Digital PDFs

DEC-11-LMFUA-A-D

May 2000

96 pages

Original

4.2MB

Document:	RSX-11M FORTRAN-IV User's Guide
Order Number:	DEC-11-LMFUA-A-D
Revision:	0
Pages:	96
Original Filename:	http://bitsavers.org/pdf/dec/pdp11/rsx11m_s/RSX11M_V1/RSX-11M-V1-Fortran-IV_Users_Guide.pdf

OCR Text

RSX-11M
FORTRAN-IV User's Guide
Order No. DEC-II-LMFUA-A-D

RSX-IIM Version 1

Order additional copies as directed on the Software
Information page at the back of this document.

digital equipment. corporation · maynard. massachusetts

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 manual.
The software described in thi. s document is furnished to the purchaser
under a license for use on a single computer system and can be copied
(with inclusion of DIGITAL's copyright notice) only for use in such
system, except as may otherwise be provided in writing by DIGITAL.
Digital Equipment Corporation assumes no responsibility for the use
or reliability of its software on equipment that is not supplied by
DIGITAL.
ASSOCIATED DOCUMENTS
Refer to User's Guide to RSX-llM Manuals, DEC-ll-OMUGA-A-D.

The HOW TO OBTAIN SOFTWARE INFORMATION page, located at the back of
this document, explains the various services available to DIGITAL
software users.
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:
COP
COMPUTER LAB
COMSYST
COMTEX
DOT
DEC
DECCOMM
DECTAPE
DIBOL

DIGITAL
DNC
EDGRIN
EDUSYSTEM
FLIP CHIP
FOCAL
GLC-8
IDAC
IDACS

INDAC
KA10
LAB-8
LAB-8/e
LAB-K
OMNIBUS
OS/8
PDP
PHA

PS/8
QUICKPOINT
RAD-8
RSTS
RSX
RTM
RT-ll
SABR
TYPESET 8
UNI~US

CONTENTS
CHAPTER

CHAPTER

OPERATING PROCEDURES

1.1
1.1.1
1.1.2

USING THE FORTRAN-IV SYSTEM
Command Strinq
File Specifications

1-1
1-1
1-2

1.2
1.2.1
1.2.2
1.2.2.1
1.2.2.2
1.2.2.3
1.2.3

USING THE FORTRAN-IV COMPILER
Compiler Switches /
List Formats
Source Listing
Storaqe Map Listinq
Generated Code Listinq
Compiler Memory Requirements

1-4
1-6
1-7
1-7
1-7
1-8
1-10

1.3
1.3.1
1.3.2
1.3.2.1
1.3.2.2
1.3.2.3
1.3.2.4
1.3.2.5
1.3.2.6
1.3.2.7
1.3.2.8
1.3.2.9
1.3.2.10
1.3.3
1.3.3.1
1.3.3.2
1.3.3.3
1.3 ct 3.4
1.3.4

USING THE TASK BUILDER TO LINK FORTRAN PROGRAMS
Task Builder Switch Options for FORTRAN Proqrams
Task Builder Options for FORTRAN Proqrams
ACTFIL
COMMON
LIBR
MAXBUF
TASK
UIC
UNITS
FMTBUF
ASG
FORTRAN Library Usage
Re1ocatab1e Libraries
Shared Libraries
System Libraries
User Libraries
Overlays

1-10
1-11
1'!'"11
1-12
1-12
1-12
1-13
1-13
1-13
1-13
1-13
1-14
1-14
1-14
1-14
1-15
1-15
1-15
1-15

1.4

USING MCR TO INITIATE TASK EXECUTION

1-17

1.5

EXAMPLES

1-18

1.6

DEBUGGING A FORTRAN PROGRAM

1-19

FORTRAN-IV OPERATING ENVIRONMENT

2.1

FORTRAN-IV OBJECT TIME SYSTEM

2-1

OBJECT CODE

2-1

2.3

SUBROUTINE LINKAGE

2-3

2.4

SUBPROGRAM REGISTER USAGE

2-4

2.5

VECTORED ARRAYS

2-5

RSX-11M FORTRAN-IV SPECIFIC CHARACTERISTICS

3.1

VARIABLE NAMES

3-1

3.2

INITIALIZATION OF COMMON

3-1

PAR

iii

APPENDIX

3.3

CONTINUATION LINES

3-1

3.4

DEFAULT LOGICAL UNIT-DEIVCE!FILE NAMES

3-1

3.5

STATEMENT ORDERING RESTRICTIONS

3-2

3.6

OTS/FCS FILE OPEN CONVENTIONS

3-3

FORTRAN DATA REPRESENTATION

A-l

A.l

INTEGER FORMAT

A-l

A.2
A.2.l
A.2.2

A-l
A-2
A-2

A. 2.3

FLOATING-POINT FORMATS
Real Format
Double-Precision Format
Complex Format

A.3

LOGICAL*l

A-3

A.4

HOLLERITH FORMAT

A-3

A.5

LOGICAL FORMAT

A-4

A.6

RADIX-50 FORMAT

A-4

SYSTEM SUBROUTINES

B-1

B.l

SYSTEM SUBROUTINE SUMMARY

B-1

B.2

ASSIGN

B-2

B-3

DATE

B-3

B.5

I DATE

B-4

B.6

ERRSET

B-4

B.?

ERRSNS

B-5

B.8

ERRTST

B-6

B.9

EXIT

B-6

B.10

USEREX

B-6

B.ll

FDBSET

B-?

B.12

RAD50

B-8

B.13

lRAD50

B-8

B.14

R50ASC

B-9

B.15

RANDU, RAN

B-10

B.16

SECNDS

B-10

B.l?

TIME

B-ll

A-3

(

APPENDIX

FORTRAN ERROR DIAGNOSTICS

C-l

C.l
C.l.l

C-l

C.l.3
C.l.4

COMPILER ERROR DIAGNOSTICS
Errors Reported by the Initial Phase of the
Compiler
Errors Reported by Secondary Phases of the
Compiler
Warning Diagnostics
Fatal Compiler Error Diagnostics

C.2
C.2.l
C.2.2
C.2.2.l
C.2.3
C.2.3.l
C.2.3.2
C.2.4

OBJECT TIME SYSTEM ERROR DIAGNOSTICS
Error Processing Algorithm
Object Time System Error Message Format
Short Message File
Object Time System Error Codes
Initial Control Bit Settings
Error Messages
Notes on OTS Error Processing Implementation

COMPATIBILITY WITH OTHER PDP-II FORTRANS

D-l

0.1
0.1.1
0.1.2
0.1.3
0.1.4

COMPATIBILITY WITH PDP-II FORTRAN V08.04 UNDER
RSX-IID V4A
Language Differences
Object Time System Differences
Implementation Differences
Operational Differences

0-1
0-1
0-2
D-2
D-3

0.2

COMPATIBILITY WITH PDP-II FORTRAN IV-PLUS

0-4

BIT STRING MANIPULATIONS

E-l

E.l
E.l.l
E.l.2
E.l.3
E.l.4

LOGICAL OPERATIONS
Inclusive OR
Logical Product
Logical Complement
Exclusive OR

E-l
E-l
E-l
E-l
E-2

E.2

SHIFT OPERATIONS

E-2

SOFTWARE PERFORMANCE REPORTS

F-l

C.l.2

APPENDIX

C-3
C-4
C-8
C-9

C-lO
C-lO
C-12
C-13
C-13
C-13

C-lS
C-2·1

(,-

P~F~E

This document provides information necessary to compile, task build,
execute, and debug a FORTRAN program under the RSX-llM operating
system. Chapter one describes the operational procedures.
Chapter
two provides information about the Object Time System (OTS). This
system is a collection of routines selectively linked to the user's
program which perform certain arithmetic input/output, and system
dependent service operations. It also detects and reports run-time
error
conditions.
Chapter
three
describes
system dependent
information not included in the PDP-ll FORTRAN Language Reference
Manual.
The Appendices provide reference information about internal
data representations, system subroutines, error diagnostics, and
compatibility of FORTRAN-IV with otherPDP-ll FORTRAN processors.
This manual should be used only after some knowledge of the FORTRAN
Language, as implemented on the PDP-ll, has been acquired. The
associated document which may be used for this purpose is titled
PDP-ll FORTRAN Reference Manual (DEC-ll-LFLRA-A-D). The user should
also be familiar with the RSX-llM Operating System as described in the
RSX-llM Operations Procedure Manual (DEC-ll-OMOGA-A-D).
NOTE
Subroutines are provided in the FORTRAN
library to enable real-time
process
control, process I/O, and
RSX-llM
system directives to be performed by
means of FOR~RAN.
These routines and
their ISA-standard
calls
are fully
described in the RSX-llM
Executive
Reference Manual and the RSX-llM
I/O
Drivers Manual.
DOCUMENTATION CONVENTIONS
All RSX-llM executive and system program command lines are terminated
by a RETURN.
Since this is a non-printing character, at certain
places in the text the notation <CR> represents the RETURN key.
Some special keyboard characters require that the CTRL (control) key
be pressed simultaneously with a second character. These characters
are denoted by t (up arrow): e.g., tz (CTRL Z).
In all examples, text printed by the RSX-llM executive or
system
program is underlined, text typed by the user is not underlined.

vii

CHAPTER 1
OPERATING PRocEDURES

1.1

USING THE FORTRAN~IV SYSTEM

Three steps are required to transform a F()RTRAN source program int.o an
executing Task. These steps are:
1.
2.
3.

Compilation
Task Build (or Linkage)
Initiation of Execution

Compilation is accomplished by invoking the FORTRAN-IV C:o*,plle:r.
Next, the Task Builder is used to construct a task image. Finally,
task execu~ion is. ini tiated by using. th~,. ~ppropri~te Monitor Console
Routine (MCR) commands. Each step in this procedure involves several
required and optional command inputs and produces an output to be used
in the next step. Other optionaL outputs are also possible.
A di.agram depicting this process is giveh in Figure l-l.The remainder
of this chapter discusses each step of' the process ih detail and gives
some simple examples.

MeR
COMMANDS

(

\.
Figure i-l
Preparing a FORTRAN Source for Execution

1.1.1

Command String

System programs rUnning under RSX-llM usually require a command string
to specify such things as input files and various options. A standard
command string has one of the following formsi
output-files-listminput-files-list
or
@command-file

1-1

CHAPTER 1.

OPERATING PROCEDURES

Both output-files-list and input-files-list are strings consisting of
file specifications separated by commas.
Each file specification
selects a file to be used as an output or input file by the system
program.
A @command-file is an indirect command file which contains
multiple command strings.
Any number of file specifiers is possible, the actual number being
determined by the system program which will use the file command
string.

1.1.2 File Specifications
Each file specifier
format:

(whether

input

output)

has

the

following

dev:[g,o]filename.type;version!switchl ••• /switchn
where:
dev:

= The physical device unit on which the

volume
containing the desired file is mounted, e.g.,
ORO:. The name consists
of
two
ASCII
characters followed by an optional one or two
digit (octal) unit number and a colon.
The
default value is SY:, the system disk. Table
1-1 depicts the legal device list.

[g ,0]

= The User Identification Code (UIC) associated

with the user file directory containing the
desired file.
This consists of a group
number and a user number. The default value
of UIC is the identification code under which
the
system
program is running, usually
[200,200] •
filename

= The name of the file.

type

= A means of distinguishing among forms of

version

/switch

= A one or two character ASCII name identifying

In RSX-llM, a filename
can be up to nine alphanumeric characters in
length.
Filename and
type
are
always
separated by a period (.).

one
file.
System programs default the file type
to an appropriate standard type, (e.g., FTN)
so
the
typical user will not need to
explicitly specify it. File type and version
are always separated by a semicolon (;).

An octal number used to differentiate among
versions of a file.
When a file is first
created using the editor, it is assigned a
version
number
of
1. If the file is
subsequently opened for editing, the file
system retains the original file for backup
and creates a new file with the same filename
and type, but with a version number of 2.
Version is in the range 0-77777.
the

switch

1-2

option.

The

switch itself may

(

CHAPTER 1. OPERATING PROCEDURES
have three forms. If the switch
for example, is SW. Then:

designator,

/SW sets the switch action;
/-SW, negates the switch action, and
/NOSW also negates the switch action.
in addition the switch identifier may be
followed
by any number of values.
The
permitted values are ASCII strings, octal
numbers, and decimal numbers. The default
for a value is octal. Decimal values must be
terminated by a decimal
point.
Values
preceded by a number sign (I) are octal; the
octal
option
is
included
for explicit
documentation purposes. Any numeric
value
may be preceded with a + or - sign; if the
number sign (I) is used, the + or
must
precede it. The following are valid
switch
specifications.
/SW:27:MAP:29.
/-SW
/NOSW:NOSWITCH:-#50
the
Although the filename command string has a standard syntax,
interpretation of the string and permissible options is program
dependent.

1-3

CHAPTER 1. OPERATING PROCEDURES
Table 1-1
RSX-llM Devices
PERIPHERAL DEVICES

DEVICE-UNIT

Analog to Digital Converter (ADO I-D)

ADnn:

Analog to Digital Converter (AFCll)

AFnn:

Card Reader (CRll)

CRnn:

Cassette (TAll)

CTnn:

DECtape (TCll)

DTnn:

Disk (RP04)

DBnn:

(RFll)

DFnn:

(RKll)

DKnn:

(RPOJ)

DPnn:

(RSOJ/RS04)

DSnn:

Laboratory Peripheral System (LPSll)

LSnn:

Line Printer (LPll/LSll/LVll)

LPnn:

Magtape (TU16)

MMnn:

Magtape (TMll)

MTnn:

Synchronous Line Interface (DP-ll)

XPnn:

(DU-ll)

XUnn:

Asynchronous Line Interface (DLIl-E)

XLnn:

Terminal (DLll/DHll/DJll)

TTnn:

Universal Digital Controller (UDell)

UDnn:

PSEUDO DEVICES

1.2

Console Listing

CL:

Console Output

CO:

Pseudo Input Terminal

TI:

System Default Device

SY:

USING THE FORTRAN-IV COMPILER

The FORTRAN-IV Compiler is an RSX-llM system program initiated via the
FOR MCR-command. It produces relocatable object modules from FORTRAN
source programs.

1-4

(

CHAPTER 1. OPERATING PROCEDURES
To invoke the Compiler enter the following
characters typed by the system are underlined).

MCR

command.

(All

MCR)FOR<CR)
The FORTRAN-IV Compiler then prints "FOR)ft
ready to accept a command string.

indicate

that

A standard command string as described in Section 1.1.1 is used to
specify input and output files to the Compiler. None, one or two
output files may be specified. The first output file is the object
module file and the second is the listing file. Only one input file
may be specified. This is the FORTRAN source file.
It may contain
one or more FORTRAN program units (Main Program and/or Subprograms).
File types default as shown in Table 1-2.
Table 1-2
FORTRAN Compiler Default File Types
File
Object
Listing
Source
Command

Default Type
OBJ
LST
FTN
CMD

Up to two levels of indirect command files are permitted.
An

example FORTRAN command sequence is shown below:
2FOR
FOR)OBJECT,LIST=FILEl

This command string directs the Compiler to take the source file
FILE1.FTN from the system device and output the files LIST.LST and
OBJECT.OBJ to the system device. When the compilation is complete the
Compiler again prompts with ftFOR)ft. At this point another command
string may be entered or tZ may be typed to cause the Compiler to
terminate execution.
alternate way to specify a single command line is to enter the
command string on the line containing the MCR command. For example
the following performs the same function as the example above:

2FOR OBJECT,LIST=FILEl
Only a single compilation per FOR command may be specified in this
manner.
Upon completion of the compilation the compiler exits to the
RSX-llM Executive which will prompt with a)".
Either of the output files can be omitted by omitting
specification from the command string. For example:
)FOR
FOR)FlLE1=FlLEl
produces FlLE1.OBJ on the system device but no listing file.

1-5

the

file

CHAPTER 1. OPERATING PROCEDURES
~,LP:=FlLEl

produces a listing on the line printer, but no object module.

1.2.1

Compiler Switches

The FORTRAN Compiler command strings may contain switch options on the
input and output file specifications. The switches are as follows:
Switch
ILl:n

Description
Specifies the listing
encoded as follows:
ILl:O or I-Ll
ILl:l or ILl:SRC
ILl:2 or ILl:MAP
ILl:4 or ILl:COD

options.

The

argument

list d~agnostics only
list source program and diagnostics
only
list storage map and diagnostics
only
list generated code and diagnostics
only

Any combination of the above list options may be
specified by summing the numeric argument values for
the desired list options. For example:
ILI:7 or ILI:ALL or ILl

requests a source listing, a storage map listing, and
a generated code listing. If this switch is omitted
the default list option is ILI:3, source and storage
map.
(See section 1.2.2.) If no listing output is
specified the following is assumed:
TI:LIST.LST/-LI
ISP

Automatically spool listing file.
is to spool.

IDE

Compile lines with a 0 in column one. These lines are
treated as comment lines by default (I-DE)
(see
section 1.6).

lEX

Read a full 80 columns of each record in the source
file.
Only the first 72 columns are read by default
(I-EX).

lID

Print FORTRAN identification and version number.
The
default (/-ID) causes the identification and version
number not to be printed.

lOP

Enable the Common Subexpression (CSE) Optimizer. In
general the CSE optimizer will make the program run
faster. However, the size of the program may be
different than with no optimization (/-OP). The default
is to optimize (lOP).

1-6

The

default

(/SP)

CHAPTER 1. OPERATING PROCEDURES
/SN

Include internal sequence numbers (ISN) • The /-SN
option reduces storage requirements for generated code
and slightly increases execution speed but disables
line number information during Traceback. The default
(/SN) is to use ISNs.

/14

Two word default allocation for integer variables.
Normally, single storage words will be the default
allocation for integer variables not given an explicit
length specification (i.e., INTEGER*2 or INTEGER*4).
Only one word is used for computation.
/-14 is the
default.

IVA

Enable vectoring of arrays (see section
default (/VA) is to vector arrays.

/WR

Enable compiler warning diagnostics.
(/WR) is to issue warning diagnostics.

2.5).
The

The

default

Switch default summary:
/LI:3/SP/-DE/-EX/-ID/OP/SN/-I4/VA/WR

1.2.2

(

List Formats

There are three optional sections that may be included in the list
file. By default the source program and the storage map are included.
The generated code may also be included.
Any combination of these
sections can be requested by using switches in the Compiler command
string (see section 1.2.1). A description of each section is given
below.
Figure 1-2 provides a sample of the information included in
each section.

1.2.2.1

Source Listing - The source program unit is listed in this
as it appeared in the input file. Internal sequence numbers
are added by the compiler for easy reference.
Note that internal
sequence
numbers are not always incremental.
For example the
statement following a logical IF will have an internal sequence number
2 greater than that of the IF. The IF statement has internally been
assigned 2 sequence numbers: one for the comparison and one for the
associated
statement.
Comments,
uncompiled
statements,
and
continuation lines do not receive internal sequence numbers.
sec~ion

1.2.2.2 Storag~p' Listin~- This section includes a list of all
symbolic names referenced in the program unit. A location offset from
the base of the program unit (subject to relocation during task
building) is given for all local symbols. There is also a description
of the symbolic name including usage, data type, and in the case of
COMMON blocks and array names, the defined size.
NOTE
Blank
BLOCK

COMMON is described as COMMON
/ / in the storage map, but is

1-7

CHAPTER 1. OPERATING PROCEDURES
located on a Task Builder map under
Program Section named .$$$$.

the

1.2.2.3 Generated Code Listin9L - This section of the list file
contains a symbolic representation of the object code generated by the
compiler (see section 2.2) including a location offset from the base
of the program unit, the symbolic Object Time System (OTS) routine
name, and routine arguments. The code generated for each statement is
labeled with the same internal sequence number as that in the source
program listing, providing easy cross reference.

(

1-8

CHAPTER 1.

OPERATING PROCEDURES

FORTRAN IV
ISN '0006
000066 LSNS
000012 MorSMS
000016 CHISIS
000102 CPt!SM
000106 eFIS
000110 AnFSIS
00011. MOFI8M
ISN .0001
000120 rSNS
000122 MO'SMS
000126 otF~IS
000132 AOFSMS
00013e CO'S
11100140 MO"SSM
ISN .0008
0e0t44 ISNS
00111146 RfLS
0001!52 RELI
000te6 CALS
000164 MODIRM

(

ISN '0009
Ul"17" ISNS
000112 MOISMS
000176 COIS
000200 AODSMS
000204 MOFSI!
000210 CDFS
000212 AODtSS
000214 MOOISM
ISN '0010
PlIU220 I·SNS
00111222 RELS
000226 RELS
01U232 IFill!
000234 RELS
0002'121 TVel
11100242 MOFSI!
0002'6 CDFS
1110.02150 MUDSMS
0002154 MOOISM
000260 ~ELS
00026. TVOS
000266 fOLI
ISN *0012
00021111 LSNS
000214 STPt

GENERATED CODE
*000006

FORTRAN IV

~000~!

*000002
ClI000e4

0001
00"2
0OO3
IIJ0fU
000S
0006
0001
121008
0009
0010
0011
0012
0013

*040640
000030

000030
*0.0400
12100020
e000.4

M01-"!

SOURCE LISTING

INTEGER tNT
REAL. REAL
COMPLEX 1M"
DOUBLE PRECISION D8L.E
DATA INT/10Q11
REAL. • INT/2 • S.
OSL.E • R[AL./2 • • 3.1415g682516
IMA' • CMPL~(REAL., 3,21)
OBL.E • DeL.E + tNT/2 • 5,
PRINT 10,I~AG,DBL[ • 10,
FORMAT(lx,3F12.6)
STOP
END

e00024
000030
*000002 CMPLX.'000000
Cl.I00034

000064
000044
UJ40640
FORTRAN IV

STORAGE MAP

INT
REAL

INTEGER.2
REAL.4
COMPL[X·8
REAL·8
COMPLEX.!

000044

000016
000010

1M',

DBI.E
CMPL)(

000006
000030
000034
0000.'
00121000

000034
*041040
000044
000C'!'e.t
9000!4

.000014

ISN .0013
000276 ISNS
000300 RETS

Figure 1-2
Sample Listings
1-9

VARIABLl
VARIABL.E
VARIABLE
VARIABLE
PROCfDURE

CHAPTER 1. OPERATING PROCEDURES

.'__ -l... 2 • ~

Compiler Memory Requirements

The FORTRAN Compiler runs in a minimum partition of 7K words. If run
in a larger partition it uses the extra space for program and symbol
storage. If available storage is exceeded during the compilation
process, a FATAL ERROR T is issued.

1.3

USING THE TASK BUILDER TO LINK FORTRAN PROGRAMS

The Task Builder links relocatable object modules together to create a
task image.
The object modules may come from user specified input
files, user libraries, or system libraries.
References to global
symbols defined in one module and referenced in other modules are
resolved. The system object module library (SY:[l,l]SYSLIB.OLB) is
automatically searched to resolve any remaining undefined symbols.
References to resident common blocks and shared libraries are also
resolved.
Thus, the task image produced is ready to be run under the
RSX-llM executive. The Task Builder also allows the building of tasks
with overlay structures.
For a more complete description of the Task Builder refer to the
Builder reference manual (DEC-ll-OMTBA-A-D).
The Task Builder is a system program invoked by the TKB
The user initiates it by typing

MCR

Task

command.

>TKB
The Task Builder then prints "TKB)" to indicate that it is ready to
accept a command string.
Alternatively the command string may be
typed on the same line as the TKB MeR command.
The command string
should be a standard string as described in Section 1.1.1.
The first output file specifies the task image file.
A second file
may be specified if a memory allocation map is desired. A third file
may be specified to contain the program sections and global symbol
definitions in relocatable object module format.
This file is
described in the RSX-llM Task Builder Reference Manual.
The input
files contain the object modules to be linked. Additional lines of
input file specifications may also be entered. After all input files
have been typed, the user should type a line consisting only of "II"
(unnecessary if command string is specified as part of MCR command).
The Task Builder will then build the task image. The default file
types used by the Task Builder are shown in Table 1-3.
Standard indirect command file specifications are also
the Task Builder.

1-10

,,-- .

qz'~~~

acceptable

(

CHAPTER 1. OPERATING PROCEDURES
Table 1-3
Task Builder Default File Types
File

Default Type

Task Image
Map
Input
Library
Overlay
Description
Command

TSK
MAP

OBJ
OLB
ODL
CMD

The following simple example builds a task image for the
OBJECT.OBJ created by the first example in Section 1.2.
HCIDTKB

object

file

2TKB TASK ,LP: -OBJECT

~TASK,LP:-OBJECT

TKB>/I
This creates the task image file TASK.TSK on the system device and
lists the memory allocation map on the line printer. Any references
, in OBJECT.OBJ to FORTRAN OTS routines are resolved automatically
because the OTS resides in the system object module library.

1.3.1

Task Builder Switch options for FORTRAN Programs

A FORTRAN programmer should be aware of several Task Builder switch
options.
Some of these are necessary for correct linking of FORTRAN
programs. Others select options which some users may find useful.
Detailed descriptions of these switches and other Task Builder
switches may be found in the Task Builder Reference Manual.
The lEA switch must be used on the task image file if
the KEll Extended Arithmetic Element.

the

task

uses

The /FP switch must be used on the task image file if
the PDP-ll/45 Floating Point Processor.

the

task

uses

abbreviated form of memory allocation map may be selected by
specifying ISH on the map file. See the Task Builder Reference Manual
for a description of both the long and short map formats.

The ILB and /MP switches when specified with an input file indicate an
object module library and overlay description file, respectively.
These two switches will be discussed in more detail in the following
sections.
The /SP switch may be used to cause automatic
Builder map file.

1.3.2

spooling

the

Task

Task Builder Options for FORTRAN Programs

The Task Builder allows numerous keyword options to be' specified in
addi tion to the swi tches described above. Several of these are of
particular interest to the FORTRAN user.
1-11

CHAPTER 1. OPERATING PROCEDURES
If keyword options are to be specified,the user must not specify the
command string with the MCR command and must terminate his command
input with a line consisting of a single slash, "I", instead of a
double slash as described previously. This causes the Task Builder to
solicit option information by printing:
ENTER OPTIONS:
~

At this point Task Builder options may be entered, one option per
line.
After each option is entered the Task Builder prompts with
"TKB)" indicating that it is ready to accept the next option.
After
specifying the desired options a single slash, "I", should be entered
to indicate no more options. The Task Builder then proceeds to build
the task and produce the requested output files. When it has finished
it again types "TKB)" and is ready to accept a new command string. To
exit from the Task Builder type tz (Control Z).
The Task Builder options particularly relevant to FORTRAN
are described below.

programmers

1.3.2.1 ACTFIL - The number of files that may be simultaneously open
by a task is defaulted to 4. Buffer space is allocated in the task
image for this many files. When multibuffering is used via a call to
FDBSET, the number of buffers needed by the FORTRAN task will be
greater than the number of files. The number of buffers may be made
smaller to conserve memory or larger to allow more files to be open
simultaneously by means of the Task Builder option:
ACTFIL .. n
where n is the decimal number of buffers desired. (An attempt to open
a file when space is not
available may cause a fatal error at
run-time.)

1.3.2.2 COMMON - If a shared common block is to be referenced by the
FORTRAN program this intention must be declared by specifying the
following option:
COMMON a name:access
where name is the name associated with the resident common block and
access is either RO for Read Only (meaningful for mapped systems only)
or RW for Read/Write. The FORTRAN common with the same name is used
to refere~ce the data in the resident common.

1.3.2.3 LIBR - If a shared library is to be referenced by the FORT~
program the following option must be used:
LIBR- name:access
where name is the library' s name and access is either RO for Read Only
(meaningful for mapped systems only) or RW for Read/Write. Libraries
are discussed in more detail in Section 1.3.3.
1-12

(

CHAPTER 1. OPERATING PROCEDURES

1.3.2.4 MAXBUF - The maximum record size that can be handled by the
FORTRAN object time system for input/output is defaulted to 132
(decimal) bytes. This may be increased by specifying the Task Builder
option:
MAXBUF = n

where n is the number of bytes. The default generally is adequate for
formatted input/output.
If direct access operations are performed
using records larger than 132 bytes, the user must employ this option
to specify the size of the largest record which will be handled.

1.3.2.5 TASK - The task image being built may be
task name by specifying the following option:

given

explicit

TASK == name
where name may be from one to six alphanumeric characters the first of
which must be alphabetic. If no task name is specified at task build
time, it may be explicitly specified at install time, or the Task
Builder will use the first six characters of the task image file name
as the task name.

1.3.2.6 UIC - The default Ule under which the task will
be set by specifying the option:

execute

may

Ule - [g,o]
where g,o is a valid group and owner number combination.
option is not specified the default Ule is [200,200].

this

1.3.2.7 UNITS - The default number of logical units available to the
FORTRAN program is 6, that is, logical units 1 through 6 inclusive.
This number may be set explicitly smaller or larger at task build time
by specifying the option:
UNITS =- n
where n is the number of logical units desired.
The default device and file name associated with a logical unit number
is discussed in Section 3.4. If a smaller number of units than the
default is desired, a UNITS command must be given before any ASG
commands.

1.3.2.8 FMTBUF - The default size (32 words) of the buffer used for
object time format compilation may be changed by use of the following
Task Builder option:

1-13

CHAPTER 1. OPERATING PROCEDURES
FMTBUF -

where n is the decimal size, in words, of the object time format
compilation buffer.
The total size needed for format compilation is
approximately equal to the number of characters in the largest object
time format used by the program.

1.3.2.9 ASG - Logical unit numbers may be assigned to physical device
units by use of the following option:
ASG - devl:nlln21 ••• ,dev2:ml:m2: ••• , •••
where each dev is a physical device unit name and each n and m i8 a
valid logical unit number.
The default device assignments would
appear as:
ASG - SY:l:2:3:4,TI:5,CL:6

1.3.2.10 PAR - A task may be built to execute in a specific partition
by use of the following option:
PAR - pname[:start:length]
where:
pname

- is the name of the partition for which this task is
being built. In unmapped systems, the task must run in
this partition.

start

- the octal starting address of the partition.
This
address must be on a 32-word boundary if unmapped, 4K
boundary if mapped.

length

- the octal length of the partition in
specified length must be divisible by 64.

bytes.

The

The optional argument's start and length must be specified when the
task is being built on a system that does not have the partition named
by the pname parameter.
Normally a task is built to execute from the default
When PAR is not specified the default assumed is:

1.3.3

partition

GEN.

FORTRAN Library Usage

RSX-llM Library consists of a collection of object modules.
Two
kinds of libraries exist, shared and relocatable. The Task Builder is
used to include modules from relocatable libraries in a task image.

1.3.3.1 Relocatable Libraries - Relocatable libraries are stored in
files on a file structured volume such as a disk. Object modules from

1-14

CHAPTER 1. OPERATING PROCEDURES
relocatable libraries are copied into the task image of each task
referencing the module. A relocatable library may be specified as an
input file to the Task Builder.
Such a file specification must
include the /LB switch to indicate that the file is a library file.
When a library specification is encountered, those modules in the
library which contain definitions of any currently undefined global
symbols are included in the task image.

1.3.3.2 Shared Libraries - Shared libraries are located in main
memory and a single copy of each library is utilized by all
referencing tasks. Access to a shared library is gained by using the
LIBR option as described in Section 1.3.2.3. Shared libraries are
built by the user with the Task Builder; they must contain shareable
(reentrant) code.

1.3.3.3 §ystem Libraries - Each RSX-llM
system
has
a
system
re10catable
library.
The system relocatable library, SY:[l,l]
SYSLIB.OLB is automatically searched by the Task Builder if any
undefined global references are left after processing all user
specified input files. If the definition of one of these undefined
global symbols is found, the appropriate object module is included in
the task. The FORTRAN OTS is included in the system object library
and hence is loaded automatically with FORTRAN programs.

1.3.3.4 User Libraries - The user can construct his own relocatab1e
libraries of assembly language and FORTRAN routines by using the
RSX-llM Librarian. These libraries are accessed by using the /LB
switch as described in preceding sections.
If MATRIXLIB.OLB is a
relocatable
library
containing
matrix
manipulation routines and PROG.OBJ is the object file of a compiled
FORTRAN program which calls the matrix routines, the following command
string might be given to the Task Builder:
'
TKB>PROG,LP:aPROG.OBJ,MATRIXLIB.OLB/LB
TKB>//

1.3.4

Overlays

The ov~rlay facility supplied by RSX-llM allows large programs to be
executed in relatively small partitions of main memory. The overlay
system is virtually invisible to the FORTRAN programmer. All he must
do
is
specify
the overall overlay structure, indicate which
subprograms are to reside in various parts of the structure and
specify, using the overlay description language COOL), which routines
are to be autoloadable.
Overlay
loading
may
be
performed
automatically (autoload) or via explicit load requests (manual load).
A complete description of manual loading can be found in the Task
Builder Reference Manual.
The overlay structure is specified as a tree structure in terms of the
Task Builder's Overlay Description Language' (ODL). If an overlay
structure is to be built then only one input file can be specified to

1-15

CHAPTER 1. OPERATING PROCEDURES
the Task Builder.
This file must contain the appropriate ODL
statements. The specification of this file in a command string must
be followed by the IMP switch to identify it as an ODL file.
Simple overlay structures may be constructed using only two ODL
statements, .ROOT and .END. The following short examples demonstrate
how to build overlays.
For a more detailed explanation of this
facility refer to the Task Builder Reference Manual.
Suppose a FORTRAN program consists of a main program (MAIN.OBJ) which
performs input and output and calls three subroutines, one which does
pre-processing of the data (PRE.OBJ), one which performs the primary
processing function of the program (PROC.OBJ), and one which does
post-processing of the data (POST.OBJ). The following ODL statements
specity an overlay structure which has a resident portion which
consists of the main program and three overlays which share the same
memory locations. Each overlay contains a single subroutine. Figure
1-4 is a diagram of the overlay structure.
The ODL statements to
create this structure, are as followS1
• ROOT
.END

MAIN-* (PRE,PROC,POST)

The .ROOT statement is used to declare the tree structure.
The • END
statement indicates the end of the ODL statements. The names specify
object file names (default file type is OBJ). Commas separate
descriptions of overlay segments which occupy the same storage.
Parentheses are used to group these descriptions.
Dashes separate
descriptions of modules which are concatenated. If automatic loading
of overlays is desired, an asterisk must precede the name of each file
which contains a subprogram invoked from a point in the overlay
structure which is closer to the root of the structure than the
subprogram.
If all files within a pair of parentheses are to be
automatically loaded, a single asterisk placed in front of the opening
parenthesis may be used instead.
This indicates that the segment
containing the module will be automatically loaded whenever a call is
made to a subroutine in the file.

MAIN

------~-------~

PRE

PROC

Figure 1-3
Simple OVerlay Structure

1-16

POST

CHAPTER 1. OPERATING PROCEDURES
A path of an overlay structure is any route from the root of the
structure which follows a series of branches to an outermost segment
of the tree.
Figure 1-4 has only three short paths, MAIN-PRE,
MAIN-PROC, and MAIN-POST. A program in one overlay segment may call a
subprogram in another segment if and only if the two segments occur on
a common path. Thus, MAIN may call PRE, PROC or POST, but the three
subroutines cannot call each other.
A more complex structure is given in Figure 1-5 and
the ODL statements
• ROOT

specified

A-B-*(C,D-(E,F,G»

• END
The paths in this structure are A-B-C, A-B-D-E, A-B-D-F, and
A possible sequence of calls is the following,
A calls G,

G calls B,

A-B-D-G.

B calls D

A
B

I
0

I
F

E
G

Figure 1-4
OVerlay Structure

1.4

USING MCR TO INITIATE TASK EXECUTION

Once a task image has been built, MCR may be utilized to start the
execution of that task.
First, the task must be installed in the
system by typing

1-17

CHAPTER 1. OPERATING PROCEDURES
lINS

Filespec

where Filespec is a specification of the file containing the task
image.
The default device is the system disk and the default file
type is TSK. The task is installed in the partition to. which it is
bound.
In an unmapped system it may be installed in this partition
only. In a mapped system the /pAR - pamame keyword may be used to
install the task in any partition of sufficient size. An error will
result if the partition is larger than the checkpoint area in which
the task was built. The default task priority is 50. The task name
may be specified by using the /TASK-name switch.
The execution of an installed task is initiated by typing the RUN
command:

MeR

2RUN tsk
where tsk is the task name which was specified when the task was
installed or, if none was given then, the name specified when the task
was built, or if none was given then, the first six characters of the
task image filename.
A task may be terminated prior to its normal termination by typing the
ABORT MeR commands

1ABO

tsk

Execution of a task may be suspended by a FORTRAN pause statement, or
ended by a STOP statement. When this occurs, the Object Time System
will type a line with the task name, the statement which caused the
execution halt, and the contents of the display (text following STOP
or PAUSE). To continue execution after a pause, type in the RESUME MeR
conunand, which takes the form s
>RES ts~

A task which terminates as a result of a CALL EXIT statement will
produce any output indicating it is terminating.

1.5

not

EXAMPLES

The following sequence might be used to compile, link
FORTRAN task consisting of:

and

execute

the FORTRAN main program MAIN.FTN,

the FORTRAN subroutine SOORl. FTN ,

several FORTRAN subroutines in the file UTILITY .FTN,

some subroutines in the object module library MATLIB.OLB, and

the resident common block named PARM.
,2 FOR

FOR)MAIN,MAIN-MAIN

~SUBR1,SOORl-SUBRl

ZQB}UTILITY,UTILITY-UTILITY

1-18

CHAPTER 1. OPERATING PROCEDURES
FOR>tz
>TKB
TKB>TSKIMAGE=MAIN,SUBRl,UTILITY,MATLIB.OLB/LB
TKB>/
ENTER OPTIONS:
TKB>COMMON=PARM:RO
TKB>TASK=FSYS
TKB>//
2INS TSKIMAGE
2RUN FSYS
The listing files produced by the Compiler, MAIN.LST, SUBRl.LST and
UTILITY.LST, are automatically printed by the line printer spooler
task and are deleted after printing. All files reside or are created
on the system disk.
The preceding procedure could be accomplished through the use of
indirect command files.
Suppose the file COMPILE.CMD contains the
following:
MAIN,MAIN=MAIN
SUBR1,SUBR1=SUBR1
UTILITY,UTILITY=UTILITY
and the file TASKBLD.CMD contains the following:
TSKIMAGE=MAIN,SUBR1,UTILITY,MATLIB.OLB/LB

COMMONaaPARM:RO
The following is then equivalent to the previous example.
2FOR @COMPILE
>TKB

~@TASKBLD

TKB>TASK=FSYS
TKB>//
2INS TSKIMAGE
2RUN FSYS

1.6

DEBUGGING A FORTRAN PROGRAM

The RSX-11M debugging program, ODT, usually cannot be effectively used
with a FORTRAN program due to the nature of the object code generated
by the FORTRAN Compiler (see section 2.2).
However, in addition to the FORTRAN OTS error diagnostics which
include the Traceback Feature (see section 2.5), there is another
debugging tool available to the FORTRAN programmer. Placing a D in
column one of a FORTRAN statement allows that statement to be
conditionally compiled. These statements are considered comment lines
by the FORTRAN Compiler unless the /DE switch is used in the Compiler
command string. In this case the lines are compiled as reg~lar
FORTRAN statements.
Liberal use of PAUSE statements and selective
variable printout can provide the user with a method of monitoring
program execution.
This feature allows the inclusion of debugging
aids that can be compiled in the early program development stages and
later treated as comment lines.

1-19

CHAPTER 2
FORTRAN-IV OPERATING ENVIRONMENT

2.1

FORTRAN-IV OBJECT TIME SYSTEM

The FORTRAN Object Time System (OTS) is composed of the following:
1.

Math routines, including the FORTRAN library functions and
other arithmetic routines (e.g., floating point routines),

Miscellaneous utility routines (ASSIGN, DATE, ERRSET, etc.),

Routines which handle various types of FORTRAN I/O,

Error handling routines which process arithmetic errors,
errors, and system errors,

Miscellaneous routines required by the compiled code,

Process I/O routines (AFC, UDC, etc),

Laboratory Peripheral Routines (LPS), and

RSX-1IM executive directives.

I/O

The FORTRAN Library is designed as a collection of many small modules
so that unnecessary routines can be omitted during task building. For
example, if the user program performs only sequential formatted I/O,
none of the direct access I/O routines are included in the task.

2.2

OBJECT COPE

Typical FORTRAN operations often require common sequences of PDP-ll
machine instructions.
For example, at the end of any DO-loop, the
index variable must be incremented, compared with the limit value, and
Other standard sequences might be
a conditional branch taken.
generated to locate an element of a multidimensional array, initialize
an input/output operation, or simulate a floating-point operation not
supported by the hardware configuration.
These common sequences of PDP-ll instructions are contained in a
library known as the Object Time System. The FORTRAN Compiler selects

2-1

CHAPTER 2. OPERATING ENVIRONMENT
a certain combination of these instruction sequences to implement a
FORTRAN program.
During program execution, these sequences are
threaded together and effect the desired result.
The Compiler references a library instruction sequence by generating a
word containing the address of the first instruction in the sequence,
followed by information upon which the instructions are to operate.
In the case of the end-of-Do-loop sequence the information required is
the location of the index variable, the limit value, and the address
of the beginning of the loop. At runtime, register R4 is used to
thread together the various references to
library
instruction
sequencesJ the last instruction executed by each instruction sequence
is JMP @(R4)+, which transfers control to the next library instruction
sequence.
The mnemonics (global names) used for the library routine names follow
a
logically consistent format.
The mnemonics are usually six
characters in length. The first two characters specify an operation.
The third character specifies the mOde of the operation, i.e.,
integer, floating, double precision, complex, or logical. The fourth
character is always a dollar sign ($). The fifth and sixth characters,
if present, specify, respectively, a source and destination for the
operation.
The source element for the operation can be a memory
location, the hardware stack, the hardware registers, or an in-line
argument which can be referenced through R4. The destination element
for an operation can be a memory location, the hardware stack, or a
location specified' as an in-line argument which can be referenced
through R4.
The library routines perform arithmetic operations, compare values,
test values, calculate subscripts, convert from one mode type to
another, and trans~.er program control. There are special routines to
handle
internal statement numbers (ISNs), enabling the FORTRAN
Traceback feature, a routine to handle subprogram control transfer,
~d
a routine to push the address of variables on the hardware stack.
There are also sever~l routines to handle special FORTRAN runtime
operations such as PAUSE, STOP, I/O initialization, and I/O data
transfers.
For example, the following FORTRAN program:

c
C
C

PROGRAM TO DEMONSTRATE THE CODE GENERATED BY
THE FORTRAN COMPILER.

0001
0002
0003
0004
0005
0006

DIMENSION RARRAY (10,10)
I ALLOCATE A REAL*4 ARRAY
I ADD ONE TO I
I - (3*2 - 5) + I
I COMPUTE AN EXPRESSION
J = (I+100)*(N**2)
I ASSIGN A VALUE TO A REAL
A - 2.0
RARRAY(2,1) = RARRAY(l,l) + A I SUM OF TWO REAL VALUES
END

would generate object code that can be symbolically
follows (the storage map is included for reference):
STORAGE MAP

FORTRAN
NAME

OFFSET

ATTRIBUTES

RARRAY

000006
000626

REAL * 4

INTEGER*2

ARRAY (10,10)
VARIABLE

2-2

represented

/
\

CHAPTER 2. OPERATING ENVIRONMENT
000630
000632
000634

N
A

IN'l'EGER*2
INTEGER*2
REAL * 4

FORTRAN

VARIABLE
VARIABLE
VARIABLE

GENERATED CODE

ISN 10002
000640

ICI$M

000626

000626
'000144
000632
000632

MOVE VALUE OF INTEGER I ONTO STACK
; ADD 100 TO VALUE ON TOP OF STACK
; MOVE VALUE OF INTEGER N ONTO STACK
, AND SQUARE IT (MULTIPLY BY ITSELF)
, MULTIPLY (I+100) AND (N**2)
1 STORE VALUE ON TOP OF STACK INTO J

INCREMENT THE INTEGER WHOSE ADDRESS
, IS 000626 (I)

ISN 10003
000644
000650
000654
000660
000664
000666

MOI$MS
ADI$IS
MOI$MS
MUI$MS
MUI$SS
MOI$SM

000630

ISN 10004
000672

MOF$IM

1040400 000634

, MOVE AN IMMEDIATE FLOATING CONSTANT
, (2.0) TO A

000006
000634

1 MOVE

ISN 10005
000700
000706

MOF$MM
ADF$MM

000012
000012

RARRAY (1,1) TO RARRAY (2,1)
, AND ADD A TO RARRAY (2,1)

ISN 10006
000714

(

2.3

RETURN TO RSX-11M
(EXIT FROM PROGRAM)

RET$

SUBROUTINE LINKAGE

All instances of subprogram linkage are performed in the same manner,
including linkage of user written FORTRAN subprograms, and Assembly
language subprograms. Control is passed to the subprogram via the
following instruction:
JSR

PC,routine

2-3

list

having

the

CHAPTER 2.

OPERATING ENVIRONMENT

I I of arquments

Undefined

address of argument 11
address of argument 12

···
address of argument In
The value -1 is stored in the argument list as the address of any null
arguments.
Null argwnents in CALL statements appear as successive
commas, e.g., CALL SUB (A"B)
Control is returned to the calling program via the instruction I
RTS

An assembly language subroutine to find

integers using the following calli
CALL IADD

the

sum

any

number

(numl,num2, ••• ,numn,isum)

might look like the following:
IADD: :
1$:

2.4

• TITLE
MOV
CLR
DECB
ADD DECB
BNE
MOV
RTS
.END

lADDER
(RS) +,RO
Rl
RO
@(R5) +,Rl
RO
1$

Rl,@(RS)+
PC

,GET t OF ARGUMENTS
, PREPARE WORKING REG.
I CALCULATE t OF TERMS TO ADD
,ADD NEXT TERM
, DECREMENT COUNTER
, LOOP IP NOT DONE
,RETURN RESULT
, RETURN CONTROL

SUBPROGRAM REGISTER USAGE

A subprogram that is called by e FORTRAN program need not preserve any
registers.
However, the contents of the hardwar~ stack must be kept
such that each 'push' onto the stack will be matched by a 'pop' from
the stack prior to exiting the routine.
User-written assembly language programs that call FORTRAN subprograms
must preserve any pertinent registers before calling the PORT RAN
routine and restore the registers, if necessary, upon return.
Function subprograms return a single result in the hardware registers.
The register assignments for returning the different variable types
are listed below:
2-4

CHAPTER 2. OPERATING ENVIRONMENT

Integer and Logical functions - result in RO
Real functions - high order result in RO, low order result in Rl
Double Precision functions - result in RO-R3, lowest order
result in R3
Complex functions - high order real result in RO, low order real
result in Rl, high order imaginary result in R2, low
order imaginary result in R3

2.5

VECTORED ARRAYS

Array vectoring is a process which decreases the time necessary to
reference elements of a multidimensional array by using additional
memory to store the array.
Multidimensional arrays, which are actually stored sequentially in
memory, require certain address calculations to determine the location
of individual elements of the array. Typically, a mapping function is
used to perform this calculation. For example to locate the element
LIST(1,2,3) in an array dimensioned LIST(4,5,6) a function equivalent
to the following may be used. This function identifies a location as
an offset from the origin of the array storage.
(sl-l) + dl * (s2 - 1) + dl * d2 * (s3 - 1) 0) + 4 * (
1 ) + 4 * 5 * (
2 ) - 44

(

where si - subscript i
di - dimension i
Since such a mapping function requires multiplication operation(s),
and since some PDP-ll hardware configurations do not have the MOL
instruction, the compiler may 'vector' some arrays and thereby reduce
execution time at the expense of memory storage.
,f

If an array is vectored, a particular element in the array can be
located by a simplified mapping function, without the need for
multiplication. Instead, a table lookup is performed to determine the
location of a particular element.
For example, a vectored, two
dimensional array B(5,5) automatically has associated with it a one
dimensional vector that would contain relative pointers to each column
of array B. The location of the element B(m.n), relative to the
beginning of the array, could then be computed as:
Vector(n) + m
using only addition operations.
array vectoring process.

Figure 2-2

graphically

depicts

the

The compiler decides whether to vector a multi-dimensional array based
on the ratio of the amount of space required to vector the array to
the total storage space required by the array.
If this ratio is
greater than 251, the array is not vectored and a standard mapping
function is used instead. Arrays with adjustable dimensions are never
vectored.
vectored arrays are noted as such in the storage map
listing.
2-5

CHAPTER 2. OPERATING ENVIRONMENT
The Compiler I-VA switch can be used to suppress all array vectoring.
The amount of memory required to vector an array can be computed as
the sum of all array dimensions except the first. Por example, the
array X(50,10,30) requires 10+30-40 words of vector table. Note that
the array V(5,100) requires 100 words of vector storage, whereas the
array Y(lOO,S) requires only 5 words of vector storage.
It is
therefore advantageous to place an array's largest dimension first if
it is to be vectored.
Wherever possible, vector tables are shared among several different
arrays.
The compiler arranges sharable vectors under the following
conditions:
1.

Arrays are in the same program unit

For the i th dimension vector to be
shared by the arrays, dimensions to
the left of the ith dimension must
be equivalent in each array_

For example, given the statement DIMENSION A(10,10),B(10,20), A and B
share a 20 word vector for the second dimension that contains the
values 0, 10, 20, 30, 40, SO, 60, 70, 80, 90, 100, 110, 120, 130, 140,
150, 160, 170, 180, 190, of which the array A uses only the first ten
elements.

Array B
B (1 ,1)
B(2,1)
B(3,1)
B(4,1)
B (5 ,1)
B(1,2)
B(2,2)
B(3,2)

Associated
Vector
P1
P2
P3
P4
P5

•
•
•

B(1,5)
B(2,5)
B(3,5)
B(4,5)
B(5,5)

The location of element B(m,n) Vector(n) + m
P5

Figure 2-2
Array Vectoring

2-6

CHAPTER 3
~X-llM

FORTRAN-IV SPECIFIC CHARACTERISTICS

'l'hie chapter deals with information specific to RSX-llH FORTRAN-IV
that ~s Q~i.tted from or relaxes r.est~iQtion8 included in the PDP-ll
FORTRAN Language Referenqe Manual.
It S~9~~d b~ noted that d~vi,ations f~om FORTRAN syntax requirements
outlil.l~d
ip the PDP-ll FORT~ Lanqu~ge Reference Manual, even if
accept~~~
~n
RSX-llM FORTRAN, 4ecrease the portability of the
prog;am, and may J?~Ohibit successful execution on another PDP-ll
system.

~ 4! ~

VNUMLEf .. ~~s.

~.~~l..~.r.t 'QR'l'~ ~.llQWt:I v~~i.~le names t.o extend Past six characters in
~@l.lg1;~.
flow,'VEtl;', Qnly the. ti,rsi; si.x Qhaxoaoters are significant and
should be upi,q~e. ~Qng· ~ll var-iable n8Jlles in the program unit.
A

warning diagnostic is given for each variable name which exceeds six
characters in length. The diagnostic will be suppressed if (/-WR) is
included in the compiler command string'.

3.2

INITIALIZATION OF COMMON VARIABLES

RSX-llM FORTRAN
be

CO~o~,
to
~tatement.

3~~

allows any
ini.ti_lize.d

variables in COMMON, i.ncluding blank
i.n any program uni.t by use of the DATA

CONTINUATION LINES

RSX-llM FORTRAN does not place any limi.ts on
continuation iines that a statement may oont.ain.

~.~ 4

DEFAUL~' LOGI.CAL

the

number

WIT .. J):EVl.CEIFlLE ASS·I,GNMBNTS

Liste.d: in table 3-1 axe the default 109ica.l. unit. - device and filename
assignments. . The default device assignments may be changed at task
build time via the ASG keyword option or prior to execution via the
REASSIGN MCR command. For example the command:

2REA tsk 3 LP:
3-1

CHAPTER 3. SPECIFIC CHARACTERISTICS

connects logical unit 3 to the physical device unit line printer in
the task named tsk.
The device and/or filename assignments may be
changed at execution time by use of the ASSIGN library routine (see
section B.2). The default filename conventions hold for logical units
not listed below, i.e., unit number 12 will have a default filename of
FOR012.DAT.
Table 3-1
FORTRAN Logical Device Assignments
Logical unit
Number

Default Device

1
2
3
4
5
6

System disk, SY:
System disk, SY:
System disk, SY:
System disk, SY:
Requestor terminal,TI:
Console listing, CL:

Default Filename
FOROO1.DAT
FOROO2.DAT
FOROO3.DAT
FOROO4.0AT
FOROO5.0AT
FOROO6.0AT

Although any combination of valid logical unit numbers may be used,
there
is
an
imposed
maximum number of units which may be
simultaneously active. By default, four file structured logical units
may
be
concurrently
active,
i.e.,
four files may be open
simultaneously. The number may be changed by use of the ACTFIL Task
Builder option (see section 1.3.2.1). It should be noted that logical
unit numbers are allocated consecutively.
For example, if logical
units 2 and 17 are used, units 1 through 17 will be allocated.
A formatted READ statement of the form:
READ f,list
is equivalent to:
READ (l,f)list
For all purposes these two forms function identically.
Assigning
logical unit number 1 to the terminal, for example, in both cases
causes input to come from the terminal.
The ACCEPT, TYPE, and PRINT statements also have similar functional
analogies.
Assigning devices to logical units 5, 5, and 6 affects
respectively the ACCEPT, TYPE, and PRINT statements.

3.5

STATEMENT ORDERING RESTRICTIONS

statement
RSX-llM FORTRAN does not impose as strict
requirements
as
those outlined in the PDP-ll FORTRAN
only
three statement
Reference
Manual.
There
are
requirements that must be met:

3-2

ordering
Language
ordering

CHAPTER 3. SPECIFIC CHARACTERISTICS
1.

In a Subprogram, the first non-comment line
FUNCTION, SUBROUTINE or BLOCK DATA statement.

The last line in a program unit must be an END statement.

Statement Functions
referenced.

must

defined

before

must

they

are

However, if the statement ordering requirements as outlined in the
POP-ll FORTRAN Language Reference Manual are not followed, a warning
diagnostic will be included with the source listing. A /-WR specified
in the Compiler command will surpress the diagnostic.

3.6

FCS/OTS FILE OPEN CONVENTIONS

A file or device is opened for I/O activity (if no file or device is
already open on that logical unit) by the execution of a READ or a
WRITE statement.
The type of File Control Services (FCS) open
operation invoked depends upon the type of file being opened. The
file may be explicitly specified as "OLD" or "NEW" in a call to
FDBSET;
these specifications are implicit in the use of the READ and
WRITE statements. When no explicit specification is made, and a READ
is the first I/O operation performed on a unit, "OLD" is assumed. If
a WRITE is the first I/O operations, then "NEW" is assumed.

3-3

(
\

{

APPENDIX A
FORTRAN DATA REPRESENTATION

A.l

INTEGER FORMAT
Sign
B-inary number
15 14

Integers are stored in a two's complement representation. If the /I4
compiler switch (see section 1.2.1) is used, an integer is assigned
two words, although only the high-order word (i.e., the word having
the lower address) is significant.
By default, integers will be
assigned to a single storage word.
Explicit
length
integer
specifications (INTEGER*2 and INTEGER*4) will always take precedence
over the setting of the /14 switch. Integer constants must lie in the
range -32768 to +32767. For example:
+22 = 00026
8
-7 = 177718

A.2

(

FLOATING-POINT FORMATS

The exponent for both 2-word and 4-word floating-point formats is
stored in excess 128
(12008) notation. Binary. exponents from -128 to
+127 are represented by the binary equivalents of 0 through 255 (0
through
377 8 ). Fractions are represented in sign-magnitude notation
with the binary radix point to the left. Numbers are assumed to be
normalized and, therefore, the most significant bit is not stored
because it is redundant (this is called -hidden bit normalization-).
This bit is assumed to be a 1 unless the exponent is 0 (corresponding
to 2- 128 ) in which case it is assumed to be O. The value 0 is
represented by two or four words of zeros. For example, +1.0 would be
represented by:
40200

in the 2-word format, or:
40200

o
o
o

A-I.

APPENDIX A. DATA REPRESENTATION
in the 4-word format.

-5 would be:

140640
0
in the 2-word format, or:
140640
0
0
0
in the 4-word format.

A.2.1

Real Format (2-Word Floating Point)

word 1:
Low-order mantissa

word 2:
15

since the high-order bit of the mantissa is always 1, it is discarded,
giving an effective precision of 24 bits, or approximately 7 digits of
accuracy. The magnitude range lies between approximately .29 X 10- 38
and .17 X 10 39 •

A.2.2

Double Precision Format (4-Word Floating Point)

word 1:
(

word 2:
word 3:
word 4:

I15

Low-order mantissa

I15
I15

Lower-order mantissa

0
0
Lowest-order mantissa
0

I
I

The effective precision is 56 bits or approximately 17 decimal digits
of accuracy.
The magnitude range lies between .29 X 10- 8 and .17 X
10 39 •

A-2

APPENDIX A. DATA REPRESENTATION

A.2.3

Complex Format

Si~n

word 1:

10B~nary excess
1=
1 8 ~xeQngnt
15 14
7

word 2:

I15

word 3:

Sir Binary excess
10=
11~8 eX2Qngnt
15 14
7

word 4:

A.3

I15

H!:h-order

m t;L&UIA
6

Real
° Part

Low-order mantissa

°
;Las
I mH~~-or:er
6

Imaginary
° Part

Low-order mantissa

LOGI'CAL*l

Data item I
7
Any non-zero value is considered to have a logical value of .TRUE.
The range of numbers from +127 to -128 can be represented in LOGICAL*l
Format. LOGICAL*l array elements are stored in adjacent bytes.

A.4

HOLLERITH FORMAT

word 1:
word 2:

char 2

char 4

char 1

]

char 3

8 7

I b1ank= 4O a
15

8 7
•
•
•

Ichar n (n~255)1
0

8 7

A-3

APPENDIX A. DATA REPRESENTATION
Hollerith constants are stored internally one charac~er per byte.
Hollerith values are padded on the right with blanks to fill the
associated data item if necessary. Hollerith constants can only be
used in DATA, FORMAT, and CALL statements. Only the quoted form of
Hollerith constants can be used in STOP and PAUSE statements.

A.5

LOGICAL FORMAT

True:

word 1

word 2

15
False:

word 1

word 2

112121212111
0

uns~cIfIed I

01010101010 I
0
unsEecified I
0

Logical (LOGlCAL*4) data items are treated as LOGICAL*l values for use
with arithmetic and logical operators. Any non-zero value in the low
order byte is considered to have a logical value of true in logical
expressions.

A.6

RADIX-50 FORMAT

Radix-50 character set
Character
ASCII

Octal
Equivalent

Radix-50
Equivalent

space
A-Z

40
101-132
44
56

0
1-32
33
34
35
36-47

•

unused
0-9

60-71

The following table provides a convenient means of translating between
the ASCII character set and its Radix-50 equivalents. For example,
given the ASCII string X2B, the Radix-50 equivalent is (arithmetic is
performed in octal):
X-113000
2=002400
B-000002
X2B=-115402

A-4

APPENDIX A. DATA. REPRESENTATION

Single Char.
or
First Char.

Second
Character

--------

A 003100
B 006200
C 011300
D 014400
E 017500
F 022600
G 025700
H 031000
I 034100
J 037200
K 042300
L 045400
M 050500
N 053600
o 056700
P 062000
Q 065100
R 070200
S 073300
T 076400
U 101500
V 104600
W 107700
x 113000
y 116100
z 121200
$ 124300
• 127400
132500
0 135600
1 140700
2 144000
3 147100
4 152200
5 155300
6 160400
7 163500
8 166600
9 171700

A 000050
B 000120
C 000170
D 000240
E 000310
F 000360
G 000430
H 000500
I 000550
J 000620
K 000670
L 000740
M 001010
N 001060
o 001130
P 001200
o 001250
R 001320
S 001370
T 001440
U 001510
V 001560
W 001630
x 001700
Y 001750
z 002020
$ 002070
• 002140
002210
0 002260
1 002330
2 002400
3 00245'0
4 002520
5 002570
6 002640
7 002710
8 002760
9 003030

A 000001
B 000002
C 000003
D 000004
E 000005
F 000006
G 000007
H 000010
I 000011
J 000012
K 000013
L 000014
M 000015
N 000016
o 000017
P 000020
Q 000021
R 000022
S 000023
T 000024
U 000025
V 000026
W 000027
x 000030
Y 000031
z 000032
$ 000033
• 000034
000035
0 000036
1 000037
2 000040
3 000041
4 000042
5 000043
6 000044
7 000045
8 000046
9 000047

------------

Third
Character

A-5

APPENDIX B
LIBRARY SUBROUTINES

-j"' ...

t_>~

B.l

LIBRARY SUBROUTINE SUMMARY

In addition to the functions intrinsic to the FORTRAN system, there
are subroutines in the FORTRAN library which the user may call in the
same manner as a user-written subroutine. These subroutines are:
ASSIGN

Allows specification at run-time of filename or device
and filename to be associated with a FORTRAN logical unit
number.

Allows the file on a specified logical unit to be closed.

DATE

Returns
a
9-byte
string
containing
representation of the current date.

the

ASCII

I DATE

Returns three integer
month, day and year.

representing

the

current

ERRSET

Allows the user to specify the
detection of certain errors.

action

ERRSNS

Allows the user to obtain information
recently detected error condition.

about

ERRTST

Allows monitoring of certain error types
execution.

during

EXIT

Terminates the execution of a program and returns control
to the RSX-I1M executive.

USEREX

Allows specification of a routine to be invoked
of program termination.

part

FDBSET

Allows specification of special
associated with a logical unit.

RADSO

Performs conversion of up to six character Hollerith
strings and returns the result as a function value.

IRADSO

Performs conversion
representation.

RSOASC

Converts Radix-50 strings to Hollerith strings.

(

values

B-1

Hollerith

I/O

options

strings

taken
the

most

program

Radix-50

APPENDIX B. SYSTEM· SUBROUTINES
RANDU,
RAN

Returns a random real number with a uniform
between 0 and 1.

SECNDS

Provides system time of day or elapsed time as a floating
point value in seconds.

TIME

Returns
an
a-byte
string
containing
the
ASCII
representation of the current time in hours, minutes and
seconds.

B.2

distribution

ASSIGN

The ASSIGN subroutine allows the association of filename information
with a logical unit number.
The ASSIGN call, if present, must be
executed before the logical unit is opened for I/O operations (by READ
or WRITE) for sequential access files, or before the associated DEFINE
FILE statement for random-access files. The device assiqnment remains
in
effect until a new CALL ASSIGN is performed.
The
filename
assignment remains in effect only until the file is closed
by
a
CALL CLOSE. The call to ASSIGN has the general form:

~I
I
I

CALL ASSIGN (n, name, icnt)
CALL ASSIGN

requires only the first argument, all others are optional,
and if omitted are replaced by the default values as noted in the
argument descriptions. However, if any argument is to be included,
all arguments that precede it must also be included.
A description of the arguments to the ASSIGN routine follows:
logical unit number expressed as an
variable, or expression.

integer

constant,

name

Hollerith or literal string containing any standard
RSX-llM device/filename specification. If the device
is not specified, then the device remains unchanged
from the d~fault assignments, or the MeR REASSIGN
command. If a filename is not specified, the default
names as described in section 3.4 are used.

ient

specifies the number of characters in the string
'name'. If 'icnt' is zero, the string 'name' will be
processed
until
the
first
null
character
is
encountered.

An argument is allowed
forms of CALL ASSIGN.

- I

after icnt to be compatible

For example, in the following program:
CALL ASSIGN (3, 'TT:')
WRITE (3,-) ••••
CALL CLOSE (3)
WRITE (3,-) ••••

B-2

with

I
~
~

APPENDIX B. SYSTEM SUBROUTINES
both WRITE operations will occur on the terminal. To cause the second
WRITE to revert back to SY:, another CALL ASSIGN must be used,
explicitly setting SY: to unit 3.

B.3

The CLOSE subroutine allows the currently open file on a logical
to be closed. The form of the call is:

unit

CALL CLOSE (n)
where n is an integer value specifying the logical unit.
When the
close is completed the buffers and FDB associated with the file are
again available for use.

B.4
The DATE subroutine can be used in a FORTRAN program to obtain the
current date as maintained within the system. The DATE subroutine is
called as follows:
CALL DATE (array)
where array is an array capable of holding a 9-~te string. The array
specification in the call may be expressed as the array name alone:
CALL DATE (a)
in which the first three elements of the real array a are used to hold
the date string, or as:
CALL DATE (a(i»
which causes the 9-byte string to begin at the i(th)
array a.

element

The date is returned as a 9-~te (9-character) string in the form:
dd-mrnm-yy
where:
dd is the 2-digit date

mmm is the 3-letter month specification
yy is the last two digits of the year

B-3

the

APPENDIX B. SYSTEM SUBROU'l'INES
For example:
l5-NOV-75
In the case where the array is a real array, 4-1/2 words are used to
contain the data string with the remaining array storage being
untouched. Therefore, the date string is stored in the first nine
bytes in the elements a(i), a(i+l), and a(i+2). The last three bytes
of a(i+2) are untouched and should be filled with blanks by the user
if he intends to print the date with a JA4 format.

B.5

I DATE

IDATE returns three integer values representing
day, and year. The call has the form:

the

current

month,

integer

CALL IDATE (i, j, k)
If the current date were March 19, 1975
variables upon return would be:

the

values

the

i = 3
j
19

k = 75

B.6

ERRSET

The ERRSET subroutine allows specification of the action to be taken
when an error is detected by the OTS. The error action to be taken is
specified individually for each error, independent of other errors.
The general form of the call is:
CALL ERRSET (number, contin, count, type, log, maxlim)
where
number

is an integer value specifying the error number to
which the following parameters apply.

con tin

is a Logical value specifying whether or not to
continue after an error. .TRUE. means continue,
.FALSE. means exit if this error occurs.

count

is a Logical value specifying whether to count
this error against the task maximum error limit •
• TRUE. means count, • FALSE.
means don • t count
it.

type

is a logical value specifying the
type
of
continuation to perform.
• TRUE. means that an
ERR a transfer is to be taken if available.
If
this action is indicated and an ERR - keyword is
not specified in the I/O statement, then the task
will exit when the error occurs. .FALSE. means

B-4

APPENDIX B. SYSTEM SUBROUTINES
return to the routine that detected the error
default error recovery.
log

is a logical value specifying whether to produce
an error message for this error. .TRUE. means
produce message, .FALSE. means don't produce a
message.

maxlim

is a positive integer value used to set the task's
maximum error limit. The default value is set to
31 at task initialization.

Consult section C.2, OTS Error Processing, for a complete
of the allowed values and meanings of these arguments.

description

Null arguments are permitted for all but the first argument and
no change in the current state of that control code.

B.7

for

cause

ERRSNS

The ERRSNS subroutine allows the user to obtain information about
most recent error that has occurred during program execution.
general form of the call is:

the
The

CALL ERRSNS (nurn, fcserr, fcserl, iunit)
where
is an INTEGER*2 variable or array element into which
will be stored the most recent error number.

num

(A zero will be returned if no error has occurred.)
If the last error occurred as a result of an error indication from
File Control Services, then the next three parameters will receive
selected values from the file descriptor block (FOB). Otherwise,
values of zero will be returned. Consult the I/O Operations Reference
Manual (DEC-ll-OXFSA-A-D) for definitions of the interpretation of the
FCS error return codes.
fcserr

is an INTEGER*2 variable or array element
will be stored the F.ERR field of the FOB.

into

which

fcserl

is an INTEGER*2 variable or array element into
will be stored the F.ERR+l field of the FOB

which

iunit

is an INTEGER*2 variable or array element
will be stored the logical unit number.

which

into

Specifying zero
From zero to four arguments may be specified.
arguments serves to clear ERRSNS from previous calls. To determine if
an error occurs in a given section of a program, the fol~owing
technique is suggested;
1.

Call ERRSNS immediately prior to
clear any previous error data:

Execute the section

B-5

the

segment

order

APPENDIX B. SYSTEM SUBROUTINES
3.

Call ERRSNS again and branch on a non-zero argument to
analysis code.

error

For example:
CALL ERRSNS
CALL ASSIGN (1, 'NAME.DAT')
CALL FDBSET (l,'OLD','SHARE')
CALL ERRSNS (IERR)
IF (IERR.NE.O) GO TO 100

B.8

ERRTST

The ERRTST subroutine allows the user program to monitor the types of
errors detected during program execution. The call is of the form:
CALL ERRTST (i, j)
where i is the error number and the value of j is returned as:
j=l if an error number i has occurred
j=2 if an error number i has not occurred
The sequence:
CALL ERRTST (43,J)
GO TO (lO,20),J
CONTINUE

transfers control to statement 10 if an error 43 has occurred.
section C.2.3. for a discussion of error codes and messages.

See

The ERRTST routine also resets to 0 the error flag for that
class (but not the error count used by ERRSET). For example:

error

CALL ERRTST (I,J)
CALL ERRTST (I,J) •
The second call is guaranteed to return J-2. The ERRTST subroutine is
independent of the ERRSET subroutine; neither directly influences the
other except that ERRSET can cause execution to terminate.

B.9

EXIT

A call to the EXIT subroutine, in the form:
CALL EXIT
terminates execution. It causes
issuing task to be terminated.

all

B-6

files

closed

and

the

(

APPENDIX B. SYSTEM SUBROUTINES
B.lO

USEREX

USEREX is a subroutine which allows specification of a routine to be
invoked as part of program termination.
This allows clean up
operations in non-FORTRAN routines. The form of the subroutine call
is:
CALL USEREX (name)
Where 'name' is the routine which will be called at exit time and
should appear in an EXTERNAL statement somewhere in the program unit.
Control is transferred with a JSR PC,name instruction after all
procedures
required
for FORTRAN program termination have been
completed. The transfer of control takes place instead of the normal
exit to the RSX-llM executive. The routine specified by USEREX may
perform the exit procedure itself or return with an RTS PC.

Boll

FDBSET

The FDBSET subroutine permits specification of special input/output
options available through File Control Services under RSX-llM.
CALL FDBSET (unit, mode, share, numbuf, initsz, extend)
where

(

unit

is an integer value specifying the logical
unit to which the subsequent arguments apply

mode

is one of the following literals specifying
the type of access to be used (only the first
letter is checked for validity):
'READONLY'
'NEW'
'OLD'
'APPEND'
'UNKNOWN'

'MODIFY'
'ISUP'

For read only access
For creating a new file
For accessing an existing file
For appending to an existing
file
(meaningful)
for
sequential files only)
When it is not known whether a
file
exists:
this option
searches for it, opens it if
it already exists, or creates
the file if it does not.
Open an old file for updating.
The file can not be extended.
Open a new file (for output)
but inhibit superseding of the
previous version.
This file
will
replace
the
current
version.

If this argument is omitted the default is
determined
by
the
first I/O operation
performed on that unit. If a WRITE operation
is the first I/O operation performed on that
unit, 'NEW' is assumed. If a READ operation
is the first I/O operation performed on that
unit, 'OLD' is assumed.
share

is the literal 'SHARE' indicating
shared access bit should be set.
B-7

the

FCS

APPENDIX B. SYSTEM SUBROUTINES
numbuf*

is an integer value indicating the number of
internal buffers to be used for multibuffered
input/output. If this argument is omitted,
one internal buffer is used.
The ACTFIL
option in the Task Builder command sequence
is used to extend the buffer area.

initsz

is an integer value indicating the initial
allocation, in disk blocks, of a new file.

extend

is an integer value specifying the number
blocks by which to extend a file.

FDBSET may only be called prior to opening the unit specified in the
first
argument.
CALL FDBSET, CALL ASSIGN, and the DEFINEFILE
statement may be used together.
The unit number argument is required. All other arguments may be null
or missing to indicate no specification for that argument.

B.12

RADSO

The RADSO function subprogram provides a simplified way to encode
RSX-ILM task names in Radix-50 notation. The form of the call is
RADSO (name)
where
name

is the variable name or array element corresponding to
an ASCII string.

Note that the RADSO function may be used as an argument to an
system directive sUbroutine, e.g.,

RSX-llM

DATA A/'JOE'/
CALL REQUES (RADSO(A), ••• )
THE RADSO function is equivalent to
subprogram:

the

following

FORTRAN

function

FUNCTION RADSO(A)
CALL IRADSO (6,A,RADSO)
RETURN
END

B.13

IRADSO

The IRADSO subprogram performs conversions between Hollerith (text)
strings and Radix-50 representation.
Radix-50 representation is

*Multi-buffering is not implemented in the current release of RSX-llM.
It will be included in a subsequent release.

B-8

APPENDIX B. SYSTEM SUBROUTINES
required by the ISA Process Control and System Directives Subroutines
for the specification of task names within the RSX-llM system. (See
the PDP-II FORTRAN Language Manual for details of the Radix-50
representation.)
lRADSO may be called as a FUNCTION subprogram if the return value is
desired, or as a SUBROUTINE subprogram if no return value is desired.
The form of the call is
n a lRADSO (icnt, input, output)
or
CALL lRADSO (icnt, input, output)
where
icnt

is the (integer) maximum number of
convert.

input

is an ASCII (Hollerith)
converted to Radix-50.

text

string

output

is the location for storing
conversion.

the

results

the

is the number of characters actually converted.

characters

Three characters of text are packed into each word of output.
The
number of output words modified is computed by the expression (in
integer mode)
(ICNT+2) /3
Thus if a count of four is specified, two words of output will be
written even if only a l-character input string is given as an
argument.
Scanning of input characters will terminate on the first
character encountered in the input string.

B.14

non-radix-SO

RSOASC

The RSOASC subprogram provides decoding of Radix-50
into ASCII strings. The form of the call is:

encoded

values

CALL RSOASC (icnt, input, output)
where
of

is the number
converted.

input

is the variable or array containing the encoded
input.
Note that (icnt+2)/3 words will be read
for conversion.

B-9

output

characters

icnt

APPENDIX B. SYSTEM SUBROUTINES
output

is the variable or array
into
characters (bytes) will be placed.

which

icnt

If the undefined Radix-50 code is detected, or the Radix-50
exceeds the maximum value 174777 (octal) then question mark(s)
will be placed in the output.

(ft?ft)·

B.15

word

RANDU,RAN

The random number generator can be called as a subroutine, RANDU, or
as an intrinsic function, RAN. The subroutine call is performed as
follows:

CALL RANDU (iI, i2, x)
where il and i2 are previously defined integer variables and x is the
real variable name in which is returned a random number between 0 and
1. il and i2 should be initially set to O. II and i2 are updated to a
new generator base during each call. Resetting il and i2 to 0 repeats
the random number sequence. The values of il and i2 have a special
form;
only 0 or saved values of il and i2 should be stored in these
variables.
Use of the random number subroutines is similar to the use of the
function where:

RAN

x = RAN (iI, i2)
is the functional form of the random number generator.

B.16

(

SECNDS

The SECNDS FUNCTION subprogram returns the system time as a single
precision floating point value less the value of its single argument.
For example:
TIM=SECNDS (0.)
will return the number of seconds since midnight, that is, the current
time of day. It may be called with a non-zero argument for performing
elapsed time computations as in
C

START OF TIMED SEQUENCE
Tl
SECNDS(O.)
II:

C
C
C

CODE TO BE TIMED
DELTA = SECNDS(Tl)

where DELTA will give the elapsed time.
B-lO

APPENDIX B. SYSTEM SUBROUTINES
The value of SECNDS is accurate to not less than the resolution of the
system clock: 0.0166 ••• seconds for a 60-cycle clock and 0.02 seconds
for a 50-cycle clock. Systems with a programmable clock have resolution not less than the clock tick as set at SYSGEN.
With 24 bits of precision for real
accurate to the clock tick for
duration.

B.17

values, this representation is
values up to about two days in

TIME

The TIME subroutine allows the user to access the current system
as an ASCII String. Its form is as follows:

time

CALL TIME (a)
The TIME call returns the time as an
colons) ASCII string of the form:

8-byte

(8-character,

including

hh:mm:ss
where
hh is the two-digit hour indication
mm is the two-digit minute indication
ss is the two-digit second indication
For example:
10:45:23
A 24-hour clock is used.
format specification.

This quantity is typically output with a 2A4

B-ll

(

APPENDIX C
FORTRAN ERROR DIAGNOSTICS

C.l

COMPILER ERROR DIAGNOSTICS

The RSX-llM FORTRAN Compiler, while reading and processing the FORTRAN
source program, can detect syntax errors (or errors in general form)
such as unmatched parentheses, illegal characters, unrecognizable key
words, missing or illegal statement parameters.
The error diagnosti'cs are generally clear in specifying the exact
nature of the error. In most cases, a check of the general form of
the statement in question as described in the PDP-II FORTRAN Language
Reference Manual will help determine the location of the error.
Some of the most common causes of syntax errors, however, are typing
mistakes.
A typing mistake can sometimes cause the Compiler to give
very misleading error diagnostics. The user should be careful of the
following common typing mistakesl
1.

Missing commas or parentheses in a complicated expression
FORMAT Statement.

Misspelling of particular instances of variable names.
If
the Compiler does not detect this error (it usually cannot),
execution may also be affected.

If the user terminal does not clearly differentiate between 0
(zero) and the letter 0, what appear to be identical
spellings of variable names may not appear so to the
Compiler, and what appears to be a constant expression may
not appear so to the Compiler.

(
\.

inadvertent line continuation signal on the line following
the statement in error.

If any errors were detected in a compilation, the message:
ERRORS DETECTED: n
will be printed on the initiating terminal;
n is the
errors, not including warnings, detected by the compiler.

number

The next three sections describe the initial phase and secondary phase
error diagnostics and the fatal FORTRAN Compiler error diagnostics.

C-l

APPENDIX C. ERROR DIAGNOSTICS
The following is an example of
issued by the Compiler.

FORTRAN IV
0001
0002
0003

0004
0005

0006

.* •• * M
0007

.* •• *

C
E

FORTRAN IV

FORTRAN

program

with

diagnostics

SOURCf LISTING

PAGE 001

DOUBLE P~!CIStON OBlE D8l£2
DATA INT/100/0BLE/500.1
OBLE2 • INT/2. 5 •• 0SlE
WRITE,(6,10) DeLE,oelE
FORMAT(lX,2F12.6)
.
STOP

MISSING COMMA
MISSING COMMA

lABEL DEFINEO PREVIOUSLY

INTEGER INT
END

NOT STANDARD ORDERING
NO END STATEMENT

M01-01

EXTRA COMMA

DIAGNOSTICS

tN lINE 0002 MSG *030

lINE 0004 MSG *049
[ ~'RNING] MSG *091
[ WARNING J MSG *093

EXTRA CHARACTERS AT END OF STATEMENT
SYNTAX ERROR
VA~I'BlE "DBlEOB" NAME EXCEEOS 6 C~'RACTERS
NON-STANDARD STATEMENT ORDERING

C-2

APPENDIX C. ERROR DIAGNOSTICS
C.l.l

Errors Reported by the Initial Phase of the Compiler

Some of the easily recognizable FORTRAN syntax errors are detected by
the initial phase of the Compiler. These errors are reported in the
source program listing. They are not reported if a source listing is
not requested.
The error diagnostics are printed after the source statement to which
they apply (the L error diagnostic is an exception). The general form
of the diagnostic is as follows:

***** c
Where c i2 a code letter whose meaning is described below:
INITIAL PHASE ERROR DIAGNOSTICS
Code Letter

Description

Columns 1-5 of continuation line are not blank.
Columns 1-5 of a continuation line must be blank
except for a possible '0' in column 1.

Illegal
continuation.
Comments
cannot
be
continued and the first line of any program unit
cannot be a continuation line.

Missing END statement.
An
supplied
by
the Compiler
encountered.

Hollerith string or quoted literal string longer
than 255 characters or longer than the remainder
of the statement.

Non-FORTRAN character used. The line contains a
character that is not in the FORTRAN character set
and is not used in a Hollerith string or comment
line.

Illegal statement label
definition.
Illegal
(non-numeric) character in statement label.

Line too long to print. There are more than 80
characters (including spaces and tabs) in a line.
Note: this diagnostic is issued preceding the
line containing too many characters.

Multiply defined label.

Statement contains unbalanced parentheses.

Syntax error.
Statement not
form.

Statement could not
FORTRAN statement.

END statement
if end-of-file

is
is

Multiple
equal
signs,
etc.
of the general FORTRAN statement

C-3

identified

legal

APPENDIX C. ERROR DIAGNOSTICS
C.l.2

Errors Reported by Secondary Phases of the Compiler

Those Compiler error diagnostics not reported by the initial phase of
the Compiler will appear immediately after the source listing and
immediately before the storage map.
Since the diagnostics appear
after the entire source program has been listed, they must reference
the statement to which they apply by using the internal sequence
numbers assigned by the Compiler.
The general form of the diagnostic is:
IN LINE nnnn'MSGlm text
Where nnnn is the internal sequence number of the statement in
question, m is an integer constant specifying the error number, and
text is a short description of the error.
Below, listed alphabetically, are the error diagnostics.
Included
with each diagnostic is a brief explanation. Refer to the PDP-ll
FORTRAN Language Reference Manual for information to help correct the
error.
The notation **** signifies that a particular variable
statement label will appear at that place in the text.

name

SECONDARY PHASE ERROR DIAGNOSTICS
ADJUSTABLE DIMENSIONS ILLEGAL FOR ARRAY ****
All arrays must be dimensioned with integer constants
except as specified in the Language Reference Manual.
ARRAY **** HAS TOO MANY DIMENSIONS
An array can have up to seven dimensions.
ATTEMPT TO EXTEND COMMON BACKWARDS
While attempting to equivalence arrays in COMMON, an
attempt was made to extend COMMON past the recognized
beginning of COMMON storage.
COMMON BLOCK EXCEEDS MAXIMUM SIZE
An attempt was made to allocate more space to
than is physically addressable ()32k words).
DANGLING OPERATOR
An operator (+,-,*,/,
Example: I=J+

etc.)

missing

COMMON

operand.

DEFECTIVE DOTTED KEYWORD
A dotted relational operator was not recognized.
possible misuse of decimal point.

Also,

DO TERMINATOR **** PRECEDES DO STATEMENT
The statement specified as the terminator of a DO
must come after the DO statement.

loop

EXPECTING LEFT PARENTHESES AFTER ****
An array name or Function name
followed by a left parenthesis.

not

C-4

reference

APPENDIX C. ERROR DIAGNOSTICS
EXTRA CHARACTERS AT END OF STATEMENT
All the necessary information for a syntactically
correct FORTRAN statement has been found on this line,
but more information
exists.
Possibly
due
to
inadvertent continuation signal on next line, or a
missing comma.
FLOATING CONSTANT TOO SMALL
A floating constant in an expression is too close to
zero to be represented in the internal format. Use
zero if possible.
ILLEGAL ADJACENT OPERATOR
Two operators (*,/, logical operators, etc.)
are
illegally placed next to each other. Example I I/*J.
ILLEGAL ELEMENT IN I/O LIST
An item, expression, or implied DO specifier in an
list is of illegal syntax.

I/O

ILLEGAL DO TERMINATOR STATEMENT ****
A DO statement terminator must not be a GO TO,
arithmetic IF, RETURN, or DO statement or logical IF
containing one of these statements.
ILLEGAL STATEMENT ON LOGICAL IF
The statement contained in a logical
another logical IF or DO statement.

ILLEGAL TYPE FOR OPERATOR
An illegal variable type has been
exponentiation or logical operator.

must

used

not

with

ILLEGAL USAGE OF OR MISSING LEFT PARENTHESIS
A left parenthesis was required but not found, or a
variable reference or constant is illegally followed by
a left parenthesis.
INTEGER OVERFLOW
An integer constant or expression value must
outside the range -32767 to +32767.

not

fall

INVALID COMPLEX CONSTANT
A complex constant has been improperly formed.
INVALID DIMENSIONS FOR ARRAY
An attempt was made while dimensioning an array
explicitly specify zero as one of the dimensions.

INVALID DO TERMINATOR ORDERING AT LABEL ****
Do loops are incorrectly nested.
INVALID EQUIVALENCE
Illegal
equivalence,
or
equivalence
contradictory to a previous equivalence.
INVALID FORMAT SPECIFIER
A format specifier is not
statement or an array name.

c-s

the

label

that

FORMAT

APPENDIX C. ERROR DIAGNOSTICS
INVALID IMPLICIT RANGE SPECIFIER
Illegal implicit range specifier, i.e., non-alphabetic
specifier, or specifier range is in reverse alphabetic
order.
INVALID LOGICAL UNIT
A logical unit reference must be an integer variable or
constant in the range 1 to 99.
INVALID OCTAL CONSTANT
An octal constant is too
other than 0-7.

large

contains

digit

INVALID OPTIONAL LENGTH SPECIFIER
A data type declaration optional length specifier is
illegal. For example, REAL*4 and REAL*8 are legal, but
REAL*6 is not.
INVALID RADIXSO CONSTANT
Illegal character detected in a RADIXSO constant.
INVALID RECORD FORMAT
The third parenthetical argument in a
statement must be the single character U.

DEFINE

FILE

INVALID STATEMENT IN BLOCK DATA
It is illegal to have any executable
statements in a BLOCK DATA Subprogram.

FORMAT

INVALID STATEMENT LABEL REFERENCE
Reference has been made to a statement number that is
of illegal construction. GO TO 999999 is illegal since
the statement number is too long.
INVALID SUBROUTINE OR FUNCTION NAME
A name used in a CALL statement or function reference
is not valid. Example: use of an array name in a CALL
statement routine name reference.
INVALID TARGET FOR ASSIGNMENT
The left side of an arithmetic assignment statement
not a variable name or array element reference.

INVALID TYPE SPECIFIER
An unrecognizable data type was used.
INVALID USAGE OF FUNCTION OR SUBROUTINE NAME
A function name cannot appear in a
DATA, OR EQUIVALENCE statement.

DIMENSION,

COMMON,

INVALID VARIABLE NAME
A variable name contains an illegal character.
LABEL ON DECLARATIVE STATEMENT
It is illegal to place
statement.

label

declarative

MISSING ASSIGNMENT OPERATOR
The first operator seen in an arithmetic assignment
statement was not an equal sign (=). Example: I+J=K.

C-6

APPENDIX C. ERROR DIAGNOSTICS
MISSING COMMA
The conuna delimiter was expected but was not found.
See the section of the FORTRAN Reference Manual that
describes the general form of the
statement
in
question.
MISSING DELIMITER IN EXPRESSION
Two operands have been placed next to each other in
expression with no operator between them.

MISSING LABEL
Expecting a statement label but one was not found.
Example:
ASSIGN J TO I. A valid statement label
reference should precede 'TO' but does not.
MISSING RIGHT PARENTHESIS
Expecting a right parenthesis but one was not found.
Example:
READ(5,100,). The first non-blank character
after the format reference
should
be
a
right
parenthesis but is not.
MISSING QUOTATION MARK
In a FIND statement, the logical unit number and record
number must be separated by a single quotation mark.
MISSING VARIABLE
Expecting a variable, but one was not found.
Example:
ASSIGN 100 TO 1. A variable name should follow the 'TO'
but one does not.
MISSING VARIABLE OR CONSTANT
Looking for an operand (variable or constant) but found
a
delimiter
(comma, parenthesis, etc.). Example:
WRITE(). A unit number should
follow
the
open
parenthesis, but a delimiter (close parenthesis) is
encountered instead.
MODES OF VARIABLE **** AND DATA ITEM DIFFER
The data type of each variable and its associated
list item must agree in a DATA Statement.

data

MULTIPLE DECLARATION FOR VARIABLE ****
A variable cannot appear in more than one
declaration
statement
or
dimensioning
Subsequent declarations are ignored.

data type
statement.

NUMBER IN FORMAT STATEMENT NOT IN RANGE
An integer constant in a FORMAT
than 255 or is zero.

greater

been

nested

statement

PARENTHESES NESTED TOO DEEPLY
Group repeats in a FORMAT statement
too deeply.

have

P-SCALE FACTOR NOT IN RANGE -127 TO +127
P-scale factors must fall in the range -127 to +127.
REFERENCE TO INCORRECT TYPE OF LABEL ****
A statement label reference that should be a label on a
FORMAT statement is not such a label, or a statement
label reference that should be a label on an executable
statement is not such a label.
C-7

REFERENCE TO UNDEFINED STATEMENT LABEL
A reference has been made to a statement number that
has not been defined anywhere in the program unit.
STATEMENT MUST BE UNLABELED
A DATA, SUBROUTINE, FUNCTION, BLOCK DATA, arithmetic
statement function definition, or declarative statement
must not be labeled.
STATEMENT TOO COMPLEX
An arithmetic statement function has more than 10 dummy
arguments.
Or the statement is too long to compile.
Break it up into 2 or more smaller statements.
SUBROUTINE OR FUNCTION STATEMENT MUST BE FIRST
A SUBROUTINE, FUNCTION or BLOCK DATA Statement, if
present, must be the first statement in a program unit.
SYNTAX ERROR
Check the general form of the statement with the
general form outlined in the Language Reference Manual
section that describes that type of statement.
TARGET MUST BE ARRAY
The third argument in an
must be an array name.

ENCODE

DECODE

statement

SYNTAX ERROR IN INTEGER OR FLOATING CONSTANT
An integer or floating constant has been incorrectly
formed.
For example, 1.23.4 is an illegal floating
constant because it contains two decimal points.
UNLABELED FORMAT STATEMENT
All FORMAT Statements must be labeled.
USAGE OF VARIABLE **** INVALID
An attempt was made to EXT!RNAL a common variable, an
array variable, or a dummy argument. Or an attempt was
made to place in COMMON a dummy argument or external
name.
VARIABLE **** INVALID IN ADJUSTABLE DIMENSION
A variable used as an adjustable dimension must
integer dummy argument in the subprogram unit.

WRONG NUMBER OF SUBSCRIPTS FOR ARRAY ****
An array reference does not have the same number of
subscripts as specified when the array was dimensioned.

C.l.3

Warning Diagnostics

Warning diagnostics report conditions which are not true error
conditions, but which may be potentially dangerous at execution time,
or which may present compatibility problems with FORTRAN Compilers
running on other PDP-ll Operating Systems. The warning diagnostics
are normally enabled, but may be suppressed by use of the /-WR

C-8

(

APPENDIX C. ERROR DIAGNOSTICS
Compiler switch.
The form and placement of the warning diagnostics
are the same as those for the secondary phase error diagnostics (see
section C.l.2) except that the line number reference is replaced with
"WARNING". A listing of the warning diagnostics follows:
ADJUSTABLE DIMENSIONS ILLEGAL FOR ARRAY ****
Adjustable arrays must be a dummy argument in a
subprogram, and the adjustable dimensions must be
integer dummy arguments in the
subprogram.
Any
variation from this rule will cause a dimension of 1 to
be used and this warning message to be issued.
STATEMENT ORDERING
Although
the
RSX-llM
FORTRAN IV
Compiler
has
less-restrictive statement ordering requirements than
those outlined in chapter 7 of the PDP-ll FORTRAN
Language
Reference
Manual,
non-adherence to the
stricter requirements may cause error conditions on
other FORTRAN Compilers.
See section 3.5 of this
document.

NON~STANDARD

VARIABLE **** IS NOT WORD ALIGNED
Placing a non-LOGICAL*l variable or array after a
LOGICAL*l variable or array in COMMON or equivalencing
non-LOGICAL*l variables
or
arrays
to
LOGICAL*l
variables or arrays may cause this condition. An
attempt to reference the variable at runtime will cause
an error condition.
VARIABLE **** NAME EXCEEDS SIX CHARACTERS
A variable name of more than six characters was
specified.
The first six characters were used as the
true variable name. Other FORTRAN Compilers may treat
this as an error condition. See section 3.1 of this
document.

(

C.l.4

Fatal Compiler Error Diagnostics

Listed below are the fatal Compiler error diagnostics.
These
diagnostics, which are sent directly to the initiating terminal,
report hardware error conditions, conditions which may
require
rewriting of the source program, and conditions which may require
attention from DEC Software Support. The form of the diagnostic is:
FATAL ERROR n
where n is an error code having one of the following values:
Code
C

Meaning
Constant subscript overflow.
Too
many
subscripts have been employed in a statement.

constant

SOLUTION - simplify the statement
L

More than 80 characters in input record.
SOLUTION - simplify
lines.

statement

C-9

use

continuation

APPENDIX C. ERROR DIAGNOSTICS

Unrecoverable error occurred while the Compiler was
writing the object file (.OBJ). Possibly, insufficient
output file space.
SOLUTION - rectify hardware problem, or make more space
available for output by deleting unnecessary files.

Optimizer push down overflow - statement too complex,
or too many common subexpressions occurred in one basic
block of a program.
SOLUTION - simplify complex statements;
report the
error to your local software support representative.

Unrecoverable hardware error
occurred
Compiler was reading ,source file.

while

the

SOLUTION - rectify hardware problem.
S

Subexpression stack overflow - statement too complex.
SOLUTION - simplify complex statements.

Memory Overflow
SOLUTION - break up program into subprograms or compile
in a larger partition.

Unrecoverable error occurred while the Compiler was
writing listing file. Possibly, listing file space is
not large enough.
SOLUTION - rectify hardware problem, or make more space
available for listing file by deleting unnecessary
files.

Code generation stack overflow - statement too complex.
SOLUTION - simplify complex statements.

Compiler error
SOLUTION - report this error to your local
support
representative.
Please
include
listing.

C.2

C.2.l

software
program

OBJECT TIME SYSTEM ERROR DIAGNOSTICS

Error Processing Algorithm

The Object Time System detects many Input/Output, arithmetic, invalid
argument and other kinds of errors and reports them on the user's
terminal via logical device TI:. The action taken for each error is
determined by an error control table within the OTS. (This table may
be modified during program execution by means of
the
ERRSET
subroutine, see section B.S.)

C-10

(
\

APPENDIX C. ERROR DIAGNOSTICS
Error processing for each error is
Significant bits are as followsz

controlled

control

byte.

Continuation Bit

If not set, this bit directs the task to
exit as a result of this error. If set,
the task will continue provided certain
other conditions are met.

Count Bit

If set, this error is counted against
the task error count limit. If that
limit is exceeded, the task will exit.

Continuation Type Bit

Two types
possible:
1)

continuation

action

are

Return to routine that reported the
error to take appropriate recovery
action and proceed, or
Take an ERR- transfer in an I/O
statement. If an ERR = transfer is
specified for the error and none
was included in the Input/Output
statement, the task will exit.

The above three conditions must all
continue.
Log Bit

satisfied

for

the

task

If the task continues, then the log bit
is tested. If the bit is set, an error
message is produced before continuing;
otherwise, the task continues.

If any of the above conditions is not satisfied, the task will exit
an error message will always be produced. In this case, the
additional text "EXITING DUE TO" is included in the error message so
that it is clear why a task is abnormally terminating.
~d

Two additional bits are of interest
acceptability of ERRSET arguments.
Return permitted bit
ERRa permitted bit

here

since

they

control

the

If set, then the continuatio~ type bit
may be set by ERRSET to specify return.
If set, then the contination type bit
may be set by ERRSET to specify that an
ERR- transfer is to occur.

These two bits are used by ERRSET to check the validity of ERRSET
arquments.
At least one of these must be set in order to set the
continuation bit. Also the continuation type argument is checked
against these bits for acceptability.
All four combinations of these two bits occur
most errors are in one of two groups.
1)
2)

the

OTS,

although

I/O errors generally permit ERR- continuation type but not
return continuation.
Most other errors per.mit return continuation but not ERRa
transfer
continuation
(even if they occur during I/O
statement processing.)

C-ll

APPENDIX Cs ERROR DIAGNOSTICS
Notable exceptions are the synchronous system trap errors (numbers 3
through 10) and recursive Input/Output error (number 40) which will
always result in task termination, and the Input and Output Formatted
Conversion Errors (numbers 63 and 64) which allow both types of
continuation.
The initial setting of the error control bits is shown
error messages in section C.2.3.

C.2.2

together

with

Object Time System Error Message Format

OTS error message
formatted as follows:

consists

several

lines

information

tsknarn
[EXISTING DUE TO] ERROR number
text
[AT PC = address]
[FCS: f.err f.errl filename unit]
IN
xxxxxx AT yyy
FROM xxxxxx AT yyy
•••
FROM • MAIN. AT yyy
(In the above message prototype, fixed parts of the message are
in capital letters and variable parts in lower case letters.)

shown

The variable parts of the message are:
tsknam

-the name of the task in which the error occurred.

number

-the error number

text

-a one-line description of the error.

(

If the OTS error resulted from one of the synchronous system traps,
then the program counter will be shown in the line "AT PC =". This
line is only produced for errors numbered 5 through 12.
If the OTS error resulted from an error reported to it by File Control
Services, the line beginning "FCS:" will be included. Consult the I/O
Operations Reference Manual for a description of the FCS error codes.
the value of the F.ERR
Descriptor Block (FDB).

f.errl

the value of the F.ERR+l field of the FDB.

filename

the name
version)

unit

the logical unit on which this error occurred.

the

file

field

(not

the

f.err

including

type

File

Next follows a traceback of the subprogram calling nest at the time of
the error.
Each line represents one level of subprogram call and
shows
xxxxxx

the name of the subprogram.

C-12

(

APPENDIX C. ERROR DIAGNOSTICS
The name of the main program is shown as .MAIN.
The name of a subprogram is the same as the name
used in the SUBROUTINE or FUNCTION statement.
Arithmetic
statement
functions,
OTS
system
routines and routines written in assembly language
will not be shown in the traceback.
the internal sequence number of the subprogram at
which the error, call statement, or function
reference occurred.
A question mark, "?", instead of
a
number
indicates that the subprogram was compiled with
the /-SN compiler switch (suppress sequence number
accounting) in effect and hence the line number is
not known for that program unit.

C.2.2.1 Short Message File - In order to save space, a short message
file may be included which prints only the error number. This may be
used by specifying the module $SHORT at task build time; e.g.,
>TKB
TKB>MAIN=[l,l]SYSLIB/LB:$SHORT,MAIN

C.2.3

Object Time System Error Codes

C.2.3.1 Initial Control Bit Settings - The following table shows the
initial settings of the significant bits in the error control byte as
described in section C.2.l.

(

C-13

APPENDIX C. ERROR DIAGNOSTICS

Table C-l
Error Control Bit Settings

ERROR
NUMBER

CONTINUE?

COUNT?

CONTINUE
TYPE

L.OG?

PERMITTED
ERRa? RETURN?

NO
NO
NO
NO
NO
NO
NO
NO
NO
NO

FATAL
FATAL
FATAL
FATAL.
FATAL,
FATAL.
FATAL
FATAL
FATAL
FATAL

YES
YES
YES
VES
VES
YES
YES
YES
YES
YES

NO
NO
NO
NO
NO
NO
NO
NO
NO
NO

NO
NO
NQ
NO
NO
NO
NO
NO
NO
NO

YES
VES
VES
YES
YES
YES
YES
VES
YES
VES
YES
YES
VES
YES
VES

YES
YES
YES
YES
YES
YES
YES
YES
YES
YES
VES
YES
YES
NO
YES

ERRERR=
ERR.
ERR.
ERRa
ERR.
ERRERR.
ERR=
ERRERR_
ERR=
ERR.
RETURN
ERRD

VES
VES
YES
YES
YES
VES
YES
VES
YES
YES
YES
YES
YES
YES
YES

YES
VES
YES
YES
YES
YES
YES
YES
YES
YES
VES
YES
YES
NO
YES

NO
NO
NO
NO
NO
NO
NO
NO
NO
NO

YES
YES
YES
NO
VES
VES
VES
YES

YES
VES
VES
NO
YES
YES
YES
YES

ERR.
ERRERR.
FATAL.
ERR=
ERR=
RETURN
ERR.

YES
YES
YES
YES
YES
VES
YES
YES

YES

2
3

4
5

ft
7
8
q

10
20
21
22
23

24
25
2ft
27
28
2q
30

31
32
33
34
37
38
3q
40
41

42
43
44

C-14

YES
YES
NO
YEI
YES
NO
YEI

NO
NO
¥€S
NO
NO
NO
NO
NO

NO
NG
YES
NO

APPENDIX C. ERROR DIAGNOSTICS

Table C-l (Cont.)
Error Control Bit Settings

ERROR
NUMBER

CONTINUE?

COUNT?

CONTINUE
TYPE

1.0G?

P(RMITTEO
ERR:? RETURN?

60
61
6c
63
64
65
66
67

YES
YES
YES
YES
YES
YES
YES
YES

YES
YES
YES
NO
YES
YES
YES
YES

ERRc
ERRc
ERR:
RETURN
ERR=
ERRc
ERRERRI:

YES
YES
YES
NO
YES
YES
YES
YES

YES
YES
YES
YES
YES
YES
YES
YES

NO
YES
NO
YES
YES
NO
NO
NO

70
71
72
73
74

YES
YES
YES
YES
YES

YES
YES
YES
YES
NO

RETURN
RETURN
RETURN
RETURN
RETURN

YES
YES
YES
YES
NO

NO
NO
NO
NO
NO

YES
YES
YES
YES
YES

80
81
8e
83
84
85
86

YES
YES
YES
YES
YES
YES
YES

RETURN
RETURN
RETURN
RETURN
RETURN
RETURN
RETURN

YES
YES
YES
YES
YES
YES
YES

NO
NO
NO
NO
NO
NO
NO

YES
YES
YES
YES
YES
YES
YES

q0
CJl

NO
YES

NO
NO

FATAl.
RETURN

YES
NO

NO
NO

NO
YES

100
101

NO
NO

FATAl.
FATAl.

YES
YES

NO
NO

C.2.3.2 Error Messages
Group 0 - Severe Errors
These messages result from severe error conditions for which no error
recovery is possible. Consult the RSX-llM Executive Reference Manual
for details of what error conditions will cause traps to the System
Synchronous Trap Table entries cited below.

C-lS

APPENDIX C. ERROR DIAGNOSTICS
1

INVALID ERROR CALL
A TRAP instruction has been executed whose low
byte is within the range used by the OTS for error
reporting (see Section C.2.4) but for which no
error condition is defined.

TASK INITIALIZATION FAILURE
Task start up has failed for one of the
reasons:

following

The directive to initialize synchronous system
trap handling (SVTK$S) has returned an error
indication.

The executive directive to
asynchronous trap (SFPA$S)
error indication.

The File Control Services initialization call
(FINIT$) has returned an error indication.

enable the FPP
has returned an

ODD ADDRESS TRAP (SST 0)

SEGMENT FAULT (SST 1)
This is most likely due to a subscript
of range on an array reference.

value

out

T-BIT OR BPT TRAP (SST 2)

lOT TRAP (SST 3)

RESERVED INSTRUCTION (SST 4)
The program has attempted to execute an illegal
instruction.
This may be caused by task building
with the wrong FORTRAN library for the given
hardware configuration.
Hardware may have been
linked.

NON-RSX EMT (SST 5)
The program has executed an EMT instruction whose
low byte is not in the range used by the RSX-llM
executive.

TRAP INSTRUCTION TRAP (SST 6)
A trap instruction has been executed whose low
byte is outside the range used for OTS error
messages (see C.2.4 below).

PDPll/40 FIS TRAP (SST 7)
A module using FIS was linked
FORTRAN library.

(

with

non-FIS

FPP HARDWARE FAULT
The FPP Floating Exception Code (FEC) register
contained the value 0 following an FPP interrupt.
This is probably a hardware malfunction.

FPP ILLEGAL OPCODE TRAP
The FPP has detected an
instruction.

C-16

illegal

floating

point

APPENDIX C. ERROR DIAGNOSTICS
13

FPP UNDEFINED VARIABLE TRAP
The FPP loaded an illegal value (-0.0). This trap
should not occur since the OTS initialization
routine does not enable this trap condition.
A
negative zero value should never be produced by
any FORTRAN operation.

FPP MAINTENANCE MODE TRAP
The FPP has interrupted with a Floating Point
Exception Code register value of 14 (octal). This
is probably a hardware malfunction.

Group 1 - General Input/Output Errors
These messages result from errors related to the file system.
20

REWIND ERROR
condition was detected by FCS during the
.POINT operation used to position to the beginning
of a file.
An error

DEFINEFILE ALREADY DONE
A DEFINEFILE statement was attempted on a unit for
which one has already been done.
The second
DEFINEFILE is ignored.
To change a DEFINEFILE
specification a CLOSE operation may be performed.

RECORD TOO LONG
A record has been read which is too large to fit
into the buffer specified by the MAXBUF TKB
option. Rebuild the task using a larger MAXBUF
specification.

BACKSPACE ERROR
One of the following errors has occurred:
a.
b.
c.

BACKSPACE was attempted on a file opened
for appending
FCS has detected an error condition
during the .POINT operation used to
rewind the file
FCS has detected an error condition
while reading forward to the desired
record.

END-OF-FILE DURING READ
Either an end-file record produced by the ENDFILE
statement or the FCS end-of-file condition has
been encountered during a READ statement and no
END= transfer specification was provided.

INVALID RECORD NUMBER
A direct-access READ, WRITE, or FIND statement has
specified a record number outside the rang~ from
one to the value specified in a
DEFINEFILE
statement.

DEFINEFILE NOT DONE
A direct access READ, WRITE, or FIND operation was
attempted before a DEFINE FILE was performed.

C-17

APPENDIX C. ERROR DIAGNOSTICS
27

MORE THAN ONE RECORD
attempt was made to read or write more than
single record in an ENCODE or DECODE statement.

CLOSE ERROR
condition has been detected by FCS during
a CLOSE operation when attempting to close a file.

An error

NO SUCH FILE
A file with the specified name could not be
during an open operation.

found

OPEN FAILURE
FCS has detected an error condition during an open
operation.
(This message is used when the error
condition is not one of the more common conditions
for which specific error messages are provided.)
MIXED ACCESS MODES
attempt was made to use both formatted and
unformatted operations, or both sequential and
direct access operations, on the same unit.

INVALID LOGICAL UNIT NUMBER
A logical unit number was used which is outside
the range specified by the TKB UNITS= option.

ENDFILE TO DIRECT ACCESS FILE
An end-file record may not be written to a
access file.

direct

UNIT ALREADY OPEN
A DEFlNEFlLE statement, CALL ASSIGN, or CALL
FDBSET was attempted which specified a logical
unit already opened for input/output.
INCONSISTENT RECORD LENGTH
An existing direct access

file has been opened
whose record length attribute is not the same as
specified in the DEFINEFILE or OPEN statement.
The record length is not changed.
38

ERROR DURING WRITE
FCS has detected an error condition while writing.

ERROR DURING READ
FCS has detected an error condition while reading.

RECURS lVE I/O ATTEMPT
An expression in the I/O list of a READ or WRITE
statement has caused initiation of another READ or
WRITE operation. This can happen if a FUNCTION
that performs I/O is referenced in an expression
in a READ or WRITE statement I/O list.

NO FCS BUFFER ROOM
There is not enough free core left in the File
Control Services buffer area to set up required
I/O buffers •. Rebuild the task with a larger
ACTFIL
declaration
or
reduce the level of
multibuffering.

C-18

APPENDIX C. ERROR DIAGNOSTICS
42

DE~CE HANDLER NOT RESIDENT
During open operation, the filename specifiction
included a device for which no handler task is
resident.

FILE NAME SPECIFICATION ERROR
The file name string used in a CALL ASSIGN is
syntactically
invalid,
contains
a
switch
specification, references an undefined
device
mnemonic -, or is _ otherwise not acceptable to the
RSX-IIM operating system.

RECORDSIZE TOO BIG FOR 'MAXBUF'
A DEFINEFILE statement has specified a record size
which exceeds the size available in the record
buffer. Rebuild the task using a larger TKB
MAXBUF specification.

Group 2 - Element Transmission Errors
These messages result from errors related to transmitting data between
a FORTRAN program and an internal record.
60

INFINITE FORMAT LOOP
The format associated with an I/O statement that
includes an I/O list has no field descriptors to
use in transferring those variables.
FORMAT/VARIABLE - TYPE MISMATCH
An attempt was made to output a real variable with

an integer field descriptor or an integer variable
with a real field descriptor.

SYNTAX ERROR IN FORMAT
A syntax error was encountered while the OTS was
scanning format specifications stored in an array.

OUTPUT CONVERSION ERROR
During a formatted output operation, the value of
a particular number could not be output in the
specified field length without loss of significant
digits. The field is filled with ·'s.

INPUT CONVERSION ERROR
During a formatted input operation an illegal
character was detected in an input field or the
input value overflowed the range representable in
the input variable. The value of the variable is
set to zero.

FORMAT TOO BIG FOR 'FMTBUP'
The OTS has run out of memory while scanning an
array format that was generated at run time. The
default internal format buffer length is 64 bytes.
This may be increased by using the FMTBUF TKB
option (see section 1.3.2.8).

RECORD TOO BIG FOR 'MAXBUP'
During an output operation a record was specifie~
that was longer than the maximum record length.
The default maximum record length is 132 (decimal)

(

C-l9

APPENDIX C. ERROR DIAGNOSTICS
bytes.
This may be changed by use of the MAXBUF
Task Builder option (see section 1.3.2.4).
67

RECORD TOO SMALL FOR I/O LIST
A READ statement has attempted to input more
than existed in the record being read.
example, the I/O list might have
too
elements.

data
For
many

Group 3 - Arithmetic Errors
These messages
conditions.
70

result

from

arithmetic

overflow

and

underflow

INTEGER OVERFLOW
During an arithmetic operation
magnitude has exceeded 32767.

integer's

INTEGER ZERO DIVIDE
During an integer mode arithmetic
attempt was made to divide by zero.

operation

FLOATING OVERFLOW
During an arithmetic operation a real value has
exceeded the largest representable real number.
The result of the operation is set to zero.

FLOATING ZERO DIVIDE
During a real mode arithmetic operation an attempt
was made to divide by zero. The result of the
operation is set to zero.

FLOATING UNDERFLOW
During an arithmetic operation a real value has
become less than the smallest representable real
number, and has been replaced with a value of
zero.

FPP FLOATING TO INTEGER CONVERSION OVERFLOW
During a type conversion, an FPP overflow
occurred.

trap

Group 4 - Argument Errors
These messages result from
functions or subprograms.

incorrect

calls

FORTRAN-IV

supplied

WRONG NUMBER OF ARGUMENTS
An improper number of arguments were used in a
call to a FORTRAN library function or system
subroutine.

INVALID ARGUMENT
One of the FORTRAN Library
Subroutines
has detected
value. See Appendix B.

Functions or System
an invalid argument

UNDEFINED EXPONENTIATION
An exponentiation has been attempted
mathematically undefined: e.g., 0.**0.

C-20

which

APPENDIX C. ERROR DIAGNOSTICS
LOGARITHM OF NEGATIVE VALUE

An attempt was made to take

negative number.

the logarithm of
The result returned is zero.

,SQUARE ROOT OF NEGATIVE VALUE
An attempt was made to evaluate the square root of
a negative value. Zero is returned as the result.

INVALID ARGUMENT TO LIBRARY FUNCTION
An invalid argument was used in a

call

FORTRAN library function.
86

INVALID ERROR NUMBER
The error number argument
to
one
of
the
subroutines ERRSET or ERRTST is not a valid error
number.

Group 7 - Miscellaneous Errors
90

COMPILER DETECTED ERROR
If an attempt is made to link and run an object
file, with errors reported during compilation,
generated by the FORTRAN Compiler, this error will
result
when the illegal source statement is
executed.

COMPUTED GO TO OUT OF RANGE
The integer variable or express~on in a computed
GO TO statement was less than 1 or greater than
the number of statement label references in the
list.
Control is passed to the next executable
statement (see
the
PDP-II FORTRAN
Language
Reference Manual).

Group 8 - System Directive Subroutine Errors
These messages result from incorrect calls to RSX-lIM system directive
subroutines.
100

DIRECTIVE: MISSING ARGUMENTS
call to a system directive subroutine was made
in which one or more of the arguments required for
directive execution was not given.

101

DIRECTIVE:

INVALID EVENT FLAG NUMBER.

A call to a system directive subroutine

was made
in
which
the argument used for event flag
specification was not in the valid range (1 to
64) •

C.2.4

Notes on OTS Error Processing Implementation

OTS error reports are initiated by execution of a TRAP instruction in
which the low byte of the instruction contains the internal error
code. Internal error codes are 128 (decimal) greater than the
externally printed error code.

C-2l

APPENDIX C. ERROR DIAGNOSTICS
UsaX's who wish to do 110 may use the fir.t 128 (0 to 127) trap codes .a
follows.
TRAP instruetions tran.fer control to the OTS error
proce.sor by means of a System Synohronou8 Trap Table locAted in the
O'l'S impure work area.
The first word of this table ha~ the q10bal
symbol $SST. Codin9 similar to the following miqht be used to
interoept aontrol.

, INITIALIZATION
,INtTI
MOV
MOV

,SAVE a~s TRAP ADDR
, PU'l' NEW ADDR IN SST
TABLE

$SST+14,S~T6

• ItfrCEP , $S8T+14

•••
SST61

• WORD

•••
, TRAP HANDLER

INTCEP.

1$.

CMP

f128.*2,@SP

BHI
JMP

1$
@SST6

,LOW BYTE *2 OF TRAP
INSTRC.
,BRANCH IF USER CODE
,00'1'0 OTS

,USER PROCESSING

•••
TST
RTI

,DISCARD EXTRA WORD
,EXIT INTERRUPT

(SP)+

Similar techniques could be used to intercept
traps.

the

other

synchronous

OTS error m8S8aqes are output to loqical devic.e TI. by means of a LON
automatically provided by the task builder in addition to those
specified by the UNITS direotive. If for any reason the QIO direotive
used to write the error lines reoeives an error condition return, then
an immediate exit is taken to the RSX-1IM Executive.

C-22

APPENDIX 0
COMPATIBILITY WITH OTHER PDP-II FORTRANS

COMPATIBILITY WITH PDP-II FORTRAN V08.04 UNDER RSX-IID V4A

0.1

FORTRAN IV (FOR) is a new implementation of FORTRAN for the PDP-II. It
has been designed to be compatible with the earlier FORTRAN (MOP, FTN)
on RSX-IID. However, some differences exist as a result of
1.

Correcting deficiencies in FTN FORTRAN.

Language specification decisions necessary to promote the
goal of an upward compatible line of FORTRANs, including
FORTRAN IV under RT-ll and RSX-IIM, FORTRAN IV-PLUS under
RSX-IID,' and FORTRAN-IO on the DECsystem-lO.

Providing significant extensions to FTN FORTRAN in a manner
consistent with the remainder of the FORTRAN language.

This section summarizes the differences that may affect conversion
programs from FTN to FOR.

0.1.1

Language Differences
1.

FTN permits transfer of control to FORMAT' statements, which
execute as CONTINUE statements. FOR does not, and will issue
compile-time error messages.

FTN V07.14 provides an uncounted form for Radix-SO constants
used in DATA statements. V08.04 added the counted form and
promised de-support of the uncounted form. Only the counted
form is provided in FOR.

FTN logical tests treat any non-zero bit pattern as .TRUE.
and an all-zero bit pattern as .FALSE •• FOR tests only the
lowest-order byte of a variable.

The FTN BYTE statement and data type do not exist' in FOR.
LOGI CAL *1 is equivalent to BYTE in most respects, LOGICAL *1
does not force alignment on a word boundary.

The syntax of the IMPLICIT statement is not the same
and FTN.

0-1

FOR

APPENDIX D. COMPATIBILITIES
6.

FTN provides expressions in FORMAT
statements
variable format expressions): FOR does not.

FTN permits Hollerith strings in certain expressions:
FOR
does not. DATA statements may be used to initialize variable
with strings to achieve the same result.

D.l.2

(called

Object Time System Differences
1.

FTN allows both formatted and unformatted records in the same
file by means of the MXDFUF subroutine call. This is not
supported by FOR.

The FTN service subroutines PDUMP and SETPDU are not provided
in FOR.
Similar but more flexible debugging output may be
obtained by using WRITE statements in debug lines.

The TRCLIB library for statement level execution
not available in FOR.

tracing

Some FTN service subroutines are not available,
function is provided in a different manner:

but

FTN

FOR

SETERR
TSTERR
APPEND

ERRSET subroutine
ERRTST subroutine
FDSSET subroutine

F-"

their

Default format width and number of digits for A, 0, E, F, G,
I, L, and 0 formats are supported in FTN. There are no
format defaults under FOR.

Under FTN, Q format returns the number of characters in the
input record (i.e., the record length) independent of the
current scan position.
Under FOR, Q format returns the
number of characters remaining in the input record following
the scan position. The record length may be obtained by
using the Q format first in the format specification.

(

{

'\
0.1.3

Implementation Differences
1.

FTN performs the following actions when opening a unit for
direct access I/O: first check if a file already exists: if
so, use it. If not, create a new file for use.
In FOR the type of open depends on whether a READ or WRITE
statement is causing the open operation. If it is a READ,
then the file must already exist, or an error will result.
If it is a WRITE, then a new file is created.
The action sequence described above for FTN may
by using 'UNKNOWN' as the second argument
FDBSET.

be obtained
in a call to

The implementation of error handling in FOR is significantly
different from that in FTN. In particular, the use of error

0-2

APPENDIX D. COMPATIBILITIES
classes fo~ controlling error handling is replaced by control
over each individual error condition. FOR provides much more
complete error information to the executing program if
desired.
3.

The FTN ASSIGN service subroutine returns a success/error
indication in its fourth argument. The FOR ASSIGN subroutine
ignores the fourth argument. Errors may be detected by means
of CALL ERRTST or CALL ERRSNS.

FTN does not compile format specifications for
run-time, whereas FOR does. Two differences result:

A special internal buffer is provided in FOR to contain
the compiled form of format specifications stored in
arrays which must be compiled at run-time.
This may
require the use of the FMTBUF TKB option to obtain a
sufficiently large buffer area for this purpose.

Format specifications stored in arrays are re-compiled at
run-time each time they are used. If an H format is used
in a READ statement to read data into the format itself,
that data does not get copied back into the original
array. Hence, it will not be available on a subsequent
use of that array as a format specification in FOR.

These considerations do not apply
defined in FORMAT statements.

(

use

format

specifications

FTN does 'not enforce the restriction against using
a
statement referring to an existing file to
DEFlNEFILE
designate a record size different from that specified when
the file was created. Most of the time the new record size
appeared to work as expected, but it was subject to obscure
errors if references were made to records nearly at the
end-of-file. This restriction is detected and enforced by
FOR.

FOR
FTN implements REWIND as a close, open sequence.
repositions the file at its beginning by means of an FCS
.POINT operation.

FTN implements the ENDFILE statement as a close operation.
FOR
implements
the
ENDFILE
statement by writing an
end-of-file record on a file.

FOR checks that the labels used in an assigned GO TO are
valid labels in the program unit, but it does not check at
run time whether an assigned label is in the list in the GO
TO statement. FTN checks at run time.

0.1.4

Operational Differences
1.

Under FTN a task which has suspended execution as a result of
a PAUSE statement resumes execution by means of the CON
(continue) MeR command. Under FOR, the RES (reswne) MCR
command must be used.

0-3

APPENDIX D. COMPATIBILITIES
2.

D.2

The default device assignments for logical units 5 and 6 are
CL:
and TI:
respectively on RSX-llD V4A. The default
assignments on RSX-llM are TI:S and CL:6.

COMPATIBILITY WITH PDP-II FORTRAN IV-PLUS
1.

FORTRAN IV-PLUS (F4P) logical
tests
check
only
the
highest-order bit of the value, and it treats a one as .TRUE.
and a zero as • FALSE.
FOR treats any non-zero bit pattern in the lowest-order byte
of a variable as .TRUE.
and an all-zero bit pattern as
.FALSE ••

The FOR implementation of DO loops does not enforce the
restrictions, stated in the language manual, concerning
changing DO loop parameters, including the control variable,
within the range of the loop. FORTRAN IV-PLUS permits DO
parameters other than the control variable to be changed.
Such changes do not affect the number of times the loop is
executed (the iteration count computed at the start for the
loop). FORTRAN programs which violate the FOR rules but
execute as desired may behave differently under F4P.

FOR defines named common blocks in terms of named .CSECTs.
F4P uses .PSECTs with an explicit D attribute. Since .CSECTs
are equivalent to .PSECTs with I attribute specification
assembly language programs which communicate with FORTRAN
programs by means of COMMON blocks must be changed in order
to
be
successfully linked by TKB.
(TKB treats such
conflicting attribute specifications as non-fatal errors.)

In F4P, INTEGER*4 causes both 32 bit allocation and 32 bit
computation. Only 16 bits are used for computation in FOR.

F4P provides the BYTE statement and data type as a synonym
for LOGICAL*l; F4P also provides LOGlCAL*27 FOR does not.

F4P checks the label list in an assigned GO TO at
FOR checks only for the validity of the labels.

D-4

run

time;

APPENDIX E
BIT STRING MANIPULATIONS

Realtime Process Control, Process I/O, and most of the RSX-IIM system
directives may be performed by means of FORTRAN callable subroutines.
These routines are described in the RSX-llM Executive Reference
Manual, DEC-ll-OMERA-A-D.
Two types of bit string ~anipulations are provided in order to support
these calls:
logical and shifting. In order to process words from
arrays it is necessary to be able to manipulate information on a bit
by bit basis.

E.l

LOGICAL OPERATIONS

These operations are implemented as function subprograms.
In the
following functions, m and n specify integer variables or array
elements. Operations are performed on a full word, bit by bit.
These operations are supported for compatibility purposes only.
FORTRAN logical operators .AND., .OR., .XOR., X!QV., and ,NOT.
simpler to use and generate better code.

E.l.l

Inclusive OR
lOR (m, n)

Where m and n designate arguments which are logically added.

E.l.2

Logical Product
lAND (m, n)

Where m and n designate arguments which are logically multiplied.

E-I

The
are

APPENDIX E. BIT STRING MANIPULATIONS

E.l.3

Logical Complement
NOT

(m)

Where m designates the argument which is logically complemented.

E.l.4

Exclusive OR
IEOR (m, n)

Where m and n designate arguments which are exclusively added.

E.2

SHIFT OPERATIONS

The logical shift is implemented as a function subprogram. A right or
left shift can be specified.
Zeros are propagated following the
shifted value and the argument's sign is not extended.
ISHFT (m, n)

m designates the argument to be shifted
n

n specifies the number of positions to
the direction of the shift:
n)O

shift right

n(O

shift left

naO

no shift

The absolute value of n, n, should not exceed fifteen.
the function result will be zero.

E-2

shifted

If it

and

does,

APPENDIX F
SOFTWARE PERFORMANCE REPORTS

From time to time problems and/or errors will be encountered in the
use of the FORTRAN-IV Compiler or Object Time System. These should be
communicated to Digital Equipment Corporation by means of a Software
Performance Report (SPR). Software Performance Report Forms, such as
the one shown at the end of this section, may be obtained from the
local Digital office.
Software Performance Reports should be submitted directly to Software
Communications, P. O. Box F, Maynard, Mass., 01754 for handling.
Software Support Specialists in each Digital office receive regular
reports on known software problems and will in many cases be able to
provide software ·patches for correcting the problem and/or
an
alternate programming technique for temporarily avoiding the problem.
Reports which appear to be new will be dealt with by the appropriate
group within the Software Engineering Department. Every SPR will be
acknowledged upon receipt, and every SPR will be answered in writing.
SPR's that are of general i~terest will be published so that other
Software Specialists and customers may benefit.
In preparing a Software Performance Report, the
are recommended:

following

guidelines

Give as complete a description as possible of the problem
encountered.
Often a detail that may seem irrelevant will
give a clue to solving the problem.

If possible, isolate the problem to a small example.
Large,
unfamiliar programs are difficult to work with and may result
in a misunderstanding of what the problem is or an inability
to duplicate the problem.

Include console samples, listings, link maps, and so on with
the SPR.
Annotations showing where the error occurred are
extremely helpful.

If a program reads input data, include sample input
and, if possible, sample output.

If an error example cannot be isolated to a single program
unit, include listings, etc., on all program units involved.

F-l

listings

APPENDIX F. SOFTWARE PERFORMANCE REPORTS
Experience shows that as many as one-third of all SPR's db not contain
sufficient information to duplicate the problem. Complete and concise
information will help Digital give accurate and timely service to
software problems.
The OTS version and edit number should be specified when communicating
with DEC Software Support concerning the OTS. The version and edit
numbers can be found in the Task Builder map. They ~ill appear as a
symbol and associated value in the <. ABS.) program section of the
$OTI module map listing and will have the following form:
VOOxc
where

nnnnnn
x - a decimal digit
c = an alphabetic character
n = an octal digit

The Compiler version number appears in the <. ABS.) section of the
module $FORT, and takes the form MF4xxx, where x is a decimal digit.
A system may crash if Task Builder switches such as
used which do not match the hardware configuration.

F-2

IFP

lEA

are

APPENDIX F. SOFTWARE PERFORMANCE REPORTS

mamalma

SPR # _ _ _ _ _ _ __
Field # _ _ _ _ _ _ __
Page _ _ _ of _ _ _ __

SOFTWARE PERFORMANCE REPORT

Name

Saftware Specialist

Company

Office

Address

Report type

o Logic/coding error'
o Documentation error
o Suggestion
o Inquiry

Zip
Phone

/Date sent

Computer

IMemory

ISystem device

System program & version
Attachments

Cost Center

Other hardware

. 0 terminal

o obJect

o source

tape

printout

Page

Document

/ Monitor & version

o Jlstlng

o example

~«,
....~V
C-.~'
-J

(

FOR SOFTWARE SPECIALIST USE ONLY
DATE REC'O
DATE ANS'O

TO S.C"
TO CUSTOMER

FOR SOFTWARE COMMUNICATIONS USE ONLY
DATE REC'O

TO MAINTAINER

DATE ANS"D

DEC 7-(380)-1044B-R373

__ ,

•• '

UARCO FORM NO. 81-307-34

F-3

INDEX
ABORT, 1-19
ACCEPT, 3-2
Accessing an existing file, B-8
ACTFIL (task builder option),
1-12
Algorithm, error processing, C-12
Appending to a file, B-8
Arrays, multi-dimensional, 2-5
Arrays, vectored, 2-5
ASG (task builder option), 1-14
Assembly language programs, 2-5
Assembly language subprograms, 2-3
ASSIGN subroutine, 3-2, B-2
Assignments, logical device, 3-2
Assignment, logical unit, 3-2
Asterisk used in ODL, 1-17
Bit string manipulations, E-l
Blank COMMON, 1-8
Buffers, 1-12

(

CALL EXIT statement, 1-19
Clock, programmable, B~12
Clocks, line frequency, B-12
CLOSE, B-3
Command string, 1-1
Command file, indirect, 1-2
Command files example, indirect,
1-20
COMMON (task builder option), 1-12
COMMON, blank, 1-8
COMMON, initialization, 3-1
Compatibility with other FORTRANs,
D-l
Compiler, 1-5
Error diagnostics, C-l
Memory requirements, 1-10
Switches, 1-6
See also Switches, compiler
Version number, F-2
Complex format, A-3
Continuation lines, 3-1
Control returned to calling
program, 2-4
Conventions, documentation, vii
Conversion
from FTN to FOR, D-l
from F4P to FOR, D-l
Creating a new file, B-8
Data representation, A-I
DATE, B-3

IDE switch, 1-6, 1-20
Debugging, 1-20
Default file types, 1-5
DEFINEFILE, B-9
Devices,
Peripheral, 1-4
Pseudo, 1-4
Diagnostics
Error, C-l
Warning, C-lO
Documentation conventions, v~~
Double precision format, A-2
lEA switch, 1-11
11/45 Floating point processor, 1-11
.END statement, 1-16
ERR=transfer, C-12
Error action, B-4
Error codes, OTS, C-15
Error control bits, C-12, C-15
Error diagnostics,
Compiler, C-l
Fatal, C-lO
Initial Phase,C-3
OTS, C-12
Secondary phase, C-4
Warning, C-lO
Error handling, OTS, B-4, B-6
Error message, OTS, C-13
Short file, C-14
Error processing,
Algorithm, C-12
Implementation, C-22
Errors, syntax, C-l
ERRSET subroutine, B-4
ERRSNS subroutine, B-5
ERRTST subroutine, B-6
lEX switch, 1-7
EXIT subroutine, B-7
Extended arithmetic element, 1-11
Fatal error diagnostics, C-lO
FCS, B-8, C-14
FCS file control conventions, 3-3
FDBSET subroutine, B-8
File control services (FCS) , B-8,
C-14
File control conventions, FCS, 3-3
File descriptor block (FOB), C-14
Filename, 1-2
File search, B-8
File specifications, 1-2
File type, 1-2

INDEX-l

File types, default, 1-5
File updating, B-8
Floating-point formats, A-I
Floating point processor, 11/45,
1-11
FMTBUF (task bui1d~r option), 1-14
FOR MCR command, 1-5
Format compilation, object time,
1-14
FORTRAN libraries, 1-15
FORTRAN subprograms, 2-3
FTN, D-1
Function subprograms, 2-5
F4P, D-1
Generated code listing, 1-8

[.-..

Hollerith,
Format, A-4
Strings, B-9
/I4 switch, 1-7, A-I
lAND, E-1
/ID switch, 1-7
IDATE subroutine, B-4
lEaR, E-2
Image, task, 1-10
Implementation differences,
FTN-FOR, D-2
Implementation, error processing,
C-22
'
Ipdirect command file, 1-2
example, 1-20
Initial phase error diagnostics,
C-3
Input file, 1-5
Installing a task, 1-18
Integer .format, A-I
Internal buffers, B-8
Internal Sequence Numbers (ISN),
1-8, 2-2
lOR, E-1
lRAD50 subroutine, B-9
ISNs, internal sequence numbers,
1-8, 2-2
Keyword options, 1-12
Languages differences, FTN-FOR,
D-l
/LB switch, 1-11, 1-15
/LI switch, 1-6
LIBR option, 1-15
Libraries, 1-13, 1-15
Relocatable, 1-15
Shared, 1-13, 1-15
System, 1-15
User, 1-15

Library subroutines, B-1
Summary, B-1
List file, 1-7
List formats, 1-7
Listing
Generated code, 1-8
Source, 1-8
Storage Map, 1-8
Local symbols, 1-8
LOGICAL*l, A-3
Logical device assignments, 3-2
Logical format, A-4
Logical operations, E-1
Logical operators, E-1
Logical shift, E-2
Logical unit numbers, 1-14, B-2
Logical units, 1-13
LUN assignments, 1-14, 3-2, B-2
Map, memory allocation, 1-10
MAXBUF (task builder option), 1-13
Maximum record size, 1-13
MCR, monitor console routine, 1-1,
1-5, 1-10
Memory allocation map, 1-10
Memory requirements, compiler, 1-10
IMP switch, 1-11, 1-16
Monitor console routine, (MCR), 1-1,
1-5, 1-10
Multi-dimensional arrays, 2-5
NOT, E-2
Null arguments, 2-4
Object code, 2-1
Object time format compilation, 1-14
Object time system (OTS), 2-1,
see OTS
ODL, overlay description language,
1-16
ODT, (on-line Debugging Technique) ,
1-20
/Op switch, 1-7
Operational differences, FTN-FOR,
0-4
Operations ,.
Logical, E-1
Shift, E-2
Operators, logical, E-1
Options, keyword, 1-12
Options, task builder, 1-12
COTS), object time system, 2-1
Differences, FTN-FOR, D-2
Edit number, F-2
Error codes, C-15
Error diagnostics, C-12
Error handling, B-4, B-6
Error message format, C-13
Version number, F-2
Output file, 1-5

INDEX-2

- :',

~
r---:..
.'
\.
j

Overlays, 1-16
Overlay description language (ODL) I
1-16
Overlay structure, 1-16
Overlay structure, examples,
1-17, 1-18
Object code, 2-1
Object time format compilation,
1-14
Object time system (OTS) , 2-1
See OTS
ODL, overlay description language,
1-16
ODT, (On-line Debugging Technique) ,
1-20
lop switch, 1-7
Operational differences, FTN-FOR,
0-4
Operations,
Logical, E-1
Shift, E-2
Operators, logical, E-l
Options, keyword, 1-12
Options, task builder, 1-12
(OTS) , object time system, 2-1
Differences, FTN-FORm 0-2
Edit number, F-2
Error codes, C-15
Error diagnostics, C-12
Error handling, B-4, B-6
Error message format, C-13
Version number, F-2
Partition, 1-14
PAUSE statements, 1-20
Peripheral devices, 1-4
PRINT, 3-2
Program termination, B-7
Pseudo devices, 1-4

SECNDS subroutine, B-ll
Secondary phase error diagnostics,
C-4
ISH switch, 1-11
Shared access, B-8
Shared library, 1-13, 1-15
Shift, logical, E-2
Shift operati9ns, E-2
Short message file, C-14
ISN switch, 1-7
Software performance reports, F-1
Source listing, 1-8
ISP switch, 1-6, 1-11
Statement ordering, 3-3
STOP, 1-19
Storage map listing, 1-8
Subprograms,
Assembly language, 2-3
Function, 2-5
FORTRAN, 2-3
Register usage, 2-4
Subroutines, library, B-1
Subroutine linkage, 2-3
Switches, compiler, 1-6
IDE, 1-6
lEX, 1-7
lID, 1-7
1I4, 1-7, A-I
ILl, 1-6
lop, 1-7
ISN, 1-7
Isp, 1-6
IVA, 1-7, 2-6
IWR, 1-7, C-10
Switch default summary, 1-7
Switch format, 1-3
Switch options, task Builder, 1-11
lEA, 1-11
IFP, 1-11
ILB, 1-11, 1-15
IMP, 1-11, 1-16
ISH, 1-11
Isp, 1-11
Syntax errors, C-1
SYSLIB.OLB (system object library),
1-10, 1-15
System Libraries, 1-15

R50ASC subroutine, B-10
RAD50 subroutine, B-9
RADIX-50
Character set, A-4
Format, A-4
Notation, B-9
RANDU, RAN subroutine, B-11
REASSIGN, 3-2, B-2
TASK (task builder option), 1-13
Read only access, B-8
Task builder, 1-10
Real format (2-word floating point),
Exit, 1-10
A-2
Task builder options, 1-12
Record size, maximum, 1-13
ACTFIL, 1-12, 3-2
Relocatable libraries, 1-15
ASG, 1-14
Register usage, subprogram, 2-4
COMMON, 1-12
RESUME, 1-19
FMTBUF, 1-14
RETURN, vii
LIBR, 1-13
.ROOT statement, 1-16
MAXBUF, 1-13
RUN, 1-19
PAR, 1-14

INDEX-3

TASK, 1-13
Ule, 1-2, 1-13
UNITS, 1-13
Task Builder switch options, 1-11
See also Switch options, Task
Builder
Task execution, 1-19
Task image, 1-10
Task Image file, 1-10
Task name, 1-13
TIME subroutine, B-12
TYPE, 3-2

Ule (task builder option), 1-13
UNITS (task builder option), 1-13
Units, logical, 1-13
Unit numbers, logical, 1-14
USEREX subroutine, B-7
User libraries, 1-15
/VA switch, 1-7, 2-6
Variable names, 3-1
Vectored arrays, 2-5
Version, 1-2
Warning diagnostics, e-10
/WR switch, 1-7, e-10

{
\

INDEX-4

HOW TO OBTAIN SOFTWARE INFORMATION
SOFTWARE NEWSLETTERS, MAILING LIST
The Software Communications Group, located at corporate headquarters in
Maynard, publishes newsletters and Software Performance Summaries (SPS)
for the various Digital products. Newsletters are published monthly,
and contain announcements of new and revised software, programming
notes, software problems and solutions, and documentation corrections.
Software Performance Summaries are a collection of existing problems
and solutions for a given software system, and are published periodically. For information on the distribution of these documents and how
to get on the software newsletter mailing list, write to:
Software Communications
P. O. Box F
Maynard, Massachusetts 01754
SOFTWARE PROBLEMS
Questions or problems relating to Digital's software should be reported
to a Software Support Specialist. A specialist is located in each
Digital Sales Office in the United States. In Europe, software problem
reporting centers are in the following cities.
Reading, England
Paris, France
The Hague, Holland
Tel Aviv, Israel

Milan, Italy
Solna, Sweden
Geneva, Switzerland
Munich, West Germany

Software Problem Report (SPR) forms are available from the specialists
or from the Software Distribution Centers cited below.
PROGRAMS AND MANUALS
Software and manuals should be ordered by title and order number. In
the United States, send orders to the nearest distribution center.
Digital Equipment Corporation
Software Distribution Center
146 Main Street
Maynard, Massachusetts 01754

Digital Equipment Corporation
Software Distribution Center
1400 Terra Bella
Mountain View, California 94043

Outside of the United States, orders should be directed to the nearest
Digital Field Sales Office or representative.
USERS SOCIETY
DECUS, Digital Equipment Computer Users Society, maintains a user exchange center for user-written programs and technical application information. A catalog of existing programs is available. The society
publishes a periodical, DECUSCOPE, and holds technical seminars in the
United States, Canada, Europe, and Australia. For information on the
society and membership application forms, write to:
DECUS
Digital Equipment Corporation
146 Main Street
Maynard, Massachusetts 01754

DECUS
Digital Equipment, S.A.
Sl Route de l'Aire
1211 Geneva 26
Switzerland

READER'S COMMENTS
NOTE:

RSX-IIM FORTRAN-IV
USER'S GUIDE
DEC-II-LMFUA-A-D

This form is for document comments only. Problems
with software should be reported on a Software
Problem Report (SPR) form (see the HOW TO OBTAIN
SOFTWARE INFORMATION page).

Did you find errors in this manual?

If so, specify by page.

Did you find this manual understandable, usable, and well-organized?
Please make suggestions for improvement.

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
[] Occasional programmer (experienced)
User with little programming experience
Student programmer
Non-programmer interested in computer concepts and capabilities

o
o

Name.,...,..---,...,---.....-________________ Oat.e _____________
Organization __--_._-.....--___- - - - - - - - - - - - - - - - - - - - Street_.....-,---___- - - - - - - - - - - - - - - - - - - - - - - - - - - - Ci ty_~-.,..----~----- State ____- - - Zip Code _ _ _ _ _ __
or
Country
If you do not require a written reply, please check here.

-------------------------------------------------------------Fold lIere------------------------------------------------------------

----------------------------------------------..- Do Not Tear - Fold lIere 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:

Software Communications
P. O. Box F
Maynard, Massachusetts 01754