallocate a previously assigned channel.
The
channel to be released
is passed as an argument to DECHN.
If the
channel to be deallocated was not assigned by ALCHN. and thus cannot
be deassigned, AC 0 is set to -Ion return.
The calling sequence for a DECHN. routine is:
XMOVEI 16, ARGBLK
PUSHJ 17, DECHN.
where ARGBLK is:
0-1

9-12

2-8

14-17

-1

1 0

18.6

18-35
0

Reserved type

Pointer to a word
containing
the channel #
to be released

FUNCTIONS TO FACILITATE OVERLAYS

FOROTS provides a subroutine (FUNCT.) to serve as an interface with
the LINK overlay handler.
This subroutine consists of a group of
functions that allow the overlay handler to perform I/O, memory
management, and error message handling.' These functions have only one
entry point, FUNCT.i and they are called by the sequence:
XMOVEI 16, ARGBLK
PUSHJ 17, FUNCT.
The format of the ARGBLK is:
ARGBLK:

-<n+3>,,0
IFIW 2,address of integer function code
IFIW l7,address of 3-letter ASCII error prefix
IFIW 2,address of status code on return
IFIW type,address of first argument
IFIW type,address of nth argument

where:
type
function code
error prefix
status

is the FORTRAN argument type (see Chapter 14)
is the number of one of the required functions
is ignored by FOROTS
is undefined on the call and set on the return
with one of the values below.
-1

l •••• n

Function not implemented
Successful return
Specific error message

When TOPS-20 extended addressing is in effect,
FUNCT.
can be used;
however, memory will be allocated in FOROTS's section instead of in
the
user's
section.
You
can
use
the
LINK
switch
/OTSEGMENT:NONSHARABLE to put FOROTS in the user's section.

18-31

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
Table 18-2:
Function
Number
(Octal)

Function Numbers and Function Codes

Function
Mnemonic

0
1
2

ILL
GAD
COR

3
4
5
6

RAD
GCH
RCH
GOT

7
10

ROT
RNT

IFS

CBC

13
14
15
16

RRS
WRS
GPG
RPG

17
20
21
22

GPSI
RPSI
MPG
UPG

Function Description
Illegal function; returns -1 status
Gets a specific segment of memory
Gets a given amount of memory from
anywhere
in the space allocated to the
overlay handler
Returns a specific segment of memory
Gets an~I/O channel
Returns an I/O channel
Gets memory from the space allocated to
the object-time system
Returns memory to the object-time system
Gets the
initial
run time
from the
object-time system
Gets
the
initial
time
file
run
specification
from
object-time
the
system
Cuts back memory (if possible) to reduce
job size
Reads RETAIN status (DBMS)
Writes RETAIN status (DBMS)
Allocates memory on a page boundary
Deallocates memory obtained by function
15
Gets TOPS-20 PSI channel
Returns TOPS-20 PSI channel
Gets a contiguous set of pages
Returns a contiguous set of pages

Each function of the FUNCT.
subroutine
is described below.
The
arguments described in each of the following functions are what should
be in the addresses pointed to by the argument block described above.
ILL CO) FUNCTION - This function is illegal.
ignored, and the status returned is -1.

The

argument

block

GAD (I) FUNCTION - This function gets memory from a specific address
in the space allocated to the overlay handler.
The arguments are:
arg 1
arg 2

address of requested memory
address of number of words of memory to allocate

A call to GAD with arg 2 equal to -1 requests all available memory.
On return, the status is one of the following:

o
1
2

successful allocation
not enough memory available
memory not available at specified address
illegal arguments (such as, address + size
than 256K)

18-32

greater

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
COR (2) FUNCTION - This function gets memory from any available
allocated to the overlay handler.
The arguments are:
arg 1
arg 2

space

undefined (address of allocated memory on return)
address of size of requested allocation

On return, the status is one of the following:

memory allocated
not enough memory available
illegal argument (that is, size is greater than 256K)

RAD (3) function - This function
returns
address within the space allocated to
arguments are:
arg 1
arg 2

memory at the specified
the overlay handler. The

address of memory to be return
address of size of memory to be returned (in words)

On return, the status is one of the following:

successful return of memory
memory cannot be returned
illegal argument (that is,the address or
greater than 256K)

I
3

GCH (4)
are:

FUNCTION - This function gets an I/O channel.

arg I
arg 2

undefined
return)
ignored

(address

channel

number

the

size

The

arguments

allocated

On return, the status is one of the following:

a
1

successful channel allocation
no I/O channels available

RCH (5) FUNCTION - This
arguments are:
arg 1
arg 2

function

returns

I/O

channel.

The

address of number of channel to be returned
ignored

On return, the status is one of the following:

a
1

channel released
invalid channel number

GOT (6) FUNCTION - This function gets memory from the space
to the object-time system.
The arguments are:
arg 1
arg 2

allocated

undefined (address of allocated memory on return)
address of size of memory requested

On return, the status is one of the following:

a
1

successful allocation
not enough memory available
illegal argument (such as, size is greater than 256K)

18-33

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
This function differs from function
1
in that
if the object-time
system has two free memory lists, then function 1 is used to allocate
space for links, and this function is used to allocate space
for
I/O
buffers.
Function 1 uses the free memory list for LINK, and function
6 uses the list for the object-time system.
ROT (7) FUNCTION - This function returns
system.
The arguments are:
arg 1
arg 2

memory

the

object-time

address of memory to be returned
address of size of memory to be returned (in words)

On return, the status is one of the following:

successful return of memory
memory cannot be returned
illegal argument (such as, the address or the
greater than 256K)

size

RNT (10) FUNCTION - This function returns the initial
run time
from
the object-time system.
(At the beginning of the program, the
object-time system will have executed a RUNT 1M UUO; that result is the
time returned by RNT.) The arguments are:
arg 1
arg 2

undefined (contains address
return)
ignored

initial

run

On return, the run time is in arg 1, and the status is O.
is

time

The

status

IFS (11) FUNCTION (TOPS-IO only) - This function returns the
initial
run-time
file
specification from
the object-time system.
(This
initial file specification is the one used to begin the program;
that
is, it was given with the system RUN command.) The arguments are:
arg 1
arg 2
arg 3

undefined (address of SIXBIT device on return)
undefined (address of SIXBIT filename on return)
undefined
(project-programmer number on return
address of path block if run from SFD

On return, the status is one of the following:

successful return
error

This function tells the overlay handler which file to read
initial RUN command.

after

CBC (12) FUNCTION - This function cuts back memory if possible,
reduces the size of the user job.
There are no arguments.

the
which

The returned status is:

o
RRS (13)
WRS (14)

always
FUNCTION
FUNCTION - These functions are reserved for use by DBMS.

GPG (15) FUNCTION - This function gets memory on a page boundary.
arguments are the same as for FUNCTION 6, GOT.
arg 1
arg 2

ignored
address of number of words of memory to allocate
18-34

The

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
On return, arg 1 has the address of the allocated memory.
It will be
on a page boundary.
(That is, it will be a multiple of 1000 octal.)
On return, the status is one of the following:

a
1
3

RPG
15.

successful
not enough memory available
illegal argument

(16) FUNCTION - This function returns memory obtained by
The arguments are the same as for FUNCTION 7, ROT.
arg 1
arg 2

FUNCTION

address of memory to be returned
address of size of memory to be returned (in words)

On return, the status is one of the following:

a
1

successful return of memory
was not allocated
illegal argument

GPSI (17) FUNCTION - This function gets the TOPS-20 PSI channel.
It
assigns a software
interrupt channel number.
GPSI provides only
controlled access to the PSI tables.
It arranges that the tables
exist and that SIR and EIR have been done, but does not do AIC or any
other JSYS necessary to set up the channel
(ATI or MTOPR,
for
example) •
arg 1
arg 2
arg 3

address of channel number to allocate,
or -1
to
allocate any user-assignable channel
address of level number
address of interrupt routine (if the FORTRAN program is
running on a system that supports extended addressing,
the interrupt routine address may be a 30-bit address
in any section,
including section O.
Otherwise, the
interrupt address must be an IS-bit address.)

On return, the status is one of the following:

a
1
2

allocated OK (arg 1 is the channel number
sent)
requested channel was already assigned
no free channels
argument error

-1

was

RPSI (20) FUNCTION - This function returns the TOPS-20 PSI channel.
It returns a PSI channel allocated by FUNCTION 17. RPSI provides only
controlled access to the PSI tables.
It
removes the given channel
from the tables.
This function does not do DIC or any other JSYS
necessary to remove an interrupt condition from a channel.
arg 1

address of channel number to return

On return, the status is one of the following:

a
1

OK
channel was not allocated
argument error

MPG (21) FUNCTION - This function gets a contiguous set of pages. The
pages requested are always allocated from the section FOROTS is in.
The user cannot depend upon this call to either create or destroy the
pages.

lS-35

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
arg 1

first page number to allocate.
in the range 0 to 777.

arg 2

number of pages to allocate

The page number must be

On return, the status is one of the following:

successful allocation of all given pages
one or more pages were already allocated
illegal argument (bad page number or count)

1
3

UPG (22) FUNCTION- This function returns a contiguous set of pages.
The pages returned are considered to be in the section FOROTS is in.
The user cannot depend upon this call to either create or destroy the
pages.
arg 1

first page number to deallocate.
be in the range 9 to 777

arg 2

number of pages to deallocate

The page number

must

On return, the status is one of the following:

successful deallocation of all given pages
one or more pages was not allocated by MPG
illegal argument (bad page number or count)

1
3

18.7

LOGICAL/PHYSICAL DEVICE ASSIGNMENTS

You make FORTRAN logical and physical device assignments at run time,
or standard system assignments are made according to a FOROTS Device
Table, that is,
DEVTB.
Table 10-3 in Section 10.4.3 shows the
standard assignments contained by the Device Table.

18.8

FOROTS AND INQUIRE BY FILE STATEMENT

See Section 11.7 for a description of the INQUIRE statement.
If no device is given for the FILE= specifier, FOROTS uses DSK: as the
default.
If no extension is given, FOROTS uses .DAT.
For TOPS-20, if
no generation number is given, FOROTS uses a generation number of O.
FOROTS determines if the device specified is a disk.
a disk, the following happens:

If the device is

if a
file
with
file
FOROTS determines
exists
the
• specification
given in the INQUIRE statement.
It returns the
answer (either .TRUE.
or .FALSE.) in the variable specified
by the EXIST= specifier,
if any.
If such a file exists,
FOROTS 'expands' the file specification as follows:
•

A logical name is translated into a physical device name.

•

For TOPS-20, the file specification, which is overlaid by
the
user-specified directory,
filename, extension, and
generation.

•

For TOPS-20,
an actual
file generation
number
substituted for a generation number of 0, -1, or -2.
18-36

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
The resultant file specification, in string form,
the 'full (expanded) file·string.'
•

called

FOROTS searches for a match between the file specification
given in the
INQUIRE statement and a file specification
associated with a
logical
unit for which there
is
a
"connection." This is to determine the values to be returned
for the INQUIRE specifiers OPENED= and NUMBER=
(see Section
11.7.3). FOROTS looks at all FORTRAN logical units for which
there is a connection in ascending order, starting with zero.
FOROTS compares the file specification given in the INQUIRE
statement
(with FILE= defaults applied) with the exact file
specification given in the OPEN statement
(with
FILE=
defaults applied) if the following is true:
•

The file does not exist on the directory.

•

An OPEN statement has been executed and STATUS='UNKNOWN'
and ACCESS='SEQUENTIAL' (see Section 11.3.1).

•

No data transfer statements have been executed using
unit.

the

If the file exists, FOROTS compares the full
file
string
associated with the unit with the full (expanded) file string
given in the INQUIRE statement.
The file exists
if the
following is true:
•

An OPEN has been executed with STATUS other
'UNKNOWN' or ACCESS other than 'SEQUENTIAL'.

•

An I/O transfer statement has been executed.

than

If neither of these two comparisions are successful,
FOROTS
returns the current unit number in the variable specified
with the NUMBER= specifier, and returns 'YES' in the variable
specified with the OPENED= specifier.
If the same file is
connected on several units, the matching technique described
will
return the smallest unit number for which there is a
match.
For non-disk devices specified in the INQUIRE statement file
string
specification, FOROTS looks at all the FORTRAN logical units for which
there is a connection in ascending order, starting with zero.
If the device in the file string specified in the INQUIRE statement is
not
the
user's controlling terminal,
FOROTS expands the
file
specification given in the INQUIRE statement by translating a
logical
name given as the device into its corresponding physical name.
It
then compares the device part of this expanded file specification with
the device part of the full (expanded) file string associated with the
logical unit.
If the device is the user's controlling
terminal
(device
'TTY'),
FOROTS determines if the device associated with the logical unit is
also the user's controlling terminal.

18-37

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
18.9

USEROPEN PROCEDURES

The USEROPEN keyword in a FORTRAN OPEN statement (see Section 11.3.39)
provides a way to access RMS-20 facilities that are otherwise not
available to FORTRAN programs.
The USEROPEN keyword specifies a user-written external procedure
controls the opening of a file.
It has the form:

that

USEROPEN = procedure-name
where procedure-name is the symbolic name of a user-written function.
The procedure must be declared in an EXTERNAL statement (see Section
7.6) •

When an OPEN statement (wi th or wi thout the USEROPEN keyword)
is
executed on an RMS file, FOROTS uses the OPEN statement keywords to
establish the RMS File Access Block (FAB), the Record Access Block
(RAB), and optionally one or more Extended Argument Blocks (XABs).
If a USEROPEN keyword is included in the OPEN statement, FOROTS
establishes defaults for some of the data structure fields, then calls
your USEROPEN routine instead of opening the file according to its
normal defaults.
The USEROPEN routine can then provide additional
parameters to RMS, as well as read values from the FAB, RAB, and XABs
as returned by RMS.
FOROTS will call a USEROPEN routine
following form:
xM:

with

argument

list

the

can

perform

the

-3, ,0
IFIW TP%INT,address of unit
IFIW TP%LBL,address of FAB
IFIW TP%LBL,address of RAB

Using these arguments, the
following operations:

USEROPEN

routine

then

Modify the FAB/RAB or XAB(s).

Issue the $OPEN or $CREATE RMS service calls (depending on
whether the file exists or is to be created), followed by the
$CONNECT RMS service call to prepare the file for record I/O.
(Required)

Check the status indicators returned by RMS after each of the
above
service
calls.
Your
procedure
should return
immediately with the RMS error code in AC0 if RMS returns a
failure status.
(Required)

Read file attribute information returned by RMS to
RAB, or XAB(s).
(Optional)

Return a success or failure status value in
AC0 must be zero on success.
(Requi red)

18.9.1

(Optional)

AC0

the
to

FAB,

FOROTS.

Example of a USEROPEN

The following example shows a FORTRAN OPEN statement calling a
USEROPEN routine UOPN, which changes the default "No duplicate"
attribute for the primary index in an indexed file.

FORTRAN-10/20 VII

18-38

February 1987

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
EXTERNAL UOPN
OPEN
(UNIT=l,FILE='IDX.DAT' ,ORGANIZATION='INDEXED',
1
STATUS='NEW' ,KEY=(1:10) ,RECL=40,
1
USEROPEN=UOPN)
A USEROPEN routine may cause a first-time
invocation of RMS.
To
ensure that RMS is invoked compatibly with FOROTS, the following code
must be placed in all USEROPEN and MACRO routines that may invoke RMS:
SEARCH

RMSINJ

iRMS/dynamic library universal file

.REQUIRE SYS:RMSZER
.REQUIRE SYS:RTLZNM
.REQUIRE SYS:ZERBOO
.REQUIRE SYS:DYNBOO

iDynamic library boot-strap files

The following function UOPN creates an RMS indexed file after setting
the FLG XAB field in the primary index XAB to allow duplicate primary
keys:
TITLE UOPN
SEARCH RMSINJ,MACSYM,MONSYM
INTERN UOPN
.REQUIRE SYS:RMSZER
.REQUIRE SYS:RTLZER
.REQUIRE SYS:ZERBOO
.REQUIRE SYS:DYNBOO
STDAC.

iStandard AC definitions

T0==0
UOPN:

XMOVEI
MOVX
$STORE

Tl,@1(16)
T2,FB$SUP
T2,FOP,«Tl»

iGet FAB address
i Get "Supersede existing" bit
i Set in FAB

$FETCH
MOVX
$STORE

Tl,XAB,«Tl»
T2,XB$DUP
T2,FLG,«Tl»

iGet address of 1st XAB in chain
iGet "allow dupe" bit
;Set in XAB

$CREATE @1(16)
T2,ER$MIN
CAlL
BADRET
JRST

;Create the file
;Error?
iYes

$CONNECT @2(16)
CAlL
T2,ER$MIN
JRST
BADRET

;CONNECT RAB
iError?
iYes

SETZ
POPJ

T2,
P,

;Return success

T0,T2
P,

;Return error code in AC0

BADRET: MOVE
POPJ
END

UOPN

To allow an existing file to be superseded with a higher generation,
the routine first turns on the FB$SUP bit in the FAB.
It then fetches
the address of the first XAB in the XAB chain (in this case the only
XAB)
and sets the XB$DUP bit in the FLG field to allow duplicate
primary keys (the RMS/FOROTS default is to disallow duplicate primary
keys) •

FORTRAN-10/20 VII

18-39

February 1987

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
For local RMS indexed files, the XAB field of the FAB will always
point to the first key XAB in any XAB chain. For all remote files,
the XAB field of the FAB points to a CONFIG XAB containing network
system information; the NXT field of the CONFIG XAB will point to the
start of the key XAB chain.
Refer to the RMS-20 programmer's
Reference Manual and the TOPS-20 RMS User's Guide for more information
of RMS data structures.

18.9.2

RMS/FOROTS Data and Control Structures

When RMS is invoked to access a file on a particular unit, FOROTS
allocates and initializes the following RMS data structures for each
such uni t:
•

FAB - File Access Block

•

RAB - Record Access Block

•

XAB - Extended Attribute Block.
the file is indexed or remote.

One or more are allocated if

In addition, a statically allocated NAM and TYP block are linked to
unit's FAB block during OPEN processing.

Values in these control structures are set or defaulted by FORTRAN as
described below. In addition, RMS may update these control structures
with information derived from the file after it is opened.
Unless an error intervenes, these data structures are not deallocated
until the unit is closed.
User-written subroutines and USEROPEN
functions may alter or examine field values in these structures, but
they must not alter their allocation. Caution should be exercised
when changing fields that have been set as a result of FORTRAN
keywords (such as MRS or FAC values), since unpredictable results may
occur during subsequent I/O.
In general, if a desired value can be
obtained through the use of a FORTRAN keyword, use the keyword.
The following RMS FAB, RAB and XAB fields are either initialized by
FORTRAN and/or examined upon return from an RMS service call. Fields
not listed here take the RMS defaults, or are returned by RMS.
FAB FIELDS
Field

Name

FORTRAN OPEN keyword/value

F$BSZ

Bytesize

n if BYTESIZE=n;
else
TOPS-20, 8 for VMS files

F$CTX

Context

Reserved for Digital

F$FAC

File access

Set at OPEN depending on STATUS,
ACCESS
and
READONLY or SHARED
keywords. Set to FB$GET for read
else
access
files;
to
FB$GET+FB$PUT+FB$DEL+FB$UPD+FB$TRN
FB$BRO
files

FORTRAN-10/20 VII

18-40

also

for

remote

..,

for

non-RMS

February 1987

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
FILE= filename + DEFAULTFILE= file
name if present, else FORnn.DAT if
neither given

F$FNA

File specification
string address

F$FOP

File-processing options

FB$SUP Supersede

Set at OPEN depending on STATUS and
ACCESS keywords

FB$DLT Delete on close

Set at CLOSE depending on DISPOSE
keyword in OPEN or CLOSE, or STATUS
keyword in CLOSE

FB$SPL Spool to printer

Set at CLOSE depending on DISPOSE
keyword in OPEN or CLOSE, or STATUS
keyword in CLOSE

FB$SCF Submit command

Set at CLOSE depending on DISPOSE
keyword in OPEN or CLOSE, or STATUS
keyword in CLOSE

F$MRS

Maximum record size

For RMS files,
set
to
RECORDSIZE=n; else to 0

F$MRN

Maximum record number

For RM.8 files,
set
MAXREC=n; else to 0

F$NAM

Name block address

Set to address of name block;
both
the expanded and resultant string
areas are set up

F$TYP

Type block address

Set to address of type block for
remote non-RMS files.
T$CLA is set
to TY$IMA

F$ORG

File organization

FB$IDX if ORGANIZATION='INDEXED'
FB$REL if ORGANIZATION='RELATIVE'
FB$SEQ if ORGANIZATION='SEQUENTIAL'
or omi tted

F$RAT

Record attributes

FB$BLK Do not cross page
boundaries

I if NOSPANBLOCKS

F$RFM

Record format

FB$FIX if RECORDTYPE='FIXED'
FB$VAR if RECORDTYPE='VARIABLE'
FB$STM if RECORDTYPE='STREAM'
FB$FIX
if
no
RECORDTYPE
and
ORGANIZATION
is
'RELATIVE'
or
'INDEXED'

F$SHR

File sharing

FB$PUT Allow other
FB$GET Allow other
FB$DEL Allow other
FB$UPD Allow other

FORTRAN-10/20 VII

PUTs
GETs
DELETEs
UPDATEs

I
I
I
I

if
if
if
if

18-41

SHARED
SHARED
SHARED
SHARED

February 1987

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
F$XAB

o if the file is not indexed or

Extended attribute

not remote.
contains the
XAB, whose
beginning of
any.
For
contains the
key XAB.

is
For all remote files,
address of a CONFIG
N$NXT points to the
the key XAB chain, if
local indexed files,
address of the first

RAB FIELDS
Field

Name

FORTRAN OPEN keyword/value

R$BKT

Bucket code

Set to n
processing

R$CTX

Context

Reserved for DIGITAL

R$FAB

FAB address

Set to
block

R$KBF

Key buffer address

Set to address of record number if
ACCESS='DIRECT' or address of key
data if ACCESS='KEYED'

R$KRF

Key of reference

Set to key of reference for indexed
files

R$KSZ

Key size

Set to
files

R$MBF

Multibuffer count

n if BUFFERCOUNT=n, else 0

R$RAC

Record access mode

RB$KEY
cKEYED'

during

address

size

remote

non-RMS

file

argument

key

for

indexed

ACCESS='DIRECT'

RB$SEQ
if
ACCESS='SEQUENTIAL',
'APPEND' or ACCESS is omitted.
RB$BLK for remote non-RMS files
R$RBF

Record buffer address

R$ROP

Record-processing options

Set during READ or WRITE

RB$EOF End-of-file
RB$KGE Key .GE.
RB$KGT Key .GT.

R$RSZ

Record size

Set during READ and WRITE

R$UBF

User record area

Set during READ and WRITE address

R$USZ

User record area size

Set during READ and WRITE

FORTRAN-10/20 VII

if ACCESS='APPEND'
If indexed READ
If indexed READ

18-42

February 1987

USING THE FORTRAN OBJECT TIME SYSTEM (FOROTS)
The following KEY XAB fields are set by FORTRAN for indexed files:
XAB FIELDS
Field

Name

FORTRAN OPEN keyword/value

X$DTP

Key datatype

XB$STG if CHARACTER (default); else
XB$IN4 if INTEGER

X$FLG

Key options flags

XB$DUPIXB$CHG
index

X$REF

Key of reference

Set depending on OPEN KEY= position
of key specifier

X$POS

Key position

From OPEN KEY= specifier

X$SIZ

Key size

From OPEN KEY= specifier

X$NXT

Link address

Address of next XAB in chain, or 0

FORTRAN-10/20 VII

18-43

for

each

secondary

February 1987

CHAPTER 19
USING THE FORTRAN REAL-TIME SOFTWARE (TOPS-10 ONLY)

19.1

INTRODUCTION

The FORRTF library subroutines are designed to allow the timesharing
FORTRAN user to do real-time programming on TOPS-IO systems.
These
subroutines, described in Section 19.3, are listed below:
LOCK
RTINIT
CONECT
RTSTRT
BLKRW
RTREAD
RTWRIT
STATO
STATI
RTSLP
RTWAKE
DISMIS
DISCON
UNLOCK
With these subroutines, the timesharing job can dynamically connect
real-time devices to the Priority Interrupt (PI) system, respond to
these devices at interrupt level,
remove the devices from
the PI
system,
and change their
PI level.
Use of these routines requires
that you have real-time privileges and are able to lock your job in
core.
The privilege bits required are:
JP.RTT
JP.LCK

(bit 13) - real-time privileges
(bit 14) - locking privileges

The number of real-time devices that can be handled at one time is an
assembly-t ime
constant
(RTDEVN)
in
the
FORRTF source.
The
DIGITAL-distributed software has RTDEVN equal to 2,
but
it can be
changed
(up to 6) by editing the statement "RTDEVN==2" in FORRTF.MAC
and reassembling.
The error messages output by FORRTF can be
in either full message
format or coded
format
(refer to Table 19-1).
Use of the code and
format saves over 100 words of run-time core.
If core is limited,
reassembly of FORRTF.MAC with the assembly-time constant SHORT changed
from the DIGITAL-distributed a (full
format)
to -1
(coded
format)
accomplishes the core saving.
On multiprocessor systems, the real-time traps apply only to the
processor
specified
by
the job's CPU specification.
If the
specification indicates more than one processor, the specification is
changed to indicate CPUO.
Note that the priority interrupt channel is
only for the indicated CPU.
19-1

USING THE FORTRAN REAL-TIME SOFTWARE (TOPS-IO ONLY)
19.2

USING FORRTF

Users of FORTRAN-IO real-time software must consider the following:
1.

Use of memory

Device control in block or single mode

Priority-interrupt levels

Masks

19.2.1

Memory

The job being executed must be locked
in memory with the LOCK
subroutine
(see Section 19.3.1). Any data being read into memory can
only be read into the low segment and above the protected
job data
area
(the first
140 locations).
The BLKRW subroutine (see Section
19.3.5) tests the validity of the locations specified to receive data
in block reading to ensure that no overwritings occur.
However, when in block mode, the block pointer must be reset before
dismissing
the end-of-block interrupt; otherwise, all memory could be
overwritten.

19.2.2

Modes

Real-time jobs can control their devices in one of two ways:
block
mode or single mode.
In block mode, an entire block of data is read
or written before the
user-interrupt
routine
is run;
whereas,
in
single mode,
the user-interrupt program is run every time the device
interrupts.
There are two types of block mode:
fast-block mode and normal-block
mode.
A device
in fast-block mode requires that a PI channel be
dedicated entirely to itself.

19.2.3

priority-Interrupt Levels

Priority-interrupt levels 1 through 6 are legal depending on the
system configuration.
The lower the number of the level, the higher
the priority of that level.
Programs that execute
for
a long
time
should not be put on high-priority interrupt levels, since they could
cause other
real-time programs on lower levels to lose
data.
Specification of the PI level as zero for a particular device causes
the device to be removed from the PI system.

19.2.4

Masks

For a description of the bits included
in the startmsk and
intmsk
parameters of RTSTRT and the status word in STATO and STATI, see the
DECsystern-l0 Hardware Reference Manual.

19-2

USING THE FORTRAN REAL-TIME SOFTWARE (TOPS-IO ONLY)
19.3

SUBROUTINES

Each of the 14 subroutines associated with FORTRAN real-time software
is described briefly in Sections 19.3.1 through 19.3.14.
These
subroutines have been programmed to be compatible with programs
written according to the TOPS-IO Monitor Calls Manual.

19.3.1

LOCK

LOCK locks the
job in memory and allocates and
initializes the
internal controlling tables for all real-time devices.
LOCK must be
called before any of the other real-time routines, and must be called
exactly once.
The form of the LOCK subroutine is:
CALL LOCK

19.3.2

RTINIT

RTINIT initializes the internal tables controlling a real-time device.
RTINIT must be called for each iDdividual device being used.
The form of the RTINIT subroutine is:
CALL RTINIT (unit, dev, pi, trpadr, intmsk)
where:
unit

is the real-time device unit number (any number from 1
to RTDEVN).
This number is not connected in any way
with the FORTRAN logical unit number.

dev

is the device code for the real-time device
DECsystem-lO Hardware Reference Manual) •

is the priority-interrupt level on which the real-time
device
is to be run.
Each individual device
in
fast-block mode must have a level dedicated to
itself.
If the level
is equal
to
zero, the device will be
removed from the priority-interrupt system altogether.

(see

the

If it is necessary to connect one device to several
levels simultaneously,
a negative value for PI tells
the system not to remove any other occurrences of the
device from any other (or the same) PI level.
(Note
that this counts as another real-time device.)
trpadr

is the address of a FORTRAN entry to which real-time
interrupts are to trap.
This can be a function or
subroutine subprogram.
Any variables that must be
shared
between
the
user-level
code
and
the
interrupt-level routine must be passed by means of
COMMON.
Passing them as parameters causes disastrous
results.

19-3

USING THE FORTRAN REAL-TIME SOFTWARE (TOPS-IO ONLY)
intmsk

is the mask of all interrupting flags for the real-time
This is actually set up by RTSTRT and should
be zero whenever the real-time device is inactive (that
is,
in a call
to RTINIT,
except
in the case of
fast-block mode).
In fast-block mode, intmsk must be
set to -1.

devic~.

19.3.3

CONECT

CONECT tells the system to connect a real-time device to the proper PI
level
and sets up several elements of the device-controlling tables.
Every device must be CONNECTED.
The form of the CONECT subroutine is:
CALL CONECT (unit, mode)
where:
unit

is the real-time device unit number (see RTINIT).

mode

is either:
-2, write a block of data, fast mode; then interrupt.
-1, write a block of data, normal mode; then interrupt.
0, interrupt every word
+1, read a block of data, normal mode; then interrupt.
+2, read a block of data, fast mode; then interrupt.

19.3.4

RTSTRT

RTSTRT can be used to start a real-time device, as well as to stop it
and
zero its interrupt mask.
A device must be started to be used and
should be stopped before it is disconnected.
The form of the RTSTRT
subroutine is:
CALL RTSTRT (unit, startmsk, intmsk)
where:
unit

is the real-time device unit number

startmsk

is the flags necessary to start the device
(see
the DECsystem-lO
Hardware Reference Manual).
If
the device is being stopped, this parameter should
be zero.

intmsk

is the mask of all
interrupting bits for
the
particular device
(see the DECsystem-lO Hardware
Reference Manual).
If the device is in fast-block
mode and being started, intmsk should equal -1;
if, however, the device
in any mode
is being
stopped, the parameter must be O.

19.3.5

(see RTINIT).

BLKRW

BLKRW is used with either of the block modes.
It sets up the size and
starting address of the data block being handled.
A new count and
starting address must be set up each time the current one runs out.
19-4

USING THE FORTRAN REAL-TIME SOFTWARE (TOPS-IO ONLY)
The form of the BLKRW subroutine is:
CALL BLKRW (unit, count, blkadr)
where:
unit

is the real-time device unit number (see RTINIT).

count

is the number of words to be read or written.

blkadr

is the array into which the data is to
from which it is to be read.

19.3.6

written

RTREAD

RTREAD, used with a device in single mode, reads a single word of data
from the device.
The form of the RTREAD subroutine is:
CALL RTREAD (unit, datadr)
where:
unit

is the real-time device unit number (see RTINIT).

datadr

is the address of the location in which
data read.

19.3.7

store

the

RTWRIT

RTWRIT sends a single word of data to a
mode.

real-time

device

single

The form of the RTWRIT subroutine is:
CALL RTWRIT (unit, datadr)
where:
unit

is the real-time device unit number (see RTINIT).

datadr

is the location of the data word
device.

19.3.8

sent

the

STATO

STATO sends the specified status word to
real-time device.
The form of the STATO subroutine is:
CALL STATO (unit, statadr)

19-5

the

status

USING THE FORTRAN REAL-TIME SOFTWARE (TOPS-10 ONLY)
where:
unit

is the real-time device unit number (see RTINIT).

statadr

is the location of the word of status bits to
to the real-time device.

19.3.9

sent

STATI

STATI reads the current device status bits into the location specified
for inspection by the FORTRAN program.
The form of the STATI subroutine is:
CALL STATI (unit, adr)
where:
unit

is the real-time device unit number (see RTINIT).

adr

is the location into which the device status
to be read.

19.3.10

bits

are

RTSLP

RTSLP is called from the timesharing level and causes the FORTRAN job
to sleep until RTWAKE is called from interrupt level.
The program
goes to sleep for the specified number of seconds (up to 60). When it
wakes up,
it checks to see if RTWAKE has been called from interrupt
level.
If RTWAKE has been called, RTSLP returns to the calling
program; otherwise the job goes back to sleep again.
The form of the RTSLP subroutine is:
CALL RTSLP (time)
where:
time

19.3.11

is the length of sleep time in seconds.

RTWAKE

RTWAKE is called at interrupt level to wake up the FORTRAN program.
The form of the RTWAKE subroutine is:
CALL RTWAKE

19.3.12

DISMIS

DISMIS dismisses the
interrupt
user-interrupt routine must be
causes its execution to begin.

currently being processed.
The
sure to dismiss the interrupt that

The form of the DISMIS subroutine is:
CALL DISMIS
19-6

USING THE FORTRAN REAL-TIME SOFTWARE (TOPS-IO ONLY)
19.3.13

OISCON

DISCON disconnects a real-time device from its PI level. All devices
should be disconnected through calls to DISCON before the job is
terminated.
The form of the DISCON subroutine is:
CALL DISCON (unit)
where:
unit

19.3.14

is the real-time device unit number (see RTINIT).

UNLOCK

UNLOCK unlocks the job from core.
When execution of a job is
complete, the job is automatically unlocked before the return to the
monitor.
The UNLOCK subroutine provides a method to unlock a
job
before execution is complete.
Note that all real-time device handling
must be finished before the job is unlocked.
The form of the UNLOCK subroutine is:
CALL UNLOCK

19.3.15

Error Messages

Table 19-1 lists real-time software error messages, including the code
format,
the full message format,
and the subroutine in which the
message occurs.

19-7

USING THE FORTRAN REAL-TIME SOFTWARE (TOPS-IO ONLY)
Table 19-1:

Error Messages - Code Format and Full Message Format

Code Format

Full Message Format

Subroutine in
which message
occurs

?ILLEGAL UNIT NUMBER.
TO HANDLE MORE DEVICES,
REASSEMBLE FORRTF WITH A
LARGER
?ERROR COMES FROM THE
SUBROUTINE "subroutine name"

" RTDEVN"

?RTINIT MUST BE CALLED BEFORE
CONECT

CONECT

?CONECT MUST BE CALLED BEFORE
RTSTRT OR BLKRW

RTSTRT,BLKRW

?REAL TIME BLOCK OUT OF BOUNDS
?END OF BLOCK TOO HIGH
BLKRW
[such as, overwrites some program
or in high segment]
?END OF BLOCK TOO LOW,
such as, start address less
than 140

4
A

?JOB CANNOT BE LOCKED IN
CORE
?JOB NOT PRIVILEGED
?NOT ENOUGH CORE AVATLABLE
FOR LOCKING

5
A

?APR ERROR AT INTERRUPT
LEVEL
?PDL OVERFLOW
?ILLEGAL MEMORY REFERENCE

6
A
B

A
B
C
D
E

o
1
8

A
B

LOCK

?RTTRP ERROR
realtime trap error of the
followi ng so r t
?ILLEGAL PI NUMBER
PI channel not available
?TRAP ADDRESS OUT OF BOUNDS
?SYSTEM LIMIT FOR REALTIME
DEVICES EXCEEDED
?JOB NOT LOCKED IN CORE OR NOT
PRIVILEGED
?DEVICE ALREADY IN USE BY
ANOTHER JOB
?OCCURRED IN THE DISCON
ROUTINE
?OCCURRED IN THE CONECT
ROUTINE
?NOT ENOUGH CORE AVAILABLE
FOR THE CONTROL BLOCKS
?NOT ENOUGH CORE AVAILABLE

19-8

DISCON
CONECT

LOCK

APPENDIX A
SUMMARY OF FORTRAN STATEMENTS

FORTRAN statements and
This appendix summarizes the forms of all
provides a
section reference where each statement is described in
detail.
Section
Reference

Form
ACCEPT(FMT=f{,END=s][,ERR=s][,IOSTAT=ios]) [iolist]
ACCEPT(FMT=*[,END=s ][,ERR=s][,IOSTAT=ios]) [iolist]
ACCEPT f{,iolist]
ACCEPT *[,iolist]

Section 10.8

ASSIGN s to i

Section 8.3

BACKFILE un
BACKFILE(UNIT=un[,ERR=][,IOSTAT=ios])
BACKFILE(un[,ERR=s][,IOSTAT=ios])

Section 11.8.8

BACKSPACE un
BACKSPACE(UNIT=un[,ERR=s][,IOSTAT=ios])
BACKSPACE(un[,ERR=s][,IOSTAT=ios])

Section 11.8.4

BLOCK DATA [sub]

Section 13.1

CALL sub [( [a1 [,a2] ... ])J

Section 13.4.2.2

CHARACTER [*len[,J] v[*len] [,v[*lenJ]

Section 7.2.2

CLOSE (closelist)

Section 10.16

COMMON [![cbJl]n~ist[[,]/[cb]/nlist] ...

Section 7.4

COMPLEX v [,v ... ]

Section 7.2

CONTINUE

Section 9.4

DATA nlist/clist/ [[,]nlist/clist/] ...

Section 7.9

DECODE(c,f,a[,ERR=s][,IOSTAT=i'os]) [iolist]

Section 10.12

DELETE(UNIT=un[,ERR=s ][,IOSTAT=ios])
DELETE(
un[,ERR=s][,IOSTAT=ios])
DELETE(UNIT=un,REC=rn[,ERR=s][,IOSTAT=ios J)
DELETE(
un,REC=rn[,ERR=s ][,IOSTAT=ios J)
DELETE(
un'rn
[,ERR=s][,IOSTAT=iosJ)

Section 10.13

DIMENSION a(d) [,a(d) ... ]

Section 7.1

s[,]

i=el,e2[,e3]

Section 9.3.1

. DO

s[,]

WHILE(e)

Section 9.3.2
Section 7.2

DOUBLE PRECISION v[,v ... ]

FORTRAN-10/20 VII

A-I

February 1987

SUMMARY OF FORTRAN STATEMENTS

Form

Section
Reference

ELSE

Section 9.2.4

ELSE IF (e) THEN

Section 9.2.4

ENCODE(c,f,a[,ERR=s][,IOSTAT=iosj)[iolist]

Section 10.12

END

Section 9.7

END DO

Section 9.4

END IF

Section 9.2.4

ENDFILE un
ENDFILE(UNIT=un[,ERR=s][,IOSTAT=iosj)
ENDFILE(un[,ERR=s][,IOST AT=iosj)

Section 11.8.5

ENTRY en [(dl [,d2 ... ])]

Section 13.4.3

EQUIVALENCE (nlist) [,(nlist) ... ]

Section 7.5

EXTERNAL proc [,proc] ...

Section 7.6

FIND(UNIT=un,REC=rn[,ERR=s][,IOSTAT=iosj)
FIND(un'rn[,ERR=s][,IOSTAT=ios ])

Section 11.8.1

FORMAT (fs)

Section 12.1.1

fun ([argl,arg2, ... argnj)

Section 13.3.4

[typ] FUNCTION fun ([argl [,arg2] ... j)

Section 13.3.2

GO TO i [[,](s [,s] ... )]

Section 9.1.3

GO TO s

Section 9.1.1

GO TO (s [,s] ... )[,] e

Section 9.1.2

INCLUDE filespec/switch

Section 6.4.2

IF (e) st

Section 9.2.2

IF (e) sl, s2, s3

Section 9.2.1

IF (e) sl, s2

Section 9.2.3

IF (e) THEN

Section 9.2.4

IMPLICIT type (a [,a ... ])[,type (a[,a ... ]) ...
IMPLICIT NONE

Section 7.3

INQUIRE(FILE=fl[DEFAULTFILE=dfi][,fiistj)

Section 11. 7.1

INQUIRE([UNIT=]u,ulist)

Section 11.7.2

INTEGER v [,v ... ]

Section 7.2

INTRINSIC fun[,fun]

Section 7.7

LOGICAL v [,v ... ]

Section 7.2

NAMELIST /name/list[jname/list] ...

Section 12.7

OPEN (openlist)

Section 10.14

PARAMETER (p=c[,p=c ... ])
PARAMETER p=c [,p=c ... ]

Section 7.8

FORTRAN-10/20 VII

A-2

February 1987

SUMMARY OF FORTRAN STATEMENTS

Form

Section
Reference

PAUSE [n]

Section 9.6

PRINT(FMT=f[,ERR=s][,IOSTAT=ios]) [iolist]
PRINT(FMT=* [,ERR=s] [,IOSTAT=ios]) [iolist]
PRINT f[,iolist]
PRINT *[,iolist]

Section 10.10

PROGRAM name

Section 6.4.1

PUNCH(FMT=f[,ERR=s] [,IOSTAT=ios]) [iolist]
PUNCH(FMT=*[,ERR=s][,IOSTAT=ios]) [iolist]
PUNCH f[,iolist]
PUNCH *[,iolist]

Section 10.11

READ (UNIT=un,FMT=f[,END=s ][,ERR=s ][,IOSTAT=ios])[iolist]
READ (
un,FMT=f[,END=s][,ERR=s] [,IOSTAT=ios]) [iolist]
READ(un,
f[,END=s][,ERR=s][,IOSTAT=ios]) [iolist]
READ (UNIT=un,FMT=*[,END=s ][,ERR=s][,IOSTAT=ios]) [iolist]
READ (
un,FMT=*[,END=s ][,ERR=s][,IOSTAT=iosj)[iolist]
READ(
un,
*[,END=s][,ERR=s][,IOSTAT=ios])[iolist]
READ (UNIT=un,FMT=name[,END=s] [,ERR=s] [,lOST AT=ios])
READ (
un,FMT=name[,END=s][,ERR=s][,IOSTAT=ios])
READ(
un,
name[,END=s][,ERR=s][,IOSTAT=ios])
READ f[,iolist]
READ *[,iolistj
READ (UNIT=*,FMT=f[,END=s ][,ERR=s][,IOSTAT=ios])[iolist]
READ(UNIT=*,FMT=*[,END=s] [,ERR=s][,IOSTAT=ios]) [iolist]
READ(UNIT=un[,END=s] [,ERR=s] [,IOSTAT=ios]) [iolist]
READ (
un[,END=s][,ERR=s][;IOSTAT=ios]) [iolist]
READ(UNIT=un,FMT=f,REC=rn[,ERR=s ][,IOSTAT=ios])[iolist]
READ (
un,FMT=f,REC=rn[,ERR=s][,IOSTAT=ios]) [iolist]
READ (
un,
f,REC=rn[,ERR=s][,IOSTAT=ios]) [iolist]
READ (
un'rn,FMT=f
[,ERR=s][,IOSTAT=ios]) [iolist]
READ (
un'rn,
f
[,ERR=s][,IOSTAT=iosj)[iolist]
READ(UNIT=un,REC=rn[,ERR=s][,IOSTAT=ios]) [iolist]
READ (
un,REC=rn[,ERR=s][,IOSTAT=iosj)[iolist]
READ (
un'rn
[,ERR=s][,IOSTAT=ios])[iolist]
READ(UNIT=un,FMT=f,key=val[,KEYID=kn] [,ERR=s][,IOSTAT=ios])[iolist]
READ(un,
FMT=f,key=val[,KEYID=kn] [,ERR=s][,IOSTAT=ios]) [iolist]
READ( un,
f,key=val[,KEYID=kn ][,ERR=s][,IOSTAT=ios]) [iolist]
READ (UNIT=un,key=val[,KEYID=kn] [,ERR=s] [,lOST AT=ios]) [iolist]
READ (
un,key=val[,KEYID=kn][,ERR=s][,IOSTAT=ios]) [iolist]

Section 10.5

REAL v [,v ... ]

Section 7.2

REREAD (FMT=f[,END=s ][,ERR=s][,IOSTAT=ios)) [iolist]
REREAD (FMT=* [,END=s][,ERR=s][,IOSTAT=ios]) [iolist]
REREAD f[,iolist]
REREAD *[,iolist]

Section 10.7

RETURN [e)

Section 13.4.4

REWIND un
REWIND(UNIT=un[,ERR=s][,IOSTAT=ios))
REWIND (un[,ERR=s] [,IOSTAT=ios])

Section 11.8.2

REWRITE(UNIT=un,FMT=f[,ERR=s] [,IOSTAT=ios]) [iolist]
REWRITE(
un,
f[,ERR=s][,IOSTAT=ios]) [iolist]
REWRITE(UNIT=un[,ERR=s] [,IOSTAT=ios]) [iolist]
REWRITE(
un[,ERR=s ][,IOSTAT=ios]) [iolist]

Section 10.14

SAVE [a[,a] ... ]

Section 7.10

SKIPFILE un
SKIPFILE(UNIT=un[,ERR=s][,IOSTAT=ios))
SKIPFILE(un[,ERR=s][,IOSTAT=ios])

Section 11.8.7

FORTRAN-10/20 VII

A-3

February 1987

SUMMARY OF FORTRAN STATEMENTS
Section
Reference

Form
SKIPRECORD un
SKIPRECORD(UNIT=un[,ERR=s ][,IOSTAT=ios])
SKIPRECORD(un[,ERR=s][,IOSTAT=ios])

Section 11.8.6

STOP [n]

Section 9.5

SUBROUTINE sub [([d1 [,d2] ... ])]

Section 13.4.2.1

TYPE(FMT=f[,ERR=s][,IOSTAT=ios]) [iolist]
TYPE(FMT=*[,ERR=s][,IOSTAT=ios])[iolist]
TYPE f[,iolist]
TYPE *[,iolist]

Section 10.9

v=e

Section 8.2

UNLOAD un
UNLOAD(UNIT""lln[,ERR=s][,IOSTAT=ios])
UNLOAD(un[,ERR=s][,IOSTAT=ios])

Section 11.8.3

UNLOCK(UNIT=un[,ERR=s][,IOSTAT=ios])
UNLOCK(
un[,ERR=s][,IOSTA T=ios])
UNLOCK un

Section 10.15

WRITE(UNIT=un,FMT=f[,ERR=s] [,IOSTAT=ios]) [iolist]
WRITE(
un,FMT=f[,ERR=s][,IOSTAT=ios]) [iolist]
WRITE(
un,
f[,ERR=s][,IOSTA T=ios]) [iolist]
WRITE(UNIT=un,FMT=*[,ERR=s][,IOSTAT=ios])[iolist]
WRITE(
un,FMT=*[,ERR-=s][,IOSTAT=ios]) [iolist]
WRITE(
un,
*[,ERR=s][,IOSTAT=ios])[iolist]
WRITE(UNIT=un,FMT=name[,ERR=s] [,IOSTAT=ios])
WRITE(
un,FMT=name[,ERR=s][,IOSTAT=ios])
WRITE(
un,
name[,ERR=s][,IOSTAT=ios])
WRITE f[,iolist]
WRITE *[,iolist]

Section 10.6

WRITE(UNIT=* ,FMT=f[,ERR=s][,IOSTAT=ios]) [iolist]
WRITE(UNIT=*,FMT=*[,ERR=s][,IOSTAT=ios]) [iolist]
WRITE(UNIT=un[,ERR=s][,IOSTAT=ios]) [iolist]
WRITE(
un[,ERR=s][,IOSTA T=ios]) [iolist]
WRITE(UNIT=un,FMT=f,REC=rn[,ERR=s][,IOSTAT=ios]) [iolist]
WRITE(
un,FMT=f,REC=rn[,ERR=s][,IOSTAT=ios]) [iolist]
WRITE(
un,
f,REC=rn[,ERR=s][,IOSTAT=ios])[iolist]
WRITE(
un'rn,FMT=f,
[,ERR=s][,IOSTAT=ios])[iolist]
WRITE (
un'rn,
f,
[,ERR-s][,IOSTAT=ios]) [iolist]
WRITE(UNIT=un;REC=rn[,ERR=s] [,IOSTAT=ios]) [iolist]
WRITE(
un,REC=rn[,ERR=s ][,IOSTAT=ios]) [iolist]
WRITE(
un'rn
[,ERR=s][,IOSTAT=ios])[iolist]

FORTRAN-10/20 V11

A-4

February 1987

APPENDIX 8
ASCII-1968 CHARACTER CODE SET

The character code set defined
in the X3.4-1968 Version of the
American National
Standard
for
Information
Interchange (ASCII) is
given in this appendix.

8-1

ASCII-1968 CHARACTER CODE SET
ASCII CODE
Control Characters
Even
Parity
Bit

7-Bit
Decimal

7-Bit
Octal

000

NUL

001

SOH

Start of heading [SaM, start of message] . Control A.

002

STX

Start of text [EOA, end of address] . Control B.

003

ETX

End of text [EOM, end of message] . Control C.

004

EaT

End of transmission; shuts off TWX machines and disconnects some
data sets. Control D.

005

ENQ

Enquiry [WRU, "Who are you?"]. Triggers identification ("Here
is ... ") at remote station if so equipped. Control E.

Acknowledge [RU, "Are you ... ?"]. Control F.

Character Class·

Remarks
Null, tape feed. Control @ (control shift p 2).

006

ACK

007

BEL

008

010

Backspace. Control H.

009

011

Horizontal tabulation. Control I.

010

012

Line feed. Control J.

011

013

HT
LF3
VT 3

Vertical tabulation. Control K.

012

014

FF3

Form feed (to top of next page). Control L.

013

015

Carriage return (to beginning of line). Control M.

014

016

Bell (audible or attention signal). Control G.

Shift out; change character set or change ribbon color to red.
Control N.

015

017

016

020

DLE

Shift in; return to standard character set or color. Control O.

017

021

DCI

Device control 1, turns transmitter (reader) on. Control Q (X ON).

018

022

DC2

Device control 2, turns punch or auxiliary on. Control R (TAPE,
AUXON).

019

023

DC3

Device control 3, turns transmitter (reader) off. Control S (X OFF).

020

024

DC4

Device control 4 (stop), turns punch or auxiliary off. Control T
(--rAPE-, AUX OFF).

021

025

NAK

Negative acknowledge [ERR, error]. Control U.

022

026

SYN

Synchronous idle [SYNC]. Control V.

Data link escape [DCO]. Control P.

023

027

ETB

024

030

CAN

End of transmission block [LEM, logical end of medium] . Control W.

025

031

End of medium [S.]. Control Y.

026

032

SUB

Substitute [S2]' Control Z.

027

033

ESC

Escape, prefix [S3]' Control [(control shift K2).
File separator [S 4] . Control \ (control shift L2).

Cancel [So], Control X.

028

034

029

035

030

036

Group separator [Ss]' Control] (control shift M2).
Record separator [S 6] . Control (control shift N2 ).

031

037

Unit separator [S71. Control- (control shift 0 2).

1 CC communication control, FE format effector, IS information separator.
20n LT33, LT35 and similar units.
3 Includes a carriage return on some equipment, but not on standard DEC units.

8-2

ASCII-1968 CHARACTER CODE SET

Graphic Characters
Figures

Upper Case

Lower Case

Even

Parity 7-Bit 7-Bit
Bit Decimal Octal Character
032
040
SP
1
033
041
o
/I
034
042
o
035
043
#
$
036
044
o
045
037
1
%
1
038
046
&
o
047
039
040
050
o
041
051
052
1
042
*
043
053
o
+
044
054
045
o
055
046
o
056
057
1
047
/
(/)1
048
060
o
049
061
1
050
2
062
051
063
3
o
4
052
064
1
053
065
5
o
054
6
066
o
055
067
7
070
8
1
056
057
9
071
o
058
072
o
1
059
073
074
060
o
<
075
061
076
062
>
063
077
o

Parity 7-Bit 7-Bit
Bit Decimal Octal Character
064
100
@
1
065
101
A
o
066
B
102
o
1
067
C
103
068
D
104
o
E
069
105
1
F
070
106
071
o
G
107
H
o
072
110
073
111
I
1
074
J
112
o
075
K
113
076
114
L
077
o
M
115
078
N
o
116
079
1
117
a
p
080
o
120

Parity 7-Bit
7-Bit
Bit Decimal Octal Character 3
o
096
140
,2

o
1

o
o
1

o
o
o
1

081
082
083
084
085
086
087
088
089
090
091
092
093
094
095

121
122
123
124
125
126
127
130
131
132
133
134
135
136
137

o
1

o
o
1

o
1

R
S
T
U
V

o
1

z
[

]

,,2

097
098
099
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127

141
142
143
144
145
146
147
150
151
152
153
154
155
156
157
160
161
162
163
164
165
166
167
170
171
172
173
174
175
176
177

a
b
c
d
e
f
g
h
j
k
1
m
n
a
p
q

u
v
w
x
y
z
{

I
}4
",,2,5

DEL6

1 Zero-slash absent on many units.
2

Under study by responsible American National Standards Committee for possible change at next revision of ASCII
(ca. 1982}.

3 Codes 140-173 first defined in 1965. For a full ASCII character set the operating system accepts codes 140-176 as

lower case. For a program requiring a character set that lacks lower case, the operating system translates input codes
140-174 into the corresponding upper case codes (100-134) and translates both 175 and 176 into 033, escape.
Early versions of the DECsystem-l0 Monitor used 175 as the escape code and translated both 176 and 033 to it.
4Unassigned control character (usually ALT MODE) before 1965. Code generated by ALT MODE key on some DEC
units, especially earlier ones; on some more recent units, the ALT key generates the standard escape code, 033.
5 Control character ESC before 1965; code generated by ESC key on some DEC units designed at that time.

6 Delete, rub out (not part of lower case set).

B-3

ASCII-1968 CHARACTER CODE SET

Remarks on Special Graphic Characters

SP Space - normally nonprinting.

Question mark.

Exclamation point.

Quotation mark, diaeresis.

Number sign. £ on some (non-DEC) units.

Dollar sign.

Percent.

(a)

Ampersand.

Circumflex, upward arrow head. t before 1965,
but used until 1972 on DEC units.
Underline, underscore.- before 1965, but used
until 1972 on DEC units.

Closing parenthesis.
Asterisk.
Plus.

Grave accent, opening single quotation mark.
(w 1965-67, but never on DEC units.
Opening brace.

Comma, cedilla.

Vertical line. Control character ACK before
1965; ---, 1965-67, but never on DEC units;
: in appearance 1968-19-77, but generally not
on DEC units.

Hyphen, minus.
Period, decimal point.
Slant, slash, solidus.

Closing brace. Unassigned control
(usually ALT MODE) before 1965.

Colon.

character

Overline, tilde, general accent. Control charafter ESC before 1965; I 1965-67, but never
on DEC units.

Semicolon.

Reverse slant. ~ 1965-67, but never on DEC
units. Shift Lon LT33, LT35 and similar units.
Closing bracket. Shift M on LT33, LT35 and
similar units.

Opening parenthesis.

Commercial at. ' 1965-67, but never on DEC
units.
Opening bracket. Shift K on LT33, LT35 and
similar units.

Apostrophe, closing single quotation mark,
acute accent. -- in appearance on some DEC
units.

Greater than.

Less than.
Equals.

8-4

APPENDIX C
COMPILER MESSAGES

The FORTRAN compiler issues two types of messages: warning and fatal
error.
While compiling the program, if the compiler encounters a
situation that does not prevent it from completing the compilation,
but does warrant your knowing about, it prints a warning message and
continues compilation. If, however, the problem in your program is
such
that
compilation cannot continue, the compiler prints a
fatal-error message and stops compilation of the program. Whenever a
fatal error is generated a relocatable object module will not be
produced.
Compiler messages are printed in the following form:
?FTNxxx LINE:n text
or
%FTNxxx LINE:n flag: text
where:
?
%
FTN
xxx
Line:n
text
flag:

indicates a fatal message
indicates a warning message
is the FORTRAN mnemonic
is the 3-letter mnemonic for the error message
is the optional line number where the error occurred
is the explanation of error
is the prefix for warning messages generated when the
compatibility flagger is invoked. This prefix describes
the type of incompatibility the message refers to, and can
be one of the following:
•

Extension to Fortran-77:

•

VMS incompatibility:

•

Fortran-77 and VMS:

See Section 16.6 for more information on the compatibility flagger.
Square brackets ([ ]) in this appendix signify variables and
output on the terminal.

C-l

are

not

COMPILER MESSAGES
Fa tal Er ro rs
Each fatal error in the following list is preceded by ?FTN on the user
terminal and on listings.
They are presented here in alphabetical
order.
ABD

[symbolname] has already been defined [definition]
The usage given conflicts with current information about the
symbol.
For example, a symbol defined in an EQUIVALENCE
statement cannot be referenced as a subprogram name.

AHE

IF at line [number] already has ELSE

AOA

Assumed-size arrays only allowed in subprograms

ASA

Assumed-size arrays cannot be used

ATL

ARRAY [name] too large
The total amount of memory necessary
array is greater than 512P.

AWN

accommodate

Array reference [name] has wrong number of subscripts
The array was defined to have more or fewer dimensions
the given reference.

BOV

this

than

Statement too large to classify
To determine statement type, some portion of the statement
must be examined by the compiler before actual semantic and
syntactic analysis begins. During this classification, the
entire portion of the required statement must fit into the
internal statement buffer (large enough for a normal 20-line
statement) •
This error message is issued when the portion of a given
statement required for classification is too larg~ to fit in
the buffer. Once FORTRAN-IO/20 has classified a statement,
there is no explicit restriction on its length.

CER

Compiler error in routine [name]
Submit an SPR for any occurrence of this message.

CEL

Character entry points must have the same length

CEN

Character expression used where numeric expression required

CEX

Constant or constant expression required

CFF

Cannot find file
The file referenced in an INCLUDE statement was not found.

CFL

Reference to character function [name]
length

CNE

Character and numeric entry points cannot be mixed

CPE

Checksum or parity
[name]

error

C-2

must

have

fixed

[source/listing/object]

file

COMPILER MESSAGES
CQL

No closing quote in character constant

CSA

Can't split string across numeric and character variables

CSF

Illegal statement function reference in CALL statement

DDA

[symbolname]

DDN

DO loops too deeply nested -

DFC

Variable dimension [name] must be scalar, defined as
or in COMMON

DFD

Double [type]

is duplicate dummy argument

formal

name illegal

Duplicate fields
specification.
DIA

reduce nesting

were

encountered

DO index variable [name]

INCLUDE

file

is already active

In any nest of DO loops, a given index variable may
defined for more than one loop.
DID

Cannot initialize a dummy parameter in DATA

DLN

Optional data value list not supported

not

The extended FORTRAN statement form that allows data
to be defined
in type specification statements
supported by FORTRAN-IO/20.
specification

without

values
is not

DNL

Implied DO
variables

DPR

Dummy parameter [name]

DSF

Argument [name]

DTI

The dimensions of [arrayname] must be of the type integer

DVE

Cannot use dummy variable in EQUIVALENCE

ECS

[variable]
statement

ECT

Attempt to enter [symbolname]

EDN

Expression too deeply nested to compile

EID

ENTRY statement illegal inside a block IF or DO loop

ElL

Expression illegal in an input list

ElM

ENTRY statement illegal in main program

ENF

Label [number] must refer to an executable statement, not
FORMAT

ETF

Enter failure [number]

EXB

EQUIVALENCE extends COMMON block [name] backwards

FEE

Found [symbol]

associated

list

referenced before definition

is same as FUNCTION name

EQUIVALENCE-d

COMMON

illegal

SAVE

into COMMON twice

[filespec]

when expecting either [symbol] or [symbol]

General syntax error message.

C-3

COMPILER MESSAGES
FER

[file error text]
An error has
specification.

occured

when

processing

command

FlO

Can't initialize character function name

FNE

Label [number] must refer to a
statement

FWE

Found [symbol] when expecting [symbol]

lAC

Illegal ASCII character [character] in source

IAL

Incorrect argument type for library function [name]

IAN

Illegal assignment between character and numeric data

IBD

Illegal substring bound in DATA statement

IBK

Illegal statement in BLOCK DATA subprogram

ICL

Illegal character [character]

ICN

Illegal combination of character and numeric data

ION

DO loop at line:

[number]

FORMAT,

not

file

executable

in label field

is illegally nested

You are attempting to terminate a DO loop before terminating
one or more loops defined after the given one.
IDS

Implicit DO indices may not be subscripted

lOT

Illegal or misspelled data type

IDV

Implied DO index is not a variable

lED

Inconsistent EQUIVALENCE declaration
The given EQUIVALENCE declaration would cause some
name to refer to more than one physical location.

IFD

INCLUDEd files must reside on disk

IFE

[INCLUDE file error]
This error occured while trying
INCLUDE file on the DECSYSTEM-20.

IFS

Illegal format specifier

110

Non-integer implied DO index

lIP

Illegal implicit specification parameter

lIS

Incorrect INCLUDE switch

ILF

Illegal statement after logical IF

open

the

symbolic

specified

Refer to Section 9.2.2 for restrictions on logical IF object
statements.
IND

Improper nesting:

DO at line [number] has not terminated

C-4

COMPILER MESSAGES
INI

Improper nesting:

INN

INCLUDE statements may not be nested

IOC

Illegal operator for character data

IOD

Illegal statement used as object of DO

ION

Numeric operand of concatenation operator

IOR

Substring bound out of range

IQB

INQUIRE -

IQN

INQUIRE - neither UNIT nor FILE keywords were specified

ISD

Illegal subscript expression in DATA statement

IF at line [number]

has not terminated

both UNIT and FILE keywords were specified

Subscript expressions may be formed only
indexes and constants combined with +, ISN

[symbolname]

with

implicit

*, or /.

is not [symbol type]

The symbol cannot be used in the attempted manner.
ISS

[var iable]

ITL

Illegal transfer into loop to label [number]

IUT

Program units may not be terminated within INCLUDEd files

IVC

Invalid character constant

IVH

Invalid hollerith constant

IVP

Invalid PPN

IXM

Illegal mixed mode arithmetic
Complex and
expression.

illegal in SAVE statement

double-precision

[OPEN specifier]

cannot

appear

the

IXS

Illegal

IZM

Illegal [datatype]

FORTRAN will not run on a KA

KAS

FORTRAN can not compile for a KA

LAD

Label [number]

LED

Illegal list directed [statement type]

LFA

Label arguments illegal in FUNCTION or array reference

LGB

Lower bound greater than upper bound for array [name]

LLS

Label too large or too small

same

specifier

size modifier [number]

already defined at line:

[number]

Labels cannot be 0 or greater than five digits.
LND

Label
[number]
declaration

must

refer

C-5

[statement] ,

not

COMPILER MESSAGES
LNI

List directed I/O with no I/O list

LTL

Too many items in list - reduce number of items
In rare instances, a combination of long lists in
statement can exhaust the syntax stack.

single

formal

label

MCE

More than 1 COMMON variable in EQUIVALENCE group

MSP

Statement name misspelled

MST

[OPEN specifier] must be [integer or array]

MWL

Attempt to
arguments

NCC

Can't store numeric constant in character variable

NCF

Not enough core for
[number]

NEX

No exponent after D or E in constant

NFS

No filename specified

define

multiple

the

file

RETURN

without

specs.

Total

K needed

The INCLUDE statement requires a filename.
NGS

Cannot get segment [segment name] - error code [GETSEG error
code]
This message means the system is unable to GETSEG one of the
compiler segments on the DECsystem-lO.

NIF

No matching IF

NIO

NAMELIST directed I/O with I/O list

NIR

Repeat count must be an unsigned integer

NIU

Non-integer unit number in I/O statement

NLF

Wrong number of arguments for library function [name]

NLS

[variable] may not be declared length star

NMD

No matching DO

NNA

NAMELIST not allowed in ENCODE, DECODE, and REREAD

NNF

No statement number on FORMAT

NNN

NML= must specify a NAMELIST

NRC

Statement not recognized

NUO

.NOT.

NWB

Numeric variable must be aligned on word boundary

NWD

Incorrect use of

NYI

Not yet implemented

OAG

Octal or logical argument illegal to generic function

is a unary operator

* or?

in [filespec]

C-6

COMPILER MESSAGES
OBO

[variable] may only be specified once

OPW

OPEN/CLOSE parameter [name]

OUB

Only upper bound of last dimension
asterisk

PD6

FORTRAN will not run on PDP6

PIC

The DO parameters of [index name] must be integer constants

PN4

project number must be 4 in ppn

is of wrong type

This error is for DECsystem-IO file
DECSYSTEM-20.
.

[arrayname]

may

specifications

PRF

Pro t e c t ion fa i I u r e [ n urn b e r ] [ f i I e s pe c ]

PTL

Program too large

the

real

The program unit takes up more than 512P.
QEF

Quota exceeded or disk full [number]

[filespec]

RDE

Rib or directory error [number]

RFC

[function name] is a recursive function call

RIC

Complex constant cannot be used to represent
imaginary part of a complex constant

RUS

Relational expression illegal as UNIT specifier

SAD

Array [name] - signed dimensions may appear only as constant
range limits

SMC

Size modifier conflict for variable [name]

SNC

Substring of non-character variable

SNL

[statement name] statements may not be labeled

SOR

Subscript out of range

STD

Sta temen t [number]

TDO

[symbol type] type declaration out of order

TFL

Too many FORMAT labels specified

UCE

User core exceeded at location [address]
while processing statement [number]

UEC

Label [number] previously used in executable context

UFC

Label [number] previously used in FORMAT context

UKW

Unrecognized keyword

UMP

Unmatched parentheses

UNS

UNIT may not be specified

[filespec]

the

is a declaration

C-7

in phase

[segment]

COMPILER,MESSAGES
USI

[symbol type]

[symbol name]

used incorrectly

The given symbol cannot be used in this way.
VNA

Subscripted variable in EQUIVALENCE, but not an array

VSE

EQUIVALENCE subscripts must be integer constants

VSO

Variable dimension allowed in subprograms only

WIF

[I/O type]

ZLD

Zero-trip DO loop illegal in DATA statement

is illegal with internal files

Warning Messages
Each warning message in the following list is preceded by %FTN on the
user
terminal
and
on listings.
They are presented here
in
alphabetical order.
ACB

Argument out of range of CHAR, high order bits ignored

ADS

Variable [name]

AGA

Opt - object variable, of
list, was never assigned

AIL

Illegal length argument for ICHAR, first character used

AIS

Extension to Fortran-77:

ANS

VMS incompatibility: ASSOCIATEVARIABLE not
OPEN

CAl

COMPLEX expression used in arithmetic IF

CAO

Consecutive arithmetic operators illegal

CAP

Extension to Fortran-77: Consecutive arithmetic operators

CCC

Fortran-77 and VMS: Carriage control character

CCN

CHARACTER constant used where numeric expression required

CIS

Conflicting INCLUDE switches

CNM

Character and numeric variables mixed

CNS

Extension to
Fortran-77:
non-specific length

COS

Extension to Fortran-77: Comment on statement

COV

Extension to
Fortran-77:
character expression

CSM

Extension to Fortran-77: Comma field separator is missing

already declared in SAVE statement

C-8

assigned

GOTO

without

optional

Apostrophe in I/O specifier

Concatenation

Assigned

set

with

variable

VMS

variable

appears

COMPILER MESSAGES
CTR

Complex terms used ina relational other than EO or NE
The result of the other relational
operands is undefined.

CUO

operators

with

complex

Constant underflow or overflow
This message
is issued when overflow or underflow is
detected as the result of building constants or evaluating
constant expressions at compile time.

DEB

Extension to Fortran-77: DEBUG lines

DFN

VMS incompatibility: Default file name on VMS
Fortran-IO/20

DII

Previous declaration of intrinsic function is ignored

DIM

possible DO index modification inside loop

differs

from

A program that does this may be incorrectly compiled by the
optimizer, since it assumes that indexes are never modified.
Note that the number of iterations
is calculated at the
beginning of the loop and is never affected by modification
of the index within the loop.
DIS

Opt - program is disconnected - optimization discontinued
Submit an SPR if this message occurs.

DOW

Extension to Fortran-77: DO WHILE statement

DPE

VMS incompatibility: Different precedence in exponentiation

DWE

Fortran-77 and VMS:
[descriptor]

DWL

Extension to Fortran-77: DO without statement label

DXB

DATA statement exceeds bounds of array [name]

EDD

Extension to Fortran-77: END DO statement

EDS

Extension to Fortran-77: DECODE statement

EDS

Extension to Fortran-77: ENCODE statement

EDX

Fortran-77 and VMS: FORMAT edit descriptor [descriptor]

EOC

Fortran-77 and VMS: Exponential operator

EXD

Extension to Fortran-77: Transfer of control into DO loop at
label [label]

FAR

Extension to Fortran-77: Format in numeric array

FIF

Extension to Fortran-77:
function in Fortran-77

FIN

Extension to Fortran-77: FIND statement

FMR

Multiple RETURNs defined in a FUNCTION

FMT

VMS incompatibility: Keyword FMT instead of NML

Default

widths

with

[function name]

C-9

edit

descriptor

is not an intrinsic

COMPILER MESSAGES
FNA

A function without an argument list

FNG

Extension to Fortran-77:
function in Fortran-77

FNS

Extension to
subroutine

FOO

Statement function
dimensioned

HCP

VMS incompatibility: Hollerith constant padded with spaces

HCN

Hollerith constant used where numeric expression required

HCU

Extension to Fortran-77: Hollerith constant

IAT

Illegal type for argument [number]

ICC

Illegal character, continuation field of initial line

[function name]

Fortran-77:
declared

not

generic

[name]

not

FORTRAN-77

out

order

array

not

for statement function

Continuation lines cannot follow comment lines.
ICD

Inaccessible code.

Statement deleted

The optimizer will delete statements that cannot be
during execution.
ICS

Illegal character in line sequence number

ION

Opt -

illegal DO nesting -

reached

optimization discontinued

A GO TO within a DO loop goes to the ending statement of an
inner, nested DO loop.
The line number printed out with the
warning message is that of the OUTER DO.
DO

GO TO

CONTINUE

IFL

Opt -

infinite loop.

Optimization discontinued

IMN

IMPLICIT NONE

INC

Extension to Fortran-77:

INS

VMS incompatibility: /NOCREF switch

INS

VMS incompatibility: /CREF switch

INS

VMS incompatibility: Default for VMS is /NOLIST

INCLUDE statement

C-IO

COMPILER MESSAGES
IUA

Illegal use of an array - use scalar variable instead

KIS

Obsolete switch /KI

KWU

Fortran-77 and VMS: Keyword [keyword name]

KWV

Fortran-77 and VMS: Keyword value for [keyword name]

LID

Identifier [name] more than six characters
The remaining characters are ignored.

LNC

Fortran-77 and VMS: Non-numeric operand in numeric context

LNE

VMS
incompatibility:
EQUIVALENCE-d

LOL

VMS incompatibility: List of labels

LSP

Extension to Fortran-77:

MBD

IMPLICIT NONE -

MLN

Fortran-77 and VMS: Mixing logical and numeric

MSL

Fortran-77 and VMS: Multi-statement line

MVC

Number of variables
[is less than/is
numbers of constants in DATA statement

NAM

Extension to Fortran-77: NAMELIST statement

NDP

Fortran-77 and VMS: No decimal places with edit descriptor

NEC

Extension to Fortran-77:
context

NED

No END statement in program

NIB

Extension to Fortran-77: Non-integer substring bounds

NIG

Extension to Fortran-77: Non-integer as
GOTO

NIK

Extension to Fortran-77: Non-integer used with [keyword]

NIS

Extension to Fortran-77: Non-integer subscript

NIX

Extension to Fortran-77: Non-integer as index to RETURN

NLC

Fortran-77 and VMS: Non-logical operand in logical context

NLK

Extension to Fortran-77: Use of NAMELIST

NOD

Global optimization not supported with /DEBUG - /OPT ignored

NOF

No output file given

NPC

VMS incompatibility: Null padding before [symbolic name]

NPP

Extension to Fortran-77:
list

NSC

Fortran-77 and VMS: Non-standard character in column 1

Logical

and

[data type]

numeric

variables

length specifier

[variable] must be explicitly declared

Numeric

C-ll

greater

expression

parentheses

index

than]

around

the

character

computed

PARAMETER

COMPILER MESSAGES
OCU

Fortran-77 and VMS: Octal constant

OHC

Octal or hexadecimal constant

010

Extension to Fortran-77:

010

Fortran-77 and VMS:

PAV

PARAMETER used as associative variable

PLP

PARAMETER list must be enclosed in parentheses

PPS

PROGRAM statement parameters ignored

[statement name]

statement

Used for compatibility purposes.
PSR

Pound sign (i)
instead

PWS

Fortran-77 and VMS: PRINT (Specifiers)

RDI

Attempt to redeclare implicit type

RIM

RETURN statement illegal in main program

RLC

Extension to Fortran-77: & used with return label

RLC

Fortran-77 and VMS: $ used with return label

RLX

Fortran-77 and VMS: Return label [label]

SSC

Extension to Fortran-77: Substring bounds not constant

SEP

VMS incompatibility:
name or entry point

SID

Slash (I)

SMD

Extension
to
Fortran-77:
Single
multi-dimensioned array [array name]

SNN

VMS incompatibility:
name

SOD

[name]

SOR

Fortran-77 and VMS: Subscript out of range for array
name]

SPN

VMS incompatibility:
PARAMETER name

SRO

Fortran-77 and VMS: Symbolic relational operator [operator]

SVN

VMS incompatibility:
function name

TLF

Fortran-77 and VMS:

TSI

Type of symbolic constant ignored

VAl

[name]

in random access - use REC= or apostrophe (I)

[symbolic name]

in dimension bound -

statement

is the same as

use colon (:)

[symbolic name]

program

instead

subscr i pt

with

is the same as NAMELIST

statement out of order

[symbolic

name]

[symbolic name]

C-12

same

is same as variable or

Two-branch logical IF

already initialized

the

[array

COMPILER MESSAGES
VFS

VMS incompatibility:
routine on VMS

[function name]

VGF

VMS incompa ti b i 1 i ty:
on VMS

[funct i on name]

is a

VIF

VMS incompatibility:
function on VMS

[function

VND

FUNCTION return value is never defined

VNF

VMS incompatibility:
function on VMS

[function name]

VNG

VMS incompatibility:
function on VMS

[func t ion

VNI

Opt - variable [name]

Fortran-supplied
gener ic

function

intrinsic

name]

not
is

name]

an
not

intrinsic
a

generic

is not initialized

The optimizer analysis determined that the given variable
was never initialized prior to its use in a calculation.
VNS

VMS incompatibility:
subroutine

[subroutine name]

VSD

VMS incompatibility: Subroutine [subroutine name] may differ

WDU

Fortran-77 and VMS: WRITE with default unit

WNA

Wrong number of arguments for statement function

WOP

Opt - warnings given in Phase
incorrect

is not a VMS-supplied

Optimized

code

may

One or more of the messages issued
prior to
this message
resulted
from situations that violate assumptions made by
the optimizer, and thus may cause it to generate code that
does not execute as desired.
XCR

Extraneous carriage return
Carriage return was not immediately preceded or followed
a line termination character.
[* or &]

XEN

Fortran-77 and VMS:

XOR

Extension to Fortran-77: Logical .XOR. operator

ZMT

Size modifier [number]

with external name

treated as [data type]

This message is issued when one of the data type size
modifiers that is accepted only for compatibility is used.
Internal Compiler Errors
An internal compiler error is an attempt by either the compiler or the
monitor
to document an error
inside the FORTRAN compiler.
An
occurrence of an internal compiler error signifies that something
is
wrong with the FORTRAN compiler.

C-13

COMPILER MESSAGES
Monitor-detected internal errors are of the form:
?
?
?

Internal compiler error
[message] at location [address] in phase [segment]
While processing statement [line-number]

where [message] can be one of the following for TOPS-lO:
Illegal memory reference
A read or write was attempted to a non-existent page
stack exhausted
Monitor detected PDL overflow
Memory protection violation
Illegal reference to high segment
or where [message] can be one of the following for TOPS-20:
Illegal memory reference
A read was attempted to a non-existent page
Non-existent memory write
A write was attempted to a non-existent page
Illegal memory read
A memory read failed
Illegal memory write
A memory write failed
Stack exhausted
Monitor detected PDL overflow
Compiler-detected errors are of the form:
? Internal compiler error-processing statement [line-number]
? Call to [routine-name] from [address]
Submit an SPR if you receive an internal compiler error.
At the end of program unit compilation, the compiler prints
summary line, which is one of the following:
[ No error detected ]
%FTNWRN
no warning{s)
%FTNWRN
[warning count] warnings{s)
?FTNFTL
[fatal count] fatal error{s)
?FTNFTL
[fatal count] fatal error{s)
warning{s)

C-14

and no warning{s)
and [warning count]

error

APPENDIX D
FOROTS ERROR MESSAGES

Errors detected
categories:

run

time

FOROTS

fall

into

the

following

System errors (SYS) -

errors internal to FOROTS

Open errors (OPN) and CLOSE

I/O errors that occur during a file

Arithmetic
fault
calculations

errors

Library errors (LIB) routines

Data errors (DAT) -

Device errors (DEV) -

Compatibility errors
compatibility flagger

RMS-returned errors -

(APR) -

errors

generated

OPEN

numer ic

FORLIB

library

errors in data conversion on I/O
I/O hardware errors
(COM) -

generated

errors

the

errors returned by RMS-20 to FOROTS

The messages generated by FOROTS contain the following elements:
1.

A 3-letter code that identifies the type of message
onl y)

The message
encountered

For I/O errors, two integer values which are retrieved by the
ERRSNS subroutine

For compatibility errors, a prefix precedes the message that
describes the type of incompatibility the messages refers to;
one of the following:

itself,

which

•

Extenstion to FORTRAN-77:

•

VMS incompatibility:

•

FORTRAN-77 and VMS:

describes

See Section 16.6 for more information
flagger.

what

the

(TOPS-10

FOROTS

has

compatibility

For RMS errors, the RMS STS and STV codes, where available,
are
included
in
the error message
itself,
and can be
retrieved by the ERRSNS subroutine.

FORTRAN-10/20 VII

D-l

February 1987

FOROTS ERROR MESSAGES
,The 3-letter code (TOPS-10 only) and the message are, by default,
printed at your terminal when an error occurs; the two ERRSNS values
are stored within the arguments you have supplied for the ERRSNS
subroutine.
If you do not include a call to the ERRSNS subroutine in
your program, your program cannot have access to the two ERRSNS
values.
When an RMS-detected error occurs, the second ERRSNS value contains
the STS code from RMS (a number between 300000 and 377777 octal). The
STV code, which further amplifies the reason for an RMS error, is
available to your program through an additional argument to the ERRSNS
subroutine. For instructions on how to use the ERRSNS subroutine, see
Section 13.4.1.15.)
Table D-I contains a list of all the 3-letter message
ERRSNS values that are generated by FOROTS.

FORTRAN-10/20 VII

D-2

codes

and

the

February 1987

FOROTS ERROR MESSAGES
Table D-l:
1st
Value

FOROTS Error Codes
2nd
Value(S)

Meaning

Code(l)

No error detected
No error detected

o
n (7)

IDC (3)

Arithmetic trap
Integer divide check

n (7)

IOV (3)

Input Conversion Error
Integer overflow

(7)

FOV (3)

Input Conversion Error
Floating overflow

n (7)

FOV (3)

Arithmetic trap
Floating overflow

n (7)

FDC (3)

Arithmetic trap
Floating divide check

n (7)

FUN

Arithmetic trap
Floating underflow

FUN (3)

1
2

3
4

6
(3)

Input Conversion Error
Floating underflow

7
9

(7)

FTS (3)

104

IDU

105
106

UNO
NOF

107

CWU

108

CLE

109

ICE

110

NCS

112

AQS

113
115

SSE
SRE
TMA

116

CGP

117

CRP

(1)

118

NSS

(2)

119
120
121

CFS (2)
CGS (2)
CPP (3)

output Conversion Error
Output field
width
too
small
FORLIB errors and warnings
DIVERT:
illegal to divert
to unit
DIVERT: unit not open
DIVERT: unit not open for
formatted I/O
DIVERT:
Can't write
to
unit
Concatenation result longer
than expected
Illegal length
character
expression
No
character
stack
allocated
First argument of SORT must
be a quoted string
Substring range error
Subscript range error
Too many arguments in call
to SORT
Can't get pages 600:677 for
SORT
Can't return pages 600:677
after call to SORT
No free section available
for SORT
Can't find SYS:SORT.EXE
Can't get SYS:SORT.EXE
Can't
preallocate
pages
600:677 for SORT

114

(3)
(3)

D-3

February 1987

FOROTS ERROR MESSAGES
Table D-l:
1st
Value

FOROTS Error Codes (Cont'd)
2nd
Va1ue(5)

Code (1)

Meaning

122
123

IPN
CCS

124

ECS

125

ALZ

126

DMA (2,3)

Illegal page number
Not
enough
memory
for
creating character stack
Not
enough
memory
for
expanding character stack
Argument less than or equal
to zero
Must give lower and
upper
bounds to dump in non-zero
sections

509

ETL (3)

532
583

ARC (3)
FVM (3)

584

RIF (3)

590

DQW (2,3)

I/O warnings
At tempt to WRITE
beyond
fixed-length record
Ambiguous repeat count
Format and variable type do
not match
Reading
into
FORMAT
non-standard
Disk full or quota exceeded
- Please EXPUNGE, then type
CONTINUE

113
114

SSE (3)
SRE (3)

bounds
check
FORLIB
warnings
Substring range error
Subscript range error

-1

EOF

End of file
End of file

24
25

Record or
record
number
error
Bad format binary file
Attempt to read a
record
that has not been written
Illegal record number
Record too large memory
full
Cannot backspace image file
with no RECORDSIZE
Cannot
skiprecord
image
file with no RECORDSIZE
Non-digit
in
record
delimiter
Record size different from
that specified in OPEN
Unexpected
continuation
LSCW found
Attempt to WRITE beyound
variable or array
Record length negative or
zero

302
510

BBF
RNR

512
517

IRN
RTL

536

CBI

536

CSI

570

ICD

572

RSM

573

FCL

576

WBA

577

SLN

502

CSF (2,3)

535

BSI (3)

D-4

OPEN/CLOSE warnings
Can't set FORTRAN carriage
control attribute
BLOCKSIZE ignored:
device
is not a maqnetic tape
February 1987

FOROTS ERROR MESSAGES
Table D-l:
1st
Value

FOROTS Error Codes (Cont.)
2nd
Value(5)

Code (1)

Meaning

541

UOA (3 )

542

NCK (3)

550
588

CQF (1,3)
BSC (2,3)

595

OGX (1,3)

609

NNI(1,3)

Unknown
keyword,
OPEN
igno red
OPEN-only keyword specified
in CLOSE, ignored
Cannot QUEUE file
Bytesize in OPEN different
from FDB, FDB value used
Galaxy
version
not
2
supported
Node name ignored

J
J

250+n
250+n
250+n
610

CLS
RNM
CLS
DEL
RNM
UDO

APP (2 )

240

UOF (2 )
FRR

240

RR1

240

RRR

242
243

NFC (1 )
CIR

244

RLB

245
246

NSD
FBR

247

BLZ

248

lAC

249

IDM

250+n
405

OPN ( 1)
PPN (2 )

503

CEF (2 )

506
540

ICA
SDO

569

TFM

585

IAV

CLOSE error
Cannot CLOSE file
Cannot RENAME file
" Close" FILOP. error n ( 4 )
"Delete" FILOP. error n (4 )
"Rename" FILOP. error n (4)
Unsupported /DISPOSE option
for RMS files

(2)
(2 )
(1 )
(1 )
(1 )
(2 )

FORTRAN-10/20 VII

OPEN error
Cannot set up to append
to
magnetic tape file
USEROPEN failed
/RECORDTYPE:FIXED requires
/RECORDSIZE
Random
requires
I/O
RECORDSIZE
specifier
in
OPEN statement
Random
requires
I/O
/RECORDSIZE
Too many OPEN units
/CARRIAGECONTROL:TRANSLATED
illegal
this
with
/RECORDTYPE
larger
than
/RECORDSIZE
/BLOCKSIZE
No such device
BLOCKSIZE must be specified
for RECORDTYPE='FIXED'
integral
BLOCKSIZE must be
multiple of RECL
Specified ACCESS
illegal
for this device
is
Specified MODE
illegal
for this device
Cannot OPEN file
JSYS error - PPN cannot be
translated
End
of
command
file
encountered
Incompatible attributes
Same device open on another
unit
conflicting
with
specifiers
Tape format conflicts with
OPEN statement or default
Illegal value
for
OPEN
specifier
D-5

February 1987

FOROTS ERROR MESSAGES
Table D-1:
1st
Value

FOROTS Error Codes (Cont.)
2nd
Va1ue(5)

Cod e (1)

Meaning

600

ATC (2)

601

URT (2)

602
603
604

URS (2)
BME (2)
MEU (2)

606
607
608

KER (2)
TKS (2)
RMS (2)

611

DFR (2)

612

KDB (2)

613

RIR (2)

615

RNA (2)

File's attribute different
from that specified in OPEN
RECORDTYPE
value
not
supported
for
this file
type
Unsupported remote system
Bytesize must be 8
Missing
EXTERNAL
for
USEROPEN routine
Missing required KEY=
Too many KEY= specifiers
RMS
file
requries
ORGANIZATION=
Default
/RECORDTYPE:FIXED
requires /RECORDSIZE
KEY
datatype
CHARACTER
conflicts with BYTESIZE
New RELATIVE or
INDEXED
file requires /RECL
RMS not available on SYS:

315

CDR

315

CDS

593

POI

594

CDF

605

KRI(2)

614

SIF(2)

239

IUN

321

CFC (3)

322

CFF (3)

323

CFR (2,3)

323

CFX (2,3)

323

CFR (I, 3)

323

CFX (1,3)

324

CFK (3)

Mixed ACCESS modes
Can't do direct I/O
to
sequential file
Can't do sequential I/O to
direct file
Illegal for DIRECT (RANDOM)
files
Can't
determine
whether
formatted or unformatted
Can't do keyed I/O to this
file type
Statement illegal for
this
type of access

Illegal logical unit number
Illegal unit number

FORTRAN-10/20 VII

D-6

Compatibility error
FORTRAN-77
and
VMS:
Carriage control character
VMS
incompatibility:
Intrinsic routine
invoked
incompatibly
FORTRAN-77
extension:
FORTRAN-20 supplied routine
invoked
FORTRAN-77
and
VMS:
FORTRAN-20 supplied routine
invoked
FORTRAN-77
extension:
FORTRAN-10 supplied routine
invoked
FORTRAN-77
and
VMS:
FORTRAN-10 supplied routine
invoked
FORTRAN-77 and VMS: Keyword
[keyword]

February 1987

FOROTS ERROR MESSAGES
Table D-l:
1st
Value

FOROTS Error Codes (Cont.)
2nd
Value(5)

Code.( 1)

Meaning

325

CFT (3 )

326

CFO (3)

327

CFG (3 )

328

CFD (3 )

and
FORTRAN-77
VMS:
Trailing spaces
in output
reco rd
FORTRAN-77
extension:
Overlap
of
character
assignments
and
FORTRAN-77:
VMS:
G
format descriptor used with
character
VMS
incompatibility:
I
Format default width used

310

RBR

REREAD error
REREAD not proceeded
READ

45
241

ESV

241
241
533
539
544

USW (1 )
ASW (1 )
DLT
EDS/EDA (2)
NDI (1 )

545
546
547
548

IPP
TMF
NSI
IDD

551

NQS (1 )

263

CDT

554

CWL

301
303
306

ILF
1FT
DLF

524

RIC

538
552
553
575

IRC
IHC
IFW
UDT

583

FVF

OPEN/CLOSE statement syntax
errors
Unknown
ambiguous
or
keyword
Unknown swi tch
Ambiguous switch
Dialog string too long
Error parsing DIALOG string
No device specified with

":"

(1 )
(1 )
(1 )
(1 )

Illegal PPN
Too many SFDs
Null SFD
Illegal character in DIALOG
str ing
PADCHAR must
be
single
character in double quotes

WRITE on READ-only file
Cannot WRITE to READ-only
file
Cannot write a file with
MODE=LINED

FORTRAN-10/20 VII

Syntax error in FORMAT
Illegal character in FORMAT
Illegal FORMAT
Data in I/O list but not in
FORMAT
Reading
into
character
format illegal
Illegal repeat count
Illegal Hollerith constant
III egal field width
Undefined data
type
or
internal FOROTS error
Format and variable type do
not match

D-7

February 1987

FOROTS ERROR MESSAGES
Table D-1:
1st
Value

FOROTS Error Codes

(Cont~)

2nd
value(5)

Code (1)

Meaning

307

ILC

Input conversion error
Illegal character in data

501
508
574
579
581
582

UNS
IOL
IMV
IDI (1)
DLL (1)
IWI

599

ICE

FOROTS calling errors
Unit not specified
Bad I/O list
Illegal MTOP value
Illegal DUMP mode I/O list
Dump mode I/O list too long
Illegal to initiate another
I/O statement
Illegal
length
for
character expression

100

I LM (2)
UMO (2,3)
NLT

101

SLT

102

ALT

530
537

UTE (1)'
UTO (1)

537

UTO (1,3)

587

ITE (1)

309
513

VNN
NEQ

514
515

NRP
ILN

516
519

ILS
CCC

521
522

RPE
SNV

580
596

NLS
NEC

597

ISS

598

SNQ

J
J

FORTRAN-10/20 VII

Error
in
magnetic
tape
operations
Unexpected MTOPR% error (2)
Error trying to set tape
<statement>
not
allowed
with
Labeled Magnetic tape
files
Input/Output switching not
allowed
with
Labeled
Magnetic tape file
/ACCESS='APPEND'
not
allowed
with
Labeled
Magnetic tape files
Unexpected TAPOP. error
Unexpected
TAPOP.
error
trying to set parameters
Unexpected
TAPOP.
error
trying to set parameters
Tape is not usable by your
job
NAMELIST data errors
Variable not in namelist
"=" not found
in namelist
data
Missing right paren
Variable or namelist does.
not start with letter
Illegal subscript
Cannot convert constant to
correct type
Illegal repeat count
Sign with null value
Null string illegal
when
Found
character
expecting ":"
Substring
descriptor
ill egal
String not within single
quotes

D-8

February 1987

FOROTS ERROR MESSAGES
Table D-l:
1st
Value

FOROTS Error Codes (Cont.)
2nd
Val ue (5)

Code (1)

Meaning

ISW
IOE
OSW
ISW
OSW
IOE
MFU
DQE

Unclassifiable
device
errors
Cannot switch to input
General purpose I/O error
Cannot switch to output
Cannot switch to input
Cannot switch to output
General-purpose I/O error
Memory full
Disk full or quota exceeded

98
J
J
J
250+n
250+n
400
586
590

(2)
(2)
(2)
(1 )
(1 )
(1 )
(2 )

No ERRSNS values
BLK (3 ,6)
CCP
CDP
CGD
DBM
DST
EFS
FFX

(1 ,6)
(1,6)
(6)
(6 )
(1 ,6)
(6)
(6)

IEM
IJE
MFU
NOR
PAG

(6)
(2,6)
(6)
(3,6)
(3 ,6)

PGD (6)
POV (6)
SNH (6 )
TDT (6 )

Blocks allocated but not
deallocated
Cannot create page
Cannot destroy page
Can't get DBMS
DBMS not loaded
Error in dialog string
Enter correct file specs
FOROP function code exceeds
range
Error in memory management
"Impossible" JSYS error
Memory full
Error number out of range
Pages allocated but
not
deallocated
Deallocating
more
pages
than allocated
PDL overflow
Internal FOROTS error
Trap occured during
trap
processing

(1) TOPS-10 only
(2) TOPS-20 only
(3) This is a- warning, not an error.
The error cannot be
trapped with an ERR= branch, but IOSTAT and ERRSNS will be
set.
(4) See the TOPS-10 Monitor Calls Manual
FILOP. error codes and their meanings.
(5) "J" means the TOPS-20 JSYS error code.
be between 600000 and 610000 (octal).

for

the

This number will

(6) No ERRSNS values

(7) Where n is the number of times the error occurs

D-9

list

FOROTS ERROR MESSAGES
APR and LIB errors are usually reported as warnings, and the program
continues.
The number of APR and LIB errors listed on the user's
terminal can be changed by the FORTRAN Library Subroutine ERRSET (see
Section 13.4.1.14).
The I/O errors either cause messages to be
printed on the terminal or can be trapped by an error exit argument
(ERR=statement number) on OPEN, ~EAD, WRITE, and CLOSE.
The FORTRAN Library Subroutine ERRSNS (see Section 13.4.1.15) allows
you to find out which I/O error occurred. When called, ERRSNS returns
one or two integer values that describe the status of the last I/O
operation
performed
by FOROTS.
(The second integer value is
opt ional .) Fo r exampl e,
CALL ERRSNS (I,J)
calls this subroutine.

D.l

J, the second integer value, is optional.

ALPHABETICAL DESCRIPTION OF FOROTS MESSAGES

This section contains alphabetical descriptions of each warning and
fatal error message that is generated by FOROTS during program
execution. Each message is first listed; then is followed by a brief
description of how it is generated; followed by, in some cases, a
recovery procedure; and finally, where applicable, followed by the
ERRSNS values associated with the message.
ALT?

/ACCESS='APPEND' not allowed with Labeled Magnetic tape files
Cause:

You specified ACCESS='APPEND' in an OPEN statement when
using a labeled magnetic tape file.
This is not
allowed.

Recovery: Change the OPEN statement.
ERRSNS values:
ALZ?

First Value = 96

Second Val ue

102

Argument less than or equal to zero
Cause:

An argument (such as a memory size) was specified
a value less than or equal to zero.

with

Recovery: Specify the correct value for the argument.
ERRSNS values:
APP?
only)

First Value = 21

Can't setup to append to magtape
Cause:

Second Value = 125
file

<JSYS

error>

(TOPS-20

The MTOPR% JSYS failed trying to position the
tape with the function .MOFWF or .MOBKR.

magnetic

Recovery: Use the information provided to determine a recovery.
ERRSNS values:

FORTRAN-10/20 VII

First Value = 30

D-10

Second Value = JSYS error
number

February 1987

FOROTS ERROR MESSAGES
AQS?

First argument to SORT -must be a quoted string
Cause:

The SORT routine detected that the first argument was
of type CHARACTER, but the string was not word aligned.

Recovery:

If the first argument to the SORT program is of type
CHARACTER, it must be word aligned and terminated by an
ASCII null.
The most reliable way to generate such an
argument is to use,a character constant.

ERRSNS values:

First Value

Second Value = ,112

ARC % Ambiguous repeat count
Cause:

In a ,FORMAT statement,
a
number between two
format
specifiers can be considered belonging to either one.

Recovery: Insert a comma before or after the number, depending on
which specifier the number belongs with.
ERRSNS values:
ASW?

First Value = 22

Ambiguous switch /<sw>
Cause:

Second Value = 532

(TOPS-10 only)

In dialog mode, a switch was specified, but not enough
of the switch was given to uniquely specify which
switch was intended.

Recovery: Retype the line, completely specifying which switch you
mean.
ERRSNS values:

First Value

ATC?
File's <attribute> conflicts with
(TOPS-20 only)
Cause:

Second Value
OPEN

statement

241
or

default

You opened
an
RMS
file
with
RECORDTYPE=
or
ORGANIZATION= attributes in your OPEN statement, which
conflict with the actual attributes of the file.

Recovery: Specify the correct attributes, or omit them from
to use the file's attributes.
ERRSNS values:
BBF?

First Value = 30

OPEN

Second Value = 600

Bad format binary file
Cause:

The control information stored
in a binary file
is
incorrect.
The
file
cannot
be
read
using
MODE='BINARY'.
This error can be caused when the file
you are
reading was not written by FORTRAN using
MODE='BINARY'.

Recovery: Make sure that you are using the correct file and
mode.
Any file can be read with MODE='IMAGE'.
Section 11.3.24)
ERRSNS values:

FORTRAN-10/20 VII

First Value

D-11

Second Value

data
{See

302

February 1987

FOROTS ERROR MESSAGES
BLK % Pages allocated but not deallocated
Cause:

Internal FOROTS error in memory management.

Recovery: Submit an SPR and include your program.
BLZ?

BLOCKSIZE must be integral multiple of RECL
Cause:

You specified a BLOCKSIZE in an OPEN statement that was
not a multiple of the RECORDSIZE you specified when
using magnetic tape files.

Recovery: Change the BLOCKSIZE value to
RECORDSIZE.
ERRSNS values:
BME?

First Value

multiple

Second Value

the

247

Bytesize must be eight (TOPS-20 only)
Cause:

You specified a BYTESIZE= other than 8 when accessing a
VMS RMS file.

Recovery: Change BYTESIZE=, or omit the specifier to default
files to 8.
ERRSNS values:

First Value

Second Value

VMS

603

BSC % Bytesize in OPEN different from FOB, FOB value used
Cause:

You specified a BYTESIZE= value in OPEN that was
different than the one in the File Descriptor Block
(FOB). The FOB value ,.is used.

Recovery: If FOB value is correct, do not specify a BYTESIZE
value in OPEN, else specify BYTESIZE value in OPEN.
ERRSNS values:

First Value = 26

BSI % Blocksize ignored:
Cause:

device is not a magnetic tape

A BLOCKSIZE specifier was given in an OPEN statement
(see Section 11.3.4)
but was not used because the
device being opened is not a magnetic tape.

ERRSNS values:
CBI?

Second Value = 588

First Value = 26

Second Value = 535

Can't BACKSPACE IMAGE file with no RECORDSIZE
Cause:

An OPEN statement with the MODE=' IMAGE' specifier (but
not the RECORDSIZE= specifier) was executed prior to a
BACKSPACE statement that referred to the open unit.

Recovery: If you are using fixed-length records, specify the
RECORDSIZE parameter in the OPEN statement (see Section
11.3.~3).
Otherwise, the BACKSPACE cannot be done.
ERRSNS values:

FORTRAN-10/20 VII

First Value = 25

0-12

Second Value = 536

February 1987

FOROTS ERROR MESSAGES
CCC?

Can't convert constant to correct type
Cause:

In NAMELIST input, a variable was assigned a value that
does not match.
For example,
if C is a complex
variable, the input:
C=.TRUE.
is in error,
number.

since

.TRUE.

not

legal

complex

Recovery: Correct the error in the source program.
ERRSNS values:
CCP?

Can't create page <n>
Cause:

CCS?

First Value = 97
(PAGE.

Second Value = 519

error <n»

(TOPS-10 only)

FOROTS attempted to use a page of memory for some task,
but was unable to.
The monitor error code gives the
reason.
This can be caused by
erroneous
MACRO
subroutines.
If no such cause
is found, it is an
internal FOROTS error.

Not enough memory for creating character stack
Cause:

A character stack was requested that was larger than 36
sections
(larger that the maximum virtual memory
available) •

Recovery: Specify correct call argument.
ERRSNS values:
COF?

124

The specified
file has had
both
formatted
and
unformatted I/O operations (or OPENs) performed on it.

Recovery: Use I/O operations
specifier.

and

OPENs

ERRSNS values:

Second Value

First Value

Can't destroy page <n>
Cause:

COR?

Second Value

Can't determine whether formatted or unformatted
Cause:

COP?

First Value = 21

(PAGE.

error <n»

with

the

same

FORM=

594

(TOPS-10 only)

FOROTS attempted to use a page of memory for some task,
but was
unable to.
The monitor error code gives the
reason.
This can be caused by
erroneous
MACRO
subroutines.
If no such cause
is found, it is an
internal FOROTS error.

Can't do direct I/O to sequential file
Cause:

An attempt was made to perform I/O to
already open in a conflicting mode.

Recovery:

file

Second Value

315

that

Open file in the appropriate mode.

ERRSNS values:

First Value = 31
0-13

February 1987

FOROTS ERROR MESSAGES
CDS?

Can't do sequential I/O to direct file
Cause:

An attempt was made to perform I/O to a
already open in a conflicting mode.

Recovery:

Open file is the appropriate mode.

ERRSNS values:
CDT?

First Value = 31

Second Value

file

that

315

Can't <read/write> an <input/output>-only file
Cause:

An attempt was made to perform I/O to a file, but the
file is not open for I/O in the appropriate direction.

Recovery: Open the file with ACCESS='SEQINOUT' or 'RANDOM', as
appropriate.
It is not possible to open a file for
APPEND access and then read from it.
(See Section
11.3.1.)
ERRSNS values:
CEF?

First Value

Second Value

263

End of command file encountered (TOPS-20 only)
Cause:

An indirect file was specified as
and the end of the file was
terminator character (line-feed).

a DIALOG argument,
encountered before a

Recovery: Edit the file and insert a line-feed.
ERRSNS values:

First Value = 30

CFC % Fortran-77 and VMS:
Cause:

Second Value = 503

Carriage control character

A carriage-control character
was
used
that
incompatible with ANSI FORTRAN and VAX FORTRAN.

Recovery: If you want the program to be compatible
FORTRAN
or
VAX
FORTRAN,
use
a
carriage-control character.
ERRSNS values:

First Value = 33

CFD % VMS incompatibility:
Cause:

wi th ANSI
compatible

Second Value = 321

I format default width used

A format descriptor was used with a default width
is incompatible with VAX FORTRAN.

that

Recovery: If you want your program to be compatible with VAX
FORTRAN, then do not default the width of this edit
descriptor.
ERRSNS values:

FORTRAN-10/20 VI)

First Value

D-14

Second Value

328

February 1987

FOROTS ERROR MESSAGES
CFF % VMS incompatibility:
Cause:

Intrinsic routine invoked incompatibly

An intrinsic routine was
invoked
in
a
method
incompatible with VAX FORTRAN (such as use of an
EXTERNAL statement for an intrinsic function).

Recovery: If you want the program to be compatible with VAX
FORTRAN, chang~ to a method of invoking intrinsic
routines that is compatible with VAX.
ERRSNS values:

First Value = 33

Second Value

322

CFG % Fortran-77 and VMS: G format descriptor used with character
Cause:

The G format descriptor was used with character data,
which is an extension to ANSI FORTRAN and VAX FORTRAN.

Recovery: If you want the program compatible with ANSI FORTRAN or
VAX FORTRAN, do not use the G format descriptor to edit
character data.
ERRSNS values:

First Value

CFK % Fortran-77 and VMS:
Cause:

Second Value

327

Keyword [keyword]

An OPEN or CLOSE keyword was used that is
with ANSI FORTRAN and VAX FORTRAN.

incompatible

Recovery: If you want the program compatible with ANSI FORTRAN or
VAX FORTRAN, use a compatible OPEN or CLOSE keyword.
ERRSNS values:

First Value =33

CFO % Fortran-77 extension:
Cause:

Second Value = 324

Overlap of character assignments

A character assignment statement was used in which the
character positions defined in the character variable,
array element or substring on the left of the equal
sign are referenced in the character expression on the
right of the equal sign.
This is incompatible with
ANSI FORTRAN.

Recovery: If you want the program to be compatible with ANSI
FORTRAN, use a character assignment statement that does
not overlap the character expression and the character
variable, array element, or substring.
ERRSNS values:
CFR % Fortran-77
(TOPS-20 only)
CFR % Fortran-77
(TOPS-10 onl y)
Cause:

First Value = 33

Second Value

326

extension:

FORTRAN-20

supplied

routine

invoked

extension:

FORTRAN-l 0

supplied

routine

invoked

A FORTRAN-10/20-supplied subroutine was invoked that is
not available with ANSI FORTRAN.

Recovery: If you want your program to be compatible
FORTRAN, use a compatible subroutine.
ERRSNS values:

First Value = 33
D-15

Second Val ue

ANSI

323
February 1987

FOROTS ERROR MESSAGES
CFS?

Can't find SYS:SORT.EXE Cause:

<JSYS error> (TOPS-20 only)

The file SORT.EXE cannot be found on SYS:.
A
supplied error message will give more detail.

Recovery: Use the
information provided by the
determine the proper course of action.
ERRSNS values:

First Value = 21

CFT % Fortran-77 and VMS:
Cause:

monitor

Second Value = 119

Trailing spaces in output record

Your program contains a FORMAT that specifies trailing
blanks
(X format and $ format).
In this case,
FORTRAN-10/20 preserves the trailing spaces.

Recovery: If you want the program compatible with ANSI FORTRAN
and VAX FORTRAN, do not use this form of the FORMAT
statement.
ERRSNS values:

First Value

Second Value

325

FORTRAN-20 supplied
routine
invoked
CFX % Fortran-77 and VMS:
(TOPS-20 onl y)
FORTRAN-10 supplied routine invoked (TOPS-10
CFX % Fortan-77 and VMS:
only)
Cause:

A FORTRAN-10/20-supplied subroutine was invoked that is
not available with ANSI FORTRAN or VAX FORTRAN.

Recovery: If you want the program to be compatible with ANSI
FORTRAN or VAX FORTRAN, use a compatible subroutine.
ERRSNS values:
CGP?

First Value = 33

Second Value = 323

Can't get pages 600:677 for SORT
Cause:

The SORT subroutine brings the SORT program into core
in pages 600 through 677.
Some of these pages were
already occupied by programs or data at the time that
SORT was called.

Recovery: Decrease the size of your program.
Having fewer
files
open
or using a BUFFERCOUNT=l specifier in OPEN
statement may help (see Section 11.3.5).
If this does
not help,
you can segment the program by using LINK's
overlay facility (see the LINK Reference Manual).
ERRSNS values:
CGS?

First Value = 21

Can't get SYS:SORT.EXE Cause:

Second Value = 116

<JSYS error> (TOPS-20 only)

The file SORT.EXE was found on SYS:, however for some
reason
it could not be merged into your program in
order to sort files.
A monitor supplied error message
will give more detail.

Recovery: Use the
information provided by the
determine the proper course of action.
ERRSNS values:

First Value = 21

0-16

Second Value

monitor

120
February 1987

FOROTS ERROR MESSAGES
CIR?

/CARRIAGECONTROL:TRANSLATED illegal with this /RECORDTYPE
Cause:

You specified CARRIAGECONTROL='TRANSLATED', which
not allowed for the RECORDTYPE you specified.

Recovery: Change the OPEN statement.
ERRSNS values:
CLE?

First Value = 30

Second Val ue

243

Concatenation result larger than expected
Cause:

The specified substring bounds are out of range.

Recovery: Specify legal substring bounds.
ERRSNS values:
CLS?

First Value = 21

Second Value

108

CLOSE failed, <I/O error message> (TOPS-10 only)
Cause:

A CLOSE UUO. or FILOP. CLOSE function failed.

Recovery: Use the information provided to determine a recovery.
ERRSNS values:

First Value = 28

Second Value = 250+n

CPP % Can't preallocate pages 600:677 for SORT
Cause:

A call to
preallocate
allocated.

the SRTINI subroutine was
SORT's pages because they

unable
to
were already

Recovery: Decrease the size of your program. Having fewer
files
open or using a BUFFERCOUNT=l specifier in the OPEN
statement may help (see Section 11.3.5).
ERRSNS values:

First Value = 21

CQF % Can't queue file:
Cause:

Second Value

QUEUE. UUO ERROR <N> (TOPS-10 only)

This error may occur when executing a CLOSE statement
in which the DISPOSE specifier is given with one of the
values:
'LIST',
'PRINT', or
'PUNCH', and
GALAXY
release 4 is running.
(See Section 11.5.4.)

Recovery: Refer to the TOPS-10 Monitor Calls Manual
explanation of the QUEUE. error number <n>.
ERRSNS values:
CRP?

121

First Value = 26

Second Value

for

= 550

Can't return pages 600:677 after call to SORT (TOPS-10 only)
Cause:

Before the SORT subroutine returns to the user,
it
tries to deallocate the pages that the SORT program
resides in.
The pages could not be deallocated.
This
is not expected to occur.

Recovery: Submit an SPR.
ERRSNS values:

First Value

21
D-17

Second Value

117
February 1987

FOROTS ERROR MESSAGES
CSF % Can't set FORTRAN carriage control attribute (TOPS-20 only)
Cause:

You are running a TOPS-20 system that does not have
Edit
2981, which allows support for the FORTRAN
carriage-control attribute.

Recovery: Autopatch the monitor to include Edit 2981.
ERRSNS values:
CSI?

First Value = 26

Second Value = 502

Can't skiprecord image file with no RECORDSIZE
Cause:

A SKIPRECORD statement was executed for a file opened
with
MODE=IMAGE,
and
the file did not have a
RECORDSIZE(RECL) specified. Since there are no record
markers in IMAGE mode files, FOROTS cannot determine
how far to skip. .

Recovery: Add a RECL= specifier (see Section 11.3.33) to the OPEN
statement.
ERRSNS values:
CWL?

First Value = 25

536

Can't write a file with MODE='LINED'
Cause:

The program attempted to execute an output operation
such as a WRITE statement after an OPEN statement for
the same unit.
The OPEN statement
contained
a
MODE=' LINED' specifier.

Recovery: Change the OPEN statement
(See Section 11.3.24.)
ERRSNS values:
CWU?

Second Value

DIVERT:
Cause:

First Value = 47

specifier

Second Value

MODE='ASCII'.
554

Can't write to unit <n>
DIVERT file is not opened for output.

Recovery: Open DIVERT file for output.
ERRSNS values:
DBM?

First Value = 21

Second Value

107

DBMS not loaded
Cause:

A DBMS call to a.sharable FOROTS was attempted
DBMS.

without

Recovery: The system manager must build DBMS into sharable FOROTS
or remove the file from FORLIB that directs the DBMS
call to' the sharable FOROTS.

0-18

February 1987

FOROTS ERROR MESSAGES
DEL?

Can't delete file:<FILOP.
Cause:

error message> (TOPS-10 only)

The "DELETE" FILOP.

failed.

The file is not deleted.

Recovery: Usually you can correct the problem when the program is
finished
and delete the file with the monitor DELETE
command.
ERRSNS values:
DFR?

First Value

Second Value

250+n

Default RECORDTYPE:FIXED requires /RECORDSIZE (TOPS-20 only)
Cause:

You attempted to create a new RMS relative or
indexed
file with a default /RECORDTYPE of FIXED, but did not
specify /RECOROSIZE.

Recovery: Specify /RECORDSIZE.
ERRSNS values:

First Value = 30

Second Value

611

DLF % Data in 10 list but not in format
Cause:

An I/O statement has requested data to be
but
the
FORMAT
statement does not
descriptor that would translate the data.

Recovery: tix the FORMAT statement.
following descriptors:

transferred,
specify any

It must contain one

the

A,B,E,F,G,I,L,O,Q,R,Z
ERRSNS values:
DLL?

First Value = 62

Second Value

306

DUMP mode I/O list too long (TOPS-10 only)
Cause:

The I/O list specified for a DUMP mode READ or WRITE
statement is too long for the FOROTS internal DUMP mode
control list.

Recovery: Split the I/O list across two or
statements.
ERRSNS values:

First Value

READ

Second Value

581

DMA % Must give lower and upper bounds to dump
(TOPS-20 only)
Cause:

non-zero

WRITE

sections

A call to PDUMP or DUMP was made without specifying
memory bounds.
In section zero, this is interpreted as
'all of memory'.
For extended addressing,
lower and
upper bounds must be specified.

Recovery: Specify lower and upper bounds for memory.
ERRSNS values:

FORTRAN-10/20 VII

First Value = 21

D-19

Second Value = 126

February 1987

FOROTS ERROR MESSAGES
DQE?

Disk full or quota exceeded
Cause:

The disk quota for the disk on which a file is being
written is exhausted, or the entire disk structure is
full.
If this error is encountered while running under
batch, the program is aborted and an attempt is made to
close all files.
If this error is encountered while
timesharing, the user is requested to type an EXPUNGE
command and then a CONTINUE command.

Recovery: EXPUNGE or create
structure.
ERRSNS values:

First Value = 98

room

the

specified

Second Value

disk

590

DQW % Disk full or quota exceeded - please EXPUNGE, then type CONTINUE
(TOPS-20 only)
Cause:

The file or files being written on the disk have either
exhausted your disk quota or filled the structure.

Recovery: FOROTS leaves the terminal at EXEC level so that you
have more options to use to provide space on the disk
structure. You can type CONTINUE to resume processing
if you don't reset the current fork.
ERRSNS values:

First Value = 22

Second Value

590

DSS % DISPOSE='SAVE' assumed - device is not a disk
Cause:

A DISPOSE value other than 'SAVE' was specified for a
file on a device other than disk.
(See Section
11.3.15.)

ERRSNS values:
DST?

Fi rst Val ue

Second Value = 549

Error in dialog string (TOPS-10 only)
Cause:

A syntax error in the DIALOG= specifier.

Recovery: Correct the error in the program
(See Section 11.5.3.)

DIALOG

mode.

DTL % Dialog string too long
Cause:

The argument to DIALOG= cannot be parsed because it
too long.

Recovery: Use a shorter
ERRSNS values:

strin~.

First Value = 45

0-20

Second Value

533

February 1987

FOROTS ERROR MESSAGES
ECS?

Not enough memory for expanding character stack
More memory than is available on a KL10 was requested
by either a dynamic concatenation (concatenation of
character variables of length*) or by a call
to the
ALCCHR subroutine.

Cause:

Recovery: Reduce the size of your concatenation or argument to
ALCCHR.
If you are running extended addressing, you
can present parts of the character stack by invoking
SORT and dynamic libraries at the beginning of the
program.
ERRSNS values:
EDS/EDA?

First value

Second Value = 124

Error in DIALOG string - <message> (TOPS-20 only)

Cause:

A syntax error in the DIALOG= specifier.
11.5.3.)

(See

Section

Recovery: Retype the specifier correctly in DIALOG mode.
ERRSNS values:

First Value = 45

Second Value = 539

EFS [ Enter correct file specs ]
Cause:

Dialog mode. You should respond to this message by
entering any information you wish to change about the
indicated file.
This can include the device, filename,
directory, or any OPEN parameter.
The form of a
response is:

TOPS-10
TOPS-20

DEV:FILE.EXT[DIRECTORY] /SWITCH:VAL /SWITCH:VAL
DEV:<DIRECTORY>FILE.EXT.GEN /SWITCH:VAL /SWITCH:VAL
All parts of this specification are optional.

EOF?

End of file
Cause:

An input statement has attempted to read more data than
the file contains.

Recovery: Use an END= specifi~r
lengthen the file.
ERRSNS values:
ESV?

First value

the

READ

Second Value

statement,

-1

<unknown/ambiguous> keyword value /<switch>:<value> (TOPS-20
only)
Cause:

A switch entered in dialog mode was not
was not specified uniquely.

recognized

Recovery: Retype the line, specifying the correct switch.
ERRSNS values:

First Value = 45

D-2l

Second Value = 241

February 1987

FOROTS ERROR MESSAGES
ETL % Attempt to write beyond fixed-length record
Cause:

In an ENCODE statement, the format specified more
characters than the string will
hold.
The excess
characters are ignored.
(See Section 10.12.)

Recovery: Shorten the format or lengthen the string.
ERRSNS values:
FBR?

First Value = 22

Second Value

= 509

BLOCKSIZE must be specified for RECORDTYPE='FIXED'
Cause:

For magnetic tape files,
a RECORDTYPE='FIXED'
specified, and no BLOCKSIZE value was specified.

was

Recovery: If you specify RECORDTYPE='FIXED' for magnetic tapes,
you must specify a BLOCKSIZE, which must be a multiple
of the RECORDSIZE.
ERRSNS values:
FCL?

Second Value

First Value = 30

246

Found unexpected continuation LSCW
Cause:

A RECL(RECORDSIZE)
has been specified
in an OPEN
statement, and FOROTS has encountered a continuation
(type 2) logical segment control word
(LSCW).
This
type LSCW is never written
in fixed-length binary
records.

Recovery: Remove the RECL(RECORDSIZE) specification in
statement.
ERRSNS values:

Seco'nd Val ue

First Value = 25

the

OPEN

573

FDC % Floating divide check
Cause:

The program contains a floating-point division in which
the divisor
is too small compared to the dividend to
yield a result that is in the floating-point range.

Recovery: Correct
program
so
floating-point range.
ERRSNS values:

FFX?

that

First Value = 5

division

within

Second Value = n
where n is
the number
of times the
error occurs

FOROP. function code exceeds range
Cause:

A library routine has called for an operation that
is
not
available.
This can be caused by using a
mismatched FOROTS and FORLIB, or by an erroneous MACRO
subroutine.

Recovery: Make sure the versions of FOROTS.EXE and FORLIB.REL
that you are loading from are matching versions.
If
this doesn't help,
find
the subroutine causing the
problem and alter or remove the erroneous call.
FORTRAN-10/20 VII

February 1987

FOROTS ERROR MESSAGES
FOV % Floating overflow
Cause:

A REAL or DOUBLE PRECISION number was read that is too
large in magnitude (see Chapter 3). This is only a
warning and does not stop execution of the program.
The results of a calculation that overflows are set to
the largest representable number with the sign of the
correct result.

Recovery: Modify the data so that its values fall in the range of
values that can be represented for the data type.
ERRSNS values:

= 3 or 4 Second Value = n

First Value

where n is
the number
of times the
error occurs

FRR?

/RECORDTYPE:FIXED requries /RECORDSIZE
Cause:

A RECORDTYPE='FIXED' was specified in an OPEN statement
without a RECORDSIZE (RECL) specifier.

Recovery: Specify RECL
11.3.33) •
ERRSNS values:

the

First Value

OPEN

statement

(see

Second Value

240

Section

FTS % Output field width too small
Cause:

The field width specified in a FORMAT statement was not
large enough to allow the printing of the value being
output. For example, this error would occur if the
number 100 is output with the format specifier "12".

ERRSNS values:

First Value = 9

Second Value

FUN % Floating underflow
Cause:

A REAL or DOUBLE PRECISION number was read that is too
small in magnitude (see Chapter 3).
This only a
warning and does not stop execution of the program.
The result of a calculation that underflows is set to
zero.

Recovery: Modify the data so that its values fall in the range of
values that can be represented for the data type.
ERRSNS values:

First Value = 6 or 7 Second Value = n
where n is
the number
of times the
error occurs

D-23

February 1987

FOROTS ERROR MESSAGES
FVF?

Format and variable type do not match
Cause:

An attempt was made to READ or
with other than A or G format.

WRITE

character

Recovery: Specify A or G edit descriptors when reading
data.
ERRSNS values:

First Value

data

character

Second Value = 583

FVM % Format and variable type do not match
Cause:

An I/O statement has been executed that uses a format
edit descriptor with a type that does not match the I/O
list item being processed.

Recovery: Specify the appropriate format edit descriptor for
I/O list item.
ERRSNS values:
lAC?

Second Value

583

/ACCESS illegal for this device
Cause:

An attempt was made to OPEN a device for
access specified (or implied) is illegal.

which

the

Recovery: Change the ACCESS specifier in the OPEN statement
data transfer statement.
(See Section 11.3.1.)

ERRSNS values:
IAV?

First Value = 22

the

First Value = 30

Second Value = 248

Illegal value for OPEN specifier
Cause:

An OPEN statement specifier has
that specifier.

value

illegal

for

Recovery: Specify a legal value for that OPEN specifier.'
ERRSNS values:
ICA?

First Value

= 30

Second Value = 585

Incompatible attributes
Cause:

An illegal combination
spec i f i ed.

open

attributes

has

been

Recovery: Change one or more of the conflicting specifiers.
ERRSNS values:
ICD?

First Value = 30

Second Value = 506

Non-digit in record delimiter
Cause:

The format of the tape being read is
not
'0'
(DELIMITED). The Record Control Word (ReW) contained a
non-digit, or the data on the tape is incompatible.

Recovery: Specify the correct TAPEFORMAT in the OPEN statement.
ERRSNS values:

First Value = 25

D-24

Second Value = 570

February 1987

FOROTS ERROR MESSAGES
ICE?

Illegal length for character expression
Cause:

A program has specified a
zero length or negative
length character substring as an I/O list element.

Recovery: Fix program to specify a positive length substring.
ERRSNS values:

First Value = 81

Second Value = 599

IDC % Integer divide check
Cause:

Program contains an integer division by 0.

Recovery: Correct division in program.
ERRSNS values:

IDD ?

First Value = 1

Second Value = n
where n is
the number
of times the
error occurs

Illegal character <chr> (TOPS-10 only)
Cause:

An illegal character was encountered in dialog mode.

Recovery: Retype the response without illegal characters.
ERRSNS values:
IDI ?

First Value = 45

Second Value = 548

Illegal DUMP mode I/O list (TOPS-10 only)
Cause:

An I/O list entry has been specified whose entry size
(number of words)
is different from its increment.
This can only happen if an implied DO loop is specified
for the I/O list, the index increment is set to a value
other than 1, and
the program is compiled
with
/OPTIMIZE.

Recovery: Use an index increment of 1,
program with /OPTIMIZE.
ERRSNS values:
IDM?

First Value = 81

not

Second Value

the

579

/MODE:<mode> illegal for this device
Cause:

Not all devices can do I/O in all modes.
terminals cannot do binary I/O.

For

Recovery: Change the MODE= specifier or the device.
11.3.24.)
ERRSNS values:
IDU?

compile

DIVERT:
Cause:

First Value

Second Value

example,

(See Section
249

illegal to divert to unit <n>
Unit specified is an input-only device.

Recovery: Specify a unit for which output is legal.
ERRSNS values:

First Value

= 21
D-25

Second Value

= 104
February 1987

FOROTS ERROR MESSAGES
IEM?

FOROTS internal error in memory management
Cause:

This is an internal error that is not expected to
occur.
It means that the memory management routines
have detected a problem with their control information.

Recovery: Submit an SPR.
1FT?

Illegal FORMAT
Cause:

Your program contains an illegal FORMAT statement.

Recovery: Correct FORMAT statement.
ERRSNS values:
IFW?

First Value = 62

Second Value = 303

Illegal field width
Cause:

An illegal (negative) field width was
FORMAT statement.

specified

Recovery: Specify a legal field width in the FORMAT statement.
ERRSNS values:
IHC?

First Value = 62

Second Value = 553

Illegal Hollerith constant
Cause:

A format specification contains an H edit descriptor
that is not preceded by a length or does not contain
enough characters.

Recovery: Use the correct format for an H edit descript~r.
Section 12.4.2.)
ERRSNS values:
IJE?

First Value = 62

(See

Second Value = 552

"Impossible" JSYS error at <PC) - <JSYS ERROR) (TOPS-20 only)
Cause:

This is an internal FOROTS error that is not expected
to occur. A monitor call failed that was not expected
to.
A monitor-supplied error message may be
of
assistance in avoiding the problem.

Recovery: Submit an SPR.
ILC?

Illegal character in data
Cause:

A format descriptor that
nonnumeric character.

requires

number

found

Recovery: Fix the input data or FORMAT statement.
ERRSNS values:

FORTRAN-10/20 VII

First Value = 64

0-26

Second Value = 307

February 1987

FOROTS ERROR MESSAGES
ILF?

Illegal character in format
Cause:

ILM?

A format specification contains
defined meaning.

character

Recovery: Correct the error in the
program.

format

ERRSNS values:

Second Value

First Value

list

rerun

the

operation

that

and
301

Unexpected MTOPR% ERROR (TOPS-20 only)
Cause:

An error was encountered during a file
FOROTS did not expect.

Recovery: This type of error should not happen.
SPR.
ERRSNS values:

ILN?

with

First Value

Please submit an

Second Value

JSYS error
number

Variable or namelist does not start with letter
Cause:

NAMELIST input contains something other than a
legal
variable or NAMELIST name in a context where a variable
or NAMELIST name is required.

Recovery: Correct the source program with a
legal
NAMELIST name.
(See Section 12.6.)
ERRSNS values:
ILS?

Second Value

515

Illegal subscript
Cause:

In NAMELIST I/O, an illegal subscript was given for
array.

ERRSNS values:
IMV?

First Value = 97

variable

First Value

Second Value

516

Illegal MTOP value
Cause:

A MARCO program has issued an MTOP call to FOROTS
an illegal value for the function.

with

Recovery: Specify a legal function value in the call.
ERRSNS values:
IOE?

First Value

= 81

Second Value = 574

<10 error message>

Cause:

An I/O error has occurred.
The monitor error code
is
given, along with an interpretation of the probable
meaning of the error bits.
This message normally
indicates that the data recorded on an external device
has been damaged and cannot be read correctly.

ERRSNS values:

First Value

= 98

Second Value = 400

(TOPS-10)

JSYS error
number
(TOPS-20)
D-27

February 1987

FOROTS ERROR MESSAGES
IOL?

Bad format IO list
Cause:

The code generated by the compiler for an I/O list is
by this version of FOROTS.
The
not
understood
This
erroneous entry in the I/O list is ignored.
probably indicates an internal error in the compiler or
in FOROTS.

Recovery: Locate the problem area of the I/O
it.
First Value

ERRSNS values:

list

and

Second Value

508

simplify

IOV % Integer overflow
Cause:

An attempt was made to read data that was out of
for an integer variable.

ERRSNS values:
IPP?

First Value

range

Second Value = 0

Illegal PPN (TOPS-10 only)
Cause:

A directory specification starts with something that is
not a legal PPN specification.
The forms of legal PPNs
are:
[n,n], [n,], [,n], or [,]
where n represents a 1- to 6-digit octal number.

Recovery: Use a legal directory specification.
ERRSNS values:
IPN?

Second Value

First Value = 45

545

Illegal page number <n>
Cause:

A call to TOPMEN or SRTINI has specified a page
outside the range 1:777.

number

Recovery: Specify a correct page number
ERRSNS values:
IRC?

= 21

First Value

Second Value

122

Illegal repeat count
Cause:

An
illegal
statement.

repeat

count

was

given

FORMAT

Recovery: Correct the FORMAT statement.
ERRSNS values:
IRN?

First Value

= 62

Second Value

538

Illegal record number <n>
Cause:

A direct-access I/O statement has
number that is zero or negative.

specified

record

Recovery: Correct the invalid record number in the program.
ERRSNS values:

First Value = 25
D-28

Second Value

= 512
February 1987

FOROTS ERROR MESSAGES
ISS?

Illegal substring descriptor
Cause:

An I/O statement refers to an illegal
substring
delimiter (substring not within bounds of string).

Recovery: Correct the substring specifier.
ERRSNS values:
ISW?

First Value = 97

Second Value

597

Can't switch to input
Cause:

A file that was being written cannot be open for
output.
The file is either protected against reading,
or has been deleted before the OPEN for
read
is
executed.

Recovery: Specify correct protection for OPEN write.
ERRSNS values:

First Value = 98

Second Value

= 250+n
(TOPS-10)
JSYS error
number
(TOPS-20)

ITE?

Tape is not usable by your job (TOPS-10 only)
Cause:

A tape unit was specified or implied that is not owned
by your job, and is probably owned by another job.

Recovery: ASSIGN the drive or MOUNT the tape.
ERRSNS values:
IUN?

First Value = 96

Second Value

587

Illegal unit number <n>
Cause:

An I/O statement has specified a unit
negative or too large.

number

that

Recovery: Change the UNIT specifier value and rerun the program.
ERRSNS values:
IWI?

First Value = 32

Illegal to initiate
< I/O stat~ment)
Cause:

another

I/O

Second Value
statement

= 239

while

processing

An I/O statement, STOP statement, or PAUSE statement
has
been
initiated while processing another I/O
statement (such as in a function reference used as an
I/O list element), or while within a subroutine called
as a result of an I/O error through ERRSET.

Recovery: Remove the offending I/O statement, STOP statement,
PAUSE statement.
ERRSNS values:

First Value

0-29

Second Value

582

February 1987

FOROTS ERROR'MESSAGES
KDB?

KEY datatype CHARACTER conflicts with BYTESIZE (TOPS-20 only)
Cause:

You specified or defaulted a
KEY=
CHARACTER, but specified a BYTESIZE
(TOPS-20 file) or 8 (VMS file).

datatype
to
not equal to 7

Recovery: Specify correct BYTESIZE or omit to use system default.
ERRSNS values:
KER?

First Value = 30

Second Value = 612

New indexed file requires KEY= (TOPS-20 only)
Cause:

You attempted to create a new RMS indexed file
specifying a primary index.

without

Recovery: Specify one or more KEY= specifiers in your OPEN.
ERRSNS values:
KRI?

First Value = 30

Second Value = 606

Can't do keyed I/O to <sequential/relative> file (TOPS-20 only)
Cause:

You attempted a keyed (indexed)
sequential or relative file.

READ

RMS

Recovery: Correct the invalid READ statement in the program.
ERRSNS values:

First Value = 31

MEU? Missing EXTERNAL declaration in
(TOPS-20 only)

call

USEROPEN

routine

Cause:

You specified a USEROPEN routine but did not declare it
EXTERNAL in the program.

Recovery:

Supply the EXTERNAL declaration.

ERRSNS values:
MFU?

Second Value = 605

First Value = 30

Second Value = 604

Memory full
Cause:

There is insufficient memory to complete
the program.

execution

Recovery: Some memory can be saved by opening fewer files at a
time, by using BUFFERCOUNT=1 in OPEN statements, and by
using minimal tape block sizes. If these techniques do
not help, you can segment the program using LINK's
overlay facility (see the LINK Reference Manual).

FORTRAN-10/20 VII

D-30

February 1987

FOROTS ERROR MESSAGES
NCA?

No memory available for character stack
Cause:

For non-overlay programs, this message
indicates that
the memory manager has allocated all available space
between the user's low segment and FOROTS.
For overlay
programs,
this messages
indicates that /SPACE:0 has
been specified to LINK.

Recovery: For non-overlay programs, LINK with /OTS:NONSHARE.
For
overlay programs, specify at least 1000 to the /SPACE
switch in LINK (see the LINK Reference Manual).
ERRSNS values:

First Value

= 21

Second Value

NCK % <keyword> in CLOSE is meaningless Cause:

NCS?

= III

ignored

Options have been included in the CLOSE statement
are meaningless for closing the file.

Recovery: Use valid CLOSE options.

(See Section 11.5.)

ERRSNS values:

Second Value

First Value = 26

No character stack allocated Cause:

that

= 542

compiler error

An internal compiler error has occurred.

Recovery: Submit an SPR.
ERRSNS values:
NDI?

No device specified with ":"
Cause:

Second Value = 110

(TOPS-10 only)

An OPEN statement has specified a null device name.

ERRSNS values:
NEC?

First Value = 21

First Value

= 45

Second Value = 544

Found "<chr>" when expecting ":"
Cause:

Substring parameters not separated by":"

Recovery: Insert a ":" between substring parameters.
ERRSNS values:
NEQ?

First Value = 97

Found "<chr>" when expecting
Cause:

Second Value

= 596

"="

NAMELIST input found an illegal character in a
that requires an equal sign.

context

Recovery: Replace illegal character with equal sign.
ERRSNS values:

First Value = 97

D-3l

Second Value = 513

February 1987

FOROTS ERROR MESSAGES
NFC?

Too many open units (TOPS-10 only)
Cause:

On TOPS-10 monitors before version 7.00, at most,
16 units can be open at the same time.

only

Recovery: Arrange the program so that it never needs to have more
than 16 simultaneously open units.
ERRSNS values:
NLS?

First Value

= 30

Second Val ue

242

Null string illegal
Cause:

An attempt was made to input to a
during list-directed input.

zero

length

string

Recovery: Insert characters into the string, or remove the string
delimiting quotes.
ERRSNS values:
NLT?

First Value = 97

Second Value = 580

<statement> not allowed with Labeled Magnetic tape files
Cause:

When using labeled magnetic tape files, you used a
BACKSPACE or ENDFILE statement in your program, which
is no tall owed.

Recovery: Remove any BACKSPACE or ENDFILE
program.
ERRSNS values:

First Value = 96

statements

from

your

Second Value = 100

NNI % Node name ignored (TOPS-10 only)
Cause:

Remote file access is not supported on TOPS-10.

Recovery: Omit the node specification from the filespec.
ERRSNS values:
NQS?

Second Value = 609

PADCHAR must be single char in double quotes (TOPS-10 only)
Cause:

In dialog mode, the PADCHAR specifier must be
by the pad character in double quotes.

ERRSNS values:
NRP?

First Value = 26

First Value = 45

followed

Second Value = 551

Missing right paren
Cause:

In NAMELIST or list-directed complex input, the closing
right parenthesis that ends a complex number was not
found.

ERRSNS values:

FORTRAN-10/20 VII

Fi rst Val ue

D-32

Second Value = 514

February 1987

FOROTS ERROR MESSAGES
NSD?

No such device <dev>
Cause:

The specified device does not exist.

Recovery: Change the device name to one that does exist.
ERRSNS values:
NSI?

Second Value = 245

Null SFD (TOPS-10 only)
Cause:

A directory specification contains a null SFD.

ERRSNS values:
NSS?

First Value = 30

First Value = 45

Second Value = 547

No free section available for SORT (TOPS-20 only)
Cause:

SORT runs it its own section on machines that support
extended
addressing.
There are no free sections
available.
simple
is not
should

Recovery: There are 31 sections normally available when a
FORTRAN program runs.
If your application
trying to use extended addressing, this error
not occur, and you should submit an SPR.
ERRSNS values:

First Value = 21

Second Value = 118

OGX % Galaxy version 2 not supported (TOPS-10 only)
Cause:

Your system is using an unsupported version of GALAXY.

Recovery: Inform the system administrator
supported version of GALAXY.
ERRSNS values:
OPN?

Second Value

upgrade

the

595

Can't OPEN file
Cause:

The specified file could not be opened.
The reason
given is taken from the monitor error code (see the
TOPS-10 Monitor Calls Manual).

ERRSNS values:
OSW?

First Value = 26

FILOP.
Cause:

First Value = 30

Second Value = 250+n

error n - can't switch to output
An attempt to open a file for write access which
previously been open for read-only access failed.

has

Recovery: Change protection code; remove other file access.
ERRSNS values:

First Value = 98

Second Value = 250+n
(TOPS-10)
JSYS error
number
(TOPS-20)

D-33

February 1987

FOROTS ERROR MESSAGES
PAG % Pages allocated but not deallocated
Cause:

Internal FOROTS error in memory management.

Recovery: Submit an SPR and include your program.
PGD?

Deallocating more pages than allocated
Cause:

Internal FOROTS error in memory management.

Recovery: Submit an SPR and include your program.
POI?

<file positioning operation> Illegal for DIRECT (RANDOM)
Cause:

A file positioning operation (such
BACKSPACE)
was
attempted
on
a
DIRECT{RANDOM) access.

file

as REWIND
file open

or
for

Recovery: Remove the file positioning statement.
ERRSNS values:
PPN?

= 31

First Value

593

<JSYS error> (TOPS-20 only)
Cause:

A TOPS-20 OPEN statement has specified a PPN instead of
a
directory
name, but the PPN has no matching
directory.

Recovery: Specify the correct PPN, or
directory name instead.
ERRSNS values:
RBR?

Second Value

First Value = 30

better

yet,

Second Value

specify

the

405

REREAD not preceded by READ
Cause:

A REREAD statement was encountered before any READ
statement.
A READ must be executed first so there is
something to reread.

Recovery: Cause a READ statement to be executed first.
ERRSNS values:
RLB?

First Value = 39

Second Value = 310

/RECORDSIZE larger than /BLOCKSIZE
Cause:

A RECORDSIZE was specified in an OPEN statement that is
larger than the specified or implied BLOCKSIZE.

Recovery: Correct either RECORDSIZE or BLOCKSIZE,
BLOCKSIZE if it is not specified.
ERRSNS values:

First Value = 30

0-34

specify

Second Value = 244

February 1987

FOROTS ERROR MESSAGES
RIC?

Reading into character format illegal
Cause:

An attempt was made to READ into a character format.

Recovery: Correct program to avoid this construct. READ into a
character variable and use this variable
(perhaps
concatenated with other character expressions)
for a
modifiable format.
ERRSNS values:

First Value

Second Value

524

RIF % Reading into FORMAT nonstandard
Cause:

A READ statement was executed that reads data into a
Hollerith or quoted string in a FORMAT statement. This
is a practice contrary to the ANSI standard and is
likely not to work in future releases of FORTRAN-10/20.

Recovery: READ into character variables
and
use
(perhaps
concatenated with character constants) as the FORMAT.
ERRSNS values:
RIR?
only)

New RELATIVE or

INDEXED

file

Second Value = 584
requires

/RECORDSIZE

(TOPS-20

Cause:

You tried to create (STATUS='NEW') an RMS relative
indexed file without specifying a record length.

Recovery:

Specify /RECORDSIZE.

ERRSNS values:
RMS?

First Value = 22

First Value = 30

Second Value = 613

RMS file requires ORGANIZATION in OPEN statement (TOPS-20 only)
Cause:

The OPEN statement did not specify an RMS file by means
of ORGANIZATION=, but the file you tried to open is an
RMS file.

Recovery: Supply an ORGANIZATION attribute.
ERRSNS values:
RNA?

First Value = 30

Second Value

608

RMS not available on SYS:
Cause:

Your program tried to invoke RMS (an OPEN specified a
remote or RMS file, or an INQUIRE-by-file specified a
remote file) when RMS is not available on your system.

Recovery: Make sure your system supports RMS V3 and/or RMSFAL VI.
ERRSNS values:

FORTRAN-10/20 VII

First Value = 30

D-35

Second Value = 615

February 1987

FOROTS ERROR MESSAGES
RNM?

Can't rename file
Cause:

An attempt to rename the specified file failed.

Recovery: Change file protection, remove other file access.
ERRSNS values:

First Value = 28

Second Value = 250+n
(TOPS-10)
JSYS error
number
(TOPS-20)

RNR?

Record <n> has not been written
Cause:

In direct-access input, an attempt was made to read a
record that was never written.
This may indicate the
use of an incorrect record number.

Recovery: Make sure you are requesting the correct record.
ERRSNS values:
RPE?

First Value = 25

Second Value = 510

Illegal repeat count
Cause:

In NAMELIST or list-directed input, a repeated constant
was found,
but the repeat count is not a positive
integer.

Recovery: Correct the input and try again.
ERRSNS values:
RRl?

First Value = 97

Second Value

521

Random I/b requires RECORDSIZE specifier in OPEN statement
Cause:

Direct-access I/O was attempted to a file that has not
been opened with the RECL or RECORDSIZE specifier to
give the size of the record(s).

Recovery: Specify a
record size.
The record size
is
in
characters for
formatted files, words for unformatted
files.
(See Section 11.3.33.)
ERRSNS values:
RRR?

First Value = 30

Second Value = 240

Random 10 requires /RECORDSIZE
Cause:

An OPEN statement was
attempted
that
specified
ACCESS=DIRECT(RANDOM) with no record size specified

Recovery: Specify a record size.
ERRSNS values:

First Value

= 30

D-36

Second Value

240

February 1987

FOROTS ERROR MESSAGES
RSM?

Record size different from that specified
Cause:

A record size found in a binary record
is
than that specified in the OPEN statement.

different

Recovery: Specify the correct record size in the OPEN statement.
ERRSNS values:
SDO?

First Value = 25

Second Value

= 572

Same device open on unit with conflicting specifiers
Cause:

An OPEN statement was attempted for a device for which
another OPEN or data transfer statement had been
already executed, and the file specifications were
in
conflict.

Recovery: Change file specifications
ERRSNS values:
SIF?

First Value = 30

Second Value

540

<statement> illegal for this file type or access (TOPS-20 only)
Cause:

Your program attempted to execute a statement that
is
legal only for
RMS files on a non-RMS file; or a
file-positioning, DELETE, UNLOCK or REWRITE statement
was attempted on an RMS file when either /ACCESS or
/ORGANIZATION does not permit the operation.

Recovery: Correct the invalid statement in the program.
ERRSNS values:
SLN?

First Value = 31

Second Value

= 614

Record length negative or zero
Cause:

An ENCODE or DECODE statement was initiated that has a
negative or zero value for the record (string) length.

Recovery: Correct program to specify a legal record length.
ERRSNS values:
SLT?
files

First Value = 25

Input/Output switching not allowed with Labeled
Cause:

Magnetic

tape

Your program attempted
to switch between input
output when using labeled magnetic tape files.

and

Recovery: Don't switch
input
magnetic tape files.
ERRSNS values:
SNH?

Second Value = 577

First Value

and

= 96

output

when

using

labeled

Second Value = 101

Internal FOROTS error at <PC>
Cause:

This error is not expected
to occur.
consistency check has found a bug.

internal

Recovery: Please submit an SPR if you get this message.

FORTRAN-10/20 VII

D-37

February 1987

FOROTS ERROR MESSAGES
SNQ?

String not within single quotes
Cause:

A character data item read as list-directed or NAMELIST
input is not enclosed in single quotes.

Recovery: Enclose character data item in single quotes.
ERRSNS values:
SNV?

First Value = 97

Second Value = 598

Sign with null value
Cause:

List-directed or NAMELIST input contains a + or not followed by a value.

sign

Recovery: Correct the input and try again.
ERRSNS values:

First Value = 97

Second Value

SRE % Subscript range error - subscript <n> of array
<n>
Cause:

522
<name>

line

An illegal subscript or range has been specified for an
array reference.

Recovery: Specify a legal array reference.
ERRSNS values:

First Value = 23

Second Value = 114

SSE % substring range error <var(bound»
Cause:

on line <n>

An illegal substring bound or range has been
in a character expression.

specified

Recovery: Specify a legal reference.
ERRSNS values:
TFM?

First Value = 23

Second Value

113

Tape format conflicts with OPEN statement or default
Cause:

The actual format of the tape (either CORE-DUMP or
INDUSTRY) conflicts with the format specified in the
OPEN statement or by the monitor. At this point, the
file is already opened in the wrong format.

Recovery: Specify the correct TAPEFORMAT in the OPEN statement or
with the EXEC.
ERRSNS values:
TKS?

First Value

Second Value = 569

Too many KEY specifiers (TOPS-20 only)
Cause:

RMS allows a maximum of
file.

255

indexes

for

indexed

Recovery: Reduce the number of KEY= specifiers.
ERRSNS values:

FORTRAN-10/20 VII

First Value = 30

0-38

Second Value = 607

February 1987

FOROTS ERROR MESSAGES
TMA?

Too many arguments in call to SORT
Cause:

When the first· argument
in a call
to SORT is a
character constant, the argument list must be copied in
order to convert the argument to a Hollerith constant.
At most, 10 arguments can be copied.

Recovery: Change the first argument to a Hollerith
use less than 10 arguments.
ERRSNS values:
TMF?

First Value

= 21

Second Value

constant,

115

Too many SFDs (TOPS-10 only)
Cause:

A directory specification contains more than five SFDs.

Recovery: Specify the correct directory.
ERRSNS values:
UDO?

First Value

= 45

Second Value

Unsupported /DISPOSE:<option> for
Cause:

546

RMS files (TOPS-20 only)

A PRINT, LIST, PUNCH, SUBMIT or PLOT was specified for
a TOPS-20 RMS file; or PUNCH or PLOT specified for a
VMS RMS file.

Recovery: Correct the specifier.
ERRSNS values:
UDT?

First Value = 28

Second Value

610

Undefined data type or internal FOROTS error
Cause:

Internal FOROTS error.

Recovery: Submit an SPR.
ERRSNS values:
UME?

First Value

Second Value

575

Unexpected MTCHR error <n> (TOPS-10 only)
Cause:

This message is not expected to occur.
It
indicates
that a MTCHR UUO has failed.
The monitor-supplied
error code may give some indication of the reason.

Recove~y:

Submit an SPR.

ERRSNS values:

First Value = 96

Second Value

= 531

UMO % <JSYS error> trying to set tape <density/parity/data mode>
(TOPS-20 onl y)
Cause:

It was not possible to set the indicated parameter of
the tape.
The monitor error message gives the reason.

Recovery: Make sure you are using
requested operations.
ERRSNS values:

FORTRAN-10/20 VII

First Value = 96

D-39

drive

that

Second Value

supports

the

JSYS error
number
February 1987

FOROTS ERROR MESSAGES
UNO?

DIVERT:
Cause:

unit <n> is not open
The file to which error messages are diverted
opened for output before DIVERT is called.

must

Recovery: Open the file for output before calling DIVERT.
ERRSNS values:
UNS?

First Value = 21

Second Value = 105

Unit not specified
Cause:

A call was made to FOROTS which did not contain a
number.

unit

Recovery: Correct calling code.
ERRSNS values:

First Value = 81

Second Value

501

UOA % Unknown OPEN keyword <n>, ignored
Cause:

The compiler has generated an OPEN call that
an unknown keyword.
The keyword is ignored.

Recovery: Make sure you are using the correct
compiler, FORLIB, and FOROTS.
ERRSNS values:
UOF?

First Value = 26

versions

Second Value

the

= 541

USEROPEN failed
Cause:

An RMS service call in a USEROPEN procedure has failed.

Recovery: Examine the STS and STV error values to
exact cause.
ERRSNS values:

URS?

contains

First Value

Second Value

determine

the

JSYS error
number

Unsupported remote system (TOPS-20 only)
Cause:

A remote file access specified a remote system that
neither TOPS-20 or VMS.

Recovery: Specify a TOPS-20 or VMS node.
ERRSNS values:
URT?
only)

First Value = 30

Recordtype <type> not supported for
Cause:

Second Value
this

file

602
type

(TOPS-20

/RECORDTYPE:STREAM was specified
for a TOPS-20 RMS
file;
or RMS returned a recordtype other than STREAM,
VARIABLE or FIXED for an existing remote file.

Recovery: Correct the invalid /RECORDTYPE.
ERRSNS Values:

FORTRAN-10/20 VII

First Value = 30

D-40

Second Value

601

February 1987

FOROTS ERROR MESSAGES
USW?

Unknown switch /<sw> (TOPS-10 only)
Cause:

In dialog mode, an unknown switch was specified.

Recovery: Retype the line, specifying the correct switch.
ERRSNS values:
UTE?

First Value = 45

Second Value = 241

Unexpected TAPOP error <n> (TOPS-10 only)
Cause:

This message is not expected to occur.
It indicates
that a TAPOP UUO has failed.
The monitor-supplied
error code may give some indication of the reason.

Recovery: Submit an SPR.
ERRSNS values:

First Value

Second Value = 530

UTO % Unexpected TAPOP.
error <n> trying to set
mode/blocksize> (TOPS-10 only)
Cause:

It is not possible to set the
indicated parameter of
the tape.
The monitor error message gives the reason.

Recovery: Make sure you are using
requested operation.
ERRSNS values:
VNN?

<density/parity/data

First Value = 96

drive

that

supports

the

Second Value = 537

Variable <var> not in namelist
Cause:

NAMELIST input contains an assignment
that is not in the namelist.

variable

Recovery: Correct the input and try again.
ERRSNS values:
WBA?

First Value = 97

Second Value

309

Attempt to WRITE beyond variable or array
Cause:

An attempt was made to write beyond the end of a
character variable or array with an internal file WRITE
statement.

Recovery: Correct program to
variable or array.
ERRSNS values:

First Value

stay

within

Second Value = 576

D-41

limits

character

February 1987

APPENDIX E
INTERACTIVE DEBUGGER (FORDDT) ERROR MESSAGES

FORDDT responds with two levels of messages - fatal error and warning.
Fatal error messages indicate that the processing of a given command
has been terminated. Warning messages provide helpful
information.
The format of these messages is:
? FDTxxx text
or
%FDTxxx text
where:
?
%
FDT
xxx
text

indicates a fatal message
indicates a warning message
is the FORDDT mnemonic
is the 3-letter mnemonic for error message
is the explanation of error

Square brackets ([ ]) in this section signify variables
output on the terminal.

and

are

not

?FDT

Fatal Errors
The fatal errors in the following list are each preceded
by
the user terminal.
They are listed in alphabetical order.
BDF

[symbol]
[symbol]

BOI

Bad octal input

is undefined
is multiply defined

An illegal character was detected in an octal input value.
CCN

Cannot continue
A pause has been placed on some
form of
causing FORDDT to loop;
should never
FORTRAN-compiled programs.

CFO

skip instruction
be encountered in

Core file overflow
The storage area for GROUP text has been exhausted.

CNU

The command [name]

is not unique

More letters of the command are required to
from the other commands.

E-l

distinguish

INTERACTIVE DEBUGGER (FORDDT) ERROR MESSAGES
CSH

Cannot START here
The specified entry point is not an acceptable FORTRAN
program entry point.

DNA

Double-precision comparisons not allowed

DTO

Dimension table overflow
FORDDT does not have the space to
dimensions until some are removed.

FCX

record

any

main

array

Format capacity exceeded
An attempt was made to specify a FORMAT statement requiring
more space than was originally allocated by the FORTRAN
compiler.

FNI

Formal not initialized, please retype
There was a reference to a
formal
subprogram that was never executed.

FNR

[array name]

parameter

some

is a formal and may not be redefined

Formal parameters may not be DIMENSIONed.
IAF

Illegal argument format [rest of user line]
The parameters to
the given command were not specified
properly.
Refer to the documentation for correct format.

IAT

Illegal argument type = [number]
An unrecognized subprogram argument
Submit an SPR if this message occurs.

ICC

type

was

detected.

Comparison of two constants is not allowed
A conditional test involves two constants.

IER

Internal FORDDT error [number]
Internal FOR DDT error -

IGN

please report through an SPR.

Invalid group number
Group numbers must be integers and in the range one
eight.

INV

through

Invalid value [rest of user line]
A syntax error was detected in the numeric parameter.

IPN

Illegal program name

IRS

Illegal range specification
The particular range specified for an array is not defined.

ITM

Illegal type modifier - S
The mode S is only valid for ACCEPT statements.

E-2

INTERACTIVE DEBUGGER (FORDDT) ERROR MESSAGES
IWI

I/O within I/O error
An attempted TYPE or ACCEPT command cannot be executed
because a
fatal
"I/O within I/O" error from FOROTS would
result, since the user program is currently processing an
IOLST call.

JSE

[JSYS error message]
Error reading program name (on TOPS-20 or some other
input) •

LGU

[array name]

lower subscript .GE.

FORDDT

upper

The lower bound of any given dimension must be less than
equal to the upper bound.
LNF

[label]

MCD

Compile program with the
statement

MLD

[array name] multi-level array definition not allowed

is not a FORMAT statement
DEBUG

switch

type

FORMAT

The same array cannot be dimensioned more than once
(by
means of the [dimensions] construct) in a single command.
MSN

More subscripts needed
The array is defined
to have more
specified in the given reference.

NAL

dimensions

than

were

Not allowed
An attempt has been made to modify something other than data
or a FORMAT.

NAR

Not after a reenter
The given command is not allowed until program integrity has
been restored by means of a CONTINUE or NEXT command.

NDT

DDT not loaded

NFS

Cannot find FORTRAN start address for

[program name]

Main program symbols are not loaded.
NFV

[symbol]

is not a FORTRAN variable

Names must be 6-character
with a letter.

alphanumeric

NGF

Cannot GOTO a FORMAT statement

NPH

Cannot insert a PAUSE here

strings

beginning

An attempt has been made to place a breakpoint at other than
an executable statement or subprogram entry point.
NSP

[symbol]

no such PAUSE

An attempt has been made to REMOVE
never set up.
E-3

breakpoint

that

was

INTERACTIVE DEBUGGER (FORDDT) ERROR MESSAGES
NUD

[symbol]

not a user-defined array

An attempt has been made to remove dimension information for
an array that was never defined.
PAR

Parentheses required, please retype
Parentheses are required for
the
statements and complex constants.

PRO

specification

FORMAT

Too many PAUSE requests
The PAUSE table has been exhausted.
10.

RGR

The

maximum

limit

Recursive group reference
A group may not reference itself.

SER

Subscript error
The subscript specified is outside the range of its
dimensions.

STL

[array name]

size too large

An attempt has been made to
256K.
TMS

define

array

larger

than

Too many subscripts [dimensions]
The array is defined
to have
fewer dimensions
specified in the given element reference.

URC

defined

than

are

Unrecognized command [command]

Warning Messages
Each warning message is preceded by %FDT on your
terminal.
warning messages are listed here in alphabetical order.
ABX

[array name]

The

compiled array bounds exceeded

FORDDT has detected another symbol defined in the specified
range of the array.
Note that this will occur in certain
EQUIVALENCE cases and can be ignored at that time.
CAB

Cannot allocate buffer for help file

CHI

Characters ignored:

"[text]"

The portion of the command string
included
thought to be extraneous and was ignored.
ECI

Buffer full -

excess characters ignored

EOH

Error opening help file

IOE

I/O error reading help file

E-4

"text"

was

INTERACTIVE DEBUGGER (FORDDT) ERROR MESSAGES
IWI

I/O within I/O
FORDDT has PAUSEd at a breakpoint while the user program is
currently
processing an
IOLST call.
TYPE and ACCEPT
commands cannot be processed at this breakpoint.

NAA

[ s ym b 0 I ]

i s no tan a r r ,a y

NHF

Cannot find help file; I'm sorry I can't help you

NSL

No symbols loaded
FORDDT cannot find the symbol table.

NST

Not STARTed
The specified command requires that a START be previously
issued to ensure that the program is properly initialized.

POV

Program overlayed by [program name]
Program overlayed by ***
The symbol table is different from the last time FORDDT had
control.
The program name
is printed only if it has
changed, otherwise '***' is printed.

SCA

Supersedes compiled array dimension
The FORTRAN generated dimension is being superseded for
given array.

SPO

Variable is single-precision only

WSP

Writing to shared page

XPA

Attempt to exceed program area with [symbol name]
An attempt has been made to access
currently defined program space.

E-5

memory

outside

the

APPENDIX F
FORTRAN-SUPPLIED PLOTTER SUBROUTINES

The FORTRAN subroutine library contains a set of subroutines that are
used with plotting devices.
To successfully use these routines, a
plotter must be connected to your system.
The FORTRAN software
contains the following plotter subroutines:
PLOTS

The PLOTS subroutine initializes the plotter or reports if
the plotter is not available.
This must be the first
plotter subroutine specified.
(See Section F.I.)

AXIS

The AXIS subroutine draws an axis with tic marks and scale
values at I-inch increments. An identifying label may also
be plotted along the axis.
(See Section F.2.)

CAXIS

The CAXIS subroutine performs the same functions as the AXIS
subroutine.
(See Section F.3.)

LINE

The LINE subroutine draws a
specified.
(See Section F.4.)

line

MKTBL

The MKTBL subroutine
(See Section F.5.)

NUMBER

The NUMBER subroutine causes floating-point
plotted as text.
(See Section F.6.)

PLOT

The PLOT subroutine moves the plotter pen to a new position.
Raising and lowering the pen is also specified in the PLOT
subroutine.
(See Section F.7.)

SCALE

The SCALE subroutine selects scale values for
CAXIS subroutine.
(See Section F.8.)

SETABL

The SETABL subroutine
Section F.9.)

SYMBOL

The SYMBOL subroutine raises the plotter pen, moves it to
the position .specified by x and y, and plots/a string of
characters.
(See Section F.ID.)

WHERE

The WHERE subroutine reports on the current position of
plotter pen.
(See Section F.II.)

specifies

F-I

special

the

points

character

set.

through

numbers

character

the

AXIS

set.

or
(See

the

FORTRAN-SUPPLIED PLOTTER SUBROUTINES

PLOTS
Subroutine

F.l

PLOTS SUBROUTINE

The PLOTS subroutine initializes the plotter or reports if the plotter
is
not available.
This must be the first
plotter subroutine
specified.
The form of the PLOTS subroutine is:
CALL PLOTS(i[,steps])
where:
is an integer variable, which is set to -1
if the
plotter is not available, or set to a if the plotter is
available.

The plotter may not be available because:
the system
does not have a
plotter;
the plotter is in use by
another user; the plotter is turned off; or the plotter
is being spooled, but you are trying to write to it
directly.
steps

is an optional floating-point variable or constant that
specifies the number of steps (per inch) used by the
plotter.
The default is 100 steps per inch, and may be
changed by the installation.
Commonly, plotters plot
100 steps per inch, 200 steps per inch, or 100 steps
per centimeter (about 254 steps per inch).

AXIS
Subroutine

F.2

AXIS SUBROUTINE

The AXIS subroutine draws an axis with tic marks and scale values at
I-inch increments.
An identifying label may also be plotted along the
axis.
NOTE
The AXIS subroutine is provided for compatibility with
previous
versions
of
FORTRAN-IO/20.
The AXIS
subroutine uses a numeric array to contain the label
that is plotted along the axis.
The CAXIS subroutine
(see Section F.3) allows a character expression to
contain the label.

F-2

FORTRAN-SUPPLIED PLOTTER SUBROUTINES
The form of the AXIS subroutine is:
CALL AXIS(x,y,asc,nasc,size,theta,xmin,dx)
where:
x,y

is a variable or constant
starting point of the axis.

pair

that

specifies

asc

is the name of a numeric array that
that is plotted along the axis.

nasc

is an integer constant or variable that specifies the
number of characters contained in array asc.
If nasc
is negative, the label in the array is placed on the
clockwise side of the axis.
If nasc is positive, the
tic marks, label, and scale values are placed on the
counterclockwise side of the axis.

size

is a constant or variable that specifies the length
the axis in inches.

theta

is a constant or variable that specifies the angle at
which the axis is plotted.
The value of theta is
usually 0.0 or 90.0.

xmin

is a variable or constant that specifies the
the scale at the beginning of the axis.

value

is a variable or constant that specifies the change
scale for a I-inch increment.

contains

the
label

NOTE
The proper values for xmin and dx may be
determined by calling the SCALE subroutine (see
Section F.8).

CAXIS
Subroutine

F.3

CAXIS SUBROUTINE

The CAXIS subroutine draws an axis with tic marks and scale values at
I-inch increments.
An identifying label may also be plotted along the
axis.
The form of the CAXIS subroutine is:
CALL CAXIS(x,y,asc,sign,size,theta,xmin,dx)

F-3

FORTRAN-SUPPLIED PLOTTER SUBROUTINES
where:
x,y

is a variable or constant
starting point of the axis.

pair

that

specifies

the

asc

is the name of a character expression that
label that is plotted along the axis.

sign

If nasc is negative, the label in the array is placed
on the clockwise side of the axis.
If nasc
is
positive, the tic marks, label, and scale values are
placed on the counterclockwise side of the axis.

size

is a constant or variable that specifies the length
the axis in inches.

theta

is a constant or variable that specifies the angle at
which the axis is plotted.
The value of theta is
usually 0.0 or 90.0.

xmin

is a variable or constant that specifies the
the scale at the beginning of the axis.

value

is a variable or constant that specifies the change
scale for a I-inch increment.

contains

NOTE
The proper values for xmin and dx may be
determined by calling the SCALE subroutine (see
Section F.8).

LINE
Subroutine

F.4

LINE SUBROUTINE

The LINE subroutine draws a continuous line through a set of points.
The form of the LINE subroutine is:
CALL LINE(x,y,n,k)
where:
x

is the name of an array that contains the
x-coordinates of the points to be plotted.

floating-point

is the name of an array that contains the
y-coordinates of the points to be plotted.

floating-point

F-4

FORTRAN-SUPPLIED PLOTTER SUBROUTINES
n

is an integer constant or variable that specifies the
number of points to be plotted.

total

is an integer constant or variable that equals the number of
elements of x and y.
Since single-precision one-dimensional
arrays are usually used, this value is usually 1.

MKTBL
Subroutine

F.S

MKTBL SUBROUTINE

The MKTBL subroutine defines a special character set to be used when
plotting;
the SETABL subroutine (see Section F.9) enables you to use
the character set defined by the MKTBL subroutine.
The form of the MKTBL subroutine is:
CALL MKTBL(setnumber,tableaddress)
where:

F.S.l

setnumber

is an integer variable from 1 to 10 that specifies
the numeric
identifier of the character set, for
example, the number of the ASCII character set is
1.
If the character set cannot be defined by a
call to MKTBL, a value of zero is returned in this
variable.

tableaddress

is a constant or variable that specifies the
starting address of a character table that has 128
(200 octal)
consecutive words.
Each character
table word contains the number of strokes (line
segments) for the character in the left half, and
the address of the table of strokes in the right
half.
See Section F.S.l for
a description of
these tables.

Character Tables

The next sections describe how to define and organize a character
table.
The character set, called by the MKTBL subroutine, enables
to create and use a character set other than the default character
that
is used to plot characters.
(Usually the default character
is ASCI I.)

set
you
set
set

To create your own character set, you need to create a character table
and a character stroke table.
These two tables are described in
Sections F.S.l.l and F.S.l.2, respectively.

F-S

FORTRAN-SUPPLIED PLOTTER SUBROUTINES
F.S.l.l Creating a Character Table - A character table contains 128
entries (200 octal).
Each entry in the character table, regardless of
whether the table is defined in the plotter subroutine library or by
you through a call to the MKTBL subroutine, indicates the character to
be plotted for the ASCII character that has that numerical value.
Figure F-l is a diagram of an entry word in a character table.

Strokes

Address

'----v-----I '----v-----I
18 bits

18 bits
MR-S-1752-81

Figure F-l:

Plotter Character Table Entry

As shown in Figure F-l, each entry in the character table contains the
number of strokes
(line segments) required to plot the character in
its left half and an address reference in the right half of the word.
The address in the right half of the entry references an entry in the
character stroke table for the character set.
The character stroke
table is described in the next section.

F.S.l.2 Creating a Stroke Table - Each character
in the character
table has a corresponding character stroke table.
The purpose of the
character stroke table is to define the number and
type of strokes
(drawn line segments) it takes to produce a character in the character
table.
Figure F-2 is a diagram of an entry in a character stroke table.

x
\

Three 5-bit bytes

Figure F-2:

Character Stroke Table Entry

As shown in Figure F-2, each stroke
is
stroke
table by three 5-bit bytes.
three bytes are:
Byte
1

MR-S-1753-81

described
in the character
The possible values for these

Value and Meaning

a
1

pen is raised (off the paper)
pen is lowered (on the paper)

x-coordinate ~alue at end of stroke

y-coordinate value at end of stroke

F-6

FORTRAN-SUPPLIED PLOTTER SUBROUTINES
When determining the height and width of each character to be plotted,
consider the following conventions:
1.

Characters are drawn within a grid that is IS units
eight units wide.

high

Characters are generally plotted six units above the base
line.
Two units are generally left blank to the right of
each character; one unit is generally left blank on top of
each character.
The spacing above, below, and on either side
of each character
provides
adequate
spacing
between
characters and between lines.

The plotter starts drawing each character at the lower left
corner of the character grid.
If the grid is set up in the
conventional manner, the lower left corner is grid position
(0,6).
Normal width characters end at the lower right corner
of the character grid after allowing
spacing
between
characters.
The last coordinate in each character grid is usually
(8,6).
The character grid for the next character has as its origin
(0,0) at the ending coordinate in the previous grid (8,6).

Accents, circumflexes, underscores, and other characters that
are to be plotted
in the same character grid as another
character should end at the same grid position as they began.
By
doing
this,
for
example, the ASCII character that
represents the accent character is plotted before
the
character that is the letter to be accented.

F.S.l.3 Sample Character Stroke Table - In this sample, you want to
plot the Greek letter beta as a normal sized character. When plotted,
the letter is drawn on a IS x 8 unit grid above the base line,
allowing for spacing between characters and between lines.
The strokes
(line segments)
for
beta,
including
the
invisible
segments,
are then determined.
The character will be plotted more
smoothly if as many of the line segments as possible are connected,
and
if doubled segments are avoided.
Note that stroke 9 (the return
stroke for the crossbar of the beta) and stroke IS (moving the pen to
ending point (8,6) are not visible; the pen is raised.

F-7

FORTRAN-SUPPLIED PLOTTER SUBROUTINES

15
4

Y
2

~
6
8,9

1V
V

lX
N°
11
1V 15

Base Line

o
Figure F-3:

MR-S-1754-81

Sample Character Stroke Table

If the address at the beginning of the table that
includes the
character in Figure F-3 is called GREEK, then the FORTRAN call to set
up the table could be:
IGREEK=4
CALL MKTBL (IGREEK, GREEK)
The above call to the MKTBL subroutine defines the GREEK character
table to be the fourth character table.
To use the GREEK table for
plotting, use the FORTRAN call:
CALL SETABL (IGREEK, IFLAG)
The entry in the character table for beta is:
RADIX
BETA:

DIO
iVALUES IN DECIMAL
IS"TBETA iCHARACTER TABLE ENTRY FOR BETA

Note that the character table entry for beta contains the starting
address of the character stroke table (TBETA) in its right half.
The character stroke table for beta is:

TBETA:

SEEN==l
iIF 'SEEN' THEN THE STROKE MARKS THE PAPER
UNSEEN==O
iIF 'UNSEEN' THEN THE STROKE IS INVISIBLE
RADIX ~DlO
iALL VALUES ARE IN DECIMAL
BYTE(S) SEEN,2,8,SEEN,2,13
BYTE(S) SEEN,3,14,SEEN,S,14
BYTE(S) SEEN,6,13,SEEN,6,12
BYTE(S) SEEN,S,11,SEEN,2,11
BYTE{S) UNSEEN,S,11,SEEN,6,10
BYTE(S) SEEN,6,9,SEEN,S,8
BYTE(S) SEEN,3,8,SEEN,2,9
BYTE(S) UNSEEN,8,6
F-8

FORTRAN-SUPPLIED PLOTTER SUBROUTINES
F.5.l.4
FORTRAN- and User-Defined Character Sets - The standard ASCII
character set is always defined and is character set 1 for calls to
the SETABL subroutine, unless character set 1 is redefined by a
user
call
to
the MKTBL subroutine.
If SETABL is not called, the ASCII
character set is the default.
The Cyrillic (Russian) character set is available as character set
2,
and the
Feanorian character set is available as character set 3
(unless character sets 2 or 3 has been redefined by a user call to the
MKTBL subroutine).
In order to use these character sets, the user
program must contain an
EXTERNAL statement for variable
PLOTF
(Feanorian) or PLOTC (Cyrillic).
User-defined character sets should use character sets 4 through 10
avoid conflicts with the standard character sets.

NUMBER
Subroutine

F.6

NUMBER SUBROUTINE

The NUMBER subroutine causes floating-point numbers to be
text.

plotted

The form of the NUMBER subroutine is:
CALL NUMBER(x,y,size,fnum,theta,ndigit)
where:
x,y

are variables or constants that specify the x and
y
coordinates of the point to be plotted.
The specified
point is the lower left corner of the
first
character
to be plotted.

size

is a variable or constant that specifies the size
(in
inches)
of the digits
to be plotted.
The specified
value should be a multiple of
.08
inches
(or
centimeters if
plotter is metric) if a small value is
used.

fnum

is a variable or constant that
number to be plotted.

theta

is a variable or constant that specifies the direction
(in degrees)
of the base line on which the characters
are plotted.

ndigit

is a variable or constant that specifies the number of
digits to be plotted to the right of the decimal point.
If ndigit is negative, only the
integer part of the
number is plotted; the resulting integer is rounded.

F-9

the

floating-point

FORTRAN-SUPPLIED PLOTTER SUBROUTINES

PLOT
Subroutine

F.7

PLOT SUBROUTINE

The PLOT subroutine moves the plotter pen to a new position.
Raising
and lowering the pen is also specified in the PLOT subroutine.
NOTE
The plotter is not released after
specified movement.

completion

the

The form of the PLOT subroutine is:
CALL PLOT(x,y,penup/down)
where:
x,y

are the variables or constants that specify the x
and
y coordinates for the point to which the pen
will be moved.

pen up/down

is an integer constant or variable that specifies
whether the pen
is on the paper or above the
paper.
The possible values for this variable are:

3
2
1

raise pen before movement
lower pen before movement
leave pen in current state (raised or lowered)

-1, -2, or -3

same as corresponding
positive
values except that on completion
of the indicated motion,
the new
pen position is taken as a new
origin and the output buffer
is
sent to the plotter~
Using the
negative values is helpful if you
are
plotting
consecutive
characters in the same program.

SCALE
Subroutine

F.8

SCALE SUBROUTINE

The SCALE subroutine scales values for the AXIS subroutine.
The form of the SCALE subroutine is:
CALL SCALE(x,n,s,xmin,dx)
F-IO

FORTRAN-SUPPLIED PLOTTER SUBROUTINES
where:

is an array name of a one-dimensional
floating-point
array to be scaled for the AXIS subroutine.

is an integer constant or variable that
length of the array in words (36-bit).

is a constant or variable that specifies the length (in
inches) of the desired axis.

xmin

is a constant or variable that specifies the smallest
element
in array x.
The value of xmin will be the
value of the scale at the beginning of the axis.

is a constant or variable that equals the change
in
scale for
a
I-inch interval
so that array x can be
plotted in inches.

specifies

the

SETABL
Subroutine

F.9

SETABL SUBROUTINE

The SETABL subroutine enables you to select the character set that
used to plot characters.

The form of the SETABL subroutine is:
CALL SETABL (setnum,status)
where:
setnum

is an integer constant or variable that equals the
number of the character set.
The standard ASCII
character set is defined
to be set 1 and
is the
default.
Character sets are defined by the MKTBL
subroutine (see Section F.5).

status

is an integer variable whose value after the call
to
the SETABL subroutine is either 0, if the character set
specified in setnum is valid, or -1, if the character
set specified by setnum is invalid.
NOTE
If you use a character set other than
the
character
sets defined by default
in the
plotter subroutine library, you must call
the
MKTBL subroutine before calling the SETABL
subroutine.

F-ll

FORTRAN-SUPPLIED PLOTTER SUBROUTINES

SYMBOL
Subroutine

F.IO

SYMBOL SUBROUTINE

The SYMBOL subroutine plots a specified string of characters
(from
either the default character set or the character set specified by the
last successful call to the SETABL subroutine) •
The form of the SYMBOL subroutine is:
CALL SYMBOL(x,y,size,asc,theta[,nasc)
where:
x

is a constant or variable that equals the x coordinate
of the lower left corner of the first character to be
plotted.

is a constant or variable that equals the y coordinate
of the lower left corner of the first character to be
plotted.
The plotter pen
is raised and moved
to
position
x,y before the string of characters is
plotted.

size

is a constant or variable that specifies the height (in
inches)
of the character to be plotted.
The specified
value should
be a multiple of
.08
inches
(or
centimeters if you have ~metric plotter).

asc

is the name of a character expression or numeric array
that contains the ASCII characters to be plotted.

theta

is a constant or variable that specifies the direction
(in degrees)
of the base line on which the characters
are to be plotted.

nasc

is an integer constant or variable that is equal to the
number of characters in numeric array asc that are to
be plotted.
This is ignored if a character expression
is specified for asc.

WHERE
Subroutine

F.II

WHERE SUBROUTINE

The WHERE subroutine reports on the current position
pen, in inches, relative to the origin.

F-12

the

plotter

FORTRAN-SUPPLIED PLOTTER SUBROUTINES
NOTE
The plotter origin is set by a call to the AXIS
subroutine or a call to the PLOT subroutine that has a
negative value for the penup/down variable. Also, the
WHERE subroutine does not allow you to determine
whether the plotter pen is raised or lowered.
The form of the WHERE subroutine is:
CALL WHERE(x,y)
where:
x

is a variable in which the subroutine
coordinate of the current print position.

returns

the

is a variable in which the subroutine
coordinate of the current pen position.

returns

the

F-13

IND.EX

A editing, 12-39
ABORT switch, 16-9
ACCEPT
sequential list-directed, 10-46
ACCEPT statement, 10-45
sequential FORMAT-statement,
10-45
Access
direct, 18-4
sequential, 18-3
ACCESS specifier
in INQUIRE, 11-54
in OPEN, 11-7
Accumulator usage, 15-11
Actual arguments, 13-49
length of character, 13-51
Adjustable dimensions, 7-3
ALCCHR subroutine, 13-24
ALL with DEBUG, 16-12
ALL with FLAG, 16-22
ALL with NOWARN, 16-15
.AND. logical operator, 5-8
ANSI standard, 1-1
ANSI with FLAG, 16-22
Apostrophe editing, 12-16
Argument lists, 15-12
Argument types, 15-14
Arguments
actual, 13-49
dummy, 13-49
Arguments to FLAG, 16-22
ARGUMENTS with DEBUG, 16-12
Arithmetic assignment statement,
8-1
Arithmetic constant expressions,
5-5
Arithmetic expressions, 5-1
writing, 5-2
Arithmetic IF statement, 9-3
Arithmetic operations, 5-2
Arithmetic operators, 5-2
Array declarator, 4-4, 7-1
Array element subscript, 4-3
Array elements
order of stored, 4-5
Array names, 4-3
Arrays, 4-3
assumed-size, 7-4
dimensioning, 4-4
dimensioning dummy, 15-5
ASCII character code, B-1
ASCII data files, 18-4
ASSIGN statement, 8-3
Assigned GO TO statement, 9-3
Assigned variables
in FORMAT statements, 12-4
Assignment statement, 8-1
arithmetic, 8-1
character, 8-4

Assignment statement (Cont.)
logical, 8-3
Assignments
mixed-mode, 8-2
ASSOCIATEVARIABLE specifier
in OPEN, 11-8
Assumed-size arrays, 7-4
AXIS subroutine, F-2
BACKFILE statement, 11-67
BACKSPACE statement, 11-64
Binary data files, 18-4
BINARY switch, 16-9
Bit data constants, 7-23
Bit manipulation functions, 13-14
Blank lines, 2-7
BLANK specifier
in INQUIRE, 11-54
in OPEN, 11-9
BLISS-36
calling routines, 15-19
interaction with, 15-18
BLKRW subroutine, 19-4
BLOCK DATA statement, 14-1
Block data subprograms, 14-1
Block IF constructs
indexed, 9-9
Block IF statements, 9-5
BLOCKSIZE specifier
in OPEN, 11-10
BN editing, 12-28
Boolean values, 3-6
BOUNDS with DEBUG, 16-12
BUFFERCOUNT specifier
in OPEN, 11-11
BYTESIZE specifier
in INQUIRE, 11-54
in OPEN, 11-12
BZ editing, 12-28
CALL statement, 13-44
Calling BLISS-36 routines, 15-19
Calling COBOL subroutines, 15-18
Calling FORDDT, 17-17
Calling sequences
FOROTS, 18-16.1
Carriage-control specifiers, 12-9
CARRIAGECONTROL specifier
in INQUIRE, 11-54
in OPEN, 11-12
CAXIS subroutine, F-3
CBC function, 18-34
CDABS function, 13-25
CDCOS subroutine, 13-25
CDEXP subroutine, 13-26
CDLOG subroutine, 13-26
CDSIN subroutine, 13-27
CDSQRT subroutine, 13-28

Index-I

February 1987

Channel allocation routines,
18-30
Channel deal location routines,
18-30
CHAR function, 13-13
Character assignment statement,
8-4
Character code
ASCII, B-1
Character comparison functions,
13-14
Character constant expressions,
5-7
Character constants, 3-4
Character expressions, 5-6
Character intrinsic functions,
13-12
Character set, 2-1
FORTRAN-defined, F-9
user-defined, F-9
Character substrings, 4-6
Character tables, F-5
Character type specification
statements, 7-6
CHARACTER*len, 7-6
CHKDIV subroutine, 13-28
CLOSE statement, 11-43
exampl es, 11-51
implicit, 11-44
CLOSE statement specifiers, 11-44
DEVICE, 11-46
DIALOG, 11-46
DIALOG=, 11-47
DIRECTORY, 11-46
DISPOSE, 11-47
ERR, 11-49
FILE, 11-46
IOSTAT, 11-49
LIMIT, 11-49
NAME, 11-46
PROTECTION, 11-46
STATUS, 11-50
summary of, 11-45
UNIT, 11-51
CLRFMT subroutine, 13-29
CLROVL in LINK, 15-20
COBOL
calling subroutines, 15-18
interaction with, 15-16
CODE with EXTEND, 16-20
Colon editing, 12-24
Commands
COMPILE, 16-1, 16-5
DEBUG, 16-1, 16-5
EXECUTE, 16-1, 16-5
FORDDT, 17-1, 17-7
LOAD, 16-1, 16-5
Comment lines, 2-5
Common block, 7-9
COMMON statement, 7-8
Common storage, 7-9
COMMON with EXTEND, 16-20

Compilation control statements,
6-4
COMPILE command, 16-1, 16-5
Compiler
Running FORTRAN-10, 16-2
Running FORTRAN-20, 16-6
using, 16-1
Compiler errors
internal, C-13
Compiler generated label, 16-23
Compiler generated variables,
16-24
Compiler listing, 16-22
Compiler messages, C-1
Compiler switches
FORTRAN-10, 16-2
FORTRAN-20, 16~9
Compiling programs, 16-1, 16-5
Complex constants, 3-4
Complex editing, 12-36
Computed GO TO statements, 9-2
Concatenation operator, 5-6
CONNECT subroutine, 19-4
Constant expressions, 5-15
arithmetic, 5-5
character, 5-7
int~ger, 5-5
logical, 5-10
Constant folding, 15-8
Constant propagation, 15-8
Constants, 3-1
bit data, 7-23
character, 3-4
compl ex, 3-4
doub1e-oetal, 3-5
double-precision, 3-2
hollerith, 3-6
integer, 3-1
logical, 3-6
octal, 3-5
real, 3-2
statement label, 3-7
Continuation lines, 2-4
CONTINUE LSCW, 18-4
CONTINUE statement, 9-17
Control information list, 10-9
Control statements, 9-1
COR function, 18-33
Creating indexed files, 15-24
CROSSREF switch, 16-3, 16-9
Current-record DELETE, 10-60
Current-record pointer, 15-29
D editing, 12-32
D-floating double-precision
format, 3-3
Data access, 10-5
direct, 10-6
keyed, 10-6
sequential, 10-6
Data magnitude on G-format, 12-35
DATA statement, 7-21
Data transfer operations, 10-5

Index-2

February 1987

Data transfer statements, 10-1
summary, 10-3
Data transfers
formatted, 10-6, 12-1
unformatted, 10-8
DATA with EXTEND, 16-20
DATE subroutine, 13-29
DEBUG
and FORDDT, 17-4
DEBUG arguments, 16-12
DEBUG command, 16-1, 16-5
Debug lines, 2-6
DEBUG switch, 16-3, 16-9, 16-11
Debugger
FORDDT, 17-1
Debugging programs, 16-1, 16-5
DECODE statement, 10-56
Default memory layout
/EXTEND, 16-16
DEFAULTFILE specifier
in OPEN, 11-14
Deferred OPEN statement, 11-2
DELETE statement, 10-59
current record, ~0-60
direct, 10-60
Deleting records
in indexed files, 15-29
DENSITY specifier
in OPEN, 11-15
Device assignments
logical, 10-12
Device control statements, 10-1,
11-61
DEVICE specifier
in CLOSE, 11-46
in OPEN, 11-15
Devices
rea1-t ime, 19-1
DFLOATING switch, 16-3, 16-9
DIALOG specifier
in CLOSE, 11-46
in OPEN, 11-16
DIALOG= specifier
in CLOSE, 11-47
in OPEN, 11-16
Dimension declarator, 7-2
DIMENSION statement, 7-1
Dimensioning arrays, 4-4
Dimensioning dummy arrays, 15-5
DIMENSIONS with DEBUG, 16-12
Direct access, 10-6, 18-4
Direct DELETE, 10-60
Direct FORMAT-statement READ,
10-30
Direct FORMAT-statement WRITE,
10-38
DIRECT specifier
in INQUIRE, 11-55
Direct unformatted READ, 10-33
Direct unformatted WRITE, 10-41
DIRECTORY specifier
in CLOSE, 11-46
in OPEN, 11-17, 11-18

DISCON subroutine, 19-7
DISMIS subroutine, 19-6
DISPOSE specifier
in CLOSE, 11-47
in OPEN, 11-19
DIVERT subroutine, 13-30
DO iteration control, 9-12
DO lists
implied, 10-25
DO loop extended range, 9-15
DO loop iterations, 15-2.1
DO loops
floating-point, 15-2
DO statements
nested, 9-14
DO WHILE statement, 9-13
Dollar sign editing, 12-22
Double-octal constants, 3-5
Double-precision constants, 3-2
Double-precision format
D-f1oating, 3-3
G-floating, 3-3
Double-precision numbers
range and accuracy, 15-1
DTOG function, 13-19
DTOGA subroutine, 13-30
Dummy arguments, 13-49
length of character, 13-51
DUMP subroutine, 13-30
E (exponential notation), 3-2
E editing, 12-32
ECHO-OPTION switch, 16-9
Edit descriptors
FORMAT-statement, 12-6
nonrepeatab1e, 12-8
numer ic, 12-29
repeatable, 12-7
Ed i ti ng
A, 12-39
apostrophe, 12-16
BN, 12-28
BZ, 12-28
colon, 12-24
complex, 12-36
D, 12-32
dollar sign, 12-22
E, 12-32
F, 12-32
G, 12-33
H, 12-17
I, 12-31
L, 12-38
numer ic, 12-29
o (octal), 12-36
P, 12-25
positional, 12-18
Q, 12-29
R, 12-40
S, 12-24
slash, 12-23
SP, 12-24
SS, 12-24

Index-3

February 1987

Editing (Cont.)
T, 12-19
TL, 12-20
TR, 12-20
X, 12-21
Z (hexadecimal), 12-37
ELSE IF THEN statement, 9-5
ELSE statement, 9-5
Embedded format specifications,
12-15
ENCODE statement, 10-56
END DO statement, 9-17
END IF statement, 9-5
END LSCW, 18-4
END statement, 9-20
END= specifier, 10-18
ENDFILE statement, 11-65
Entry points
FOROTS, 18-15
ENTRY statement, 13-45
EQUIVALENCE and extended
addressing, 7-17
EQUIVALENCE statement, 7-11
.EQV. logical operator, 5-8
ERR specifier
in CLOSE, 11-49
in INQUIRE, 11-55
in OPEN, 11-20
ERR= specifier, 10-19
Error codes
FOROTS, D-3
Error messages
FORDDT, E-l
FOROTS, D-l
real-time, 19-7
Error processing
FOROTS, 18-2
Error reporting, 16-33
Errors
fatal, C-2
fatal compiler, 16-34
ERRSET subroutine, 13-31
ERRSNS subroutine, 13-32
ERRSNS values, D-2
Evaluation of expressions, 5-13
Executable statements, 1-1, 6-1
EXECUTE command, 16-1, 16-5
Executing programs, 16-1, 16-5
EXIST specifier
in INQUIRE, II-55,
EXIT subroutine, 13-33
EXPAND switch, 16-3, 16-9
Expressions
arithmetic, 5-1
arithmetic constant, 5-5
character, 5-6
character constant, 5-7
constant, 5-15
evaluation of, 5-13
integer constant, 5-5
log ical, 5-7
logical constant, 5-10
mixed-mode, 5-15

Expressions (Cont.)
relational, 5-11
EXTEND
applications with large arrays,
16-17
applications with large code,
16-17
EXTEND arguments, 16-19, 16-20
EXTEND psect placement, 16-18
EXTEND switch, 16-9, 16-16
Extended addressing, 16-16
and EQUIVALENCE, 7-17
and FORDDT, 17-17
default memory layout, 16-16
memory layout, 15-22
Extended attribute block
fields, 18-43
Extended attribute block (XAB) ,
18-40
Extended range DO loop, 9-15
External functions, 13-18
FORTRAN-supplied, 13-18
user-defined, 13-20
EXTERNAL statement, 7-17
F editing, 12-32
F66 switch, 16-3, 16-9
F77 switch, 16-3, 16-9
FAB block, 18-40
fields, 18-40
Fatal compiler errors, 16-34
Fatal errors, C-2
FFUNIT subroutine, 13-33
Field
line continuation, 2-3
remark, 2-4
statement, 2-4
statement label, 2-3
File access block
fields, 18-40
File access block (FAB), 18-40
File control statements, 10-1,
11-1
File identifier
internal, 10-13
File specifications
remote, 11-21
FILE specifier
in CLOSE, 11-46
in OPEN, 11-20
Files
ASCII data, 18-4
binary data, 18-4
internal, 10-8, 10-54
linking TWOSEG REL, 16-21
non-FORTRAN, 15-10
FILESIZE specifier
in OPEN, 11-21
FIND statement, 11-62
FLAG arguments, 16-22
FLAG switch, 16-3, 16-21
FLAG switch arguments, 16-22
FLAG-NON-STANDARD switch, 16-9

Index-4

February 1987

Floating-point
DO loops, 15-2
FORDDT
and extended addressing, 17-17
call i ng, 1 7 -1 7
loading, 17-5
starting, 17-5
FORDDT and DEBUG, 17-4
FORDDT commands~ 17-1, 17-7
FORDDT debugger, 17-1
FORDDT error messages, E-1
FORM specifier
in INQUIRE, 11-56
in OPEN, 11-22
Format control, 12-11
Format editing, 12-16
Format list, 12-3
I/O list interaction, 12-11
order and interpretation, 12-5
Format specifications, 12-3
embedded, 12-15
Format specifier, 10-14
FORMAT statement, 12-2
assigned variables in, 12-4
FORMAT-statement ACCEPT
sequential, 10-45
FORMAT-statement edit descriptors,
12-6
FORMAT-statement formatting,
10-15, 12-2
FORMAT-statement PRINT
sequential, 10-50
FORMAT-statement PUNCH
sequential, 10-53
FORMAT-statement READ
direct, 10-30
sequential, 10-29
FORMAT-statement REREAD
sequential, 10-43
FORMAT-statement TYPE
sequential, 10-48
FORMAT-statement WRITE
direct, 10-38
sequential, 10-37
Formatted data transfers, 10-6,
12-1
Formatted indexed READ, 10-34
Formatted input, 12-14
Formatted output, 12-15
Formatted READ transfers, 10-29
Formatted REWRITE, 10-61
FORMATTED specifier
in INQUIRE, 11-56
Formatted WRITE transfers, 10-37
Formatting
FORMAT-statement, 10-15, 12-2
list-directed, 10-16, 12-41
NAMELIST-statement, 10-17,
12-44
FOROTS, 18-1
and INQUIRE by file, 18-36
MACRO calls for, 18-17
RESET in, 15-11

FOROTS (Cont.)
sharable, 15-10
using, 18-15
FOROTS calling sequences, 18-16.1
FOROTS entry points, 18-15
FOROTS error codes, D-3
FOROTS error messages, D-l
FOROTS error processing, 18-2
FOROTS features, 18-1
FOROTS in~ut/output facility,
18-2.1
FOROTS memory management, 15-20
FOROTS/RMS
data and control structures,
18-40
FORRTF subroutine, 19-1
FORTRAN-10
compiler switches, 16-2
running the compiler, 16-2
FORTRAN-10/20, 1-1
FORTRAN-20
compiler switches, 16-9
running the compiler, 16-6
FORTRAN-20 compiler
using, 16-5
FORTRAN-77 standard, 1-1
FORTRAN-defined character set,
F-9
FORTRAN-supplied external
functions, 13-18
FORTRAN-supplied plotter
subroutines, F-1
FORTRAN-supplied subroutines,
13-24
Function effects, 15-10
FUNCTION statement, 13-20
Function subprogram, 13-20
restrictions, 13-21
Function subprograms
using, 13-21
Functions, 13-1'
bit manipulation, 13-14
CBC, 18-34
CDABS, 13-25
CHAR, 13-13
character comparison, 13-14
character intrinsic, 13-12
COR, 18-33
DTOG, 13-19
external, 13-18
FORTRAN-supplied external,
13-18
GAD, 18-32
GCH, 18-33
generic, 13-2
GOT, 18-33
GPG, 18-34
GPSI, 18-35
GTOD, 13-19
ICHAR, 13-13
IFS, 18-34
ILL, 18-32
INDEX, 13-12

Index-5

February 1987

Functions (Cont.)
intrinsic, 13-1
LEN, 13-12
LSNGET, 13-19
MPG, 18-35
RAD, 18-33
RAN, 13-19
RANS, 13-19
RCH, 18-33
RNT, 18-34
ROT, 18-34
RPG, 18-35
RPSI, 18-35
RRS, 18-34
SECNDS, 13-19
specific, 13-2
statement, 13-15
TIM2GO, 13-19
UPG, 18-36
user-defined external, 13-20
using intrinsic, 13-2
using statement, 13-16
WRS, 18-34
Functions for overlays, 18-31
G editing, 12-33
G-f10ating double-precision
format, 3-3
G-format
data magnitude on, 12-35
GAD function, 18-32
GCH function, 18-33
Generic functions, 13-2
GETOVL in LINK, 15-20
GFLOATING switch, 16-3, 16-9
Global optimization, 15-6
Global register allocation, 15-9
GO TO statement
ass ig ned, 9-3
GO TO statements, 9-1
computed, 9-2
unconditional, 9-2
GOT function, 18-33
GPG function, 18-34
GPSI function, 18-35
GTOD function, 13-19
GTODA subroutine, 13-34
H editing, 12-17
Hierarchy of operators, 5-14
High segment
sharable, 16-35
Hollerith constants, 3-6
I editing, 12-31
I/O
optimization, 15-9
I/O list, 10-23
I/O list and format list
interaction, 12-11
ICHAR function, 13-13
IF statement, 9-3
arithmetic, 9-3

IF statement (Cont.)
block, 9-5
logical, 9-4
logical two-branch, 9-4
IF THEN statement, 9-5
IFS function, 18-34
ILL function, 18-32
ILL subroutine, 13-34
Implicit CLOSE, 11-44
Implicit OPEN, 11-3
IMPLICIT statements, 7-7
Implied DO lists, 10-25
INCLUDE statement, 6-4
INCLUDE switch, 16-3, 16-9
INDEX function, 13-12
Index variable, 9-11
INDEX with DEBUG, 16-12
Indexed f i 1 es
creating, 15-24
defining keys, 11-24
deleting records in, 15-29
reading, 15-28
updating records in, 15-29
us i ng, 15 - 24
writing, 15-26
Indexed .HEAD, 10-33
formatted, 10-34
sequential, 10-34
unformatted, 10-35
Indexed sequential access method,
15-24
.
INIOVL in LINK, 15-20
Initial lines, 2-4
Initial tab, 2-3
INITIALIZE specifier
in OPEN, 11-21
Input
formatted, 12-14
NAMELIST-controlled, 12-45
Input/output facility
FOROTS, 18-2.1
Input/output statements, 10-1
INQUIRE by file and FOROTS, 18-36
INQUIRE specifiers
ACCESS, 11-54
BLANK, 11-54
BYTESIZE, 11-54
CARRIAGECONTROL, 11-54
DIRECT, 11-55
ERR, 11-55
EXIST, 11-55
FORM, 11-56
FORMATTED, 11-56
IOSTAT, 11-56
KEYED, 11-56
NAME, 11-57
NAMED, 11-58
NEXTREC, 11-58
NUMBER, 11-58
OPENED, 11-58
ORGANIZATION, 11-59
RECL, 11-59
RECORDSIZE, 11-59

Index-6

February 1987

INQUIRE specifiers (Cont.)
RECORDTYPE, 11-59
SEQUENTIAL, 11-60
UNFORMATTED, 11-60
INQUIRE statement, 11-52
by file, 11-52
by unit, 11-53
INQUIRE statement specifiers,
11-54
Integer constant expressions, 5-5
Integer constants, 3-1
Interaction with BLISS-36, 15-18
Interaction with COBOL, 15-16
Internal compiler errors, C-13
Internal file identifier, 10-13
Internal files, 10-8, 10-54
Internal READ
statement, 10-55
Internal WRITE statement, 10-55
Intrinsic functions, 13-1
character, 13-12
using, 13-2
INTRINSIC statement, 7-18
Invoking RMS, 11-29
IOSTAT specifier, 10-20
in CLOSE, 11-49
in INQUIRE, 11-56
in OPEN, 11-23
Iteration control
DO, 9-12
Key attributes, 10-23
KEY specifier
in OPEN, 11-24
Key-fie1d-va1ue specifier, 10-21
Key-of-reference specifier, 10-23
Keyed access, 10-6
KEYED specifier
in INQUIRE, 11-56
Keywords, 1-1
L editing, 12-38
Labels
compiler generated, 16-23
LABELS with DEBUG, 16-12
Large arrays
and extended addressing, 16-17
Large code applications, 16-19
Large executable code
and extended addressing, 16-17
LEGAL subroutine, 13-34
LEN function, 13-12
Length of character actual
arguments, 13-51
Length of character dummy
arguments, 13-51
Length specification, 7-6
LIMIT specifier
in CLOSE, 11-49
in OPEN, 11-25
Line continuation field, 2-3
LINE subroutine, F-4
Line types, 2-4

Line-sequence number, 16-22
Line-sequenced source files, 2-7
Lines
blank, 2-7
comment, 2-5
continuation, 2-4
debug, 2-6
initial, 2-4
multi-statement, 2-5
LINK overlay facilities, 15-19
LINK overlay handler, 18-31
LINK subroutines, 15-20
Linking TWOSEG REL files, 16-21
List-directed ACCEPT
sequential, 10-46
List-directed formatting, 10-16,
12-41
List-directed PRINT
sequential, 10-51
List-directed PUNCH
sequential, 10-53
List-directed READ
sequential, 10-30
List-directed REREAD
sequential, 10-44
List-directed TYPE
sequential, 10-49
List-directed WRITE
sequential, 10-39
LISTING switch, 16-9
LNMAP switch, 16-3, 16-10
LOAD command, 16-1, 16-5
Loading FORDDT, 17-5
Loading programs, 16-1, 16-5
LOCK subroutine, 19-3
Logical assignment statements,
8-3
Logical constant expressions,
5-10
Logical constants, 3-6
Logical device assignments, 10-12
Logical expressions, 5-7
Logical IF statements, 9-4
Logical operators, 5-8
Logical segment control words,
18-4
Logical two-branch IF statement,
9-4
Logical unit identifier, 10-11
LOGOVL in LINK, 15-20
LSCW, 18-4
LSNGET function, 13-19
MACHINE-CODE switch, 16-10
MACRO calls for FOROTS, 18-17
MACROCODE switch, 16-3
MAXREC specifier
in OPEN, 11-25
Memory allocation routines, 18-29
Memory layout
extended addressing, 15-22
Memory management
FOROTS, 15-20

Index-7

February 1987

Memory management (Cont.)
overlay, 15-20
Message summary, 16-34
Messages
compiler, C-1
warning, 16-34, C-8
Mixed-mode assignments, 8-2
Mixed-mode expressions, 5-15
Mixed-mode operations, 5-3
MKTBL subroutine, F-5
MODE specifier
in OPEN, 11-26
MPG function, 18-35
Multi-statement lines, 2-5
MVBITS subroutine, 13-35

.NOT. logical operator, 5-8
NOVMS with FLAG, 16-22
NOWARN arguments, 16-15
NOWARN switch, 16-3, 16-10, 16-15
NUMBER specifier
in INQUIRE, 11-58
NUMBER subroutine, F-9
Numeric edit descriptors, 12-29
Numeric editing, 12-29
Numeric type specification
statements, 7-5

NAME specifier
in CLOSE, 11-46
in INQUIRE, 11-57
in OPEN, 11-28
NAMED specifier
in INQUIRE, 11-58
NAMELIST statement, 12-44
NAMELIST-contro11ed input, 12-45
NAMELIST-contro11ed output, 12-46
NAMELIST-statement formatting,
10-17, 12-44
NAMELIST-statement READ
sequential, 10-31
NAMELIST-statement WRITE
sequential, 10-40
Names
array, 4-3
common block, 7-9
symbol ic, 4-1
variable, 4-2
.NEQV. logical operator, 5-8
Nested Block IF constructs, 9-9
Nested DO statements, 9-14
Next record pointer, 15-29
NEXTREC specifier
in INQUIRE, 11-58
NOANSI with FLAG, 16-22
NOBINARY switch, 16-10
NOCOMMON with EXTEND, 16-20
NODATA with EXTEND, lE-20
NOERRORS switch, 16-3, 16-10
NOEXTEND switch, 16-10
NOF77 switch, 16-3, 16-10
NOFLAG switch, 16-3
NOFLAG-NON-STANDARD switch, 16-10
Non-FORTRAN files, 15-10
Non-FORTRAN programs, 15-10
NONE wi th
DEBUG, 16-12
NONE with FLAG, 16-22
NONE with NOWARN, 16-15
Nonexecutable statements, 1-1,
6-2
Nonrepeatable edit descriptors,
12-8
NOSPANBLOCKS specifier
in OPEN, 11-28

o (octal) editing, 12-36
Object time system, 18-1
Octal constants, 3-5
OPEN
TAPEFORMAT specifier in, 11-41
OPEN statement, 11-1
deferred, 11-2
examples, 11-51
implicit, 11-3
on connected unit, 11-4
OPEN statement specifiers, 11-4
ACCESS, 11-7
ASSOCIATEVARIABLE, 11-8
BLANK, 11-9
BLOCKSIZE, 11-10
BUFFERCOUNT, 11-11
BYTESIZE, 11-12
CARRIAGECONTROL, 11-12
DEFAULTFILE, 11-14
DENSITY, 11-15
DEVICE, 11-15
DI ALOG, 11-16
DIALOG=, 11-16
DIRECTORY, 11-17, 11-18
DISPOSE, 11-19
ERR, 11-20
FILE, 11-20
F I L E S I Z E, 11- 21
FORM, 11-22
INITIALIZE, 11-21
IOSTAT, 11-23
KEY, 11-24
LIMIT, 11-25
MAXREC, 11-25
MODE, 11-26
NAME, 11-28
NOSPANBLOCKS, 11-28
ORGANIZATION, 11-29
PADCHAR, 11-29
PARITY, 11-30
PROTECTION, 11-30, 11-32
READONLY, 11-34
RECL, 11-34
RECORDSIZE, 11-34
RECORDTYPE, 11-36
SHARED, 11-38
STATUS, 11-39
summary of, 11-6
TYPE, 11-39
UNIT, 11-42
USEROPEN, 11-42

Index-8

February 1987

OPEN statement specifiers (Cont.)
VERSION, 11-43
OPENED specifier
in INQUIRE, 11-58
Operators
arithmetic, 5-2
concatenation, 5-6
hierarchy of, 5-14
logical, 5-8
relational, 5-11
Optimization
global, 15-6
I/O, 15-9
Optimization techniques, 15-6
OPTIMIZE switch, 16-3, 16-10,
17-17
OPTION switch, 16-10
.OR. logical operator, 5-8
Order of stored array elements,
4-5
Ordering of statements, 6-2
ORGANIZATION specifier
in INQUIRE, 11-59
in OPEN, 11-29
Output
formatted, 12-15
NAMELIST-controlled, 12-46
OVERFL subroutine, 13-35
Overlay facilities
LINK, 15-19
Overlay handler
LINK, 18-31
Overlay memory management, 15-20
Overlays
functions for, 18-31
P editing, 12-25
PADCHAR specifier
in OPEN, 11-29
PARAMETER statement, 7-20
Parenthetical subexpressions,
5-13
PARITY specifier
in OPEN, 11-30
PAUSE statement, 9-19
PDUMP subroutine, 13-36
PLOT subroutine, F-10
PLOTS subroutine, F-2
positional editing, 12-18
PRINT
sequential FORMAT-~tatement,
10-50
sequential list-directed, 10-51
PRINT statement, 10-50
priority interrupt levels, 19-2
Priority interrupt system, 19-1
PROGRAM statement, 6-4
Programs
compiling, 16-1, 16-5
debugging, 16-1, 16-5
executing, 16-1, 16-5
loading, 16-1, 16-5
non-FORTRAN, 15-10

Programs (Cont.)
source, 1-1
wr it i ng, 15-1
PROTECTION specifier
in CLOSE, 11-46
in OPEN, 11-30, 11-32
Psect placement
/EXTEND, 16-18
PSECT with EXTEND, 16-20
PUNCH
sequential FORMAT-statement,
10-53
sequential list-directed, 10-53
PUNCH statement, 10-52
Qed i t i ng, 12 - 2 9
QUIETX Subroutine, 13-36
R editing, 12-40
RAB block, 18-40
fields, 18-42
RAD function, 18-33
RAN function, 13-19
RANS function, 13-19
RCH function, 18-33
READ
direct FORMAT-statement, 10-30
direct unformatted, 10-33
formatted indexed, 10-34
indexed, 10-33
indexed sequential, 10-34
sequential FORMAT-statement,
10-29
sequential list-dJrected, 10-30
sequential NAMELIST-statement,
10-31
sequential unformatted, 10-32
unformatted indexed, 10-35
READ state~ent, 10-27
internal, 10-55
READ transfers
formatted, 10-29
unformatted, 10-32
Reading indexed files, 15-28
READONLY specifier
in OPEN, 11-34
Real constants, 3-2
Real-time devices, 19-1
Real-time error messages, 19-7
Real-time software, 19-1
RECL specifier
in INQUIRE, 11-59
in OPEN, 11-34
Record access block
fields, 18-42
Record access block (RAB) , 18-40
Record specifier, 10-13
Records
deleting, 15-29
updating, 15-29
RECORDSIZE specifier
in INQUIRE, 11-59
in OPEN, 11-34

Index-9

February 1987

RECORDTYPE specifier
in INQUIRE, 11-59
in 0 PEN, 11- 3 6
Register allocation
global, 15-9
Relational expressions, 5-11
Relational operators, 5-11
Remark field, 2-4
Remarks, 2-6
Remote file specifications, 11-21
Remote links, 15-2
REMOVL in LINK, 15-20
Reordering computations, 15-4
Repeat specification, 7-22
Repeatable edit descriptors, 12-7
REREAD
sequential FORMAT-statement,
10-43
sequential list-directed, 10-44
REREAD statement, 10-43
RESET in FOROTS, 15-11
RETURN statement, 13-47
REWIND statement, 11-63
REWRITE
formatted, 10-61
unformatted, 10-61
REWRITE statement, 10-60
RMS
data files, 18-14
defining keys, 11-24
invoking, 11-29
RMS data structures, 18-40
RMS files
writing, 10-42
RMS indexed files
reading, 10-33
RMS/FOROTS
data and control structures,
18-40
RMSUTL utility, 15-24
RNT function, 18-34
ROT function, 18-34
Routines
calling BLISS-36, 15-19
channel allocation, 18-30
channel deallocation, 18-30
memory allocation, 18-29
RPG function, 18-35
RPSI function, 18-35
RRS function, 18-34
RTINIT subroutine, 19-3
RTREAD subroutine, 19-5
RTSLP subroutine, 19-6
RTSTRT subroutine, 19-4
RTWAKE subroutine, 19-6
RTWRIT subroutine, 19-5
Running the FORTRAN-10 compiler,
16-2
Running the FORTRAN-20 compiler,
16-6
RUNOVL in LINK, 15-20
S ed i t i ng, 12-24

SAOVL in LINK, 15-20
SAVE statement, 7-24
SAVFMT subroutine, 13-37
SAVRAN subroutine, 13-37
SCALE subroutine, F-10
SECNDS function, 13-19
Sequential access, 10-6, 18-3
Sequential FORMAT-statement
PUNCH, 10-53
Sequential FORMAT-statement
ACCEPT, 10-45
Sequential FORMAT-statement PRINT,
10-50
Sequential FORMAT-statement READ,
10-29
Sequential FORMAT-statement
REREAD, 10-43
Sequential FORMAT-statement TYPE,
10-48
Sequential FORMAT-statement WRITE,
10-37
Sequential indexed READ, 10-34
Sequential list-directed ACCEPT,
10-46
Sequential list-directed PRINT,
10-51
Sequential list-directed PUNCH,
10-53
Sequential list-directed READ,
10-30
Sequential list-directed REREAD,
10-44
Sequential list-directed TYPE,
10-49
Sequential list-directed WRITE,
10-39
Sequential NAMELIST-statement
READ, 10-31
Sequential NAMELIST-statement
WRITE, 10-40
SEQUENTIAL specifier
in INQUIRE, 11-60
Sequential unformatted READ,
10-32
Sequential unformatted WRITE,
10-41
SETABL subroutine, F-11
SETRAN subroutine, 13-38
Sharable high segment, 16-35
SHARED specifier
in OPEN, 11-38
SKIPFILE statement, 11-67
SKIPRECORD statement, 11-66
Slash editing, 12-23
Software
Real-time, 19-1
SORT subroutine, 13-38
Source files
line-sequenced, 2-7
Source program, 1-1
SP editing, 12-24
Specific functions, 13-2
Specification statements, 7-1

Index-10

February 1987

Specifiers
carriage-control, 12-9
CLOSE statement, 11-44
END=, 10-18
ERR=, 10-19
format, 10-14
INQUIRE statement, 11-54
IOSTAT, 10-20
key-field-value, 10-21
key-of-reference, 10-23
OPEN statement, 11-4
record, 10-13
SRTINI subroutine, 13-39
SS editing, 12-24
START LSCW, 18-4
Starting FORDDT, 17-5
Statement block, 9-8
Statement definition, 2-2
Statement field, 2-4
Statement format, 2-2
Statement function
restrictions, 13-17
Statement functions, 13-15
using, 13-16
Statement label constants, 3-7
Statement label field, 2-3
Statement numbers, 2-3
Statements
ACCEPT, 10-45
arithmetic assignment, 8-1
arithmetic IF, 9-3
ASSIGN, 8-3
assigned GO TO, 9-3
assignment, 8-1
BACKFILE, 11-67
BACKSPACE, 11-64
BLOCK DATA, 14-1
block IF, 9-5
CALL, 13-44
character assignment, 8-4
character type specification,
7-6
CLOSE, 11-43
COMMON, 7-8
compilation control, 6-4
computed GO TO, 9-2
CONTINUE, 9-17
control, 9-1
DATA, 7-21
data transfer, 10-1
data transfer summary, 10-3
DECODE, 10-56
DELETE, 10-59
device control, 10-1, 11-61
DIMENSION, 7-1
DO WHILE, 9-13
ELSE, 9-5
ELSE IF THEN, 9-5
ENCODE, 10-56
END, 9-20
END DO, 9-17
END IF, 9-5
ENDFILE, 11-65

Statements (Cont.)
ENTRY, 13-45
EQUIVALENCE, 7-11
executable, 1-1, 6-1
EXTERNAL, 7-17
file control, 10-1, 11-1
FIND, 11-62
FORMAT, 12-2
FUNCTION, 13-20
GO TO, 9-1
IF, 9-3
IF THEN, 9-5
IMPLICIT, 7-7
INCLUDE, 6-4
indexed DO, 9-10
input/output, 10-1
INQUIRE, 11-52
internal READ, 10-55
internal WRITE, 10-55
INTRINSIC, 7-18
logical assignment, 8-3
logical IF, 9-4
logical two-branch IF, 9-4
NAMELIST, 12-44
nested DO, 9-14
nonexecutable, 1-1, 6-2
numeric type specification, 7-5
OPEN, 11-1
ordering of, 6-2
PARAMETER, 7-20
PAUSE, 9-19
PRINT, 10-50
PROGRAM, 6-4
PUNCH, 10-52
READ, 10-27
REREAD, 10-43
RETURN, 13-47
REWIND, 11-63
REWRITE, 10-60
SAVE, 7-24
SKIPFILE, 11-67
SKIPRECORD, 11-66
specification, 7-1
STOP, 9-18
SUBROUTINE, 13-43
Summary of, A-I
TYPE, 10-47
type specification, 7-4
unconditional GO TO, 9-2
UNLOAD, 11-64
UNLOCK, 10-62
WRITE, 10-35
STATI subroutine, 19-6
STATO subroutine, 19-5
STATUS specifier
in CLOSE, 11-50
in OPEN, 11-39
STOP statement, 9-18
Storage
common, 7-9
Subexpressions
parenthetical, 5-13

Index-II

February 1987

Subprogram
function, 13-20
subroutine, 13-22
Subprograms
block data, 14-1
Subroutine calls
writing, 15-11
SUBROUTINE statement, 13-43
Subroutine subprogram, 13-22
Subroutines, 13-1, 13-22
ALCCHR, 13-24
AXIS, F-2
BLKRW, 19-4
calling COBOL, 15-18
CAXIS, F-3
cncos, 13-25
CDEXP, 13-26
CDLOG, 13-26
CDSIN, 13-27
CDSQRT, 13-28
CHKDIV, 13-28
CLRFMT, 13-29
CONNECT, 19-4
DATE, 13-29
DISCON, 19-7
DISMIS, 19-6
DIVERT, 13-30
DTOGA, 13-30
DUMP, 13-30
ERRSET, 13-31
ERRSNS, 13-32
EXIT, 13-33
FFUNIT, 13-33
FORRTF, 19-1
FORTRAN-supplied, 13-24
FORTRAN-supplied plotter, F-l
GTODA, 13-34
ILL, 13-34
LEGAL, 13-34
LINE, F-4
LOCK, 19-3
MKTBL, F-5
MVBITS, 13-35
NUMBER, F-9
OVERFL, 13-35
PDUMP, 13-36
PLOT, F-10
PLOTS, F-2
programming considerations,
15-3
.
QUITEX, 13-36
RTINIT, 19-3
RTREAD, 19-5
RTSLP, 19-6
RTSTRT, 19-4
RTWAKE, 19-6
RTWRIT, 19-5
SAVFMT, 13-37
SAVRAN, 13-37
SCALE, F-10
SETABL, F-11
SETRAN, 13-38
SORT, 13-38

Subroutines (Cont.)
SRTINI, 13-39
STATI, 19-6
STATO, 19-5
SYMBOL, F-12
TIME, 13-39
TOPMEM, 13-40
TRACE, 13-41
UNLOCK, 19-7
user-defined, 13-42
WHERE, F-12
Subscript
array element, 4-3
Substrings
character, 4-6
Summary of CLOSE statement
specifiers, 11-45
Summary of OPEN statement
specifiers, 11-6
Summary of statements, A-I
Switches
ABORT, 16-9
BINARY, 16-9
CROSSREF, 16-3, 16-9
DEBUG, 16-3, 16-9, 16-11
DFLOATING, 16-3, 16-9
ECHO-OPTION, 16-9
EXPAND, 16-3, 16-9
EXTEND, 16-9
F66, 16-3, 16-9
F77, 16-3, 16-9
FLAG, 16-3, 16-21
FLAG-NON-STANDARD, 16-9
FORTRAN-20 compiler, 16-9
GFLOATING, 16-3, 16-9
INCLUDE, 16-3, 16-9
LISTING, 16-9
LNMAP, 16-3, 16-10
MACHINE-CODE, 16-10
MACROCODE, 16-3
NOBINARY, 16-10
NOERRORS, 16-3, 16-10
NOEXTEND, 16-10
NOF77, 16-3, 16-10
NOFLAG, 16-3
NOFLAG-NON-STANDARD, 16-10
NOWARN, 16-3, 16-10, 16-15
OPTIMIZE, 16-3, 16-10, 17-17
OPTION, 16-10
SYNTAX, 16-3, 16-10
SYMBOL subroutine, F-12
Symbolic names, 4-1
SYNTAX switch, 16-3, 16-10
T editing, 12-19
TAPEFORMAT specifier
in OPEN, 11-41
TIM2GO function, 13-19
TIME subroutine, 13-39
TL editing, 12-20
TOPMEM subroutine, 13-40
TR editing, 12-20
TRACE subroutine, 13-41

Index-12

February 1987

TRACE with DEBUG, 16-12
TYPE
sequential FORMAT-statement,
10-48
sequential list-directed, 10-49
Type declarators, 7-5
Type specification statements,
7-4
character, 7-6
numeric, 7-5
TYPE specifier
in OPEN, 11-39
TYPE statement, 10-47
Unconditional GO TO statements,
9-2
Unformatted data transfers, 10-8
to ASCII devices, 10-8
Unformatted Indexed READ, 10-35
Unformatted READ
direct, 10-33
sequential, 10-32
Unformatted READ transfers, 10-32
Unformatted REWRITE, 10-61
UNFORMATTED specifier
in INQUIRE, 11-60
Unformatted WRITE
direct, 10-41
sequential, 10-41
Unformatted WRITE transfers,
10-40
Unit identifier
logical, 10-11
UNIT specifier
in CLOSE, 11-51
in OPEN, 11-42
UNLOAD statement, 11-64
UNLOCK statement, 10-62
UNLOCK subroutine, 19-7
Updating records
in indexed files, 15-29
UPG function, 18-36
User-defined character set, F-9
User-defined external functions,
13-20
User-defined subroutines, 13-42
USEROPEN example, 18-38
USEROPEN procedures, 18-38
USEROPEN specifier
in OPEN, 11-42

Using FOROTS, 18-15
Using FORRTF, 19-2
Using function subprograms, 13-21
Using indexed files, 15-24
exceptions, 15-30
Using intrinsic functions, 13-2
Using statement functions, 13-16
Using the compiler, 16-1
Using the FORTRAN-20 compiler,
16-5
Variable names, 4-2
Variables, 4-2
compiler generated, 16-24
VERSION specifier
in OPEN, 11-43
VMS with FLAG, 16-22
Warning messages, 16-34, C-8
WHERE subroutine, F-12
WRITE
direct FORMAT-statement, 10-38
direct unformatted, 10-41
sequential FORMAT-statement,
10-37
sequential list-directed, 10-39
sequential NAMELIST-statement,
10-40
sequential unformatted, 10-41
WRITE statement, 10-35
internal, 10-55
WRITE transfers
formatted, 10-37
unformatted, 10-40
Writing arithmetic expressions,
5-2
Writing indexed files, 15-26
Writing programs, 15-1
for use on other computers,
15-2
Writing subroutine calls, 15-11
Writing to RMS files, 10-42
WRS function, 18-34
X editing, 12-21
XAB block, 18-40
fields, 18-43
Z (hexadecimal) editing, 12-37

Index-13

February 1987

TOPS-10/TOPS-20
FORTRAN Language Manual
AA-N383B-TK

READER'S COMMENTS
NOTE:

This form is for document comments only. DIGITAL will use comments submitted on
this form at the company's discretion. If you require a written reply and are eligible to
receive one under Software Perfo'rmance Report (SPR) service, submit your comments on an SPR form.

Did you find this manual understandable, usable, and well-organized? Please make suggestions for improvement.

Did you find errors in this manual? If so, specify the error and the page number.

Please indicate the type of reader that you most nearly represent.

o Assembly language programmer
o Higher-level language programmer
o Occasional programmer (experienced)
o User with little programming experience
o Student programmer
o Other (please specffy)~~~~~~~~~~~~~~~~~~~~~~
Name

Date _~_ _~_~~~~_

Organization _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ Telephone _ _ _ _ _ _ _ __
Street _____________________________________
City _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ State ____ Zip Code _ _ __
or Country

- -- - -- - ~.- -ODO ~.oto
Tear t- F~HeIr e and Tape- - - - - - - - - -- - - - - - - - - - -

-r ~-111--------~~~~;:;~~ --

if Mailed in the
United States

BUSINESS REPLY MAIL
FIRST CLASS PERMIT NO. 33 MAYNARD MASS.
POSTAGE WILL BE PAID BY ADDRESSEE

SOFTWARE PUBLICATIONS

200 FOREST STREET

MR01-2/L 12

MARLBOROUGH, MA

01752

I
1

Do Not Tear - Fold Here and Tape - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - __ I
1
1
1

1
1

UPDATE NOTICE

TOPS-10/TOPS-20
FORTRAN Language Manual
AD-N383B-T1

February 1987

Insert this Update Notice in the TOPS-10/TOPS-20
FORTRAN Language Manual to maintain an up-to-date record
of changes to the manual.

Changed Information:
The changed pages contained in this update package reflect
Version 11 of FORTRAN-1 0/20.
The instructions for inserting this update start on the next page.

~DmDDmD

wore

INSTRUCTIONS
AD-N383B-T1
The following list of page numbers specifies which pages are to be placed in the
TOPS-10/TOPS-20 FORTRAN Language Manual as replacements for, or additions to,
current pages.
Title Page
Copyright Page

Entire
Chapter 11

17-3
17-4

18-37
18-43

Entire
Contents

12-1
12-2

17-7
17-8

Entire
Appendix A

Entire
Preface

13-31
13-34

17-11
17-12

Entire
Appendix 0

2-1
2-4

15-1
15-2.2

17-15
17-16

Entire
Index

4-1
4-2

15-15
15-20

18-1
18-4

Readers Comment
Mailer

6-1
6-2

15-23
15-30

18-13
18-18

Entire
Chapter 10

16-11
16-35

18-29
18-30

KEEP THIS UPDATE NOTICE IN YOUR MANUAL TO MAINTAIN AN
UP-TO-DATE RECORD OF CHANGES.
TYPE AND IDENTIFICATION OF DOCUMENTATION CHANGES.
Five types of changes are used to update documents contained in the TOPS-10/TOPS-20
software manuals. Change symbols and notations are used to specify where, when,
and why alterations were made to each update page. The five types of update changes
and the manner in which each is identified are described in the following table.
The Following Symbols and/or Notations

Identify the Following Types of Update Changes

Change bar in outside margin; version
number and change date printed
at bottom of page.

Changes were required by a new version
of the software being described.

Change bar in outside margin; change date
printed at bottom of page.

Changes were required to either clarify or
correct the existing material.

Change date printed at bottom of page.

Changes were made for editorial purposes
but use of the software is not affected.

Bullet (e) in outside margin; version number
and change date printed at bottom of page.

Data was deleted to comply with a new ver'sion of the software being described.

Bullet (e) in outside margin; change date
printed at bottom of page.

Data was deleted to eit~er clarify or correct
the existing material.

February 1987