Digital PDFs

AA-4158B-TM

May 1977

294 pages

Original

11MB

Document:	FORTRAN-20 Reference Manual Ver 5 Apr77
Order Number:	AA-4158B-TM
Revision:	0
Pages:	294
Original Filename:	AA-4158B-TM_FORTRAN-20_Reference_Manual_Ver_5_Apr77.pdf

OCR Text

FORTRAN Reference Manual
Order No. AA-4158B-TM

April 1977

This document describes the language elements of the
FORTRAN-20 compiler for the DECSYSTEM-20.
This document supersedes the document of the
same name, Order No. DEC-20-LFMRA-A-D,
. published January 1976.

OPERATING SYSTEM AND VERSION:

Any Digital-supported operating system for the
DECSYSTEM-20.

SOFTWARE VERSION:

FORTRAN-20, Version 5

To order additional copies of this document, contact the Software Distribution
Center, Digital Equipment Corporation, Maynard, Massachusetts 01754

diaital
eauioment
corcoration
· maunard,
massachusetts
......,
..
.
.......,

First Printing, January 1976
Revised:
April 1977

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 be used or copied only in accordance with the terms of such
license.
Digital Equipment Corporation assumes no responsibility for the use
or reliability of its software on equipment that is not supplied by
DIGITAL.

The postage prepaid READER'S COMMENTS form on the last page of this
document requests the user's critical evaluation to assist us in preparing future documentation.
The following are trademarks of Digital Equipment Corporation:

DIGITAL
DEC
PDP
DECUS
UNIBUS
COMPUTER LABS
COMTEX
DDT
DECCOMM

DECsystem-10
DEC tape
DIBOL
EDUSYSTEM
FLIP CHIP
FOCAL
INDAC
LAB-8
DECsystem-20

MASSBUS
OMNIBUS
OS/8
PHA
RSTS
RSX
TYPESET-8
TYPESET-10
TYPESET-II

5/77-15

CONTENTS

Page
CHAPTER

CHAPTER

PROLOGUE

1-1

1.1

BACKGROUND

1-1

CHARACTERS AND LINES

2-1

2.1
2.2
2.2.1

CHARACTER SET
STATEMENT, DEFINITION, AND FORMAT
Statement Label Field and Statement
Numbers
Line Continuation Field
Statement Field
Remarks
LINE TYPES
Initial and Continuation Line Types
Multi-Statement Lines
Comment Lines and Remarks
Debug Lines
Blank Lines
Line-Sequenced Input
ORDERING OF STATEMENTS

2-1
2-2

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
2.3.6
2.4
CHAPTER

CHAPTER

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

DATA TYPES, CONSTANTS, SYMBOLIC NAMES,
VARIABLES, AND ARRAYS

3-1

3.1
3.2
3.2.1
3.2.2
3.2.3
3.2.4
3.2.5
3.2.6
3.2.7
3.2.8
3.3
3.4
3.5
3.5.1
3.5.2
3.5.3

DATA TYPES
CONSTANTS
Integer Constants
Real Constants
Double-Precision Constants
Complex Constants
Octal Constants
Logical Constants
Literal Constants
Statement Label Constants
SYMBOLIC NAMES
VARIABLES
ARRAYS
Array Element Subscripts
Dimensioning Arrays
Order of Stored Array Elements

3-1
3-2
3-2
3-2
3-3
3-3
3-4
3-5
3-5
3-6
3-6
3-7
3-7
3-8
3-9
3-10

EXPRESSIONS

4-1

4.1
4.1.1

ARITHMETIC EXPRESSIONS
Rules for Writing Arithmetic
Expressions
LOGICAL EXPRESSIONS
Relational Expressions
EVALUATION OF EXPRESSIONS
Parenthesized Subexpressions

4-1

4.2
4.2.1
4.3
4.3.1

iii

4-2
4-4
4_-7
4-9
4-9

CONTENTS

(CONT.)

Page
4.3.2
4.3.3
4.3.4
CHAPTER

Hierarchy of Operators
Mixed Mode Expression
Use of Logical Operands in Mixed Mode
Expressions

4-9
4-10

COMPILATION CONTROL STATEMENTS

5-1

5.1
5.3
5.4

INTRODUCTION
PROGRAM STATEMENT
INCLUDE STATEMENT
END STATEMENT

5-1
5-1
5-1
5-1

SPECIFICATION STATEMENTS

6-1

6.1

6-1
6-1

6.8

INTRODUCTION
DIMENSION STATEMENT
Adjustable Dimensions
TYPE SPECIFICATION STATEMENTS
IMPLICIT STATEMENTS
COMMON STATEMENTS
Dimensioning Arrays in COMMON
Statements
EQUIVALENCE STATEMENT
EXTERNAL STATEMENT
PARAMETER STATEMENT

DATA STATEMENT

7-1

7.1

INTRODUCTION

7-1

ASSIGNMENT STATEMENTS

8-1

8.1

INTRODUCTION
ARITHMETIC ASSIGNMENT STATEMENTS
LOGICAL ASSIGNMENT STATEMENTS
ASSIGN (STATEMENT LABEL) ASSIGNMENT
STATEMENT

8-1
8-1
8-4

CONTROL STATEMENTS

9-1

9.1
9.2
9.2.1
9.2.2
9.2.3
9.3
9.3.1
9.3.2
9.3.3
9.4
9.4.1
9.4.2
9.4.3
9.5
9.6
9.7
9.7.1

INTRODUCTION
GO TO CONTROL STATEMENTS
Unconditional GO TO Statements
Computed GO TO Statements
Assigned GO TO Statements
IF STATEMENTS
Arithmetic IF Statements
Logical IF Statements
Logical Two-Branch IF Statements
DO STATEMENT
Nested DO Statements
Extended Range
Permitted Transfer Operations
CONTINUE STATEMENT
STOP STATEMENT
PAUSE STATEMENT
T (TRACE) Option

9-1
9-1
9-1
9-2
9-2
9-3
9-3
9-4
9-4
9-5
9-6
9-8
9-9
9-10
9-10
9-11
9-12

5.2

CHAPTER

6.2
6.2.1
6.3
6.4
6.5
6.5.1
6.6
6.7
CHAPTER

CHAPTER

8.2
8.3
8.4
CHAPTER

4-11

6-2
6-3
6-5
6-5
6-7
6-7

6-8
6-9

8-4

CONTENTS (CONT.)

Page
CHAPTER

I/O STATEMENTS

10-1

10.1
10.2
10.2.1
10.2.2
10.2.3
10.3

DATA TRANSFER OPERATIONS
TRANSFER MODES
Sequential Mode
Random Access Mode
Append Mode
I/O STATEMENTS, BASIC FORMATS AND
COMPONENTS
10.3.1
I/O Statement Keywords
10.3.2
FORTRAN Logical unit Numbers
10.3.3
FORMAT Statement References
10.3.4
I/O List
10.3.4.1
Implied DO Constructs
10.3.4.2
Formatted Record Handling
10.3.5
The Specification of Records for
Random Access
10.3.6
List-Directed I/O
NAMELIST I/O Lists
10.3.7
10.4
OPTIONAL READ/WRITE ERROR EXIT AND
END-OF-FILE ARGUMENTS
READ STATEMENTS
10.5
10.5.1
Sequential Formatted READ Transfers
10.5.2
Sequential Unformatted Binary READ
Transfers
10.5.3
Sequential List-Directed READ
Transfers
10.5.4
Sequential NAMELIST-Contro11ed READ
Transfers
10.5.5
Random Access Formatted READ Transfers
10.5.6
Random Access Unformatted READ
Transfers
10.6
SUMMARY OF READ STATEMENTS
10.7
REREAD STATEMENT
WRITE STATEMENTS
10.8
10.8.1
Sequential Formatted WRITE Transfers
10.8.2
Sequential Unformatted Binary WRITE
Transfer
10.8.3
Sequential List-Directed WRITE Transfers
10.8.4
Sequential NAMELIST-Contro11ed WRITE
Transfers
10.8.5
Random Access Formatted WRITE Transfers
10.8.6
Random Access Unformatted WRITE
Transfers
10.9
SUMMARY OF WRITE STATEMENTS
10.10
ACCEPT STATEMENT
10.10.1
Formatted ACCEPT Transfers
ACCEPT Transfers Into FORMAT Statements
10.10.2
10.11
PRINT STATEMENT
10.12
TYPE STATEMENT
10.13
FIND STATEMENT
10.14
ENCODE AND DECODE STATEMENTS
10.14.1
ENCODE Statement
10.14.2
DECODE Statement
10.14.3
Example of ENCODE/DECODE Operations
10.15
SUMMARY OF I/O STATEMENTS

10-1
10-1
10-1
10-1
10-2
10-2
10-3
10-3
10-3
10-6
10-6
10-7
10-7
10-8
10-10
10-10
10-11
10-11
10-12
10-12
10-13
10-13
10-13
10-14
10-14
10-16
10-16
10-16
10-17
10-17
10-17
10-17
10-18
10-18
10-18
10-19
10-19
10-20
10-21
10-21
10-22
10-22
10-23
10-24

CONTENTS (CONT.)

Page
CHAPTER

CHAPTER

NAMELIST STATEMENTS

11-1

11.1
11.2
11.2.1
11.2.2

INTRODUCTION
NAMELIST STATEMENT
NAMELIST-Controlled Input Transfers
NAMELIST-Controlled Output Transfers

11-1
11-1
11-2
11-3

FILE CONTROL STATEMENTS

12-1

12.1
12.2
12.2.1
12.2.2

INTRODUCTION
OPEN AND CLOSE STATEMENTS
Options for OPEN and CLOSE Statements
Summary of OPEN/CLOSE Statement Options

12-1
12-1
12-2
12-10

FORMAT STATEMENT

13-1

13.1
13.1.1
13.2
13.2.1
13.2.2

INTRODUCTION
FORMAT Statement, General Form
FORMAT DESCRIPTORS
Numeric Field Descriptors
Interaction of Field Descriptors
With I/O Variables
G, General Numeric Conversion Code
Numeric Fields with Scale Factors
Logical Field Descriptors
Variable Numeric Field widths
Alphanumeric Field Descriptors
Transferring Alphanumeric Data
Mixed Numeric and Alphanumeric Fields
Multiple Record Specifications
Record Formatting Field Descriptors
$ Format Descriptor
CARRIAGE CONTROL CHARACTERS FOR PRINTING
ASCII RECORDS

13-1
13-1
13-2
13-4

DEVICE CONTROL STATEMENTS

14-1

14.1
14.2
14.3
14.4
14.5
14.6
14.7
14.8
14.9

INTRODUCTION
REWIND STATEMENT
UNLOAD STATEMENT
BACKSPACE STATEMENT
END FILE STATEMENT
SKIP RECORD STATEMENT
SKIP FILE STATEMENT
BACKFILE STATEMENT
SUMMARY OF DEVICE CONTROL STATEMENTS

14-1
14-1
14-2
14-2
14-2
14-3
14-3
14-3
14-3

SUBPROGRAM STATEMENTS

15-1

15.1
15.1.1
15.2
15.3

INTRODUCTION
Dummy and Actual Arguments
STATEMENT FUNCTIONS
INTRINSIC FUNCTIONS (FORTRAN DEFINED
FUNCTIONS)
EXTERNAL FUNCTIONS
Basic External Functions (FORTRAN-20
Defined Functions)
Generic Function Names

15-1
15-1
15-3

13.2.3
13.2.4
13.2.5
13.2.6
13.2.7
13.2.8
13.2.9
13.2.10
13.2.11
13.2.12
13.3
CHAPTER

CHAPTER

15.4
15.4.1
15.4.2

13-6
13-7
13-7
13-10
13-10
13-11
13-12
13-14
13-14
13-15
13-16
13-16

15-3
15-6
15-7
15-7

CONTENTS (CONT.)

Page
SUBROUTINE SUBPROGRAMS
Referencing Subroutines (CALL Statement)
FORTRAN-20 Supplied Subroutines
RETURN STATEMENT AND MULTIPLE RETURNS
Referencing External FUNCTION
Subprograms
MULTIPLE SUBPROGRAM ENTRY POINTS (ENTRY
STATEMENT)

15-8
15-11
15-12
15-12

BLOCK DATA SUBPROGRAMS

16-1

16.1
16.2

INTRODUCTION
BLOCK DATA STATEMENT

16-1
16-1

APPENDIX A

ASCII-1968 CHARACTER CODE SET

A-I

APPENDIX B

USING THE COMPILER

B-1

RUNNING THE COMPILER
Switches Available with FORTRAN-20
The /DEBUG Switch
COMPIL-C1ass Commands
READING THE LISTING
Compiler-Generated Variables
ERROR REPORTING
Fatal Errors and Warning Messages
Message Summary
CREATING A REENTRAN'I' FORTRAN PROGRAH
WITH LINK

B-1
B-1
B-3
B-4
B-5
B-6
B-17
B-17
B-18

WRITING USER PROGRAMS

C-1

GENERAL PROGRAMMING CONSIDERATIONS
Accuracy and Range of Double-Precision
Numbers
Writing FORTRAN-20 Programs for
Execution on Non-DEC Machines
Using Floating-Point DO Loops
Computation of DO Loop Iterations
Subroutines - Programming Considerations
Reordering of Computations
Dimensioning of Formal Arrays
FORTRAN-20 GLOBAL OPTIMIZATION
Optimization Techniques
Elimination of Redundant Computations
Reduction of Operator Strength
Removal of Constant Computation From
Loops
Constant Folding and Propagation
Removal of Inaccessible Code
Global Register Allocation
I/O Optimization
Uninitia1ized Variable Detection
Test Replacement
Improper Function References
Programming 'I'echniques for Effective
Optimization

C-1

15.5
15.5.1
15.5.2
15.6
15.6.1
15.7
CHAPTER

B.1
B.1.1
B.1.1.1
B.1.2
B.2
B.2.1
B.3
B.3.1
B.3.2
B.4
APPENDIX C
C.1
C.1.1
C.1.2
C.1.3
C.1.4
C.1.5
C.1.6
C.1.7
C.2
C.2.1
C.2.1.1
C.2.1.2
C.2.1.3
C.2.1.4
C.2.1.5
C.2.1.6
C.2.1.7
C.2.1.8
C.2.1.9
C.2.2
C.2.3

vii

15-14
15-15

B-18

C-1
C-1
C-2
C-2
C-2
C-3
C-4
C-4
C-5
C-5
C-5
C-6
C-7
C-7
C-7
C-8
C-8
C-8
C-8
C-9

CONTENTS (CONT.)

Page
C.3
C.3.1
C.3.2
C.3.3
C.3.4
C.3.S
C.3.6
C.3.7
C.3.7.1
C.3.7.2
C.3.8
C.3.8.1
APPENDIX D

INTERACTING WITH NON-FORTRAN PROGRAMS
AND FILES
Calling Sequences
Accumulator Usage
Argument Lists
Argument Types
Description of Arguments
Converting Existing MACRO Libraries
for use with FORTRAN-20
Interaction with COBOL
Calling FORTRAN-20 Subroutines from
COBOL Programs
Calling COBOL Subroutines from
FORTRAN-20 Programs
LINK Overlay Facilities
Conventions
FOROTS

C-9
C-9

C-IO
C-ll
C-12
C-13
C-14
C-20
C-21
C-22
C-22
C-23
D-l

D.l
D.2
D.3
D.4
D.4.1

HAROWARE AND SOFTWARE REQUIREMENTS
FEATURES OF FOROTS
ERROR PROCESSING
INPUT/OUTPUT FACILITIES
Input/Output Channels Used Internally by
FOROTS
File Access Modes
D.4.2
D.4.2.1
Sequential Transfer Mode
D.4.2.2
Random Access Mode
ACCEPTABLE TYPES OF OATA FILES ANO THEIR
D.S
FORMATS
ASCII Data Files
D.S.l
D.S.2
FORTRAN Binary Oata Files
D.S.2.1
Format of Binary Files
D.S.3
Mixed Mode Oata Files
D.S.4
Image Files
D.6
USING FOROTS
FOROTS Entry Points
D.6.1
Calling Sequences
D.6.2
MACRO Calls for FOROTS Functions
D.6.3
D.6.3.1
I/O Statements, Sequential Access
Calling Sequences
D.6.3.2
NAMELIST I/O Sequential Access Calling
Sequences
D.6.3.3
Array Offsets and Factoring
D.6.3.4
I/O Statements Random Access Calling
Sequences
D.6.3.S
Calling Sequences for Statements That
Use Oefault Oevices
Statements to position Magnetic
D.6.3.6
Tape Units
D.6.3.7
List Oirected Input/Output Statements
Input/Output Oata Lists
D.6.3.8
OPEN and CLOSE Statements,
D.6.3.9
Calling Sequences
D.6.3.10
Memory Allocation Routines
Software Channel Allocation and
D.6.3.11
Oe-allocation Routines
FUNCTIONS TO FACILITATE OVERLAYS
D.7
LOGICAL/PHYSICAL OEVICE ASSIGNMENTS
D.8

viii

D-l
0-2
0-3
0-3
D-3
0-4
0-4
0-4
0-4
D-4
D-S
O-S
D-12
0-13
0-13
0-14
0-14
O-IS
0-16
0-17
D-18
0-20
0-20
0-22
0-22
0-23
0-26
0-27
0-28
0-29
0-32

CONTENTS

(CONT.)

Page
FORDDT

E-l

INPUT FORMAT
Variables and Arrays
Numeric Conventions
Statement Labels and Source Line Numbers
NEW USER TUTORIAL
Basic Commands
FORDDT AND THE FORTRAN-20/DEBUG SWITCH
LOADING AND STARTING FORDDT
SCOPE OF NAME AND LABEL REFERENCES
FORDDT COMMANDS
ENVIRONMENT CONTROL
FORTRAN-20/0PTIMIZE SWITCH
FORDDT MESSAGES

E-2
E-2
E-3
E-3
E-3
E-3
E-7
E-7
E-8
E-8
E-17
E-17
E-17

APPENDIX F

COMPILER MESSAGES

F-l

APPENDIX G

FOROTS ERROR MESSAGES

G-l

APPENDIX H

DECSYSTEM-10 COMPATABILITY

H-l

APPENDIX E
E.l
E.l.l
E.l.2
E.l.3
E.2
E.2.1
E.3
E.4
E.S
E.6
E.7
E.8
E.9

INDEX

Index-l
TABLES

TABLE

1-1
2-1
3-1
3-2
4-1
4-2
4-3
4-4
4-S
4-6
4-7
8-1
10-1
10-2
10-3
10-4
12-1
13-1
13-2
13-3
13-4
13-S
14-1
lS-l

FORTRAN Statement Categories
FORTRAN Character Set
Constants
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-10 Operators
Rules for Conversion in Mixed Mode
Assignments
FORTRAN-20 Logical Device Assignments
Summary of READ Statements
Summary of WRITE Statements
Summary of I/O Statements
OPEN/CLOSE Statement Arguments
FORTRAN-20 Conversion Codes
Action of Field Descriptors On
Sample Data
Numeric Field Codes
Descriptor Conversion of Real and Double
Precision Data According to Magnitude
FORTRAN-20 Print Control Characters
Summary of FORTRAN-20 Device Control
Statements
Intrinsic Functions (FORTRAN-20 Defined
Functions)

1-2
2-1
3-1
3-6
4-1
4-3
4-4
4-S
4-6
4-7
4-10

8-2
10-4
10-lS
10-17
10-24
12-11
13-3
13-S
13-6

13-8
13-17
14-4
lS-4

CONTENTS (CaNT.)

Page

15-2
15-3
B-1
B-2
C-l
D-l
D-2
E-l
G-l
G-2

Basic External Functions (FORTRAN-20
Defined Functions)
FORTRAN-20 Library Subroutines
FORTRAN-20 Compiler Switches
Modifiers to /DEBUG Switch
Argument Types and Type Codes
Function Numbers and Function Codes
FORTRAN Device Table
Table of Commands
FOROTS I/O Error Messages and ERRSNS
Returned Values
FOROTS Arithmetic and Library Error
Messages

15-9
15-17
B-2
B-3
C-12
D-30
D-33
E-l
G-2

G-5

CHAPTER 1
PROLOGUE

1.1

BACKGROUND

A FORTRAN source program consists of statements constructed using the
language elements and the syntax described in this manual.
A
statement performs one of the following functions:
1.

Causes operations such as
branching to be carried out.

multiplication,

division,

Specifies the type and format of data being processed.

Specifies the characteristics of the source program.

and

FORTRAN statements are composed of keywords,
i.e., 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
action
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.
In this manual, the FORTRAN statements are grouped into 12 categories,
each of which is described in a separate chapter.
The name,
definition, and chapter reference for each statement category are
given in Table 1-1.
The basic FORTRAN language elements,
(constants, variables,
and
expressions), the character set from which they may be formed, and the
rules that govern their construction and use are described in Chapters
2 through 4.

1-1

PROLOGUE
Table 1-1
FORTRAN Statement Categories
Chapter
Reference

Category
Name

Function

Compilation Control
Statements

Identify programs
their
beginning
points.

Specification
Statements

Declare
the
properties
of
variables, arrays, and functions.

DATA
Statements

Assign
initial
values
variables and array elements.

Assignment
Statements

Assign values to named variables
and array elements.

Control
Statements

Determine the order of execution
of
the
object
program
and
terminate its execution.

Input/Output
Statements

Transfer data between internal
storage
and
specified
input/output devices.

NAME LIST
Statements

Establish lists that are used
with
certain
input/output
statements to transfer data that
appears in a special type of
record.

File Control
Statements

Identify, open, and close files
and set parameters for input and
output operations between files
and the processor.

FORMAT
Statements

Specify
formats for
input/output devices.

Device Control
Statements

Control
the
positioning
of
records or
files
on
certain
input/output devices.

Subprogram
Statements

Define functions and subroutines
and their entry points.

BLOCK DATA
Statements

Define
data
specification
subprograms that may initialize
common storage areas.

1-2

and
and

indicate
ending

data

PREFACE

This manual describes the FORTRAN language as implemented for
the
DECsystem-20 FORTRAN Language Processing System.
In the text, the
language is called FORTRAN-20 (to distinguish it from ANSI FORTRAN),
or simply FORTRAN.
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.
Your use of FORTRAN may also require use of other DECsystem-20
programs:
the monitor,
the CREF program, the debugging program, a
text editor, and the BATCH program.
These are described in the
following manuals:
User's Guide
DEC-20-0UGAA-A-D
Monitor Calls User's Guide
DEC-20-UMUGA-A-D
EDIT User's Guide
DEC-20-UEUGA-A-D
BATCH Reference Manual
DEC-20-0BRMA-A-D
The standard for FORTRAN is the American National Standards Institute
(ANSI)
FORTRAN,
X3.9-1966.
FORTRAN-20 extensions and additions to
ANSI FORTRAN are gray shaded.

CHAPTER 2
CHARACTERS AND 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 (1) ,
is
acceptable within literal constants or comment text,
but these
characters cause fatal errors in other contexts. An exception is
CONTROL-Z, which, when used in terminal input, means end-of-file.
NOTE
Lower-case
alphabet
characters
are
treated
as
upper-case
outside the
context of Hollerith constants, literal
strings, and comments.
Table 2-1
FORTRAN Character Set
Letters
A,a
B,b
C,c
D,d
E,e
F,f
G,g
H,h
I,i

S,s
T,t
U,u
V,v
W,w
X,x
Y,y
Z,z

J, j

K,k
L,l
M,m
N,n
0,0
P,p

Q,q
R, r
Digits

0
1
2
3
4

5
6
7
8
9

1. The complete ASCII-1968 character set is defined in the X3.4-l968
version
of
the
"American
National
Standard for
Information
Interchange," and is given in Appendix A.
2-1

CHARACTERS AND LINES
Table 2-1
(Cont.)
FORTRAN Character Set
Symbols

, Comma
- Hyphen (Minus)
Period (Decimal Point)
/ Slant (slash)
: Colon
; Semicolon
< Less Than
= Equals
Greater Than
>
"Circumflex

! Exclamation Point

Quotation Marks
# Number Sign
$ Dollar Sign
& Ampersand
I
Apostrophe
( Opening Parenthesis
) Closing Parenthesis
* Asterisk
+ Plus
1/

Line Termination Characters
Line Feed
Form Feed
vertical Tab
Line Formatting Characters
Carriage Return
Horizontal 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
includes or starts in character position 6.
(Refer to Section 2.3.1
for a description of initial and continuation line types.) Tabs within
literal 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.
Each line is divided into four fields:

Line Character Positions
~

234

-----------------+1_\

__~~--~'~'~---------y----------J
Statement
Label Field

Continuation
Field

Statement Field

2-2

73
Remarks

CHARACTERS AND LINES
2.2.1

Statement Label Field and

State~ent

Numbers

You may 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, e.g., the numbers 00105 and 105 are both accepted as
statement number 105.
You may 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. You cannot label non-executable statements other than
FORMAT and END statements.
A main program and a subroutine may 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:
Assume that main module MAINMD and subprogram SUBI both
A GO TO statement, for
contain statement number 105.
instance, in MAINMD will refer to statement 105 in MAINMD,
NOT to 105 in SUBI. A GO TO in SUBI will transfer control
to 105 in SUBI.
When you enter source programs into the system via a standard user
terminal, you may use an initial tab to skip all or part of the label
field.
If an initial tab is encountered during compilation, FORTRAN examines
the character immediately following the tab to determine the type of
line being entered. If the character following the tab is one of the
digits 1 through 9, FORTRAN considers the line as a continuation line
and the second character after the tab as the first character of the
statement field.
If the character following the tab is other than one
of the digits 1 through 9, FORTRAN 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 in a continuation
line, and in character position 7 in an initial line.

2.2.2

Line Continuation Field

Any alphanumeric character (except a blank or a zero) placed in
field
(position 6) identifies the line as a continuation line.
Section 2.3.1 for description.)

this
(See

Whenever you use a 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.

2.2.3

Statement Field

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

2-3

CHARACTERS AND LINES
2.2.4

Remarks

In lines consisting of 73 or more character positions, only the first
characters are interpreted by FORTRAN.
(Note that tabs generally
occupy more than one character position, usually advancing the counter
to the next character position that is an even multiple of eight.) All
other characters in the line (character positions 73, 74 ... etc.) are
treated as remarks and do not affect compilation.
72

Note that remarks may also be added to a line in character positions 7
through 72, provided the text of the remark is preceded by the symbol
"!" (Refer to 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 m'u"i'tI~:";'sWtatement line,

A comment line,

A debug line, or

A blank line.

Each of these line types is described in the following paragraphs.

2.3.1

Initial and Continuation Line Types

A FORTRAN statement may occupy the statement fields of up to 20
consecutive lines. The first line in a multi-line statement group is
referred to as the initial line; the succeeding lines are referred to
as continuation lines.
An initial line may be assigned a statement number and must have
either a blank or a zero in its continuation line field, i . e • ,
character position 6.
If you enter an initial line via a keyboard input device, you may use
an initial tab 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
non-numeric character, i.e., the first character of the statement
,field must be non-numeric.
Continuation lines cannot be assigned statement numbers;
they are
identified by any alphanumeric character (except for a blank or zero)
placed in character position 6 of the line, i.e., continuation line
field.
The label field of a continuation line is treated as remark
text.
If vou are entering a continuation line via a keyboard, you may use an
;initial tab to skip all or part of the label field; however, the tab
imust be followed immediately by a numeric character other than zero.
lThe tab-numeric combination identifies the line as a continuation
jline.

2-4

CHARACTERS AND LINES
Note that blank lines, comments, and debug lines that are treated like
comments, i.e., debug lines that are not compiled with the rest of the
program (refer to Section 2.3.4) terminate a continuation sequence.
Following is an example of a 4-1ine
initial tabs:
105

FORTRAN

FORMAT

statement

using

FORMAT (lHl,17HINITIAL CHARGE = ,FIO.6,10H
COULOMB,6X,
213HRESISTANCE = ,F9.3,6H
OHM/ISH CAPACITANCE = ,FIO.6,
38H
FARAD,11X,13HINDUCTANCE = ,F7.3,8H
HENRY///
421H
TIME
CURRENT/7H
MS,10X.2HMA///)
Continuation Line Characters, i.e., 2, 3, and 4

2.3.2

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 semicolons (;).

Only the first statement in the series can have
number.

Statements following the first statement
continuation of the preceding statement.

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

statement

cannot

a
next

An example of a multi-statement line is:
450

DIST=RATE * TIME ;TIME=TIME+0.05 iCALL PRIME(TIME,DIST)

2.3.3

Comment Lines and Remarks

Lines that contain descriptive text only are referred to as comment
lines.
Comment lines are commonly used to identify and introduce a
source program, to describe the purpose of a particular set of
statements, and to introduce subprograms.
To structure a comment line:
1.

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

You may write the text into character positions 2 through the
end of the line.

You may place comment lines anywhere in the source program,
but they cannot precede a continuation line because comments
terminate a continuation sequence.

You may write a large comment as a sequence of any number of
lines;
however, each line must carry the identifying
character (C,$,/,*, or 1) in its first character position.

2-5

CHARACTERS AND LINES
The following is an example of a comment that occupies more
line.

than

one

CSUBROUTINE - A12
CTHE PURPOSE OF THIS SUBROUTINE IS
CTO FORMAT AND STORE THE RESULTS OF
CTEST PROGRAM HEAT TEST-llOl
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 character group "Stop if card is blank" is identified as a remark
by the preceding! symbol. Remarks do not result in the generation
of object program code, but they will appear on listings. The symbol
1,
indicating a remark, must appear outside the context of a literal
specification.
Note that characters appearing in character positions 73 and beyond
are automatically treated as remarks, so that the symbol
need not
be used.
(Refer to 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, i.e.,
not compiled with the rest of the program unless the /INCLUDE switch
is present in the command string.
(Refer to Appendix C for a
,description of the file switch options.) When the /INCLUDE switch is
f present in the command string, the D (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.
Note
that the initial and all
continuation lines of a debug statement must contain a D (or d)
in
character position 1.

2.3.5

Blank Lines

You may insert lines consisting of only blanks, tabs, or no characters
anywhere in your source program except immediately preceding a
continuation line, because blank lines are by definition initial lines
and as such terminate a continuation sequence. 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.

2.3.6

Line-Sequenced Input

pr\nmn7\,...,.

__ ..... ..: ___ , , _ ...

_____ i.._

,..: _ _ _ _ _ ... _ _ _ _ ..:1

..c.:, __

_ _ _ * _ _ ..:1 .... _ _ ...:1

' - __

M1""\Trrt

L·V~"J.J:\.r\LII

V1:''-.LVllct.L.1.,Y

ctl,,;l,,;t::1:''-;:'

.1..Lllt::-i:)t::YUt::lll,,;t::U

LJ..1.t::i:)

cti:)

U,Y

I:.1.J.1.J.

1:'LVUUI,,;t::U

BASIC.
These sequence numbers are used in place of the listing line
numbers normally generated.

2-6

CHARACTERS AND LINES
2.4

ORDERING OF STATEMENTS

The order in which you place statements in a program unit is
important.
That is, certain types of statements have to be processed
before others to guarantee that compilation takes place as you expect.
The proper sequence for
statements is summarized by the following
diagram.
PROGRAM, FUNCTION, SUBPROGRAM, or
BLOCK DATA Statements
IMPLICIT Statements
PARAMETER Statements

Comment Lines

DIMENSION, COMMON,
EQUIVALENCE, EXTERNAL
NAMELIST, or Type
Specification Statements

FORMAT Statements

Statement
Function
Definitions
DATA Statements
Executable
Statements
END Statement
Horizontal lines indicate the order in which statements must appear.
That is, you cannot intersperse horizontal sections.
For example, all
PARAMETER statements must appear after all IMPLICIT statements and
before any DATA statements,
i.e.,
PARAMETER,
IMPLICIT, and DATA
statements cannot be interspersed.
Statement function definitions
must
appear
after
IMPLICIT
statements and before executable
statements.
Vertical lines indicate the way in which certain types of statements
may be interspersed.
For example, you may intersperse DATA statements
with statement function definitions and executable statements.
you
may intersperse FORMAT statements with IMPLICIT statements, parameter
statements, other specification statements, DATA statements, statement
function definitions, and executable statements. The only restriction
on the placement of FORMAT statements is that they must appear after
any PROGRAM,
FUNCTION,
subprogram, and BLOCK DATA statements, and
before the END statement.

2-7

CHARACTERS AND LINES
Special Cases:
1.

The placement of an INCLUDE statement
types of statements to be INCLUDEd.

The ENTRY statement is allowed only in
functions
or
subroutines.
All executable references to any of the dummy
parameters must physically follow the ENTRY statement unless
the references appear in the function definition statement,
the subroutine, or in a preceding ENTRY statement.

BLOCK DATA subprograms cannot
contain
any
executable
statements,
statement functions, FORMAT statements, EXTERNAL
statements, or NAMELIST statements.
(Refer to Section 16.1.)

When statements are out of place, FORTRAN
which may indicate fatal errors.

2-8

issues

dictated

messages,

some

the

CHAPTER 3
DATA TYPES, CONSTANTS, SYMBOLIC NAMES, VARIABLES, AND ARRAYS

3.1

DATA TYPES

The data types you may use in FORTRAN source programs are:
1.

integer,

real,

double-precision,

complex,

octal,

d~llble-oct:al, ,

literal t

statement label, and

logical.

The use and format of each of the foregoing data types are
in the descriptions of the constant having the same
(Sections 3.2.1 through 3.2.8).

3.2

discussed
data type

CONSTANTS

Constants are quantities that do not change value during the execution
of the object program.
The constants you may use in FORTRAN are listed in Table 3-1.
Table 3-1
Constants
Category

Constant(s) Types

Numeric

Integer, real, double-precision, complex, and
octal
Logical
Literal
Address of statement lap.~l

Truth Values
Literal Data
S.tatement~. Label

3-1

DATA TYPES, CONSTANTS, SYMBOLIC NAMES, VARIABLES, AND ARRAYS

3.2.1

Integer Constants

An integer constant is a string of from 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 be signed.
You cannot use decimal points, commas, or other symbols on integer
constants (except for a preceding sign, + or -).
Examples of valid
integer constants are:
345
+345
-345
Examples of invalid integer constants are:
+345.
3,450
34.5

3.2.2

(use of decimal point)
(use of comma)
(use of decimal point;

not a whole number)

Real Constants

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

A basic real constant: a string of decimal digits followed
immediately by a decimal point followed optionally by a
decimal fraction, e.g., 1557.42.

A basic real constant followed immediately by a decimal
integer exponent written in E notation (exponential notation)
form, e.g., 1559.E2.

An integer constant (no decimal point) followed by a decimal
integer exponent written in E notation, e.g., l559E2.

Real constants may be of any size; however, each will be
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 form
cannot be empty
(blank);
it must be either a zero or an integer
constant. The magnitude of the exponent must be greater than -38 and
equal to or less than +38 (i.e., -38<n<+38). The following are
examples of valid real constants.
-98.765
7.0E+0
.7E-3
5E+5
50115.
50.El

(7 • )

(.0007)
(500000.)
(500.)

The following are examples of invalid real constants.
72.6E75
.375E
500

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

3-2

DATA TYPES, CONSTANTS, SYMBOLIC NAMES, VARIABLES, AND ARRAYS
3.2.3

Double-Precision Constants

Constants of this type are similar to real constants written in E
notation form;
the direct differences between these two constants
are:
1.

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

Each
double-precision
locations.

The letter D, instead of E,
is used in
constants to identify a decimal exponent.

constant

occupies

two

storage

double-precision

You must use both the letter D and an exponent
(even of zero)
in
writing a double-precision constant. The exponent need only be signed
if it is negative;
its magnitude must be greater than -38 and equal
to or less than +38
(i.e.,
-38<n~+38).
The range of magnitude
permitted a double-precision constant is 0.14 X 10**(-38)
to 3.4 X
10**(+38)
The following are examples of valid double-precision constants.
7.9D03
7.90+03
7.9D-3
79D03
79DO

(=
(=
(=
(=
(=

7900)
7900)
.0079)
79000)
79)

The following are examples of invalid double-precision constants.
7.9D99
7.9ES

3.2.4

(exponent is too large)
(" E" denotes a single-precision constant)

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) real constant of the pair
represents the real part of the number;
the second real constant
represents the imaginary part of the number.
Both the real and
imaginary parts of a complex constant can be signed.
The real constants that represent the real and imaginary parts of a
complex constant occupy two consecutive storage locations in the
object program.

3-3

DATA TYPES, CONSTANTS, SYMBOLIC NAMES, VARIABLES, AND ARRAYS

3.2.5

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 processor storage location.
;When necessary, standard octal constants are padded with leading zeros
to fill their storage location.
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; zeros are added to fill any unused digits.

~If

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 starting in the low-order
(rightmost) word and proceeds leftwards into the high-order word.
All octal constants must:
1.

be preceded by a double quote (") to identify the
octal, e.g., 11777, and

digits

be signed if negative, but optionally signed if positive.

contain one or more of the digits

a through 7, but not

The following are examples of valid octal constants:
"123456700007
" 1234567000 a7
+"12345
-"7777
"-7777
The following are examples of invalid octal constants:
"12368
7777

(contains an 8)
(no identifying double quotes)

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+"202 400 000
000) has as its result the sum of A with the floating point number
2.0; while the'subexpression (1+'1202 400 000 OOO) has as its result
the sum of I with a large integer.
iOctal constants may not be used 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 either:
. an integer or real variable, only the contents of the high order:
location (leftmost) are used.

3-4

DATA TYPES, CONSTANTS, SYMBOLIC NAMES, VARIABLES, AND ARRAYS
3.2.6

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.
Logical quantities may be operated on in arithmetic and logical
statements.
Only the sign bit of a numeric used in a logical IF
statement is tested to determine if it is true (sign is negative)
or
false (sign is positive) .

3.2.7

Literal Constants

A literal constant may be either of the following:
1.

A string of alphanumeric and/or special characters
within apostrophes, e.g., 'TEST#S'.

contained

A Hollerith literal, which is written as a string of
alphanumeric and/or special characters preceded by nH (e.g.,
nHstring).
In the prefix nH,
the letter n represents a
number
that
specifies the exact number of characters
(including blanks) that follow the letter Hi
the letter H
identifies the literal as a Hollerith literal.
The following
are examples of Hollerith literals:
2HAB
14HLOAD TEST #124
6H#124-A

NOTE
A tab (-I) in a Hollerith literal is counted
character, e.g., 3H -I AB.

one

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

up to ten 7-bit ASCII characters
precision type variables, and

up to five
variables.

7-bit

ASCII

for

characters

complex
for

all

double

other

type

The 7-bit ASCII characters that comprise a literal constant are stored
left-justified
(starting in the high-order word of a 2-word precision
or complex literal) with blanks placed in empty character positions.
Literal constants that occupy more than one variable are stored as
successive variables in the list.
The following example illustrates
how the string of characters
A LITERAL OF MANY CHARACTERS
is stored in a six-element array called A.
DIMENSION A(6)
DATA A/' A LITERAr. OF MANY CHARAC.TERS '/.,<.
3-5

DATA TYPES, CONSTANTS, SYMBOLIC NAMES, VARIABLES, AND ARRAYS
A (1)
A (2)
A (3)
A (4)
A (5)
A (6)

3.2.8

is
is
is
is
is
is

set
set
set
set
set
set

to
to
to
to
to
to

'A LIT',
'ERAL
'OF MA'
'NY CHi
'ARACT'
'ERS

Statement Label Constants

Statement labels are
statement numbers.

numeric

identifiers

that

represent

program

: You write statement label constants as strings of from one to five
: decimal digits, which are preceded by either a dollar sign ($) or an
ampersand (&).
For example, either $11992 or &11992 may be used as a
statement label constant.
<

You use statement label constants only in the argument list of CALL
statements to identify the statement to return to in a multiple RETURN
statement.
(Refer to Chapter 15.)

3.3

SYMBOLIC NAMES

Symbolic names may cq~sist of ~~y ~lphanumer~c .~ombin~tion of from one
to six~ characters. You may use more than six characters I but FORTRAN
will ignore all but .th~ first six. The first character of a symbolic
name must be an alphabetic "6haiacter.
The following are examples of legal symbolic names:
A12345
IAMBIC
ABLE
The following are examples of illegal symbolic names:
iAMBIC
lAB

(symbol used as first character)
(number used as first character)

You use symbolic names to identify specific items of a source program;
Table 3-2 lists these items, together with an example of a symbolic
name and text reference for each.
Table 3-2
Use of Symbolic Names

Symbolic Names
Can Identify
1.
2.
3.
4.
5.
6.

Variables
Arrays
Array elements
Functions
Subroutines
External library
functions
7. COMMON block names

For Example

For a Detailed
Description
See Section

PI, CONST, LIMIT
TAX
TAX (NAME, INCOME)
MYFUNC, VALFUN
CALCSB, SUB2, LOOKUP
SIN, ATAN, COSH

3.4
3.5
3.5.1
15.2
15.5
15.4

DATAR, COMDAT

6.5

3-6

DATA TYPES, CONSTANTS, SYMBOLIC NAMES, VARIABLES, AND ARRAYS
3.4

VARIABLES

A variable is a datum (storage location)
that is identified by a
symbolic name and is not a constant, an array or an array element.
Variables specify values that are assigned to them by
either
arithmetic statements (Chapter 8), DATA statements (Chapter 7), or at
run time via I/O references (Chapter 10). Before you assign a value
to a variable, it is termed an undefined variable, and you should not
reference it except to assign a value to it.
If you reference an undefined variable,
will be obtained.

unknown

value

(garbage)

The value you assign to a variable may 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 lAB=5+B, however, the value
of lAB at a given time will depend on the value of variable B at the
time the statement was last executed.
The type of a variable is the type of the contents of the
it identifies. Variables may be:
1.

integer

real

logical

double-precision, or

complex.

datum

that

You may declare the type of a variable by using either implicit or
explicit type declaration statements (Chapter 6).
However, if you do
not use type declaration statements, FORTRAN assumes the following
convention:
1.

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

Variable names that begin with any letter other than I, J, K,
L, M, or N are normally real variables.

Examples of determining the type of a variable according
foregoing convention are given in the following table:
Variable
ITEMP
OTEMP
KA123
AABLE

3.5

Beginning Letter

the

Assumed Data Type
Integer
Real
Integer
Real

0
K

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 given in
Section 3.3 for writing symbolic names.

3-7

DATA TYPES, CONSTANTS, SYMBOLIC NAMES, VARIABLES, AND ARRAYS
Each datum within an array is called an array element.
As with
variables,
you may assign a value to an array element. Before you
assign a value to an array element it is considered to be undefined;
you should not reference it until you have assigned it a value.
If
you reference an undefined array element, the value of the element
will be unpredictable.
Name each element of an array by using the array name together with a
subscript that describes the position of the element within the array.

3.5.1

Array Element Subscripts

Give the subscript of an array element identifier within parentheses,
as either one subscript quantity or a set of subscript quantities
delimited by commas.
write the parenthesized subscript immediately
after the array name.
The general form of an array element name is AN
(Sl, S2, ... Sn), where AN is the array name and Sl through Sn represent
nnumber of subscript quantities. 'You may use any number of subsriiip~~
: quantities in an element 'name; however, the number used must always
. eq~al ~he' number of dimensions
(Section 3.5.2) specified for the
array.
A subscript can be any compound expression (Chapter 4), for example:
1.

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 used in array subscripts may be of any
type,
but noninteger expresslons
(including complex) are
~onverted to integer when the subscript is evaluated.

A subscript may contain function references
(Chapter 14).
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 quantity given is a
nested 3-level subscript.

Here are examples of valid array element subscripts:
1.

IAB(l,S,3)

ABLE(A)

! 3.

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

'4.

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

3-8

DATA TYPES, CONSTANTS, SYMBOLIC NAMES, VARIABLES, AND ARRAYS
3.5.2

Dimensioning Arrays

You must declare the size (number of elements) of an array in order to
reserve the needed amount of locations in which 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, the following figure represents
a 3-row, 3-column, 2-plane array.

3 ROWS

~~
~C:J

'/,

<?"

3 COLUMNS

10-1058

You specify the size of an array by an array declarator written as a
subscripted array name.
In an array declarator, however, each
subscript quantity is a dimension of the array and must be either an
integer variable or an integer constant.
For example,
declarators.

TABLE(I,J,K)

and

MATRIX

(10,7,3,4)

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 lAB (2,3,4) has 24 elements (2 X 3 X 4
24) .
Dimension arrays only in the specification statements DIMENSION,
COMMON, and type declaration (Chapter 6). Subscripted array names
appearing in any of the foregoing statements are array declaratorsi
subscripted array names appearing in any other statements are always
array element identifiers. In array declarators the position of a
given subscript quantity determines the particular dimension of the
array (e.g., row, column, or plane) that it represents.
The first
three subscript positions specify the number of rows, columns, and
planes that comprise the named arraYi each following subscript given
then specifies a set comprised of n-number (value of the subscript) of
the previously defined sets.

3-9

For

exa~ple:

The Dimension Declarator

Specifies the Array(s)

TAB (2)
TAB (2,2)

TAB (2,2,2)

TAB (2,2,2, 2) r-::.=-=--r----t:;-;-:;~;_:;_:;t1

NOTE
FORTRAN-20
permits
any
number
dimensions in an array declarator.

3.5.3

Order of Stored Array Elements

The elements of an array are arranged in storage in ascending order.
The value of the first subscript quantity varies between its m1n1mum
and maximum values most rapidly.
The value of the last given
subscript quantity 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)

(2,2)

(1,3)

(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.

B (1,1,1)

Thus B(3,1,1)

B (2,1,1)

B (3,1,1) - ,

is stored before B(1,2,1), and so forth.

3-10

CHAPTER 4
EXPRESSIONS

4.1

ARITHMETIC EXPRESSIONS

Arithmetic expressions may be either simple or compound.
arithmetic expressions consist of an operand that may be:

Simple

a constant

a variable

an array element

a function reference (see Chapter 14 for description), or

an arithmetic
parentheses.

logical

expression

written

within

Operands may be of integer, real, double precision, complex, ~9£!;'~:l<L .9£1

JjJ:j~ru~lj type.

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

(integer constant)
(integer variable)
(array element)
(function reference)
(a parenthetical expression)

A compound arithmetic expression consists of two or more operands
combined by arithmetic operators.
Table 4-1 lists the arithmetic
operations permitted in FORTRAN and the operator recognized for each.
Table 4-1
Arithmetic Operations and Operators
Operation

Operator
''>'., ,

1. Exponentiation
2. Multiplication
3. Division
4. Addition
5. Subtraction

**~or
~ v> .~~ A

*
/
+

4-1

Example
A**B o r "~~"""_~~~_~uj
A*B
A/B
A+B
A-B

EXPRESSIONS
4.1.1

Rules for Writing Arithmetic Expressions

Observe the
expressions:
1.

following

rules

structuring

compound

arithmetic

The operands compr1s1ng a compound arithmetic expression may
be of different types. Table 4-2 illustrates all permitted
combinations of data types and the type assigned to the
result of each.
NOTE
one combination of data types, double-precision
complex, is prohibited in FORTRAN-20.

An expression cannot contain
operators.
For
example,
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 obtain a
multiplication of the elements.

When you use exponentiation, the base quantity and its
exponent may be of different types.
For example, the
expression ABC** 13 involves a real base and an integer
exponent.
The permitted base/exponent type combinations and
the type of the result of each combination are given in Table
4-3.

4-2

two adjacent and unseparated
the expression A*/B is not

Table 4-2
Type of the Result Obtained From Mixed Mode Operations
Type of Argument 2
For operator;

+. -. *. /
I. Type of operation
used
2. Type associated
with result
3. Conversion on
Argument I

Integer

Real

Double
Precision

Real

Complex

Logical

Octal

Double Octal

1. Integer

I. Real

1. Douhle Precision

I. ~'Ol1lplex

Integer

I. Integer

2. Integer

2. Double Precision

_. Complex

Integer

3. None

3. From Integer to

3. From lnte~er to

3. None

3. l\one

Double Precision

::omplex. Value
used as Real part
4. ','one

..\. None

4. t\onc

4. lIigh order word
is used dirc('Uy:
lov; order word
is ignored.

4. !11;:h orJ ~r word
is u:;ed directly;
further \lords
::::.re i;nor~Ll.

Real

I. Real

L Real

1. Real

Real

4. Conversion on
Argument 2

4. None

4, None

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

I. Real

I. Double PreCision

Real

J. None

Real

4. None

Double Precision

3. None

3. Used directly as

4. None

the high order
word: low order
word is zero.
4. None

I. Complex
'1

('ompJe~

3. Used directly as
the Real part:
,OlaVnar,. part
is zero.
4. None

1. Intc~cr

Integer

I. Ir.:c;;:r
_. Integer

:--one

Snnc

2. Real
~one

J. None

3. \"onc

4. ~onc

4. \"on"

... lIi;:h order word
is used directly;
10"" order word
is ignored.

4. llj;~ ardor word
is lli('d directly;
further v.ords
are 19nored.

4. Conversion on
Argument 2

4. From Integer to

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

I. Double Precision

I. Double l'recision

I. Double Precision

1. Douhle Precision

I. Double PreCISion

I. Double Precision

I. Double I'recision

., Double Precision

Double Precision

2. Double Precision

2. Double PreCISion

Double Precision

2. Double Precision

2. Double !'recision

I. Type of operation

Complex

Integer

used
2. Type associated
with result
3. Conversion on
Argument I
4. Conversion on
Argument 2

Real

3. None

4. From Integer to
Double Precision

4. Used directly 'as
the hit:h order
,vend ~ low order
\'rard is zero.

4. None

4. Used directly as

4. Used directly as
the Jllgh order
v.onl: low order

4. None

4. First two words
are used directly;
furth':! lNords
:lIe i::;ncred.

I. Complex

the high crder
word: low order
word is zero.

I. Complex

1. Complex

I. Complex

I. Compkx

I: Complex

I. Complex

C. Complex

_. Complex

2. Complex

_. Complex

3. None

3. 1\one

3. None

3. t"one

3. None

3. NOM

3. None

4. From Integer to

4. Used diredly as

4. None

4. Used d [rectlr as

4. Used d itectly as
the Real rart:
imaginary part

4. None

4. Firsl two "iords
are U"ed dir ectly.
Further words
are ignored.

Complex. Value
used as Real part.

the Real part:
unar-mary pari

the Real port;

i mag;~..!iY part

I. Into~er

2. Type associated

Intq~er

Double Precision

Complex

lero

Ie; 7l.'fO.

I. Complex

I. Integer

!. Integer

I. Integer

I. Integor

2. Octal

1")

I. Type of operation
used

word I~: leTO.

2. Rf:tl

_. [)l)uble PreciSion

3. None

Complex

J. lsed directly as

J. Used directly as

3. None

4. None

the high order
word; low order
word is zero.
1'>'one

the R cal part;
imal'inary part
is zero
4. None

4. None

Octal

with result

3. Conversion on

3. None

Argument I
Logical

Octal

4. Conversion on
Argument 2

4. None

L Typ~ of operation

I. Integer

I. I{eal

2. Integer

3. None

J. Nont'

4. None

used
, Type associated
with result
3. Conversion on
Argument I

4. Conversion on
Argument 2

Non\?

Real

.t.

i\on~

3. None

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

4. High order word

Double Precision

1. Complex

I. Integer

1. Integer

Double Precis,on

Complex

Octal

_. Octal

2. Octal

Used dire:tI>' 35
the high order
word: low order
word is zero.
4. l\one

.~.

Used directly 3'
the Real part:
ima,:inary part

3. None

4. None

4. :-';one

4. High order word
is used directly;
low order ,,"ord

4. High order word
is used directly;
further words
are i~norcd.

tS-ll'Tu

4. NOM

is ignored.

I. Type of operation
,
Double
Oclal

used
Type associated
with result
Con\"erslon on
Argument I

4. Conversion on
Argument 2

I. Integer

I. Real

I. Double Precision

I. Complex

I. Integer

I. Inter,er

L Integer

2. Integer

..,

Real

2. Double Precision

:!. Complex

2. Octal

lIi,h order "mel
is used directly;
low order word

3. :\ono

3. Sone

4. None

3. High order word
is used directly;
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;

3. .1i~h mUe r worol
is u,cd directly:
lo,w order worlh

4. None

3. lIi[t." order word
is used diredly:
low order word
is ignored.
4. None

3. lIigh order word
is used directly:
low ord er word

is ignored.
4. None

Octal

low ordIT word
is ignored.

J. Type of operation

Literal

used
, Type associated
with result
3. Conversion on
Argument 1

Conversion on
Argument 2
~

is used directly;
ftuther words
are ignored.

He it,IHHt:U.

4. rlii'h 'order word
is lIsed directly;
Ill'" ,,,dcr worJ,
'fC

L:llor;'?d.

I. Integer

I. Real

I. Double Precision

1. Complex

I. Integer

2. Integer

_. Real

2. Double Precisi,'n

2. Oelal

2. Octal

3. High order word

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

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

3. First two words
are used directly;
further words
are ignored.
4. t"one

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

3. lIigh order wortl

is used directly;
further words
ar e ignored.
4. None

3. lIig:h order word
is used Girectly;
further words
arc ignored.
4. High order word
is used directly;
further words'
are ignored.

Complex

Octal

3. High order word

is used directly;
further words

are ignored.
4. None

is used directly:
further words
are ignored_
4. lIigh order word
is used directly:
low order word
is ignored.

Octal

EXPRESSIONS
Table 4-3
Permitted Base/Exponent Type Combinations
Base Operand

Integer
Real
Double
Precision
Complex

4.2

Exponent Operand
Integer

Real

Integer

Real

Double
Precision

Complex

Double
Precision
Real
Real
Double
Precision
Double
Double
Double
Precision Precision Precision
Complex
Complex
(Prohibited)

Complex
Complex
(Prohibited)
Complex

LOGICAL EXPRESSIONS

Logical expressions may be either simple or compound. Simple logical
expressions consist of a logical operand, which may be a logical type:
1.

constant

variable

array element

function reference (see Chapter 15), or

another expression written within parentheses.

Compound logical expressions consist of two or more operands
by logical operators.

combined

Table 4-4 gives the logical
operation each provides.

operators

4-4

and

description

the

EXPRESSIONS
Table 4-4
Logical Operators
Operator

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.

• XOR.

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

.EQV.

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

.NOT.

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

write logical expressions in the general form P .OP.
Q, where P and
Q are logical operand and .OP.
is any logical operator but ".NOT.II.
The .NOT.
operator complements the value of a logical operand;
you
must write it immediately before the operand that it modifies, e.g.,
.NOT.P. Table 4-5 is a truth table illustrating all possibie logical
combinations of two logical operands (P and Q) and the resultant of
each combination.
When an operand of a logical expression is double-precision or
complex, only the high-order word of the operand is used in the
specified logical operation.
The assignment of a .TRUE. or a .FALSE. value to a given operand is
based only on the sign of the numeric representation of the operand.

4-5

EXPRESSIONS
Table 4-5
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
,,"

Then the Expression:

P .OR. Q

False
.

,. ,

Is:

'>' .. ,

'v ,

False
y,,-

'" ~

True

P .XOR. Q

False

True

False

P .XOR. Q

True

False

True

P . XOR. Q

True

False

P .XOR. Q

False

True

P .EQV. Q

True

False

P .EQV. Q

False

True

P. EQV. Q

False

P .EQV. Q

True

Examples
Assume the following variables:
variable
REAL, RUN
I,J,K
DP,D
L, A, B

CPX,C
Examples of valid logical
variables are:

Type
Real
Integer
Double Precision
Logical
Complex
expressions

L.AND.B

(RE"ACkI)"~"x6R'~"(Dp'+kf

L.AND'.A.OR .. NOT. (I"":K)

4-6

consisting

the

foregoing

EXPRESSIONS
Logical functions are performed on the full 36-bit binary processor
representation of the operands involved.
The result of a logical
operation
is
found
by
performing
the
specified
function,
simultaneously, for each of the corresponding bits in each operand.
For example, consider the expression A=C.OR.D, where C=1I456 and
D=1I20l. The operation performed by the processor and the result are:
Word
Bits
o
Operand C 0
Operand D 0
Result A o

1 ---t.~ 24

o
o
o

...
t:o
t:o

25 26

0
0
0

o
o

0
0
0

27
1

1
1

o
o

o
o
o

o
1

34 35
1 0

1
1

Table 4-5 also illustrates all possible logical combinations of two
one-bit binary operands (P and Q) and gives the result of each
combination. Just read 1 for true and 0 for false.

4.2.1

Relational Expressions

Relational expressions consist of two expressions combined by a
relational operator.
The relational operator permits the programmer
to test, quantitatively, the relationship between two arithmetic
expressions.
The result of a relational expression is always a
false value.

logically

true

In FORTRAN-20, you may write relational operators either as a 2-letter
.mnemonic enclosed. within ,periods, e.g., .GT., or _sy~bplicallyusing;
.thesymbols, ~, <, =_a~d #~ Table 4-6 lists both the mnemonic and
symbolic forms of the relational operators and specifies the type of
quantitative test performed by each operator.
Table 4-6
Relational Operators and Operations
Relation Tested

Operators
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

4-7

EXPRESSIONS
Write relational expressions in the general form A(l) .OP.A(2), where
A represents an arithmetic operand and .oP. is a relational operator.
You may mix arithmetic operands of type
precision in relational expressions.

integer,

real,

and

double

You may compare complex operands using only the operators .EQ.
(==)
and .NE.
(#) .
Complex quantities are equal if the corresponding
parts of both words are equal.
Examples
Assume the following variables:
Variables

Type

REAL, RON
I,J,K
DP,D
L,A,B
CPX,C

Real
Integer
Double Precision
Logical
Complex

Examples of valid relational expressions consisting of
variables are:

the

foregoing

( REAL) . GT . 10
I == 5
C.EQ.CPX
Examples of invalid relational expressions consisting of the foregoing
variables are:
(REAL) .GT 10 (closing period missing from operator)
C>CPX

(complex operands can only be
.NE. operators)

combined

.EQ.

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

and

relational

(I.GT.IO).AND·t(J<=K);
«I*RON):;;~:(I/J»

.OR*."K

(I • AND. Kf#', ( (REAL) . OR. (RON) )
C:#CPX.OR.'R'ON

4-8

and

EXPRESSIONS
4.3

EVALUATION OF EXPRESSIONS

The following determine the order of computation of an expression:
1.

the use of parentheses

an established hierarchy for the execution
relational, and logical operations and

the location of operators within an expression.

4.3.1

arithmetic,

Parenthetical Subexpressions

In an expression, all subexpressions written 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 by a parenthetical subexpression, the required computations
are performed according to the hierarchy of assigned operators
(Section 4.3.2).
Example:
The separate computations performed in evaluating the expression
A+B/((A/B)+C)-C are:
1.

Rl=A/B

2=Rl+C

R3=B/R2

R4=R3-C

RS=A+R4

where:

Rl through RS represent the interim and final results
computations performed.

4.3.2

Hierarchy of Operators

The following hierarchy
(order
classes of FORTRAN operators:

execution)

first,
arithmetic operators,
second, relational operators, and
third, logical operators.

4-9

assigned

the

EXPRESSIONS
Table 4-7 specifies the precedence
operators of the foregoing classes.

assigned

the

individual

With the exception of integer division and exponentiation,
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 4-7
Hierarchy of Operators
Symbol or Mnemonic

Class

Level

EXPONENTIAL

First

ARITHMETIC

Second
Third
Fourth

-(unary minus) and + (unary plus)
*,/
+,-

RELATIONAL

Fifth

~~T.,~GE,.".LT~".LE., .EQ., .NE.

"> , >= , < ,~= , == , #, .
LOGICAL

Sixth
Seventh
Eighth
'Nint'h

,~,

,."'>"'

"'-

.NOT.
.AND.
.OR .

• EQV .. 1. ~X6if.

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} ,which is evaluated as written here.
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).
Similarly, here too,
parentheses alter
the evaluation
expression. (A**2}**B is evaluated in these two steps:
first Rl = A**2 (intermediate result)
second R2 = Rl**2 (final result)

4.3.3

the

Mixed Mode Expressions

Mixed
mode
expressions
are
evaluated
on
a
subexpression-by-subexpression basis, with the type of the results
obtained converted and combined with other results or terms according
: to the conversion procedures described in Table 4-2.

4-10

EXPRESSIONS
Example
Assume the following:
Variable
D
X
I,J

Type
Double-Precision
Real
Integer

The mixed mode expression D+X* (I/J) is
manner:
1.Rl

I/J

2.R2 =X*Rl
3.R3

D+R2

evaluated

the

following

Rl is integer
Rl is converted to type real and is multiplied
to produce R2

R2 is converted to type double precision and is added
to D to produce R3

where

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

4.3.4

Use of Logical Operands in Mixed Mode Expressions

final

results

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

4-11

CHAPTER 5
COMPILATION CONTROL STATEMENTS

5.1

INTRODUCTION

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 descr ibed in this chapter are: . PROGRAM . state.ment I.
,"IN.CLUPE. statem~l'}t,: and END statement.

5.2

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.
(Refer to Section 3.3 for a description of symbolic
names. )

: The following rule governs the use of the PROGRAM statement:
The PROGRAM statement must be the first statement in a program
unit.
(Refer to Section 2.4 for a discussion of the ordering of
statements.)

5.3

INCLUDE STATEMENT

This statement allows you to include code segments or predefined
· declarations in a program unit without having them reside in the same
i physical
file as the primary program unit. The general form of the
INCLUDE statement is
i

INCLUDE 'dev:filename.ext[proj,prog]/NOLIST'
where:
dev:

is a device name.
When no
specified, DSK: is assumed.

5-1

device

name

COMPILATION CONTROL STATEMENTS
filename.ext

is the filename and extension of the statements
that you wish to include. The name of the file is
required; the extension is optional.
If you
specify
filename" only, .FOR is the assumed
extension. If you specify the filename and period
(filename.), the null extension is assumed. You
may not specify wild (*) information.
II

[proj , pr og]

is the project-programmer number.
Your projectprogrammer number is assumed if none is specified.
You cannot specify subdirectory information.

/NOLIST

is an optional switch indicating that the included
statements
are
not
to be included in the
compilation listing.

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

The INCLUDEd file may contain any legal statement except;
another INCLUDE statement, or a statement that terminates the
current program unit, such as the END, PROGRAM, FUNCTION,
SUBROUTINE, or BLOCK DATA statements.

The proper placement of the INCLUDE statement within a
program unit depends upon the types of statements to be
INCLUDEd.
(Refer to Section 2.4 for information on the.
ordering of FORTRAN statements.)

The file(s) 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, provided /NOLIST is.
; ~ot specified.

5.4

END STATEMENT

Use this statement to show the physical end of
subprogram. END is a nonexecutable statement.
END statement is:

a source program or
The general form of an

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

This statement must be the last
source program or subprogram.

:2.

When used in a main program, the END statement implies a STOP
statement operation;
in a subprogram, END implies a RETURN
statement operation.

You may label an END statement.

5-2

physical

statement

CHAPTER 6
SPECIFICATION STATEMENTS

6.1

INTRODUCTION

Use specification statements to specify the type characteristics,
storage allocations, and data arrangement. There are seven types of
specification statements:
1.

DIMENSION

Statements that explicitly specify
INTEGER

IMPLICI'r

COMMON

EQUIVALENCE

EXTERNAL

7•

PARAMETER:

type,

such

REAL

Specification statements are nonexecutable and conform to the ordering
guidelines described in Section 2.4.

6.2

DIMENSION STATEMENT

DIMENSION statements identify and allocate the space required for
source program 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 Sl, S2, .•. ,Sn
where Si is an array declarator.
following form:
name(max, ... ,max)

Array declarators are names

the

name (min :m~x ' . ' .. ,min : max);

where name is the symbolic name of the array, and each min:max
represents the lower and, upper bounds of an array dimension.

6-1

value

SPECIFICATION STATEMENTS
min:max value for an array dimension may be either an integer
constant or, if the array is a dummy argument to a subprogram, an
integer variable. The value given the mlnlmum specification for a
dimension must not exceed the value given the maximum specification.
Minimum values of 1 with their following colon delimiters may be
omitted from a dimension subscript. This is because minimum values
are assumed to be ~ inth~ f~rst place.
Examples
DIMENSION EDGE (':"1:1,"4:8), NET (5,10,4), TABLE (567)
DIMENSION TABLE (lAB:J,K,M,XO£tO)
(where lAB, J, K, and M are of type integer).
~ ..O'

"«

Note that you may use a slash in place of a colon as the
between the upper and lower bounds of an array dimension.

delimiter

Adjustable Dimensions
When used within a subprogram, an array declarator may use type
integer parameters· as dimension subscript quantities. The £ollowing
rules govern the use of adjustable dimensions in a subprogram:
1.

For single entry subprograms, the array name and each
subscript variable must be given by the calling program as
parameters when the subprogram is called.
The subscript
variables may also be in COMMON.

For multiple entry subprograms in which the array name is a
parameter, any subscript variables may be passed. If all
subscript variables are not passed or in COMMON, the value of
the subscript as passed for a previous entry will be used.

The type of the arra~ dimension variables cannot
within the program.

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

The original size of the array cannot exceed the array
dimensions assigned within a subprogram, i.e., the size of an
array is not dynamically expandable.

Examples

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
SUBROUTINE SBI (ARRl,M,N)
DIMENSION ARRl(M,N)
ARRl(M,N)=VALUE
ENTRY SB2(ARRl,M)
ENTRY SB3(ARRl,N)
ENTRY SB4JARRl)
>

6-2

altered

SPECIFICATION STATEMENTS
In the foregoing c~~mplc, the first call made to the subroutine must
be made to SBI. Assuming that the call is made at SBI with the values
M=ll and N=13, any succeeding call to SB2 should give M a new value.
If a succeeding call is made to SB4, the last values passed through
entries SBl, SB2, or SB3 will be used for M and N.
Note that for the calling program of the form:
CALL SBl(A,ll,13)
M=15
CALL SB3(A,13)
the value of M used in the dimensionality of the
execution of SB3 will be 11 (the last value passed).

6.3

array

for

the

TYPE SPECIFICATION STATEMENTS

Type specification statements declare explicitly the data type of
variable, array, or function symbolic names. You may give an array
name in a type statement either alone (unsubscripted) to declare the
type of all its elements or in a subscripted form to specify both its
type and dimensions.
write type specification statements in the following form:
type list
where type may be anyone of the following declarators:
1.

INTEGER

REAL

DOUBLE PRECISION

COMPLEX

LOGICAL

NOTE
In order to be compatible with the type
statements used by other manufacturers,
the data type size modifier, *n, is
accepted by FORTRAN-20. 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-20:

6-3

SPECIFICATION STATEMENTS
Declarator

Form of Variable Specified

INTEGER*2 Full word integer with warning message
INTEGER*4 Full word integer
LOGICAL*l Full word logical with warning message
LOGICAL*4 Full word logical
REAL*4
Full word real
REAL*8
Double-precision real
COMPLEX*8 Complex
COMPLEX*16 Complex with warning message
In addition, you may 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(10), D
A and D are single-precision (one full
word)
real,
and
Band
Care
double-precision (two full words) real.
The list consists of any number of variable, array, or function names
that are 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 (5:10,10:20,5)
NOTE
Variables, arrays, and functions of a
source program, which are not typed
either implicitly or explicitly by a
specification statement, are typed by
the following conventions:
1.

Variable names, array names, and
function names that begin with the
letters I, J, K, L, M, or N are type
integer.

Variable names, array names, and
function names that begin with any
letter other than I, J, K, L, M, or
N are type real.

If a name that is the same as a predefined FORTRAN-20 function name
appears in a conflicting type statement, it is assumed that the name
refers to a user-defined routine of the given type. If you place a
generic function name in an explicit type statement, it loses its
generic properties.

6-4

SPECIFICATION STATEMENTS
6.4

IMPLICIT STATHMHNTS

IMPLICIT statements declare the data type of variables and functions
IMPLICIT
according to the first letter of each variable name.
statements are written in the following form:
IMPLICIT type (Al,A2, ... ,An), type (Bl,B2, ... ,Bn), ... ,type .....
As shown in the foregoing
form statement, an IMPLICIT statement
consists of one or more type declarators separated by commas.
Each
type declarator has the form
type (Al,A2, ... ,An)
where type represents one of the declarators listed in Section 6.3,
and the parenthetical list represents a list of different letters.
Each letter in a type declarator list specifies that each source
program variable
(not declared in an explicit type specification
statement) starting with that letter is assigned the data type named
in the declarator.
For example, the IMPLICIT type declarator REAL
(R,M,N,O) declares that all names that begin with the letters R, M, N,
or a 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 the
FORTRAN supplied functions.
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, e.g.,
A-E for A,B,C,D,E.
For example,
the statement IMPLICIT INTEGER
(I,L-P)
declares that all variables which begin with the letters
I,L,M,N,O, and P are INTEGER variables.
You may use more than one IMPLICIT statement, but they must appear
before any other declaration statement in the program unit.
Refer to
Section 2.4 for a discussion on ordering FORTRAN statements.

6.5

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 implicitly
transfer arguments between a calling program and a subprogram.
Write
COMMON statements in the following form:
COMMON/Al/Vl,V2, ... ,Vn ... /An/Vl,V2, ... ,Vn
where the enclosed letters /Al/, ... , /An/ represent optional
constructs (referred to as common block names when used).

6-5

name

SPECIFICATION STATEMENTS
The list (e.g., VI,V2 ... ,Vn) appearing after each name construct lists
the names of the variables and arrays that are to occupy the common
area identified by the construct. The items specified for a common
area are ordered within the storage area as they are listed in the
COMMON statement.
Either label COMMON storage areas or leave them blank (unlabeled).
If
the common area is to be labeled, give a symbolic name within slashes
immediately before the list of items that is to occupy the names area.
For example, the statement
COMMON/AREAI/A,B,C/AREA2/TAB(13,3,3)
establishes two labeled common areas (i.e., AREAl and AREA2).
Common
block names bear no relation to internal variables or arrays that have
the same name.
If a common area is to be declared as unlabeled, give either nothing
or two sequential slashes (//) immediately before the list of items
that is to occupy blank common.
For example, the statement
COMMON/AREAI/A,B,C//TAB(3,3,3)
establishes one labeled
(AREAl)
and one unlabeled
Unlabeled common area is also called "blank common".

common

area.

A given labeled common name may appear more than once in the same
COMMON statement and in more than one COMMON statement within the same
program or subprogram.
Each labeled common area is treated as a separate, specific storage
area.
The contents of a common area, i.e., variables and arrays, may
be assigned initial values by DATA statements in
BLOCK
DATA
subprograms.
Declarations of a given common area in different
subprograms must contain the same number, size, and order of variables
and arrays as the reference area.
Items to be placed in a blank common area may also be given in
statements throughout the source program.

COMMON

During compilation of a source program, FORTRAN will string together
all items listed for each labeled common area and for blank common
areas in the order in which they appear in the source program
statements.
For example, the series of source program statements:
COMMON/STI/A,B,C/ST2/TAB(2,2)//C,D,E
COMMON/STI/TST(3,4)//M,N
COMMON/ST2/X,y,Z//O,P,Q
has the same effect as the single statement
COMMON/STI/A,B,C,TST(3,4)/ST2/TAB(2,2) ,X,Y,Z//C,D,E,M,N,O,P,Q
All items specified for blank common are placed into one area.
Items
within blank common are ordered as they are given throughout the
source program. Common block names must be unique with respect to all
subroutine, function, and entry point names.
The largest definition of a given common area must be loaded first.

6-6

SPECIFICATION STATEMENTS
6.5.1

Dimensioning Arrays in COMMON Statements

Subscripted array names may be given in COMMON statements as array
dimension declarators.
However, variables cannot be used as subscript
quantities in a declarator appearing in a COMMON statement;
variable
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.
Example
COMMON /A/B(lOO) , C(lO,lO)
COMMON X ( 5 , 15) ,Y ( 5 )

6.6

EQUIVALENCE STATEMENT

The EQUIVALENCE statement enables you to control the allocation of
shared storage within a program or subprogram. This statement causes
specific storage locations to be shared by two or more variables of
either the same or different types. Write the EQUIVALENCE statement
in the following form:
EQUIVALENCE (Vl,V2, ... ,Vn) , (Wl,W2, ... ,Wn), (Xl,X2, ... ,Xn)
where each parenthetical list contains the names of variables
array elements that are to share the same storage locations.
example, the statements

and
For

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

for example,

the

two

EQUIVALENCE (A,B), (B,C)
EQUIVALENCE (A,B,C)
When you use array elements in EQUIVALENCE statements, they must have
either as many subscript quantities as dimensions of the array or only
one subscript quantity.
In either of the foregoing cases,
the
subscripts must be integer constants.
Note that the single case
treats the array as a one-dimensional array of the given type.
You may use the items given in
EQUIVALENCE
statement and in
following rules are observed:
1.

EQUIVALENCE list in both
COMMON statement providing

You cannot set two quantities declared in a COMMON
to be equivalent to one another.

6-7

the
the

statement

SPECIFICATION STATEMENTS
2.

Quantities placed in a common area by means of an EQUIVALENCE
statement are permitted to extend the end of the common area
forwards.
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
follows:

x
Y A (I)
Z A (2)
A (3)

(shared location)
(shared location)

A(4)
3.

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 by the
following diagram:

A (I)
A (2)

X A(3)
Y A (4)

6.7

EXTERNAL STATEMENT

Any 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. Write the
EXTERNAL statement in the following form:
EXTERNAL namel,name2, ... ,namen
where each name listed is declared to be a subprogram name.
desired, these subprogram names may be FORTRAN defined functions.

You may also use FORTRAN defined function names for your subprograms
by prefixing the names by an asterisk (*) or an ampersand (&) within
an EXTERNAL statement.
For example,
EXTERNAL *SIN, &COS

6-8

SPECIFICATION STATEMENTS
declares SIN and COS to be user subprograms.
(If a prefixed
not a FORTRAN defined function, then the prefix is ignored.)

name

Note that specifying a predefined FORTRAN function
in an EXTERNAL
statement without a prefix, i.e., 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 of 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
other declarator statement, with the exception of a type statement.

6.8

PARAMETER STATEMENT

The PARAMETER statement allows you to
during compilation.

define

constants

symbolically

The general form of the PARAMETER Statement is as follows:
PARAMETER

PI=Cl,P2=C2, ...

where
pi

is a standard user-defined identifier
section as a parameter name)

(referred to

this

is any type of constant (including literals) except a label
or complex constant.
(Refer to Chapter 3 for a description
of FORTRAN constants.)

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

Place parameter names only within the statement field of an
initial or continuation line type, i.e., not within a comment
line or literal text.

Place parameter names only where constants are acceptable.

Place parameter name references after the PARAMETER statement
definition.

Use parameter names that are unique with respect to all other
names in the program unit.

Do not redefine
statements.

Do not use parameter names as part of some larger syntactical
construct
(such as a Hollerith constant count or a data type
size modifier).

parameter

6-9

names

subsequent

PARAMETER

CHAPTER 7
DATA STATEMENT

7.1

INTRODUCTION

DATA statements are used to supply the initial values of variables,
arrays, array elements, and labeled common. (1) write DATA statements
as follows:
DATA Listl/Datal/,List2/Data2/, ... ,Listn/Datan/
where the List portion of each List/Data/ pair identifies a set of
items to be initialized and the /Data/ portion contains the list of
values to be assigned the items in the List.
For example, the
statement
DATA IA/S/,IB/IO/,IC/1S/
initializes variable IA to the value S, variable IB to the value 10,
and the variable IC to the value IS. The number of storage locations
you specify in the list of variables must be less than or equal to the
number of storage locations you specify in its associated list of
values.
If the list of variables is larger
(specifies more storage
locations)
than its associated value list, a warning message is
output. When the value list specifies more storage locations than the
variable list, the excess values are ignored.
The List portion of each List/Data/ set may contain the names of one
or more variables, array names, array elements, or labeled common
variables.
You may specify an entire array (unsubscripted array name)
or a portion of an array in a DATA statement List as an implied DO
loop construct.
(See Section 10.3.4.1 for a description of implied DO
loops.) For example, the statement
DATA (NARY(I),I=1,S)/1,2,3,4,S/
initializes the first five elements of array
. NARY(2)=2, NARY(3)=3, NARY(4)=4, NARY(S)=S.

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 loop Initial, Terminal, and
Increment parameters must also be of type INTEGER.
In a DATA
statement,
references to an array element must be integer expressions
in which all terms are either integer constants or indices of
embracing implied DO loops.
Integer expressions of the foregoing
types cannot include the exponentiation operator.
The /Data/ portion of each List/Data/ set may contain one or more
numeric,
logical, ,literal, or octal constants and/or alphanumeric
strings.
1.

Refer t6 Section 6.S for a description of labeled common.
7-1

DATA STATEMENT
~ You must identify octal constants by
; quote (n) symbol, e.g, "777.

preceding

them

with

double

You may specify literal data as either a Hollerith specification,
e.g.,
SHABCDE, or a string enclosed in single quotes, e.g., 'ABCDE'.
Each ASCII datum is stored left-justified and is padded with blanks up
to the right boundary of the variable being initialized.
When you assign the same value to more than one item in List, a repeat
specification may be used.
Write the repeat specification as N*D
where N is an integer that specifies how many times the value of item
D is to be used.
For example, a /Data/ 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-20 converts the datum to the type of
the variable. The type conversion is performed using the rules given
for
type conversion in arithmetic assignments.
(Refer to Chapter 8,
Table 8-1.)
Octal, logical, and literal constants are not converted.
Use

Sample Statement

DATA PRINT,I,O/'TEST' ,30,:1177/, (TAB(J) ,J=1,30)/30*S/ The
first
30
elements of array
TAB
are
initialized
to

S.O.
DATA((A(I,J),I=1,S),J=l,6)/30*1.0/

No
conversion
required.

DATA((A(I,J),I=S,10),J=6,lS)/60*2.0/

No
conversion
required.

When a literal string is specified that is longer
than one variable
can hold, the string will be stored left-justified across as many
variables as are needed to hold it.
If necessary, the last variable
used will be padded with blanks up to its right boundary.
Example
Assuming that X, Y, and Z are single-precision, the statement
DATA X,Y,Z/'ABCDEFGHIJKL'/
will cause
X to be initialized to 'ABCDE'
Y to be initialized to 'FGHIJ'
Z to be initialized to 'KL~~~'
When a literal 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.

7-2

DATA STATEMENT

Example
Assuming that the variable C is complex, the statement
DATA C/'ABCDE ' , 'FGHIJ'/
will cause the first word of C to be initialized to 'ABCDE ' and its
second word to be initialized to I~~~~~I. The string 'FGHIJ ' is
ignored.

7-3

CHAPTER 8
ASSIGNMENT STATEMENTS

8.1

INTRODUCTION

Use assignment statements to assign a specific value to one or more
program variables. There are three kinds of assignment statements:

8.2

Arithmetic assignment statements

Logical assignment statements

Statement Label assignment (ASSIGN) statements.

ARITHMETIC ASSIGNMENT STATEMENT

You use statements of this type to assign specific numeric
variables
and/or
array
elements.
Write arithmetic
statements in the form

values to
assignment

v=e
where v is the name of the variable or array element that is to
receive the specified value and e is a simple or compound 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 correctly interpreted as "the current
contents of the location identified as v are to be replaced by the
final value of expression e; the current contents of v are lost."
When the type of the specified variable or array element name differs'
from that of its assigned value, FORTRAN-20 converts the value to the
type of its assigned variable or array element. Table 8-1 describes 1
the type conversion operations performed by FORTRAN-20 for each
possible combination of variable and value types.

8-1

Table 8-1
Rules for Conversion in Mixed Mode Assignments
Variable Type (v)

Expression Type (e)
Real

Integer

Complex

Double-Precision

Logical

REAL

R,I

H,L

INTEGER

R,C,I

H/C/L

COMPLEX

C,R

prohibited

til
til
H

Cil

ex>
I

DOUBLEPRECISION

C,H,L

prohibited

§!
tzl

til

t\.)

LOGICAL

R,I

H,L

D,H

:;;
8

tzl

3:
tzl

OCTAL

R,I

H,C,L

LITERAL

D,H%

C,H%

DOUBLE
OCTAL*

til

Table 8-1 (Cont.)
Rules for Conversion in Mixed Mode Assignments
Legend
D
Direct replacement
C = Conversion between integer and floating-point with truncation
R
Real part only
I
Set imaginary part to 0
H
High-order only
L
Set low-order part to 0

:J:>I
til
til
H
G)

::::
tzl

Notes

(X)

1-3

til

1-3

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.

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

Use the first word of the literal.

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.

only

1-3

tzl

:::
tzl

1-3

ASSIGNMENT STATEMENTS
8.3

LOGICAL ASSIGNMENT STATEMENTS

Use this type of assignment statement to assign values to variables
and array elements of type logical. Write the logical assignment
statement in the form
v=e
where v is one or more variables and/or array element names, and e
a logical expression.

Examples
Assuming that the variables L, F, M, and G are of
following statements are valid:

type

logical,

the

Sample Statement
L=.TRUE.

The contents of L is replaced by logical
truth.

F=.NOT.G

The contents of L is replaced by the
logical complement of the contents of G.

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

If A is greater than T, the contents of
M is replaced by logical truth: if A is
less than or equal to T, the contents of
M is replaced by logical false. This
can also be read: If A is greater than
T, then M is true, otherwise, M is
false.

L= ( (I. GT. H)' • AND. (J<=K) )

The contents of L are replaced by either
the true or false resultant of the
expression.

8.4

ASSIGN (STATEMENT LABEL) ASSIGNMENT STATEMENT

Use the ASSIGN statement to assign a statement label constant, i.e., a
1- to 5-digit statement number, to a variable name. Write the ASSIGN
statement in the form
ASSIGN n TO I
where n represents the statement number and I is a variable name.
example, the statement

For

ASSIGN 2000 TO LABEL
specifies that the variable
2000.

LABEL

represents

the.

statement

with the exception of complex and double-precision, you
type of variable in an ASSIGN statement.

may

number
use

any

Once a variable has been assigned a statement number, FORTRAN-20 will
consider it a label variable.
If a label variable is used in an
arithmetic statement, the result will be unpredictable.

8-4

ASSIGNMENT STATEMENTS
Use the ASSIGN statement in conjunction with assigned GO TO control
statements
(Chapter 9).
The ASSIGN verb sets up statement label
variables that are then referenced in subsequent GO TO control
statements.
The following sequence illustrates the use of the ASSIGN
statement:
555 TAX=(A+B+C) *.05

ASSIGN 555 TO LABEL

GO TO LABEL

8-5

CHAPTER 9
CONTROL STATEl·tENTS

9.1

INTRODUCTION

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

9.2

GO TO

CONTINUE

STOP

PAUSE

GO TO CONTROL STATEMENTS

There are three kinds of GO TO statements:
1.

Unconditional

Computed

Assigned

A GO TO control statement causes the statement that it identifies to
be executed next, regardless of its position within the program. The
following paragraphs describe each type of GO TO statement.

9.2.1

Unconditional GO. TO Statements

write GO TO statements of this type in the form
GO TO n
where n is the label, i.e., statement number, of an executable
statement, e.g., GO TO 555. When executed, an unconditional GO TO
statement transfers control of the program to the statement that it
specifies.

9-1

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

9.2.2

the

Computed GO TO Statements

Write GO TO statements of this type in the form
GO TO (Nl,N2, ••• ,Nk)E
where the parenthesized list is a list of statement numbers and E is
an arithmetic expression.
You may include any number of statement
numbers in the list of this type of GO TO statement;
however, each
number you give must be used as a label within the program or
subprogram containing the GO TO statement.
NOTE
A comma may optionally
parenthesized list.

the

The value of the expression E must be reducible to an integer value
that is greater than 0 and less than or equal to the number of
statement numbers given in the statement list. If the value of the
expression E does not compute within the foregoing range, the next
statement is executed.
When a computed GO TO statement is executed, the value of its
expreSS1on, i.e., E, is computed first.
The value of E specifies the
position within the given list of statement numbers of the number 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.

9.2.3

Assigned GO TO Statements

Write GO TO statements of this type in either of the following forms:
GO TO K
GO TO K, (Ll,L2, ••• Ln)
where K is a variable name and the parenthesized list of the second
form contains a list of statement labels, i.e., statement numbers.
The statement numbers you give must be within the program or
subprogram containing the GO TO statement.
Assigned GO TO statements of either foregoing form must be logically
preceded by an ASSIGN statement that assigns a statement label to the
variable name represented by K.
The value of the assigned label
variable must be in the same program unit as the GO TO statement in
which it is used. In statements written in the form
GO TO K, (Ll,L2, ••• Ln)
9-2

CONTROL STATEMENTS
if K is not assigned one of the statement numbers given
statement list, the next sequential statement is executed.

the

logical,

and

Examples
GO TO STATI
GO TO STATl, (177,207,777)

9.3

IF STATEMENTS

There are three kinds of
logical two-branch.

9.3.1

statements:

arithmetic,

Arithmetic IF Statements

Write IF statements of this type in the form
IF(E)Ll,L2,L3
where (E) is an expression enclosed within parentheses and Ll, L2, L3
are
the
labels, i.e., statement numbers, of three executable
statements.
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 Ll;

transferred

the

statement

Equal to 0, control
identified by L2;

transferred

the

statement

Greater than 0, control
identified by L3.

the

statement

transferred

You must give all three statement numbers in arithmetic IF statements;
the expression given may not compute to a complex value.
Examples
Sample Statement
IF (E TA) 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.

9-3

CONTROL STATEMENTS

NOTE
You must label the statement following
an
arithmetic
IF;
otherwise
the
statement can never be executed.

9.3.2

Logical IF Statements

Write IF statements of this type in the form
IF(E)S
where E is any expression enclosed in parentheses and S is a
executable statement.

complete

Logical IF statements transfer control of the program either to the
next sequential executable statement or to the statement given in the
IF statement, i.e., S, according to the computed logical value of the
given expression.
If the value of the given logical expression is
true (negative), control is given to the. executable statement within
the IF statement. If the value of the expression is false (positive
or zero), control is transferred to the next sequential executable
program statement.
The statement you give in a logical IF statement may be any executable
statement except a DO statement or another logical IF statement.
Examples
Sample Statement

9.3.3

IF (T.OR.S) X=Y+l

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

IF (Z.GT.X(K)) CALL SWITCH(S,Y)

Performs a subroutine call
the result of IF is true.

IF (K.EQ.INDEX) GO TO 15

Performs
transfer
is true.

an
unconditional
if the result of IF

Logical Two-Branch IF Statements

Write IF statements of this type in the form
IF (E) Nl, N2
where E is any expression, and Nl and N2 are statement labels
within the program unit.

defined

Logical two-branch IF statements transfer control of the program to
either statement Nl or N2, 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 Nl.
If the value
of the expression is false (positive or zero), control is transferred
to statement N2.
9-4

CONTROL STATEMENTS
Note that you must number
the statement immediately following the
logical two-branch IF so that control can later be transferred to the
portion of code that was skipped.
Examples
Sample Statement

9.4

IF (LOGl) 10,20

Transfers control to statement 10
if LOGI is negative;
otherwise
transfers control to statement 20.

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

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

31
C;
32
to

DO STATEMENT

DO statements simplify the coding of iterative procedures;
in the following form:

write them

Indexing Parameters
DO N I

= Ml,M2,M3

~fL ~~

TERMINAL
STATEMENT
LABEL
INDEX
VARIABLE

PARAMETER
TERMINAL
PARAMETER

INI-TIAL
PARAMETER
where
1.

Terminal Statement Label N is the statement number of the
last statement of the DO statement range. The range of a DO
statement is defined as the series of statements that follows
the DO statement up to and including its specified terminal
statement.

Index Variable I is an unsubscripted variable whose value is
defined at the start of the DO statement operations. The
index variable is available for use throughout each execution
of the range of the DO statement, but its value should not be
altered within this range.
It is also available for
use in
the program when:
a.

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

a CALL is executed from within the DO statement
that uses the index variable as an argument, and

9-5

range

CONTROL STATEMENTS
c.

if an input-output statement with either or both the
options END= or ERR= (Chapter 10) appears within the DO
statement range.

Initial Parameter Ml assigns the index variable,
I,
its
initial value.
This parameter may be any variable, array
element, or expression.

Terminal Parameter M2 provides the value that determines how
many repetitions of the DO statement range are performed.

Increment Parameter M3 specifies the value to be added to the
initial parameter (Ml) on completion of each cycle of the DO
loop.
If M3 and its preceding comma are omitted, M3 is
assumed to be equal to 1.

An indexing parameter may be any: ar ifhme'E:i.c expression resul ting in
either a positive or negativ~- ~~lti~: "The values of the indexing
parameters are calculated only once, at the start of each DO-loop
operation.
The number of times that a DO loop will execute is
specified by the formula:
MAX(l, ((M2-Ml)/M3)+1)
Since the 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
(Ml), and a count variable
(generated by the
compiler)
is set to the negative of the calculated count. At the end
of each DO loop cycle, the value of the increment parameter
(M3)
is
added to the index variable, and the count variable is incremented by
1. If the number of specified iterations has not been performed
(i.e., if the count variable is still negative), another cycle of the
loop is initiated.
One execution of a DO loop range is always performed regardless of the
initial values of the index variable and the indexing parameters.
Exit from a DO loop operation on completion of the number of
iterations specified by the loop count is referred to as a normal
exit.
In a normal exit, control passes to the first executable
statement after the DO loop range terminal statement, and the value of
the DO statement index variable is considered undefined.
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 (Section 9.4.3).

9.4.1

Nested DO Statements

One or more DO statements may be contained, i.e., nested, within the
range of another DO statement. The following rules govern the nesting
of DO statements.

9-6

CONTROL STATEMENTS
1.

The range of each nested DO statement must be entirely within
the range of the containing DO statement.
Example
Valid

Invalid

001

DO 1

~~
2.

The range of
DO 2 is outside
that of DO 1.

The ranges of nested DO statements cannot overlap.
Example
Valid

Invalid

001

DO 1

002

c= ~
C

003

The ranges of
loop DO 2 and
DO 3 overlap.

More than one DO loop within a nest of DO loops may 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. The statement label 4 of the shared
terminal statement cannot be used in any GO TO or arithmetic
IF statement that occurs anywhere other than within the range
of the DO statement to which it belongs.
Example
004

All the DO statements
share the same terminal
statement, however, it
belongs to the first
DO 4.

004
004

9-7

CONTROL STATEMENTS
9.4.2

Extended Range

The extended range of a DO statement is defined as the set of
statements that execute between the transfers out of the innermost DO
statement of a set of nested DOs and the transfer back into the range
of this innermost DO statement. The extended range of a nested DO
statement is as follows:
DO 1
002
003

Extended- Range

The following rules govern the use of a DO statement extended range:

The transfer out statement for an extended range operation
must be contained by the most deeply nested DO statement that
contains the location to which the return transfer is to be
made.

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.

CONTROL STATEr·tENTS

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

You may use and return from a subprogram within
range.

9.4.3

index

extended

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 executes, the value of the
controlling DO statement's index variable is defined as the
current value.

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

You may use and return from
range of any:

the

a.
b.
c.

subprogram

from

within

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

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

•

extended range

[

~4-<I- - , '

Invalid Transfer
Dl

9-9

operations

permitted

CONTROL STATEMENTS
9.5

CONTINUE STATEMENT

You may place CONTINUE statements anywhere in the source program
without
affecting the program sequence of execution.
CONTINUE
statements are commonly used as the last statement of a DO statement
range in order to avoid ending with a GO TO, PAUSE, STOP, RETURN,
arithmetic IF, another DO statement, or a logical IF statement
containing any of the foregoing statements. Write this statement as

12 CONTINUE
Example
In the following sequence, the labeled CONTINUE statement
legal termination for the range of the DO loop.

provides

DO 45 ITEM=l,lOOO
STOCK=NVNTRY (ITEM)
CALL UPDATE (STOCK,TALLY)
IF (ITEM.EQ.LAST) GO TO 77
45 CONTINUE

77 PRINT 20, HEADING,PAGENO

9.6

STOP STATEMENT

Execution of the STOP statement causes the execution of the object
program to be terminated and returns control to the monitor.
A
descriptive message may optionally be included in the STOP statement
to be output to your I/O terminal immediately before program execution
is terminated. Write this statement like this:
STOP

:'8,'1'01',' 'N

where 'N' is a string of ASCII characters enclosed by single quotes
and n is an octal string up to 12 digits. The string N or the value n
is printed at your I/O terminal when the STOP statement executes.
The'
string N may be of any length.
(Continuation lines may be used for
1 arge messages.)

9-10

CONTROL STATEMENTS
Examples
STOP 'Termination of the Program'
or
STOP 7777

9.7

PAUSE STATEMENT

Execution of a PAUSE statement suspends the execution
program and gives you the option to:
1.

Continue execution of the program

Exi t

Initiate a TRACE operation (Section 9.7.1).

the

object

The permitted forms of the PAUSE statements are:
1.

PAUSE

PAUSE 'literal string'

PAUSE n, where n is an octal string up to 12 digits.

Execution of a PAUSE statement of any of the
the standard instruction:

foregoing

forms

caUSes

TYPE G TO CONTINUE, X TO EXIT, T TO TRACE
to be printed at your terminal.
If the form of the PAUSE statement
contains either a literal string or an integer constant, the string or
constant prints on a line preceding the standard message.
For
example, the statement
PAUSE 'TEST POINT A'
causes the following to be printed at your terminal:
TEST POINT A
TYPE G TO CONTINUE, X TO EXIT, T TO TRACE
The statement
PAUSE 1
causes the following to be printed at your terminal:
PAUSE 000001
TYPE G TO CONTINUE, X TO EXIT, T TO TRACE

9-11

CONTROL STATEMENTS
9.7.1

T(TRACE) Option

, The entry of the character T in response to the message output by the
execution of a PAUSE statement starts a TRACE routine. This routine
causes a complete history of all subroutine calls made during the
execution of the program, up to the execution of the PAUSE statement
to be printed at your terminal. The history printed by the TRACE
routine consists of:
1.

The names of all subroutines called, arranged in the
order of their call;

reverse

The absolute location (written
called subroutine;

The name of the calling subroutine plus an offset factor and
the absolute location (written within parentheses) of the
statement within the routine that initiated the calli

The number
brackets);

within

angle

An alphabetic code (written within square brackets)
specifies
the
types
of each argument involved.
alphabetic codes used and the meaning of each are:

that
The

arguments

within

involved

I
F

o
S
D

C
K

(written

the

Type Specified

Code Char acter.
U

parentheses)

Undefined typei
the use of the
argument will determine its type.
Logical
INTEGER
Single-precision REAL
Octal
Statement Number
Double-precision REAL
COMPLEX
A literal or constant

: Example
,The following printout illustrates the execution of the
PAUSE
: statement "PAUSE 'TEST POINT AI", the entry of a T character to
initiate the TRACE routine, the resulting trace printout, and the
, entry of the character G to continue the execution of the program.
TEST POINT A
TYPE G TO CONTINUE, X TO EXIT, T TO TRACE.
*T
NAME
(LOC)
«--- CALLER (LOC)
<#ARGS> [ARG TYPES]
TRACE.
(414056) «--PAUSe +141 (376) <#1>
[U]
PAUSe
(235)
«--- MAIN.+4(151)
<#1>
[U]
TYPE G TO CONTINUE, X TO EXIT, T TO TRACE.
*G

9-12

CONTROL STATEMENTS

In addition to its use with the PAUSE s\:.a\:.emenl:,
TRACE routine directly, using the form

you

hlay

call

the

CALL TRACE
or as a function, using the form
X=TRACE(x)
Execution of the foregoing statements starts the TRACE routine, which
prints the history of all subprogram calls made during the execution
of the program, up to the execution of the CALL statement or up to the
execution of the function, respectively. The history printed by the
TRACE routine under these circumstances is as described in the
preceding paragraph.

9-13

CHAPTER 10
I/O STATEMENTS

10.1

DATA TRANSFER OPERATIONS

FORTRAN I/O statements permit the transfer of data between processor
'storage
(memory)
and peripheral devices and/or between storage
locations. Data in the form of logical records may be transferred by
use of an a) sequential, b) random access, c) append ,transfer mode, or
d) dump mode., The areas in memory from which data is to be taken
during output (write) operations and into which data is stored during
input (read) operations are specified by:
1.

A list in the I/O statement that initiated the transfer

A list defined by a NAMELIST statement, or

Between a specified FORMAT statement and the external medium.

The type and arrangement of transferred data may be specified by
format specifications located in either a FORMAT statement or an array
(formatted I/O), or by the contents of an I/O list
(list-directed
I/O).
The following sections describe the statements
required to initiate I/O transfer operations.

10.2

and

data

format

TRANSFER MODES

The characteristics and requirements of the a) sequential, b)
random
access, and c)
append data modes are described in the following
paragraphs.

10.2.1

Sequential Mode

Records are transferred during a sequential mode of operation in the
same order they appear in the external data file.
Each I/O statement
executed ln a sequential mode transfers the record immediately
following the last record transferred from the accessed source file.

10.2.2

Random Access Mode

This mode permits access to and transfer of records from a file in any
desired order. Random access transfers, however, may be made only to
(or from)
a device that permits
random-type
data
addressing
operations, i.e., disk, and to files that have previously been set up
10-1

I/O STATEMENTS
for random access transfer operation. Files for random access must
contain a specified number of identically sized records that may be
· accessed, individually, by a record number.
You may use the FORTRAN-20 OPEN statement - see Chapter 12 · subroutine call to DEFINE FILE to set up random access files.

Use the OPEN statement to establish a random access mode to permit the
· execution of random access data transfer operations.
The OPEN
statement should logically precede the first I/O statement for the
'specified logical unit in the user source program.

10.2.3

Append Mode

· This mode is a special version of the sequential transfer mode:
Use
· it only for sequential output (write) operations. The append mode
permits you to write a record immediately after the last logical
· record of the accessed file. During an 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 use an OPEN statement to establish
· append I/O operations can be executed.

10.3

append

mode

before

I/O STATEMENTS, BASIC FORMATS AND COMPONENTS

The majority of the I/O statements described in this chapter are
written in one of the following basic forms or in some modification of
these forms:
Use

Basic Statement Forms
Keyword (u,f}list
K~yw~rd (u#R,tjlist
Keyword (u,*}list
Keyword (u,N)
Keyword (u}list
: keyword (u#R)list

F.orma t ted I/O Transfer...
...
Random Access Formatted I/O Transfer
List-Directed I/O Transfer
NAMELIST-Controlled I/O Transfer
Binary I/O Transf~r
~and~m_Acceis-Binary I/b fianife~

where
Keyword

the statement name (READ or WRITE)

logical unit number

list

FORMAT statement number in the current program
unit or the name of an array that contains the
desired format specifications
I/O list

= the delimiter # followed by the number
record in an established random-access file

= symbol specifying a list-directed I/O transfer

= the name of an I/O list
statement

defined

NAMELIST

The following paragraphs provide details of the foregoing components.
10-2

I/O STATEHENTS
10.3.1

I/O Statement Keywords

The keywords (names) of the FORTRAN-IO
this chapter are:
1.
2.
3.
4.

10.3.2

READ
REREAD
ACCEPT
FIND
DECODE

7.
8.
9.

I/O

statements

described

WRITE
PRINT
TYPE
ENCODE

FORTRAN Logical Unit Numbers

Decimal numbers identify the physical devices used for most FORTRAN
I/O operations.
During compilation, the compiler assigns default
logical unit numbers for the REREAD, READ, ACCEPT,
PRINT, and TYPE
statements.
Default unit numbers are negatively signed decimal
numbers that you cannot access.
You may make the logical device assignments at run time, or you may
use the standard assignments contained by the FORTRAN-20 Object Time
System (FOROTS).
Table 10-1 lists the standard logical device
assignments.
We recommend that you specify the device explicitly in
the OPEN statement.

10.1.3

FORMAT Statement References

A FORMAT statement contains a set of format specifications that
defines the structure of a record and the form of the data fields
comprising the record. Format specifications may also be stored in an
array rather than in a FORMAT statement.
(Refer to Chapter 13 for a
complete description of the FORMAT statement.)
The execution of an I/O statement that includes either a FORMAT
statement number or the name of an array that contains format
specifications causes the structure and data of the transferred record
to assume the form specified in the referenced format.
Records
transferred under the control of a format specification are referred
to as "formatted" records.
Conversely, records transferred by I/O
statements that do not reference a format specification are referred
to as "unformatted" records. During unformatted transfers, data is
transferred on
a
one-to-one
correspondence
between
internal
(processor)
and external
(device)
locations, with no conversion or
formatting operations.
Unformatted files are binary files divided into records by FORTRAN-20
embedded control words; the control words are invisible to you. You
cannot prepare files of this type without using FOROTS.
Unformatted
files are for use only within the FORTRAN environment.

10-3

Table 10-1
FORTRAN-20 Logical Device Assignments
Device/Function

Default Filename

FORTRAN Logical Unit Number

Use

Standard Devices*

o
DSK
CDR
LPT
CTY
TTY
f-'

I
~

MTAO
MTAl
MTA2
FORTR
DSK
DSK
DSK
DSK
DSK

FORxx.DAT

00
01
02
03
04
05
06 through 15 not valid
16

ILLEGAL
Disk
Card Reader
Line Printer
Console Teletype
User's Teletype

*The total number of standard devices permitted is
parameter.

Magnetic Tape

Assignable Device
DISK

j
an

(J)

17
18
19
20
21
22
23
24

.........

installation

J-3
ttl

3:
ttl

J-3

(J)

Table 10-1 (Cont.)
FORTRAN-20 Logical Device Assignments
Device/Function

Default Filename

FORTRAN Logical Unit Number

Use

Standard Devices*
DEVl
DEV2
DEV3
DEV4

FORxx.DAT

DE(S

Assignable Devices

25
26
27
28

C/l

DEV63

FOR63.DAT

Disk

1-3
:t:t-:3

tI:l

3:
tI:l

Default Devices (inaccessible to the user)

:z:
t-:3

REREAD
CDR
TTY

Current file
FORCDR.DAT
FORTTY.DAT

LPT
TTY

FORLPT.DAT
FORTTY.DAT

REREAD statement
READ statement
ACCEPT statement
Not Valid
PRINT statement
TYPE statement

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

*The total number of standard devices permitted is
parameter.

installation

C/l

I/O STATEMENTS
10.3.4

I/O List

An I/O list specifies the names of variables,
arrays,
and array
elements to which input data is to be assigned or from which data is
to be output.
Implied DO constructs (Section 10.3.4.1), which specify
sets of array elements, may also be included in I/O lists.
The number
of items in a statement list determines the amount of data to be
transferred during each execution of the statement.

10.3.4.1 Implied DO Constructs - When an array name is given in an
I/O list, all elements of the array are transferred in the order
described in Chapter 3 (Section 3.5.3).
If only a specific set of
array elements is involved,
they may be specified in the I/O list
either individually or in the form of an implied DO construct.
Write implied DOs within parentheses in a format similar to that of DO
statements.
They may contain one or more variable, array, and/or
array element names, delimited by commas and followed by indexing
parameters that are defined as for DO statements.
The general form of an implied DO is
(name{SL) ,I=Ml,M2,M3)
where
name

an array name

the subscript list
element identifier

the index control variable that may represent a
subscript appearing in a preceding subscript list

Ml,M2,M3

the
indexing
parameters
that
specify,
respectively,
the
initial,
terminal,
and
increment values that control the range of I.
If
M3 is omitted (with its preceding comma), a value
of 1 is assumed.

array

Examples
S must be an integer variable
(A{S) ,S=1,5)

Specifies the first five elements of the
one-dimension array A, i.e., A{l), A(2),
A(3), A(4), A(5).

(A{2,S) ,S=1,10,2)

Specifies the elements A{2,l), A{2,3),
A{2,5), A{2,7), A{2,9) of array A.

(I, 1=1 , 5)

Specifies the integers 1,2,3,4, and 5.

As stated previously, implied DO constructs may also
more variable names.

contain

one

Example (B and C must be integer variables):
({A{B,C) ,B=l,lO),C=l,lO) ,I,J

Specifies a 10 X 10 set of elements
of array A, the location identified
by I, and the location identified
oy J.
10-6

I/O STATEMENTS
You may also nest implied DO constructs.
one or more sets of indexing parameters.

Nested implied DOs may share

Example
((A(J,K),J=1,5) ,D(K) ,K=l,lO)

Specifies a 5 X 10 set of elements
of
array
A and the first 10
elements of array D.

When you specify an array or set of array elements as either a storage
or transmitting area for I/O purposes, the array elements involved are
accessed in ascending order with the value of the
first
subscript
quantity varying most rapidly and the value of the last given
subscript increasing to its maximum value least rapidly.
For example,
the elements of an array dimensioned as TAB(2,3) are accessed in the
order:
TAB(l,l)
TAB(2,l)
TAB(1,2)
TAB(2,2)
TAB(1,3)
TAB(2,3)

10.3.4.2 Formatted Record Handling - Data is processed under
format
control so that each item in the I/O list is matched with a field
descriptor in the FORMAT statement.
If the end of the FORMAT
specification is reached and more items remain in the I/O list, a new
line or record is established and the data processing is restarted,
either:
1.

at the first item in the FORMAT specification or,

(if parenthesized sets of FORMAT specifications exist within
the FORMAT specification) with the last set within the FORMAT
specification.

On input, if the record is exhausted before the data transfers are
completed, the remainder of the transfer is completed as if the record
were extended with blanks.
See Section 13.2.2 for more details.

10.3.5

Specification of Records for Random Access

You must identify records to be transferred in a random access mode in
an I/O statement by an integer expression or variable preceded by an
apostrophe used as a delimiter, e.g., 1101.
NOTE
You may use a pound sign (#) in place of
the apostrophe (I), e.g., both #101 and
'101 are accepted by FORTRAN-20.

10-7

I/O STATEMENTS
10.3.6

List-Directed I/O

'The use of an asterisk in an I/O statement in place of a FORMAT
'statement number causes the specified transfer operation to be
"list-directed".
In a list-directed transfer, the data to
be
'transferred and the type of each transferred datum are specified by
,the contents of the I/O list included in the I/O command used.
The
· transfer of data in this mode is performed without regard for column,
,card, or line boundaries. The list-directed mode is specified by the
substitution of an asterisk (*) for the FORMAT statement reference,
i.e., f, of an I/O statement. The general form of a list-directed I/O
statement is
keyword (u,*}list
: Example
READ (5,*)I,IAB,M,L
· You may use list-directed transfers to read data from
· input device, including an input keyboard terminal.

any

acceptable

NOTE
Do not use device positioning commands,
such as BACKSPACE, SKIP RECORD, etc., in
conjunction
with
list-directed
I/O
operations.
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
, characters:
1.

Input constants must be of a type acceptable to FORTRAN-20.
Octal constants, although acceptable, are not permitted in
list-directed I/O operations.

Literal constants must be enclosed within single quotes,
e.g.,
'ABLE'.
A quoted string which is too long to fit in
one element of the input list will be placed in adjacent
elements and will be padded with blanks. If a quoted string
is being placed in an array and it fills more than one
element of the array, the remaining elements of the array
will be unchanged. In this case, it is assumed that the user
meant for the long string to go into the array and for any
following data to go into the rest of the input list. If the
string fits into one element of the array, the array will
continue to be filled.

Blanks are delimiters; therefore, they are not permitted
any but literal constants.

You may omit decimal points from real constants that do not
have a fractional part. FORTRAN-20 assumes that the decimal
point follows the rightmost digit of a real constant.

Complex constants must be enclosed in parentheses.

10-8

I/O STATEMENTS
Delimiters in data
following:

for

list-directed

input

must

comply

with

the

Delimiters may be either commas or blanks.

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 considered by
FORTRAN-20 as being only a single delimiter.

Represent a null (the complete absence of a datum)
by two
consecutive commas that have no intervening constant(s). You
may place any number of blanks, tabs, carriage return/line
feed characters, or end-of-input conditions between the
commas of a null. Each time you specify a null item in the
input
data, its corresponding list element is skipped
(unchanged). The following illustrates the effect of a null
input:
INPUT Items

101, 'A ,tab, 'NOl',
I

I I I /

Corresponding
I/O List Items

\ 'Lr'7B'NjBER

Resulting
Contents of
List Items

101.A un- NOI
changed
lAB

Slashes (/) cause the current input operation to terminate
even if all the items of the directing list are not filled.
The contents of items of the directing I/O list that either
are skipped (by null inputs) or have not received an input
datum before the transfer is terminated remain unchanged.
Once the I/O list of the controlling I/O 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 nulls in data for list-directed input may be
repetition factor so that an item is repeated.

assigned

The repetition of a constant is written as
r*K
where r is an integer constant that specifies the number of times
. constant represented by K is to be repeated.

the

The repetition of a null is written
asterisk.

integer

Examples
10*5
3*'ABLE'
3*

represents 5,5,5,5,5,5,5,5,5,5
represents 'ABLE', 'ABLE', 'ABLE'
represents null,null,null

10-9

followed

I/O STATEMENTS
10.3.7

NAMELIST I/O Lists

You may define one or more lists by a NAMELIST statement (Chapter 11).
Each I/O list defined in a NAMELIST statement is identified by a
unique (within the routine)
1- to 6-character name that may be
referenced by one or more READ or WRITE statements.
The first
character of each I/O list name must be alphabetic.
By using the
NAMELIST statement,
you eliminate the need for specifying the entire
I/O list.
I/O statements that reference a NAMELIST-defined I/O list cannot
contain either a FORMAT statement reference or an I/O list.
You
canno~ use NAMELIST-controlled I/O operation to transfer octal numbers
or literal strings.
You may use only NAMELIST-controlled READ/WRITE statements to bring
in/write out records formatted in the
following manner.
Format
records for NAMELIST-controlled input operations as follows:
$NAME Dl,D2,D3 ... Dn$
\vhere
1.

$ symbols delimit the beginning and end of the record.
The
first
$ must be in column 2 of the input record;
column 1
must be blank.

NAME is the name of a NAMELIST-defined input list.
The named
list identifies the processor storage locations that are to
receive the data items read from the accessed record.

Dl through Dn are pairs of the form "variable=value"
the value is assigned to the associated variable.
items cannot be octal numbers or literal strings.

where
These

NOTE
Do not use device positioning commands
such as BACKSPACE, SKIP RECORD, etc., in
conjunction with NAMELIST-controlled I/O
operations.
If you do, the results are
unpredictable.
See Chapter 11 for more information on NAMELIST I/O transfers.

10. 4

OPTIONAL READ/WRITE ERROR EXIT AND END-OF-FILE ARGUr.lENTS

You may optionally add either or both an error exit or an end-of-file
argument to the portion in parentheses of any form of the READ and
WRITE statements when a unit is specified.
Write the error exit argument as ERR=c where c is a statement number
in the current program unit.
Using this argument terminates the
current I/O operation and transfers program control to the statement
identified by the argument if an error is detected.
For example, the
detection of an error during the execution of
READ(10,77,ERR=101)TABLE,I,M,J

10-10

I/O STl\TEMENTS
terminates the input operation anu transfers program control to
statement 101. See the Library Subroutine ERRSNS (Chapter 15) to find
out how to identify the actual error that occurred.
When an ERR= transfer occurs, all items on the input list and
implied DO indexes on input or output lists become undefined.

all

Write the end-of-file argument as END=d, where d is a statement number
in the current program unit.
This branch, when taken, stops the
current I/O operation and transfers program control to the statement
identified by the argument.
In the example below, the detection of an
end-of-file condition during the execution of
READ(10,77,END=50)TABLE,I,M,J
results in the transfer of control to statement 50.
When an END= transfer occurs, all items on the input list receive the
value zero and all implied DO indices on input lists become undefined.
If the END= argument is not present, but an ERR= argument is, an
end-of-file (EOF) condition is treated as a user-trappable error. If
neither the ERR= nor the END= argument is present and an end-of-file
condition is detected, a message is printed, the file is closed,
program execution is terminated, and control is returned to the
monitor.

10.5

READ STATEMENTS

READ statements transfer data from peripheral devicgs into specified
processor storage locations.
The permitted forms of this type of
input statment permit READ statments to be used in both sequential and
random
access
transfer
modes
for
formatted,
unformatted,
list-directed, and NAMELIST-controlled data trans

10.6

SEQUENTIAL FORMATTED READ TRANSFERS

Descriptions of the READ statements that
sequential transfer of formatted data follow:
1.

may

used

for

the

Form:

READ (u,f)list

Use:

Input data from logical
unit
u,
formatted
according to the specification given in f, into
the processor storage locations identified in
input list.

Example:

READ (10,555)TABLE(10,20) ,ABLE, BAKER, CHARL

Fo rm:

READ (u, f)

Use:

Input the data from logical unit u directly into
either a Hollerith (H)
field descriptor or a
literal field descriptor given within the format
specifications of the referenced FORMAT statement.
If the referenced FORMAT statement does
not
contain either of the foregoing types of format
field descriptors, the input record is skipped.
If a required field descriptor is present, its
contents are replaced by the input data.

Example:

READ{15,101)
10-11

I/O STATEMENTS
3.

10.6.1

Form:

READ f

Use:

Input the data from the READ default device
(card
reader) directly into either a Hollerith (H) field
descriptor or a literal field descriptor given
within the format specifications of the referenced
FORMAT statement.
If the
referenced
FORMAT
statement does not contain either of the foregoing
types of format field descriptors, the input
record is skipped.
If a required field descriptor
is present, its contents are replaced by the input
data.

Example:

Form:

READ f,list

Use:

Input the data from the READ default device
(card
reader)
into the processor storage locations
identified in the input list.
The input data is
formatted as specified in f.

Example:

READ 15, ARRAY (20,30)

Sequential Unformatted Binary READ Transfer

You may use only the following form of the READ statement for
sequential transfer of unformatted input of FORTRAN binary data:

the

Form:

READ (u)list

Use:

Input one logical record of data from logical unit
u into processor storage as the value of the
location identified in list.
You may read only
binary files output by a FORTRAN-20 unformatted
WRITE statement with this type of READ statement.
NOTE
If you use the form READ
(u) ,
one
unformatted input record will be skipped.

Example:

(10.6.2

READ (10) BINFIL (10,20,30)

Sequential List-Directed READ Transfer

;You may use the following forms of the READ statements
sequential, list-directed input transfer:
1.

control

Form:

READ{u,*)list

Use:

Read data from logical device u into processor
storage as the value of the locations identified
in list.
Each input datum is converted, if
necessary,
to the type of its assigned list
variable.

Example:

READ(10,*)IARY(20,20},A,B,M
10-12

I/O STATEHENTS
2.

10.6.3

Fo rm:

READ *,list

Use:

Read the data from the READ default device
(card
reader)
into the processor storage locations
identified in the input list.
Each input datum is
converted,
if necessary,
to
the type of its
assigned list variable.

Example:

READ *,ABEL(10,20),I,J,K

Sequential NAMELIST-Controlled READ Transfers

You may use only the following form of the READ statement to
a sequential NAMELIST-controlled input transfer:

10.6.4

Form:

READ(u,N)

Use:

Read data from logical unit u into processor
storage as the value of the locations identified
by the NAMELIST input specified by the name N.
The input data is converted to the type of'
assigned variable if type conflicts occur.
Only
input files that contain records formatted and
identified for NAMELIST operations
(Paragraph
10.3.7) may be read by READ statements of this
form.

Random Access Formatted READ Transfers

You may use only the following form of the READ statement to
a random access formatted input transfer:

10.6.5

initiate

initiate'

Form:

READ (u#R,f)list

Use:

Input data from record R of logical unit u.:
Format each input datum according to the format
specifications of f
and place into processor
storage as values of the locations identified in.
list. Only disk files that have been set up by
either an OPEN or DEFINE FILE statement may be
accessed by a READ statement of this form.
(If
record R has not been written, an error results.)

Example:

READ(1#20,100) I, X(J)

Random Access Unformatted READ Transfers

You may use only the following form of the READ statement to
a random access unformatted input transfer:

10-13

initiate

I/O STATEMENTS
Form:

READ (u#R)list

Use:

Input data from record R of logical unit u.
Place
the input data into processor storage as the value
of the locations identified in list. Only binary
files that have been output by an unformatted
random access WRITE statement may be accessed by a
READ statement ·of this form.
(If record R has not
been written, an error results.)

Example:

READ (1#20) BINFIL
Read record number 20 into array BINFIL.
NOTE
If the form READ (u#R) is used,
it will
cause
logical
input record R to be
skipped.

10.7

SUMMARY OF READ STATEMENTS

Table 10-2 summarizes the various forms of the READ statements.
Table 10-2
Summary of READ Statements
Type of Transfer

Sequential

Random Access

Formatted

READ(u,f)list
READ(u,f)
READ f,list
READ f

READ(u#R,f)list

Unformatted

READ(u)list
READ(u)

READ(u#R)list
READ(u#R)

List-Directed

READ(u,*)list
READ *,list

NAMELIST

READ{U,N)

Note:

10.8

Transfer Mode

You may include the ERR=c and END=d arguments in any of
the ,above
READ
statements.
When included, the
foregoing arguments
must
be
last,
e.g.,
READ
(lO,20,END=lOl,ERR=500)ARRAY(50,lOO) .

REREAD STATEMENT

. The REREAD statement causes the last record read from the last
input device to again be accessed and processed.

10-14

active

I/O STl':.TEMENTS
You cannot use the REREAD feature of FORrl'RAi~-20 until iill inpul (READ)
transfer from a file has been accomplished.
If you use REREAD
prematurely, an error results.
Once a record has been accessed by a formatted READ statement, the
record transferred may be reread as many times as desired. In a
formatted transfer, you may use the same or new format specification
by each successive REREAD statement.
You may use the REREAD statement only for sequential
transfers. The form of the REREAD statement is:

formatted

data

Form:

REREAD f,list

Use:

Reread the last record read during the last
initiated
READ operation and input the data
contained by the record into the processor storage
locations specified in the input list. Format the
data read according to the format specificatiOns
given in statement f.

Example:

DIMENSION ARRAY(lO,lO) ,FORMA(lO,lO) ,FORMB(lO,lO),
1
FORMC(lO,lO)
90 READ(16,100)ARRAY

100 FORMAT (-----)

110 REREAD 100,FORMA
115 REREAD 150,FORMB
120 REREAD 160,FORMC
150 FORMAT(-----)
160 FORMAT (-----)
In the above sequence, statement 90 inputs data formatted according to
statement 100 into the array ARRAY. Statement 110 reads the record
read by statement 90 and inputs the data formatted as in the initial
READ operation into the array FORMA.
Statement 115 reads the record read by statement 90 and inputs
data formatted according to statement 150 into the array FORMB.

the

Statement 120 reads the record read by statement 90 and inputs
data formatted according to statement 160 into the array FORMC.

the

NOTE
If you try to REREAD a record input from
the teletype, you will get either the
current
record
or
the
last
150
characters
of
the
current record,
whichever is the lesser.

10-15

I/O STATEMENTS
10.9

WRITE STATEMENTS

WRITE statements transfer data from specified processor storage
locations to peripheral devices.
The various forms of the WRITE
statement enable it to be used in sequential,
appen~,
and random
'access transfer modes for formatted, unformatted, list-directed, and
'NAMELIST-controlled data transfers.

10.9.1

Sequential Formatted WRITE Transfers

You may use the following forms of the
sequential transfer of formatted data:
1.

10.9.2

WRITE

statement

for

the

Form:

WRITE(u,f)list

Use:

Output the values of the
processor
storage
locations
identified
in list into the file
associated with logical unit u.
Convert and
arrange
the
output
data
according to the
specifications given in f.

Example:

WRITE(06,500)OUT(10,20) ,A,B

Form:

WRITE f,list

Use:

Output the values of the
processor
storage
locations identified in list to the default device
(line printer).
Convert and arrange the output
data according to the specifications given in f.

Example:

WRITE 10,SEND(5,10) ,A,B,C

Form:

WRITE f

Use:

Output the contents of any Hollerith
(H)
literal (' I) field descriptor(s) contained by f
the default device (line printer).
If neither
the foregoing types of field specifications
found in f, no output transfer is performed.

Example:

WRITE 10

or
to
of
is

Sequential Unformatted Binary WRITE Transfer

You may use the following form of the
sequential transfer of unformatted data:

WRITE

statements

for

the

Form:

WRITE (u)list

Use:

Output the values of the
processor
storage
locations
identified
in list into the file
associated with logical unit u.
No conversion or
arrangement of output data is performed.

Example:

WRITE(12)ITAB(20,20) ,SUMS(lO,5,2)

10-16

I/O STATEMENTS
10.9.3

Sequential List-Directed WRITE Transfers

You may use the following form of the WRITE statement
sequential list-directed output transfer.

10.9.4

initiate

Form:

WRITE(u,*)list

Use:

Output the values of the
processor
storage
locations
identified
in list into the file
associated with logical unit u.
The conversion of
each datum from internal to external form is
performed according to the type of the list
variable from which the datum is taken.

Example:

WRITE(12,*)C,X,Y,ITAB(lO,10)

Sequential NAMELIST-Controlled WRITE Transfers

You may use only the following form of the WRITE statement to initiate
a sequential NAMELIST output transfer.

10.9.5

Form:

WRITE(u,N)

Use:

Output the values of the
processor
storage
locations
identified by the contents of the
NAMELIST-defined list specified by name N into the
file associated with logical unit u.

Example:

WRITE(12,NMLST)

Random Access Formatted WRITE Transfers

You may use only the following form of the WRITE statement to initiate
a random access type formatted output transfer:

10.9.6

Form:

WRITE(u#R,f)list

Use:

Output the values of the
processor
storage
locations identified by the contents of list to
record R of the file associated with logical
device u.
Only disk files that have been set up
by either an OPEN statement or a call to the
subroutine DEFINE FILE may be accessed by a WRITE
transfer of this form.
The data transferred will
be formatted according to the specifications given
in f.
Only those records
that
have
been
specifically written are available to be read.

Random Access Unformatted WRITE Transfers

You may use only the following form of the WRITE statement to initiate
a random access unformatted output transfer:
Form:

WRITE(u#R)list

Use:

Output the values of the
processor
storage
locations identified by the contents of list to
record R of the file associated with logical
10-17

I/O STATEMENTS
device unit u. Only disk files that have been set·
up by either an OPEN or a call to the DEFINE FILE
subroutine may be accessed by a WRITE transfer of·
this form. Only those records that have been
specifically written are available to be read.

10.10

SUMMARY OF WRITE STATEMENTS

Table 10-3 summarizes the various forms of the WRITE statements.
Table 10-3
Summary of WRITE Statements
Type of Transfer

Transfer Mode
Sequential

Random Access

Formatted

WRITE(u,f)list
WRITE f,list
WRITE f

WRITE(u#R,f)list

Unformatted

WRITE(u)list

WRITE(u#R)list

List-Directed

HRITE(u,*)list

NAMELIST-controlled

WRITE(u,N)

Note:

10.11

You may include the ERR=c and END=d arguments in
any WRITE statement which has a unit number;
however, they must be last.

ACCEPT STATEMENT

The ACCEPT statement enables you to input data via either a terminal
keyboard or a batch control file directly into specified processor
storage locations. Use this statement only in the sequential transfer
mode for the formatted transfer of inputs from your terminal keyboard
during program execution.
The following paragraphs describe the
permitted forms of the ACCEPT statement.

• 10.11.1

Formatted ACCEPT Transfers

: Use the following forms of the ACCEPT
transfer of formatted data.
1.

statement

for

the

sequential

Form:

ACCEPT f,list

Use:

Input data character-by-character from the user's
terminal into the processor storage locations
identified by the contents of list.
Format the
input data according to the format specifications
given in f.

Example:

ACCEPT lOl,LINE(73)
10-18

I/O STATEMENTS
2.

10.11.2

Form:

ACCEPT *,list

Use:

Input data character-by-character from the user's
terminal
into the processor storage locations
identified by the contents of list.
Convert the
input characters, where necessary, to the type of
its assigned list variable.

Example:

ACCEPT *,IAB,ABE,KAB,MAR

ACCEPT Transfers Into FORMAT Statements

You may use the following form of the ACCEPT statement to input data
from your terminal keyboard directly into a specified FORMAT statement
if the FORMAT statement has either or both a Hollerith
(H),
or
a
literal
('s')
field descriptor.
If the
referenced statement has
neither of the foregoing descriptors, the input record is skipped.

10.12

Form:

ACCEPT f

Use:

Replace the contents of the appropriate fields
of
statement f with the data entered at the user's
terminal keyboard.

Example:

ACCEPT 101

PRINT STATEMENT

The PRINT statement causes data from specified processor storage
locations to be output on the standard output device (line printer).
Use this statement only for
sequential formatted data transfer
operation; write it in either of the three following forms:
1.

Form:

PRINT f,list

Use:

Output the values of the
processor
storage
locations identified by the contents of list to
the line printer.
The values output are
to be
formatted and arranged according to the format
specifications given in f.

Example:

PRINT 55,TABLE(10,20) ,I,J,K

Form:

PRINT *,list

Use:

Output the values of the
processor
storage
locations identified by the contents of list to
the line printer.
The conversion of each datum
from
internal to external form is performed
according to the type of the list variable from
which the datum is taken.

Example:

PRINT *,C,X,Y,ITAB(lO,lO)

Form:

PRINT f

Use:

Output the contents of the FORMAT
statement
Hollerith
(H) or literal field descriptors to the
line printer.
If neither an H nor a literal field
10-19

I/O STATEMENTS
descriptor
is present in the referenced FORMAT
statement, no operation is performed.
Example:

PRINT 55

The third form of the PRINT statement is particularly useful when
employed with ACCEPT f statements to cause desired data (comments or
headings) to be inserted into reports at program execution time.
Example
The sequence
55

FORMAT(' END OF ROUTINE')

PRINT 55
results in the printing of the phrase
printer.

10.13

"END OF .ROUTINE"

the

line

TYPE STATEMENT

The TYPE statement causes data from specified processor storage
locations to be output to your (control) terminal printing or display·
device.
Use this statement only for sequential formatted data
transfers;
write it in one of the following forms:
1.

Form:

TYPE f,list

Use:

Output the values of the
processor
storage
locations identified by the contents of list to
the user's terminal.
The values output are to be
formatted according to the format specifications
given in f.

Example:

TYPE 101,TABLE(10,20)I,J,K

Form:

TYPE f

Use:

Output the contents of the referenced FORMAT
statement
Hollerith
(H)
or
literal
field
descriptors to the user's terminal device.
If the
referenced
FORMAT statement does not contain
either an H or a literal field descriptor,
no
operation is performed.

Example:

TYPE 101

Form:

TYPE *,list

Use:

Output the values of the
processor
storage
locations identified by the contents of list to
the user's terminal.
The conversion of each datum
from
internal to external form is performed
according to the type of the list variable from
which the datum is taken.

Example:

TYPE *,IAB(l,S) ,A,B

10-20

I/O STATEMENTS
10.14

FIND STATEMENT

The FIND statement does not initiate a data transfer operation;
use
it during random access read operations to locate the next record to
be read while the current record is being input.
The program does not
have access to the "found" record until the next READ statement is
executed.
The form of the FIND statement is
FIND(u#R)
Example:
In the sequence
READ(Ol#90)
FIND(Ol#lOl)

READ(Ol#lOl)
the FIND statement will locate record #101 on device 01 after
record
90 has been retrieved.
Record #101 is not processed until the second
READ statement in the sequence is executed.

10.15

ENCODE AND DECODE STATEMENTS

Use the ENCODE and DECODE statements to perform sequential formatted
data transfer between two defined areas of processor storage, i.e., an
I/O list and a user-defined buffer;
no peripheral I/O device is
involved in the operations performed by these statements.
The ENCODE statement transfers data from the variables of a specified
I/O list into a specified storage area.
ENCODE operations are similar
to those performed by a WRITE statement.
The DECODE statement transfers data from a specified storage area into
the processor storage locations identified by the variables of an I/O
list.
DECODE operations are similar to those performed by a READ
statement.
write the ENCODE and DECODE statements in the following forms:
ENCODE(c,f,s)list
DECODE(c,f,s)list
where
c specifies the number of characters to be in each internal
storage area.
This argument may be an integer, an integer
expression, or either a real or double precision expression that
is converted to an integer form.
NOTE
5 characters per storage location are stored in the
buffer without regard to the type of variable given as
the starting location.

10-21

I/O STATEMENTS
f specifies either a FORMAT statement or an array
format specifications.

that

contains

s specifies the address of the first storage location that is to
be used in the transfer operations.
When multiple records are
specified by the format being used, the succeeding records follow
each other in order of increasing storage addresses.
list specifies an
10.3.4).

I/O

list

the

standard

form

(Paragraph

When multiple records are stored by ENCODE, each new record starts on
a new storage location boundary rather than there being a CRLF
inserted between records.

10.15.1

ENCODE Statement

A description of the form and use of the ENCODE statement follows:
Form:

ENCODE(c,f,s)list

Use:

The values of the processor storage locations
identified by the contents of list are converted
to ASCII character strings according to the format
specifications
given
in
f.
The converted
characters are then written into the destination
area starting at location s.
If you try to
transfer more characters than the specified area
can contain, the excess characters are ignored.
If you transfer fewer characters than specified
for the record size, the empty character locations
are filled with blanks.

Example:

10.15.2

ENCODE(500,101,START)TABLE

DECODE Statement

A description of the form and use of the DECODE statement follows:
Form:

DECODE(c,f,s)list

Use:

The character strings are taken starting
at
location s, converted (decoded) according to the
format specifications given in f,
and stored as
the values of the processor storage locations
identified in list.
If the
format
specification
requires
more
characters from a record than are specified by c,
the extra characters are assumed to be blanks.
If
fewer characters are required from a record than
are specified by c,
the extra characters are
ignored.

Example:

DECODE(50,50,START)GET(5,10)

10-22

I/O STATEMENTS
10.15.3

Example Of ENCODE/DECODE Operations

The following program illustrates the
DECODE statements:

use

both

the

ENCODE

and

Example
Assume the contents of the variables to be as follows:
A(l)
A(2)
J
B
C
(1 )
( 2)
( 3)
( 4)
( 5)
( 6)

contains the floating point number 300.45
contains the floating point number 3.0
is an integer variable
is a 4-word array of indeterminate contents
contains the ASCII string 12345
DO 2 J=l,2
ENCODE(16,10,B)J,A(J)
FORMAT(lX,2HA(,I1,4H)
TYPE 11,B
FORMAT (4A5)
CONTINUE
DECODE(5,12,C)B
FORMAT(3Fl.0,lX,F1.0)
TYPE 13,B
FORMAT ( 4 F 5. 2)
END

10
11
2

( 7)
( 8)

12
(9)
(10) 13
(11)

,FB,2)

Array B can contain 20 ASCII characters. The result of
statement after the first iteration of the DO loop is:
B(l)

'A (1)

B (2)
B (3)
B (4)

the

Typed at line 4 as

'300.4'

A (1)

300.45

•5

The result after the second iteration is:
B(l)

'A (2)

Typed at line 4 as

B (2)
B (3)
B (4)

'=
'3.0

A(2)

3.0

The DECODE statement:
1.
2.
3.
4.
5.
6.
7.

Extracts the digits 1, 2, and 3 from C
Converts them to floating point values
Stores them in B(l), B(2), and B(3)
Skips the next character (the digit 4)
Extracts the digit 5 from C
Converts it to a floating-point value, and,
Stores it in B(4)

The output from the TYPE statement at line 9 is:
1.00 2.00 3.00 5.00

10.16

SUMMARY OF I/O STATEMENTS

Table 10-4 summarizes all permitted forms of the I/O statement.

10-23

ENCODE

'fable 10-4
Summary of I/O Statements
I/O Statements
Formatted
READ
Sequential

Random
WRITE
Sequential or
Append(l)
f--J
0

Transfer Format Control
Unformatted
Namelist

READ (u,f)list
READ f,list
READ f

READ(u)list

READ (u#R,f)list

READ(u#R)list

WRITE (u,f)list
WRITE f,list
WRITE f

WRITE(u)list

List-Directed

READ(u,N)

READ(u,*)list
READ *,list

WRI'rE (u,N)

,\'lRITE (u, *) list

Random(2)

v~RI'rE

(u#R,f)list

t-3

WRITE (u#R)list

t-3

.t:.

trl

REREAD
Sequential

REREAD f,list

FIND
Random-only

FIND (u#R)

ACCEPT
Sequential only

Z
t-3
m

FIND ( u#R)

ACCEPT f,list
ACCEPT" f

1. You must use an OPEN statement to set up an append mode.
2. You must use either the OPEN statement or a call to the DEFINE
FILE subroutine to set up a random 'access mode.

ACCEP'f *, list

Table 10-4 (Cont.)
Summary of I/O Statements
I/O Statements
Formatted
PRINT
Sequential only

Transfer Format Control
Unformatted
Namelist

List-Direction

PRIN1' *, list

PRINT f,list
PRINT f

TYPE
Sequential only
f-'

TYPE *,list

TYPE f,list
TYPE f

lJ1

1-3
~

I
tv

ENCODE
Sequential only
DECODE
Sequential only
Legend:
u
f

list
N

1-3
tlj

ENCODE (c,f,s)list

Ei3

1-3
Ul

I DECODE (c,f,s)list

logical unit number
statement number of FORMAT
statement or name of array
containing format information
I/O list
name of specific NAMELIST
I/O list

*
#R
c
s

symbol used to specify list-directed I/O
operator
variable which specifies logical record
position
number of characters per internal record
address of the first storage location to
be used

CHAPTER 11
NAMELIST STATEMENTS

11.1

INTRODUCTION

Use the NAMELIST statement to define I/O lists similar to those
described
in Chapter 10
(Paragraph 10.3.4).
Reference defined
NAMELIST I/O lists in special forms of the READ and WRITE statements
to provide a method of transferring and converting data without
referencing format specifications or specifying an I/O list in the I/O
statement.

11.2

NAMELIST STATEMENT

Write NAMELIST statements in the following form:
NAMELIST/Nl/Al,A2, ... ,An/N2/Bl,B2, ... ,Bn/Nn/ ...
where
/Nl/ through /Nn/

represents names of
Always enclose the
(/N/)

individual lists.
name with slashes

Al through An
and
Bl through Bn

are the
items of the lists identified,
respectively, by names Nl and N2.
A
list may contain one or more variable,
array,
or array element names.
Delimit
the items of a list by commas.
Each
list
of
a
NAMELIST
statement is
identified (and referenced) by the name
immediately preceding the list.

Example
DIMENSION C(2,4)
NAMELIST/TABLE/A,B,C/SUMS/TOTAL
In the foregoing example,
the name TABLE identifies the
list
A,B,C(2,4),
and the name SUMS identifies the list consisting of the
array TOTAL.
Once a list has been defined in a NAMELIST statement, one or more
statements may reference its name.

11-1

I/O

NAMELIST STATEMENTS

The rules for structuring a NAMELIST statement are:
1.

You may use a maximum of six characters for a NAMELIST name.

You must begin it with an alphabetic character.

You must enclose it in slashes.

The NAMELIST name must precede the list of entries
it refers.

The NAMELIST name must be unique within the program.

You may define a NAMELIST name only once, and you must define
it by a NAMELIST statement. Once defined, you may use a name
only in READ or WRITE statements.

You must define the NAMELIST name
statement in which it is used.

You cannot use a variable used in a NAMELIST statement
dummy argument in a SUBROUTINE definition.

You must define any dimensioned variable contained in a
NAMELIST
statement
in
an array declaration statement
preceding the NAMELIST statement.

11.2.1

advance

which

the

I/O
as

NAMELIST-Controlled Input Transfers

During input (READ) transfer operations in which a NAMELIST-defined
name is referenced, the records are read until a record is found that
begins with the sequence 1 $1 (a space followed by a dollar sign)
followed by the referenced name. The dollar sign must be the second
character in the record; the first character in the record must be a
blank.
Once the proper symbol-name combination is found, the data
items following it are transferred on a one-to-one basis to the
processor
storage locations identified by the contents of the
referenced list. The input data is always converted to the type of
the list variable when there is a conflict of types. 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. Data items of records to be input
(read) using NAMELIST-defined lists must be separated by commas and
may be of the following form:
V=Kl,K2, ... ,Kn
where
1.

V may be a variable, array, or array element name.

Kl through Kn are constants of type integer,
real, double
precision, complex (written as (A,B) where A and B are real),
or logical (written as T for true or F for false). A series
of identical constants may be represented as a single
constant preceded by a repetition factor
(5*5 represents
5,5,5~5,5) .

11-2

NAMELIST STATEMENTS
In transfers of this type,
logical and complex constants must be
equated to variables of their own type.
Other type constants (real,
double-precision, and integer) may be equated to any other
type of
variable
(except logical or complex), and will be converted to the
variable type.
For example, assume A is a 2-dimensional real array, B
is a
I-dimensional integer array, C is an integer variable, and that
the input data is as follows:
$FRED A(7,2)=4, B=3,6*2.8, C=3.32$
A READ statement referring to the NAMELIST defined name FRED will
result
in the following:
The integer 4 will be converted to floating
point and placed in A(7,2).
The integer 3 will be placed in B(l), and
the
integer
2 (converted) will be placed in B(2) ,B(3), ... ,B(7).
The
floating point number 3.32 will be converted to the
integer
3 and
placed in C.
NOTE
"&"
may be used
instead
NAMELIST-controlled input.

11.2.2

11$"

NAMELIST-Controlled 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.
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.

statement

contain

all

referencing

For example, if JOE is a 2 X 3 real array, the statement
NAMELIST/NAMl/JOE,Kl,ALPHA
(u, NAMl)

~~RITE

generates the following form of output:
Column

$NAMEI
JOE= -6.750000

,
0.2340000E-04,
-1970000.
Kl=
, $
3.000000

O.OOOOOOOE+OO,
ALPHA=

680.0000
73.10000

NOTE
Do not use device positioning commands
such as BACKSPACE, SKIP, RECORD, etc.,
with NAMELIST-controlled I/O operations.
If
you
do,
the
results
are
unpredictable.
11-3

-17.80000

CHAPTER 12
FILE CONTROL STATEMENTS

12.1

INTRODUCTION

This chapter describes the OPEN and CLOSE statements.
They are file control statements used to set up files and establish
parameters for I/O operations and to terminate I/O operations.

12.2

OPEN AND CLOSE STATEMENTS

Both the OPEN and CLOSE statements are unique to FORTRAN-20;
they
both use the same format and have the same options and arguments.
The OPEN statement enables you to define all of the important aspects
of each desired data transfer operation; it provides an extensive.
list of required and optional arguments that define in detail:
.
1.

the name and location of the data file

the type of access required

the data format within the file

the protection code(l) to be assigned an output data file

the disposition of the data file

data file record, block and file sizes

a data file version identifier

In addition, a DIALOG argument is provided that permits you to
establish a dialogue mode of operation when the OPEN statement
containing it is executed.
In a
dialogue
mode,
interactive
terminal/program communication is established. This enables you to
define, redefine, or defer the values of the optional arguments
contained by the current OPEN statement during program ,execution.
The general form of the OPEN statement is:
OPEN(Argl,Arg2, ... ,Argn)

1. Refer to the Monitor Calls Manual, for a description of file access
protection codes.

12-1

FILE CONTROL STATEMENTS
Use the CLOSE statement in the termination of an I/O operation to
dissociate the I/O device being used from the active file and
file-related information, and to restore the core occupied by I/O
buffers and other transfer-related operations. All required device
dependent termination functions are also performed on the execution of
a CLOSE statement. Note that the CLOSE statement can change the name,
protection, directory, and disposition of the file being closed.
Once a CLOSE statement has been executed, you must
statement to regain access to the closed file.

use

another

OPEN

The general form of the CLOSE statement is:
CLOSE(Argl.,Arg2., ... ,Argn}
CAUTION
If you use a filename argument in a
CLOSE statement that is different from
the current filename, the file will be
renamed.

12.2.1

Options for OPEN and CLOSE Statements

The options and their arguments, which you may use in
and CLOSE statements, are:
1.

UNIT

both

the

OPEN

This option is required;
it defines the
FORTRAN I/O unit number to be used. FORTRAN
devices are identified by assigned decimal
numbers within the range 1-63; however, UNIT
may be assigned an integer variable
or
constant.
The general form of this argument
is:
UNIT=

An integer variable or constant
NOTE

FORTRAN-20
standard
logical
unit
assignments are described in Chapter 10
(Table 10-1). The range, i.e., 1-63, of
the
possible
UNIT
numbers
is an
installation-defined parameter.
2.

DEVICE

This option may specify either the physical
or
the logical name of the I/O device
involved.
(A logical name always
takes:
precedence over a physical name.) The DEVICE
arguments may specify I/O devices located at
remote stations, as well as logical devices.
The general form of the DEVICE argument is:
DEVICE=

12-2

A literal constant or variable

FILE CONTROL STATEMENTS
If you omit this option, the logical name u
(where u is the decimal unit number) is
tried;
if this
is not successful,
the
standard (default) device is attempted.
3.

ACCESS

ACCESS describes the type of
input and/or
output statements and the file access mode to
be used
in a
specified
data
transfer
operation.
You may assign ACCESS anyone of
six possible names, each of which specifies a
specific
type
of
I/O
operation.
The
assignable names and the operations specified
are:
a.

SEQIN

The specified data file
is
to be
read
in sequential
access mode.

SEQOUT

The specified data file
to
be
written
in
sequential access mode.

SEQINOUT

The specified data file
may
be first read, then written
(READ/WRITE
sequence)
record-by-record
in
a
sequential
access
mode.
When you specify SEQINOUT, a
WRITE/READ
sequence
is
illegal.
If no access is
specified,
SEQINOUT
is
assumed.

RANDOM

The specified data file may
be either
read or written
into, one record at a
time.
In a random access mode of
operation,
the
relative
position of each record is
independent of the previous
READ
or WRITE statement;
all records accessed must
have a fixed logical record
length.
The RECORD
SIZE
option
is
required
for
random access
operations.
You
must specify a disk
device
when
the
random
argument is used.

RANDIN

This argument enables you to
establish
a
special,
read-only random access mode
with a named file.
During a
RANDIN mode,
you may read
the
named
file
simultaneously with
other
users
who
have
also
established a RANDIN mode
and with the owner of the
file.
The use of RANDIN
enables a data base to be
shared by more than one user
at the same time.

12-3

is
a

FILE CONTROL STATEMENTS
f.

APPEND

The record specified by a
corresponding
WRITE
statement is to be added to
the logical end of a named
file. You must close and
then
reopen the modified
file to permit it to be
read.
The general form of
ACCESS argument is:

the

'SEQIN'
'SEQOUT'
'SEQINOUT'
ACCESS= 'RANDOM'
'RANDIN'
'APPEND'
variable (set to
literal)
4.

MODE

This option defines the character set of an
external file or record.
The use of this
argument is optional; if you do not use it,
one of the following is assumed:
a.
b.

ASCII for a formatted I/O file transfer
Binary for an unformatted
I/O
file
transfer.
NOTE

Refer to the Monitor Calls Manual for a
detailed description of the data modes
given in the following list.
You must use one of the following character
set specifications with the MODE argument:
Literal

Action Indicated

'ASCII'

Specifies an ASCII character set.

'BINARY' Specifies data
formatted
FORTRAN binary data file.

'IMAGE'

Specifies an image (I) mode data
transfer for the associated READ or
WRITE statements.
IMAGE
is
an
unformatted binary mode.

'DUMP'

The data file to be
to be handled in
operation.

transferred is
a DUMP mode of

The general form of the MODE argument is:
'ASCII'
'BINARY'
'IMAGE'
'DUMP'
variable (set to literal)

MODE=

12-4

FILE CONTROL STATEMENTS

DISPOSE

This option specifies an action to be
taken
regarding a
file at close time.
When used,
DISPOSE must be either a variable or one of
the following literals:
Literal
'SAVE'

Action Indicated
Leave the file on the device.

'DELETE' If the device involved is a disk,
remove the file:
otherwise, take no
action.
'PRINT'

If the file is on disk, queue it for
printing;
otherwise,
take
no
action.

'LIST'

If the file is on disk, queue it for
printing
and
delete
the
file:
otherwise take no action.

'RENAME' Change filename.
(This is redundant
if a new filename is given.)
If the DISPOSE argument is not given,
argument DISPOSE
'SAVE' is assumed.
general form of the DISPOSE argument is:

DISPOSE=

FILE

'SAVE'
'DELETE'
'PRINT'
'LIST'
'RENAME'
variable

the
The

(set to literal)

This option specifies the name of the
file
involved in the data transfer operation.
FILE
must
be
either
a
literal,
double-precision,
complex,
or
single-precision variable.
Single-precision
variables are assumed to contain a 1- to
5-character
file
specification:
double-precision
variables
permit
10-character file specification.
The
format
is a 1- to 6-character filename optionally
followed by a period and a 0- to 3-character
extension.
Any excess characters in either
the name or extension are
ignored.
If you
omit the period and extension, the extension
.DAT is assumed;
if just the extension is
omitted,
a null extension is assumed.
So if
you want a
filename without an extension,
remember to use the period.
If a filename is not specified or is zero,
a
default name is generated that has the form
FORxx.DAT
where xx is the FORTRAN logical unit number
(decimal) or is the logical unit name for the
default statements ACCEPT,
PRINT,
READ,
or
TYPE.
The general form of a FILE argument
is:
12-5

FILE CONTROL STATEMENTS

FILE=
7.

PROTECTION

A literal or variable set to a
literal

This option specifies a protection code to be
assigned the data file being transferred.
The protection code determines the level of
access
to the file
that three possible
classes of users (owner, member,
or other)
will have.
PROTECTION may be a 3-digit octal
literal or a variable;
if the argument is
assigned a zero value or is not given, the
default protection code established for
the
installation is used.
The general form of
the PROTECTION argument is:
PROTECTION= 3-digit octal
constant
integer variable

DIRECTORY

Use this option for disk files only.
It
specifies the location of the user
file
directory (UFD)
or the sub-file directory
(SFD) that contains the file specified in the
OPEN statement. A directory identifier may
consist of either:
a. Your project
programmer
number
that
identifies the UFD, for example, 10,7, or
b. A UFD/SFD directory path specification. A
path specification lists the UFD and the
names of its SFDs that form a path to the
desired SFD.
For example, the following
path specification identifies the path
leading to SFD 1234:
lO,7,SFDA,SFDB,1234

NOTE
Refer to the Monitor Calls Manual for
a
complete description of directories and
multilevel directory structures.

The general form of a DIRECTORY argument is:
DIRECTORY=

array
Literal
or
containing
directory
specification

name
path

You may also establish an array containing
the directory specification as its elements,
and reference the array in the DIRECTORY
argument.
Single-precision arrays permit
5-character directory names to be
used;
double-precision arrays permit 6-character:

12-6

FILE CONTROL STATEHENTS
(0 )
names to be used.
You must use a
zero
entry
to
terminate
a
directory
path
specification given in an array.

Examples of
the
use
of
singleand
double-precision arrays in an OPEN statement
DIRECTORY specification follow:
a. Single-Precision Array
5, DIRECTORY = PATH, ... )

OPEN (UNIT

where PATH and its elements are:
DIMENSION PATH (5)
PATH (1) = "10 ! (PROJECT NUMBER)
PATH (2)= "7
! (PROGRAHMER NUMBER) UFO
PATH (3)='SFDA'
Names of sub-file
PATH (4)='SFDB'
directories (SFD's)
PATH (5)=0
b.

Double-Precision Array
OPEN (UNIT=5, DIRECTORY = PATH, ... )
where PATH and its elements are:
DOUBLE PRECISION PATH (5)
PATH (1) =11000000000010000000000007
! (PROJ.,PROG. NUMBERS=UFD)
PATH
PATH
PATH
PATH

(2)='SFDABC'
(3)='MYAREA'
(4)='YOURIT'
(5)=0

!names of sub-file
!directories (SFDs)

The elements of a directory specification
may then be either a literal or a singleor double-precision array.
The following is
specification:

example

literal

DIRECTORY='lO,7,SFDl,SFD2,SFD3'
'---..--l \

ProJect
Programmer
Number

'/l'v"'"

Su -Fl e
Directory
Path

Whenever the specification is an array,
you
may
specify
the
required
project and
programmer numbers either of two ways.
You
can use one word with the project number in
the left half and the programmer number
in
the right half, or, use the right halves of
separate sequential word locations.

12-7

FILE CONTROL STATEMENTS
9.

BUFFER COUNT

This option enables you to specify the number
of I/O buffers to be assigned to a particular
device. If this argument is not given or is,
assigned a value of zero, the Monitor default
is assumed.
The general form
of
this
argument is:
BUFFER

10.

FILE SIZE

COUNT=

VERSION

BLOCK SIZE

An
integer
variable

RECORD SIZE

constant

An octal constant
variable

integer

You can use this option for all storage media
except disk.
It enables you to specify a i
physical storage block size for a device.
The value assigned the BLOCK SIZE arguments
may be an integer constant or variable.
The
size specified must be greater than or equal
to 3 and less than or equal to 4095.
The
general form of this argument is:
BLOCK SIZE= An
integer
variable

13.

Use this option for disk operations only; it
enables
you to assign a 12-digit octal
version number to a file when it is output.
The quantity assigned to the VERSION argument
may be either an octal constant or variable.
The general form of the argument is:
VERSION=

12.

constant

Use this option for disk operations only: it
enables you to estimate the number of words
that an output file is going to contain. The
use of FILE SIZE enables you to ensure at the
start of a program that enough space is
available for its execution.
If the size
specified is found to be too small during
program executions, the Monitor allocates
additional space according to the normal
Monitor algorithms.
The value assigned to
the FILE SIZE arguments may be an integer
constant or variable and will be rounded up ,
to the next higher block boundary (multiple
of 200 octal).
The general form of this
argument is:
FILE SIZE=

11.

An integer
variable

constant

This option enables you to force all logical
records to be a specified length.
If a
logical record exceeds the specified length,
it is truncated; if a logical record is less'
than the specified size, nulls are added to
pad the record to its full size. The RECORD
SIZE argument is required whenever a random'
access mode is specified. The value assigned
to this argument may be either an integer
constant or variable, and may be expressed as

12-8

FILE CONTROL STATEMENTS
the number of words or characters,
depending
on the mode of the file being described.
The
general form of this argument is:
RECORD
14.

ASSOCIATE
VARIABLE

SIZE=

An integer
variable

constant

Use this option for
disk
random
access
operations only.
It provides storage for the
number of the record to be accessed next if
the program being executed were to continue
to sequentially access records starting from
the current READ.
For example, if record
number 3 were read,
the ASSOCIATE VARIABLE
would be equal to 4.
The general form of
this argument is:
ASSOCIATE VARIABLE = Integer variable

15.

PARITY

Use this option for magnetic tape operations
only;
it permits you to specify the type of
parity to be observed (odd or even)
during
the transfer of data.
The general form of
this option is:
PARITY=

16.

DENSITY

DIALOG

(set to literal)

Use this option for magnetic tape operations
only;
it permits you to specify any of four
possible bit-per-inch
(bpi)
tape
density
parameters
for
magnetic
tape
transfer
operations.
The general form of this option
is:
DENSITY=

17.

'ODD'
'EVEN'
variable

'200'
'556'
'800'
'1600'
variable (set to literal)

The use of this option in an OPEN statement
you
to supersede or defer,
at
enables
time,
the
values
previously
execution
assigned to the arguments of the statement.
There are two forms of this argument.
The
first is:
DIALOG
This form establishes a dialogue with your
terminal when the OPEN statement is executed.
FOROTS outputs the following messages at the
user's terminal.
UNIT=n:/ACCESS=SEQINOUT/MODE=ASCII
ENTER NEW FILE SPECS.
END WITH A $ (ALT)
Once
the
message
and
defined
file
specification are output, you may enter any
desired changes.
You need enter only the
arguments that are to be changed.
12-9

FILE CONTROL STATEMENTS

The second form of the argument is:
DIALOG=

Literal or array

The value assigned to DIALOG may be a literal
or an array containing a file specification
with the desired information.
18.

ERR

The use of this option in an OPEN or CLOSE
statement enables you to transfer program
control to an executable statement when an
error is detected during the processing of
the OPEN or CLOSE statement.
The general
form of this option is:
ERR= s
where s
is the statement label
of
an
executable statement
(that appears in the
same program unit as the error specifier)
to
which program control is transferred when an
error is detected.
Associated with the ERR= option on OPEN/CLOSE
is the subroutine ERRSNS that enables you to
pinpoint the error.
See Appendix H for
FOROTS error values returned by ERRSNS.

Examples:
OPEN (UNIT= 1, 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', RECORD SIZE= 80,
2
ASSOCIATE VARIABLE= I, ERR= 240)
causes a disk file named PAYROL.DAT to be opened on unit 3 for
random
input/output 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
will terminate and control will transfer to the statement labeled 240.
CLOSE (UNIT= 3, DISPOSE= 'DELETE')
causes the file on unit 3 to be closed and removed if the file
disk.

12.2.2

Summary of OPEN/CLOSE Statement Options

Table 12-1 summarizes the options permitted and required in the
and CLOSE statements and the type of value required by each.

12-10

OPEN

FILE CONTROL STATEMENTS

Table 12-1
OPEN/CLOSE Statement Arguments
Argument
ACCESS=

Possible Value

Open*

Close*

SEQIN I , I SEQOUT I , I SEQINOUT I ,
RANDIN I , I RANDOM I , I APPEND I ,
or variable
Integer variable
Integer constant or variable
Integer constant or variable
Literal constant or variable
Literal constant or variable
Literal or array or none
Literal or variable or array
Literal constant or variable
Statement Number
Literal constant or variable
Integer constant or variable
Literal constant or variable
Literal constant or variable
An octal constant or
integer variable
Integer constant or integer
variable
Integer variable or constant
Octal constant or variable

0
0
0
0
0
0
0
0
0
0
0

I
I
I
I
I
I

ASSOCIATE VARIABLE=
BLOCK SIZE=
BUFFER COUNT=
DENSITY=
DEVICE=
DIALOG=
DIRECTORY=
DISPOSE=
ERR=
FILE=
FILE SIZE=
MODE=
PARITY=
PROTECTION=
RECORD SIZE=
UNIT=
VERSION=

R = Required
o = Optional
I = Ignored

12-11

0
0
0
0

I
I
I

0
0

CHAPTER 13
FORMAT STATEMENT

13.1

INTRODUCTION

Use FORMAT statements in conjunction with the I/O list of I/O
statements during formatted data transfer operations. The FORMAT
statements contain field descriptors that, together with the list
items of associated I/O statements, specify the forms of the data and
data fields that comprise each record.
FORMAT statements may appear almost anywhere in a source program. The
only placement restrictions are that they follow PROGRAM, FUNCTION,
SUBPROGRAM, or BLOCK DATA statements;
and that they precede the END
statement.
(Refer to Section 2.4.)
You must label FORMAT statements so that I/O statements can
them.

13.1.1

reference

FORMAT Statement, General.Form

The general form of a FORMAT statement follows:
k FORMAT(SAl,SA2, ... ,SAn/SBl,SB2, ... ,SBn/ ... )
where
k

= the required statement label (which
be referenced by I/O statements).

SAl through SAn
and
SBI through SBn

= individual field descriptor sets

can

only

In the foregoing statement form, the individual field descriptors are
delimited by commas
(,).
Field descriptor sets and records are
delimited by slashes (/).
For example, a FORMAT statement of the
form:
FORMAT(SAl,SA2/SBl,SB2/SCl,SC2)
contains format specifications for
three
containing two field descriptor sets.

records

with

each

record

Adjacent slashes (//) in a FORMAT statement specify that a record is
to be skipped during input or is to consist of an empty record on
output.
For example, a FORMAT statement of the form:
FORMAT (SA1,SA2///SBl,SB2)

13-1

FORMAT STATEMENT
specifies four records are to be processed;
third records are to be skipped.

however, the

second

and

You may represent repeated field descriptors or groups of field
descriptors by using a
repeat form.
Indicate the repetition of a
single field descriptor by preceding the descriptor with an integer
constant that specifies how many times the descriptor
is to be
repeated.
For example, a FORMAT statement of the form:
FORMAT(SA1,SA2,SA3,SA1,SA2,SA3,SA1,SA2,SA3)
may be written as
FORMAT(3(SA1,SA2,SA3))
You may nest the repeat forms of field descriptors to any depth.
example~ a FORMAT statement of the form:

For

FORMAT(SA1,SA2,SA2,SA3,SA1,SA2,SA2,SA3)
may also be written in the form:
FORMAT(2(SA1,2SA2,SA3))
The following paragraphs discuss the manner in which you may
foregoing statement forms and the effect each has on
involved.

13.2

use the
the data

FORMAT DESCRIPTORS

FORMAT statement descriptors describe the record structure of the
data,
the format of fields within the record, and the conversion,
scaling, and editing of data within specific fields.
The following
descriptors can be used with FORTRAN-20:
Descriptors

Comments

rFw.d
rEw.d
rDw.d
rGw.d

Floating point numeric field descriptors

rIw

Integer field descriptor

rLw

Logical field descriptor

rAw
rRw

Alphanumeric data field descriptor

kHs
'text'

Alphanumeric data in a FORMAT statement field
descriptor
Field formatting descriptors

Numerical scale factor descriptor

Record delimiter

Carriage return suppression for terminal

raw

Octal field descriptor
13-2

FORHAT STATEt·1ENT

where
r

an optional unsigned integer representing a repeat count.
This option enables a field descriptor to be repeated r
times.

an optional integer constant representing the width (total
number of characters contained) of the external form of
the field being described.
All characters,
including
digits,
decimal points,
signs,
and blanks that are to
comprise the external form of the field, must be included
in the value of w .

an optional unsigned
integer specifying the number of
fractional digits
that are to appear
in the external
representation of the field being described.
Note that w
must be specified if .d is included in the descriptor.

an unsigned integer specifying the number of characters to
be processed during the transfer of alphanumeric data.

represents a string of ASCII

a signed integer constant (plus signs are optional).

(alphanumeric) characters.

The characters A, D, E, F, G, H, I, L, 0, P, and R indicate the manner
of conversion and editing to be performed between the internal
(processor) and external representations of the data within a specific
field;
these characters are referred to as conversion codes.
Table
13-1 gives the FORTRAN-20 conversion codes and a brief description of
the function of each.
Table 13-1
FORTRAN-20 Conversion Codes
Code
A
D

E
F
G
H
I
L

Q
P
R

Function
Transfer alphanumeric data
Transfer real data with a D exponent{l)
Transfer real data with an E exponent{l)
Transfer real data without an exponent
Transfer integer, real, complex, or logical data
Transfer literal data
Transfer integer data
Transfer logical data
Transfer octal data
Numer icals'caling' factor
Transfer alphanumeric data

1. An exponent of 0 is assumed if none is given.
The use of commas to delineate format descriptors within a
format
specification is optional as long as no ambiguity exists.
For
example,
FORMAT ( 3X, A2)
can be written as
FORMAT (3XA2)
13-3

FORMAT STATEMENT
Since interpretation of a format specification
the specification

left

associative,

FORMAT(I22,I5)
can be written as
FORMAT(I22I5)
However, a comma is required when you wish to specify
FORMA T ( 12 , 2 I 5 )
The following paragraphs provide detailed descriptions of the "various
types of format descriptors, the manner in which they are written and
employed, and their use in FORMAT statements.

13.2.1

Numeric Field Descriptors

The forms of the field descriptors used
conversion of numeric data follow.

Fw.d
Fw.d,Fw.d
Iw

'" ','

Gw
Gw.d,Gw.d

the

format

and

Double-precision data with a D exponent
Real data with an E exponent
For the real and imaginary parts of a complex
datum
Real or double-precision data without an exponent
For the real and imaginary parts of a complex
datum
Integer data

r"Ow-"'-~~-"~-"'~~"'~ ~~'~'~':.J;:j'~I~~f·~"?I'at:~

~Gw'.'d

specify

Type of Data Used For

Description
Dw.d
Ew.d
Ew.d,Ew.d

Real or double-precision data
For integer (or logical) data
For the real and imaginary parts of a complex
datum
NOTE
The G conversion code may be used
all but octal numeric data types.

Examples
Consider the; following program segment:
INTEGER 11,12
REAL Rl,R2,R3
DOUBLE PRECISION Dl,D2
II
506
12
8
Rl
506.0
R2
18.1
R3
506001.0
Dl
18.0
D2
-504.0

13-4

for

FORMAT STATEMENT
Table 13-2 describes the actions performed by several types of
formatted WRITE statements on the data given in the foregoing program
segment.
Table 13-2
Action of Field Descriptors On Sample Data
Item Descriptor
Form

Sample
Descriptor

1
Dw.d
Ew.d
2
Fw.d
3
4
Iw
Iw
S
~_ 6 , , Ow",~_"
Gw.d
7
S
Gw.d

DS.2
ES.2
FS.2
IS
I2
OS

WRITE(-,-)Dl
WRITE(-,-)Rl
WRITE(-,-)R2
WRITE(-,-)Il
WRITE(-,-)Il
_,_, WRITE '(-: '-: }1?'

GS.2
GS.2
GS

WRITE(-,-)R3
WRITE(-,-)R2
WRITE(-,-)Il

Gw.d
Gw

where:

"'-GS' ~-i"'

WRITE
Statement
Using the
Sample
Descriptor

"WRITE"( ::~ :Yb'i"

External
Form
of Sample
Field
Described
Z.nnD
Z.nnE
aa.nn
aaaan
an
nnnnn
Z.nnD
Z.nnE
aa.nn
aaan

nn
nn

External
Appearance
of Sample
Data
0.lSD+02
0.SlE+03
lS.lO
,kS,l6s 0 6

,9,Q~Q<~ ,9~ .,~ "A_~'_~M"Y'A'~h~J

-.SOD+02
0.SlE+06
)5}6iS1S.lO

iSiS 50 6

n represents a numeric character.

Z represents either a - or O.
(Note that if
a negative number cannot be output.)

a represents a digit, leading blank (¥) or
sign depending on the numeric output.

n-d>6,
a

minus

Notes:
1.

In Item 1, the value Dl has only two significant digits and
d=2, so no rounding will occur on input.

In Item 2, since Rl has 3 significant digits, it is rounded
to fit into the specified field.

In Item S, the width
(w)
part of a format descriptor
specifies an exact field that permits no rounding of its
contents.
If the w specification is too small for
the
datum to be transferred, asterisks are output to indicate
that the transfer was not made.

In Items Sand 9, the relationship between G and fixed
floating real data is discussed in Paragraph 13.2.3.

In Items 1, 2, 3, 7, and S, the D and E exponent prefixes
are optional in the external form of the floating point
constants. For example, 1.lE+3 may be written as 1.1+3.

Table 13-3 summarizes the internal and external
specified by the numeric format conversion code.

13-S

forms

the

and

data

FORMAT STATEMENT
Table 13-3
Numeric Field Codes
Internal Form

Conversion
Code

External Form

Binary floating-point
double-precision

Decimal floating-point with D
exponent

Binary floating-point

Decimal floating-point with E
exponent

Binary floating-point

Decimal fixed-point

Binary integer

Decimal integer

Binary word

Octal value

One of the following:
single-precision
binary floating-point,
binary integer, binary
logical, or binary
complex

Single-precision decimal
floating-point, decimal
integer, logical (T or F), or
complex (two decimal
floating-point numbers),
depending upon the internal
form

Complex quantities transfer as two independent real quantities.
The
format specification for
complex quantities consists of either two
successive real field descriptors or one repeated
real
field
descriptor.
For example, the statement
FORMAT(2E15.4,2(F8.3,F8.5))
may transfer up to three complex quantities.
The equivalent of the foregoing statement is
FORMAT(E15.4,E15.4,F8.3,F8.5,F8.3,F8.5)

13.2.2

Interaction of Field Descriptors With I/O Variables

The execution of an I/O statement that specifies a
formatted data
transfer operation initiates format control. The actions performed by
format control depend on information provided by the elements of the
I/O statement's list of variables and the field descriptors that
comprise the referenced FORMAT statement's format specifications.
In processing each FORMAT controlled I/O statement that has an I/O
list,
FORTRAN scans the contents of the list and the format
specific?tions in step.
Each time another variable or array element
name is obtained from the list,
the next field specification is
obtained from the format specification.
If the end of the format
specification is reached and more items remain in the list, a new line
or record is established and the scan process is restarted, either at
the first item in the format specification or, if parenthesized, sets
of format specifications exist within the format specification, with
the last set within the format specification.

13-6

FORMAT STATEMENT
When the I/O list is exhausted, control proceeds to the next statement
in the program, but not before the FORMAT statement is scanned either
to its end or to the next variable transfer format descriptor.
(That
is,
the FORMAT statement is scanned past slashes, literal constants,
Hollerith field descriptors, and spacing descriptors,
but not past
data field descriptors.)
A record is terminated by one of the following:
1.

a slash in the FORMAT specification

the delimiting right parentheses, ), of the FORMAT statement

a lack of items in the I/O list

a lack of Hollerith
FORMAT statement

literal

field

descriptors

the

On input, an additional record is read only when a single slash, /, is
encountered in the FORMAT statement. A record is skipped when two
slashes, //, are encountered or a slash is followed by the end of the
FORMAT statement.
If the FORMAT statement finishes a record by a
slash or the end of the FORMAT statement, any data left in the input
record is ignored.
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.
is
On output, an additional record is written only when a slash, /,
encountered in the FORMAT statement.
If a pair of consecutive
slashes, //, or a single slash followed by the end of the FORMAT
statement is encountered, an empty record.is written.

13.2.3

G, General Numeric Conversion Code

You may use the G conversion code in field descriptors for the format
control of real, double-precision, integer, logical, or complex data.
With the exception of real and double-precision data,
the type of
converSlon performed by a type G field descriptor depends on the type
of its corresponding I/O list variable.
In the case of real and
double-precision data, the kind of conversion performed is a function
of the external magnitude of the datum being transferred.
Table 13-4
illustrates the conversion performed for various ranges of magnitude
(external form) of real and double-~recision data.

13.2.4

Numeric Fields with Scale Factors

You may add scale factors to 0, E, F, and G conversion codes in
descriptors.
The scale factor has the form

field

nP
where n is a signed integer (+ is optional)
and P identifies the
operation.
When used,
a scale factor is added as a prefix to field
descriptors.

13-7

FORMAT STATEMENT
Examples
-2PFIO.5
IPE8.2
When you add a scale factor to an type F field descriptor (or
type G
if the external field
is a
fixed point decimal) a power of 10 is
specified so that
External Form of Number

= (Internal Form)*lO**(scale factor)

For example, assuming the data involved to be the real number
the field descriptor

26.451,

F8.3
produces the external field
,k>J2f26.451
Table 13-4
Descriptor Conversion of Real and Double-Precision
Data According to Magnitude
Equivalent Method of
Conversion Performed

Magnitude of Data in
External Form (M)

F (w-4) . d , 4X

0.1 M~l
1 M~lO

F (w- 4) • (d -1) ,4 X

10d-2 M<lOd-l
10d-l M<lOd
ALL OTHERS

F (w-4) .1, 4X
F(w-4).0,4X
Ew.d
NOTE

In all numeric field conversions,
the
field width
(w)
you specify should be
large enough to include the decimal
point,
slgn,
and exponent character in
addition to the number of digits.
If
the specified width is too 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.

13-8

FORMAT SThTEMENT
The addition of the scale factor of -IP
-lPF8.3
produces the external field
~~~2.645

When you add a scale factor to D, E,
and G (external field not a
decimal fixed-point) type field descriptors, it 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 specify no scale factor, it is understood to be zero.
Once
you specify a scale factor, however, it holds for all subsequent types
D, E, F, and G field descriptors within the same format specification
unless another scale factor is specified. A scale factor is reset to
zero when you specify a scale factor of zero. Scale factors have no
effect on I and 0 type field descriptors.
When you add a scale factor to a D or E field 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, assuming the data involved to be the real number
the field descriptor

12.493,

Ell.3
produces the external field
~~0.125E+02

The addition of the scale factor 2P
2PEll.3
produces the external field
~~12.49E+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
13-9

FORMAT STATEMENT
d+l significant digits).
If n~d+2,
digits and n-d-l insignificant trailing
decimal point.

there will be d+l significant
zeros on the left of the

If the data to be printed is 12.493, these formats produce results
follows:
FORMAT

OUTPUT

SIGNIFICANT
DIGITS

REASON

ElS.3
IPElS.3
-lPElS.3
2PElS.3
-3PElS.3
4PElS.3
6PElS.3

~~~~~~0.12SE+02
~~~~~~1.249E+Ol
~~~~~~.012E+03
~~~~~~12.49E+00
~~~~~~0.OOOE+05
~~~~~~1249.E-02
~~~~124900.E-04

3
4
3
4

n=O
n<d+2
-d<n
n<d+2

n~-d

4
4

n<d+2

13.2.S

n~d+2

Logical Field Descriptors

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 input/output list.
On input, the first non-blank character in the logical data field must
be T or F,
the value of the logical variable is stored in the list
variable as true or false, respectively.
If the entire 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.

13.2.6

the

Variable Numeric Field Widths

Several of the conversion codes are acceptable in FORMAT statements
without
field
width
specifications,
the w.d portion of the
specification so that can be omitted(l).
On input, the conversion codes D, E, F, G, I, L, and 0 are acceptable
without field width specifications. The field begins with the first
non-blank character encountered and ends with the first illegal
character
in the given field.
(Blanks and tabs also terminate a
field.)
Note that for conversion code L (logical
data),
all
consecutive alphabetics following a T (true)
or an F (false) are
considered part of the field and are ignored.
In succeeding fields
the
input
stream
is scanned until a non-blank character is
encountered.
If the character is a comma
(,),
the next field is
skipped,
and the following input field begins with the character
following the comma. Any character other than a comma is assumed to
be the first character in the next input field.
Null fields are

1. If d is given, w must also be specified.
13-10

FORMAT STATEMENT
denoted by successive commas optionally separated by blanks or tabs.
A null field
is equivalent to a fixed-field input of blanks.
For
example, the source code
READ 1, X, Y, Z, L, I, J
1 FORMAT (3F, L, I, A3)
with data as follows
,1.OE+S"TRUEXXXl~~~~ABC

results in
X
Y

Z
L

I
J

0.0
1.OE+S
0.0
TRUE
1
'ABC'

Note that if a comma is included in the input data after the XXXI
before the blanks, i.e., the data is
,1.OE+S "
then J

and

TRUEXXXl,~~~~ABC

= '~~~'

I,
L, 0, and Rare
On output, the format codes A, D,
E,
F, G,
acceptable without field width specifications. The following defaults
are assumed:
Format Code

Assumed Default

A single-precision
A double-precision

AS
AIO
02S.18
ElS.7
FlS.7
GlS.7
G2S.18
lIS
LIS
OIS
RS
RIO

o
E
F

G single-precision
G double-precision
I

R single-precision
R double-precision

13.2.7

Alphanumeric Field Descriptors

You may accomplish the formatted transfer of alphanumeric data in a
manner similar to the formatted transfer of numeric data by use of the
field descriptors Aw and Rw, where A and R are the control characters
and w is the number of characters in the field.
The A and R descriptors both transfer alphanumeric data into or from a
variable in an input/output list depending on the I/O operation. A
list variable may be of any type.
For example,
READ (6,S) V
S FORMAT (A4)

13-11

FORMAT STATEMENT
causes four alphanumeric characters to be read from unit 6 and
in the variable v.

stored

The A descriptor deals with variables containing left-justified,
blank-filled characters;
the R descriptor deals with variables
containing right-justified, zero-filled characters.
The following
paragraphs summarlze the result of alphanumeric data transfer (both
internal and external representations) using the A and R descriptors.
These paragraphs assume that w represents the field width and m
represents the total number of characters possible in the variable.
Double precision variables contain 10 characters (m=lO); all other
variables contain 5 (m=5).
A Descriptor
1.

INPUT, where w ~ m -- The rightmost m characters of the field
are read in and stored left-justified and blank-filled in the
associated variable.

INPUT, where w < m A l l w characters are read in and stored
left-justified and blank-filled in the associated variable.

m
characters
are
output
and
OUTPUT, where w ~ m
right-justified in the field. The remainder of the field is
blank-filled.

OUTPUT, where w < m -- The left
associated variable are output.

most

characters

the

R Descriptor
1.

INPUT, where w ~m -- The right most m characters of the field
are read in and stored right-justified, zero-filled in the
associated variable.

INPUT, where w < m -- All w characters are read in and stored
right-justified, zero-filled in the associated variable.

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 right most
associated variable are output.

13.2.8

characters

the

Transferring Alphanumeric Data

You may transmit alphanumeric data directly into or from the FORMAT
stat:,eme,Ilt9Y ..' t:wo di~ferent . methods: .' ... H~conversion, r:QI">Ili~~"u~~~(),f]
• si ngTe' .9~U 0 t e s ',"1 .wEi: w;w.,:a" .~ f~f ~ y~ii'~'> ~ rEird~~dei~ ciI~ ~ 0 r . :
In H-conversion, the alphanumeric string is specified in the form nH,
where H is the control character and n is the total number of
characters (including blanks) in the string. For example, you may use
the following statement sequence to print the words PROGRAM COMPLETE
on the device LPT:
PRINT 101
101 FORMAT (17H~PROGRAM~COMPLETE)

13-12

FORMAT STATEMENT
Read and write operations of this type are initiated by I/O statements
that reference a format statement and a logical device, but do not
contain an I/O list (see preceding example).
write transfers from a FORMAT statement cause the contents of the
statement field descriptor to be output to a specified logical device.
The contents of the field descriptor, however, remain unchanged.
Read transfers with a FORMAT statement cause the contents of the field
descriptors involved to be replaced by the characters input from the
specified logical device.
Alphanumeric data is stored in a field descriptor left-justified.
If
the data input into a field has fewer characters than the field,
trailing blanks are added to fill the field. If the data input is
larger than the field of the descriptor, the excess rightmost
characters are lost.
Examples
WRITE (1,101)
101 FORMAT (l7H~PROGRA~COMPLETE)
cause the string PROGRAM COMPLETE to be output to the file
1.

device

Assuming the string START on device 1, the sequence
READ (1,101)
101 FORMAT (l7H~PROGRAM~COMPLETE)
would change the contents of statement 101 to
101 FORMAT

(17HSTART~~~~~~~~~~~~)

The foregoing functions may also be accomplished by a literal field
descriptor consisting of the desired character string enclosed within
apostro~hes, i.e., 'string'.
For example, you may use the descriptors
101 FORMAT

(l7H~PROGRAM~COMPLETE)

101 FORMAT

('~PROGRA~COMPLETE')

and

in the same manner.
The result of literal conversion is the same as H-conversion.
On
input, the characters between the apostrophes are replaced by input
characters, and on output, the characters between the apostrophes
(including blanks) are written as part of the output data.
An apostrophe character within a literal field should be represented
by two successive apostrophe marks; otherwise, the statement will not
compile. For example, the statement sequence
50 FORMAT ('DON' 'T')
PRINT 50
will compile and will cause the word DON'T to be output
printer. The statement
50 FORMAT ('DON'T')
however, will cause a compile error.
13-13

the

line

FORMAT STATEMENT
13.2.9

Mixed Numeric and Alphanumeric Fields

You may place an alphanumeric field descriptor among other
the format.
For example, you may use the statement:
FORMAT

fields

(I4,7H~FORCE=FlO.5)

to output the line:
~~22~FORCE=~~l7.6890l

You may omit the separating comma after an alphanumeric format
as shown in the foregoing statement.

field,

When you omit a comma delimiter from a format specification,
format
control associates as much information as possible with the leftmost
of the two field descriptors.

13.2.10

Multiple Record Specifications

To handle a group of input/output records where different records have
different field descriptors,
use a slash to indicate a new record.
For example, the statement
FORMAT (308/I5,2F8.4)
is equivalent to
FORMAT (308)
for the first record, and
FORMAT (I5,2F8.4)
for the second record.
You may
appear
written
middle
records

omit separating commas when you use a slash. When n slashes
at the end or beginning of a format, n blank records will be
on output or skipped on input. When n slashes appear in the
of a
format,
n-l blank records are written on output or n-l
skipped on input.

Both the slash and the closing parenthesis at the end of the format
indicate the termination of a record.
If the list of an input/output
statement dictates that the transmission of data is to continue after
the closing parenthesis of the format
is reached, the format is
repeated, starting with:
1.

that group repeat specification terminated by the last
parenthesis of the next lower level group, or

level zero if no higher level group exists.

Thus, the statement

! /-vet

FORMAT (F7.2, (2{E15.5,E15.4) ,17»

leveI6'

level 1

level 2
13-14

right

FORMAT STATEMEN'l'
causes the format
2(E15.5,E15.4) ,17
to be used after the first record.
As a further example, consider the statement
FORMAT (F7.2/(2(E15.5,E15.4) ,17))
The first record has the format
F7.2
and the next 5 records have the format
2( E15 . 5 , E 15 . 4) ,17

13.2.11

Record Formatting Field Descriptors

You may use two field descriptors, Tw and nX, to position data
a record.

within

You may use the field descriptor Tw to specify the character position.
(external form) in which a record begins.
In the Tw field descriptor, .
the letter T is the control character, and w is an unsigned integer
constant that specifies the character position, in a record, where the;
transfer of data is to begin.
When the output is printed, w
corresponds to the (w-l)th print position, since the first character>
of the output buffer is a forms control character and is not printed.
It is recommended that the first field specification of the output
format be IX, except where a forms control character is used.
.
NOTE
Two successive T field specifications
in
the
second field
will
result
field
if
the
overwriting the first
fields overlap.
Examples
The statement sequence
PRINT 2
2 FORMAT (T50, 'BLACK',T30, 'WHITE')
causes the following line to be printed
WHITE

(print position 29)

BLACK

(print position 49)

The statement sequence
1 FORMAT (T35, 'MONTH')
READ (2,1)

13-15

FORMAT STATEMENT
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~~~XYZ

is read with the format specification
10 FORMAT (T7,A3,Tl,A3)
then the characters XYZ and ABC are read in that order.
You may 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).
Example
The statement
FORMAT

(5H~STEP,I5,lOX,2Hy=,F7.3)

may be used to print the line

13.2.12

$ Format Descriptor

A $ format descriptor at the end of an output FORMAT is used to
suppress the carriage return at the end of the current record.
It is
mainly used on terminal output but will work on non-terminal devices.
A $ format descriptor is ignored in input FORMATs and has no effect if
embedded in an output FORMAT.
The $ format descriptor must be the
next format descriptor to be processed when the corresponding output
list is exhausted for the $ descriptor to have the defined effect.

13.3

CARRIAGE CONTROL CHARACTERS FOR PRINTING ASCII RECORDS

You may use the first character of an ASCII record to control the
spacing operations of the line printer or Teletype terminal printer
unit on which the record is being printed.
Specify the control
character desired by beginning the FORMAT field specification for the
ASCII record to be output with IHa .•. where a is the desired control
character.
Table 13-5 describes the control characters permitted in
FORTRAN-20 and the effect each has on the printing device.

13-16

FORMA'l' S'l'ATEMENT
Table 13-5
FORTRAN-20 Print Control Characters
FORTRAN Character

Printer Character

Effect

Octal Value

space

012

Skip to next
line with form
feed after
60 lines

o zero

LF,LF

012

Skip a line

lone

014

Form feed - go
to top of next
page

+ plus

Suppress
skipping overprint the
line

* asterisk

DC3

023

Skip to next
line with no
form feed

- minus

LF,LF,LF

012

Skip two lines

2 two

DLE

020

Space 1/2 of a
page

3 three

013

Space 1/3 of a
page

/ slash

DC4

024

Space 1/6 of a
page

period

DC2

022

Triple space
with a form
feed after
every 20 lines
printed

, comma

DCl

021

Double space
with a form
feed after
every
30 lines
printed

NOTE
Printer control characters DLE, DCl,
DC2, DC3, and DC4 affect only the line
printer.

13-17

CHAPTER 14
DEVICE CONTROL STATEMENTS

14.1

INTRODUCTION

You may use the following
source programs:
1.

REWIND

, 2.

UNLOAD

BACKSPACE (1)

ENDFILE

, 5.

SKIPRECORD (1) :

SKIPFILE

BACKFILE

device

control

statements

FORTRAN-20

The general form of the foregoing device control statements is
keyword u
keyword (u)
where
keyword
u

is the statement name
is the FORTRAN logical device number
10-1)

(Chapter 10, Table

The operations performed by the device control statement are normally
used only for magnetic tape devices (MTA).
In FORTRAN-20, however,
the device control operations are simulated for disk devices.
The following paragraphs describe the
control statements.

14.2

form

and

use

the

device

REWIND STATEMENT
Form:

REWIND u

Use:

Move the file contained by device u to its initial
(load)
point.
If the medium is already at its load
point, this statement has no effect.
Subsequent READ

1. The results of these commands are unpredictable when used on
directed and NAMELIST-controlled data.
14-1

list-

DEVICE CONTROL STATEMENTS
or WRITE statements that reference device u will
transfer data to or from the first record located on
the medium mounted on device u.
Example:

14.3

14.4

REWIND 16

UNLOAD STATEMENT
Form:

UNLOAD u

Use:

Move the medium contained on device u past its load:
point until it has been completely rewound onto the i
source reel.

Example:

UNLOAD 16

BACKSPACE STATEMENT
Form:

BACKSPACE u

Use:

Move the medium contained on device u to the start of
the record that precedes the current record.
If the
preceding record prior to execution of this statement
was an endfile record, the endfile record becomes the
next record after execution.
If the current record is
the first record of the file, this statement has no
effect.
NOTE
You cannot use this statement for
files
set up for random access, list-directed,
or NAMELIST-controlled I/O operations.

Example:

14.5

BACKSPACE 16

END FILE STATEMENT
Form:

END FILE u

Use:

Write an endfile record in the file located on device
u. The endfile record defines the end of the file that
contains it.
If an endfile record is reached during an
I/O operation initiated by a statement that does not
contain an END= option, the operation of the current
program is terminated.

Example:

END FILE 16

14-2

DEVICE CONTROL STATEMENTS
14.6

SKIP RECORD STATEMENT
Form:

SKIP RECORD u

Use:

In accessing the file located on
record
immediately
following
accessed) record.

skip the
device u,
the
current
(last

NOTE
You cannot use this statement for files set
for
random
access,
list-directed,
NAMELIST-controlled I/O operations.
Example:

14.7

14.8

up
or

SKIP RECORD 16

SKIP FILE STATEMENT
Form:

SKIP FILE u

Use:

In accessing the medium located on unit U ,
skip the
file immediately following the current (last accessed)
file.
If there is no file after the current file,
an
error will occur.

Example:

SKIP FILE 01

BACKFILE STATEMENT
Form:

BACKFILE u

Use:

Move the medium mounted on device u to the start of the
file that precedes the current (last accessed) file.
If there is no file before the current file, completion
of the last operation will move the medium to the start
of the first file on the medium.

Example:

14.9

BACKFILE 20

SUMMARY OF DEVICE CONTROL STATEMENTS

Table 14-1 summarizes the form and use of device control statements.

14-3

DEVICE CONTROL STATEMENTS
Table 14-1
Summary of FORTRAN-20 Device Control Statements
Use

Statement Form
REWIND u
UNLoAb"u
END FILE u
SKIP RECORD u
SKIP FILE u
BACKFILE u
BACKSPACE u

Rewi~d medium to its loadpoi~t
Rewind medium on't:o' its s'olirce reel
Write an endfile record into the curient ,file,
Skip the next record
Skip the next file
Move medium backwards one file
Move medium back 6ne recoid

14-4

CHAPTER 15
SUBPROGRAM STATEMENTS

15.1

INTRODUCTION

Procedures you use repeatedly in a program may be written once and
then referenced each time you need the procedure.
Procedures that may
be referenced are either internal (written and contained within the
program in which they are referenced) or external (self-contained
executable procedures that may be compiled separately). The kinds of
procedures that may be referenced are:
1.

statement functions,

intrinsic functions

external functions, and

subroutines.

(FORTRAN-defined functions),

The first three of the foregoing categories are
referred
to
collectively as functions or function procedures;
procedures of the
last category are referred to as subroutines or subroutine procedures.

15.1.1

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.
Functions and subroutines use dummy arguments to indicate the type of
the actual arguments they represent and whether the actual arguments
are variables, array elements, 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, array element,
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.
(Examples
of argument lists are given in the following paragraphs.)
The position, number, and type of each dummy argument in a subprogram
list must agree with the position, number, and type of each argument
in the argument list of the subprogram reference.

15-1

SUBPROGRAM STATEMENTS
Dummy arguments may be:
1.

variables,

array names,

subroutine identifiers,

function identifiers, or

statement labei identi"fiers that are den()ted
"*", "$", or u&".

the

sym"S"of

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 literal 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.

15-2

SUBPROGR~M

15.2

STATEMENTS

STATEMENT FUNCTIONS

Statement functions define an internal subprogram in
statement. The general form of a statement function is:

single

name(argl,arg2, ... ,argn)=E
where
name

is a name you assign that consists of one to six
characters.
The name you use must conform to the
rules for symbolic names given in Section 3.3.
The type of a statement function is determined
either by the first character of its name or by it
being explicitly declared in a type statement.

(argl ... argn)

represents a list of dummy arguments.

is an arbitrary expression.

The expression E of a statement function may be any legitimate
arithmetic expression that may use the g1ven dummy arguments and
indicates how they are combined to obtain the desired value.
You may
use the dummy arguments as variables or indirect function references;
but you cannot use them as arrays. The dummy argument names bear no
relation to their use outside the context of the statement function
except for their data
type.
The
expression
may
reference
FORTRAN-defined
functions
(Section 15.3) or any other defined
statement function,
or call an external function.
It may not
reference any function that directly or indirectly references the
given statement function or any subprogram in the chain of references.
That is,
recursive references are not allowed. Statement functions
produce only one value, the result of the expression that it contains.
A statement function cannot reference itself.
You must define all statement functions within a program unit before
the first executable statement of the program unit. When used, the
statement function name must be followed by an actual argument list
enclosed within parentheses and may appear in any arithmetic or
logical expression.
Examples:
SSQR(K)=(K*(K+l)*2*K+l)/6
ACOSH(X)=(EXP(X/A)+EXP(-X/A»/2.0

15.3

INTRINSIC FUNCTIONS (FORTRAN DEFINED FUNCTIONS)

Intrinsic functions are subprograms supplied by FORTRAN. Reference an
intrinsic function by using its name as an operand in an expression.
The name always refers to the intrinsic function unless it is preceded
by an asterisk (*) or ampersand (&) in an EXTERNAL statement, declared
in a conflicting explicit type statement, or specified as a routine
dummy parameter.
Table 15-1 describes FORTRAN-20 intrinsic functions
and
their
arguments. Notice that octal constants are not allowed as arguments.

15-3

Table 15-1
Intrinsic Functions (FORTRAN-20 Defined Functions)
Mnemonic

Function

I-'
U1

I
~

Absolute value:
Real
Integer
Double- precision
Complex to real

ABS*
IABS*
DABS*
CABS

Conversion:
Integer to real
Real to integer

FLOAT*
IFIX*

Definition

Number of
Arguments

arg
arg
arg
c=(x**2+Y**2)**(1/2)

1
1
1
1

Real
Integer
Double
Complex

Real
Integer
Double
Real

1
1

Integer
Real

Real
Integer

Sign of arg *
largest integer
arg
~

til

c::
~

tU
!:O

o
1
1
1
1

SNGL
DBLE*
DFLOAT
REAL*

Double to real
Real to double
Integer to double
Complex to real
(obtain real part)
Complex to real
(obtain imaginary
part)
Real to complex

Type of
Argument
Functl.on

Double
Real
Integer
Complex

Real
Double
Double
Real

til

t-3

AIMAG

Complex

til

Real

3:
til

Z
t-3

Truncation:
Real to real

CMPLX*

c=Arg + i*Arg

Real

Complex

AINT

Sign of arg*
largest integer
arg
~

Real

1
1

Real
Double

Integer
Integer

INT*
IDINT

Real to integer
Double to integer

til

* In line functions.
--------

---

Table 15-1 (Cont.)
Intrinsic Functions (FORTRAN-20 Defined Functions)
Function

Remaindering:
Real
Integer
Double- precision

Mnemonic

AMOD
MOD*
DMOD

Definition

{The remainder

when Arg 1 is
divided by Arg 2

Number of
Arguments

}

Type of
Argument
Functlon

2
2
2

Real
Integer
Double

>1
>1
>1
>1
>1

Integer
Real
Integer
Real
Double

Real
Real
Integer
Integer
Double

Maximum value:

til

AMAXO
AMAXl*
MAXO*
MAXI
DMAXI

I-'

!Max (Argl ,Arg2, • • . )1

I
Ul

c::
OJ

~
G)

3!
((J

Minimum Value:

1-3

AMINO
AMINl*
MINO*
MINI
DMINI

!Min(Ar g l'Ar g 2'···)1

Transfer of Sign:
Real
Integer
Double precision

SIGN*
ISIGN
DSIGN

Positive Difference:
Real
Integer

DIM*
IDIM

>1
>1
>1
>1
>1

Integer
Real
Integer
Real
Double

Real
Real
Integer
Integer
Double

{Si gn(Ar g 2) * Argl}

2
2
2

Real
Integer
Double

{Ar g l-Min(Ar g l,Ar g 2)}

2
2

Real
Integer

* In line functions.
-----

I-J
ttJ
3:
ttJ

1-]

til

SUBPROGRAM STATEMENTS
15.4

EXTERNAL FUNCTIONS

External functions are function subprograms that consist of a FUNCTION
statement followed by a sequence of statements that define one or more
desired operations;
subprograms of this .type may contain one or more
RETURN statements and must be terminated by an END statement.
Function subprograms are independent programs that may be referenced
by other programs.
The FUNCTION statement that identifies an external
form:

function

has

the

type FUNCTION name (argl,arg2, ... ,argn)
where
type

is an
optional
type
specification
as
described in Section 6.3.
These include
INTEGER, REAL, DOUBLE PRECISION, COMPLEX or
LOGICAL (plus the optional size modifier, *n,
for compatibility with other manufacturers.)

name

is the name you assign to the function.
The
name
may
consist
of from one to six
characters,
the first of which must
be
alphabetic.
You may include the optional
size modifier (*n) with the name if the type
is specified.
(Refer to Section 6.3.)

(argl, .•. ,argn)

is a list of dummy arguments.

If you omit type in the FUNCTION statement, the type of the function
may be assigned, by default, according to the first character of its
name, or may be specified by an IMPLICIT statement or by an explicit
statement given with the subprogram itself.
Note that if you want to use the same name for a user-defined function
and the name of a FORTRAN-20 defined function (library basic external
function), the desired name must be declared in an EXTERNAL statement
and prefixed by an asterisk (*) or ampersand (&) in the referencing
routine.
(Refer to Section 6.7 for a description of the EXTERNAL
statement.)
The following rules govern the structuring of a FUNCTION subprogram:
1.

You must use the symbolic name assigned a FUNCTION subprogram
as a variable name in the subprogram.
During each execution
of the subprogram, this variable must be defined and, once
defined, may be referenced or redefined. The value of the
variable at the time of execution on any RETURN statement is
the value of the subprogram.
NOTE
A RETURN statement returns control to the calling
statement
that
initiated the execution of the
subprogram. See Section 15.6 for a description of
this statement.

15-6

SUBPROGP~M

SThTEMENTS

You may 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 may not appear
in any EQUIVALENCE,
COMMON, or DATA statement used within the subprogram.

The function subprogram may define or redefine one or more of
its arguments so as to effectively return results in addition
to the value of the function.

The function subprogram may contain any FORTRAN statement
except BLOCK DATA,
SUBROUTINE PROGRAM,
another FUNCTION
statement, or any statement that directly or
indirectly
references the function being defined or any subprogram in
the chain of subprograms leading to this function.

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 may have more than one RETURN statement.
The END statement specifies the
subprogram and implies a return.

15.4.1

physical

end

the

Basic External Functions (FORTRAN-20 Defined Functions)

FORTRAN-20 contains a group of predefined external functions that are
called basic functions.
Table 15-2 describes each basic function, its
name, and its use. These names always refer to the basic external
functions unless declared in an EXTERNAL or conflicting explicit type
statement.

15.4.2

Generic Function Names

The compiler generates a call to the proper FORTRAN-20 defined
function,
depending on the type of the arguments, for the following
generic function names:
ABS
AMAXI
AMINI
ATAN
ATAN2
COS
INT
MOD
SIGN
SIN
SQRT
EXP
ALOG
ALOGIO
In the following example
K=ABS(I)
15-7

SUBPROGRAM STATEMENTS
the type of I determines which function is called.
If I
is an
integer,
the compiler generates a call to the function lABS.
If I is
real, the compiler generates a call to the function ABS.
If I
is
double precision, the compiler generates a call to the function DABS.
The function name loses its generic properties if it appears in an
explicit type statement,
if it is specified as a dummy routine
parameter, or if it is prefixed by n*n or
n&" in an EXTERNAL
statement. When a generic function name that was specified unprefixed
in an EXTERNAL statement is used as a routine parameter, it is assumed
to reference a FORTRAN-20 defined function of the same name, or if
none exists, a user-defined function.
Note that IMPLICIT statements
have no effect upon the data type of generic function names unless the
name has been removed from its class by use of an EXTERNAL statement.

15.5

SUBROUTINE SUBPROGRAMS

A subroutine is an external computational procedure that is identified
by a SUBROUTINE statement and mayor may not return values to the
calling program.
The SUBROUTINE statement used to identify
a
subprogram of this type has the form:
SUBROUTINE name(argl,arg2, ... ,argn)
where
name

is the symbolic name of the subroutine to
defined.

(argl, ... ,argn)

is an optional list of dummy arguments.

15-8

Table 15-2
Basic External Functions (FORTRAN-20 Defined Functions)
Function

Exponential:
Real
Double
Complex
Logarithm:
Real
Double
I--'
Ul

I
\D

Complex
Square Root:
Real
Double
Complex

Number of
Arguments

Type of
Funct~on
Argument

EXP
DEXP
CEXP

t9}

1
1
1

Real
Double
Complex

ALOG
ALOGIO
DLOG
DLOGIO
CLOG

In (Arg)
log
(Arg)
In (Arg)
log
(Arg)
In (Arg)

1
1
1
1
1

Real
Real
Double
Double
Complex

UJ
C
ttl
It!

G'l

SQRT*
DSQRT
CSQRT

(Arg)**1/2
(Arg)**1/2
(Arg)**1/2

1
1
1

Real
Double
Complex

t-3

:tJ

t-3

j:l1

Z
1-3

Sine:
Real (radians)
Real (degrees)
Double (radians)
Complex

SIN*
SIND
DSIN
CSIN

Cosine:
Real (radians)
Real (degrees)
Double (radians)
Complex

COS*
COSD
DCOS
CCOS

*Generic functions

Definition

Mnemonic

{Sin (Ar 9 )}

1
1
1
1

Real
Real
Double
Complex

{COS (Ar 9 )}

1
1
1
1

Real
Real
Double
Complex

(fl

Table 15-2 (Cont.)
Basic External Functions (FORTRAN-20 Defined Functions)
Function

I-'
U1

I
I-'

Mnemonic

Definition

Hyperbolic:
Sine
Cosine
Tangent

SINH
COSH
TANH

sinh(Arg}
cosh(Arg)
tanh(Arg}

Arc sine

ASIN

Arc cosine

ACOS

Number of
Arguments

Type of
Argument
I Function

Real
Real
Real

asin(Arg}

Real

acos(Arg}

Real

til

III
td

Arc tangent
Real
Double
Two REAL arguments
Two DOUBLE arguments

ATAN*
DATAN
ATAN2*
DATAN2

atan(Arg}
datan(Arg}
atan(Argl/Arg2}
atan(Argl/Arg2}

1
1
2
2

Real
Double
Real
Double

Complex Conjugate

CONJG

Arg=X+iY,CONJG=X-iY

Complex

Random Number

RAN

Result is a random
number in the range
of 0 to 1.0

Real

1 Dummy
Argument

Integer,
Real,
Double,
or Complex

~
(j)

~
til

1-3

*Generic functions

1-3
t%j

1-3
til

SUBPROGRAM STATEHENTS
The following
subprogram:

rules

control

the

structuring

subroutine

You may not use the symbolic name of the subprogram in any
statement within the defined subprogram except the SUBROUTINE
statement itself.

You may not use the given dummy arguments in an EQUIVALENCE,
COMMON, or DATA statement within the subprogram.

The subroutine subprogram may define or redefine one or
of its arguments so as to effectively return results.

The subroutine subprogram may contain any FORTRAN statement
except BLOCK DATA, FUNCTION, 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.

Dummy arguments that represent statement labels may be either
an *, $, or &.

The subprogram should contain at least one RETURN statement
and must be terminated by an END statement.
The RETURN
statements indicate the logical end of a computational
routine;
the END statement signifies the physical end of the
subroutine.

Subroutine subprograms may have as many entry points as
desired
(see description of ENTRY statement given in Section
15.7) .

15.5.1

Referencing Subroutines (CALL Statement)

You must reference subroutine subprograms by using a CALL statement of
the following form:
CALL name(argl,arg2, ... ,argn)
where
of

the

name

is the symbolic
name
subroutine subprogram.

(argl, ... ,argn)

is an optional list of actual arguments.
If
the
list is included,
the given actual
arguments must agree in order,
number,
and
type with the corresponding dummy arguments
given in the defining SUBROUTINE statement.

The use of literal constants is an exception to the rule
agreement of type between dummy and actual arguments.
argument in a CALL statement may be:
1.

a constant

a variable name

15-11

desired

requiring
An actual

SUBPROGRAM STATEMENTS
3.

an array element identifier

an array name

an expression

the name of an external subroutine, or

a statement label.

Example:
The subroutine
SUBROUTINE MATRIX (I, J , K, M':*)

END
may be referenced by
CALL MATRIX(lO,20,30,40,§lOl)

15.5.2

FORTRAN-20 Supplied Subroutines

FORTRAN-20 provides you with an extensive group of
predefined
subroutines.
Table 15-3 gives the descriptions and names of these
predefined subroutines.

15.6

RETURN STATEMENT AND MULTIPLE RETURNS

The RETURN statement causes control to be returned from a
to the calling program unit.
This statement has the form:
RETURN

(standard return)

RETURN e

(multiple returns)

subprogram

: where e represents anintege~constant,variable! or expression.
The
executi6n of this st~t~m~nt in the first of the friiegoing f6rm~ '(i.e.,
standard return) causes control to be returned to the statement of the
calling program that follows the statement that called the subprogram.
,...

,."

The multiple returns form of this statement, i.e., RETURN e, enables
you to select any labeled statement of the calling program as a return'
point. When the multiple returns form of this statement is executed,'
the assigned or calculated value of e specifies that the return is to;
be made to the eth statement label in the argument list of the calling
,statement.
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,
' l i s t of the calling statement.
If e is less than 1 or is larger than
the number of available statement labels, a standard return operation:
is performed.

15-12

SUBPROGRl\M STl\TEMENTS

NorrE

A dummy argument for a statement label
must be either a *, $, or & symbol.
You may use any number of RETURN (standard return) statements in any
subprogram.
The use of the multiple returns form of the RETURN
statement, however, is restricted to subroutine subprograms.
The
execution of a RETURN statement in a main program will terminate the
program.
Example
Assume the following statement sequence in a main program:

CALL EXAMP(1,$lO,K,$15,M,'$20)

GO TO 101

15-13

SUBPROGRAM STATEMENTS
Assume the following
subprogram:

statement

sequence

the

called

SUBROUTINE

SUBROUTINE EXAMP (L, ;*,M, ;*,N,::X)

RETURN

:RETURN(CjD) .

END
Each occurrence of RETURN returns control to the statement GO
in the calling program.
If, on t'he exe~ution of' th~ RETURN (C/D) stateme'nt,'0 the value of
is:
Less than or equal to:

o
1
2
3

Greater than or equal to:
4

15.6.1

TO
0.

101
(C/I)f*~

The following is performed:
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

101

The following is performed:
a standard return to the GO TO 101
statement is made.

Referencing External FUNCTION Subprogram

Reference an external function subprogram by using its assigned name
as an operand in an arithmetic or logical expression in the calling
program unit. The name must be followed by an actual argument list.
The actual arguments in an external function reference may be:
1.

a variable name,

an array element identifier,

an array name,

an expression,

a statement number, or

15-14

SUI3PROGRAr·1 STATEMENTS
6.

the name of
SUBROUTINE) .

another

external

procedure

FUNCTION

NOTE
Any subprogram name to be used as an
argument
to another subprogram must
first appear in an EXTERNAL statement
(Chapter 6) in the calling program unit.
Example
The subprogram defined as:
INTEGER FUNCTION ICALC(IX,IY,IZ)

RETURN
END
may be referenced in the following manner:

TOTAL=ICALC(IAA,IAB,IAC)+SOO

15.7

MULTIPLE SUBPROGRAM ENTRY POINTS (ENTRY STATEMENT)

FORTRAN-20 provides an ENTRY statement that enables you to specify
additional entry points into an external subprogram.
This statement<
used in conjunction with a RETURN statement enables you to employ only
one computational routine of a subprogram that contains several such
routines.
The form of the ENTRY statement is:
ENTRY name(argl,arg2, ... ,argn)
where
name

is the symbolic name
desired entry point

(argl, ... ,argn)

is an optional list of dummy arguments.
list may contain
1.

variable names,

array declarators,

15-15

assigned

the
This

SUBPROGRAM STATEMENTS
3.

the name of
an
external
(SUBROUTINE or FUNCTION), or

procedure

are
statement label identifiers that
denoted by either a *, $, or & symbol.

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 any number of ENTRY statements in
an external subprogram.

Execution is begun at the first
following 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 may not use an ENTRY statement in a main program or
a subprogram reference itself through its entry points.

have.

You may not use an ENTRY statement in the range of a DO or an
extended DO statement construction.

The dummy arguments in the ENTRY statement need not agree in
order, number, or type with the dummy arguments in SUB~OUTINE
or FUNCTION statements of any other ENTRY statement 1n 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 may not reference a dummy argument unless it appears in
the dummy list of an 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 the function must be
current entry name.

15-16

executable

and

returned

statement

not

use

arguments

the

SUBPROGRAM STATEMENTS

Table 15-3
FORTRAN-20 Library Subroutines
Subroutine Name

Effect
CALL AXIS(X,Y,ASC,NASC,S,THETA,XMIN,DX)

AXIS

AXIS causes an axis with tick marks and scale values
at I-inch increments to be drawn. An identifying
label may also be plotted along the axis. Parameters
X and Y specify the start of the axis. The axis is
plotted, starting at X, Y, at an angle of THETA
degrees for a distance of S inches. The angle THETA
is usually either 0 (X axis) or 90.0
(Y axis).
Characters ASC of array ASC are plotted as a label
for the axis drawn.
If NASC is positive, the tic
marks, label, and scale values are placed on the
counterclockwise side of the axis;
if NASC is
negative, the foregoing items are placed on the
clockwise side of the axis.
Parameter XMIN is the value of the scale at the
beginning of the axis; parameter DX is the change in
scale for a I-inch increment. The values of XMIN and
DX may be determined by subroutine SCALE.
DATE

CALL DATE (array)
This subroutine places today's date as left-justified
ASCII characters into a dimensioned 2-word array.
The date is in the form:
dd-mmm-yy
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, e.g., Mar, and yy is a 2-digit year.
The data is stored in ASCII code, left-justified, in
the two words.

DEFINE FILE

CALL DEFINE FILE (u,s,v,f,pj,pg)
The arguments
follows:
u =

this

subroutine

are

defined

logical device numbers.

the size of the records comprising the file
being defined. The argument s may be an integer
constant or variable.

an associated variable. The associated variable
is an integer variable that is set to a value
that points to the record that immediately
follows
the last record transferred.
This
variable is modified by the FIND statement
(Chapter
10) .
At
the end of each FIND
operation, the variable is set to a value that
points to the record found.
The variable v
cannot appear in the I/O list of any I/O
statement that accesses the file set up by the
DEFINE FILE statement.
15-17

SUBPROGRAM STATEMENTS

Table 15-3 (Cont.)
FORTRAN-20 Library Subroutines
Subroutine Name

Effect

= filename to be given the file being defined.
pj = your project number.
f

your programmer's number.
NOTE
Numbers pj and
Directory.

identify

your

File

Example
The statement
CALL DEFINE FILE (l,lO,ASCVAR, 'FORTFL.DAT',O,O)
establishes a file named FORTFL.DAT on device 01, a
disk,
which
contains
ten
word records.
The
associated variable is ASCVAR, and the file is in
your area.
A DEFINE FILE call can be used to establish and
define the structure of each file to be used for
random access 1/0 operations.
NOTE
The OPEN statement may be used to perform the
same functions as DEFINE FILE.
DUMP

CALL DUMP (L ( I) , U(1) , F (I) , ... , L (n) I U (n) ,F (n) )
DUMP causes particular portions of memory to be
dumped.
L(l) and U(l} are the variable names that
give the limits of memory to be dumped. Either L(l}
or U(l} may be upper or lower limits. F(l) is a
number indicating the format in which the dump is to
be performed: 0 = octal, 1 = real, 2 = integer, and
3 = ASCII.
If F is not 0, 1, 2, 3, the dump is in octal.
If
F(n) is missing, the last section is dumped in octal.
If U(n} and F(n} are missing, an octal dump is made
from L to the end of the job area. If L(n}, U(n),
and F(n) are missing, the entire job area is dumped
in octal.
The dump is terminated by a call to EXIT.

15-18

SUBPROGR~M

STATEMENTS

Tctble lS-3 (Cant.)
FORTRAN-20 Library Subroutines
Subroutine Name
ERRSET

Effect
CALL ERRSET(N)

ERRSET allows you to control
the
typeout
of
execution-time arithmetic error messages. ERRSET is
called with one integer argument.
Typeout of all arithmetic and library error messages
is suppressed after N occurrences of these error
messages. If ERRSET is not called, the default value
of N is 2.
ERRSNS

CALL ERRSNS(I,J)
ERRSNS allows you to determine the exact nature of an
error on READ, WRITE, OPEN, or CLOSE that was trapped
with the "ERR= statement label" option.
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 optional.)
CALL ERRSNS(I,J)
returns a FORTRAN-standardized number in I and a
processor-dependent number in J to describe the last
I/O operation. See Appendix H and Table H-l for more
information and a detailed description of the values
returned.

EXIT

EXIT
EXIT returns control to the Monitor and,
terminates the execution of the program.

ILL

therefore,

CALL ILL
If the flag is set and an
ILL sets the ILLEG flag.
illegal
character
is
encountered in floatingpoint/double- precision input, the corresponding word
is set to zero.

LEGAL

CALL LEGAL
LEGAL clears the ILLEG flag. If the flag is set and
an illegal character is encountered in the floatingpoint/double- precision input, the corresponding word
is set to zero.

LINE

CALL LINE (X,Y,N,K)
LINE causes a line to be drawn through the N points
specified
by (X(l) ,Y(l», (X(2) ,Y(2» ... (X(N) ,Y(N»
where the elements of X and Y are spaced K words
apart in storage.

15-19

SUBPROGRAM STATEMENTS

Table 15-3 (Cont.)
FORTRAN-20 Library Subroutines
Subroutine Name
MKTBL

Effect
CALL MKTBL{I,J)

MKT~L specifies a special character set

where I
is
the number to be assigned the set and J contains the
starting address of a character table of 200(8)
consecutive words.
In each character table word, the
left half contains the number of strokes in the
character
(0 if nothing is to be plotted fOL the
wOLd) and the right half contains the address of the
table of strokes for the character.
NUMBER

CALL NUMBER(X,Y,SIZE,FNUM,THETA,NDIGIT)
NUMBER causes a floating- point number to be plotted
as text.
PaLameters X, Y, SIZE, and THETA have the
same meanings as for
the SYMBOL call.
PaLameter
NDIGIT is the number of digits plotted to the right
of the decimal point.
If NDIGIT is a negative value,
only the integeL part of the number is plotted.
FNUM
specifies the number to be plotted.

PDUMP

CALL PDUMP (L (1) I U ( 1) ,F (1) , •.. I L (n) I U (n) ,F (n)
The arguments of PDUMP are the same as those of
DUMP.
PDUMP is the same as DUMP except that control
returns to the calling program after the dump has
been executed.

PLOT

CALL PLOT(X,Y,IPEN)
PLOT moves the pen in a straight line from its
current position to the position specified by X,Y.
If IPEN=3, the pen is raised before the movement;
if
IPEN=2 the pen is loweLed before movement;
if IPEN=l
the pen is left unchanged from its previous state.
If the value of IPEN is negative (-1, -2 or -3) the
pen action is the same as for
the 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
plotteL.
The plotter is not released
specified movement.

PLOTS

completion

the

CALL PLOTS (I)
PLOTS is the plotter setup routine.
If the plotter
LS
not available,
I
is set to -1;
if it is
available, I is set to O.
This call must be the
first plotter routine called.

15-20

SUBPROGRAH STATm·mNTS
Table 15-3 (Cont.)
FORTRAN-20 Library Subroutines
Subroutine Name
RELEAS

Effect
CALL RELEAS(unit)

RELEAS closes out I/O on a device initialized by the
FORTRAN Object Time System and
returns it to the
uninitialized state.
RELEAS should be the last call
referencing that device.
SAVRAN

CALL SAVRAN(I)
SAVRAN is called with one integer
argument.
SAVRAN
sets
its
argument
to
the last
random number
(interpreted as an integer) that has been generated
by the function RAN.

SCALE

CALL SCALE(X,N,S,XMIN,DX)
SCALE selects scale values for an AXIS call where
X
and N specify a I-dimensional array X with the length
N.
Parameter S specifies the length of
the desired
axis, SCALE determines a value of DX that allows X to
be plotted in S inches.
XMIN
is selected as
the
smallest element of the array X, and is truncated to
be a multiple of DX.

SETABL

CALL SETABL(I,J)
SETABL specifies a
character set where
I
is an
integer
that
gives the number of the desired
character set.
If a character set has
been defined
by I,
the value of J is set to 0;
if not, J is set
to -1.
The standard ASCII character set is defined
as 1.

SETRAN

CALL SETRAN (I)
SETRAN has one argument, which must be a non-negative
integer <2**(31).
The starting value of the function
RAN is set to the value of this argument, unless the
argument is zero.
In this case, RAN uses its normal
starting value.

SORT

CALL SORT ('OUTPUT/SWS=INPUT/SWS,INPUT/SWS')
SORT sorts one or more files using the SORT program.
The argument is
an ASCII
string that represents
(version
3 or
later)
the standard SORT command
string.
Its components are:
OUTPUT
INPUT
SWS

file specification of the output file.
file specification of the input file(s).
one or more switches for the output file,
the input file(s), the sorting process, and
sometimes SCAN.
The switches not allowed
in the
FORTRAN call are:/BLOCK, /COMP3,
/EBCDIC, /INDUSTRY,
/LABEL,
/SIXBIT,
and
/VERSION.

15-21

SUBPROGRAM STATEMENTS
Table 15-3 (Cont.)
FORTRAN-20 Library Subroutines
Subroutine Name

Effect

Wild card format is not allowed in the SORT call.
For information about using the SORT program, see the
SORT Userls Guide. Example:

CALL SORT('SRTFIL.SRT=INSTRT/REC:80/KEY:l:2'}
SYMBOL

CALL SYMBOL(X,Y,SIZE,ASC,THETA,NASC)
SYMBOL raises the plotter pen and moves it to
position specified by X and Y. Lower pen and plot'
characters found in array ASC.
Parameter
~IZE
specifies the height of the characters plotted in
inches (floating- point values); THETA specifies the
direction of the base line on which the text of array
ASC is to be plotted, and NASC specifies the number
of characters in array ASC.'

TIME

CALL TIME(X) or CALL TIME(X,Y)
TIME returns the current time in its argument(s)
in
left-justified ASCII characters.
If TIME is called
with one argument,
CALL TIME(X)
the time is in the form
hh:mm
where hh is the hours (24-hour time) and mm
minutes.
If a second argument is requested,

the

CALL TIME(X,Y)
the first argument is returned in the same form as
the one-argument call, and the second has the form
bss.t
where b is a blank, ss is in seconds,
tenths of a second.
WHERE

and

in.

CALL WHERE(X,Y)
Variables X and Yare set to the values
identify the current position of the pen.

15-22

which

CHAPTER 16
BLOCK DATA SUBPROGRAl>1S

16.1

INTRODUCTION

Use block data subprograms to initialize data to be stored in any
common areas.
You may use only specification and DATA statements,
i.e., DATA, COMMON, DIMENSION, EQUIVALENCE, and TYPE,
in BLOCK DATA
subprograms.
A subprogram of this type must start with a BLOCK DATA
statement.
You may enter initial values into more than one labeled
in a single subprogram of this type.

common

block

An executable program may contain more than one block data subprogram.

16.2

BLOCK DATA STATEMENT

The form of the BLOCK DATA statement is:
BLOCK DATA name
where
name

is a symbolic
subprogram.

16-1

name

given

identify

the

APPENDIX A
ASCII-1968 CHARACTER CODE SET

The character code set defined in the X3.4-l968 Version of the
Information Interchange (ASCII) is
American National Standard for
given in the following matrix.
1st 2
octal
digits

Last octal digit
o
1

OOX
Olx
02x
03x

NUL
BS
DLE
CAN

04x

05x

(
0
8

06x

07x
lOx
llx
l2x
13x
l4x
l5x
l6x
17x

SOH
HT
DCl
EM
!
)
1
9
A
I
Q
Y
a
i
q
y

H
p
X
grave
h
p
x

STX
LF
DC2
SUB

ETX
VT
DC3
ESC

ENQ
CR
NAK
GS

ACK
SO
SYN
RS

BEL
SI
ETB
US

#
+

EOT
FF
DC4
FS
$

;

B
J
R
Z
b
j
r
z

C
K
S
[
c
k
s

D
L
T

E
M
U

d
I
t

e
m
u

{

}

F
N
V
A(t)
f
n
v
-(ESC)

*
2

/
7
?
G
0
W
(*-)
g

Characters inside parentheses are ASCII·1963 Standard.
NUL
SOH
STX
ETX
EOT
ENQ
ACK
BEL
BS
HT
LF
VT
FF
CR
SO
SI

Null
Start of Heading
Start of Text
End of Text
End of Transmission
Enquiry
Acknowledge
Bell
Backspace
Horizontal Tabulation
Line Feed
Vertical Tabulation
Form Feed
Carriage Return
Shift Out
Shift In

DLE
DCl
DC2
DC3
DC4
NAK
SYN
ETB
CAN
EM
SUB
ESC
FS
GS
RS
US
DEL

A-I

Data Link Escape
Device Control 1
Device Control 2
Device Control 3
Device Control 4
Negative Acknowledge
Synchronous Idle
End of Transmission Block
Cancel
End of Medium
Substitute
Escape
File Separator
Group Separator
Record Separator
Unit Separator
Delete (Rubout)

w
DEL

Graphic
subsets
64 95

APPENDIX B
USING THE COMPILER

This appendix explains how to access FORTRAN-20 and how to make use of
the information it provides.
You should be familiar with the
FORTRAN-20 language and the DECsystem-20 TOPS-20 monitor.

B.l

RUNNING THE COMPILER

The command to run FORTRAN-20 is:
@FORTRA
The compiler responds with an asterisk (*) and is then ready to accept
a command string. A command is of the general form:
object filename, listing filename=source filename(s)
You are given the following options:

B.l.l

The filenames can be fully specified with SFD paths.

You may specify more than one input file in the compilation
command string.
These files will be logically concatenated
by the compiler and treated as one source file.

Program units need not be terminated at file
may consist of more than one file.

If no object filename is
file is generated.

If no listing filename is specified, no listing is generated.

If no extension is given, the defaults are
.LST
.REL
(relocatable binary),
and .FOR
(source)
respective files.

specified,

boundaries

relocatable

and

binary

(listing) ,
for
their

Switches Available with FORTRAN-20

Switches to FORTRAN-20 are accepted anywhere in the command string.
They are tQtally position- and file-independent.
Table B-1 lists the
switches.
'.

B-1

USING THE COMPILER
Table B-1
FORTRAN-20 Compiler Switches
Switch

Meaning

Defaults

CROSSREF

Generates a file that can be input to
the CREF program

OFF

DEBUG

(See Section B.l.l.l.)

OFF

EXPAND

Includes the octal-formatted version of
the object file in the listing.

OFF

INCLUDE

Compiles a D in 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

NOERRORS

Does not print error messages
on the terminal.

OFF

NOWARNINGS

Does not output warning messages.

OFF

OPTIMIZE

Performs global optimization.

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.
Example
@FORTRA
*OFILE,LFILE=SFILE/MAC,S2FILE
The /MAC switch will cause the MACRO code equivalent of SFILE.FOR
S2FILE.FOR to appear in LFILE.LST.
All switches used or implied are printed at the top
page.

B-2

each

and

listing

USING THE COMPILER
B.l.l.l The /DEBUG Switch - The /DEBUG switch tells FORTRAN-20
to
compile a series of debugging features into your program.
Several of
these features are specifically designed to be used with FORDDT.
Refer
to Appendix E for more information.
By adding the modifiers
listed in Table B-2, you can include specific de~ugging features.
Table B-2
Modifiers to /DEBUG Switch
Meaning

Modifiers
:DIMENSIONS

Generates dimension information in .REL
FORDDT.

: 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 may be used without
FORDDT.)

:INDEX

Forces DO LOOP indices to be stored at the
beginning of each iteration rather than held in a
register for the duration of the loop.

:BOUNDS

Generates the bounds checking code for
all array
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.)

:NONE

Do not include any debug features.

:ALL

Enable all debugging aids.

file

for

The format of the /DEBUG switch and its modifiers is as follows:
/DEBUG:modifier
or
/DEBUG: (modifier list)
Options available with the /DEBUG modifiers are:
1.

No debug features - Either do not specify the
or include /DEBUG:NONE.

All debug features - Either /DEBUG or /DEBUG:ALL.

Selected features - Either a
i.e.,
/DEBUG:BOU/DEBUG:LAB
or a list of modifiers
/DEBUG: (BOU,LAB, •.. )

B-3

series

/DEBUG

modified

switch

switches;

USING THE COMPILER
4.

Exclusion of features
(if you wish all but one or
modifiers and do not wish to list them all, you may use
prefix "NO" before the switch you wish to exclude).
exclusion of one or more features implicitly includes all
others,
i.e.,
/DEBUG:NOBOU
is
the
same
/DEBUG: (DIM,TRA,LAB,IND).

two
the
The
the
as

If you include more than one statement on a single line, only the
first statement will receive a label
(/DEBUG:LABELS)
or FORDDT
reference (/DEBUG:TRACE).
(The /DEBUG option and the /OPTIMIZE option
cannot be used at the same time.)
NOTE
If a source file contains line sequence
numbers that occur more than once in the
same subprogram,
the
/DEBUG
option
cannot be used.
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
instruction for
some
index of the loop.

inner
of the

loop
plus
one
references to the

:BOUNDS

For each array,
DIMENSIONS:.

formula

the

same

For each reference to an array element,
use 5+N
words 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.

B.l.2

LOAD-Class Commands

You can invoke FORTRAN-20 by using LOAD-class commands.
These
commands cause the monitor to interpret the command and construct new
command strings for
the system programs actually processing the
command.
COMPILE
LOAD
EXECUTE
DEBUG

B-4

USING THE COMPILER
Example
.EXEC ROTOR
The compiler switches OPT, CREF,
and DEBUG may be specified
LOAD-class commands and may be used globally or locally.

Example
.EXECUTE/CREF Pl.FOR,P2.FOR/DEBUG:NOBOU
The other compiler switches must be passed
specific source file.

parentheses

for

each

Example
.EXECUTE Pl.FOR(M,I)
Refer to the Monitor Calls User's Guide for further information.

B.2

READING THE LISTING

When you request a listing from the FORTRAN compiler, it contains
following information:

the

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 via the INCLUDE statement, all INCLUDEd lines will
have an asterisk (*) appended to their line-sequence number.

A summary of the names and relative
octal)
of scalars and arrays in
compiler generated variables.

All COMMON blocks and the relative locations
the variables in each COMMON block.

A listing of all equivalenced variables or arrays
relative locations.

A listing of the subprograms referenced
and FORTRAN defined library functions).

A summary of temporary locations generated by the compiler.

A heading on each page of the listing containing the program
unit name (MAIN., 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:

B-5

program locations
(in
the source program plus

(both

(in

octal)
and

user

their
defined

USING THE COMPILER
LINE:
This column contains the internal sequence number
of the line corresponding' to the mnemonic code.
It
appears on the first instruction 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 "P" appended.
Labels
generated by the compiler are followed by the letter
i'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 used the /LNMAP switch and did NOT use the /MACRO
switch,
a line number/octal location map is appended to the
listing. This section lists the line numbers
in 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 EDIT.
NOTE
One FORTRAN line
can
produce
multiple
octal
locations.
In this case the line number map lists
only the first location.

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
during compilations.

B.2.l

detected

warning

messages

issued

Compiler Generated Variables

In certain situations the compiler will generate 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
i.e., .50001

B-6

USING THE COMPILER
where:
Letter

Function of Variable

Arithmetic statement function formal parameters.

Result of a DO LOOP initial value expression
parameter of an adjustably dimensioned array.

Result of a common subexpression (see Section
or constant computation (C.2.1.3).

Temporary storage for expression values.

Result of
(C.2.1.2).

Result of the DO LOOP step size expression of
iteration count for a loop.

reduced

operator

strength

C.2.1.1)

expression
computed

You may find these variables on the listing under SCALARS and ARRAYS.
The following example shows a listing where
pointed out.

B-7

all

these

features

are

compiler Compiled for KI processor
verlion
~CRO Code equivalent included

Program Source
Name
Filename

MAIN.

T If<l 1

f'OHTKAfII

v.~(~15)

IKI/M

1o-MAk-77

Kl=r·J
IfCK1.LT.500.uR.Kl.GT.1500) Kl=O

00009
00010
00011
00012
00013
00014

A(I,J)=Kl
K2=1+J

00016
00017
00018
00019

PAGE 1

00 100 1=1,100

QOOOH

0001~

16:05

IMPLICIT IhTEG~R (A-ZJ
DIMENSION A(lUO,2u0J,H(100,200)
.':iUt-i1=O
SIJM2=O
DO 100 J=1,20()

00001
00002
00003
00004
00005
00006
00007

c::

1 f (1\2. ElJ • 100. OR • K2 • E(~ • 200 • OI~ .1\ 2. EO. 300 ) K2=K2+1

8(I,J)=K2
SUM1=SLJM1+Kl

tJl

CUNTlfllLiE

TYPE 10,SUMl,5UM2
FORMAT(7H SUM1= ,19,10H
EfIIV

1-3
tI::

S{)H2=SUM2+K2
100
C

(')

3:
"0

SUfv12= ,19)

t"t
tJl
~

5VI:1Pi-<lJGRAr"lS CALLi(;

SCALARS AND ARRAYS

"*"

NO r.:Xt-'l.JICl'l' Ot:l'"I{liI'l'lUN

n%" NOT Rt:Ft:RENCED J

~Compiler Generated Variable
47042
A
47043
.50001 116103
*K1
1
B
2
*J
116106 *"'2
11b107 *SUM1
116110
.50000 110104 *suM2
116105 *1
~_________________________~Internal sequence n~ber of first
instruction that goes with this line

LINE

IJUC

LABt:L

GENt:kATEO COD£<.:

•

3
4
5
tJj

JFCL

1
2
3
4

JSP

2,SUMl
2,SUM2

2, (7774', 0000001 J

HLkEM

2,.50000

HRkZH

2,J

MUVE

2,[777b3400(001)

MOVE

.3,J
3,0(2)

t""

til
H

:z
1-3

::c
[tl

2 {Ill :

()

3~1 :

4M:

J.MULI

MOVEt-l
CAlL

14
15
16
17
20

16,RE5ET.

l·iOVEN
HOVt:

SETZ~

0,0
0,0

5
6

Octal displacement of instruction

Jk5T

3,1<1
3,764
3,2734
O,oM

JRS'I

O,5""J

SETZB

4,Kl

CAIL~

6rYl: ..

t::l
~

Compiler
Generated
Label

l-lAlfII.

TIMl

fUKTKAN

51>"1 :

10
11

3 , ~J
3,0(2)

MOV~

4,Kl
4,A-145(3)

"',uVE

3,J

3,OCD

AOD!
;v,uVt:M

i·IOVE

3:l

CAlE

CAIN

5,K2
5,144
5,310
O,BM

JRS'f

7 (vI:

I--'

41
42
43
44
45
46

PAGE 1-1

3,K2

c::
til

CAIN

5,454

AUS

3,K2

NOV~: 1

3,144

II1UL
ADDI
tviOVE

3,0(2)

MUVEtvl

5,8-145(3)

ADlJM
A()DM

.....

zG')
8

:c
ttl

(")

3,J

0
3:
I'C

.....

5,K2

t'1
ttl

4,SUlw11

5,SlJM2
Program label

lOOP: ..

53
54
55
56

AOHJN
AOS
AOSG£
JRSl'
i'tlQVEI
PUSHJ
MOVEl
PUSHJ
MUVE!

PUSrlJ

47
50
51

1b:05

Yt-\ ;

Hr-. :

3,144

IMUI,
AlJDI

HOVEM

14
15

MOV~l

1 b-MAk--' I

2!J

IKI/M

:t3
2,*

34
35

'v.~(51~)

2,4M
2,J

O,.SOOOO
0,3/1'1
16,lOM
17,UUT.

16,11M
17,IOLST.

Ib,lM
17,t:XI'r.

ARGUMENT BLOCKS:
60
61
62

1M:

0,,0
0, ,0

10M:

777773,,0
0,,777777
0, ,0
0,,0

340,,10P
0,,7

67
70
71
72
73

0,,0

11M:

ll00"SUM1
1l00"SUM2
4000,,0
c:

til
H

MAIN.

TIMl

FORTRAN

V.~(515)

IKI/M

16-MAR-77

16:05

PAGE 1-2

G1
8

tIl

tJ:l
I

tZl

I-'
I-'

()

:J:

FORMAT STATEMENTS (IN LOW SEGMENT):
18

116111

lOP:

(7H S

116112
116113

UM1=
,19,1

116114
116115
116116

SUM2

= ,19

116117

MAIN.

[ NU ERRORS DETECTED ]

td
t-t

t-t

tZl
~d

MAIN.

TIM1

I
I-'
t\J

V,S(S15) IKI

16-rAAH-77

16:09

PAGE 1

IMPLICIT INTEGER (A-Z)

00001
00002
00003
00004
00005
00006
00007

tJj

FORT~AN

DIMENSION A(10Q,200),B(lOO,200)

SUMl=O
SUM2=O
DO 100 J=1,200
DO 100 1=1,100

Kl=l*J

00008

IF(Kt.LT.500.0R,Kl.GT,1500) Kl=O

00009
00010
00011
00012
00013
00014
00015
00016
00017
00018
00019

ACI,J)=Kl
K2=I+J
IF(K2,~Q,100.0R.K2,EQ.200,OR.K2.EQ,300)

K2=K2+1

BCI,J)=K2
100
C

SUM1=SUM1+Kl
SUM2=SUM2+K2
CONTINUt

C
til
H

G'l

TYPE 10,SUM1,SUM2
fORMAT(7H SUMl= ,I9,lOH
END

t-3

::c

SUN2;: ,19)

tzJ

n
o

"0
H

SUBPROGRAMS CALLED

t""
tzJ
~

"%" NOT REFERENCED]

SCALARS AND ARRAYS [ "*" NO EXPLICIT DEFINITION •
*Kl

,50000 116104

*SUM2

2
116105

*~l

47042
11610b

*K2

47043
11b107

,SOOOI 116103
*SUM1
116110

LINE NUMBER/OCTAL LOCATION MAP
I 0

.·~·-~I·---·-----··-·---------·-------·---------·--~··~~.-5
3
10
00000
4
46
44
45
00010
26
37
31
[ NO ERRORS DETECTED
MAIN,

21
56

...~ .. --.----.. ---------... 52

Line number 11 starts at location 31.
The previous listing shows that line 11
uses locations 31 through 36, but only
the first location is shown here.

MAIN.

TIM1

FORTRAN V,5(515) /KI/OPT/M

16.MAR .. 77

16:07

PAG~

til

to
I
~

00001
00002
00003
00004
00005
00006
00007
00008
00009
00010
00011
00012
00013
00014
00015
00016
00017
00018
00019

100
C
10

IMPLICIT INTEGER CA~Z)
DIMENSION A(100,200),BC1OO,200)
SU"l1 =0
SUM2=O
DO 100 J=1,200
DO 100 1=1,100
1\1=1*J
IF(Kl.LT.500,OR.Kl.GT,1500) K1=0
ACI,J)=Kl
K2=I+J
IFCK2,EQ.I00.QR,K2.EQ.200.0R.K2,EQ.300) K2=K2+1
B(I,J)=K2
SUM 1 =sur-1 1 +K 1
SUM2=SUM2+K2
CONTINUE
TYPE 10,SUM1,SUM2
FORMAT(7H SUM1= ,I9,10H
END

SUM2= ,19)

H
~
G')

==
tlj

n
0

::
to

tot

tlj
~

SUBPROGRAMS CALLED

SCALARS AND ARRAYS ( "*" NO EXPLICIT DEfINITION -

*Kl

47045
,00001 116111

2
~.ROOOI 47042
,50001 116105
,SOOOO 116106
116112 *SUM1
116113
*K2

LINE

LABEL

*
5
ll:I

~
~

1_-

LOC

Jf"CL

0,0

1
2

JSP

16,RESET,
0,0

3
4
5
6

SETZB
MOV£<:l
MOVE:M
M()VNI

10,11
12,144

MOVe;!

12"ROOQl
l2,3tO
7,1

MOVEM

12,,50000

14
15
16
17

20
21
22

Optimizer
Created
Statements

4M.

47044
116110

L Created
Optimizer
Variables

til
t-t

(j)
~

tIl
~

()

MOVE
MOVE

6,7
2,[777634000001]

MOV!':I
ADD

4,0(2)
4"ROOOl
5,6
5,764
5,2734
0,7M
O,6 M

SMa

MOVE
CALL
CAlLE
JRST
JRST
7M,

,ROOOO 47043
* 5 Ut-1 2 + 11 6 1 0 7

GENERATED CODE

"I" NOT REfERENCED

I"d
t-t
t"t

MAIN,

TIM1

FORTRAN V,5(515) IKI/OPT/N

6M:

24
25
26
27
30

lJj

I
I--'
U'1

10MI

9MI

8MI

:-.10VEI

5,0

16 .. t"'AR"7 7

16:07

PAGE 1-1

~10VEM

5,1\""145(4)

MOVE
ADDI
CAIE:
CAIN
JRST

3,7
3,144
3,310
O,9M

CAIN

3,454

ADDI

3,1

HOVEM

3,0(2)

ADD

3,a"'145(4)
11,5

ADD

10,3

*15

ADD

6,7

1-3

AOBJN
MOVEl

ADDM

2,5 11
12, 144
12, IRooot

1-1

*
*

tOOPI

1MI

*
*
*

*17
*
19

44
45
46
47
50

51
52

53
54
55
56

2M.

ADOI
AOSGE
JRST
MOVEM
MOVEM
MOVEM
MOVEM
MOVEl
PUSHJ
MOVEI
PUSHJ
MOVEI
PUSHJ

lJl

0, IS 0000

O,4M

11, UM14
10, UM24

5,K

7,1

.),K

(J)

•

16, 1M
17,0 UTI
16, 2M 4
17,IOLST.
16,3M
17,EXIT.

tzl
(')

I'C

Optimizer
Created
Statements

t'1

tzl
:;0

BLOCKS:

ARGUMr~NT

3t-i:

0, ,0
0,,0

U,M;

777773,,0
0,,777777

60
61

62
63

0, ,0

0,,0

65
66
67
70

340,,10P
0,,7
0,,0

12HZ

71
72

1100"SUMl
1100"SUM2
4000,,0

c:::

I-f

MAIN,

FORTRAN V.5(515) /KI/OPT/M

TIM1

IJ:I
I

16",r';AR",77

(j)

16:07

PAGF: 1-2

1-3

tzl

......

0'1

(')

FORMAT STATEMENTS (IN LOW SEGMENT):
18

MAIN.

116114
116115
116116
116117
116120
116121
116122
[ NO

lOP:

(7H S
UM1=
,19,1
OH

5UM2

=
,19
)

ERRORS OETECTED ]

I-f

t""
tzl
~

USING THE COMPILER
B.3

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.
Each error references the internal sequence number of
the incorrect line. The error messages along with the statement in
error are output to the user terminal.
For example:
.EXECUTE DAY.FOR
FORTRAN: DAY
01300
?FTNNRC LINE:01300
01500 100
?FTNMSP LINE:01500
01600 ?
?FTNICL LINE:01600
?FTNFTL MAIN.
LINK:
LOADING
[LNKNSA NO START ADDRESS]

Kl
STATEMENT NOT RECOGNIZED
CONTINE
STATEMENT NAME MISSPELLED
ILLEGAL CHARACTER C IN LABEL FIELD
3 FATAL ERRORS AND NO WARNINGS

EXIT
If errors are detected after the initial pass of the compiler,
they
appear in the list file after the end of the source listing.
They are
output to your terminal without the statement in error, but they may
reference its internal sequence number.

B.3.1

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 will probably 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. 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

fatal
warning
FORTRAN mnemonic
3-letter mnemonic for the error message
line number where error occurred
explanation of error

B-17

USING THE COMPILER
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 /NOWARNINGS switch will
suppress warning messages on both user terminal and listing.

B.3.2

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 NO/number FATAL ERRORS AND NO/number WARNINGS
or

when no messages were issued
name [NO ERRORS DETECTED]

where name is the program or subprogram name.
([NO ERRORS DETECTED]
appears on the listing only.) Appendix G is a complete list of fatal
errors and warning messages.

B.4

CREATING A REENTRANT FORTRAN PROGRAM WITH LINK

To produce a sharable program from the .REL file,
give one of the following commands to LINK:
1.

/SEG:DEFAULT MAIN/G

/OTS:SHAR MAIN/G

The resulting core image can be SSAVEd or the
used to produce a .SHR file.

B-18

such

/SSAVE

MAIN.REL,

switch

can

APPENDIX C
WRITING USER PROGRAMS

This appendix is a guide for writing effective programs
with
FORTRAN-20.
It contains techniques for optimization, interaction with
non-FORTRAN programs, and other useful programming hints.

C.l

GENERAL PROGRAMMING CONSIDERATIONS

The following paragraphs describe programming considerations you
should observe when preparing a FORTRAN program to be compiled by
FORTRAN-20.

C.l.l

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 of this manual.
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-20 can only test numbers of up to eight
significant digits (REAL precision) and 16 significant digits (DOUBLE
precision) •
You must also be careful when testing the floating-point computation
for a result of O.
In most cases the anticipated result, i.e., 0 will
be obtained; however, in some cases the result may be a very small
number that approximates o.
Such an approximation of o will cause
tests within statements, i.e., an arithmetic IF, to fail.

C.l.2

Writing FORTRAN-20 Programs for Execution On Non-DEC Machines

If you prepare a program to run on both a DECsystem-20 computer and
non-DIGITAL machine, you should:

Avoid using the non-ANSI standard features of FORTRAN-20, and

Consider the accuracy and size of the numbers
non-DIGITAL machine is capable of handling.

C-l

that

the

WRITING USER PROGRAMS
C.l.3

Using Floating-Point DO Loops

FORTRAN-20 permits you to employ non-integer single- or doubleprecision numbers as the parameter variables in a DO statement. This
enables you to generate a wider range of values for the DO loop index
variables,
which
may,
in turn, be used inside the loop for
computations.
Be sure to consider the loss of precision that may
occur.

C.l.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
MAX

(1, ((M2-Ml)/M3)+l)=Number of cycles

The values of the parameters Ml, M2, M3 may be of any type;
however,
you must consider the foregoing formula, particularly when using
logicals. One pass through each DO loop is always performed EVEN IF
THE RESULT OF THE FOREGOING CALCULATION IS LESS THAN OR EQUAL TO ZERO.

C.l.5

Subroutines - Programming Considerations

Consider the following items when preparing and executing subroutines:
1.

During execution, no check is made
number of parameters was passed.

see

the

proper

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.

If the number of actual arguments passed to a subroutine is
greater than the number of dummy arguments given, the excess
arguments are ignored.

If an actual parameter is a constant and its corresponding
dummy argument is set to another value, all references made
to the constant in the calling program may be changed to the
value of the dummy argument.

No check is made to see if the parameters passed are of the
same type as the dummy parameters.
If an actual parameter is
a constant and the corresponding dummy is of type real,
be
sure to include the decimal point with the constant.
If the
dummy is double-precision, be sure to specify the constant
with a "D".
Examples
If the function F(A) is called by inputting F(2)
and A is
type real,
F interprets the integer 2 as an unnormalized
floating-point number.
In this instance, F(A)
should be
called with F(2.0).
Similarly, if the function Fl(D)
is called by inputting
Fl(2.5) and D is double-precision, Fl assumes that its
C-2

HRITING USER PROGRAr·1S

parameters have been specified with two words of precision
and picks up whatever follows the constant 2.5 in memory.
The proper method is to use Fl(2.5DOO).
NOTE
You are given no notice if any of the
described in items 1,2,3,4, ana 5 occur.

C.l.6

situations

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 will cause an
overflow condition, and

RSI is a very small number, i.e., less than
sequence

the

program

A=RSl*RLl*RL2
B=RS2*RL2*RLI

will not produce an overflow when evaluated left to right,
since the first computation in each expression, i.e., RSl*RLl
and RS2*RL2, will produce an interim result that is smaller
than either large number (RLI or RL2).
However, the compiler will recognize RLl*RL2 as a common subexpression
(see Section C.2.I.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=(RSI*RLl)*RL2
B=(RS2*RL2)*RLI
Computations may be reordered even when
selected.

C-3

global

optimization

not

WRITING USER PROGRAMS
C.l.7

Dimensioning of Formal Arrays

When you specify an array as a formal parameter to a subprogram unit,
you must indicate to the compiler that the parameter is an array.
Dimension 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.
Designating the array with a dimension of
1 is a common practice.
Example
SUBROUTINE SUBl(A,B)
DIMENSION A(l)
the above technique
is not adequate in

There are disadvantages to using
dimension
information provided
specifically:
1.

because the
some cases,

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(l)
READ(l)A
This binary read will cause one word to be read into A.
2.

Reading the array as a format
SUBROUTINE SUB2 (FMT)
DIMENSION FMT(l)
READ (l,FMT)
This will cause one word of the array FMT to be written
with the characters read from the record on unit 1.

over

When you use the /DEBUG:BOUNDS compilation switch, the dimension
information used is that which is specified in the array declaration.
SUBROUTINE DO IT(A)
DIMENSION A(l)
A(2)=0
The reference to A(2) will cause the out-of-bounds warning message
be generated.

C.2

FORTRAN-20 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.
C-4

WRITING USER PROGRAMS
C.2.l

Optimization Techniques

C.2.l.l 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, and 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 foregoing sequence as:

T=B*C
A=T+E*F
H=A+G-T
DIF((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.

C.2.l.2 Reduction of Operator Strength - The time required to execute
arithmetic operations will vary 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 (1), 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

1. Numeri~al analysis considerations
cases where this is possible.
C-5

severely

limit

the

number

WRITING USER PROGRAMS
The execution time of the foregoing
loop would be shorter
if the
time-consuming multiply operation, i.e., 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 foregoing
sequence equivalent to:

loop

would

replaced

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 foregoing situation occurs frequently in subscript calculations
that
implicitly contain multiplications whenever the size is two or
greater.

C.2.l.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 redefinea 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 1=1,100
F=2.0*Q*A(I)+F

the value of the computation 2.0*Q,
once calculated on the first
iterations,
will
remain unchanged during the remaining 99 iterations
of the loop.
Reforming the foregoing 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).

C-6

~1RITING

USER PROGR.z\MS

C.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=3.14IS9

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=3.14IS9

X=6.28318*Y

thereby eliminating a multiply operation from the object code program.
The computation of determinate constant values at compile time is
termed "folding";
the use of the defined value of a constant for
replacement purposes throughout a program
sequence
is
termed
"propagation of the' constants." The execution time saved by the
foregoing type of compile time optimization is particularly important
when the modified instruction occurs in a loop.

C.2.I.S 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 will reduce the size of the object
program. A warning message is generated for each inaccessible line
removed.

C.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.

C-7

WRITING USER PROGRAMS
C.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 READ,
WRITE, ENCODE, DECODE, and REREAD statements. The formats of these
special blocks are described in Appendix E.
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.

C.2.1.S Uninitialized Variable Detection
A warning message is
generated when a scalar variable is referenced before it has received
a value.

C.2.1.9 Test Replacement - If the only use of a DO loop index is to
reduce operator strength (D.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 occurs frequently in subscript computation.

C.2.2

Improper Function References

Consider this statement:
P = F(X) + Q(Y)
If:
1.

the evaluation of F(X) defines or changes the variables A, B,
and C, and

the evaluation of Q(Y) defines or changes the values of B, C,
and D,

then it is possible that different values of P could result, depending
on which function is evaluated first.
Let's see how this works.
Let's assign some values (to begin with) to A, B, C, and D and define
the functions F{X) and Q(Y):
Let:
F (X) :
A
B

C
D

2.
3.
4.
5.

Q (Y) :

7.
S.
D + 9.

C
D

C
F

C-S

10.
11.
12.
A + 13.

WRITING USER PROG~~MS
Now play computer and evaluate P, calling first F(X), then Q(Y).
Now
re-evaluate P, calling Q(Y)
first, then F(X). Notice that you got
different values for P because the variables A, B, C,
and D changed
value depending on the order in which the functions were called.
(Our
answers were 33 when F(X) was called first and 36 when Q(Y) was called
first.)
The ANSI FORTRAN standard prohibits this kind of situation.
But the
compiler won't catch it unless you mention the affected variables in
the function call itself. The compiler depends on strict adherence to
this rule.
There's a strong possibility that you won't get the
results you want if you don't look for situations of this type and
avoid them.
Your best bet is to define your variables OUTSIDE the
function and not change them in the course of the evaluation of the
function itself.

C.2.3

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.

C.3

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 input/output variables.

Avoid unnecessary use of COMMON and EQUIVALENCE.

one

with

the

INTERACTING WITH NON-FORTRAN PROGRAMS AND FILES

C.3.1

Calling Sequences

The following paragraphs describe the standard procedures for
subroutine calls.
1.

writing

Procedure
half
of
the first

The calling program must load the right
accumulator
(AC)
16 with the address of
argument in the argument list.

The left half of AC 16 must be set to zero.

The subroutine is then called by a PUSHJ
AC 17.

The return will be made to the
after the PUSHJ 17 instruction.

instruction

immediately

If you use the FOROTS trace
sequence to a routine F must be

facility,

the

MOVEI 16,AP
PUSHJ 17,F

C-9

instruction

calling

WRITING USER PROGRAMS
where AP is the pointer to the argument list. If you use
the trace facility, the word preceding the first word of
an entry point should have its name in SIXBIT.
2.

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:
JSP l6,RESET.

C.3.2

Accumulator Usage

The specific functions performed by accumulators (AC) 17,16,0,
are as follows:

and

Pushdown Pointer - AC 17 is always maintained as a pushdown
pointer.
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-I) allocated to the unused remainder of the
stack.
{A trap occurs when something is pushed into the next
to last location. A positive left half is not permitted.

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 callee. AC 16
cannot point to the ACs or to the stack.

Temporary and Value Return Registers - AC
and 1 are used as
temporary registers and for returning values. The called
subprogram does not need to preserve the contents of AC
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 - At the option of the designer of a called
subprogram, a subroutine may pass back results by modifying
the arguments, returning a single-precision value in AC
or
a double-precision or complex value in AC
and 1. A
combination of the above may be used.
However,
two
single-precision values cannot be returned in AC
and 1,
since FORTRAN would not be able to handle it.

C-lO

WRITING USER PROGRAMS
5.

Preserved ACs - FORTRAN-20 FUNCTION subprograms preserve
2 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-20 or COBOL.

C.3.3

Argument Lists

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
entries.
These are reserved and must be O.

The format of an arg entry is as
word) :
bits 0-8
bits 9-12
bit 13
bits 14-17
bits 18-35

follows

(each

number

entry

Reserved for future DEC development
now) .
Arg type code.
Indirect bit if desired.
Index field, must be a for present.
Address of the argument.

arg

single

(set

a for

The following restrictions should be observed:
1.

Neither the argument list nor the arguments themselves can be
on the stack. This restriction is imposed so that the stack
can be moved.
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,
i.e.,
the
argument count word with a a in it and a zero argument word.

C-ll

WRITING USER PROGRAMS
Example
MOVEI l6,AP
PUSHJ 17,SUB

iSET UP ARG POINTER
iCALL SUBROUTINE
iRETURN HERE

iARGUMENT LIST
-3,,0
A

AP:

iSUBROUTINE TO SET THIRD ARG TO SUM OF FIRST TWO ARGS
SUB:

C.3.4

MOVE
ADD
MOVEM
POPJ

iGET FIRST ARG

T,@0(16)
T,@1(16)
T,@2(16)
17,

iADD SECOND ARG

iSET THIRD ARG
iRETURN TO CALLER

Argument Types
Table C-l
Argument Types and Type Codes
Type Code

o
1

2
3
4
5
6
7

10
11

12
13
14
15
16
17

Description
FORTRAN Use

COBOL Use

Unspecified
FORTRAN Logical
Integer
Reserved
Real
Reserved
Octal
Label
Double real
Not applicable
Double Octal
Reserved
Complex
Not applicable
Reserved
ASCIZ string

Unspecified
Not applicable
I-word COMP
Reserved
COMP-l
Reserved
Reserved
Procedure address
Not applicable
2-word COMP
Reserved
Reserved
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-20 compiler makes a local of all
non-array elements and copies all formals back to the caller's
arguments.
All unused type codes are reserved for future DIGITAL
development.

C-12

WRITING USER PROG~~MS
C.3.5

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 ..

specify

Type 2 - Integer and l-word-COMP
A 36-bit 2's complement signed binary integer.

Type 4 - Real and COMP-l
A 36-bit DECsystem-20 format floating-point number.

bit 0
bits 1-8
bits 9-35
5.

sign
excess 128 exponent
mantissa

Type 6 - Octal
A 36-bit unsigned binary value.

Type 7 - Label and procedure address
A 23-bit memory address.
bits 0-12
bit 13
bits 14-17
bits 18-35

always 0
indirect flag

the address

Type 10 - Double precision real

Type 11 -

2-word COMP

A 2-word (72-bit)
word
word
word
word
9.

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

Type 12 - Double octal
A 72-bit unsigned binary value.

C-13

WRITING USER PROGRAMS
10.

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.

11.

Type 15 - Byte String Descriptor
The format of the byte string descriptor is:
word 1:
word 2:

ILDB-type pointer,
i.e.,
aimed at the
preceding the first byte of the string
EXP byte count

byte

The byte descriptor may not be modified by the called
program.
The byte string itself must consist of a string of
contiguous bytes of uniform size. The byte size may be any
number of bits from 1 to 36. The byte count must be large
enough to encompass 256K words of storage, i.e., 24 bits for
I-bit bytes.
(See COBOL Program Reference Manual.)
12.

Type 17 - ASCIZ string
A string of contiguous 7-bit ASCII bytes left justified on
the word boundary of the first word and terminated by a null
byte in the last word.
The length of the string may be from
1 to 256K words.

C.3.6

Converting Existing MACRO Libraries for use with FORTRAN-20

The following
sequence.

simple

example

illustrates

C-14

the

FORTRAN-20

calling

(")

1 n.. f~AR-7 7

MAIN"

EX1

FORTRAN V,5(515)

00001
00002
00003
00004
00005
00006
00007
00008
00009
00010
00011.
00012
00013
00014
00015
00016
00017

AN EXAMPLE OF A CALL TO A SUBROUTINE WITH A VARIETY'OF ARGUMENTS

IKt/~

16:02

PAGE: 1

DOUBLE PRECISION DP
DIMENSION B(10)
C

C
C

C
C
C

C
C

THE ARGUMENTS ARE,
1 , A REAL VARIABIJE
2, AN ARRAY NANE
3, AN APRAY ~LEMF.NT
4. AN INTEGER VARIABLE
5, A DOUBLE PRECISION VARIABLE
6, AN OCTAL CONSTANT
7, A LITERAL

~
~
H

CALL SUS1(A,B,B(I),K,DP,"777,'ABC')

END

tf)

t=l

I-'
U1

SUBPROGRAMS CALLED

to
~

0
Gl

SUB1

SCALARS AND ARRAYS

~
(fl

"*" NO EXPLICIT DEFINITION • "%" NOT REFERENCED
3

LINE

Loe

LABEL

JFCIJ
JSP

1
2
3
4

0,0

16,RESET.
0,0

MOVE:

MOVEl
MOVEM
MOVEI
PUSHJ

5
17

GENERATED CODE

6
7
10

MOVEI

PLJSHJ

2,1
2,S-1(2)
2,,00000
16,2M

17,SUBl
1.6,lM

17,EXIT.

ARGUMENT BLOCKS,
~
~

(')

13
14

1M:

2M;

......

Cj)

en
trl

220",QOOOO
100"K
400"DP

)00,,[000000000777]

17
20

0,,0
0,,0
777771,,0
200"A
200"S

"tI

oCj)

:::c

FORTRAN V.5(515) IKI/M

MAIN,

EXl

MAIN.

23
740,,[406050320100)
[ NO ERRORS DETECTED 1

16-MAR-77

16102

PAGE 1-1

MAIN.

'EX1

fOHTRAN V.5(515) IKI/M

16:02

16~MAR~77

PAGE 1

00001

SUBROUTINE BUB1(REAL1,ARYNAM,ARYELM,INT1,DBLPRC,OCT,LIT)
DOUBLE PRECISION DBLPRC
DIMENSION ARYNAM(lO)

00002

(')

I-'

00003
00004
00005
00006
00007
00008
00009
00010
00011
00012
00013
00014
00015
00016
00017
00018
00019
00020
00021
00022
00023
00024
00025

AN EXAMPLE Of THE USE AND MODIFICATION OF FORMAL PARAMETERS
Xl=REALl
X2=ARYNAM(J)
X3=ARY~;LM

11=INTl
X4=DBLPRC
12::0CT
13=LIT

:e:
~t:I

.-i
.-3
.-i
t~

G'l

REALt=Xl
APYNAM(J)=X2
ARYELM=X3
INT1=11

tJ)

tr;]
~t:I

ttl

OBLPRC=CMPLX(X4,0.)

OCT="55
LIT::'Z¥XW'

G'l
.t:I

:1::
tJ)

RETU~N

END

SUBPROG~AMS

CALLED

CMPLX.
SCALARS AND ARRAYS [ "*" NO EXPLICIT DEFINITION •
*LIT

DBLPRC 6
14
*INTl

*OCT
*13

*12

2
10

*X4
*REAL!

*X1

"%" NOT REFERENCED

*ARYELM 4
12
17

*J
*11

.X3

*X2

ARYNAt<\ 20

LINE

LOC

LABEL

636542,,210000

SUB1:
2

GENERATED CODE

0
1
2
3
4
5
6
7
10

MOVEM

lb,.AOOlf>

r-iOVE

O,@O(16)

,',mVEt1

O,REALl

r.10Vf-~I

1,@1(16)

MOVr:r"

1,ARY.NAM

MOVI':

i.,OVEM

1,@2(16)
1 , ARYEI.JH

.MOVE

2,@3(16)

MOVEM
DHOVE
DMOVEM

2,INTl
4,@4(t6)

4,OBLPRC

lX'

Ci)

SUB1 .

EX1

FORTRAN V,5(515) IKJ/M

lbI"HAR",,77

()

16:02

PAGE bill

en
tzl
~

I
I-'
ex>

'>10VE
MOVEM
MOVE
MOVEM

14
8
9

15
16
17

20
21

22
23
24

3MI

3,oer

6,@6(16)

6,LIT

"'0

lX'

Ci)

MOVEM

O,Xl

MOVE

7, \. 1
7, ARYNAt,1

ADO
t·lOVE
MOVEM

MOVEM

11
12

~~OVEM

26
27

PUSHJ
MOVEM

30
31
32

MOVEN

~~nVEM

14
16

3,@5(16)

FIX

MOVr~M

7,777777(7)
7,X2
I,X]
2,11
17,SNG.4
4,X4
3,3
3,12
6,13
0, lU:ALl

r·mVE
ADD

35
18
19

20
21
22

~\oVEM

36
37

MOVEM
MOVEM
MOVEr

41
42
43
44

45
46
47
50

54
55
56
57

....,

5,0
5,0

D~1OVEM

4,DBLPRC

r10VEI

2,55

MOVEM
MOVE
MOVEM

2,OC'l'

MOVE
HOVE
MOVEM
MOVE
MOVEM

16,.AOO16
O,REALl
O,ARYt:LM
O,ca2(16)

MOVE

O,INT!

Cil

MOVEM

0,(a3(16)
O,DRLPRC

c:::

or-.10VE

2,LIT

O,@O(16)

~
~

til

DMOVEN
MOVE
MOVEM
MOVE

O,@4(16)

a,OCT

I'd

~1OVEM

POPJ

O,@6(16)
17,0

62
63

ARGUMENT BLOCKS.
66
67

SUBl

2,[552633053500]

tIj
~

f.l0VEI

2M;

51
52

()

3, ,1
3,ARYNAH
7,777777(3.1
1,ARYELM
2,INTl

H1=

0,,0
0,,0

[ NO ERRORS DETECTED

O,@5(16)

O,LIT

Cil

WRITING USER PROGRAMS
To convert existing MACRO programs conveniently so that they
still load and execute correctly when called from FORTRAN-20:
1.

Transfer the initial entry sequence for a routine to
entry:

will

CAIA
PUSH l7,CEXIT.##

Change all returns to POPJ 17,0

These are the functions performed by the HELLO and GOODBY macros.
These macros
(available in the file FORPRM.MAC, part of the FOROTS
release) were successfully used to convert the library routines to run
with FORTRAN-20.
In addition, since the FORTRAN-20 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
MOVEI R,@1(l6)
The latter operation will work when interfacing with FORTRAN-20.
Refer to the previous example, which illustrates the code generated by
the FORTRAN-20 compiler, for specific details of how each argument IS
accessed. Note that in the case of the formal array,
it is the
address of the array that is accessed.

C.3.7

Interaction with COBOL

The FORTRAN programmer may call COBOL programs as subprograms, and,
conversely,
the COBOL programmers may call FORTRAN-20 programs as
subprograms.
In either of the foregoing cases, I/O operation must not be
in the called subprogram.

C-20

performed

WRITING USER PROGRAHS
C.3.7.l Calling FORTRAN-20 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-20 subroutines. The form of ENTER is as follows:
ENTER FORTRAN program name

rUSING

{if~~;!iierl
} [ {if~~;!i~er2}J ]
procedure namel
procedure2

The USING clause of the foregoing forms 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-20.
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 O. Each entry in the list is one 36-bit word at the form:
0-8

9-12

13-35

type

address

Bits 0-8 are always O.
Bits 9-12 contain a type code that indicates the USAGE
argument.

the

Bits 13-35 contain the address of the argument of the first
word of the argument;
the address can be indexed or indirect.
Following is a description of the types, their codes, how the codes
appear
in the argument list, and the locations specified by the
addresses.
1.

For I-word COMPUTATIONAL items
CODE:
IN ARGUMENT LIST:
ADDRESS:

XWD 100, address
that of the argument itself

C-21

WRITING USER PROGRAMS
2.

For 2-word COMPUTATIONAL items
CODE:
IN ARGUMENT LIST:
ADDRESS:

11
XWD 440, address
that of the high-order
argument

word

the

For procedure names (which cannot be used for calls to
subprograms)

COBOL

For COMPUTATIONAL-l items
CODE:
IN ARGUMENT LIST:
ADDRESS:

CODE:
IN ARGUMENT LIST:
ADDRESS:

XWD 200, address
that of the argument itself

7
XWD 340, address
that of the procedure

The return from a subprogram (via POPJ 17,) is to the statement
the call.

after

C.3.7.2 Calling COBOL Subroutines from FORTRAN-20 Programs - To
COBOL subroutines use the standard subroutine calling mechanism:

call

CALL COBOLS (args ... )
X=COBOLS (args ... )

subroutine call
function call

You must have compiled the COBOL subroutine using the COBOL compiler
described in the DECsystem-20 COBOL Programmer's Reference Manual.

C.3.8

LINK Overlay Facilities

LINK provides several routines that are accessible directly from a
FORTRAN-20 program.
These routines are presented here briefly,
together with the FORTRAN-20 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.

C.3.8.1 Conventions - The following terms are used
parameters to LINK overlay routines.
File spec
Name
List of link names

describe. the

A literal constant consisting of device:
filename. ext [directory]
A LINK name or number that is a literal
constant or variable.
A sequence of name items separated by
commas.

The routines available are:
INIOVL

(File spec) Used to specify the overlay
file
to
be
found
if the load time
specification is to be overridden.

GETOVL

(List of link names)
Used to
overlay structure in memory.
C-22

change

the

WRITING USER PROGRAi-iS
RUNOVL

(Name)
Loads the
specified
transfers to that LINK.

LINK

REMOVL

(List of link names) Removes the
LINKs from memory.

specified

LOGOVL

(File spec) Used to specify where the log
file is to be written.
If no arguments are
given, the log file is closed.

For a full description of these routines, refer to the LINK
Manual.

C-23

and

Reference

APPENDIX D
FOROTS

This appendix describes the facilities that FOROTS provides for
the
FORTRAN user.
FOROTS implements all standard FORTRAN I/O operations
as set forth in the IIAmerican National Standard FORTRAN, ANSI
X3.9-l966. 11
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 DECsystem-20 monitor during input and
output operations. Other capabilities include:

D.l

Job initialization

Channel and memory management

Error handling and reporting

File management

Formatting of data

Mathematical library

User library (non-mathematical)

Specialized applications packages

Overlay facilities

HARDWARE AND SOFTWARE REQUIREMENTS

FOROTS may interface with all DECsystem-20 peripheral devices.
In
addition to monitor or user program requirements, a minimum of 14
pages of user memory is needed to run FOROTS.

D-1

FOROTS
The software required with FOROTS is the TOPS-20 monitor,
version
Other software items that can be associated with FOROTS include:

D.2

The MACRO assembler

The LINK loader

The FORTRAN-20 compiler

FEATURES OF FOROTS

The following list briefly describes many specific features;
more
detailed information concerning the implementation of these features
is given later in this appendix.
1.

Your program may 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.

Your programs may access both
devices in the same manner.

FOROTS helps provide complete data file compatibility between
all system devices.

FOROTS does not require line-blocking
(a requirement that
each output buffer must contain only an integral number of
lines) .

Up to 15 data files
number or all of
randomly.

FOROTS treats devices located at remote stations similarly to
local devices.

Programs written for
magnetic tape operations will
run
correctly on disk under FOROTS supervision.
FOROTS simulates
the commands needed for magnetic tape operations.

You may change or specify object program device and
specifications via a FOROTS interactive dialogue mode.

Non-FORTRAN binary data files may be read in
FOROTS.

10.

FOROTS provides interactive
processing routines.
These
execution of the program
routines whenever designated

11.

An error traceback facility for
fatal
errors provides a
history of all subprogram calls made back to the main program
at the address of the point where the error occurred.

may
the

directory

and

non-directory

be accessed simultaneously.
Any
open data files may be accessed

D-2

image

mode

file
by

program/operating system error
routines permit you to route the
to specific error
processing
types of errors are detected.

FOROTS
12.

FOROTS provides a trap handling system for
arithmetic
functions, including default values and error reports.

13.

You may mix ASCII and binary records in the same file,
and
both may be accessed in either sequential or random access
mode.

14.

FOROTS permits your program to switch from READ to WRITE
the same I/O device without loss of data or buffering.

15.

Although primarily designed for use with the FORTRAN-20
compiler, you may also use FOROTS as an independent I/O
system, as an I/O system for MACRO object programs,
and for
FORTRAN-20 object programs.

D.3

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.

D.4

INPUT/OUTPUT FACILITIES

FOROTS uses monitor-buffered I/O during all modes of access except
DUMP mode.
Display devices are supported in dump mode;
formatted
text is handled in ASCII line mode;
unformatted files are accessed as
FORTRAN binary files.
(Refer to the Monitor Calls User's Guide.)
The following paragraphs describe I/O data channel and access modes.

D.4.1

Input/Output Channels Used Internally by FOROTS

Fifteen software channels
(1 through 15)
are available in I/O
operations.
Software channel 0 is reserved for the following system
functions:
1.

The printing of error messages, and

The loading
operations)

Software
transfer
table is
channel
assigned
FOROTS.

and

initialization

FOROTS

(GETSEG

UUO

channels 1 through 15 are available for
user program data
operations.
When a request is made for a data channel, a
scanned until a free channel is found.
The first free
is assigned to the requesting program;
on completion of the
transfer, control of the software channel is returned to

D-3

FOROTS
D.4.2

File Access Modes

Data may be transferred between processor storage
devices in two major modes - sequential and random.

and

peripheral

D.4.2.1 Sequential Transfer Mode
In sequential data transfer
operations, the records involved are transferred in the same order as
they appear in the source file.
Each I/O statement executed in this
mode transfers the record immediately following the last record
transferred from the accessed source file.
A special version of the
sequential mode
(referred to as APPEND)
is available for output
(write) operations. The special APPEND mode permits you to write a
record immediately after the last logical record of the accessed file.
During the APPEND operation, 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 modes (other than SEQINOUT) by setting the
ACCESS option of a FORTRAN-20 OPEN statement to one of several
possible arguments.
For the sequential mode, the arguments are
ACCESS='SEQIN' (sequential read-only mode)
ACCESS='SEQOUT ' (sequential write-only mode)
ACCESS='SEQINOUT' (sequential read followed by a sequential
write)
ACCESS='APPEND' (sequential Append mode)

D.4.2.2 Random Access Mode - This transfer mode permits records to be
accessed and transferred from a source file in any desired order.
Random access transfers must be made between processor memory and a
device
(disk) that permits random addressing operations to files that
have been set up for random access.
Files for
random access must
contain a specified number of identically sized records that may be
individually accessed by a record number.
You may accomplish random access transfers in either a read/write mode
or a special read-only mode. You must specify random transfer modes
by setting the ACCESS option of an OPEN statement to one of several
possible arguments.
ACCESS='RANDOM'
ACCESS='RANDIN'

D.S

(random read/write mode)
(random special read-only mode)

ACCEPTABLE TYPES OF DATA FILES AND THEIR FORMATS

The following paragraphs describe the types of
acceptable to FOROTS.

D.S.l

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, such as, a Form
Feed, a vertical Tab, or a Line Feed, terminates each set. All ASCII
records start on a word boundary;
the last word in a record is padded
with nulls, if necessary, to ensure that the record also ends on a
word boundary.
Logical records may be split across physical blocks.
There is no implied maximum length for logical records.
D-4

FOROTS

NOTE
On sequential input, FOROTS does not
require conformation to word boundaries;
it reads what it sees.
Therefore, any
file that is written by FOROTS will
conform
to
the
foregoing
format
requirements.

D.S.2

FORTRAN Binary Data Files

Each logical record in a FORTRAN binary data file contains data that
the executing program may reference with either a READ or WRITE
statement. A logical record is preceded 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.

D.S.2.1 Format of Binary Files - A FOROTS binary file may contain
three forms of Logical Segment Control Words (LSCW). These LSCWs give
FOROTS the ability to distinguish ASCII files from binary files.
LSCW
001+ the number of words in the segment (exclusive of
any "ENDII LSCWs)
002 indicates that the segment of a disk
block
boundary continues
003+ number of words in the preceding segment including
LSCWs.

START
CONTINUE
END

If the access you specify for a file
(through the OPEN statement
ACCESS
parameter)
is 'SEQIN', 'SEQOUT', or 'SEQINOUT', all three
LSCWs may appear in a record. If the access you specify is 'RANDIN',
or
'RANDOM', 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

~OOK AT A BINARY rl~E AND SEE THE
CONTROL WOROS.

The random access
~OGICAL

OP~N(UNIT=1,ACCESS='RANDOM',MODt~i81NA~V'.

RECORD~l~~)

I~5

WRIT~(l'~)

J='

WRtTE(1'2)

(I,

J=l,l~~)

(J,K=1,10~)

END

D-5

SEGMENT

binary

FOROTS

000000
007.001
00001212
210k-.;003
002-004
002005
"000"6
000007
002~1"
00~1011

000012
00Z013
00 ~.H2l14
000~15
0"~;016

00Jel17

00ll?l20
0~h~021
00;~022
~~h;023
2lrch~024

",,(3025
00 ;'1026
2l0(:H~27

rlI2I1000 21 00145 -Number of words
2100""5 in record counting
LSCW or the
02100210 0"0"05 END
number of words
21021121"0 21"'00215 following this
~"~0021

~0210Z21
00~0021

"~00"5

21~21~05

~00""0

000"0"

:2J000~3

""H,040
"02041
000042
000043
"00044
210'~045
~0'~046

"0~12147
3"~~2I50

00~060

00212161
"0;i~062

210?-063

"0,,1212121 02'0005
00"'''''21 21~21005
0000rl10 eHHHH"5
0"0"021 211210"05
"""""0 12100"215
"000210 0210005
0"H:HH~0 02'0"05
0"21 07H!J 2Iel0"05
~00121~0 200"1215
~0121021" 0""005
00210~21 ael0"'''5
12100""0 ~0""05
';:'0210"0 0"121"1215
"0001210 ~"02105
r2J0~"0" 0"001215
"0210"0 0e0005

21~hH~ 7 6
01210077
00011210
00~11211

0121 ;:1102
"0~1r213

""01214
007105
00?1216
""~~107

"~H~110

""00111
""00112
00['113
"0'~ 114
00<1115
000116

00~00"

0121e12l55
000056
210012157

"0~037

~H"0~05

0e;~07"

0"~"75

~00005

~2100~21

002035
00(::036

a"21~05

~021000

00001~

""0"1215

210.~054

"01.034

121000021

"'1210073

~0"005

210012153,
2100052
1210:3053

2l01Z'032
"0r.033

21O')066
2121f!067
00\~072

0"'21"05
202121021 0002105
~HHH?J00
~""'''10 5
~"0"'021 0"'21005
r2l000021 000"05
'" 0 ~"'" 0 """"05
0000"0
01210"05
"021-02121 ""0"05
1210002121 0"0"05
0000"21 ~CJl2I"05
"021000 0~"'005
0021121"" ~"""05
"000"" 000005
"002100 0"0"'1215
0~H.~0~0

kH"H~~0
0~HHHJ0

0lrH'12I7l

~0210"21

21"0"05
012100"0 0"0"05
"0"0t1l0 0r2l0"05
"000"" 000"1215
000"'~H~ ""0"05
0000021 0"0"05
012100"" 0"001215
0021000 0"'0"05
"0"000 0"012105
00~02l0 ""0"05
0121"""'0 0"0"05
00"0'-"0 0 0 00"5
000"03 0"0"05
01210003 0"021215
02102100 121~0035
"'0~~00 210"01215
0121~H'2I2I 0"0'1'05
000121~121 1210001215
"0"'''00 0021"215
"0"9'rlI2I ""'001215
0012103121 000005

00~~030
"0~J031

word to the
END LSCW.

00L~117

l2Iek3120
000121
"0~122

000123
000124
00C125
r210~126

210"127
"0~13"

"0"131
21e~lJ2

21121121133
"0121134

00~135

000136
21"~131

012101421
01210141

~"'I2I01215
~21000" 0~00"5
~I2I0"~3 0000215

00~142

2101.143
210·1)144

0~I,H~12I5

02112100" rc'J00el05
~0002l21 000005
00~rc'J~0 002101215
~HH"~~3 002101215

00r}145

00 ;'~146
r210 ..~147
00 (.~ 15121

D-6

~0"005

210~064
0'H~065

21"0005

""HI""" 0"0""5
2l"0~00

~""'r2105

"rcHJ" 2l" 0 0 ""05
"121"000 0021"215
o rcHc~ 0 0 21 21"'0"05
"1210"00 121 0 0005
1i10,,0"0 12100"05
"00000 0"0"05
""0012121 ~r2J0005
00000121 0"'001215
00121000 0"0005
0121"00" 00""11,,5
"0"0""0 121"0"1215
0"0121215
""BeeQJ 21"~~05
21"10e10 021QJ"05
""00000 2100"215
"210"021 3021"21'
1211210021121 21"2101215
2100121el0 21£1"""'5
"3121~0121 0021"1215
2121!2112100 0021"05
02100321 1210121"121'
1212100"21 o"'HH!I 5
rc'J000 121" 0""~05
[2!2I0000 0021"05
0121"0"121 i2l"'121~"5
012100021 ~Q!0005
0~0"00 00121"05
"03000 121"0146 -END LSCW
Containing the
02110~121 0021145
number of words
"000~"" 000"1217 in the record
rc'J00eJt'l2I eJ~21007 including LSCW's.

"""""' '

FOROTS
'210'~ 151

00((H1~0

00:~152
00~153

00t'10210 0"21001
0021"00 21(210~07
o 2"H~ ~0 0121~HHP
000000 ~0121QJ01

QJ0;~154

00/1155
"0:~156
21rzL~157

1210216"
'00/161
0rcH'.162
00 '1163
00J164
002'165
00·;166

012!QJI2!~0

~H'002J7

2l0~HHH'

0"0007

~"121007

012!012100

0~0"'07

00~253
00.~ 2 54

~kH"0012!

~~001211

0000~0

00.'~177

eJ~W201
1210,~ 202

00~~ 203
00/2214
1?10?205
007206
0~H2217

00,'210
00'~1211

00 ~~ 212
1210~J213

0r,P214
00!~215
~ 0 ~,~ 216

002'0~12!

00121007

00Q!l2ltJl2! ~00"07
012!0eJ2!0 "l'0007
eJ000al2! ""'0007
2ll2!el000 2"0~01
2100000 ~00eJ01

000255
00;'256
00?257
00;1260
00Z261
1lJ~ ~') 262
00{1263
00/264
00~A 2 65
0eh'.i266
1?10~· 267

00k""~eJ ,,00~07
2!12!~H~0" 000"07

0~H'271

~00Q!~0

e!00~07

"'''0~~12I

""0007

2l0~000
~0021~0
~0el000

0~00"7
000~01

0~0~07

210012l~0

0~0eJ07

0"'0~01

i2l0~2221

002221

2112121030

~~00"1

00~222
00~;223

0"Ql0~21

0~000?

00~'227
0i21~230
00~231

00J232

00,)212
002273
1?10t1214
2102275
eJ0J276
2102277
(2)0J30121
001.:301

21012102121 0"'0"1217
<H'" 21 121 0 ~H"0"07
a00001
;;'021007

O00224
00;3225
0210226

00~27121

2100000 2100"1217
o 121 I(HHJ 0 eJ00~"7
000000 2100"01

i2l~~~~0
0000~0
i2l0i2l0~0

00(1217

00i~240

00iJ241
001/,242
000243
00 ~~ 244
2100245
00 ~~ 246
0,,,;::247
002.250
00J251
00Z252

00~1170

00 ;;200

00e237

00QJI2!QJ0
12100000
0000210 00001217
00121000 ~"'00"7
0000"121 ~00007
0012J0012! 0 0 0007

00~1167

000173
00(1114
00D175
001176

00~234
00~)235
00~236

~00~01
~fZl0012J7
~e'0012!7

00012!0121 00121007
000000 ~H~I2I001

00"~ 171
00t~172

00J233

rai2l~302

i2lraZ303
00~3i2l4

"0"""0 0 0 0210'
0121012100 0000"7
o 121121 12! 1210 0(2)012101
00"12130 121"001211
00210121121 ~~012J01
00001210 0(2)012)07
i2l 0 r,Hl' 1210 000~i217
000000 000007

(2)0;~305

0(0312)6
i2l0~307
00~31I2J

000311
000312
000313

D-7

0000321 0021007
021000121 121012101217
00210021 000001
0""~00121 000001
0000121121 ""121"1217
2112!02100 2101210121"
0"H:10,-,,, 1210121007
0012!00121 0012!001
0121000121 0"0001
00~0et0 01210007
0021021121 0"0001
00Ql0012! ~0001211
000000 121~12112!2J1
00e!~eJl2! 00121007
21000121121 \I5~0e101
00000121 1210121~~"
001211210121 0210007
ell2J~0~0

VJI2!0"~0

0000012!

00121~flI7
0~0007
0~12101217

02)12!007
0210000 12!" ',H~ 0 7
2!12I0~0121 0"0007
~12100"0 0"'02107
0"~0"" 0"0"07
210210210 21"0007
0000"0 0"121007
"00~00

21000210

0e'0~07
0~0001

2l0"0~0
012!~00"

,,012! 012! 7
2100001

2!(CH100121
"0"210121

1t1~121007

0000~"

~HHH'!"12!

~"'~"~1211

~0121007

eJ00001
012!02121I2J 0 0 02107
21I2!eJ00121 000012!7
i2l12!~0012! 00012107
00fZ10eJ0 0~0"'01
00 ra i2l 0121 000007
0000121121 0~0e!01
~00~012! 00121007
000000 12100""7
12l0I(H~~H' 00"0,,7
0~H~0~0 121"0007
i2leJ"0"0 0021'1107
012100021 ,,00 121 12! 1
2102112100 12100007
"121300121 00121146
~00""0

FOROTS
In the sequential access binary file, the second record crosses
l28-word disk boundary and contains a 002 (CONTINUE) LSCW.

C
C

~OOK

AT A BINARY r!~E AND SEE THE ~O$!CA~ SE~MENT
CONTROL WOReS.
OPENCUNIT z 1,MODEa I BINARY' )

1-'

WRIT(C1) (I,

J-,
WRIT((1)

J=~'l~~)

(J,K=1,100)

END

~ fi" "21 0 0
001}001

~010'H~ ~0~145
0~~"00 ~"0005

00~}043

0~2'~00

~00005

~000"0

00'''~05

000~"5
~'HHH'0 ~00""5

(J00005

00.3003

~000~0
00~00~

00k'044
00(1045

~0121"05

00.:.~046

0"V004

0~"H"~0

21012101215
0"12101215

00~H~47

00,~002

00;J005
00 ;~006
0~.~007
00,~01~
00'.~P.!11

00 ..:012
~0~013
0~/014

00;'.015
00\1;016
00'017
2101,020
210,~021

21~H:'022
fZl0;~fZl23
I2lfZl.~024

~"0tJ~0

00~Ql30

0~'HH~5

00r.00~
~000~0
000~00
e,"0Ql0~

00121005
0 0 0005

002J"~0

00 '''Hl 5

0"'''H~5

a"'0~05

0~H~ra00

,,00005

0"0121"0
00000121

~00005
~01210~5
~0012105

0~H'HH'0

00;~055

000000 000005
0'HH31210 0~0005
000~H35
r2l~12I005

0000~H3
~0~000
~H'0t100

0 0 031215

0000210

0~00"5

0"0005
0002105
0P102105

00:'063

~"'0021121

~Z0005

2121;~2164

0000",,,

~000"5

210.1065

021~210121

~0121005

2l2l00(Hl 017'0~05

00~2166

0"'121005

2!0~02J0

0~0~05
~H'!" eI 05

"'022167

0021005

021:'071
00?kl72

000000
0012!r2!00
02100P121
0002'021
021~0~0

12l"'H~"5

000~~0

0'-"0005

00210~21
~21e;0~0
2l0"'0~121
2l2l~0~21

"~0~215

021~~031

2102101210 ~H'0"'''5
2121212100 0~00"5
0I{H~0021 02'0005
0000021 0"0005
0210000 00121305
21121000121 00210215
00210"''' ~e!0~215
v"H'I000 1Zl12!12l005
~elI2I0W3e1 2100005
~00000 0"-'0",,5

021;)2141
12'21;;'042

0000eJ0

00~' 060
00l~61
021 ~·;2162

2100212'21

00~:~0421

~00005

00012100

00,~054

00 . H2!56

1210001215

12l0ll2l25
00/026
"0.~ 027
"0~~ 12130
00:JfZl32
000033
0fZlG034
021\32135
021I.H3-36
00?2137

00i1050
00J051
00J052
002053

0"~2J'HI 000~05
00~0~" ,,01212J0S

000~210

0000~21
21000~0

1210.:~057

0r2!~070

~0121005

12l0r~~073
12!0~~74

0"00~21 ZI~0"21S

12l0r075
210?076
0rcH~2177

12l0;:l100
210"101
021;~1"2

210,}'103
a001f(}4
~0Jl215

D-8

2J00005

1Zl"0005
0 0 121005

"liH' " 0121 0"'~H3215
002'000 21°"1210'
r2!02'~30 02121"'"
0210"0121 0~12"'05
00e'''021 3"0005
00~012l0
~000021
012l~000

~~00"5
~0"12l2l5

0012l0et5

°21tH'I2I 21 a0"'121005
0 00e15
2J0~0"0
"r3021~121 ~0e1005

the

FOROTS

001106
00Jl"7
"'0~11121
"'12I~~111

"'00112
01Zl~11J
1Zl0~l114

00;:115
000116
01ZlJ117
1Zl1Zl012121
12I1Zl~1121

1Zl0~122
12I1Zl~123

"'1Zl;.'124
003125
00~126

12100127
00013"
"0~131

"00132
"0~133

"00134
"0~135

00~a36

"00137
"0~~14"

"0r.141
00e142
"0~143
"'0~144
"'''~~145

210[1146
00J147
00~-150

"00151
1Zl0~~ 152
"02153
"0'~154
"0~' 155

000156
00?151
001160
00~161

00;1162
00 ~~ 163
"'f2J;~164

001165
1Z102166
"00167
1Z10~l1'0
"~H3171
00~.l172

01210~~121
~12I0~eJl2I

~0Z17J
1Zl0;~174

21"'121"1215
;301211211215
012100"121 121012101215
IZl 121 121121 0 121 121"'1211211215
~121"k:1121121 121 0 1211Zl!2l5
0121"121"121 1210121"1215
12112101210121 0012101215
001211210121 12100"05
1Zl01Zl1Zl0121 121012101215
121012112100 12101211211215
0001210121 121"'0""5
"0",121"0 000""5
"0121""121 00001215
0121001('JI2I ~0001215
1Zl01Zl1Zl1Z1121 2'012112105
""1210210 0"'121121"5
1210121""5
"""0"121 121"12101215
" " QJ 'Hila
0"~00" """00'
00"""" 0"'''00!5
"1210""0 0"001215
"121"""" ~"'00"5
~0121"0"

0~H'1'5

00Z176
"0~177

"'121 ~120121
"'1Zl~~201

1Zl0i21Zl2
1Zl0e:21ZlJ
"'002(214
00,..'121215
1Zl1Zl0206
I2I'H~ 21217
1Zl1Zl;: 210
00~211
1Zl0~212

00C213
002214
002215
"0~216

00Z21'

1Zl0~22~
00~221

"""""5

"00"1ZI0 0"0"05
0"0"00 021""05
~""02!" 0 0 0"05

0"0222
00e22:5
000224
"0e225
2100226
002227
002!230
000231
002232
000233

0~"005
""""""
21""0",,
121"001215
""121000 """""5

"0"""0 0"12101215

"",,0"0
el03""" 021121146

~2!"0"5

"1211121"" ~ ~ " " 3 2 -

~"""00

"'~""'"

"0,,000 0"'0"1217
el0eJ0"0 00~HH:'!7
"000"0 121"'''''''7
"000"0 21~0rJ07
el02'~00 "n0"01
IZl~HlJ021121

elf2Jel012J12I
~0000k'1

Number of
words to
next LSCW.

0~h~234

002235
00~236

00;3237
00;)240
00e241
000242

"~"0~7
~00007
0~f2J0f2J7

00~24J

002244

00000121 00"'12101
0~0~00
i2lf2J0~~0
i21000~121

00~245

~01210'"
'-"~0~12l7
~eJ00"7

f2J0l246
0~H~247

00/250
00e251

21121000121 ~012101217
0000021 (10rcHl!{2J7
021210"'121 ~0"12107

00~l252

12'0;1253

~00~"12I ~0"'001

00~~254
1Zl0~255

1210121000 ,,000~7
0121021210 121~"0217
000021121 0"0"'1217

00J256
00;~257

D-9

1Zl0~H'0121 2' '1211211217
0121 00 el 121 00 ItH' 121 7
0121 IZllZl eJ 121
~"I2I"Ql'
12112101Zlell2l i£I 0 121 "121 7
IZl~H~"0121 (30121"1211
1Zl12l2~"12I 00121114 ... -Continue LSCW.
1210012121" 3"'12101217
"'1211210"121 ~0""12I7
1Zl01Zl1Zl1Zl" 00~HHH
000~HHI 0"12101217
"0,,0121" ~1Zl"1Zl1217
""1211210121 1210121eJI2I7
0"2'0~121 121"'121007
01211Zl1Zl1Zl121 121012112107
"",,21"121 121"121"07
01211210121121 00"0"7
1211210000 1210121"1217

elI2I00~"

0~HH'12I7

"0Q10~" ~00~"7

~"2!~00

""0"2'1
rJl2I"0"0 0"121"'"
""rJ0 21 121 "~121"'"
""130~" 0"""0'
""e"~" 021""0'
121"0"01
o
" '''' 121 " 0 0 ""07
0""0"0
01210~0" 0"0""7
0""""" 0""007
00"0~" 0~"00?

01210""0
0000""
"0"121""
"121"0""
el0e10"0

0"'0"07
0"121"01
0"00"7
~""007

2100211217
~0""121" ""0007
2l"HH'~" 0 0 121"07
"0"""0 0"121"1217
el0210021 ,,00~01
O000"" "r1121~01
""0"~21 (1301210211
"00000 0"0"07
0eJet0021 0 0 001211
000{2J",,,, "''''''''01
~H:H"0"" 0"0e107
00210021 002121121'
012100"'0 21"'''00'
J12I0"00 f2J~"~01
000~00 0"1212101
12112100021 00"001
"1210000 000Ql01
121121000121 2100"'"
0000"" 00012107
~"00"" 0~~0'"

FOROTS
~0026"
1Z10~261

"E"~262

"0Z263

"~"3264

IlIrtH~265
00~~266

"0.,267
00:~ 27"
000271
00i272
00;1;273
12l0~~214

00Z275
"0~j276

""""""
~,,~"""

"~"eJ07
0~"""7

"0i!27?
"0e3"0
"003"1
"0;;'3""2

000""" """"07
~"00~" 0~"""7

"0"0,,,

"~HH'J""
"0021"" ""'0"0'
r2J ~HHH'I " 0 '" '''' 0 7
""""""
00""00 000""7
000"07
"1c:I00"" 0""21211
"0"0"" 0~0"07
"000"0 0"'0"0'
"""~"0 000"21'
""""""
"'00000 0"0"07
0"0"'''7

0~H~303

"00304
000305
00J306
00~307

"0~310

"00311
"00312
00i'l313
0el314

Image mode files contain no LSCWs.

C
C

"01210"" 21"00",

0000"" 0"""0'
2100""0 ~"0"0'
"00000 0"0"07
"0"1lI1ZI0 "et0"0'
""r2J0"" ,,~""'"
"""00"" 00 rtH'l 07
"0""00 21"""07
"0"0"" ""0"07
0210""" 0~1c:I"07
00"""" 0""""7
"0"0021 0""""7
"0"0210 ,,0"2107
"Ql3"~" ""0147

You cannot backspace this file.

AT AN IMAGE MODE rI~E AND SEE NO LOGJCA~ SEGMENT
CONTROL WORDS.

~OOK

OPENCUNI Tc 1,MODED'IMAGE')
Ia5

WRITE(1) (I, J=l,l21'"
J='7

WRITE(1) (J,K=1,1210)

END

"0
k~"" 0
121121~12I01
00"01212
0121001213
00?01214
0~HH'''5

0001211216
00.~1210?

1210001121
0121~011
l2I~hJ12I12
0121~12113
00~12I14
1210~312115

00:1016
00012117
0121002121
12l0J021
001022
e10.?1023

00J024
12100025
1210e12l26

210""00 0 0 0""5
1211210121121121 0012112105
"121~00121 1210''12105
~121"00121 001211211215
1211211211210121 00121005
ell2l00 121121 0001211215
QI~}210 ~H' 0"'01211215
I21121e'000 0012101215
121121012100 0 0 01!ll2l5
el0000121 ~2!"12l05
1211211211210121 el2ll2l~12I5
e!12I000121 12100"1215
0121000121 ~00005
01210kl"0 ~001211215
0121121000 12Ie!~H!l12I5
0"001211Zl 12101Zl01215
~000~0 0 0 0r2J05
~121"000 1210001215
e1121121121~0 0121121005
"0I2HH,,,, 0r;,0~05

0121~12121

1210 Il 121 3121
00012131
12102!12132
0121012133
1!ll2l~034

121021035
00J036
00012131
11'001214121
121 0012141
000042
00;312143
00""044

D-IO

00001210

fc.1"0121121~

012100~121

~12l001215

12101210121121 0 0 21005
0000121" ~O12l005
"0000" 0"1211211215
12100121121121 001Zl00S
~12It:!l000 021121005
rtHHH~12I0 12112112101215
"00121"121 ~~001215
00001210 001211211215
1210121121121121 0 0 01211215
01Zl00~0 0"0"'1215
01211211210121 0001211215
001210~0

~00"05

00,,1212'121
0012leJI2I0
00021121121
01Zl1210121"

o "'1Zl 121 1215
0"'012105
01Z1001Zl5
00001215

00~12145
00~046

"l2IkHHHJ

000"41

0002!~0

~0001215

el2!1Zl005

FOROTS
~~0"5~
~00"51
~~0052

00~"53

0"H'!054
00;'055
210J056
12100057
~0~060

1210?061
0~H~062
00~063
~0~~064
00\~065

00.1066
1210~·067

1210~07"

121 02071
121 0 ;~072
"0:'2173
00 ,;~" 7 4
00~075

1210~0'6
00~077

l2Ie<110"
00()101
1210?lel2
"0011213
003104
00~~105

00?11216
000107
1210ell'"
01210111
002112
"01113
1210~114
0121~115
"0v~116
00~11'
00~122J
0~H'121

01210122
000123
000124
"'12I{l12,
00~126
00~127

0"~132J

1210e131
0121~132

1210fl13J
12100134

"""'0~9

~00135
~00136

k""H~
~'" ~0"00~
~~02100 0r2l0QJ05

00e137

~~~0~~

k"iH'l 'H~" Qlk'J"'''05

00~14"
00~141

0~0000
~0210QJ0

.,00005
21 0 02105
0001210121 0001211215
000121~0
~0"0~121

~00"0~

000"~121

01210"00

0e1riHH~5
~0000~

2'00e'~0

~00"05

00001210

JeJ00~5

12100142
1210ta43
0r,h~144
I2I~H~145
0~.a46

eI r2I 0121 05
000000 000005

210{,147
00~150

0"~~~" .J~"~05
0'H'''~0 ~"~0005

1210,1;151
2100152
00:a53
"'H:154

0riHHHljl2l

~~0005

"0~"'~0

0"'00215
J"0005

12102156

012100~0

00~1155

0~H'.157

00k')16121
00\?161
008162

~""'0~12I "~12I~0~
0fZl~~~0 '" el 0 v,,~ 5
~000"0 ~Q!0~12I5

00~16J
0~W164

00e'1fZl00 0011:H'l12l5
0000e10 ,,00(;'l05
et12l00~121
0~H'I00121

el0J165
12102166
1210 ~~ 16 7
1210 ~~ 17"

0~0~12I5
~~~01215

002101210 01210005
0~Hl000 000005
~1,Hl!00~ 3 eJ 012105
el00000 210012105

"' ' ~H'"''
~1,H~12I00

1210;·~171
00.~172

1210l17J

0~"'0fZl5
000121~5

1210~174
00~175
00(~176

02100121121 eHi'l0 0 1215
000000 ~eJ001215
~00"00 0 e1 00215
0~HH'12I121 01211211211215
00000121 21"12112105
~02112100

2100177
001~2~"

12100201
00~202

0~001215

1210:'1203

02Jel002J ~0"12I2J5
02J0"12I0 2J001212J5
1212J12I'''~0 000121"9
0"121121121121 3"'''005
"0012100 ~0"0"5
1212J0121~0 0"'01211215
~012121"'0 121"''''005
21 Q""~ I2IPJ 012101211219
""1210121121 121 e 0121 121 5

00~2Q14

0(21('.205
00;)206
k"h~21217
0121"~21121
00~211

1210\-1212
12100213
00~214

~0"0"9
121""""",
0"1211211210 0~2J2I1215

01210215
1211210216

012112101210 0"'~H"05
12l121I212JPJ0 121012112105
121012101210 0121012105

0"~217
121121~22121

00iJ221

0"12I~12I121 121~"005

D-ll

0~HI~00

~"HH'I00
~00~00

0"e~05
~"~Hl05

000005

~"~"000 00"~05
~0012100 0"'0~"'5
0~"1211210 21i21ICHH;'15

121002100 ~00"H!l5
00000a 000007
1Zl00000 ,,0"~"7
~0000'" 211210001
1Zl!,H~"00 Ql0012107
0210"1210 0002107
~0"01210

0~01210'

002101210
0021121"0
00121f2Jel0

0~012107

~1210"1210

00012107

21"'0~07
~HHHl07

00121"00

~eJ"12I07

0~00210
~12I0~~21

el00007

0012101217

"'''~000 ~~00~1
~"~0k"121
000"~"
0~0Q11210

~12I0~~0

""'0007
00121007
0~012107

0et0007
012121021121 0e1121007
000000 000007
0012100~ 00121007
'" 0 2J 0 ~H?J 000~07
001210"'0 ~""0007
Ql~H~000 0012112107
021 0r,H~"~ 000"1217
000~~0 0"'001217
0000~H?J
~~0000

~Ql0eJ"'7

0"'~12102J

~"2J007

00012107
021021210 21121001217
0021121e10 00121007
0021000 Qj02Jk'l07
el2J00~0 121210"07
012!00~12I (21121012107
2I0~i2I~0 ""'12100'
00e1i21"0 00121e!07
0000~'" 0121001211
12I"'0121~0 121"'012107
02J 121121 121 0 1ZI"'''l'l07
211210000 0Q\l2Iet07
~01211211212J (21001211211
0012121"0 0~"et07
~ 2J 121 ~H"'J 012112112112!1
12112I~l2Iell2I 20"007
~0121000 elet0"'217

""~12I1210121 ~21012107

FOROTS
1lJ"~222

0"~223
~00224

0"~225

"eJ226
"0~22'

00023"
""H~231

"0~232
"0~233
""H~234

"0Ql235
"0~236
"0~~237
"0~240

"0l241
"0·~ 242
000243
12101'244
12J0~245

"1210246
00,3247
00e25121

I2IrcH'251
001252
12102253
"0?254

0.5.3

0""~~1
"""""~
0~"~0" 0""0",

"",,,,,,,,,

""""' ' ' ""''''''''
""""""
"'0""'"
"""""" 0"''''''''

0",,0Z10 ,,00"'"

~0000~

0"""'"
"0"",,.,
0!2JeI""" 0"0007
000264 ~""""0 0 1l1 """1
"0""""
0010266 "'kHI""0
"'0"0210 0 0"07
:2'!2J~0"!2J

0"0~"~

"0~~265

0~"""0 0~""07

"0?26,
000270

0"0""0 0"""'"
0~"000 "'-'''''''7
00(('100" 0"0"0'
00"0e'0 0"""0'
"00000 ""'~0"7
21000"0 "~~00'

"0~~2'1

"0:?272
"0~27J

"0r1274
00~'275

"000~12J ~00~12I7
0~Q!00" ~002l07
211210000 ~01210'"

I2IlZlil276
12J0'~217

000300
0020301
001~ 30 2

c,~21e!1Zl7

00(2100121 Z0121eJ07
000121~0 121 0 01210'

eJ0~~3eJJ
0~h~304
00~305
1210 ;~306

00~0"7

0121021121121 0 0 0007
~0000~
el02'0~121

"'01('261
00'~262
"0~~26J

0~"",,,,, ""0"07
"0"""" Z"""07
1210"""" 000"07
"~00~" r210iH'",

~00000

000007
00""'0'
2!kHH' 210 ""'021!2J'1

"0~26"

riHH.H~""

00k'0~121

0~""00

"'02255
002256
"'02257

~000"7
02!el~e'i

012103",.,

""21021~ ,,00~0T

"0000" 0P.!00217
"0"0"" 0"'''007
""0f21~"~ 0"'''''0'
0"""~Hl' "'''''''07
"000"0 0"'0"07
000000 ~eI"0~7
I2IkHH:'l2l0 0"""01
01210"0121 12I~01210'
"121000121 """'0'"
0121"0~", a0121l'11217
etl2l1210"0 0'='''1210''
0"0"00 1210"121121'
21001210121 1210121001
01210"00 1Zl~00"7
12102101210 0001210'
01211Zleel0 12I~"007

Mixed Mode Data Files

FOROTS permits files containing both ASCII and binary data records to
be read.
Mixed files may be accessed in either sequential or random
access mode.
Logical ASCII and binary records have the same format as
described in the preceding paragraphs.
In random access mode, the
record size must be large enough to contain the largest record, either
ASCII or binary.

0-12

FOROTS
D.S.4

Image Files

The image data transfer mode is a buffered mode in which data is
transferred in a blocked format consisting of a word count located in
the right half of the first data word of the buffer
followed by the
number of 36-bit data words.
The devices that permit image data
transfers and the form in which the data is read or written are:
Device

Data Forms

Card Reader

All 12 punches in all 80 columns are packed into
the buffer as 12-bit bytes.
The first 12-bit byte
contains column 1. The last word of the buffer
contains columns 79 and 80 as the left and middle
bytes, respectively.
Cards are not split between
two buffers.

Disk

Data is written on the disk exactly as it appears
in the buffer.
Data consists of 36-bit words.

Magnetic Tape

Data appears on magnetic tape exactly as it
appears
in
the
buffer.
No processing or
checksumming of any kind is performed by the
serV1ce routine.
The parity checking of the
magnetic tape system is sufficient assurance that
the data is correct. All data, both binary and
ASCII, is written with odd parity and at 800 bits
per inch unless changed by the installation.

Plotter

Six 6-bit characters per word are transmitted to
the plotter exactly as they appear in the buffer.

D.6

USING FOROTS

FOROTS has been designed to lend 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, CLOSE, RELEASE, etc.
FOROTS will provide automatic memory allocation, data conversion,
buffering, and device interface operations to the MACRO user.

D-13

I/O

FOROTS
D.6.l

FOROTS Entry Points

FOROTS provides the following entry points for
FORTRAN compiler or a non-FORTRAN program:
Entry Point
ALCHN.
ALCOR.
CLOSE.
DBMS.
DEC.
DECHN.
DECOR.
ENC.
EXIT.
FIN.
FIND.
FORER.
FUNCT.
IN.
IOLST.
MTOP.
NLI.
NLO.
OPEN.
OUT.
RELEA.
RESET.
RTB.
TRACE.
WTB.

D.6.2

calls

from

either

Function
Allocate software channels
Allocate dynamic memory blocks
Close a file
DBMS interface
DECODE routine
De-allocate software channels
De-allocate dynamic memory blocks
ENCODE routine
Terminate program exeuction
Input/Output list termination routine
position to the next record (RANDOM ACCESS)
Error processor
Overlay interface
Formatted input routine
Input/Output list routine
File utility processing routine
NAMELIST input routine
NAMELIST output routine
Open a file
Formatted output routine
Release a device (CLOSE implied)
Job initialization entry
Binary input routine
Trace subroutine calls
Binary output routine

Calling Sequences

You must use the following general form for all calls made to FOROTS:
MOVEI
PUSHJ

l6,ARGBLK
l7,Entry Point
(control is returned here)

where:
1.

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 list given in
Paragraph D.6.l)
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:

D-14

FOROTS
1.

An error return to a specified statement number,
or WRITE statement ERR=option,

An end-of-file return to a statement number,
WRITE statement END=option,

i.e.,

READ

A fatal error that returns to
package.

the

monitor

i.e.,

READ
or

debug

Paragraphs 0.6.3.1 through D.6.3.ll give the MACRO calls and required
argument block formats needed to initialize FOROTS and FOROTS I/O
operations.
Argument blocks conform to the subprogram calling convention described
in Appendix C.
However, there is one exception in dealing with the
first word of an I/O initialization call, i.e., WTB., ENC., RTW.,
etc., for a FORTRAN logical unit number. In previous versions of
FOROTS and FORTRAN-20, if the indirect bit was not set, the argument
was immediate; if it was set to 1 (one), the argument was the address
of the variable. The type field was always a (zero).
with Version 4 of FORTRAN-20 and Version 4 of FOROTS this convention
has been changed.
If the type field of the first word of an I/O
initialization call for the FORTRAN logical unit number is 0 (zero),
the argument is an immediate mode (18 bit) constant wherever possible.
If the type field is integer, the argument is indirect
(see Appendix
C, Table C-l, Type 2).
This exception should not cause any upward compatibility problems,
since all previously working programs will still function. An added
feature with this convention is that it permits the following
construct to be correctly implemented:

100

0.6.3

N=-4
READ (N, 100 ) I, J
FORMAT (215)

!SET FOR TERMINALS

MACRO Calls for FOROTS Functions

The following paragraphs describe the forms of the MACRO calls to
FOROTS that are made by the FORTRAN-20 compiler. The calls described
are identified according to the language statement
that
they
implement.
The following terms and abbreviations may be 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.)
u

a FORTRAN logical unit number

FORMAT statement address,

v·

the name of an array containing ASCII characters,

list

an Input/Output list,

D-15

FOROTS

the statement to which control is transferred on an "END
OF FILE" condition,

the statement to which
"ERROR" condition,

name

a NAMELIST name,

a variable specifying
random access mode,

list directed I/O;

type

type specification of a variable or constant,

control

the

logical

transferred

record

number

for

the FORMAT statement is not used,

where ARGBLK is
0-8

9-12

14-17

-6
Reserved

Reserved

18-35
0

type

Format Size (in words)
v

D.6.3.l I/O Statements, Sequential Access Calling Sequences
The
READ and WRITE statements for
formatted sequential data transfer
operations and their calling sequences are:
READ(u,f,END=C, ERR=d) list
MOVEI 16, ARGBLK
PUSHJ 17, IN.
and
WRITE(u,f,END=C, ERR=d) list
MOVEI 16, ARGBLK
PUSHJ 17, OUT.

D-16

POROTS
where ARGBLK is
0-8

9-12

14-17

18-35

-5

Reserved

type

Format Size (in words)

The READ and WRITE statements for unformatted sequential data transfer
operations and their calling sequences are:
READ(u,END=C, ERR=d) list
MOVEI 16, ARGBLK
PUSHJ 17, RTB.
and
WRITE(u,END=c, ERR=d) list
MOVEI 16, ARGBLK
PUSHJ 17, WTB.
where ARGBLK is
0-8

9-12

14-17

-3

18-35
0

Reserved

type

1
Reserved

0.6.3.2 NAMELIST I/O, Sequential Access Calling Sequences - The READ
and WRITE statements for NAMELIST-directed sequential data transfer
operations and their calling sequences are:
RE.AD (u,name)
READ (u, name, END=c, ERR=d)
MOVEI 16, ARGBLK
PUSHJ 17, NLI.
and
,-

WRITE (u, name)
WRITE (u, name, END=c, ERR=d)
MOVEI 16, ARGBLK
PUSHJ 17, NLO.
0-17

FOROTS
where ARGBLK is
0-8

9-12

14-17

18-35

-4

Reserved

j
Reserved

type

NAMELIST table address

The NAMELIST table is generated from the FORTRAN NAMELIST.
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+3)-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
form,
first
in the table.
Each name is followed by a list of
arguments associated with the name;
this argument list may be of any
length and is terminated by a zero entry.
The name argument list may
be in either a scalar or an array form
(refer
to the following
diagrams).

D.6.3.3 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 = Ul-Ll+l
52 = U2-L2+l
etc.
In order to calculate the address of an element referenced by
A (Il,I2,I3, ... In)
the following formula is used:
A+{Il-L1)+(I2-L2)*Sl+{I3-L3)*S2*Sl+ ... +{In-Ln)*S[n-l]* ... *S2*Sl
The terms are factored out depending on the dimensions of
and not on the element referenced to arrive at the formula

the

array

single

A+{-L-L2*Sl-L3*S2*Sl ... )+Il+I2*Sl+I3*S2*Sl ...
The parenthesized part of this formula is the offset for
precision array and it is referred to as the Array Offset.

D-18

FOROTS
For each dimension of a given 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*(-Ll-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.
SCALAR ENTRY in a NAMELIST Table

.11

12.

.14 15.

.17

. .35

18.

SIXBIT/SCALAR NAME/

Scalar addr

12-14

15-17

18-35

ARRAY ENTRY in a NAMELIST Table
0-8

9-11

SIXBIT/ARRAY NAME/
#DIMS

type

ARRAY SIZE

OFFSET
I

Factor 1

Factor 2

Factor 3

.
.

D-19

Factor n

FOROTS
D.6.3.4 I/O Statements, Random Access Calling Sequences
The READ
and WRITE statements for random access data transfer operations and
their calling sequences are:
READ (u#R,f,END=c, ERR=d) list
READ (u#R,END=c, ERR=d) list
MOVEI 16, ARGBLK
PUSHJ 17, RTB.
and
WRITE
wRITE
MOVEI
PUSHJ

(u#R,f,END=c, ERR=d) list
(u#R,END=c, ERR=d) list
16, ARGBLK
17, WTB.

where ARGBLK is
0-8

9-12

14-17

18-35
0

-6
type

type

format size (in words)

Reserved

address of
Record Number

Reserved

f and the format size in words are 0 if the I/O statement is
unformatted.

D.6.3.5 Calling Sequences for Statements That Use Default Devices The FORTRAN-20 statements that require the use of a reserved system
default device and their calling sequences are:
Default Device
ACCEPT f, list
READ f, list
REREAD f, list

(TTY)
(CDR)
(REREAD)

UNIT=-4
UNI'I'=-5
UNIT=-6

MOVEI 16, ARGBLK
PUSHJ 17, IN.

D-20

FOROTS
where ARGBLK is
0-8

9-12

14-17

18-35

-5
Reserved

type

Reserved

type

Format Size
(in words)

Default Device
PRINT f, list
TYPE f, list

UNIT=-3
UNIT=-l

(LPT)
(TTY)

MOVEI 16, ARGBLK
PUSHJ 17, OUT.
where ARGBLK is
\'-....._.

0-8

9-12

14-17

-5
Reserved

Reserved

18-35

type

D-21

format size (in words)

FOROTS
D.6.3.6 Statements to Pos~ition Magnetic Tape Units The FORTRAN-20
statements that may be used to control the positioning of a magnetic
tape device and their calling sequences are:
Function
(FORTRAN Statement)

FOROTS Code

SKIPFILE (u)
BACKFILE (u)
BACKSPACE (u)
ENDFILE (u)
REWIND (u)
SKIPRECORD (u)
UNLOAD (u)

3
2
4

5
1

CALL:
MOVEI 16, ARGBLK
PUSHJ 17, NTOP.
where ARGBLK is
0-8

9-12

14-17

-4

18-35
0

Reserved

type

Reserved

type

FOROTS code

D.6.3.7 List Directed Input/Output Statements You may write any
form of a sequential Input/Output statement as a list-directed
statement by replacing the referenced FORMAT stateme~t number with an
asterisk
(*) .
The list-directed forms of the READ and WRITE
statements and their calling sequences are:
READ (u, *, END=c, ERR=d) list
MOVEI 16, ARGBLK
PUSHJ 17, IN.
and
WRITE (u, *, END=c, ERR=d) list
MOVEI 16, ARGBLK
PUSHJ 17, OUT.

D-22

POROTS
where ARGBLK is
0-8

9-12

14-17

18-35

-5
Reserved

Reserved

a
a

D.6.3.8 Input/Output Data Lists - The compiler generates a calling
sequence to the runtime system if an I/O list is defined for the 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:
MOVEI 16, ARGBLK
PUSHJ 17, IOLST.
Any number of elements may be included in the ARGBLK.
The end of the
argument block is specified by a zero entry or a call to the FIN.
entry.
Mnemonic Name

FOROTS Value

DATA
SLIST
ELIST
FIN

1
2
3
4

The elements of an I/O list are:
1.

DATA
The DATA element converts one sing1e- 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.
0-8

9-12

14-17

18-35

DATA

type

SCALAR ADDR

D-23

FOROTS
2.

SLIST
The SLIST argument converts an entire array from internal to
external form or vice versa, depending on the type of
statement, i.e., READ or WRITE, involved. An SLIST table has
the following form:
0-8

9-12

SLIST

type

14-17

18-35

#ELEMENTS

INCREMENT

8ASE ADDR1.

For example, the sequence:
DIMENSION A(100) ,8(100)
READ(-,-)A
or
READ(-,-) (A(I) ,1=1,100)

!only when the IOPT switch is used

develops an SLIST argument of the form:
0-8

9-12

14-17

0
0
2
0

0
0
0
0

18-35

0
2
0
0
4

144
1
A
0

More than one base address may appear in a SLIST as
the increment is the same.
The sequence
DIMENSION A(lOO), 8(100)
WRITE (-,-) (A(I) ,8(1) ,1=100)

! only when the IOPT
switch is used

develops a SLIST argument of the form:

0-8

9-12

14-17

18-35

0
0
2
2
0

0
0
0
0
0

144
1
A
8
0

0
2
0
0
0
4

D-24

long

FOROTS
3.

ELIST
The SLIST format permits only a single increment for a number
of arrays to be specified while the ELIST permits different
increments to be specified for different arrays.
The format of the ELIST is
0-8

9-12

14-17

18-35

ELIST

type

No. Elements to
transfer
increment 1

type

Base ADDR 1
increment 2

type

Base ADDR 2
increment N

type

Base ADDR N

For example, the FORTRAN sequence
DIMENSION IC (6,100), IB (100)
WRITE(-,-) (IB(I) ,IC(l,I) ,1=1,100)
produces the ELIST
0-8

9-12

14-17

18-35

3
0
0
0
0
4

0
0
2
0
2

0
0
0
0
0
0

144
1
IB
12
IC
0

The increment may be zero.
sequence
DIMENSION A(lOO)
WRITE(-,-) (K,I=100)

This could

produced

the

!on1y when the IOPT switch is used

The zero may not appear as an immediate constant in the
argument block. The ELIST for the previous example would be
0-8

9-12

14-17

18-35
144

type

D-25

Pointer to a word
containing a zero

FOROTS
4.

FIN
The end of an I/O list is indicated by a call to the FIN
routine in the object time system. This call must be made
after each I/O initialization call, including calls with a
null I/O list. The FIN rout~ine may be entered by an explicit
call or by an argument in the I/O list argument block.
If
both calls are used, the explicit call has no meaning.
The
FIN .element has the following format:
EXPLICIT CALL:
PUSHJ 17, FIN.

D.6.3.9 OPEN and CLOSE Statements, Calling Sequences - The
calling sequences for the OPEN and CLOSE statements are:

form

and

OPEN STATEMENT CALL
MOVEI 16, ARGBLK
PUSHJ 17, OPEN.
CLOSE STATEMENT CALL
MOVEI 16, ARGBLK
PUSHJ 17, CLOSE.
where ARGBLK is
0-8

9-12

14-17

Negative of
the number
of words in
block not
including
this one.
0
0
0
G
G
G

·
·

2
7
7
type
type
type

type

18-35

I
I
I
I
I
I

X
X
X
X
X
X

·
·

·
·
·
X

u
c
d
H
H

The G field (bits 0 through 8) contains a 2-digit numeric that defines
the argument name;
the H field
(bits 18 through 35) contains an
address which points to the value of the argument.

D-26

FOROTS
The numeric codes that may appear in the G field and the argument that
each identifies are:
G Field

Open Argument

G Field

01
02
03
04

DIALOG
ACCESS
DEVICE
BUFFER COUNT
BLOCK SIZE
FILENAME
PROTECTION
DIRECTORY

12
13
14
15
16
22
23
24

06
07
10

Open Argument
MODE
FILE SIZE
RECORD SIZE
DISPOSE
VERSION
ASSOCIATE VARIABLE
PARITY
DENS I'I'Y

D.6.3.10 Memory Allocation Routines - The memory management module is
called to allocate or de-allocate memory blocks.
There are two entry
points, ALCOR.
and DECOR.,
that control memory allocation and
de-allocation.
Use the ALCOR.
entry to allocate the number of words specified in the
argument block variable.
Upon return, AC a will contain either the
address of the allocated memory block or a -1 value, which indicates
that memory is not available.
The calling sequence for ALCOR.
call
is:
MOVEI 16, ARGBLK
PUSHJ 17, ALCOR.
where ARGBLK is
0-8

9-12

14-17

-1
Reserved

18-35

type

Address of
Number of Words

Use the DECOR.
entry to de-allocate a previously allocated block of
memory;
the argument variable must be loaded with the address of the
memory block to be returned.
Upon return AC a is set to O.
If the number of desired words is N, ALCOR.
actually removes N+l
words from free storage. The pointer returned points to the second
word (word 1 as opposed to word 0) removed from free storage.
The a
word contains the negative value of N in its left half. This word is
used by FOROTS to maintain linked lists of allocated
(using ALCOR.)
and free storage.
The calling sequence for a DECOR.

call is:

MOVEI 16, ARGBLK
PUSHJ 17, DECOR.

D-27

FOROTS
where ARGBLK is
9-12

0-8

18-35

14-17

-1

type

Reserved

Pointer to word
containing
address of block
to be returned

D.6.3.ll Software Channel Allocation and De-allocation Routines - You
may allocate software channels in MACRO programs via calls to the
ALCHN.
routine and de-allocate 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 to ALCHN.
in the argument block variable.
Zero is passed in the argument block
variable 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 a contains the assigned number.
The calling sequence of an ALCHN.

routine is:

MOVEI 16, ARGBLK
PUSHJ 17, ALCHN.
where ARGBLK is
0-8
-1

Reserved

9-12

14-17

18-35

type

Pointer to a word
containing
the channel #
or zero

Use the DECHN.
entry to de-allocate a previously assigned channel.
The channel to be released is passed to DECHN.
in the argument block
variable.
If the channel to be de-allocated was not assigned by
ALCHN.
and thus cannot be de-assigned, AC a is set to -1 on return.
The calling sequence for a DECHN.

routine is:

MOVEI 16, ARGBLK
PUSHJ 17, DECHN.

D-28

FOROTS
\vhere ARGBLK is
0-8

9-12

14-17

-1

Reserved

D.7

18-35
0

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., and they are called by the sequence
MOVEI 16, ARGBLK
PUSHJ 17, FUNCT.
The general form of the ARGBLK is
0-17

ARGBLK~

18-35

Negative of the
number of words
in block
type
type
type
type
type
type

0
function number
error code
status
argument 1
argument 2
argument 3

.
.

type

argument n

where
type
function number
error code
status

the FORTRAN argument type (see Appendix C)
the number of one of the required functions
the 3-letter mnemonic output by the object
time system after ?,
%,
or [.
(See Table
D-l. )
undefined on the call and set on the return
with one of the values below.
-1

1 .... n

D-29

Function not implemented
Successful return
Specific error message

FOROTS
Table D-l
Function Numbers and Function Codes
Function
Number

Function
Mnemonic

8
9

ILL
GAD
COR
RAD
GCH
RCH
GOT
ROT
RNT
IFS

CBC

0
1
2

3
4

5
6

Function Description

Illegal function
Allocates memory from a specific address
Allocates memory from available core
De-allocates memory
Gets or assigns an I/O channel
Releases an I/O channel
Allocates memory from FOROTS
De-allocates memory from FOROTS
Returns the initial runtime from FOROTS
Returns initial runtime file spec.
from
FOROTS
Cuts back memory if possible

FUNCTION 0 (ILL) - This function is illegal.
The argument
ignored, and the function always returns a status of -1.
FUNCTION 1 (GAD) - This function
address.
The arguments are:
arg 1
arg 2

allocates

memory

from

block

specific

and

arg

and

arg

address at which to begin core allocation
number of words of memory to allocate

The return statuses are:

o core allocated (arg 1 and 2 unchanged)
1 not enough memory available in system
(arg 1
unchanged)
2 cannot allocate memory at specified address (arg 1
unchanged)
3 illegal arguments (i.e., address + size is greater
(arg 1 and arg 2 unchanged)
FUNCTION 2 (COR) - This function allocates memory
The arguments are:
arg 1
arg 2

from

than
any

256K)

address.

undefined
size of core to allocate

The returned statuses are:

o core allocated (arg 2 unchanged, arg 1 beginning address of the
allocated memory)
1 not enough memory available in system (arg 2 unchanged)
3 illegal argument (i.e., size is greater than 256K)
FUNCTION 3 (RAD) - This function de-allocates memory at the
address.
The arguments are:
arg 1
arg 2

specified

address of core to be de-allocated
number of words to be de-allocated

The returned statuses are:

o memory de-allocated
1 memory cannot be de-allocated
3 illegal argument (i.e., both the
greater than 256K)
D-30

address

and

the

size

are

FOROTS
FUNCTION 4 (GCH) - This function assigns an I/O channel.
is:
arg 1

The argument

undefined

The returned statuses are:

o I/O channel assigned (arg 1 channel number)
1 no I/O channels available
FUNCTION 5
argument is:
arg 1

(RCH) - This

function

releases

I/O

channel.

The

object

time

I/O channel number to be released

The returned statuses are:

o channel released
1 invalid channel number
FUNCTION 6 (GOT) - This function gets
system list. The arguments are:
arg 1
arg 2

memory

from

the

address at which to allocate memory
number of words of memory to allocate

The returned statuses are:

0 memory allocated (arg 1 and arg 2 unchanged)
(arg 1
1 not enough memory available in system
unchanged)
2 cannot allocate memory at specified address (arg 1
unchanged)
3 illegal argument(s)

and

arg

and

arg

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.
FUNCTION 7 (ROT) - This function returns memory
system. The arguments are:
arg 1
arg 2

the

object

time

address of memory to be de-allocated and returned
size of memory to be de-allocated and returned

The returned statuses are:

o memory de-allocated
1 memory cannot be de-allocated
3 illegal argument
FUNCTION 8 (RNT) - This function returns the initial runtime from
object time system. The argument is:
arg 1

undefined

The returned status is:

o always (arg 1 - runtime from the object time system)
This function is used only if the user desires a log file.

D-31

the

FOROTS
FUNCTION 9(IFS) - This function returns the initial runtime file
specification from the object time system.
The specification is
obtained from accumulators 0, 7, and 11 after the initial RUN command.
The arguments are:
arg 1
arg 2
arg 3

undefined
undefined
undefined

The returned status is:

o always (arg 1 - device from accumulator 11, arg 2 - filename
from accumulator 0, and arg 3 - directory from accumulator 7)
This function tells the overlay handler which file to read
initial RUN command.

after

FUNCTION 10 (CBC) - This function cuts back memory if possible and
used to reduce the size of the user job. There are no arguments.

the
is

The returned status is:

o always

0.8

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,
i.e.,
DEVTB.
Table D-2 shows the standard
assignments
contained by the Device Table.

0-32

FOROTS
Table D-2
FORTRAN Device Table

Device/Function
REREAD
CDR
TTY
LPT
TTY

DSK
CDR
LPT
CTY
TTY

FORTRAN Logical
Unit Number
-6
-5
-4
-3
-2
-1
00
01
02
03
04
05

Use
REREAD statement
READ statement
ACCEPT statement
PRINT statement
Not valid
TYPE statement
ILLEGAL.
DISK
Card Reader
Line Printer
Console Teletype
User's Teletype

06 through 15 not valid

MTAO
MTAl
MTA2
FORTR
DSK
DSK
DSK
DSK
DSK
DEVl
DEV2
DEV3
DEV4
DEV5

DEV39

t1agnetic Tape
Magnetic Tape
Magnetic Tape
Assignable Device
DISK
DISK
DISK
DISK
DISK
Assignable Devices

17
18
19
20
21
22
23
24
25
26
27
28
29

D-33

APPENDIX E
FORDDT

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
appendix.
Table E-I lists all the commands available to the user of FORDDT.
Table E-I
Table of Commands
Command

Purpose

Data Access Commands
ACCEPT

Modifies data locations.

TYPE

Displays data locations.

Declarative Commands
GROUP

Defines indirect lists for TYPE statements.

MODE

Specifies format of typeout.

OPEN

Accesses program unit symbol table.

PAUSE

Places pause requests.

REMOVE

Removes pause requests.

DIMENSION

Defines dimensions of arrays for
FORDDT
references.
(Unnecessary
if
/DEBUG:DIMENSIONS was used.
See
Table
B-2.)
Defines dimensions
of
double-precision
arrays for FORDDT references.
(Unnecessary
if /DEBUG:
DIMENSIONS was used.
See Table
B-2.)

DOUBLE

E-l

FORDDT
Table E-l (Cont.)
Table of Commands
Purpose

Command
Control Commands
START

Begins execution of FORTRAN program.

CONTINUE

Continues execution after a pause.

GOTO

Transfers control to some program statement
within the open program unit.

Traces execution of the program.

STOP

Terminates program and returns
mode.

monitor

Other Commands
LOCATE

Lists program unit names in which
symbol is defined.

STRACE

Displays routine
program status.

WHAT

Displays current DIMENSION,
PAUSE information.

E.l

backtrace

given
current

GROUP,

and

INPUT FORMAT

FORDDT commands are made up 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 or tab is required
following the command name.
FORDDT expects a parameter if a delimiter
(i.e., space or tab) is found.
Comments may be appended to command
lines by preceding the comment with an !.

E.l.l

Variables and Arrays

FORDDT allows you to access and modify the data locations in your
program by using standard FORTRAN symbolic names.
Variables are
specified simply by name.
Array elements are specified in the
following format:
name (Sl, ... ,Sn)
where
name
(Sl, ... ,Sn)

a FORTRAN variable or array name
the subscripts of the particular array.

You may reference an entire array simply by its unsubscripted name;
you may specify a range of array elements by inputting the first and
last array.elements of the desired range, separated by a dash(-).

E-2

FORDDT
Examples
ALPHA
ALPHA(7)
ALPHA (PI)
ALPHA(2)-ALPHA(5)

E.l.2

Numeric Conventions

FORDDT accepts optionally
FORTRAN-20 input formats:

signed

numeric

data

the

standard

INTEGER - A string of decimal digits.

FLOATING-POINT
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 (II).

COMPLEX - An ordered pair of integer or real
separated by a comma and enclosed in parentheses.

E.l.3

optionally

preceded

constants

Statement Labels and Source Line Numbers

FORTRAN statement labels are input and output by straightforward
numeric reference,
i.e., 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.

E.2

NEW USER TUTORIAL

The new FORDDT user can rely on the commands described below as a
basis for debugging FORTRAN programs.
These commands are easy to
understand and apply.

E.2.l

Basic Commands

The easiest method of loading and starting FORDDT is:
@DEBUG filename.ext/FORTRAN/DEBUG
FORDDT will respond with
ENTERING FORDDT

Just as an asterisk (*)
signifies FORTRAN-20's readiness,
the two
angle brackets signify that FORDDT is awaiting one of the following
commands:
OPEN

Makes available to FORDDT the symbol names in a
particular program unit of the FORTRAN program.
When a
program unit symbol table is opened, the previously
E-3

FORDDT
open program unit is automatically closed. When FORDDT
is entered,
the MAIN program is automatically opened.
The command format is:
OPEN name
This will open the particular program unit named and
allow all variables within that subprogram to be
accessible to FORDDT.
OPEN

START

with no arguments will reopen the symbol table
main program unit.

Starts your program at the main
The command format is:

point.

program

entry

the

START
STOP

Terminates program execution, causes all files to be
closed, and exits to the monitor.
The command format
is:
STOP

MODE

Defines the display format for succeeding FORDDT TYPE
commands.
You need type only the first character of
the mode to identify it to FORDDT.
The modes are:
Mode
A
C
D
F
I

Meaning
ASCII (left-justified)
COMPLEX
DOUBLE-PRECISION
FLOATING-POINT
INTEGER
OCTAL
RASCII (right-justified)

Unless the MODE command is given, the
mode is the floating-point format.

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:
F,D,C,I,O,A,R
MODE
with no arguments will reset FORDDT
setting of floating-point format.
TYPE

the

original

Allows you to display the contents of one or more data
locations.
They
are displayed on your terminal
formatted according to the last MODE specification.
The command format is:
TYPE list
E-4

FORDDT
where list may contain one or more arrays, variables,
array elements,
or array element ranges separated by
commas.
For example:
TYPE I, ALPHA, BETA(2) ,J(3)-J(5)
Each item will be displayed in each of the currently
active typeout modes as set by the last MODE command.
ACCEPT

Allows you to change the contents of a
FORTRAN
variable, array, array element, or array element range.
The command format is:
ACCEPT name/mode value
where
name

array,
array
the name of the variable,
element, or array element range to be
contains
modified.
If the field
an
unsubscripted array name or an element
range, it causes all the elements to be
set to the given value (see special case
for ASCII in Section E. 6) .

mode

the format of the data value to
be
entered.
If given, it must be preceded by
a slash (/)
and immediately follow the
name.
(Note tha t /mode does not apply to
FORMAT modification.)

value

the new value to be assigned.
It must
correspond in format to the given mode.
Data Modes

You need type only the first character of a data mode
to identify it to FORDDT.
If not specified, the
default mode is REAL.
The following
input modes are
available:
Mode

Meaning

A
C
D
F
I

ASCII(left-justified)
COMPLEX
DOUBLE-PRECISION
REAL
INTEGER
OCTAL
RASCII(right-justified)
SYMBOLIC

R
S

Example
/FOO/
(l.25,-78.E+9)
123.4567890
123.45678
1234567890
76543210
\BAR\
PSI(2,4)

An example of the ACCEPT command format is:
ACCEPT ALPHA 100.6
This changes the value of the variable ALPHA to 100.6
with the default input mode of REAL, since mode was not
specified.
PAUSE

Allows you to set a breakpoint at any label,
line
number, or subroutine entry in your program.
You may
set up to ten pauses at one time.
When one of these
pauses is encountered, execution of the FORTRAN program
E-5

FORDDT
is suspended and control
is transferred to FORDDT.
Also,
when a pause is encountered, the symbol table of
that subprogram is automatically opened.
The command
format is:
PAUSE P
where P is a statement label number,
line
routine entry point name;
for example,

number,

PAUSE 100
will cause a breakpoint at statement label 100
currently open program unit.

the

Note that subprogram parameter values will be displayed
when a
pause is encountered at a subprogram entry
point.
CONTINUE

Allows the program to resume execution after a
FORDDT
pause.
After
a CONTINUE is executed,
the program
either runs to completion, or
it
runs until another
pause is encountered.
If you include a value with this
command, the program will run until the nth occurrence
of the given pause or until
a different pause is
encountered.
The command formats are:
CONTINUE
or
CONTINUE n
Example
CONTINUE 15
will continue execution until the fifteenth
of the pause.

REMOVE

occurrence

Used to remove those pauses from the program previously
set up by the PAUSE command.
The command format is
REMOVE P
where P is the number of the statement label where
pause was set, i.e.,

the

REMOVE 100
will remove the pause at statement label 100.
Note that REMOVE with no arguments will
remove all
pauses;
therefore,
no abbreviation of the command is
allowed in this instance.
This precaution prevents the
accidental removal of all pauses.
WHAT

Displays on your terminal the name of the currently
open program unit and any currently active pause
settings.
The command format is:
WHAT

E-6

FORDDT
E.3

FORDDT AND THE FORTRAN-20/DEBUG SWITCH

Most facilities of FORDDT are
features; however, if you do
a FORTRAN program, the trace
available, and several of the

available without the FORTRAN-20 /DEBUG
not use the /DEBUG switch when compiling
features
(NEXT command) will not be
other commands will be restricted.

Using the /DEBUG switch tells FORTRAN-20 to compile extra information
for FORDDT.
(See Appendix B, Using the Compiler, for a complete
description of each feature.) The additional features include:
1.

/DEBUG:DIMENSIONS, which will generate dimension information
to the REL file for all arrays dimensioned in the subprogram.
The dimension information will automatically be 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 will
generate
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 will also generate 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
IIformat-label F". If they are generated, you will be able to
display and modify FORMAT statements via 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 will generate 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 has been placed
on a single input line, only the first statement will have a
FORDDT reference and line-number label associated with it.
This also applies to the :LABELS switch.

/DEBUG:INDEX, which will force 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 able 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.
(See Section D.l.5.)
Note that this switch has no direct
commands in FORDDT.

E.4

affect

any

the

FORDDT

with

LOADING AND STARTING FORDDT
1.

The simplest method of loading and starting
the following command string:
@DEBUG filename.ext/FORTRAN/DEBUG
E-7

FORDDT
FORDDT responds with
ENTERING FORDDT

The angle brackets indicate that FORDDT is ready to receive a
command,
just as an asterisk
(*)
signifies FORTRAN-20's
readiness.
The DEBUG command to the monitor will also load DDT (standard
system debugging program).
DDT can be used or ignored.
2.

You may wish to load your compiled program and FORDDT
directly with the LINK loader.
(Loading with LINK was
accomplished implicitly in the previous command string.)
The
command sequence is as follows:
@LINK
*filename.ext /DEB/G
*filename.ext /DEB:
FORDDT
FORTRA
*filename.ext /DEB: (DDT,

(loads DDT)
(loads FORDDT)

FORDDT
FORTRA

)/G

loads both DDT
and FORDDT

If the total FORTRAN program consists of many subroutines and
insufficient memory is available to complete loading with
symbols, it is possible to load with symbols just those
sections expected to give trouble.
The remaining routines
need not be loaded.

E.5

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 via 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
Chapter 5, Section 5.2), FORDDT will ask for the defined program name.
After you enter the program name, FORDDT will open the appropriate
symbol table. At this point, symbol tables in programs other than the
main program can be opened by using the OPEN command.
(See Section
F.5.)
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.

E.6

FORDDT COMMANDS

This section gives a detailed description of all commands
The commands are given in alphabetical order.

E-8

FORDDT.

FORDDT
ACCEPT

Allows you to change the contents of a FORTRAN
array,
array element,
array element range,
statement. The command format is:

variable,
or FORHAT

ACCEPT name/mode value
where
name

the variable array,
array element,
element range,
or FORHAT statement
modified.

array
to be

mode

the format of the data value to be entered.
The mode keyword must be preceded by a slash
(/)
and
immediately
follow
the
name.
Intervening blanks are not allowed.
(Note
that /mode
does
not
apply
to
FORHAT
modification.)

value

the new value to be assigned.
The format of
the
input
value must correspond to the
specified mode.
DATA LOCATION MODIFICATION
Data Hodes

The following data modes are accepted:
Mode

Heaning
ASCII (left-justified)
COMPLEX
DOUBLE-PRECISION
REAL
INTEGER
OCTAL
RASCII (right-justified)
SYMBOLIC

A
C
D

F
I

o
R
S

Example
/FOO/
(l.25,-78.E+9)
123.4567890
123.45678
1234567890
76543210
\BAR\
PSI(2,4)

If not specified, the default mode is REAL.
Two-Word Values
For the data modes ASCII,
RASCII,
OCTAL,
and SYMBOLIC,
FORDDT will accept a II/LONG II modifier on the mode switch.
This modifier indicates that the variable and the value
are to be interpreted as two words long.
Example
ACCEPT VAR/RASCII/LONG '1234567890'
will assume that VAR is two words long and store the given
10-character literal into it.
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 will be set to the given
value.

E-9

FORDDT
Example
ACCEPT ARRAY/F 1.0
or
ACCEPT ARRAY(5)~ARRAY(10)/F 1.0
Note that this applies only to modes other than ASCII
RASCII.

and

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 or RASCII,
the value
field
is expected to contain a long literal string.
ACCEPT will store the string linearly into the array or
array range.
If the array is not filled, the remainder of
the array or range will be set to zero.
If the literal is
too long the remaining characters will be ignored.
Example
ACCEPT ARRAY/RASCII 'ABCDEFGHIJKLMNOPQRSTUVWXYZ'
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,FlO.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).
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, will be assumed
to be one.
If a value is provided, it may be a numeric
constant or program variable, but it will be treated as an
integer.
When the value n is specified, the program will
continue execution until the nth occurrence of this pause.
For example,
CONTINUE 20
will continue execution after the 20th occurrence
pause.

E-IO

the

FORDDT
DIMENSION Sets 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
will allow you to redimension an array to have a larger
scope than that of the source program.
If this is done, a
warning is given. The command format is:
DIMENSION S
where S is the name of the array specified.
For example:
DIMENSION ALPHA(7,5/6,lO)
FORDDT will remember the
is redefined or removed.

dimen~ions

of the array until

The command
DIMENSION
will give a full list of all the
for all arrays.

user-defined

dimensions

DIMENSION ALPHA
will display the current information for the
only.

array

ALPHA

DIMENSION ALPHA/REMOVE
will remove any user defined
array ALPHA.

array

information

for

the

Arrays, Array Elements, and Ranges
Array elements are specified in the following format:
name [dl/d2, ... ] (Sl, ... )
where
name
[

... ]

(

... )

the name of the array
optional, and contains dimension information.
This form is equivalent in effect to the
DIMENSION statement.
the subscripts
desired.

the

specific

element

The entire array is referenced simply by its unsubscripted
name. A range of array elements is specified by inputting
the first and last array elements of the desired range
separated by a dash (-) (A(5)-A(lO)).

E-ll

FORDDT
DOUBLE

Defines the dimensions of a double-precision 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 entries and,
therefore, reserves twice the space.
The command format
is:
DOUBLE arrayname

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 code-generating source
line number provided that the /DEBUG:LABELS switch has
been used or the contents of a symbol previously ASSIGNed
during the program execution.
Note that the program must be STARTed before this command
can be used,
and also note that a GOTO is not allowed
after the ~C~C REENTER sequence.
(See Section E.7.)
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

the group number 1-8

list

a string of TYPE statements to be called in
future accessing of the current group number.

GROUP
with no arguments will cause FORDDT
current contents of all the groups

type

out

the

GROUP n
will type
requested.

out

the

contents

the

Note that one group may call another.

E-12

particular

group

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 format
is:
LOCATE n
where n may be any FORTRAN variable,
number, or FORMAT statement number.

MODE

array,

label,

line

Defines the default formats of typeout from FORDDT.
In
initial
default
mode,
variables will be typed
in
floating-point format.
If you wish to change the
typeout
modes, the command format is:
MODE list
where list contains one or more of the modes in the
following
table.
(Only the first character of each mode
need be typed to identify it to FORDDT.)
Mode
F
D
C
I

Meaning
FLOATING-POINT
DOUBLE-PRECISION
COMPLEX
INTEGER
OCTAL
ASCII (left-justified)
RAseII (right-justified)

A
R

A typical command string might be:
MODE A,I,OCTAL
NEXT

Allows you to cause FORDDT to
trace source
lines,
statement labels,
and entry point names during execution
of your program.
This command will only provide trace
facilities if the program was compiled with the FORTRAN-20
If this switch was not used,
the NEXT
/DEBUG switch.
command will
act as a CONTINUE command.
The command
format is:
NEXT n/sw
where
n

a program variable or
and

integer

one of the following switches

numeric

value

/S= statement label
/L= source line
/E= entry point
The default starting value of n is 1, a
trace.
The default switch is /L.
The command
NEXT 20/L

E-13

single

statement

FORDDT
will trace the execution of the next
20 source
numbers or until another pause is encountered.
Note that if no argument is specified, the
given will be used.
For example,

last

line

argument

NEXT /E
will change 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 will be 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
FORDDT automatically opens the MAIN program.
The command
format is:
OPEN name
where name is the subprogram name.
will reopen the MAIN program.

OPEN with no arguments

If the PROGRAM statement was used in the FORTRAN program,
the name supplied by you will be requested upon entering
FORDDT.
PAUSE

Allows you to place a pause request at a statement number,
source line number, or subroutine entry point.
Up to ten
pauses may be set at anyone time.
When a pause
is
encountered,
execution is suspended at that point and
control is returned to FORDDT.
Also,
when a pause
is
encountered,
the symbol table of that subprogram is
automatically opened.
The command formats include:
PAUSE
PAUSE
PAUSE
PAUSE
PAUSE
PAUSE

P
P AFTER n
P IF condition
P TYPING /g
P AFTER n TYPING /g
P IF condition TYPING /g

where
P

n
g

the point where the pause is requested,
an integer constant or variable or
element
a group number

array

PAUSE 100
will set a pause at statement label 100, cause execution
to be suspended,
and cause FORDDT to be entered on
reaching 100 in the program.
PAUSE #245 AFTER MAX(5}
will cause a pause to occur at source line number
245
after
encountering
this point the number of times
specified by MAX(5}.
Note that AFTER may
not
be
abbreviated.
E-14

FORDDT
PAUSE DELTA IF LIMIT(3,l} .GT.2.5E-3
If the variable LIMIT(3,l}
is greater than the value
2.5E-3, the pause request will be granted.
The IF may not
be abbreviated, but all the usual
FORTRAN
logical
connectives are allowed.
PAUSE 505 TYPING /5
will request a pause to be made at the first occurrence of
the label 505, and the variables in group 5 will be
displayed.
The
TYPING
specification
may
not
be
abbreviated.
PAUSE LINE#24 AFTER 16 TYPING 3
will place a request at source line number
24 after 16
(octal)
times through;
however, the contents of group 3
will be displayed every time.
When 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 control C REENTER
sequence, a START command, or a control C START sequence.
REMOVE

Removes the
format is:

previously

requested

pauses.

The

command

REMOVE P
For example,
REMOVE #123
will remove a pause at program source line number 123.
REMOVE ALPHA
will remove a pause at the subroutine entry to ALPHA.
REMOVE with no arguments will remove all your pause
requests,
and, in this case, no abbreviation of REMOVE is
allowed.
This prevents the unintentional removal of
pauses.
START

Starts your program at the normal FORTRAN
entry point.
The command format is:

main

program

START
STOP

Terminates the program, requests FOROTS to close all open
files,
and causes an exit to the monitor.
The usual
command format is:
STOP
STOP/RETURN
will allow a return to monitor mode without releasing
devices or closing files so that a CONTINUE can be issued.

E-15

FORDDT
STRACE

Displays a subprogram level backtrace of the current state
of the program. The command format is:
STRACE

TYPE

Causes one or more FORTRAN defined variables, arrays,
or
array elements to be displayed on your terminal.
The
command format is:
TYPE list
where list may be one or more variable 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 will use the
last argument list submitted to FORDDT.
An array
example:

element

range

can

also

specified.

For

TYPE PI(S)-PI(13)
will display the values from PI(S)
to PI(13)
inclusive.
If an unsubscripted array name is specified, the entire
array will be typed.
There are several methods of choosing the form of
in conjunction with the MODE command.
the

typeout

If you do not specify a format,
floating-point form.

You can specify a format via
described in this appendix.

You can change the format previously designated
by the MODE command by including print modifiers
in the TYPE or GROUP string. The print modifiers
are:

the

default
MODE

command

/A,/C,/D,/F,/I,/O,/R
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 /IK,L/O,M,N/A,/2
the typeout mode is integer until another mode is
specified. Therefore,
K,M,and/2
L
OCTAL
N = ASCII
wHAT

= Integer

Displays the information saved
format is:
WHAT

E-16

FORDDT.

The

command

FORDDT
E.7

ENVIRONMENT CONTROL

If a program enters an indefinite loop, you can recover by typing a
~C~C
REENTER sequence.
This action will cause FORDDT to simulate a
pause at the point of reentry and allow you to control your
run-away
program.
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
~C~C REENTER sequence will not
destroy the user profile.
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 was used, a NEXT
1 command can be issued to restore program int8grity.

E.8

FORTRAN-20/0PTIMIZE SWITCH

You should never attempt to use FORDDT 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.
Also, since the
optimizer moves statements around in your program, attempts to trace
program flow will lead to great confusion.

E.9

FORDDT 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
fatal
warning
FORDDT mnemonic
3-letter mnemonic for error message
explanation of error

%
FDT
XXX
text

Square brackets ([ ]) in this section signify variables
output on the terminal.

and

are

not

Fatal Errors
The fatal errors in the following list are each preceded by ?FDT on
the user terminal and on listings. They are listed in alphabetical
order.
BDF

[symbol] IS UNDEFINED OR IS MULTIPLY DEFINED

BOI

BAD OCTAL OUTPUT
An illegal character was detected in an octal input value.

E-l7

FORDDT
CCN

CANNOT CONTINUE
Pause has been placed on some form of skip instruction
causing FORDDT to loop;
should never be encountered in
FORTRAN-20 compiled programs.

CFO

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.

CSH

acceptable

FORTRAN-20

DIMENSION TABLE OVERFLOW
FORDDT does not have the space to
dimensions until some are removed.

FCX

CANNOT START HERE
The specified entry point is not
main program entry point.

DTO

distinguish

record

any

array

FORMAT CAPACITY EXCEEDED
An attempt was made to specify a FORMAT statement requiring
more space than was originally allocated by FORTRAN-20.

FNI

FORMAL NOT INITIALIZED
Reference to a FORMAL parameter of some subprogram that
never executed.

FNR

was

[array name] IS A FORMAL AND MAY NOT BE RE-DEFINED
FORMAL parameters may not be DIMENSIONed.

IAF

ILLEGAL ARGUMENT FORMAT
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.

COMPARE TWO CONSTANTS IS NOT ALLOWED
Conditional test involves two constants.

IER

E (number)
Internal FORDDT error - please report via an SPR.

IGN

INVALID GROUP NUMBER
Group numbers must be integral and in the range 1 through S.

INV

INVALID VALUE
A syntax error was detected in the numeric parameter.
E-lS

FORDDT
ITM

ILLEGAL TYPE MODIFIER - S
The mode S is only valid for ACCEPT statements.

LGU

[array name] LOWER SUBSCRIPT.GE.UPPER
The lower bound of any given dimension must be less than
equal to the upper bound.

LNF

[label] IS NOT A FORMAT STATEMENT

MLD

[array name] MULTI-LEVEL ARRAY DEFINITION NOT ALLOWED

The same array cannot be dimensioned more than once (via 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 RE-ENTER
The given command is not allowed until program integrity has
been restored via 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 pause at other
executable statement or subprogram entry point.
NSP

[symbol] NO SUCH PAUSE
An attempt has been made to REMOVE a pause
set up.

NUD

than

that

was

never

[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 ( .. )
Parentheses are required for
the
statements and complex constants.

E-19

specification

FORMAT

FORDDT
PRO

TOO MANY PAUSE REQUESTS
The PAUSE table has been exhausted.
10.

SER

The

maximum

limit

SUBSCRIPT ERROR
The subscript specified is outside the range of its
dimensions.

STL

define

array

larger

than

TOO MANY SUBSCRIPTS
The array is defined to have fewer dimensions
specified in the given element reference.

URC

defined

[array name] SIZE TOO LARGE
An attempt has been made to
256K.

TMS

than

are

UNRECOGNIZED COMMAND

Warning Messages
Each warning message in this list is preceded by %FTN on your terminal
and on listings. They are given here in alphabetical order.
ABX

[array name] 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.

CHI

CHARACTERS IGNORED:

"[text]"

The portion of the command string included
thought to be extraneous and was ignored.
NAR

[symbol] IS NOT AN ARRAY

NSL

NO SYMBOLS LOADED

"text"

was

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
The symbol table is different from the last time FORDDT
control.

SFA

had

SUPERSEDES FlO ARRAY
The generated dimension is being superseded
array.

for

the

given

outside

the

SPO

VARIABLE IS SINGLE-PRECISION ONLY

XPA

ATTEMPT TO EXCEED PROGRAM AREA WITH [symbol name]
An attempt has been made to access
currently defined program space.
E-20

memory

APPENDIX F
COMPILER MESSAGES

FORTRAN-20 responds with two levels of messages
fatal error and
warning.
If a warning message is received, the compilation will
continue, but a fatal error will stop the program f+om being compiled.
The format of messages is:
?FTNXXX LINE:n text
or
%FTNXXX LINE:n text
where
?

%
FTN
XXX
LINE:n
text

fatal
warning
FORTRAN mnemonic
3-letter mnemonic for the error message
line number where error occurred
explanation of error

Square brackets ([ ]) in this appendix signify variables and
output on the terminal.

are

not

Fatal Errors
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.

ATL

ARRAY [name] TOO LARGE
The total amount of memory necessary
array is greater than 5l2P.

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
F-l

COMPILER MESSAGES
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
large to fit in the buffer. Once FORTRAN-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.

CFF

CANNOT FIND FILE
The file referenced in an INCLUDE statment was not found.

CPE

CHECKSUM OR PARITY
[name]

CQL

NO CLOSING QUOTE IN LITERAL

CSF

ILLEGAL STATEMENT FUNCTION REFERENCE IN CALL STATEMENT

DDA

[symbolname] IS DUPLICATE DUMMY ARGUMENT

DFC

VARIABLE DIMENSION [name] MUST BE SCALAR, DEFINED AS
OR IN COMMON

DFD

DOUBLE [type] NAME ILLEGAL
Duplicate fields
specification.

DIA

ERROR

were

[source/listing/object]

encountered

FILE

FORMAL

INCLUDE

file

DO INDEX VARIABLE [name] 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-20.
SPECIFICATION

ASSOCIATED

values
is not

DNL

IMPLIED DO
VARIABLES

DPR

DUMMY PARAMETER [name] REFERENCED BEFORE DEFINITION

DSF

ARGUMENT [name] IS SAME AS FUNCTION NAME

DTI

THE DIMENSIONS OF [arrayname] MUST BE OF THE TYPE INTEGER

DVE

CANNOT USE DUMMY VARIABLE IN EQUIVALENCE

DWL

[source/listing/object] DEVICE [[device]] WRITE LOCKED

ECT

ATTEMPT TO ENTER [symbolname] INTO COMMON TWICE

EDN

EXPRESSION TOO DEEPLY NESTED TO COMPILE

EID

ENTRY STATEMENT ILLEGAL INSIDE A DO LOOP

ElM

ENTRY STATEMENT ILLEGAL IN MAIN PROGRAM
F-2

WITHOUT

LIST

COMPILER MESSAGES
ENF

LABEL [number] MUST REFER TO AN EXECUTABLE STATEMENT, NOT
FORMAT

ETF

ENTER FAILURE [filename]

EXB

EQUIVALENCE EXTENDS COMMON BLOCK [name] BACKWARD

FEE

FOUND [symbol] WHEN EXPECTING EITHER [symbol] OR A [symbol]
General syntax error message.

FNE

LABEL [number] MUST REFER TO A
STATEMENT

FWE

FOUND [symbol] WHEN EXPECTING [symbol]

HDE

HARDWARE DEVICE
[ [device] ]

lAC

ILLEGAL ASCII CHARACTER [character]

IAL

INCORRECT ARGUMENT TYPE FOR LIBRARY FUNCTION [name]

IBK

ILLEGAL STATEMENT IN BLOCKDATA SUBPROGRAL1

ICL

ILLEGAL CHARACTER [character] IN LABEL FIELD

IDN

DO LOOP AT LINE:

ERROR

FORMAT,

NOT

EXECUTABLE

[source/listing/object]

DEVICE

IN SOURCE

[number] IS ILLEGALLY NESTED

You are attemping to terminate a DO loop before
one or more loops defined after the given one.
IDS

IMPLICIT DO INDICES MAY NOT BE SUBSCRIPTED

IDT

ILLEGAL OR MISSPELLED DATA TYPE

IDV

IMPLIED DO INDEX IS NOT A VARIABLE

lED

INCONSISTENT EQUIVALENCE DECLARATION

terminating

The given EQUIVALENCE declaration would cause some
name to refer to more than one physical location.
IFD

INCLUDED FILES MUST RESIDE ON DISK

lID

NON-INTEGER IMPLIED DO INDEX

lIP

ILLEGAL IMPLICIT SPECIFICATION PARAMETER

lIS

INCORRECT INCLUDE SWITCH

ILF

ILLEGAL STATEMENT AFTER LOGICAL IF

symbolic

Refer to Section 9.3.2 for restrictions on logical IF object
statements.
INN

INCLUDE STATEMENTS MAY NOT BE NESTED

IOD

ILLEGAL STATEMENT USED AS OBJECT OF DO

ISD

ILLEGAL SUBSCRIPT EXPRESSION IN DATA STATEMENT
Subscript expressions may be formed only
indices and constants combined with +, F-3

with implicit
*, or /.

COMPILER MESSAGES
ISN

[symbolname] IS NOT [symbol type]
The symbol cannot be used in the attempted manner.

IUT

PROGRAM UNITS MAY NOT BE TERMINATED WITHIN INCLUDED FILES

IVP

INVALID PPN

IXM

ILLEGAL MIXED MODE ARITHMETIC
Complex and
expression.

IZM

double-precision

cannot

appear

the

same

ILLEGAL [datatype] SIZE MODIFIER [number]
Refer to Section 6.3.

LAD

LABEL [number] ALREADY DEFINED AT LINE:

[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
Labels cannot be 0 or greater than 5 digits.

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 I COMMON VARIABLE IN EQUIVALENCE GROUP

MSP

STATEMENT NAME MISSPELLED

MWL

ATTEMPT TO
ARGUMENTS

NCF

NOT ENOUGH CORE FOR FILE SPECS.

NEX

NO EXPONENT AFTER D OR E CONSTANT

NFS

NO FILENAME SPECIFIED

DEFINE

MULTIPLE

RETURN

WITHOUT

TOTAL K NEEDED= [number]

The INCLUDE statement requires a filename.
NIO

NAMELIST DIRECTED I/O WITH I/O LIST

NGS

CANNOT GET SEGMENT [name] - ERROR CODE:

[number]

Refer to the Monitor Calls User's Guide for full description
of codes.
NIR

REPEAT COUNT MUST BE AN UNSIGNED INTEGER

NIU

NON-INTEGER UNIT IN I/O STATEMENT

NLF

WRONG NUMBER OF ARGUMENTS FOR LIBRARY FUNCTION [name]

F-4

COMPILER MESSAGES
NNF

NO STATEMENT NUMBER ON FORMAT

NRC

STATEMENT NOT RECOGNIZED

NUO

.NOT.

NWD

INCORRECT USE OF

OPW

OPEN PARAMETER [name] IS OF WRONG TYPE

PD6

FORTRAN WILL NOT RUN ON A PDP-6

PIC

THE DO PARAMETERS OF [index name] MUST BE INTEGER CONSTANTS

PRF

PROTECTION FAILURE [filename]

PTL

PROGRAM TOO LARGE

IS A UNARY OPERATOR

* OR?

IN [filename]

The program takes up more than 5l2P
QEF

QUOTA EXCEEDED OR DISK FULL [filename]

QEX

BLOCK
TOO
LARGE
OR
QUOTA
[source/listing/object] FILE [name]

RDE

RIB OR DIRECTORY ERROR [filename]

RFC

[function name] IS A RECURSIVE FUNCTION CALL

RIC

COMPLEX CONSTANT CANNOT BE USED TO REPRESENT
IMAGINARY PART OF A COMPLEX CONSTANT

SAD

ARRAY [name] - SIGNED DIMENSIONS MAY APPEAR ONLY AS CONSTANT
RANGE LIMITS

SNL

[statement name] STATEMENTS MAY NOT BE LABELED

SOR

SUBSCRIPT OUT OF RANGE

TFL

TOO MANY FORMAT LABELS SPECIFIED

TOF

MORE THAN 2 OUTPUT FILES ARE NOT ALLOWED
Only a listing and a relocatable
specified as output files.

UCE

USER CORE EXCEEDED

UMP

UNMATCHED PARENTHESES

USI

[symbol type]

EXCEEDED

binary

THE

file

[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

F-5

FOR

REAL

may

COMPILER MESSAGES
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.
AGA

OPT - OBJECT VARIABLE, OF
LIST~ WAS NEVER ASSIGNED

ASSIGNED

CAl

COMPLEX EXPRESSION USED IN ARITHMETIC IF

CTR

COMPLEX TERMS USED IN A RELATIONAL OTHER THAN EQ OR NE
The result of the other relational
operands is undefined.

CUO

GOTO

WITHOUT

operators

OPTIONAL

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 expessions at compile time.

DIM

POSSIBLE DO INDEX MODIFIED INSIDE LOOP
A program that does this may be incorrectly compiled by the
optimizer, since it assumes that indices 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.
DXB

DATA STATEMENT EXCEEDS BOUNDS OF ARRAY [name]

FMR

MULTIPLE RETURNS DEFINED IN A FUNCTION

FNA

A FUNCTION WITHOUT AN ARGUMENT LIST

ICC

ILLEGAL CHARACTER, CONTINUATION FIELD OF INITIAL LINE
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 SEQ#

IDN

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

F-6

COMPILER MESSAGES

CONTINUE

CONTINUE
IFL

OPT - INFINITE LOOP.

OPTIMIZATION DISCONTINUED

LID

IDENTIFIER [name] MORE THAN SIX CHARACTERS
The remaining characters are ignored.

MVC

NUMBER OF VARIABLES DOES NOT EQUAL THE NUMBERS OF
IN DATA STATEMENT

CONSTANTS

NED

NO END STATEMENT IN PROGRAM

NOD

GLOBAL OPTIMIZATION NOT SUPPORTED WITH /DEBUG - /OPT IGNORED

NOF

NO OUTPUT FILES GIVEN

PPS

PROGRAM STATEMENT PARAMETERS IGNORED
For compatibility purposes.

RDI

ATTEMPT TO REDECLARE IMPLICIT TYPE

SOD

[name] STATEMENT OUT OF ORDER

VAl

[name] ALREADY INITIALIZED

VND

FUNCTION RETURN VALUE IS NEVER DEFINED

VNI

OPT - VARIABLE [name] IS NOT INITIALIZED
The optimizer analysis determined that the given variable
was never initialized prior to its use in a calculation.

WOP

OPT - WARNING GIVEN IN PHASE 1.
CORRECT

OPTIMIZED CODE MAY

NOT

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.

ZMT

SIZE MODIFIER [number] TREATED AS [data type]
Message is issued when one of the data type size
is used that is accepted only for compatibility.

F-7

modifiers

COMPILER MESSAGES
Internal Compiler errors
An internal compiler error is either an attempt by 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-20 compiler.
Monitor-detected internal errors are of the form
[message] AT LOCATION [address]

IN PHASE [segment]

WHILE PROCESSING STATEMENT [line-number]
where [message] can be one of
ILLEGAL MEMORY REFERENCE
STACK EXHAUSTED
MEMORY PROTECTION VIOLATION
Compiler-detected errors are of the form
? INTERNAL COMPILER ERROR PROCESSING STATEMENT NUMBER [line-number]

? CALL TO [routine-name] FROM [address]
Submit an SPR if you received an internal compiler error.

F-8

APPENDIX G
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

arithmetic
fault
calculations

errors

library errors
routines

data errors (DAT) - errors in data conversion on I/O

device errors

(LIB) - errors

during

(APR) - errors
generated

file

OPEN

numeric

FORLIB

library

(DEV) - I/O hardware errors

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
Table 15-3 for details.
The I/O errors (SYS, OPN, DAT, and DEV)
either cause messages to be printed on the terminal or can be trapped
by an error exit argument (ERR=statement label) on OPEN, READ, WRITE,
and CLOSE.
Table G-l gives the text of the messages which can be printed for SYS,
OPN,
DAT,
and DEV errors.
The included footnotes give additional
information. Table G-2 gives the text of the messages which can be
printed for APR and LIB errors.
The FORTRAN Library Subroutine ERRSNS 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 optional.)
CALL ERRSNS (I,J)
calls this subroutine.

J is the second, optional integer value.

G-1

FOROTS ERROR MESSAGES
Table G-l
FOROTS I/O Error Messages and ERRSNS Returned Values
First
Value

Second
Value

o
101

243
246
23
312
24
308
25
302

26
311
28
252
254
262
268
29
250
30
237
238
240
242
245
248
249
251
253

Explanation

No error detected
Satisfactory completion (no error detected)
Normal end of job (1)
Invalid error call
Unidentified entry in FORERR (3)
Unidentified entry in FORERR (3)
Backspace error
BACKSPACE illegal for device (9)
End-of-file during READ
Attempt to READ beyond valid input (8)
Invalid record number
LSCW illegal in binary record or reading
ASCII;
or attempt to read unwritten ASCII
RANDOM ACCESS record or
unwritten
or
destroyed record number
Direct access not specified
Cannot RANDOM ACCESS a SEQUENIAL file
CLOSE error
DTA directory is full
(2) or protection
error
Rename file already exists (2)
No room or quota exceeded (2)
Cannot delete or
rename
a
non-empty
directory (2)
No such file
File was not found
OPEN failure
DUMP mode RANDOM or APPEND access not
implemented; try IMAGE MODE
DIALOG file cannot be opened (3)
Record length missing for RANDOM ACCESS
Too many devices open:
fifteen maximum
Device not available (2)
Illegal ACCESS for device (2)
Illegal MODE or MODE switch (2)
No directory for project, programmer
number (2)
File was being modified (2)

1. Not currently implemented.
2. OPEN errors 251 through 276 map directly onto error
returned by the OPEN UUO; see the Monitor Calls Manual.

numbers

3. Error cannot currently occur.
8. Occurs when simulating mag tape output; SKIP RECORD and SKIP FILE
are illegal. Also occurs when a non-existent file is opened in MODE=
SEQINOUT and the first operation on that file is a READ.
9. Occurs if OPEN output with BACKSPACE is not a mag tape or disk.

G-2

FOROTS ERROR MESSAGES
Table G-l (Cont.)
FOROTS I/O Error Messages and ERRSNS Returned Values
First
Value

Second
Value
255
256
259
265
266
267
269
270
271
272
274
277

31
315
32
239
39
310
42
244
260
45
241
47
263
59
313
62
301
306
314
63
305

Explanation

Illegal sequence of Monitor Calls (11)
Bad UFD or bad RIB (2)
Device not available (2)
Partial allocation only (2)
Block not free on allocation (2)
Cannot supersede an existing directory (2)
SFD not found (2)
Search list empty (2)
SFD nested too deeply (2)
No CREATE flag for specified UFD (2)
File cannot be updated (2)
LOOKUP ENTER or RENAME error (2)
Mixed access modes
Cannot do SEQUENTIAL ACCESS on a RANDOM
file
Invalid logical unit number
Illegal FORTRAN unit number (2)
Error during READ
REREAD before first READ is illegal (1)
Device handler not resident
No such device (2)
No such device (2)
OPEN statement keyword error
Switch error
during
DIALOG
or
OPEN
statement scan (2)
write on read-only file
Write-lock error (2)
List-directed I/O syntax error
Illegal delimiter in LIST DIRECTED input
Syntax error in FORMAT
Illegal character in FORMAT statement (4)
I/O list without data conversion in FORMAT
Missing width field for A or R on input
Output conversion error
Optional * fill:
unidentified entry in
FORERR (7)

1. Not currently implemented.
2. OPEN errors 251 through 276 map directly onto error
returned by the OPEN UUO;
see the Monitor Calls Manual.

numbers

4. In runtime FORMAT.
7.

* fill controlled by compile-time variable ASTFIL.

11. Can occur on OPEN (MODE= 'APPEND') when file is found in LIB:
or
on
[1,4]
when device specified was SYS:
and /NEW was in your search
list.

G-3

FOROTS ERROR MESSAGES
Table G-l (Cont.)
FOROTS I/O Error Messages and ERRSNS Returned Values
First
Value

Second
Value

64
303
307
67
304
81
102
261
699
247
257
258
264
273
275
276
799
309
899
400
401
402
403
404
407
999
100
103
104
105
106

Explanation

Input conversion error
Checksum error reading binary records (5)
Illegal character in data
Record too small for I/O list
I/O list greater than record size (6)
Invalid argument
Argument block not in correct format
Argument block not in correct format (2)
Unclassifiable error on OPEN
FOROTS system error (2,3)
FOROTS system error (2)
FOROTS system error (2)
Not enough monitor table space (2)
FOROTS system error (2)
FOROTS system error (2)
FOROTS system error (2)
Unclassifiable data error
Variable cannot be found in NAMELIST block
Unclassifiable device errors
write protected
Device error
Parity error
Block too large, quota exceeded,
or
file
structure full.
Nonexistent CDR reader.
Spooled CDR file does not exist.
End-of-file (10)
End-of-tape
Unclassified system error
FOROTS system error
Monitor not build to support FOROTS
Fatal error
User program has requested more code than
is available
Run time memory management error

2. OPEN errors 251 through 276 map directly onto error
returned by the OPEN UUO;
see the Monitor Calls Manual.
3. Error cannot currently occur.
5. Checksumming controlled by compile-time variable CHKSUM.
6. Occurs when a type 2 LSCW is found in a FORSE binary record.
10. Trappable if there is no END= clause.

G-4

numbers

FOROTS ERROR MESSAGES
Table G-2
FOROTS Arithmetic and Library Error Messages
APR

LIB

Integer Overflow

Attempt to take DLOG of Negative Arg.

Integer Divide Check

Attempt to take DSQRT of Negative Arg.

Illegal APR Trap

ACOS of Arg.

> 1.0 in Magnitude

Floating Divide Check

ASIN of Arg.

> 1.0 in Magnitude

Floating Underflow

Attempt to take SQRT of Negative Arg.
Attempt to take LOG of Negative Arg.

G-5

APPENDIX H
DECSYSTEM-lO COMPATIBILITY

The following items are included in the DECsystem-20 FORTRAN software
for compatibility with the DECsystem-lO. They are not supported on
the DECsystem-20. Users must not specify these items because their
actions are undefined and the results cannot be guaranteed.
1.

Logical Device Assignments.
(Refer to pages 10-4 and E-27.)
Device

Logical unit number

Use

PTR
PTP
DIS
DTAI
DTA2
DTA3
DTA4
DTA5
DTA6
DTA7

06
07
08
09

Paper Tape Reader
Paper Tape Punch
Display
DEC tape

10
11
12
13
14
15

DEC tape

PUNCH Statement

KAlO and KIlO compiler switches

The following Library Subroutines:
SLITE(i)
SLITET(i,j)
SSWTCH(i,j)

DDT command to FORDDT.

H-l

INDEX

A format descriptor, 13-12
ABS, 15-4
ACCEPT statement, 10-18
ACCEPT transfer,
formatted, 10-18
into FORMAT statement, 10-19
ACCESS in file control
statement, 12-3
Accumulator usage, C-10
Accuracy of double-precision
numbers, C-1
ACOS function, 15-10
Addition, 4-1
Adjustable dimensions, 6-2
AlMAG, 15-4
AINT, 15-4
ALL with DEBUG, B-3
Allocation,
register, C-7
ALOG function, 15-9
ALOG10 function, 15-9
Alphanumeric data transfer,
13-12
Alphanumeric FORMAT field
descriptor, 13-11
AMAX,0, 15-5
AMAX1, 15-5
AMIN,0, 15-5
AMIN1, 15-5
AMOD, 15-5
.AND., 4-5
ANSI standard, 1-1
APPEND with ACCESS, 12-4
Argument,
subprogram, 15-1
Argument type,
COBOL/FORTRAN, C-12
Arithmetic,
mixed-mode, 4-2
Arithmetic assignment statement,
8-1
Arithmetic expression, 4-1
Arithmetic IF statement, 9-3
Arithmetic operator, 4-1
Array, 3-7
dimensioning, 3-9, C-4
Array elements,
storage of, 3-10
Array subscript, 3-8
ASCII with MODE, 12-4
ASCIZ string, C-14
ASIN function, 15-10
ASSIGN statement, 8-4
Assigned GOTO statement, 9-2

Assignment statement,
arithmetic, 8-1
label, 8-4
logical, 8-4
mixed-mode, 8-1
ASSOCIATE VARIABLE in file
control statement, 12-8
ATAN function, 15-10
ATAN2 function, 15-10
AXIS subroutine, 15-17

BACKFILE statement, 14-3
BACKSPACE statement, 14-2
BASIC,
input from, 2-6
Basic external function subprogram, 15-7
BINARY with MODE, 12-4
Blank line, 2-6
BLOCK DATA statement, 16-1
BLOCK SIZE in file control
statement, 12-8
Block data subprogram, 16-1
BOUNDS with DEBUG, B-3
BUFFER COUNT in file control
statement, 12-8

CABS, 15-4
Call,
FUNCTION, 15-14
subroutine, 15-11
CALL statement, 15-11
Carriage control character, 13-16
Category,
statement, 1-1
CCOS function, 15-9
CEXP function, 15-9
Character code, A-1
Character set, 2-1
Character set with MODE, 12-4
Characters,
line formatting, 2-2
line termination, 2-2
CLOG function, 15-9
CLOSE statement, 12-1
CLOSE statement summary, 12-10
CMPLX, 15-4
COBOL,
interaction with, C-18

Index-1

INDEX (CONT . )

COBOL/FORTRAN argument type, C-12
Command,
COMPILE, B-4
DEBUG, B-4
EXECUTE, B-4
LOAD, B-4
Comment line, 2-5
Common block name, 6-5
COMMON statement, 6-5
Compatibility with FORTRAN-10,
H-l
COMPIL in FOROTS, D-2
Compilation control statement, 5-1
COMPILE command, B-4
Compiler commands, B-4
Compiler generated variable, B-6
Compiler switches, B-1
Compiler version, B-8
Complex constant, 3-3
Complex format, 13-4
COMPLEX statement, 6-3
Computation,
redundant, C-5
reordering, C-3
Computation in DO-loop,
constant, C-6
Computed GO TO statement, 9-2
CONJG function, 15-10
Constant, 3-1
complex, 3-3
double-precision, 3-3
integer, 3-2
label, 3-6
literal, 3-5
logical, 3-5
octal, 3-4
real, 3-2
Constant computation in DO-loop,
C-6
Constant folding, C-7
Constant propagation, C-7
Continuation field,
line, 2-3
Continuation line, 2-4
Continue (G) option after PAUSE,
9-11
CONTINUE statement, 9-10
Control statement, 9-1
compilation, 5-1
Control-Z, 2-1
COS function, 15-9
COSD function, 15-9
COSH function, 15-10
CROSSREF switch, B-2
CSIN function, 15-9
CSQRT function, 15-9

D (double-precision notation) ,
3-3
D format descriptor, 13-4
DABS, 15-4
.DAT extension, 12-5
Data files,
FOROTS, D-4
DATA statement, 7-1
Data transfer operations, 10-1
Data type, 3-1
DATAN function, 15-10
DATAN2 function, 15-10
DATE subroutine, 15-17
DBLE, 15-4
DCOS function, 15-9
DEBUG command, B-4
Debug line, 2-6
DEBUG switch, B-2, B-3
Debugger,
FORDDT, E-l
Debugger code size, B-4
DECODE statement, 10-21
DEFAULT, B-15
DEFINE FILE subroutine, 15-17
DELETE with DISPOSE, 12-5DENSITY in file control statement, 12-9
Descriptor,
G format, 13-7
Device control statement, 14-1
Device control statement
summary, 14-3
DEVICE in file control statement, 12-2
Device number,
logical, 10-3
DEXP function, 15-9
DFLOAT, 15-4
DIALOG in file control statement, 12-9
DIM, 15-5
DIMENSION statement, 6-1
Dimensioning,
array in COMMON, 6-7
Dimensioning array, 3-9, C-4
Dimensions,
adjustable, 6-2
DIMENSIONS with DEBUG, B-3
Directory,
sub-file, 12-6
user file, 12-6
DIRECTORY,
in file control statement,
12-6
DISPOSE in file control statement, 12-5

Index-2

INDEX (CONT . )

Division, 4-1
DLOG function, 15-9
DLOGIO function, 15-9
DMAXl, 15-5
DMINl, 15-5
DMOD, 15-5
DO statement, 9-5
DO-loop,
constant computation in, C-6
execution, 9-6
extended range, 9-S
floating-point, C-2
implied in I/O list, 10-5
nested, 9-6
parameters, 9-6
permitted transfers, 9-9
range, 9-5
DO-loop iteration, C-2
DO-loop replacement, C-S
DOUBLE PRECISION statement, 6-3
Double-precision constant, 3-3
Double-precision format, 13-4
Double-precision numbers,
accuracy of, C-l
range of, C-l
DSIGN, 15-5
DSIN function, 15-9
DSQRT function, 15-9
Dummy argument,
subprogram, 15-1
DUMP subroutine, IS-IS
DUMP with MODE, 12-4

E (exponential notation), 3-2
E format descriptor, 13-4
EDIT program, 2-6
ENCODE statement, 10-21
END argument in I/O statement,
10-10
END FILE statement, 14-2
END statement, 5-2, 15-6
ENTER in FOROTS, C-lS
ENTRY statement, 15-15
.EQ., 4-7
EQUIVALENCE statement, 6-7
.EQV., 4-5
ERR argument in I/O statement,
10-10
ERR in file control statement,
12-10
Error,
fatal, B-17
Error processing,
FOROTS, 0-3
Error reporting, B-17
ERRSET subroutine, 15-19

ERRSNS subroutine, 15-19
Evaluation of expression, 4-9
EVEN with PARITY, 12-9
Executable statement, 1-1
EXECUTE command, B-4
Execution on non-DEC machines,
C-l
Exit (X) option after PAUSE,
9-11
EXIT subroutine, 15-19
EXP function, 15-9
EXPAND switch, B-2
Exponential notation, 3-2
Exponentiation, 4-1
permitted, 4-4
Expression,
arithmetic, 4-1
evaluation of, 4-9
logical, 4-4
mixed-mode, 4-10, 4-11
nested, 4-9
relational, 4-7
External function subprogram,
15-6
basic, 15-7
EXTERNAL statement, 6-S

F format descriptor, 13-4
. FALSE., 3-5
Fatal error, B-17
Field,
label, 2-3
line continuation, 2-3
remarks, 2-4
statement, 2-3
Field descriptor,
alphanumeric FORMAT, 13-11
FORMAT, 13-2
logical FORMAT, 13-10
numeric FORMAT, 13-4
File,
directory subfile, 12-6
FOROTS data, 0-4
non-FORTRAN, C-9
File control statement, 12-1
File directory,
user, 12-6
FILE in file control statement,
12-5
FILE SIZE in file control statement, 12-S
FIND statement, 10-21
FLOAT, 15-4
Floating-point DO-loop, C-2
Folding,
constant, C-7

Index-3

INDEX (CaNT.)

FORDDT debugger, E-l
FORDDT messages, E-17
FORMAT field descriptor,
alphanumeric, 13-11
logical, 13-10
numeric, 13-4
record formatting, 13-15
FORMAT statement, 13-1
ACCEPT transfer into, 10-19
transfer into, 10-3
FORMAT statement descriptor,
13-2
Formatted ACCEPT transfer, 10-18
Formatted READ transfer,
random access, 10-13
sequential, 10-11
Formatted WRITE transfer,
random access, 10-17
sequential, 10-16
FOROTS,
using, D-13
FOROTS data files, D-4
FOROTS error processing, D-3
FOROTS features, D-2
FOROTS hardware requirements,
D-l
FOROTS input/output facility,
D-3
FOROTS messages, G-l
FOROTS software requirements,
D-l
FOROTS/LINK interface, D-28
FORTRAN compiler, B-1
FORTRAN messages, F-l
FUNCTION call, 15-14
Function references,
order, C-8
FUNCTION statement, 15-6
Function subprogram,
basic external, 15-7
external, 15-6
intrinsic, 15-3
statement, 15-3
Function subprogram structure,
15-7
FUNCTION type, 15-6

G (option after PAUSE), 9-11
G format descriptor, 13-4, 13-7
.GE., 4-7
General (G) numeric format,
13-7
GETOVL in LINK, C-20
Global optimization, C-4

GOTO statement, 9-1
assigned, 9-2
computed, 9-2
unconditional, 9-1
• GT., 4-7

H (literal notation), 3-5
H format descriptor, 13-12
Hardware requirements,
FOROTS, D-l
Hierarchy of operators, 4-9
Hollerith literal, 3-5

I format descriptor, 13-4
I/O list, 10-5
lABS, 15-4
IDIM, 15-5
IDINT, 15-4
IF statement, 9-3
arithmetic, 9-3
logical, 9-4
IF IX , 15-4
ILL subroutine, 15-19
IMAGE with MODE, 12-4
IMPLICIT statement, 6-5
In I/O list DO-loop,
implied, 10-6
Inaccessible code, C-7
INCLUDE statement, 5-1
INCLUDE switch, B-2
INDEX with DEBUG, B-3
INIOVL in LINK, C-20
Initial line, 2-4
Input,
line-sequence, 2-6
Input from BASIC, 2-6
Input from EDIT, 2-6
Input/output facility,
FOROTS, D-3
Input/output list, 10-2
NAMELIST, 10-10
Input/output optimization, C-8
Input/output,
format, 10-2
list-directed, 10-8
statement, 10-1
summary, 10-24
INT, 15-4
Integer constant, 3-2
Integer format, 13-4

Index-4

INDEX (CONT.)

Logical FORMAT field descriptor,
13-10
Logical IF statement, 9-4
Logical operator, 4-5
LOGICAL statement, 6-3
Logical unit number, 10-3
LOGOVL in LINK, C-20
. LT., 4-7

INTEGER statement, 6-3
Intrinsic function subprogram,
15-3
ISIGN, 15-5
Iteration,
DO-loop, C-2

Keyword, 1-1

L format descriptor, 13-10
Label assignment statement, 8-4
Label constant, 3-6
Label field, 2-3
LABELS with DEBUG, B-3
• LE., 4-7
LEGAL subroutine, 15-19
Line,
blank, 2-6
comment, 2-5
continuation, 2-4
debug, 2 -6
definition, 2-2
fields, 2-2
initial, 2-4
multi-statement, 2-5
Line continuation field, 2-3
Line formatting characters, 2-2
Line sequence number, B-5
LINE subroutine, 15-19
Line termination characters, 2-2
Line types, 2-4
Line-sequence input, 2-6
LINK overlay facility, C-20
LINK/FOROTS interface, D-28
List,
input/output, 10-2
NAMELIST input/output, 10-10
LIST with DISPOSE, 12-5
List-directed input/output statement, 10-8
List-directed transfer,
sequential READ, 10-12
sequential WRITE, 10-17
Listing,
program, B-5
Literal constant, 3-5
Literal format conversion, 13-13
LNMAP switch, B-2
LOAD command, B-4
Location in object program, B-5
Logical assignment statement,
8-4
Logical constant, 3-5
Logical expression, 4-4
°

MACRO in listing, B-5
MACRO libraries, C-14
MACROCODE switch, B-2
MAX 0 , 15-5
MAXI, 15-5
Messages,
FORDDT, E-17
FOROTS, H-l
FORTRAN, F-l
realtime, G-7
MINO, 15-5
MINI, 15-5
Mixed-mode arithmetic, 4-2
Mixed-mode assignment statement,
8-1
Mixed-mode expression, 4-10,
4-11
MKTBL subroutine, 15-20
MOD, 15-5
MODE in file control statement,
12-4
Multi-statement line, 2-5
Multiple record transfer, 13-14
Multiplication, 4-1

Name,
symbolic, 3-6
NAMELIST input/output list,
10-10
NAMELIST statement, 11-1
NAMELIST-controlled transfer,
input, 11-2
output, 11-3
sequential READ, 10-13
sequential WRITE, 10-17
. NE., 4-7
Nested DO-loop, 9-6
Nested expression, 4-9
NOERRS switch, B-2
NONE with DEBUG, B-3
Nonexecutable statement, 1-1
Non-FORTRAN files, C-9
Non-FORTRAN programs, C-9
NONSHAR, B-18

Index-5

INDEX (CONT . )

.NOT., 4-5
NOWARNINGS switch, B-2
NUMBER subroutine, 15-20
Numeric,
field width variable, 13-10
Numeric format,
general (G), 13-7
Numeric FORMAT field descriptor,
13-4

o format descriptor, 13-4
Object program,
location in, B-5
Octal constant, 3-4
ODD with PARITY, 12-9
OPEN statement, 12-1
OPEN statement summary, 12-10
Operator,
arithmetic, 4-1
hierarchy, 4-9
logical, 4-5
Operator strength, C-5
Optimization,
global, C-4
program, C-9
OPTIMIZE switch, B-2
.OR., 4-5
Order of statements, 2-7
OTS, B-18
Overflow, C-3
Overlay facility,
LINK, C-20

PARAMETER statement, 6-9
PARITY in file control statement, 12-9
PATH with DIRECTORY, 12-7
PAUSE statement, 9-11
PDUMP subroutine, 15-20
PLOT subroutine, 15-20
PLOTS subroutine, 15-20
Precision for real constant, 3-2
PRINT statement, 10-19
PRINT with DISPOSE, 12-5
Program listing, B-5
PROGRAM statement, 5-1
Programs,
non-FORTRAN, C-9
optimizing, C-9
writing, C-1
Propagation,
constant, C-7
PROTECTION in file control statement, 12-6

R format descriptor, 13-12
RAN function, 15-10
RANDIN with ACCESS, 12-3
Random access data transfer,
10-1
Random access record specification, 10-7
Random access transfer,
formatted READ, 10-13
formatted WRITE, 10-17
unformatted READ, 10-13
unformatted WRITE, 10-17
RANDOM with ACCESS, 12-3
Range of double-precision
numbers, C-1
READ statement, 10-11
READ statement summary, 10-14
READ transfer,
random access, 10-13
sequential, 10-11, 10-12, 10-13.
REAL function, 15-4
Real constant, 3-2
Real format, 13-4
REAL statement, 6-3
Record formatting (T and X),
13-15
RECORD SIZE in file control
statement, 12-8
Record specification,
random access, 10-7
Reentrant program, B-18
Register allocation, C-7
Relational expression, 4-7
RELEAS subroutine, 15-21
Remarks field, 2-4
REMOVL in LINK, C-20
RENAME with DISPOSE, 12-5
Repeat for format descriptor,
13-3
Replacement,
DO-loop, C-8
REREAD statement, 10-14
RESET in FOROTS, C-10
RETURN statement, 15-7, 15-12
REWIND statement, 14-1
RUNOVL in LINK, C-20

SAVE with DISPOSE, 12-5
SAVRAN subroutine, 15-21
Scale factor in FORMAT statement,
13-7
SCALE subrouti·ne, 15-21
SEG, B-18
SEQ IN with ACCESS, 12-3
SEQINOUT with ACCESS, 12-3
SEQOUT with ACCESS, 12-3

Index-6

INDEX (CONT.)

'-"

Sequence number,
line, B-5
Sequential data transfer, 10-1
Sequential transfer,
READ, 10-11, 10-12, 10-13
WRITE, 10-16, 10-17
SET RECORD statement, 14~3
SETABL subroutine, 15-21
SETRAN subroutine, 15-21
SFD, 12-6
Sharable program, B-18
.SHR extension, B-18
SIGN function, 15-5
SIN function, 15-9
SIND function, 15-9
SINH function, 15-10
SKIP FILE statement, 14-3
SNGL, 15-4
Software requirements,
FOROTS, D-1
SORT subroutine, 15-21
Specification statement, 6-1
SQRT function, 15-9
SSAVE switch, B-18
Statement,
ACCEPT, 10-18
Arithmetic assignment, 8-1
arithmetic IF, 9-3
ASSIGN, 8-4
assigned GOTO, 9-2
BACKFILE, 14-3
BACKSPACE, 14-2
BLOCK DATA, 16-1
CALL, 15-11
CLOSE, 12-1
COMMON, 6-5
COMPLEX, 6-3
computed GOTO, 9-2
CONTINUE, 9-10
control, 9-1
DATA, 7-1
DECODE, 10-21
device control, 14-1
DIMENSION, 6-1
DO, 9-5
DOUBLE PRECISION, 6-3
ENCODE, 10-21
END, 5-2, 15-6
END FILE, 14-2
ENTRY, 15-15
EQUIVALENCE, 6-7
executable, 1-1
EXTERNAL, 6-8
file control, 12-1
FIND, 10-21
FORMAT, 13-1
FUNCTION, 15-6
GOTO, 9-1

Statement (Cont.),
IF, 9-3
IMPLICIT, 6-5
INCLUDE, 5-1
input/output, 10-1
INTEGER, 6-3
label assignment, 8-4
list-directed,
input/output, 10-8
LOGICAL, 6-3
logical assignment, 8-4
logical IF, 9-4
mixed-mode assignment, 8-1
NAMELIST, 11-1
nonexecutable, 1-1
OPEN, 12-1
PARAMETER, 6-9
PAUSE, 9-11
PRINT, 10-19
PROGRAM, 5-1
READ, 10-11
REAL, 6-3
REREAD, 10-14
RETURN, 15-6, 15-12
REWIND, 14-1
SET RECORD, 14-3
SKIP FILE, 14-3
STOP, 9-10
SUBROUTINE, 15-8
TYPE, 10-20
type specification, 6-3
unconditional GOTO, 9-1
UNLOAD, 14-2
WRITE, 10-16
Statement category, 1-1
Statement field, 2-3
Statement function subprogram, 15-3
Statement label constant, 3-6
Statement numbers, 2-3
Statement summary,
CLOSE, 12-10
device control, 14-3
input/output, 10-24
OPEN, 12-10
READ, 10-14
WRITE, 10-18
Statements,
order of, 2-7
STOP statement, 9-10
Storage of array elements, 3-10
Sub-file directory, 12-6
Subprogram,
basic external function, 15-7
block data, 16-1
external function, 15-6
intrinsic function, 15-3
multiple entries to, 15-15
multiple returns from, 15-12

Index-7

INDEX (CONT.)

Subprogram, (Cont.)
statement function, 15-3
subroutine, 15-8
Subprogram argument, 15-1
Subprogram dummy argument, 15-1
Subprograms, 15-1
Subroutine,
DATE, 15-17
ERRSET, 15-19
ERRSNS, 15-19
EXIT, 15-19
FORTRAN supplied, 15-12
ILL, 15-19
LEGAL, 15-19
LINE, 15-19
programming consideration, C-2
Subroutine call, 15-11
SUBROUTINE statement, 15-8
Subroutine structure, 15-11
Subroutine subprogram, 15-8
Subscript,
array, 3-8
Subtraction, 4-1
Switches,
compiler, B-1
SYMBOL subroutine, 15-22
Symbolic name, 3-6
SYNTAX switch, B-2

T (trace after PAUSE), 9-12
T format descriptor, 13-15
TANH function, 15-10
TIME subroutine, 15-22
Trace (T) option after PAUSE,
9-12
TRACE function, 9-13
TRACE subroutine, 9-13
TRACE with DEBUG, B-3
Transfer operations, 10-1
.TRUE., 3-5
Type,
FUNCTION, 15-6
Type specification statement,
6-3
TYPE statement, 10-20

UFD, 12-6
Unconditional GOTO statement,
9-1
Uninitia1ized variable, C-8
UNIT in file control statement,
12-2
unit number,
logical, 10-3
Unformatted transfer,
random access,
READ, 10-13
WRITE, 10-17
sequential binary,
READ, 10-12
WRITE, 10-16
UNLOAD statement, 14-2
User file directory, 12-6

Variable, 3-7
compiler generated, B-6
uninitia1ized, C-8
VERSION in file control statement, 12-8

Warning message, B-17
WHERE subroutine, 15-22
WRITE statement, 10-16
WRITE statement summary, 10-18
WRITE transfer,
random access, 10-16, 10-17
sequential, 10-16, 10-17
Writing programs, C-1

X (option after PAUSE), 9-11
X format descriptor, 13-15
. XOR., 4-5

tz, 2-1

Index-8

DECsystem-20
FORTRAN Reference Manual
AA-4l58B-TM
READER'S COMMENTS

NOTE:

This form is for document comments only. DIGITAL will
use comments submitted on this form at the company's
discretion. Problems with software should be reported
on a Software Performance Report (SPR) form.
If you
require a written reply and are eligible to receive
one under SPR service, submit your comments on an SPR
form.

Did you find errors in this manual?

c
I:::

If so, specify by page.

Did you find this manual understandable, usable, and well-organized?
Please make suggestions for improvement.

I.~

1-£

Ig>
I~

1°

la
1(1)

I~
Q)
10:

Is there sufficient documentation on associated system programs
required for use of the software described in this manual? If not,
what material is missing and where should it be placed?

Please indicate the type of user/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 Non-programmer interested in computer concepts and capabilities

Name

Date __________________________

Organization ________~-----------------------------------------------------Street ____________, __________________________________________________________
City ____________________________ State ______________ Zip Code _______________
or
Country

-------------------------------------------------------------Fold flcrc------------------------------------------------------------

------------------------------------------------ Do Not Tear - Fold Here and Staple -----------------------------------------------

FIRST CLASS
PERMIT NO. 33
MAYNARD, MASS.
BUSINESS REPLY MAIL
NO POSTAGE STAMP NECESSARY IF MAILED IN THE UNITED STATES

Postage will be paid by:

_.j

•
Software Documentation
146 Main Street ML5-5/E39
Maynard, Massachusetts 01754