Digital PDFs

DEC-15-GFWA-D

August 1971

110 pages

Original

34MB

Document:	FORTRAN IV Language Manual
Order Number:	DEC-15-GFWA-D
Revision:
Pages:	110
Original Filename:	http://bitsavers.org/pdf/dec/pdp15/XVM/DEC-15-GFWA-D-PDP-15-FORTRAN_IV_LANGUAGE_PROGRAMMERS_MANUAL.pdf

OCR Text

,'....
/~~' f

' " ,' , ' "C

(

PDP-15
- FORTRAN IV LANGUAGE
PROGRAMMER'S REFERENCE MANUAL

For additional copies, order No. D EC-15-GFW A-D from Program Library, Digital
Equipment Corporation, Maynard, Massachusetts 01754
Price $5.00

DIGITAL EQUIPMENT CORPORATION • MAYNARD, MASSACHUSETTS

1st Edition August 1971

The material in this manual is for informational purposes and is subject to change
without notice_

The following are trademarks of Digital Equipment
Corporation, Maynard, Massachusetts:
DEC
FLIP CHIP
DIGITAL

PDP
FOCAL
COMPUTER LAB

CONTENTS
Page
CHAPTER 1

BASIC ELEMENTS OF A FORTRAN-IV PROGRAM

1.1

The Character Set

1-2

1.2

Program Struch.... e

1-2

1.3

Expressing Data Values

1-3

1.3.1

Constants

1-5

1.3.2

Variables

1-8

1.3.3

Express ions

1-10

(

CHAPTER 2

ASSIGNMENT STATEMENTS

2.1

The Arithmetic Statement

2-1

2.2

The ASSIGN Statement

2-3

CHAPTER 3

CONTROL STATEMENTS

3.1

The GO TO Statement

3-1

3.1.1

The Unconditional GO TO Statement

3-1

3.1.2

The Computed GO TO Statement

3-2

3.1.3

The Assigned GO TO Statement

3-2

3.2

The DO Statement

3-3

3.2.1

Execution of a DO Range

3-4

3.2.2

Nested DO Statements

3-5

3.2.3

The CONTINUE Statement

3-6

3.3

The IF Statement

3-6

3.3.1

The Arithmetic IF Statement

3-7

3.3.2

The Logical IF Statement

3-8

3.4

Execution Control

3-9

3.4.1

The PAUSE Statement

3-9

3.4.2

The STOP Statement

3-9

CHAPTER 4

SPECIFICATION STATEMENTS

4.1

Mode Specification

4-1

4.1.1

Mode-Dec laration Statement

4-2

4.1.2

The IMPLICIT Statement

4-2

4.1.3

The EXTERNAL Statement

4-3

iii

CONTENTS (Cont)
Page
4.2

Storage Allocation

4-3

4.2.1

The DIMENSION Statement

4-3

4.2.2

The COMMON Statement

4-4

4.2.3

The EQUIVALENCE Statement

4-5

4.3

The DATA Statement

4-7

4.4

Block Data Subprograms

4-8

CHAPTER 5

SUBPROGRAM STATEMENTS

5.1

Functions

5-2

5.1.1

Statement Functions

5-2

5.1.2

External Functions

5-6

5.1.3

DEC library Functions

5-7

5.2

Subroutines

5-7

5.2.1

Subroutine Definition

5-8

5.2.2

Subroutine Calls

5-9

5.3

Multiple Entries and Returns

5-9

CHAPTER 6

DATA TRANSMISSION STATEMENTS

6.1

The FORMAT Statement

6-2

6.1.1

Statement Syntax

6-3

6.1.2

Field Descriptors

6-4

6.1.3

Object Time FORMAT Specifications

6-16

6.2

Data-Directed Input-Output

6-16

6.3

Input-Output Statements

6-17

6.3.1

Input-Output lists

6-18

6.3.2

Sequential Input-Output Statements

6-19

6.3.3

Direct-Access Input-Output

6-20

6.3.4

The ENCODE/DECODE Statements

6-23

6.3.5

Auxi I iary Input-Output

6-23

- ."

CONTENTS (Cont)
Page
APPENDIX A

LANGUAGE SUMMARY

A-1

APPENDIX B ERROR MESSAGES
B.1

Compiler Error Messages

B-1

B.2

OTS Error Messages

B-7

B.3

OTS Error Messages in FPP Systems

B-9

APPENDIX C

PDP-15 FORTRAN FACILITIES

C-1

ILLUSTRATIONS
Figure No.

Title

Page

1-1

DEC FORTRAN-IV Coding Form

1-4

3-1

Rules for Nested DO Statements

3-6

5-1

Main Program Sample

5-3

5-2

Subprogram Sample

5-4

6-1

Programming Example - Auxiliary I/O to Disk

6-25

TABLES
Table No.

Title

Page

1-1

Sequence Rules for FORTRAN Statements

1-2

Modes of Mixed Expressions

1-12

6-1

Physical Record Description for Formatted and
Unformatted Records

6-1

6-2

Field Descriptor Control Characters

6-4

6-3

The READ Statement

6-19

6-4

The WRITE Statement

6-20

6-5

Arguments for CALL DEFINE

6-21

C-1

Versions of the Extended Compiler

C-1

C-2

Versions of the OTS Libraries for the Extended Compiler

C-2

C-3

Compilers and Libraries for Exten~ed FORTRAN
Distributed with PDP-9/15 Systems

C-2

PREFACE

This manual describes the elements, syntax and use of the FORTRAN IV language as implemented for
the PDP-15 computer. Three versions of the PDP-15 compi ler are avai lable; their use is governed by
the hardware/software configuration of the system on which FORTRAN is to be run. The most comprehensive version of PDP-15 FORTRAN IV is described in this manual. See Appendix C for overall outlines and descriptions, and tabularized descriptions of the differences between the various versions of
the FORTRAN IV compilers and their associated libraries.
All versions of PDP-15 FORTRAN IV are based on USASI Standard FORTRAN (X3.9-1966); the following features were added to the PDP-15 version of FORTRAN IV:
ENCODE/DECODE Statements, data-directed Input/Output, multiple
entries and returns from subroutines.
double integer constants and variables, part-word notation for Arithmetic
statements;
direct access input/output statements.
END and ERR input-output options
octa I format descr i ptors
The following standard features are not available:
complex arithmetic
adjustable arrays (done via subroutine in this version)
One additional difference from the standard is that the blank COMMON and labeled COMMON are
treated the same.
A companion manual, "PDP-15 FORTRAN IV OPERATING ENVIRONMENT", order code DEC-15GFZA-D, describes the system software facilities needed to support the various versions of the PDP-15
FORTRAN IV compiler and hardware features which affect the FORTRAN programmer. Included in
this manual are descriptions of the FORTRAN IV Object Time System (OTS) and Science Library.

-.,

CHAPTER 1
BASIC ELEMENTS OF A FORTRAN-IV PROGRAM

A FORTRAN-IV source program is a sequence of symbolic statements which are translated by the
(

FORTRAN-IV compiler into an object program; that is, a program which may be executed by a computer. A statement, the basic unit of expression in a FORTRAN source program, may represent computer
instructions and program data or may provide the compiler with instruttions required in the translating
process, such as the size of an array, the number of times a loop is to be executed, or whether the program is a subroutine to be called by other programs.
A FORTRAN statement consists of a command portion which characterizes its function and may, in addition, require arguments. For example, the GO TO statement, which transfers control from one statement to another, requires an argument specifying the source-program statement whi ch is the target of
the transfer. The five functional categories into which all FORTRAN-IV statements fall are given
below.

General Function

Category
Assignment Statements

Assi gn val ues to symbol i c representations

Control Statements

Govern the sequence in which operations are performed

Specification Statements

Describe data the object program will
process

Subprogram Statements

Estab Iish subprograms

Data Transmission Statements

Govern the transfer of information between the computer and peripheral devices (I/O)

The format of each statement and the arguments required for each are described in detai I in subsequent
chapters.

1-1

1. 1

THE CHARACTER SET

The character set from which FORTRAN statements may be constructed consists of the 26 letters (A-Z),
the 10 digits (0-9), and the following special characters:
[

]

Left bracket
Right bracket
Colon
Semi-colon
Sharp sign
Sing Ie quote
Blank
Equals
Plus

(
)

Minus
Asterisk
Slash
Left parenthesis
Right parenthesis
Comma
Decimal point
Dollar sign
Quotes

Other characters may appear on Iy ina text str i ng (Ho II er ith constant).

1.2

PROGRAM STRUCTURE

A FORTRAN source program's beginning is simply the first statement encountered. Its end must be denoted by an END statement consisting of the characters END.
Comments may precede the body of the program or be inserted between statements by means of a comment line which begins with the character C and is followed by text.
The over-all rule for ordering statements is that non-executable statements (no machine code generated)
must precede executable statements. The precise order in which statements may appear is given in
Table 1-1 below.
Table 1-1
Sequence Rules for FORTRAN Statements
Statements

Order

BLOCK DATA; FUNCTION; SUBROUTINE

IMPLICIT

INTEGER; REAL; LOGICAL; DOUBLE PRECISION; DOUBLE INTEGER

DIMENSION

COMMON

EQUIVALENCE; EXTERNAL

DATA

Statement functions

All other

1-2

The form in which statements and comments are entered is governed by an 80-character line on a
standard FORTRAN coding form, as shown in Figure 1-1. If the source program is input in card form,
the columns on the coding sheet correspond to card columns. If paper tape is used, the columns refer
to characters.
Each line in a program is organized into the following fields, some of which may be blank:
Field Name

Columns

Contents

Statement number

1-5

A decimal number from 1-9999 identifying
the statement. May be in any order.

Line continuation field

If non-blank, indicates that the statement
portion is a continuation of the preceding
line.

Statement field

7-72

FORTRAN statement or portion thereof.

Identification field

73-80

Ignored by the compiler, filled at the user's
discretion .

A comment I ine is indicated by a C in column 1; comment text may be placed anywhere in col umns

2-72.
With the exception of the DO statement, which must be on one line, any statement may have as many
continuation lines as desired. Any statement except the Arithmetic statement and the Arithmetic IF
statement may be broken at any point. Continuation rules for these two statements are described in
Chapters 2 and 3, respectively. In general, blanks may be freely imbedded within statements to improve their legibility.
For non-card input, the first character is equivalent to the first column and a line is terminated by a
carriage return. A statement field may begin with the seventh character or may be indicated by a
TAB followed by an alphabetic character. A continuation line may begin with the sixth character or
may be indicated by a TAB followed by a numeric character.

1.3

EXPRESSING DATA VALUES

Program data may be expressed in a variety of ways in a FORTRAN-IV program. The basic units, constants and variables, represent single values - a constant has the same value throughout program execution, a variable has whatever value it is currently assigned. New values may be computed from
known values of these units via expressions, which are composed of constants, variables and FORTRAN
operators which indicate the computation to be performed.

1-3

PROGRAMMER NAME

,_H_
PROGRAM TITLE

FORTRAN

DATE

DIGITAL EQUIPMENT CORP.
MAYNARD, MASS. 01754

PAGE

.J,.

DEC 7·1075

Figure 1-1

DEC FORTRAN-IV Coding Form

)

1 .3. 1 Constants
A constant is, as indicated above, a value which does not change from one execution of a program to
another. Six types of constants, each representing a different PDP-15 interna I data format, may appear
,
in a FORTRAN-IV program. These are: INTEGER, DOUBLE INTEGER, REAL, DOUBLE-PRECISION,
LOGICAL, and HOLLERITH. The form of each is described below.
INTEGERS - An integer constant represents a single word of PDP-15 storage. Its value may be expressed in the source program as a decimal or octal number.
A decimal integer consists of one to six decimal digits with no decimal point. A negative quantity is
indicated by a minus sign. A positive quantity may optionally be preceded by a plus sign. All of the
following are legal decimal constants:
+97

-2176
576
Leading zeroes are ignored; thus -0010 is equivalent to -10. The magnitude of the integer must be less
17
than or equal to 131071
(i.e., 2 _1).
10
An octal integer is indicated by a sharp sign (#) followed by one to six octal integers. The following
are legal octal integers:

DOUBLE INTEGERS - A double integer represents two words of storage (1 sign bit and 35 bits of magnitude) and has a range between -34,359,738,367

and +34,359,738,367 , The notation D, pre10
10
ceding an octal integer, indicates that it is double integer, as in: #D7777777. Note that an octal
integer value not preceded by D may be assigned to a double integer variable (i .e., DI = #130000).
However, the value of the assigned integer must NOT exceed the maximum size permitted integers
(i.e., 377777 ), If an assigned octal integer does exceed the maximum permitted value, its most
8
significant digits will be truncated before it is assigned to the double integer variable regardless of
the fact that the double integer would accept its original value. Decimal integers whose absolute
values exceed 131071

are taken as double integers. Tf-e following are examples of legal double

integers:
#D400000
141520
#DO
# D4oo000000

400000
1-5

REAL CONSTANTS - A real constant is a string of decimal digits with a decimal point, optionally
followed by a decimal exponent. It may be a whole number (i .e., 10.), a fraction (i .e., .10), or a
mixed format number (i .e., 10.10). The programmer may supply any number of digits in a real constant but only the leftmost seven are significant. Thus,
10.111550
and
10.111552
are equivalent.
A plus or minus sign may precede the constant - plus being optional for positive quantities. All of the
following are valid:
325.
0.0
+325.

-9.8
999999.0
999999999.
If a decimal exponent is present, it is indicated by the letter E immediately following the constant.
The decimal point may be omitted if immediately followed by an exponent. The exponent itself, which
immediately follows the E, is an optionally signed one- or two-digit number indicating the appropriate
power of 10, as in:
5.E-3 (i .e., 0.005)
5.0E3 (i.e., 5000.)
5E2
(i.e., 500.)
The adjusted absolute value of the exponent cannot exceed 75. Thus, the constant. 99999E75 is legal,
but 999. 999E73 is not.
Areal constant occupies two words of PDP-15 storage in the following arrangement.
Low order
mantissa

Exponent
(2's complement)

8 9

High order mantissa

The mantissa and its sign are represented in signed magnitude form.

1-6

DOUBLE-PRECISION CONSTANTS - A double-precision constant is interpreted like a real constant
but with greater accuracy (nine-digit). It is written as a string of decimal digits, including a decimal
point immediately followed by the letter D and a signed decimal exponent no greater than 75. (Plus
is optional.) The field following the D may not be blank but may be zero. For example:
-3.0DO
987.6542D15
32. 123D+7

Double-precision constants are stored in three words arranged as follows.

Exponent (2's complement)

Sign
of
mantissa

High order mantissa

Low order mantissa

The mantissa and its sign are represented in signed magnitude form.
LOGICAL CONSTANTS - There are two logical constants - . TRUE., which is stored as 777777 , and
8
.FALSE., which is stored as O. Logical quantities may be operated upon both by arithmetic and
logical operators yielding, respectively, arithmetic and logical results.
HOLLERITH CONSTANTS - A Hollerith constant is a string of 1 to 5 characters. They are packed in
7-bit ASCII in two words of storage with the rightmost bit of the second word always zero. A Hollerith
constant may be used in CALL and DATA statements and, if the programmer exercise~ caution, in an
Arithmetic statement. There are four forms for writing a Hollerith constant:
(1)

nH characters

where n is the number of characters (1 to 5). Examples of this format are:
1HA
4HA$CD

1-7

When the above notation is used, the string is stored as a real constant.
(2)

'charac ters'

Examples of this are:

'A'
'A$CD'

If quotation marks are to be included in the character string itself, this may be indicated by having
two single quotes in sequence, as in:
'A""CD'
which stores the string A"CD. This and the following forms of string constants are stored as unsigned
double integers and may be used wherever a double integer may.

Double quotes may be used instead

of single quotes.
(3)

"characters"

Examples of this form are:

"A BC"
which yields string A BC and

which yields string A" "B.
(4)

$characters$

Examples are:
$A$
$A$$CD$
where the second example yields the string A$CD.
Blanks within a Hollerith constant are considered as characters. Thus,
'AB'
is a three-character string.

1.3.2

Variables

The term variable refers to a symbolic name which represents a location in memory and to the values
which are stored there during program execution. A variable name in FORTRAN is a string of from one
to six characters, the first of which must be alphabeti c. Thus, ALPHA, MAX, A34, and A are legal
variable names while 2A and MAXIMUM are not.

1-8

The kind of value which may be associated with a given variable name must be specified so that appropriate storage is allocated. This specification is referred to as the mode of the variable, where mode
is INTEGER, DOUBLE INTEGER, REAL, DOUBLE PRECISION, or LOGICAL.
The implicit mode assumptions of the compiler are that all variables beginning with the letters I through
N are integers and all others are real. For any modes other than integer and real, the programmer is
responsible for establishing a variable's mode. The programmer may also establish a different set of
mode assumptions via the IMPLICIT statement or explicitly declare a variable's mode via one of the
FORTRAN-IV mode-declaration statements. Chapter 5 describes these statements in detail.
A variable may also name an array, an ordered set of data whose elements are referred to by means
of subscripted variables. A subscripted variable has the form:
V(n)
where n is a list of from one to three expressions which yield positive (non-zero) integer values.
The variable name is, in effect, the name of the entire array. For example, the subscripted variable:
A(3)

refers to the third element of a one-dimensional array named A. Arrays in FORTRAN-IV may have up
to three dimensions; consequently, subscripted variables (also referred to as array elements) may have
up to three subscripts as in:

A(l,2,2)
which represents the value located in the first row, second column, and second plane of a threedimensional array named A. *
A variable which represents an array must be assigned adequate storage to contain all elements. To ensure this, the programmer must provide dimensioning information giving the maximum value each of the
array's subscripts can obtain. This may be done via several of the specification statements described

*Arrays are stored in column order in ascending absolute storage locations. For example, a 2 by 2 by
2 array is stored in the following sequence:

A(1,1,1)
A(2, 1, 1)
A(1,2,1)
A(2,2,1)
A(1, 1,2)
A(2, 1,2)
A(l,2,2)
A(2,2,2)

1-9

in Chapter 5. Note that when an array has been defined to have a certain number of dimensions, all
references to it must contain that number of subscripts. Note also that an array must be of a given
mode; i.e., each element is of the same specified mode.

1.3.3

Expressions

The term expression may broadly refer to the whole range of value descriptions which can be made in
FORTRAN. This includes the primary units discussed so far (constants and variables, function references
discussed in Chapter 5), and combinations which relate several units via FORTRAN operators. The
value of this latter type of expression is, in reality, the result of the computations represented by its
operators.
Two types of compound expression - arithmetic and logical - may be constructed in FORTRAN. Either
type may be enclosed in parentheses and function as a primary unit (or operand) in another expression.

ARITHMETIC EXPRESSIONS
An arithmetic expression is any configuration which yields a numeric value. It may be a single arithmetic unit or combination of arithmetic operands and the arithmetic operators given below.
Operation

Operator

Addition (or unary plus)
Subtraction (or unary minus)

Multiplication

Division

Exponentiation

An operand may be a constant, variable, function reference, or a parenthesized expression.
The following are examples of legal arithmetic expressions:
2.71828
XYZ
A+B*C
(A+B)*C

Precedence of Operations
Arithmetic operations are carried out according to the following rules of precedence:
(1)

function reference

(2)

** (exponentiation)

1-10

(3)

unary minus

(4)

* (multiplication), / (division)

(5)

+ (addition), - (subtraction)

At the same precedence level, operations are carried out from left to right. For example, the expression:
-I + J/2 * 10 + SQRT(A) ** 3

'-"" \.,./
(2)

(3)
~
(4)

(1)

is evaluated as follows:
(1)

the square root of A is raised to a power of 3;

(2)

the value of 1 is complemented;

(3)

J is divided by 2 and

(4)

the result multiplied by 10.

The remaining operations [(2) + (4) + (l)J are carried out from left to right.
When an expression enclosed in parentheses appears within an expression, it is evaluated before being
used as an operand, thus overriding the rules of precedence. Some examples are:
Regular Precedence

Wi th Parentheses

4 + 2 ** 2 = 8

(4 + 2) ** 2 = 36

8-4*2=0

(8 - 4) * 2 = 8

-10+4=-6

-(10+4)= -14

18/2 * 3 = 27

18/(2*3)=3

-1 ** 2 = -1

(-1) ** 2 = 1

Mode of Expressions
Expressions, like variables, have modes. In the case of an expression, however, the mode determines
its accuracy. When an expression is composed of operands of the same mode, this mode applies to the
entire expression . For example, an expression consisting of integer constants or variables is in integer
mode, and so on. Different mode operands may also be used to form expressions. All legal combinations
and the resultant mode are given in Table 1-2.

1-11

Table 1-2
Modes of Mixed Expressions
A plus sign represents any of the operators (+, -, *, or /).
Expression

Mode

1+1
R-tR

Integer
Real

I+DI
DI+I
DI+DI
R+D
D+R
D+D

}
}

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

Double integer

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

Double precision

1**1

Integer

R**1 or DI
R**R

}

- - - - - - - - - -

Real

R**D
D**I or DI
D**R
D**D

}

- - - - - - - - - -

Double precision

- - - - - - - - - - -

Double integer

I**DI
DI**I
DI**DI

}

LOGICAL EXPRESSIONS
Logical expressions consist of any configuration which yields a logical value (i .e., .TRUE. or .FALSE.).
This may be a combination of arithmetic expressions and relational operators, a logical constant or
variable, or a combination of logical operands and logical operators.
An expression using relational operators has the form:

A operator B
where A and B are arithmetic expressions and operator is one of those listed below.

Relation

Relational Expression

A .LT. B

A less than B

A •LE. B

A less than or equal to B

A .EQ. B

A equal to B

A .NE. B

A not equal to B

A .GT. B

A greater than B

A .GE. B

A greater than or equal to B

1-12

An expression has the value .TRUE. if the relation expressed is true; otherwise, it has the value .FALSE.
For example, assuming a variable A with the value 10 and a variable B with the value 20:
A . LT. B has the value. TRUE.
while
A .GE. B has the value .FALSE.
The following mode combinations are permitted in a relational expression:

Mode

May be Related to

Integer

Integer, doub Ie integer

Double integer

Double integer, integer

Real

Rea I, double precision

Double precision

Double precision, real

Logical operators can combine logical or integer operands. The operators and their meanings are given
below (T indicates a value of . TRUE. for logical operands and non-zero for integers, F, . FALSE. or
zero).

Logical Operator

Meaning

.NOT.

logical negation

.AND.

.OR.

.XOR.

logical and

inclusive or

exclusive or

Examples
Expression
Result
• NOT. T
. NOT. F

F
T

T .AND. T
T .AND. F
F .AND . T
F .AND. F

T
F

T .OR. T
T .OR. F
F .OR. T
F .OR. F

T .XOR. T

F
T

T .XOR. F
F .XOR. T
F .XOR. F

F
F
T
T

Logical expressions are, like arithmetic expressions, evaluated according to precedence rules. These
are:
(1)

relationals

(2)

.NOT.

(3)

.AND.

1-13

(4)

.OR.

(5)

.XOR.

Thus, T .XOR. F .AND. F yields the value .TRUE.
The arithmetic operands of a relational expression are evaluated before the relation is tested. At the
same level of precedence, operations are carried out from left to right. In addition, logical expressions can be parenthesized, thus overriding precedence. For example:
F .AND. F .XOR. T = . TRUE.
but
F .AND. (F .XOR. T) = .FALSE.
Following. NOT. a compound expression must be parenthesized, as in:
. NOT. (F .AND. T)
which has the value. TRUE.

1-14

CHAPTER 2
ASSIGNMENT STATEMENTS

An assignment statement permits the programmer to assign a value to a symbolic name. Two FORTRANIV statements, the Arithmetic statement and the ASSIGN statement, perform this function. In the case
of the Arithmetic statement, the symbolic name identifies a variable or an array element and the value
is a constant data value or the result of a computation. For the ASSIGN statement, the name is a
symbolic address label which may be referred to by a GO TO or arithmetic IF statement (Chapter 3)
and the value is a statement number within the source program. For either statement, the val ue assigned
to a particular name may be changed by subsequent assignment statements; other statements or expressions which refer to them will automatically operate on the most recent value assigned.

2.1

THE ARITHMETIC STATEMENT
Genera I Form

var = value
or
array (i) = value

Where

value = any FORTRAN constant or expression

Examples

COUNT = 1
TABLE (COUNT) = 100
COUNT = COUNT + 1
TABLE (COUNT) = 200

Effect

The value to the right of the equal sign is assigned to the
variable or array element to the left

Note that the equal sign in an Arithmetic statement indicates replacement rather than equivalence;
this permits constructions such as COUNT = COUNT + 1. If an Arithmetic statement requires a continuation line, the = sign must appear on the first line.

If an expression of one mode is assigned to a variable of another mode, the expression is converted
before assignment. That is, integers may be converted to real, real to double-precision, and so on.
There are, however, situations in which the value obtained will be meaningless. For example, if the
integer variable I is assigned the value of the double-integer variable J, when J = loo, the assignment
will be as expected. When J = 10000000, however, an unpredictable value assignment will result.

2-1

Conversions between logical and integer obey the following convention. Any non-zero integer is
. TRUE. (777777), zero is .FALSE. (OOOOOO).
In addition to the basic Arithmetic statement form, the programmer may use a part-wurd notation of
the form:
[m : n]

where m and n are integer constants indicating a range from 0 to 35 (O:::'m:::'n :::,35). This construction
may optionally follow any variable, array element, or parenthesized expression in the value portion of
an Arithmetic statement (to the right of =) and/or the variable or array element being assigned. In the
former case, the expression will be of type integer if (n-m) ~16 and type double integer if (n-m) ~17;
its value is bits m through n of the actual value {right adjusted}. For ~xample, the statement:

assigns A the value 23, and

assigns A the value 2. If A were a double integer, the statement

would assign A the value 23. Note that #2300 is represented internally as 002300.
If this notation is used to the left of the equal sign, it indicates that only bits m through n of the
variable are to be replaced by the value of the right hand side. For example, if the integer variable
IYAR had previously been assigned the octal value 77, the statement:

would make the new value of IYAR the octal integer 177. Also, the statements:
I YAR= 100
IYAR[9:1lJ=IYAR+l
leave the value of IYAR unchanged (i.e., 100). The programmer must be careful not to specify a
double integer range (n > 17) for an integer variable. For example:
A=#D77000000[ 19:35]
yields the single integer value O.
Note that only the first two words of a double-precision floating variable (the exponent and first-order
mantissa) may be manipulated via this notation.

2-2

2.2

THE ASSIGN STATEMENT
General Form

ASSIGN n TO label

Examples

ASSIGN 27 TO ITEST ,
ASSIGN 10 TO LOOP

Effect

The symbolic label (a variable of type integer) represents
the specified statement number in an assigned GO TO or
arithmetic IF statement

The ASSIGN statement provides a symbolic addressing capability for the two control statements mentioned above, GO TO and arithmetic IF. The statement number assigned must be that of an executable
statement. Note that the integer variable is a symbolic label only within the context of an ASSIGN
and its associated statements and may function as an integer variable elsewhere in the source program.
Before this variable appears in such an expression, however, it must be redefined by an Arithmetic
statement. For example, the statement:
ASSIGN 20 TO IVAR
does not assign the value 20 to IVAR but the memory location associated with the source-program
statement 20. The sequence:
ASSIGN 20 TO IVAR
IVAR=20
NEWVAR=IVAR+l
is permitted, but IVAR may not be used as a label until redefined by another ASSIGN statement.

2-3

CHAPTER 3
CONTROL STATEMENTS

Statements in a FORTRAN-IV program are normally executed in the sequence in which they appear.
The user may alter this sequence in two basic ways - by invoking a subprogram (Chapter 5) which returns control to the normal sequence after execution, or via one of the control statements described
in this chapter. These are:

3.1

the GO TO statement, which transfers control to a specified statement, thus
originating a new sequence of execution;

the DO statement, which establishes an iterative sequence of statements within
the norma I sequence;

the If: statement, which specifies conditi ons for the execution of a statement in
sequence or transfer to a new sequence; and

PAUSE and STOP, which, respectively, interrupt and halt program execution.

THE GO TO STATEMENT

Three forms of the GO TO statement are described below - unconditional, computed, and assigned.
All of these forms transfer control to a statement in the source program; the difference between them is
the manner in which that statement is specified. Any GO TO statement may appear at any point in
the executable portion of the source program except as the terminal statement of a DO loop (3.2).

3.1.1

The Unconditional GO TO Statement
General Form

GOTO n

Where

n == the number of an executable statement

Example

GO TO 27

Effect

Control is transferred to statement n

The simplest form of GO TO statement, the unconditional, is a direct branch to another location in
the source program. Program execution proceeds from this point in the usual sequence.

3-1

3.1.2

The Computed GO TO Statement

Where

GOTO (n ,n , .•. ,n ),i
1 2
k
n = the number of an executable statement
i = an integer variable

Example

GO TO (3, 17, 25, 50, 66),ITEM

Effect

Control is transferred to the statement whose number is the
nith in the list. If ITEM = 2, control passes to statement
17

General Form

A maximum of 64 numbers may be listed in a computed GO TO statement. The value of the integer
variable i must fall within the range from 1 to the number of statement numbers listed. If the value
falls outside of this range, an OTS error statement is generated and control passes to the next statement
in sequence.

3.1.3

The Assigned GO TO Statement
General Form

Where

GO TO label
or
GO TO label,(n ,n ,··· ,n )
k
1 2
label = an integer variable assigned a statement number
value
n .•• n = statement numbers which the ASSIGN statek
1
ment may legally assign to label

Examples

ASSIGN 13 TO KAPPA
GO TO KAPPA
GO TO KAPPA,(l, 13,100)
GO TO KAPPA,(l,72, 100)

Effect

Control is transferred to the location specified by label

The assigned GO TO statement permits symbolic addressing of statements and execution-time modification of control transfer, for example:
ASSIGN 30 TO LOOP
20 GO TO LOOP
30 ASSIGN 45 TO LOOP
GO TO 20
In this sequence, the same statement which branched to statement 30 will next branch to statement 45.

3-2

3.2

THE DO STATEMENT
General Form

Where

Examples

Effect

DO n ~=ml,m2,m3
DO n l=m ,m
1 2
or
DO n i=m , m , -m
1 2
3
n = a statement number
i = an integer variable
m ,m ,m = variables or constants
1 2 3
DO 101=2,10,2
(iterations for 1=2,4,6,8,10)
J=1
DO 1,1=5, 1 , -J
(1=5, 4, 3, 2, 1)
DO 21=1,5
(1=1,2,3,4,5)

Statements following the DO up to and including statement n are executed repeatedly for values of i starting
wi th m l' incremented {or decremented} by m3 unti I i has
surpassed the limit m2. If m3 is not present, an increment of one is assumed.

The series of statements which are executed as the result of a DO statement are called the range
of the DO. The variable i is called the index. The values m , m , and m3 are, respectively, the
2
1
initial, limit, and increment values of the index.
If the increment variable, m , is preceded by a minus sign, it is actually a decrement. For a con3
stant m , this is simply something of the form -3. For a variable m , the value of the variable itself
3
3
must be positi ve and may be preceded by a minus sign. Thus, J= 1 and -J as on increment is correct.
J=-l with J as an increment is invalid.
The initial (m ) and limit (m ) values of a DO statement may be positive, negative, or zero provided
2
1
the difference between them is less than 131072. Positive values for m and m may be expressed as
2
1
positive integers or as variables assigned positive values. Explicit minus signs, however, may not precede the integer (constant or variable) given within a DO statement for initial and limit index values.
Negative initial and limit values must be expressed as a variable whose assigned value is negative.
For example, the statements:
DO 10 1=2, -10, -2
and
DO 101=2, -A, -2

3-3

are both incorrect since a minus sign is not permitted before the integer constants or variables given
for limit values. The series:
A= -2
DO 101=2,A,-2
is correct.
Loop termination, as indicated in the model, occurs when m3 has a value beyond the limit. For a
positive increment, this occurs when I is greater than m ; for a negative increment when I is less than
2
m . For example, the loop initiated by :
2
DO 10 1= 1 , 100,2
will not be executed when 1=101. The loop initiated by:
DO 101=100,1,-2
will be terminated when 1=0.

It is the programmer's responsibility to ensure that the limit value specified will ultimately be reached.

J = 10
K = 100
M = 10
MINUSJ = -J
MINUSK = -K
The statements:
DO 10 I=J,K,-M
DO 10 J=J,MINUSK,M
specify infinite loops which are not detected by the compiler.
The statements:
DO 10 I = MINUSJ,K,M
DO 101 = J,MINUSK,-M
specify finite loops.
The range of a DO may contain any statement with one exception. That is, the terminal statement may
not be a GO TO, RETURN, STOP, PAUSE, or numerical IF statement. A logical IF statement is permitted provided that it does not include any of the statements given above.

3.2.1

Execution of a DO Range

The processing of a simple DO range is shown below.
DO 10 1= 1,10, 1
ARRAY(I) = TAB(I*2)
10TAB (1*2) = 0

3-4

The range of the DO here consists of the two Arithmetic statements, which use the 'index variable as a
subscript index. The range may have any number of statements and the index variable may be used as
an ordinary variable provided that its value is not changed. The statement 1=1*2, for example, would
be illegal within the range given above, but TAB(I)=I is valid.
The exit from the range of the DO to the next statement in sequence is referred to as the normal exit.
In this case, the value of the index variable becomes undefined. Exit may also be accomplished by
the occurrence of a control statement within the range, leaving the index variable with its current
value available for use as a variable. Control may also be transferred from outside the range of a DO
to any statement within. For example:
DO 20 1= 1, 100
10 IF (TAB(I) .EQ. 0) GO TO 50
20 CONTINUE
50 TABO)= TAB(I+l)
GO TO 10
Here, a table is consolidated by replacing a zero entry with the next entry. Control is transferred out
of the DO loop to move the entry and returned to check for zero. Note that the above example permits branching into the range of a DO which standard FORTRAN does not permit. PDP-15 FORTRAN
considers statement 50 and the following GO TO as the "extended range" of the DO.

3.2.2

Nested DO Statements

When the range of a DO statement contains another DO statement (and its range), it is referred to as
nesting. Nesting may occur to a depth of 9. * The ranges of nested DO's must not overlap; that is,
the range of an inner DO must be contained entirely within the outer DO statement as shown in Figure
3-1. They may, however, end on the same statement.
Execution of nested DO's proceeds as follows.

Each time the outermost DO is executed for one of its

index values, the DO within it is executed for all of its index values. If this range contains another
DO, that range is executed completely for each of the values of the second-level DO. For example,
using the legal nesting example above, when range 1 is executed first for 1=1, range 2 is initiated
with K=2 and range 3 is initiated with J=l and iterated (until J>5) five times. Then range 2 is iterated
with K=3 and range 3 iterated until J>5. This process continues until K>10. That is, range 2 is repeated 9 times and each of these times, range 3 is repeated five times (a total of 45 times) while I of
range 1 is still equal to 1. At this point range 1 is repeated for 1=2 and the whole procedure recurs range 2 is done 9 times, range 3, 45 times. When I >10 for range 1, range 2 will have been performed
90 times and range 3, 450 times.
*This restriction includes any implied DO in an input-output list (Chapter 6).

3-5

Legal Nesting

Illegal Nesting

DO 50 1= 1, 10 - - - - ,

DO 50 I = 1, 10 - - - - - ,

DO 20 K = 2, 10, 1

10:

J = 2,

10'l

DO 20 K = 1,5
10

50
Figure 3-1

3.2.3

Rules for Nested DO Statements

The CO NT! NUE Statement

The CONTINUE statement is a dummy statement, which does not generate code or cause any action.
It consists simply of the text:
CONTINUE
The CONTINUE statement may appear anywhere within a FORTRAN program but is especially useful
for terminating DO loops when the last statement would otherwise be one of the illegal terminal statements listed previously. For examp Ie:
DO 10 K=ST ART, END
7 PAUSE
10 CONTINUE
Here, the user can interrupt program execution at every iteration of the loop although the PAUSE statement cannot be the terminal statement.

3.3 THE IF STATEMENT
An IF statement causes control to be transferred or a statement to be executed contingent on the value
of a test expression. Two forms of the IF statement are available - arithmetic and logical. They differ
both in general form as well as in type of expression tested.

3-6

3.3.1

The Arithmetic IF Statement
Genera I Form
Where

IF(expr)n l,n ,n
2 3
expr = an arithmetic or logical expression
n = a statement number or symbolic label established by an
ASSIGN statement

Examples

IF(COUNT) 10,20,30
ASSIGN 20 TO MID
ASSIGN 30 TO FIN
IF(AO)*B) 10, MI DI FIN

Effect

The parenthesized expression is evaluated. Control is
transferred to:
n if expr <0
1
n if expr =0
2
n3 if expr >0

As shown in the model, the Arithmetic IF statement transfers control to one of three statements according to the value of the expression given. Thus, if COUNT=3, control in the first example, would be
transferred to statement 30.

If an Arithmetic IF statement requires a continuation line, the line must be broken at a comma.
IF(E)10
10,101,102
will not compile. (The desired result is IF(E)100, 101,102.)
Logical values .TRUE. and .FALSE. have the following decimal values:
a . . TRUE.=-l
b.

.FALSE. = 0

Since logical values have specific arithmetic values, logical expressions may be used in place of
arithmetic expressions in FORTRAN statements. For example, the statement:
IF (A.GT. B.AND.A. LT. C) 3,4,999
is equivalent to the two statements:
IF (A.GT.B.AND.A.LT.C) GO TO 3
GOT04
Note that in the above example the branch to 999 will never be executed. Logical expressions, however, do not always yield the values 0 or -1; for example, in the statement:
IF (l.XOR .J) 1,2,3

3-7

the branch is made to statement:
a.2ifI=J,

3.3.2

3 if I 1 J but both have the same sign,

1 if the signs of I and J are different.

The Logical IF Statement
General Form

IF(expr )s

Where

expr = any expression
s = any executable statement except a DO or logical IF

Examples

IF(L 1 • LE. L2)GO TO 17
IF(L 1)IF(X)3,5,5
IF(L .AND. (.NOT. Ll»A=A+l
IF(I-J)A=A+ 1

Effect

If the expression is .TRUE. (or non-zero), statement s is
executed; if .FALSE. (zero), the statement is ignored

Unless the statement executed as the result of a logical IF statement transfers control (i. e., GO TO),
control continues in the normal sequence with the statement following the IF. For example, in the
statement
IF(L 1)IF(X)3,5,5
10 A=B
when L1 is . TRUE., the numeric IF is executed and control transferred to statement 3 or 5. When L1
is .FALSE., control passes directly to statement 10.
Non-logical (arithmetic) expressions are permitted within logical IF statements. In such cases, nonzero values are regarded as being logically .TRUE. and zeros as being logically .FALSE. For example,
the statement:
IF (X-3.0) GO TO 5
causes a branch to statement 5 if X 13.0 (i. e., x -3.010).

3-8

3.4

EXECUTION CONTROL

3.4.1

The PAUSE Statement
General Form

PAUSE
or
PAUSE n

Where

n = an octal interger <;;"7777778

Examples

PAUSE
PAUSE 100

Effect

Execution is suspended and the number, if any, is printed

The PAUSE statement interrupts program execution, but maintains the current state of all values. Execution may be resumed by typing CTRL P (tP) on the console teletype. The integer n, when supplied,
is printed on the console teletype and may be used to identify which of several PAUSE statements was
encountered.

3.4.2

The STOP Statement
Genera I Form

STOP
or
STOP n

Where

n = an octal integer '7777778

Examples

STOP
STOP 20

Effect

Control returns to the MONITOR after n is printed

A STOP statement is used to signify the logical end of a program. If several occur (the logical end
depending on processing results), they may be numbered to indicate where the program ended.
For example, the statements:
10 IF (COUNT .GE. 100) STOP 10
50IF (COUNT .GE. 100) STOP 50
make program termination dependent on the value of the variable count at two different points.

3-9

CHAPTER 4
SPECIFICATION STATEMENTS

Specification statements provide the compiler with information regarding the data mode, size, and, if
desired, initial values of variables in the source program. All specification statements must precede
the executable portion of the program (see Table 1-1 in Chapter 1).
Data mode specification is accomplished either explicitly via the statements INTEGER, DOUBLE
INTEGER, REAL, DOUBLE PRECISION, and LOGICAL: or implicitly according to the initial character in the variable name. The FORTRAN-IV compiler contains implicit mode assumptions, but the
user may override these via an IMPUCIT statement. In addition, the EXTERNAL statement specifies a
subprogram name which will appear in a subprogram call.
Variable size is implicit in the mode assigned to a scalar variable but must be specified in the case of
arrays so that the compiler can allocate adequate storage space. The most common way of declaring
the size of an array is the DIMENSION statement. The programmer may also control the way in which
memory is allocated via the COMMON and EQUIVALENCE statements.
Initial values may be assigned within a program via the DATA statement. A set of initial values may
also be obtained at run time by using a BLOCK DATA subprogram.

4.1

MODE SPECIFICATION

Any data mode may be specified in a mode-declaration statement as described below. If INTEGER
and REAL are the only data modes used in a program, the programmer need not have any modespecification statements since he may use the compiler's implicit mode assumptions.

4-1

4.1.1

Mode-Declaration Statement
General Form

m a ,a , .•. a
1 2
n
m = INTEGER, DOUBLE INTEGER, REAL, DOUBLE
PRECISION, or LOGICAL
a = variable name, array name with dimensions, or function name

Where

Examples

INTEGER A,B,CYZ
LOGICAL TTAB(10, 10), T ,F
REAL XYZ

Effect

Elements in the argument list are declared to be of the
given mode. An array is, in addition, allocated storage
to the dimensions given

A mode-declaration statement overrides any implicit mode assumptions. Thus, the statement:
REAL ITAB,J
overrides the basic compi ler assumption that IT AB and J are INTEGER. This rule also appl ies to any
mode assumption specified in an IMPLICIT statement (4.1.2). An item may be assigned a mode only
once in a given program.

Note that any function which has not been assigned a mode in the defini-

tion statement and which does not have an implicit mode must appear in a mode declaration statement.
Note also that arguments must have the appropriate mode, as in:
DOUBLE PRECISION B,X,DABS,DATAN
B=DAT AN (DABS (X ))
This declaration ensures the proper working of the external and intrinsic functions (Section 5.1.3)
DAT AN and DABS.

4.1.2

The IMPLICIT Statement
General Form
Where

IMPLICIT m (ll),m (12)' ... mn(ln)
1
2
m = INTEGER, DOUBLE INTEGER, REAL, DOUBLE
PRECISION, or LOGICAL
I = a list of one or more alphabetic characters and/or consecutive ranges of alphabetic characters (e.g., A-G)

Examples

IMPLICIT REAL (A-E, N,X-Z),INTEGER(F-L)

Effect

Establishes a new assumption for mode of non-declared
variables

4-2

The IMPLICIT statement governs the implicit mode assumptions for a single source program. After the
occurrence of the IMPLICIT statement shown above, for example, all variables beginning with F, G,
H, I, J, K, or L will be assumed INTEGER while those that begin with ABC
DEN X Y and
1111""
Z are assumed REAL. In this case, the compiler will not assume that all letters not specified INTEGER
are REAL. Only those listed as INTEGER are REAL. The initial mode assumption may be stated as:
IMPLICIT REAL(A-N, O-Z), INTEG ER(I-N)

4.1.3

The EXTERNAL Statement
General Form

EXTERNAL a ,a , ..• a
1 2
n

Where

a = name of a subprogram

Examples

EXTERNAL ISUM,ISUB

.
CALL DEBUS(ISUM,A, B)

.
CALL DEBUG(ISUB,A, B)
Effect

The listed symbols are defined as the names of subprograms
for use as arguments of other subprograms

A statement function (Section 5.1.1) may be passed as an argument to a subprogram without being declared EXTERNAL. If, however, a subprogram requires the name of a FUNCTION or SUBROUTINE
subprogram as an argument, the calling program must declare the name in an EXTERNAL statement.
The transmission of arguments to subprograms is discussed more fully in Chapter 5. In brief, a subprogram uses dummy symbols in statements which obtain values when called; these values must all be defined in the calling program. If these values are program variables, they are already defined within
the calling program. The EXTERNAL statement ensures that subprogram names are also defined.

4.2
4.2.1

STORAGE ALLOCATION
The DIMENSION Statement
General Form

DIMENSION a (11),a (12)" "an(ln)
1
2

Where

a = array name
I = a list of from one to three integer constants giving the
maximum value of each dimension of the array (i.e.,
column, row, plane)

Examples

DIMENSION A(100),B(50,50)

Effect

Consecutive storage is allocated; the sum of the dimensions and the array type determine the amount

4-3

Each array specification in a DIMENSION statement gives the maximum value which each of its subscripts may assume. Array dimensions may, instead, be given within a type-declaration or COMMON
statement using the same notation, as in:
COMMON ARRAY(l0,4),Y,Z
INTEGER A(10, to, 10),B
Dimensioning information should appear only once in a given program. An array which will be passed
as an argument to a subprogram must be declared in both the subprogram and the call ing program; with
dummy array names and with real array names, respectively (see Chapter 5).

4.2.2

The COMMON Statement
General Form

COMMON/b l/vl istl/b2/vlist2/' ..

Where

b = blank or a label
vlist = a list of variable and arrays

Examples

COMMON/BLKA/X(3,3), Y(2,5)
COMMON A,B,C/XX/X, y,Z//D,E

Effect

The items in each vlist are allocated to the specified block
of memory where they may be shared by other programs

The COMMON statement allows the data of a main program and/or its subprograms to share a common
storage area. A common area may be divided into separate blocks. In this case, they are distinguished
from one another by a label which is a unique variable name. One block may be left unlabeled; this
area is referred to as blank COMMON. If the first block referred to in a COMMON statement is a
blank COMMON block, the slashes may be omitted, as shown in the second example above. A blank
COMMON area is otherwise denoted by two consecutive slashes.
Items are assigned to a COMMON block in the order in which they appear in a COMMON statement.
All items to be stored in a given block need not be listed at once, however, but may be given later in
the same statement, as in:
COMMON/BLK 1/A, B,C/BLK2/X, Y, Z/BLK 1/M, N,O
or in a subsequent COMMON statement as in:
COMMON/BLK 1/COUNT/BLK2/T/BLK 1/W,F
COMMON/BLK3/ARRAY(1 O)/BLK 1/G ,R
Entries are linked sequentially so the items assigned to BLK 1 will be A, B, C, M, N, 0, COUNT,
W, F, G, and R, in that order.

4-4

The sharing of COMMON blocks is made possible by the fact that storage is allocated at the same location for blocks of the same name in all programs executed together. For example, if PROG1 contains
the statement:
COMMON A,B;R;X, Y,Z
and SUBPROG 1 has:
COMMON/R;'tJ, V,W//D,E,F
the variables A and D will share the same location in blank COMMON as will the variables Band E;
likewise, COMMON block R will store X and U, Yand V, and Z and W at the same locations. A
COMMON block may be of any length but no program may attempt to enlarge a block declared by a
previousl y compi led program.

4.2.3

The EQUIVALENCE Statement
Genera I Form

EQUIVALENCE (11),(12)'··' (In)

Where

1 = a list of two or more variables or array elements with
constants as subscripts

Examples

EQUIVALENCE(RED, BLUE,GREEN,COLOR)
EQUIVALENCE(RED, BLUE), (BLUE, GREEN,COLOR)

Effect

The elements of each I are assigned to the same storage
location

Note in the second example that the relation of equivalence, once applied to a variable, holds for
subsequent equivalences.
Variables located in a COMMON block may be made equivalent to other variables but not to one
another. Quantities placed in a COMMON block via an EQUIVALENCE statement which change the
size can only occur at the current end of the block.
For example, the statements
COMMON/BLK 1;X, Y,Z
DIMENSION A(4)
EQUIVALENCE(A, Y)
cause BLK 1 to extend from X to A(4), arranged as follows:

X
Y
A(l )

)

same location

Z
A(2)
A(3)
A(4)

)

same location

4-5

Note that if, for example, A(l) were previously made equivalent to a variable, M, that Y, A(l), and
M would all share the same location.
In the following example, three arrays of different dimensions occupy the same storage locations.
DIMENSION A(lOO),C(50), D(200)
EQUIVALENCE(A,C, D)
Here, of course, only the first 50 locations apply to all three, as shown.
location 1 A(l), C(l), D(l)

location 51 A(51), D(51)

location 101 D(101)
location 200 D(200)
In the following sample program, arrays containing different numbers of subscripts are made equivalent
using the EQUIVALENCE statement.
C
C
C
C
C

EQUIVALENCE TWO ARRAYS SUCH THAT THE FOLLOWING
PAIRS OF SUBSCRIPTED VARIABLES WILL OCCUpy THE SAME
MEMORY LOCATIONS;
A(l)ANDB(l,l)
A(2) AND B(2, 1)

DIMENSION A(2), B(2,2)
EQUIVALENCE (A, B)
C
C
C
C

FOLLOWING DATA STATEMENT WILL SET B(l,1) AND B(2,1) TO
THE VALUES 3.0 AND 4.0 AS WELL.

C
DATA A(l),A(2)/3.0,4.0/
C
C
STOP
END
All variables equivalenced to COMMON variables are treated as COMMON variables with regard to
subsequent EQUIVALENCE statements. EQUIVALENCE statements which would require extension of
the beginning of a COMMON block are not allowed, as, for example:
COMMON/BLK1/X, Y,Z
DIMENSION A(4)
EQUIVALENCE(X,A(3))

4-6

4.3

THE DATA STATEMENT
General Form

OAT A vlist1/clistl,vlist !clistl, ... vlist/clist/

Where

vlist = a list of variables, array elements, or array identifiers (no dummy variables permitted)
clist = a list of constants with optional signs or a construction of the form n*c where n is the number of
variables for which c is to be assigned

Examples

DATA A,B,C/1,2,3/
DATA A, B,C,D,E,F/1 ,2,3,3*0/

Effect

Each listed constant is assigned as the value of the corresponding variable

Any type of constant may appear in a "clist". A double-precision constant must be written explicitly
in D format (e.g., 1.0D+01 or 10+01 - not 1.D+01). The mode of the variable and the constant to be
assigned must agree with one exception - an integer constant may be assigned to a double-integer
variable.
The values specified in a DATA statement are compiled into the object program and become the values
assumed by the listed variables when program execution begins. DATA statements may be given in
BLOCK DATA subprograms described below. Note that COMMON variables may not be initial i zed
using DATA except in the context of a BLOCK DATA subprogram; in this context the COMMON statement ihelf may also include data as in:
COMMON/B 1/X, Y(5),I;X, I, Y(3)/2.0, 3,5.0/
Array identifiers without subscripts may be specified in the I ist of variables in a OAT A statement. The
occurrence of an unsubscripted array identifier is equivalent to the occurrence of all of the elements of
that array listed in ascending order. Note that in respect to the I imitations of the number of variables
in the variable list permitted each version of FORTRAN (refer to Appendix C) an unsubscripted array
identifier counts as one variable. For example:
DIMENSION I (3),A(200)
DATA I,J ,A, B/1, 2,3, 4, 201 *3. 14/

4-7

4.4

BLOCK DATA SUBPROGRAMS

A BLOCK DATA subprogram is used to enter data into a single labeled COMMON block at run time;
it is identified by a statement consisting of the text:
BLOCK DATA
Between this and the END line is the body of the subprogram which may contain only DATA, COMMON,
EQUIVALENCE, DIMENSION, and TYPE statements.
A BLOCK DATA subprogram may not be used to initialize variables in a blank COMMON block area.
When a given labeled block is initialized in this manner, all elements of it must be listed in a
COMMON statement within the subprogram even if they do not appear in a DATA statement.
More than one COMMON block may be stated in a COMMON statement but only the last one can be
initialized with DATA statements. The following program, for example, will not give a compilation
error but will not function properly.
BLOCK DATA
COMMON/N1/I/N2/J
DATAI,J/1,2/
END
Two BLOCK DATA subprograms are required to initialize the two blocks, as shown below:
BLOCK DATA
COMMON/N1/I
DATA 1/1/
END
BLOCK DATA
COMMON/N2/J
DATA J/2/
END
A more extensive example of a BLOCK DATA subprogram is given below.
BLOCK DATA
DIMENSION X (4), Y(4)
COMMON/NAME/A, B,C,I,J,X, Y
DATA A,B,C/3*2. 0/
DATA X(1),X(2),X(3),X(4)/0.0,0. 1,0.2,0.3/,
1Y(l), Y(2), Y(3), Y(4}/1. O,L21, 1. OE-4, O. 2/
END

4-8

CHAPTER 5
SUBPROGRAM STATEMENTS

A subprogram is a program which is invoked by name from other programs whenever the operations it
performs are required. It is a convenient and efficient means for encoding frequently used or complex
operations since the statements appear only once in the object program regardless of the number of
times they are used. In addition, a subprogram may be designed to handle a variety of different values
which may be transmitted as arguments whenever the subprogram is invoked. The process of establishing a subprogram is referred to as subprogram definition; the statements to be executed are referred to
as the body of the subprogram; the process of invoking the subprogram and transmitting arguments is
referred to as a subprogram call.
Two basic types of subprogram* may be defined and called in FORTRAN-IV - functions and subroutines.
A function is a subprogram which always returns a result - a single numeric value. This value is, by
convention, represented in an expression by the function call. A subroutine may return several or no
values; when results are obtained, they are assigned to variables in the calling program. The programmer may call a variety of predefined subprograms from the science library. {Refer to "Operating
Environment" manual DEC-15-GFZA-D for a description of the science library.}
The transmission of arguments to a function and to or from a subroutine requires the use of dummy
variables in the definition. That is, those variables in the subprogram which are to acquire calling
values {real arguments} are listed in parentheses following the subprogram name in the subprogram definition statement. For example, if a subroutine were designed to merge two arrays {one with 500
elements and the other with 100} into a third array, the arrays might be referred to by dummy variables
A, B, and C in the subroutine statements which refer to them, such as:
IF{A(I} .GE. B(I)}C(I}=A{I}
These dummy variables would appear in the definition statements as follows:
SUBROUTINE MERGE(A,B,C)

*BlOCK DATA subprograms, described in Chapter 4, are a third type of subprogram in that they may
by comp i led separate I y.
5-1

When the subroutine is called, the user indicates the actual arrays to be merged, as in:
CALL MERGE(FILA,FILB,FILC)
The statement shown above then becomes:
IF(FILA(I) .GE. ALB(I))FILC(I)=FILA(I)
Note that the real arguments must appear in the same order as the corresponding dummy variables in
the subprogram definition. Real arguments must also agree in mode with the dummy arguments to which
they correspond. When arrays are involved, DIMENSION statements must be given for the real arrays
in the calling program and for the corresponding dummy arrays in the body of the subprogram. In the
above example, the following DIMENSION statements would be required:.
Ca Iii ng Program

Subprogram

DIMENSION FILA(500), FILB(l 00), FILC(600)

DIMENSION A(500),B(lOO),C(600)

Figures 5-1 and 5-2 show a main program and a subprogram which it invokes, respectively.

5.1

FUNCTIONS

Two types of function may be defined in FORTRAN-IV - statement and external. A statement function
is defined within another program via a form of the Arithmetic statement and may be called only by
the program in whi ch it is defined (except when passed as an argument to a subprogram). An external
function, defined via the FUNCTION statement, is an independent subprogram which may be called
by any program. Both types of function must have names which conform to the rules for variable names
and require at least one argument. A function call may appear only within an expression and, like
other elements of an expression, must have an associated mode.

5.1.1

Statement Functions
General Form
Where

f(a, ,a , .•. an)=e
2
f = function name
a = dummy argument
e = an expression

Examples

SUM(A, B, C)=A+B+C
FUNC(A, B)=2. *A/B
A(X)=3. 2+SQRT (X)

Effect

Defines function f(a"a , ..• a ) to have the value of
2
.
n
expression e

5-2

C
C
C
C
C
C
C
C
C
C
C

THIS MAIN PROGRAM CALLS SUBR TO MAKE ALL ELEMENTS OF
THE ARRAY ICOL POSITIVE RELATIVE TO THE
SMALLEST ELEMENT OF THE ARRAY. THE ELEMENTS OF THE
ARRAY ARE ASSUMED > OR = TO ZERO.

A COLUMN OF DATA IS READ INTO ICOL VIA LOGICAL 1. THEN THE
SMALLEST ARRAY ELEMENT IS FOUND. THE ARRAY NAME AND SMALLEST
ELEMENT IS PASSED TO SUBR. SUBR SUBTRACTS THIS
SMALLEST ELEMENT FROM EVERY ELEMENT OF THE ARRAY.

DIMENSION ICOL(20)
C
C

C
C
99
2

READ IN COLUMN OF DATA
READ(1, 2)ICOL
FORMATM)

C
C
C
C

FOLLOWING LOOP FINDS SMALLEST ARRAY ELEMENT AND
REMEMBERS ITS SUBSCRIPT IN JMIN.

6
4

JMIN=l
DO 4 1=1, 19
IF(ICOL(JMIN)-ICOL(I+ 1))4,4,6
JMIN=I+1
CONTINUE

C
C

CALL SUBROUTINE TO PERFORM THE SUBTRACTION OF ICOL(JMIN)
FROM EVERY ELEMENT OF ICOL.

C
C

CALL SUBR(ICOL,ICOL(JMIN))
C
C

LIST MODIFIED ELEMENTS OF IeOL VIA LOGICAL 2.

5
C

WRITE(2,5)ICOL
FORMAT(lX,I6)

GO TO 99
END
Figure 5-1

Main Program Sample

5-3

C
C
C
C
C
C
C
C
C
C
C
C
C
C
C

SUBROUTINE SUBTRACTS VALUE OF ONE ELEMENT FROM ALL THE
ELEMENTS OF A ONE DIMENSIONAL INTEGER ARRAY.
CALLING SEQUENCE IS:
CALL SUBR(ARRAY,ARRAY ELEMENT)
DUMMY VARIABLES OF SUBROUTINE SUBPROGRAM ARE IDUMA,IDUMAE
WHERE IDUMA CORRESPONDS TO ARRAY NAME AND IDUMAE
CORRESPONDS TO ARRAY ELEMENT TO BE SUBTRACTED.
IDUMA MUST BE A 20 ELEMENT SINGLY DIMENSIONED INTEGER
ARRAY.
DECLARE SUBROUTINE AND ITS DUMMY VARIABLES

SUBROUTINE SUBR (IDUMA,IDUMAE)
C
C
C
C
C
C
C

-,
DIMENSION IDUMA
DIMENSION IDUMA(20)

C
C
C
C
C
C
C
C
C
C
C
C

-..,

FOLLOWING LOOP SUBTRACTS VALUE OF ARRAY ELEMENT PASSED
TO SUBR FROM EVERY ARRAY ELEMENT OF THE ARRAY PASSED TO
SUBR.
SET ELEMENT INTO TEMPORARY. CANNOT USE IDUMAE IN LOOP SINCE
ELEMENTS OF THE PASSED ARRAY ARE ALL MODIFIED

I DECR=I DUMAE
C
C
C
C
DO 11=1,20
I DUMA(I)=IDUMA(I)-IDECR
RETURN
END
Figure 5-2

Subprogram Sample

S-4

A statement function definition is a single non-executable statement in a FORTRAN source program. It
may not precede any specification statement but must precede any executable statement of the program
in which it appears.
The expression which defines a function may include dummy variables, ordinary variables, nonHollerith constants, and previously defined external or statement functions. The function name may
be associated with a data mode by any of the conventions specified for variables. The dummy variables
used to define the function may appear in a specification statement but in no other context. Up to 10
dummy arguments may be used in a single definition.
To call a statement function, the programmer simply includes the name, with real arguments as required,
in an arithmetic expression. For example, assume the definition:
OFFSET(A,B)=A+B+ 100
where A and B are dummy variables. The function might be called as follows:
TABO)= 100+0FFSET (CT 1, CT2)
which yields the same result as:
TABO)=200+CTl+CT2
A statement function, in effect, permits the programmer to extend the symboli c language available to
him. By assigning a mnemonic name to a frequently used calculation, he obtains a more readable
source program. For example, one might define the integer function percent as:
PERCENT(I, J)=(I*l oo)/J
This function is used in the following statements to express the values in TABl as a percentage of each
of the values in TAB2.
DO 10 K=l, 10
TOT=TAB2(K)
DO 101=1,100
VAL=TAB10)
J = 100*(K-1}+1
10 TAB3(J)=PERCENT(VAL, TOT)

5-5

5.1.2

External Functions
General Form

m FUNCTION f(a1 ,a2,· •. a n )

Where

m = an optional mode specification
f = function name
a = dummy argument

Example

INTEGER FUNCTION TOT(A,B)
DIMENSION A(100)
TOT=O
DO 101=1 ,B
10 TOT=TOT+A(I)
RETURN
END

Effect

Defines an external function

An external function is an independently written program which is executed when its name appears in
an expression in another program. The optional mode specification (m) may be INTEGER, DOUBLE
INTEGER, REAL, DOUBLE PRECISION, or LOGICAL. If no mode is given the function is associated
with a mode in the same manner as a variable.
The function name must conform to the rules for variable names and must appear at least once as a
variable within the body of the function; that is, the function name must be defined before control is
returned to the calling program. Only non-subscripted variable names may appear as arguments in an
external function. If an array name is used, an argument must appear in a DIMENSION statement
within the subprogram. At least one argument must be specified for all functions.
The body of a function may contain any FORTRAN statement with the exception of BLOCK DATA,
SUBROUTINE, FUNCTION, or a statement containing any reference to itself. It must contain at
least one RETURN statement, which signifies the logical end of the subprogram. When a RETURN
statement is executed, control is returned to the calling program. The function name must have been
assigned a value before the occurrence of a RETURN.
An external function is called when its name and arguments appear in an arithmetic or logical expression. Real arguments may be variables, array elements, array names, any other expression, or the
name of an external function or subroutine. A sample external function is given below.
INTEGER FUNCTION CODE(J)
COUNT=l
DO 20 1= 10, 100, 10
IF(J .LT. I)GO TO 30
20 COUNT=COUNT+1
30 CODE=COUNT
RETURN
END

5-6

This function encodes values of J as follows: If J has a value between 0 and 9, it is encoded as one; if
its value is between 10 and 19, it is encoded as 2 and so on up to 100. It is invoked by the following
statements:
DO 30 1= 1, 100
YAL=TAB(I)
IF(YAL . LT. O)YAL=ABS(J)
30 TABO)=CODE(YAL)
Negative numbers are avoided by calling the library function ABS which returns the absolute value of
its argument. A table, TAB, is converted here to contain the codes associated with its original values.

5.1.3

DEC Library Functions

Two types of predefined functions are part of the FORTRAN Science Library - intrinsic and external.
Both types are called in the same manner as a programmer-defined external function, as in:
Y=2*SQRT (C)
and
J=2+IFIX(R)
where SQRT is an external library function and IFIX is an intrinsic library function. In each case, the
value of the expression will, if possible, be converted to the mode of the variable. (All library functions are listed and described in the "Operating Environment" manual DEC-15-GFZA-D).

5.2

SUBROUTINES

A subroutine is defined external to the program which invokes it in a manner similar to an external
function. The important distinctions between the two are: a subroutine need not have any arguments
at all but will accept numerical, logical, or Hollerith constants as arguments; a subroutine need not
return any values to the caller but may return several; and a subroutine is called via an explicit CALL
statement.

5-7

5.2.1

Subroutine Definition
General Form

SUBROUTINE name(a ,a , ••• a )
1 2
or
n
SUBROUTINE name
(FORTRAN statements)

RETURN
END
Where
Example

Effect

a = dummy argument
SUBROUTINE STORE(A, B)
DIMENSION A(100), B(100)
DO 10 1= 1,100
10 A(I)=B(I)
RETURN
END
Defines an external subroutine

A subroutine name must conform to the rules for variable names and may not appear within any statement
in the body of the subroutine. A subroutine's arguments may represent any FORTRAN expression (this
includes any constant, variable, array element, or an external subroutine or function name). Dummy
arguments may represent values supplied by the calling program or values returned by the subroutine.
In the latter case, the dummy variable must appear on the left side of an Arithmetic statement or in an
input list within the subroutine body.
For example, if the function CODE given in Section 5.1.2 were defined as a subroutine rather than as
a function, it would have the following form:
SUBROUTINE CODE(J,C)
C=l
DO 201=10,100,10
IF(J .LT. I)GO TO 30
20 C=C+1
30 RETURN
END
The statements which invoked the function would be modified to invoke the subroutine as follows:
DO 301=1,100
VAL=TAB(I)
IF(J . LT. O)J=ABS(J)
CALL CODE(VAL, TAB(I))
An array name argument must appear in a DIMENSION statement. EQUIVALENCE, COMMON, and
DATA statements are permitted but may not include any dummy variables.

5-8

The logical termination of a subroutine is signal led by a RETURN statement. The physical end is indicated by an END line.

5.2.2

Subroutine Calls
General Form

CALL subr(a 1 ,a , .•• an)
2
or
CALL subr

Where

subr == the subroutine name
a == a real argument

Examples

CALL COMPUTE(A,B,ANS)
CALL ERROR

Effect

Control is transferred to the subroutine; real arguments, if
any, are substituted for dummy variables at execution

The arguments of a CALL statement may be of any type but must agree in number, order, type, and,
in the case of arrays, dimensions with the corresponding arguments in the SUBROUTINE statement of
the called subroutine. When subroutine execution has been completed (a RETURN has been encountered),
control returns to the statement following the CALL.

5.3

MULTIPLE ENTRIES AND RETURNS
General Form

SUBROUTINEjl=UNCTION
ENT~Y name(a ,a , ..• a )

END
Where
Examples

Effect

name = the entry name
a = a dummy argument
SUBROUTINE MOVE(A, B,C)
DIMENSION A(lOO), B(100),C(100)
DO 10 1= 1 , 100
10 C(I)=B(I)
ENTRY MOVEl (A, B)
DO 201=1,100
20 B(I)=A(I)
RETURN
END
A call to the name associated with ENTRY will invoke a
portion of the subprogram in wh i ch it occurs

An ENTRY statement establishes a separately callable subprogram within a subroutine or external function. Thus, as in the model, the programmer bypasses one of the operations of a subroutine or function

5-9

by calling MOVEl rather than MOVE. An ENTRY statement may not occur within a DO loop. The
dummy arguments of an ENTRY may appear in the body of the subprogram prior to the entry statement
only if they are also arguments of a prior ENTRY or of the SUBROUTINE or FUNCTION definition.
In a multiple-entry function, the return value is always assigned to the function name itself. For example:
INTEGER FUNCTION ADD10(A)
DIMENSION A(10)
ADD10=0
J=10
10 DO 20 1= 1, J
20 ADD10=ADD10+A(I)
RETURN
ENTRY ADD5(A)
DIMENSION A(5)
ADD10=0
J=5
GO TO 10
END
The user calls ADD5, as in
SUBTOT=ADD5(TAB)
which is given the value assigned to ADDlO in the body of the function.
Multiple returns may be specified via the construction:
RETURN I
where I is an integer variable for which a legal statement number will be substituted when the subroutine or function is called. The variable I appears in the real argument list preceded by @, as in:
CALL COMP(X, Y,@lO)
The use of multiple returns is illustrated below.
SUBROUTINE COMPARE (A, B,I,J)
IF(A • EQ. B)RETURN
IF(A .LT. B)RETURN I
RETURN J
END
Here the user need not examine the result of the comparison before transferring control but control is
automatically transferred by the subroutine.

5-10

CHAPTER 6
DATA TRANSMISSION STATEMENTS

Data transmission statements control the transfer of data between internal storage and peripheral devices.
The numerous details involved in this process, while greatly simplified in FORTRAN-IV, require that
the programmer be familiar with the way in which data are stored externally and, to varying extent,
with the way data are stored internally. In general, the external data may be thought of as a continuous string of information (logical record) which consists of one or more physical records on the device
being used as a particular instance. The size of a physical record varies from device to device and,
while not specifically mentioned in any data transmission statement, influences the selection of a particular type of logical record. Table 6-1 describes the type of physical record, for each device,
which is associated with each of the two types of logical record - formatted and unformatted.

Table 6-1
Physical Record Description for Formatted and Unformatted Records
Device

Formatted

Unformatted

Teletypewriter

one line of up to 72 characters
terminated by carriage return

not applicable

Line printer

one line of up to 80, 120, or
132 characters

not applicable

Card reader

one card (up to 80 characters)

not applicable

Paper tape reader and punch

one typewritten line image

50 words

Magnetic tape

one line image of 628 characters

251 words

Disk/DECtape

one line image of 628 characters

251 words

The terms formatted and unformatted refer to the way data are stored externally since this governs the
techniques used in input/output. A formatted record is an ASCII character string which is interpreted
on input and constructed on output according to conversion rules defined in a FORMAT statement. An
unformatted record is a string of 18-bit words which may be stored into and read from program variables
specified in an input/output list in a format determined by the compiler.

6-1

PDP-15 FORTRAN-IV provides a third way of treating data called Data-Directed I/O which uses
default FORMATS for output and converts data on input to conform to the program variables which
will contain them. A logical record need not conform in size to the size of a physical record associated with a particular device as the input/output statements which accomplish data transmission will
automatically read or write the quantity of physical records required to accommodate the logical
record.
A file (a complete collection of related logical records) is identified in a source program by a logical
device number, an integer constant associated with a particular type of device in the monitor Device
Assignment Table* (DAT). At run time, this number is assigned to a physical unit and all program references to it interpreted accordingly.
There are two types of input/output which may be performed via FORTRAN-IV I/O statements - sequential and direct access. Sequential statements direct records to and from memory in the sequence
in which they are physically recorded on the device in question.

Direct access (6.3.3) statements

permit the sequential transmission of logical records to and from a direct-access device where the
physical records are not stored sequentially.
In addition to the statements which perform input/output or describe a file for direct access, FORTRANIV provides a number of auxiliary data transmission statements. The most important of these, the
FORMAT statement, provides program control over the conversion of data between program-required
and device-required forms. Other auxiliary statements control mechanical aspects of device control
such as opening and closing files and the ENCODE and DECODE statements permit conversion between
formatted and unformatted (ASCII-binary) information in memory. ENCODE and DECODE permit the
programmer to build a logical record in core, and to transmit data from this record into program variables.
6.1

THE FORMAT STATEMENT
General Form

n FORMAT(sl,s2'" .sn)

Where

n = statement number
s = field specification of the form: nkw.d
n = number of successive fields involved
k = type of conversion**
w = field width
d = number of places to the right of the decimal point

Effect

FORMAT n is establ ished as a reference for formatted I/O
operations

*Refer to the FORTRAN "Operating Environment" manual, DEC-15-GFZA-D for a detailed description of the Device Assignment Table.
**The argument k, which characterizes the type of FORMAT, is always required; others may be optional, depending on the value of k.

6-2

A FORMAT statement describes one or more records to be read or written. It may be used with or without an input/output list. Without the list, data are input to and output from the FORMAT specification
itself. Otherwise, the listed variables correspond to the list of field descriptions and the appropriate
conversions performed.

6. 1 • 1

Statement Syntax

A FORMAT statement may describe one or more physical records. If more than one is described, the
character (/) denotes the end of a record, as in:
10 FORMAT (s1' s2/s 1' s2' s3)
'--..--J~

Record
1

Record
2

When a number of slashes appear at the beginning or end of a statement, that number of records will be
skipped on input and blank records written on output. When a sequence of slashes occur within the
statement, one is considered to be the record indicator and the rest interpreted as above. Thus:

spec ifies three skipped or blank records before record 1, one after record 1, and three after record 3.
Each record description, as shown above, may consist of one or more field descriptors, a field being a
consecutive series of characters within the record. Field descriptors are, as shown above, separated
by commas. If a particular descriptor applies to a set of consecutive fields, it may be preceded by a
number indicating the number of fields to which it applies. For example, 3s1 applies the descriptor s1
to three consecutive fields. In addition, a group of field descriptors may apply to a set of consecutive
fields. In this case, the group of descriptors is enclosed in parentheses and preceded by a group count
indicating the number of times it will be used. The statement:
FORMAT (s1, s2, 3{s3, s4))
divides the record into eight fields associated with descriptors as follows:
S 1 - field 1
S2 - field 2
S3 - field 3) 1)
S4 - field 4
S3 - field 5 ) 2)
S4 - field 6
S3 - field 7) 3)
S4 - field 8

6-3

6.1.2

Field Descriptors

The form of a field descriptor is as given in the statement model:
nkw.d
where n is an optional repeat count, k is a control character specifying conversions to be performed or
special formatting functions, w is field width, and .d indicates number of decimal places to the right
as required. Table 6-2 below summarizes the control characters and their meanings.

Table 6-2
Field Descriptor Control Characters
Conversion
Control

Output Field

Input Field

decimal integer

integer or double integer variable

floating-point, scaled

real variable

floating-point, mixed

real variable

floating-point, mixed/scaled

real variable

floating-point, scaled

double-precision variable

Tor F

logical variable

a Iphanumeri c

ASCII characters

alphanumeric, right-adjusted

ASCII characters

Special
Purpose

Meaning

sets scale factor for E, F, D conversions

skips characters on input or output blanks

I I
H
}

designate Hollerith field

octal field

tab to specified position

Each of these descriptors is described in detail below. All conversion descriptors must specify the field
width. Note that the field width must be large enough to contain all characters {including decimal
point and sign} required to constitute the data value and any blank characters needed to separate it
from other data values. Note that during input, an inadequately sized input field enables only the
leftmost (most significant) characters of each item input up to the width specified to be input; all other
input characters are lost. If during output, a variable (the value of which exceeds the format specified

6-4

field width) is found; a string of asterisks (*) including any indicated decimal point is output in its
place. For example, if the output field specified is 14,- the integer 12345 is output as ****.
The interaction of the FORMAT descriptors with items in a standard or data-directed I/O I ist are described in Section 6.3. An output FORMAT for a printing device must allow for the fact that the first
character of a record is used as a carriage control indicator. Carriage control is described later in
this chapter; it is ignored in the examples given below to clarify the conversion descriptors illustrated.

I-TYPE CONVERSION
General Form:

Iw or nlw
Examples (b indicates blank)

Descriptor

Input

Output

Internal

---

bbbbb

+00000

bbbbO

-b5

-05

b-5

bbb12345

+12345

bbb12345

113

bbb1234567890

+1234567890

bbb1234567890

I-conversion descriptors are used for both single and double integer variables. On input, the numbers
specified by ware converted to a binary integer before being stored. A minus sign is required for negative numbers; plus is optional. The field may not contain a decimal point. Blanks may precede the
sign or first digit of a number; embedded blanks are interpreted as zeros.
On output, a binary integer is converted to a string of numeric characters and right-justified in the
field. A minus sign is provided for negative numbers. If the number does not fit within the field, a
string of asterisks is output instead.
The following examples illustrate the use of the I descriptor. Program variables K 1, K2, K3, and K4
will obtain values via the input FORMAT and their values will be output via the output FORMAT. The
two format statements are given below - fa input:
10 FORMAT (15,13,18,12)
and for output:
15 FORMAT (lX,14,12,14,Il)
Note that the output format specification contains a carriage control indicator (lX). The input data
is the string below:

6-5

12345b99bb12345689
~'-./

which will be interpreted in the fields shown and will have the following internal values (in octal);
K 1 = 030071
K2 = 000143
K3 = 361100
K4 = 000131
Using the output FORMAT given to output these variables gets the following printout:

/i~~

K1
(14)

K2
(12)

K3
(14)

K4
(11)

Only the second value, as shown, fits into the allotted field width. Unpredictable results may be obtained when a number is too large for an internal integer but fits in the field width. For example, the
integer 131073 (stored internally as 400001) is printed out -131071 if the output field width is large
enough.

E-TYPE CONVERSION
General Form:

Ew.d or nEw.d
Examples

Format Descriptor

Input

Internal

Output

Ell.4

00. 2134E03

213.4

bO.2134E+03

E9.2

0.2134E02

21.34

bO.21E+02

E10.3

bb-23.0321

-23.032

-0. 230E+02

The input format of an E-type number is a string of digits optionally preceded by a sign. A decimal
point and an exponent may be included in the string. The number itself should be restricted to 11
35
digits to ensure that the number, if given as an integer, will not exceed 2 _1 in magnitude. Results
are otherwise unpredictable. The input string is converted to a floating-point number with d spaces
reserved for digits to the right of the decimal point.
E-type output of a floating-point number consists of a minus sign for negative numbers followed by the
digit zero, a decimal point, d significant digits, and an exponent of the form E~nn, as shown in the
model. The field width must be at least d+7 characters long to accommodate this notation.
E-conversion is illustrated in the following examples.

6-6

Example 1.
I nput format:

10 FORMAT(E6.0,E9.3, El0.3,E16.4)

Output format:

20 FORMAT(1X, E6.0,E9. 3,E1 0.3, E16.4)

Input data:

-1.E2b-0. 12E-01-0. 123E-00-123456.
7891 E+01J
\
__
E6.0

E9.3

E10.3

E16.4

Program variables:

They are stored internally, each in two words of storage, as follows:
R1

000007
710000

227772
704467

704467
662775

773716
374025

Output of these variables using the given format will give the printout:
*.*****.*******-0.123E+00bbbbb-0.1235E+07
Rl
(E6.0)

R2
(E9. 3)

R3
(E10.3)

R4
(E16.4)

Note that the output field was not large enough to accommodate the E-format of output, for R1,
-0.E+03[W=7], or R2, -0.120E-Ol(W=10).

Example 2.
Input format:

10 FORMAT(E12. 5, E7. 2, E10. 2, E12.4)

Output format:

20 FORMAT (lX, E12. 5/1X, E12. 5/1 X , E20. 9/1X, E30. 10)

(Note the carriage control character preceding each output value.)
Input data:

123456.000000012345-. 01E+03bb+loo. OE-2 )
Rl

Output:

bO. 12346E+06
bO.12345E+03
bbbb-0.100000000E+02
bbbbbbbbbbbbbbO.10000oooo0E+01

6-7

F-TYPE CONVERSION
General Form:

Fw.d or nFw.d
Examples

Format Descriptor

Input

Internal

Output

a) F6.3

bl3457

13.457

b) F6.3

313457

313.457

** ***

c) F9.2

-21367.

-21367.00

F-type conversion is the same as that described for E-type conversion on input. On output, however,
the number is written with a minus sign if negative, an integer portion, a decimal point and the fractional part rounded to d significant digits. Note in the example an instance (item b) where the variable
exceeds the output field width. It is important, on output, to have a field width large enough for a
leading zero for values less than 1.0. For example, the descriptor F4.2 cannot output the value -.12
which must be printed as -0.12.

G-TYPE CONVERSION
General Form:

Gw.d or nGw.d
Examples

Format Descriptor

Internal
. 12345678x1 0- 1
0
. 12345678x 10
4
.12345678x10
8
.12345678x10

G14.6
G14.6
G14.6
G14.6

Output
bbO.123457E-0
bbO. 123457bbbb
bbb1234.57bbbb
bbO. 123457E+08

G-type conversion on input is the same as for E and F described above. On output, the format used
depends on the magnitude of the internal data. If the exponent of the normalized value is greater than
the number of decimal places specified in the output format (d), E-type output format is used. Otherwise, a modified F-type notation is used; that is:
F{w-4). {d-e),4X

[e = value of exponent]

where 4X indicates four blanks to be appended to the val ue. The programmer must be sure to provide a
field width sufficient to include these. E format is always used for values less than 0.1.
Some examples of G-conversion follow.

6-8

Example
Input format:

10 FORMAT(G 16. 5, G 16. 5,G 16.4; G 16.4)

Output format:

20 FORMAT(lX, G 16.5/1X,G 16. 5/1X, G 16.4/1X,G 16.4)

Input data:

.12345E+5
.12345E+6
.12345E+4
.12345E+5

Output:

bbbbbb12345.bbbb
bbbbbO.12345E+06
bbbbbbb1234.bbbb
bbbbbbO.1234E+05

D-TYPE CONVERSION
General Form:

Dw . d or n Dw •d
Examples
Input

Internal

Output

Format Descriptor

D13.6

bb+21345D03

21.345

bO.213450D+02

D13.6

b+3456789012

3456 . 789012

bO.345689E+o4

D12.6

-12345.6D-02

-123.456

* **********

In D-type conversion, an input field conforming to the format of an E-type input field is converted to
a double-precision floating-point number. The output format is also the same as for E-type output,
with the exception that the E is replaced by a D.
Example 1.
Input format:

10 FORMAT(D30. 12)

Output format:

20 FORMAT(lX, D30.12)

Input data:

34359738367.*
34359738370.
34359738371 .
. 34359738367*
-34359738367.0000000000D+02
-34359. 738367D+2*

Output:

bbbbbbbbbbbbO.343597383680D+ll*
bbbbbbbbbbbbO.200000000000D+Ol
bbbbbbbbbbbbO.299999999986D+01
bbbbbbbbbbbbO.343597383680D+OO*
bbbbbbbbbbb-0.243597383669D+03
bbbbbbbbbbb-0.343597383657D+07*

*Starred values have significant digits within the limit (34359738367) representable without truncation.

Others obtain meaningless values .
6-9

Example 2.
Input format:

10 FORMAT(D 15.9, D13.3, D20. 10, D16.4)

Output format:

20 FORMAT(lX, D15. 9, D13. 3, D20. 10, D16.4)

Program variable:

Dl ,D2, D3, D4

Input data:

-1234.56789D+12-1.00DObbbbbb34359738367 )

\..

)~~V

Dl
Output:

----_

*.*************bbbbO. 100D+OlbbbbO.3435973836D+03bbbbbbO.OOOOD+OO
\.

)

...../

P-SCALE FACTOR
General Form:

nP or -nP
Examples

Format Descriptor

Input

Scale Factor

Internal

Output
---

lPD10.4

12.3456

+1.23456

1.2345D+00

-3PF6.3

123.456

-3

123456

123.456

The control character P indicates a scale factor (n or -n) to be applied to E-, F-, G-, and D-type
conversions, for which a scale factor of zero is assumed if P is not present. When a scale factor P
occurs in a statement, it applies to all subsequent conversions of the type to which it applies within
that statement unless another expl i cit scale factor is encountered.

If an exponent appears in the external field, the scale factor is not used for input conversions. Othern
wise, the internal value will be the external value times lO- .
n
On output, for D and E conversions, the fractional part of the number is multiplied by 10 and the
exponent is reduced by n. For G-type output, the scale factor is used only if the magnitude of the
number is such that E-type output notation is to be used. For F-type output conversion, the external
value = internal value times 10n.
Example 1.
Input format:

10 FORMAT(l PF12.4/-3PE8. 3/3pE12. 5/-3PF15.5)

Output format:

20 FORMAT(lX, F12. 4/1 X , E8.3/1X, E12. 5/1X, F15. 5)

Input Data:

-1234.5678
-123.456
98.76
12345.6789

6-10

Output:

-123.4568

*******
0.98760E-Ol
12345678.74995
Example 2.
Input format:

10 FORMAT(F12.4/E8. 3/F15. 5/E12.5}

Output format:

20 FORMAT (lX, 1PF12.4/1X,3PE 12. 3/1 X ,3PF15.5/1X,-3PE12. 5)

Input Data:

-12.34
123.456
987.654
123.456

Output:

-123.4000
123.505E+00
987653.99167
0.00012E+06

A-TYPE CONVERSION
General Form:

Aw or nAw
Examples

Format Descriptor

Input

Internal

Output

ABeD

ABCDb

ABeD

ABCDEF

BCDEF

bBCDEF

A-type conversion accepts an alphanumeric field of five characters and stores it as an unsigned double
integer variable. If a field width greater than five is specified, the excessive characters are dropped
from the left. If a field width less than five is specified, trailing blanks are appended.
On output, if field width is greater than five, W-5 leading blanks are output followed by the five internally stored characters. For a field width less than five, the leftmost w characters are output.

R-TYPE CONVERSION
General Form:

Rwor nRw
Examples

Format Descriptor

Input

Internal

Output

ABCD

bABeD

ABeD

ABCDEF

BCDEF

bBCDEF

bbbAB

6-11

R-type conversion accepts an alphanumeric field of five or less characters and right-adjusts them in an
unsigned double integer. The descriptor R4, as shown, stores the string bASCO whereas A4 would store
ABCDb. For strings equal to or in excess of five characters, R works exactly like A.

H-ALPHANUMERIC FIELD TRANSMISSION
General Form :

nHtext, 'text', $text$, "text"
Examples

Format Descriptor

Format Descriptor after
Input of the String LETTERS

Output

3HASC

ABC

3HLET

'TEXT'

TEXT

'LETT'

$MESSAGES$

MESSAGES

$LETTERS$

Character strings described as Hollerith fields are transmitted directly to and from an external device.
Since input changes a Hollerith fjeld, a subsequent output operation using the same FORMAT statement
will output a text string different from that specified in the source program.
Hollerith field descriptors may be placed among other field descriptors as in:
FORMAT (5HbANS=I3)
which could be used to output the value of a variable (assume V1=100) in the form:
ANS=l00
Note that the separating comma may be omitted after a Hollerith descriptor if it is of the form nHxxx.

L-TYPE CONVERSION
General Form:

Lwor nLw
Examples

Fie Id Descriptor

Input
--

Internal

Output

bTbb

777777 8

bbbT

When a value is read in using an L field descriptor, it is assumed to have the value 0 (.FALSE.) unless
the letter T appears (and is not preceded by the letter F) in which case it has the value 777777 , On
8
output, a zero value is output as F and all others as T.

6-12

Example 1.
Input format:

10 FORMAT{Ll,L2,L5,L8)

Output format:

20 FORMAT(lX,Ll,L2,L5,L8)

Input data:

TbFbbFbbbb3456
~~

Ll L2 L3

Internal values:

L1=777777
L2=0
L5=0
L8=0

Output:

TbFbbbbFbbbbbbbF
~~

Ll L2 L3

Example 2.
Input format:

10 FORMAT06)

Output format:

20 FORMAT{lX, L1, L2, L3, L4)

Input data:

000000
777777
1
2345

Internal values:

000000
757061
000001
004451
FbTbbTbbbT

Output:

L1 L2 L3 L4

OCTAL FIELDS
General Form:

Ow or nOw
Examples
Input

Field Descriptor

Internal

Output

4000000

012

400000000000

-1

777777

012

777777777777

7m77777777

The programmer may enter values as octal numbers using the descriptor O. For integers and double integers, as shown in the model, the programmer may enter an integer value in octal whose magnitude
exceeds the specified maximum (see 1.3.1). This provides a facility for establishing masking variables.

6-13

T-TAB
General Form:

Tn (n must be 2,.2 for output; >1 for input)
Examples

Format Descriptor
T35, 'ABC'

Input

Output

Characters 35-37 of data

To print positions 34-36

The descriptor T is used to control the emplacement of data on an output record (in terms of print position) and the acquisition of data from an input string (in terms of character position). Print position
refers to n-1 since the first character in an output buffer governs carriage control.
Example 1.
Input format:

10 FORMAT (f4, $ABC$)

Input string:

ABCDEFGHI

New format:

10 FORMAT(f4,$DEF$)

Example 2.
Input format:

10 FORMAT(f 1O,A3, Tl,A2)

Input string:

bbABCDbEFGbABC

Values read:

A3=GbA
A2=bb

Example 3.
Output format:

10 FORMAT(lX, T2,$1 WON'T$, T4,$////I'NILL$)

Output:

Print position 1
I 'w~wrWILL

Example 4.
Output format:

10 FORMAT(lX, T50, 'DATE', TlO,$NAME$)

Output:

Print position 9
NAME

Print position 49
DATE

X-BLANK
General Form:

nX
Examples

Format Descriptor

Output

Input

3X,$A$

bbA

4th character of input string

6-14

The descriptor X has appeared in previous examples of output formats to introduce a blank as the first
character of a record to accommodate the carriage control conventions described below. Any number
of blanks may be introduced within an output record using X. On input, X indicates that characters
are to be skipped.
Example 1.
Input format:

10 FORMAT(3X,A 1,5X,A2)

Input data:

ABCDEFGHIJK

Values assigned:

A1=D
A2=JK

Example 2.
Output format:

10 FORMAT(lX, $A$ ,2X ,$B$)

Output:

AbbB

CARRIAGE CONTROL
As stated previously, when a formatted record is output to a printing device, its first character is not
printed but governs vertical spacing. The use of blank to initiate a line feed has been illustrated.
Other characters may also be used to indicate other than line feed. Carriage control characters and
their effects are shown below.
Effect

Character
blank

line feed

double space
form feed

no advance (overprinting)

all others

line feed

Whatever action is specified occurs before the I ine is printed.
Example
Input format:

10 FORMAT~3)

Output format:

20 FORMAT ($0$,13)

Program variables: V1, V2, V3
Input data:

300400500

1.....,,-'1.....,,-'1.....,,-'

Vl V2 V3

6-15

Output:

[empty line]
300
[empty line]
400
[empty line]
500

6.1.3

Object Time FORMAT Specifications

Format specifications may be entered along with the input data they describe. If the programmer wishes
to do this, his formatted input/output statements, instead of referencing a FORMAT statement number,
reference the name of an array into which one or more FORMAT specifications will be entered. The
array must appear in a DIMENSION statement even if its size is 1.
An object time FORMAT specification has the same general form as the source-program statement with
two exceptions - the word FORMAT is omitted and Hollerith fields are not permitted. It can be entered
into the appropriate internal array via the DATA statement or by using a formatted input statement
which references an A-type FORMAT statement in the source program.
For example, the object time format specification (I7,F10.3) may be entered via a READ statement as
follows:
DIMENSION AA(lO)
13 FORMAT(10A5)
READ(3, 13)(AA(I),I=1, 10)
READ(3,AA)JJ, BOB

6.2

enters the FORMAT

-- enters the data according to format
stored in array AA

DATA-DIRECTED INPUT-OUTPUT

ASCII data may be entered or written without a FORMAT statement if the data-directed (format-free)
input-output option is indicated in an input-output statement. Externally stored data fields in this
case are determined not by field width but by the occurrence of a del imiter (multiple carriage returns,
spaces, ALT -MODES, or commas). Each data item will be assigned to the variable indicated after
conversion to the variable's type. A value will be assigned in all cases even if the value cannot be
converted properly. For example, if the value read for an integer variable is too large, the largest
lega I integer val ue (377777 8) will be assigned to it.
For a data-directed output statement, both the variable name (subscripted as appropriate) and its value
are written in the form: 'NAME'=value. The format in which the value is written is selected according
to the variable's type as follows:

6-16

Variable Type

Output Format

LOGICAL

INTEGER

REAL

G16.8

DOUBLE PRECISION

D20.11

DOUBLE INTEGER

112

On input, the data-directed option permits a variety of input items, an item being that portion of the
input data delimited by a space, comma, carriage return, or ALT MODE. One item is accepted for
each program variable specified in the input-output list. Acceptable input items are:
(1)

string constants - delimited by a set of dollar signs ($), single-quotes ('), or
double-quotes ("); may include any characters except carriage return and
ALT MODE; rules for including the string-delimiter characters within the
string are as given in Section 1.3.1 for Hollerith constants. If a Hollerith
constant contains more than five characters, only the first five are stored.
One of the item delimiters given above must follow the string terminating
character. To ensure accuracy, a string constant should be associated with
a double-integer variable.

(2)

octal constants - preceded by the character # (#D for double integers). Noninteger values may also be given in octal form. Some examples of octal input
items are:
#123
#D 1234567
#1.23
#-1.00E+02
An octal integer whose magnitude exceeds 131071 will be considered a double
integer even if D is not supplied, except when the value is within the range
400000 and 777777. This exception permits the programmer to perform explicit
octal masking.

(3)

logical constants - T or F followed by any number of characters up to a legal
item delimiter. The values -l(T) and O(F) are stored.

(4)

decimal numbers - such as:
123
-12.3
-1. 23E+2
+ 10. 234D+I5

If an illegal character appears in any data item, an error message is printed and input proceeds . For
Teletype input, the user may reenter the erroneous item.

6.3

INPUT-OUTPUT STATEMENTS

Input-output statements perform the actual transfer of data. The READ statement specifies input. Output is specified by one of the three synonymous statements VVRITE, PRINT I or TYPE.

6-17

The general format of these statements is given below (all arguments with the exception of dare
optional).
READ}
WRITE
{ PRINT
TYPE

(~}r, f, END=m, ERR=n) list

where:
d = integer constant or variable giving the logical device number (. OAT slot)
{

~ }r = integer expression indicating record number (# or ' indicate direct.

access I/O)

f = FORMAT statement number or array reference
m, n = statement number
list = an input-output Iist (must be present unless information is transferred
directly to and from a FORMAT statement)
When ,f is absent, unformatted I/O is indicated. If the comma appears alone, data-directed I/O, as
discussed in the preceding section, is performed. The END=m and ERR=n options permit the user to
spec ify, respecti ve Iy, a statement to wh i ch contro I shou Id be transferred when an end-of-fil e or error
condition occurs. If these opti ons are not present, OTS end-of-fi Ie or error routines are used.

6.3.1

Input-Output lists

An input-output list contains the names of variables, arrays, and array elements which are to be assigned data values on input or whose values are to be output.
An input-output list has the form:

where each c is a variable, subscripted variable, or an array name. When an array name appears
(with no subscripts), the effect is the same as if all elements of the array had been listed.

Note that

during input the new values of listed variables which appear in a subscript are used. Thus, for:

I, B, C, ARRAY(I)
the array reference will use the value input for I in that statement. For example, if the value 100 is
read into I, the fourth data field will be read into ARRAY(100).

If only a portion of an array is to be specified, indexing similar to that of a DO statement may be used
to indicate several items in an array. For example, the list element:
(X(K),K=l,4)

6-18

specifies X(l), X(2), X(3), and X(4). The indexing may also be compounded by nesting, as in:
«A(I,J),I=l ,4),J=1 ,5)
which specifies:
A(l, l),A(2, 1), .•• ,A(4, l),A(l ,2), .•. ,A(4,5)
The order of the list must be that in which the input data associated with its elements is stored. For
example, if the list is A, B, C, the first three data items (as defined by the associated FORMAT statement or data-directed I/O delimiter) are stored in variables A, B, C. On output, the list specifies
variables whose values are to be written either in accordance with a FORMAT statement or, for datadirected I/O, the default formats given previously.
If the data fields in a physical record exceed the number of variables in an input-output list, excessive
fields are simply ignored. If the data fields are not sufficient to accommodate all listed variables,
successive records are automatically read until all list elements have been satisfied.

6.3.2

Sequential Input-Output Statements

The form and effect of all sequential input-output statements are given in Tables 6-3 and 6-4 below.
The END and ERR option, which may be used in any form of these statements, is omitted in the following models for the sake of clarity.

Table 6-3
The READ Statement
Data are transferred from external device (d) to internal storage.
Form

Example

Effect

READ(d,f)list

READ(3, 10)A, B,C

ASCII data are read, converted
according to FORMAT reference
10 and stored in variables A, B,
and C

READ(d)list

READ(3)A, B,C

Binary data are read into variables A, B, and C

READ(d,)list

READ(3, )A, B, C

ASCII data are read, converted
to the appropriate type, and
stored in A, B, and C

READ(d,f)

READ(3,10)

Data are read into FORMAT
reference 10

READ(d)

READ(3)

A binary record is read and ignored

6-19

Table 6-4
The WRITE Statement
Data are transferred from internal storage to external device (d).
PRINT or TYPE may be I,Jsed as synonyms of WRITE.

6.3.3

Form

Example

Effect

WRITE(d, f)1 ist

WRITE(3, 10)A, B,C

The values of A, B, and Care
converted to ASCII according
to FORMAT reference 10 and
written

WRITE(d)list

WRITE (3)A, B,C

The values of A, B, and Care
written (in binary)

WRITE(d, )list

WRITE(3, )A, B, C

The variable nomes A, B, and
C are written, each followed
by its value in the form: 'A' =
value

WRITE(d,f)

WRITE (3, 10)

FORMAT reference 10 is written

Direct-Access Input-Output*

Direct-access input-output statements permit the user to directly reference any record in a fi Ie without
indexing from the file's beginning up to the desired record. In addition, data may be changed in a
single record without creating a new file.
Before direct-access input-output may be performed, the programmer must initialize the file via a
CALL DEFINE statement. (The format of direct-access files is given in Part II, Chapter 2). The form
of the statement is:
CALL DEFINE (d, rsiz, fsiz, fnam, v, mode, a deod)
where d is a logical device number and other arguments are described in Table 6-5.
The purpose of DEFINE is to initialize a file for direct-access operations. If the file exists already,
arguments supplied must correspond to the characteristics of the given file. If the file specified in
fnam does not exist on the specified logical device, a file of the given name will be created in accordance with the parameters supplied by the other arguments. Or if fnam=O, a file is created and assigned
the name TMOab OTS (where ab are the decimal digits of a logical device number ~). If the deletion
code is set to one, the file will be deleted when the file is closed. A direct-access file is closed by
either of the statements: EN DFI LE d or CALL CLOSE (d).

*For use with DOS-15 file structure only.

6-20

Table 6-5
Arguments for CALL DEFINE
Argument
rsiz

Value
Decimal integer

<628

Represents
Record length in
characters

[ASCII characters)
Decimal integer

<131071

Record length in
words

(Binary words)

Comments
For formatted records this is given as the
number of ASCII characters and for unformatted as binary words. The maximum
for rsiz is based on the fact that a physical block size is 256 words. However,
two of these are link words, two are
header words, and the first and last
characters of the ASCII data {automatically inserted} are, respectively, a
forms control character and a carriage return. The maximum number of words rsiz
may specify is 251 *. The additional five
words in the block are allocated as follows: two header words, one I.D. word,
and two link words.
Only fixed length records may be accessed.

fsiz

Decimal integer

<131071

The number of records in the file

The I. D. word of a record contains its
record number and the O-th bit is used as
a last-record indicator. Fsiz must be less
than or equal to the number of records
in the existing file unless file size is to
be adjusted {argument a, described below}. When fsiz is less than the actual
number of records, no input-output operations may be performed on records
with greater numbers.

fnam

Array name

An array containing
a six-character file
name and threecharacter extension
(in sixbit ASCII)

Fnam may also be 0 indicating that standard default names are to be used.

Integer variable
name (referred to
as the assoc iated
variable)

The number of the
record just after
the last one accessed

V is assigned a value at the conclusion
of an input-output operati on.

mode

Integer variable

I/O mode

For unformatted I/O, mode=O; for formatted I/O, mode=non-O

Integer code

Indicates file adjustment options

a=O indicates no adjustment; otherwise,
the number of records in the fi Ie specified is adjusted to the number indicated
by the curren t fs iz •

dcod

Integer code

Applies only to
temporary fi les

If dcod=O, it means lido not delete. II If
dcod=l, temporary file .TMOab OTS is
to be deleted on device ab (decimal
digits of device number d)

*Logical records greater than 1 physical block are termed long records. These are specified when rsiz
is greater than 251.
6-21

Two conventions are avai lable for specifying the record size for an unformatted direct-access record.
One, for records whose size is less than one physical block, has a maximum of 251 words available
for user data. Since the overhead components of an unformatted record comprise three words and since
the total record. size must be even, the number of user data words must be an odd number. If an even
number is specified, DEFINE will treat it as though rsiz +1 were specified.
Another convention is used for "long records" - records greater in size than one physical block. All
blocks of a long record except the last will contain unformatted records 251 words long. The last block
will contain an unformatted record with an odd number of data words so that the total length is equivalent to fsiz (or fsiz +1 if even).
When a formatted file is created, all data words are filled with 7-bit ASCII spaces (040 ); for unfor-

matted, all data words are set to zero. The I. D. word for all binary records is set to 400000

a (see

Part II for more detail). The associated variable is set to one.
If the user arguments have requested adjustment of an existing file, a temporary file named •• TEMP OTS
is created on the device specified and the existing file (temporarily associated with a system device) is
copied into it a record at a time. When a file is lengthened, size parameters are used to add null records (spaces or zeros) after the last record is transmitted and the associated variable is set to the old
number of records plus one. If file size is being reduced, data from the end of the old file are lost and
the associated variable is set to one.
Direct-access input-output transfer is accomplished by READ and WRITE statements of the following
form (a single quote) (') may be used instead of # and the END and ERR options may be included}:
(1)

READ (d#r,f) list

(2)

READ (d#r) list

(3)

READ (d#r,) list

(4)

READ (d#r, f)

(5)

READ (d#r)

(6)

WRITE (d#r,f) list

(7)

WRITE (d#r) list

(a)

WRITE (d#r,) list

(9)

WRITE (d# r, f)

(1O)

WRITE (d#r)

These statements have the same effect as their counterparts do on sequential input-output. The major
distinction is that record number (r) must be supplied. This indicates the record number relative to the
beginning of the file. The same device number may be given for direct-access READ's and WRITE's; if
a file is to be read sequentially and direct-access read or written, two logical device numbers must be
specified.
6-22

----

6.3.4

The ENCODE/DECODE Statements
General Form:

tENCODE}
DECODE

Where

(c,v,f,ERR=n) list

= number of ASCII characters per logical record
v = name of array containing ASCII record
f = optional format reference
ERR=n is as for input-output and optional list is a list of
internal variables

The ENCODE statement converts binary data stored in the variables listed, converts them to characters,
and stores them in the array (v). If more variab les are listed than can be stored in that array, the values are ignored. If the values are not sufficient to fill the array, blanks are added. Conversion is
performed according to a FORMAT specification (f) or, in its absence, according to the data-directed
input-output rules. In the latter case, the data-directed output form 'VAR' = value is stored. Formscontrol characters are not analyzed on ENCODE and will be stored explicitly if in the FORMAT used.
The DECODE statement converts character data stored in the array (v) into binary and assigns them to
variables in the I/O list.

6.3.5

Auxiliary Input-Output

Two library subroutines, SEEK and ENTER, are available for both sequential and direct-access inputoutput to disk. ENTER is called as follows:
CALL ENTER(N,A)
where N is a device number and A is the name of an array containing an ASCII file name and extension.
The effect of ENTER is to initialize and open a named output file. For example:
DIMENSION FILEN(2)
DATA FILEN(l),FILEN(2)j5HFILNA,4HM EXT/
CALL ENTER (1, FILEN)
establishes a file named FILENAM EXT.
SEEK finds and opens a named file for input. It is called as follows:
SEEK(N,A)
where N and A are as for ENTER. To input the file established by ENTER, the call would be:
CALL SEEK(l ,FILEN)
The device-control statements listed below are applicable to sequential files on magnetic tape or disk
with the effects stated.

6-23

Statement

Magnetic Tape

Disk*

BACKSPACE d

Repositions unit to the preceding record (if at beginning, does nothing)

Repositions unit one ASCII
line or one logical unformatted record (INF'UT only)

REWIND d

The tape is positioned to its
initial point

The file is closed and reopened for input or output
by the next sequential I/O
statement to that device.
If no file is opened, does
nothing

ENDFILE d

Writes an end-of-file record

Closes the named file which
is open on device d

The ENDFILE statement may be used to create segmented files with magnetic tape; that is, using the
end-of-file indicator to separate the segments. For example, the statements:
WRITE (3) list
WRITE (3) list
ENDFILE 3
WRITE (3,10) list
ENDFILE 3
WRITE (3) list
ENDFILE 3
creates three segments. A segmented fi Ie can be read using the END=n option of the READ statement
as follows:
READ (3, END=lO) list
10 READ (3, END=20) list

20 READ (3, END=30) list
30 REWIND 3
Figure 6-1 shows a program using auxiliary input-output statements for disk. Note that the first series
of WRITE statements cause a default file (.TM001 OTS) to be created. This file is read by subsequent
READ statements specifying logical device 1.

*Note that the first sequential I/O statement to the disk opens a default file if no file has been explicitlyopened.
6-24

199
1

70
8
71
9

12
90
91

DIMENSION 1(10),11(10)
DO 1 K=l, 10
I(K)=K
REWIND 1
REWIND 7
WRITE (1) I
DO 2 K=11,20
I(K-10)=K
WRITE (1) I
DO 3 K=21 ,30
I (K-20)=K
WRITE (1) I
ENDFILE 1
READ (1) II
JT=l
JV=II(5)
IF (II (5)-5)90, 4, 90
READ(1 )11
JT=2
JV=II(5)
IF (II (5)-15)90,5,90
READ(1) II
JT=3
JV=II(5)
I F(II (5)-25)90,6,90
BACKSPACE 1
BACKSPACE 1
READ (1) II
JT=4
JV=II (7)
IF(II (7)-17)90,7,90
CONTINUE
PAUSE 1
JT=5
WRITE(7,70)1
FORMAT(lX,16)
REWIND 7
D08K=1,7
READ (7,71)JV
FORMAT06)
IF (JV-27)90, 9,90
JT=6
DO 11 K=1,3
BACKSPACE 7
READ (7,71) JV
IF (JV-25)90, 12,90
PAUSE 2
GO TO 199
WRITE(2, 91) JT, JV
FORMAT (lX,I6, 2X, 16)
STOP
END
Figure 6-1

/file .TM001 OTS created
/. TM001 OTS sought automatically

/backspacing binary data

/fi Ie •TM007 OT S created

/file .TM007 OTS sought

/backspacing ASCII data

Programming Example - Auxiliary I/O to Disk

6-25

APPENDIX A
LANGUAGE SUMMARY

Statement

Model

Text
Reference

Effect

Arithmetic

var-value
array(i)=value

value is assigned to var
or array (i)

2.1

ASSIGN

ASSIGN n TO label

Statement n is assigned
the symbol name label

2.2

BLOCK DATA

Identifies subprogram
which enters data into a
labeled COMMON
block at run time

4.4

CALL

CALL subr(a ,a ,·· .a )
n
1 2
CALL subr

Control is transferred to
the subroutine; a ,a , ...
2
a are substitute for
dUmmy variables

5.2.2

COMMON

COMMON;b l/vl ist l;b!vlist! ...

vlist items are allocated
to b blocks where they
are shared by other programs

4.2.2

CONTINUE

Dummy statement used to
prevent illegal termination of DO loops

3.2.3

DATA

DATA vlistlclistl,vlist!c1ist2/""
vlist /clist /
n
n
DECODE(c, v, f ,ERR=n)1 ist

clist is assigned to its corresponding vlist

4.3

Converts character data
stored in the array (v)
into binary and assigns
them to variables in list

6.3.4

DIMENSION

DIMENSION a (11),a (1 ),· .. an(ln)
2 2
1

Storage is allocated for
array (a) to the dimensions specified by the
subscript list (I)

4.2.1

DO n i=m1,m2,m3

Statements following the
DO are executed repeatedly for values m1
through m2 in increments
or decrements of m3

3.2

DECODE

DO n i=m1,m2
DO n i=m ,m ,-m
1 2
3

(continued on next pase)

A-1

ENCODE

Text
Reference

Effect

Model

Statement

ENCODE{c, v, f, ERR=n)1 ist

Converts binary data represented by variables in
list into characters according to FORMAT
specification (f) or datadirected I/O rules and
stores them in the array

6.3.4

(v)
EQUIVALENCE

EQUIVALE NCE(l1)' (12)' ••• (I n)

Elements of each list (I)
are assigned to the same
storage location

4.2.3

EXTERNAL

EXTERNAL a ,a ,· •• a
n
1 2

Defines subprograms
named a for use as arguments of other subprograms

4.1.3

FORMAT

n FORMAT{s1 ,s2'" .sn)

FORMAT statement n established as fieldspec i ficati on references

6. 1

FUNCTION

m FUNCTION f(a ,a , ... a )
1 2
n

Defines FUNCTION
named f with dummy arguments a and optional
mode specification m

5.1.2

GOTO

GOTO n

Control is unconditionally
transferred to statement n

3.1.1

GO TO(n ,n , ••• n ), i
1 2
k

Control is transferred to
the ith statement in the
list of n's

3.1.2

GO TO label
GO TO label,(n ,n ,· .. n )
k
1 2

Control is transferred to
the location specified by
label; the list of n's may
specify legally ASSIGNable statement numbers

3.1.3

IF(expr)n l' n 2 , n3

Control is transferred to
statement number or
ASSIGNed label n1' n2'
or n3 if evaluated expr is
<0, =0, or )() respectively

3.3.1

IF {expr)s

Statement s is executed if
expr is . TRUE. (non-zero),
ignored if . FALSE. (zero)

3.3.2

IMPUCIT m (11),m {1 ), ... mn(ln)
1
2 2

Declares mode (m) for variables beginning with alphabetic characters in
list (I)

4.1.2

IMPUCIT

(continued on next page)

A-2

Statement

Model

Effect

Text
Reference

PAUSE

PAUSE
PAUSE n

Interrupts program execution; if present, integer
n is printed on the console to distinguish one
PAUSE from another

3.4.1

PRINT(d,f)list

The values of variables in
I ist are converted to
ASCII according to
FORMAT reference (f) and
transferred to extern a I device (d)

6.3.2

PRI NT (d)1 ist

The values of variables in
list are written in binary
on external device (d)

6.3.2

PRINT(d,)list

The variable names in list
are written on external
device (d), each followed
by its value in the form
'A'=value

6.3.2

PRINT(d,f)

FORMAT reference (f) is
written on external device (d)

6.3.2

READ(d, f)1 ist

The values represented by
variables in list are read
from external device (d)
and converted according
to FORMAT reference (f)

6.3.2

READ(d)list

The binary values represented by variables in list
are read from external device (d)

6.3.2

READ(d, )list

The values represented by
variables in list are read
from external device (d)

6.3.2

READ(d, f)

Values are read into
FORMAT reference (f)

READ(d)

A binary record is read
from external device (d)
and ignored

6.3.2

STOP
STOP n

Signifies the logical end
of a program and returns
control to the MONITOR
after n is printed; if present, n distinguishes one
STOP from another

3.4.2

READ

STOP

(continued on next page)

A-3

Model

Statement

Effect

Text
Reference

SUBROUTINE

SUBROUTINE name (a ,a ,· .. a )
1 2
n
SUBROUTINE name

Defines an external subroutine named name; a's
are dummy argum~nts representing values supplied
by the calling program or
returned by the subroutine

5.2.1

TYPE

TYPE(d, f)1 ist

The values of variables in
list are converted to ASCII
according to FORMAT reference (f) and transferred
to external device (d)

6.3.2

TYPE(d)list

The values of variables in
list are written in binary
on external device (d)

6.3.2

TYPE(d, )Iist

The variable names in list
are written on external
device (d), each followed
by its value in the form
'A'=value

6.3.2

TYPE(d,f)

FORMAT reference (f) is
written on external device (d)

6.3.2

WRITE(d,f)list

The values of variables in
list are converted to ASCII
according to FORMAT reference (f) and transferred
to external device (d)

6.3.2

WRITE(d)list

The values of variables in
list are written in binary
on external device (d)

6.3.2

WRIT E(d,)1 ist

The variable names in list
are written on external
device (d), each followed
by its value in the form
'A'=value

6.3.2

WRITE(d,f)

FORMAT reference (f) is
written on external device (d)

6.3.2

WRITE

A-4

APPENDIX B

ERROR MESSAGES

B.l

COMPILER ERROR MESSAGES

In the F4X version of FORTRAN, compiler error messages are printed in the form:
>mnA<
where:
mn is the error number
A is the alphabetic mnemonic
characterizing the error class.
In F4I and F4A versions, only the alphabetic character is printed, in the form:
>A<
All error messages and the version(s} of FORTRAN to which they are applicable are given below.

Number

Letter

Meaning
Common, equivalence, data errors:

No open parenthesis after variable name in DIMENSION
statement

No slash after common block name

Common block name previously defined

Variable appears twice in COMMON

EQUIVALENCE list does not begin with open parenthesis

Only one variable in EQUIVALENCE class

EQUIVALENCE distorts COMMON

EQUIVALENCE extends COMMON down

Inconsistent EQUIVALENCing

EQUIVALENCE extends COMMON down

Illegal delimiter in EQUIVALENCE list
(continued on next page)

B-1

Number

Letter

Meaning
Common, equivalence, data errors: (cont)

Non-COMMON variables in BLOCK DATA

III egal repeat factor in DATA statement

DATA statement stores in COMMON in non-BLOCK DATA
statement or in non-COMMON in BLOCK DATA statement
DO errors:

Statement with unparenthesized = sign and comma not a DO
statement

DO variable not followed by = sign

DO variable not integer

Initial value of DO variable not followed by comma

Improper delimiter in DO statement

Illegal terminating statement for DO loop
External symbol and entry-point errors:

Variable in EXTERNAL statement not simple non-COMMON
variable

ENTRY name non-unique

ENTRY statement in main program

No = sign following argument list in arithmetic statement
function

No argument list in FUNCTION subprogram

Subroutine list in CALL statement already defined as variable

Function or array name used in expression without open
parenthesis

Function or array name used in expression without open
parenthesis
Format errors:

Bad delimiter after FORMAT number in I/o statement

Missing field width, illegal character or unwanted repeat
factor

Field width is 0

Period expected, not found

Period found, not expected

Decimal length missing (no "d" in "Fw.d")

Unparenthesized comma

-.
(continued on next pose)

B-2

Number

Letter

Meaning
Format errors: (cont)

Minus without number

No P after negative number

No number before P

No number or 0 before H

No number or 0 before X

Too many Ieft parentheses
Hollerith errors:

Number preceding H not between 1 and 5

Carriage return inside Hollerith field

Number preceding H not an integer

More than five characters inside quotes

Carriage return inside quotes
Various illegal errors:

Unidentifiable statement

Misspell ed statement

Statement out of order

Executable statement in BLOCK DATA subroutine

Illegal character in I/O statement, following unit number

Illegal delimiter in ASSIGN statement

Illegal type in IMPLICIT statement

Logical IF as target of logical IF

RETURN statement in main program

Semicolon in COMMON statement outside of BLOCK DATA

Illegal delimiter in IMPLICIT statement

Misspelled REAL or READ statement

Misspelled END or ENDFILE statement

Misspelled ENDFILE statement

Statement function out of order or undimensioned array

Typed FUNCTION statement out of order

Illegal character in context

Illegal logical or relational operator
(continued on next page)

B-3

Number

Letter

Meaning
Various illegal errors: (cont)

Illegal letter in IMPLICIT statement

Illegal letter range in IMPLICIT statement

Illegal del imiter in letter section of IMPLICIT statement

Illegal character in context

Illegal comma in GOTO statement

Illegal variable used in multiple RETURN statement

-,I

Pushdown list errors:
01

DO nesting too deep

Illegal DO nesting

Subscript/function nesting too deep

Backwards DO loop (also caused by some illegal I/O lists),
Appears after END statement.
Overflow errors:

EQUIVALENCE class list full

Program size exceeds 8K

Array length larger than 8K

Element position in array larger than 8K (EQUIVALENCE,
DATA)

Integer negative or larger than 131071

Exponent of floating point number larger than 76

Overflow accumulating constant - too many digits

Overflow accumulating constant - too many digits
Statement number errors:

Multiply defined statement number or compiler error

Statement erroneously labeled

Undefined statement number

FORMA T statement without statement number

Statement number expected, not found

Statement number more than five digits

Illegal statement number

(continued on next page)

8-4

Number

Letter

Meaning
Partword errors:

Expected colon, found none

Expected close bracket, found none

Last bit number larger than 35

First bit number larger than last bit number

First and last bit numbers not simple integer constants
Subscripting errors:

Illegal subscript delimiter in specification statements

More than three subscripts specified

Illegal delimiter in subroutine argument list

Non-integer subscript

Non-scalar subscript

Integer scalar expected, not found

Two operators in a row

Close parenthesis following an operator

Non-integer subscript

Non-scal or subscri pt

Two arguments in a row

Digit or letter encountered after argument conversion

Number of subscripts stated not equal to number declared
Table overflow errors:

Arithmetic statement, computed GOTO list, or DATA statement list too large

Too many dummy variables in arithmetic statement function

Symbol and constant tables overlap
Variable errors:

Two modes specified for same variable name

Variable expected, not found

Constant expected, not found

Array defined twice

Error: variable is EXTERNAL or argument (EQUIVALENCE,
DATA)

More than one dimension indicated for scalar variable
(continued on next page)

B-5

Number

Letter

Meaning
Variable errors: (cont)

First character after READ or WRITE not open parenthesis in

I/o statement
09

Illegal constant in DATA statement

Variables outnumber constants in DATA statement

Constants outnumber variables in DATA statement

Illegal dummy variable (previously used as non-dummy variable)

Logical operator has non-integer, non-logical arguments

Illegal mixed mode expression

Logical operator has non-integer, non-logical arguments

Signed variable left of equal sign

Illegal combination for exponentiation

• NOT. operator has non-integer, non-logical argument

Function in specification statement

Two exponents in one constant

Illegal redefinition of a scalar as a function

No number after E or D in a constant

Non-integer record number in random access I/o

Illegal delimiter in I/o statement

Illegal syntax in READ, WRITE, ENCODE, or DECODE
statement

END and ERR exists out of order in I/o statement

Constant and variable modes don't match in DATA statement

ENCODE or DECODE not followed by open parenthesis

Illegal delimiter in ENCODE/DECODE statement

Array expected as first argument of ENCODE/DECODE
statement

Illegal delimiter in ENCODE/DECODE statement
Expression errors:

Carri age return expected, not found

Binary WRITE statement with no I/o list

Illegal element in I/o list

Illegal statement number list in computed or assigned GOTO

Illegal delimiter in computed GOTO

Illegal computed GOTO statement

-,
(contmued on next page)

B-6

Number

Letter

Meaning
Expression errors: (cont)

B.2

Illegal delimiter in DATA statement

No close parenthesis in IF statement

Illegal delimiter in arithmetic IF statement

Expression on left of equals sign in arithmetic statement

Too many right parentheses

Illegal open parenthesis (in specification statements)

III egal open parenthesis

Too many right parentheses

Illegal alphabetic in numeric constant

Symbol contains more than six characters

• TRUE., • FALSE., or • NOT. preceded by an argument

Unparenthesized comma in arithmetic expression

Unary minus in I/o list

Illegal delimiter in I/o list

Unterminated implied - DO loop in I/o list

Illegal equals sign in I/o list

Illegal partword operator

Illegal arithmetic expression

OTS ERROR MESSAGES

Following is a list of OT5 error messages. (R) indicates a recoverable error; (T) a terminal error.

Error Number

Possible Source

Error Description

05 (R)

Negative REAL square root argument

SQRT

06 (R)

Negative DOUBLE PRECISION square root
argument

DSQRT

07 (R)

Illegal index in computed GO TO

.GO

10 (T)

Illegal I/O device number

. FR , . FW, . FS, • FX ,
DEFINE, RANCOM

11 (T)

Bad input data - lOPS mode incorrect

.FR, .FA, .FE, .FF, .FS,
RANCOM, RBINIO,
RBCDIO
(continued on next page)

B-7

Error Number

(T)
13 (T)
12

Error Description

Possible Source

Bad FORMAT

.FA, .FE, .FF

Negative or zero REAL logarithmic argument
(terminal)

.BC, .BE, ALOG

(R)

Negative or zero DOUBLE PRECISION logarithmic argument

. BD, • BF, • BG, • BH,
DLOG, DLOG10

15 (R)

Zero raised to a zero or negative power (zero
result is passed)

.BB, .BC, .BD, .BE, . BF,
.BG, . BH

Fatal I/O error (RSX only)

FlOPS

Undefined file

RANCOM

Illegal record size

DEFINE

Size discrepancy

RANCOM

Illegal record number

DEFINE, RAN COM

Mode discrepancy

RAN COM

Too many open fi les

DEFINE

Single integer overflow*

RELEAE, .FPP

Extended (double) integer overflow****

DBUNT, JFIX, JDFIX,
ISNGL

(R)
**33 (R)

Single fit. overflow

RELEAE

**34 (R)

Single fit. underflow

(f)
1""21 (f)
22 (f)
direct
23 (T)
access <
errors
24 (T)
25 (T)
26 (T)
30 (R)
**31 (R)
20

**32

**35

(R)

**36 (R)

Double fit. overflow t

Double fit. underflow

RELEAE
t

Fit. divide check

RELEAE

(R)
40 (f)

Integer divide check

INTEAE

Illegal number of characters specified [legal:
0<c<625]

ENCODE

41 (R)

Array exceeded

ENCODE

(T)
**50 (T)
51 (f)

Bad input data

DD10

***37

FPP memory protect/non-existent memory
(READ to WRITE Illegal I/O Direction Change
to Disk) without intervening CLOSE or REWIND

BCOIO, BINIO

*Only detected when fixing a floating point number.
**Also prints out PC with FPP system
***If extended integer divide check, prints out PC with FPP system.
****With software F4 system only detected when fixing a floating point number.
tNot detected by software system (only by FPP system).

B-8

B.3

OTS ERROR MESSAGES IN FPP SYSTEMS

In software systems, arithmetic errors resulting in the OTS error messages summarized above are detected in the arithmetic package (RELEAE and INTEAE). In the hardware FPP systems, these errors
are detected by the hardware (with the exception of single integer divide check) and serviced by a
trap routine in the FPP routine. FPP.
Where applicable, on such error conditions, the result is patched for both software and hardware systems as summarized in the following table.

PATCHED VALUE***

Error

FPP Hardware System

Software System

Single Floating Overflow
(. OTS 32)

± largest single floating value

same

Double Floating Overflow
(.OTS 33)

± largest single floating value

not detected

Single Floating Underflow
(.OTS 34)

zero

same

Double Floating Underflow
(.OTS 35)

zero

not detected

Floating Divide Check
(.OTS 36)

± largest single floating value

same

Integer Overflow
(.OTS30)

limited detection*

same

Double Integer Overflow
(.OTS 31)

none**

limited detection*

Integer Divide Check
(.OTS 37)

none

same

*When fixing a floating point number, integer and extended integer overflow is detected. In these
instances, plus or minus the largest integer for the data mode is patched as result.
**With the FPP system all extended integer overflow conditions are detected, but the results are
mean ing less.
***Where "none" is specified, the result is meaningless unless otherwise indicated.
Further, when converting an extended integer, the magnitude of which is >217_1, toa single
integer, no error is indicated and the high order digits are lost.

B-9

APPENDIX C
PDP-15 FORTRAN FACILITIES

The extended FORTRAN language described in this manual and in the companion manual (Operating
Environment Manual DEC-15-GFZA-D) is available only on the systems described below. The
FORTRAN existing on other PDP-15 systems is described in a manual entitled "PDP-15 FORTRAN IV
Programmer's Reference Manual" (DEC-15-K FZB-D).
The following tables describe the existing versions of the extended compiler, the extended Object
Time System Libraries, and the compiler-library pairs available for different systems. All versions of
the compiler are written in PDP-9 code, however, 'PDP-9 mode' versions produce only PDP-9 code as
output while 'PDP-15 mode' versions may produce PDP-15 instructions where suitable. Page and Bank
Mode libraries differ not only in the use of the PDP-15 versus PDP-9 code, but also in the values of
address masking constants used in a few of the routines.

Note that the Floating Point Processor (FPP)

is supported only on the PDP-15, thus there is no PDP-9 mode version.
The library names used in the following tables are given for designational purposes within this appendix
only and do not necessarily reflect the names under which the libraries are distributed.

Table C-1
Versions of the Extended Compiler

Main
Version

F4X

F4B

F4RX

Features

All

System

Version

Approx.
Size (8)

F4X9
FPF4X

Non-FPP, PDP-15 mode DOS-15
Non-FPP, PDP-9 mode DOS-15
FPP, PDP-15 mode DOS-15

15406
15363
15661

F4B9
FPF4B

Non-FPP, PDP-15 mode, ADSS (V5B)
Non-FPP, PDP-9 mode ADSS (V5B)
FPP, PDP-15 mode ADSS (V5B)

15251
15226
15522

All except
direct-access I/O

{F4RX
FPF4RX

Non-FPP, PDP-15 mode RSX
FPP, PDP-15 mode RSX

C-1

Table C-2
Versions of the OTS libraries for the Extended Compiler
System

Contents

Libraries

D05-15 (B055-15)

Contains all routines, assembled for D05-15
operation.

AD5S

Contains all routines except direct-access
(DEFINE, RANCOM, RBINIO, RBCDIO)
assembled for ADS5 operation.

R5X

Contains all routines except direct-access
(DEFINE, RANCOM, RBINIO, RBCDIO)
and magtape subroutines (UNIT, EOF), assembled for RSX operation and incl udes
added routines appl icable to R5X only.

xp
.LBXB
.LBXPF
· LBXBF

RP
.LBRB
• LBRPF
· LBRBF

{UB~

• LIBFX

Subsystem

Non-FPP, Page
Non-FPP, Bank
FPP, Page
FPP, Bank
Non-FPP, Page
Non-FPP, Bank
FPP, Page
FPP, Bank
Non-FPP, Pagel
Bank
F PP, Page/Bank

Table C-3
Compilers and Libraries for Extended FORTRAN
Distributed with PDP-9/15 Systems

5ystem

Page

Non-FPP
Bank

FPP
Page

Bank

D05-15

Compiler

F4X

F4X or F4X9

FPF4X

(B05S-15

Library

. LBXP

.LBXB

. LBXPF

. LBXBF

AD55 V5B

Compiler

F4B

F4B or F4B9

FPF4B

Library

.LBRP

. LBRB

. LBRPF

.LBRBF

Compiler

F4RX

FPF4RX

Library

. LIBRX

.LIBFX

. LIBFX

R5X

C-2

INDEX

B (Cont)

A (control character), 6-4

Blanks in statements, 1-3

ABS library function, 5-7

BLOCK DATA statement, 4-1, 5-6

Addition, 1-10

subprograms, 4-7

Addressing, symbolic, 2-3

Block size, 6-21

Alphanumeric conversion, 6-11

Branching, 3-1

ALT -MODE field delimiter, 6-16
.AND. logical operator, 1-13
Arithmetic

CALL CLOSE statement, 6-20

expressions, 1-10, 1-12
IF statement, 3-7
mode conversion, 2-1, 2-2
operators, 1-10
statement, 2-1
Array

CALL Statement, 5-7, 5-9
CALL DEFINE statement, 6-20, 6-21
Card coding, 1-3
Card reader record, 6- 1
Carriage control, 6-5

format, 6- 16
merging, 5-1
names, 1-9, 4-4
specification, partial, 6-18
storage, 4-4, 4-5
unsubscripted, 4-7, 6-18

characters, 6-15
Carriage returns, 1-3
multiple returns, 6-6
Characters, lost input, 6-4

ASCII data, 6-16

Character set, 1-2

Assigned GO TO statement, 3-1, 3-2

Coding form, 1-3, 1-4

Assignment statements, 2-1

Commas as field delimiters, 6-16

ASSIGN statement, 2-1, 2-3

Comments, 1-2, 1-3

Asterisks (**) in output, 6-5

COMMON block extension, 4-5

A-type conversion, 6-11

COMMON statement, 4-1, 4-4, 4-8, 5-8

At sign (@) usage, 5-10

COMMON variables, 4-6

Auxil iary input-output, 6-23

Compiler, versions of the extended, C-1
Compound expressions, 1-10

BACKSPACE statement, 6-24

Computed GO TO statement, 3-1, 3-2
Constants, 1-5
double integers, 1-5
double precision, 1-7
Hollerith, 1-7
integers, 1-5
logical, 1-7
real, 1-6

Beginning of program, 1-2
Binary integers, 6-5
Blank COMMON, 4-4
Blank field spaces descriptor (X), 6-15
Blank records, 6-3

CONTINUE statement, 3-6

Blanks in I-conversion descriptors, 6-5

Continuation lines, 1-3, 2-1
1-1

INDEX (Cont)
C (Cont)

D (Cont)

Control characters, field descriptors, 6-4

Decimal point in decimal digit, 1-6

control statements, 3-1

DEC Iibrary functions, 5-7

arithmetic IF, 3-7
assigned GO TO, 3-2
computed GO TO, 3-2
CONTINUE, 3-6
DO, 3-3
logical IF, 3-8
PAUSE, 3-8
STOP, 3-9
unconditional GO TO, 3-1

DECODE statement, 6-2, 6-23
DEFINE statement, 6-20
Definition of
object program, 1-1
source program, 1-1
statement, 1-1
Definition, statement function, 5-5

Conversion, arithmetic mode, 2-1, 2-2

Definition, subroutine, 5-8

Conversion of data, ASCII+o--binary, 6-23

Delimiters, field, 6-16

Conversion

Device Assignment Table (DAT), 6-2

A-type, 6-11
D-type, 6-9
E-type, 6-6
F-type, 6-8
G-type, 6-8
I-type, 6-5
L-type, 6-12
R-type, 6-11

Device number, 6-2
D format (double precision), 4-7
DIMENSION statement, 4-1, 4-3, 4-4, 5-6, 5-8,
6-16
Direct access input-output, 6-2, 6-20, 6-22
Disk/DECtape record, 6-1
Division, 1-10

D notation, 1-5

Data changing in single record, 6-20

Dollar sign ($) control character, 6-4

Data conversion, ASCII+o--binary, 6-23

DO loops, 3-5, 5-9

Data-directed input/output, 6-2, 6-16

nested DO statements, 3-5

Data fields wrong size, 6-19

DO range execution, 3-4

Data I ist order, 6-19

DO statement, 3-1, 3-3

Data mode specification, 4-1

DOUBLE INTEGER
mode, 5-6
statement, 4-1

DATA statement, 4-1, 4-7, 5-8,6-16
Data transmission statements, 6-1
input-output statements, 6-17
Data value expression, 1-3
D control character
double precision constant, 1-7
double precision variable, 6-4

Double integer constants, 1-5, 2-2
Double integers, octal, 6-13
Double integer variable format, 6-17
Double integer variables, 6-5, 6-11
DOUBLE PRECISION
mode, 5-6
statement, 4- 1

Decimal exponent, 1-6
Decimal numbers, 1-5

1-2

INDEX (Cont)
D (Cont)

E (Cont)

Double precision constants, 1-5, 1-7,4-7

Exponentlation, 1-10

Double precision floating point number format,
6-9

Express ion modes, 1- 12
Express ions

Double precision floating variable, 2-2

arithemtic, 1-10
logical, 3-7

Double precision variable format, 6-17
D-type conversion, 6-9

Externa I data storage, 6-1

Dummy arguments, 5-1, 5-2, 5-5, 5-8, 5-9

External functions, 5-2, 5-6, 5-7

Dummy names, 4-3

EXTERNAL statement, 4-1, 4-3

Dummy variables, 5-1

Extended range DO loop, 3-5

E control character, 6-4

.FALSE., 1-7, 1-12, 3-7

E (decimal exponent), 1-6

F control character, 6-4

E format, 6-8

Field descriptors, 6-3, 6-4
control characters, 6-4

ENCO DE statement, 6-2, 6-23
ENDFILE statement, 6-20, 6-24

Fields, 1-3

End of program, 3-9

Field width, 6-4, 6-6

END option, 6-18, 6-19

File
identification, 6-2
name, 6-21
size, 6-21

END statement, 1-2, 5-8
ENTER library subroutine, 6-23
ENTRY statement, 5-9

Files, 6-22

Equal sign, replacement, 2-1

segmented, 6-24
temporary, 6-20, 6-21, 6-22

EQUIVALENCE statement, 4-1, 4-5, 4-6, 5-8
ERR option, 6-18, 6-19

Floating point number conversion, 6-6

Error messages, B-1

Form feed, 6-15

Compiler, B-1
OTS, B-7
OTS in FPP Systems, B-9
Error statement, 3-2
E-type conversion, 6-6
Exclusive OR, 1-13, 3-7
Executable statements, 1-2
Execution control, 3-9
Execution sequence, control statements, 3-1
Exit DO loop, 3-5

FORMAT output, 6-5
FORMAT specification, Object Time, 6-16
FORMAT statement, 6-1 through 6-4
Formats, variable type, 6-17
Formatted record, 6-1
FORTRAN Science Library, 5-7
FPP processor, C-1, C-2
Fracti onal number output, 6-8
F-type conversion, 6-8

1-3

INDEX (Cont)
F (Cont)
Function name, 5-6

I (Cont)
Input-output

FUNCTION statement, 5-2, 5-6

auxiliary, 6-23
data directed, 6-16
direct access, 6-20, 6-22
format, 6-2
lists, 6-18
sequential statements, 6-19
statements, 6-17, 6-18

Functions, 5-1, 5-2
external, 5-6, 5-7
intrinsic, 5-7

G control character, 6-4
GO TO statement, 3-1, 3-2

INTEGER mode, 5-6
INTEGER statement! 4-1
Integers, 2-2

G-type conversion, 6-8

constants, 1-5
octal, 6-13
variable format, 6-17

H
H alphanumeric field transmission, 6-12

H control character, 6-4
Hollerith constants, 1-5, 1-7

Internal data storage, 6-1
Intrinsic functions, 5-7
I -type con vers ion descr i ptors, 6-5

Hollerith fields, 6-12, 6-16
L
Labels, 2-3
I control character, 6-4
Identification field, 1-3
IF statement, 3-6

Language summary, A-1
L control character, 6-4
Library
functions, 5-7
subroutines, 6-23

arithmetic, 3-7, 3-8
logical, 3-7, 3-8
IMPLICIT statement, 1-9, 4-1, 4-2, 4-3
Inclusive OR, 1-13
Increment value of DO loop, 3-3
I ndexi ng an array, 6- 18
Index of DO loop, 3-3
Index variable, 3-5
Infinite loop, 3-4
Initialization, 4-7, 4-8

Limit value of DO loop, 3-3
Line continuation field, 1-3
Line printer record, 6-1
Line spacing, 6-15
Lists, input-output, 6-18
LOGICAL mode, 5-6
LOGICAL statement, 4-1
Logical
AND statement, 1-13
constants, 1-5, 1-7, 1-12

Initial value DO loop, 3-3

1-4

INDEX (Cont)

L (Cont)
Logical (Cont)

Object program definition, 1-1

device number, 6-2
expressions, 1-12
IF statement, 3-8
negation, 1-13
operators, 1-13
records, 6-1, 6-2
values, 3-7
variable format, 6-17
variables, 1-12

Object Time FORMAT specifications, 6-16

o control character, 6-4
Octal fields, 6-13
Octal numbers, 1-5
Operands, 1-10, 1-13
Operators
arithmetic, 1-10, 1-11
logical, 1-13
relational, 1-12

Long records, 6-21, 6-22
Loop termination, 3-4
L-type conversion, 6-12

.OR., 1-13
OTS error statement, 3-2

OTS libraries, C-1, C-2

Magnetic tape record, 6-1, 6-24

Output record data placement, 6-14

Main program sample, 5-3

Output - see Input-output

Masking, 6-13, 6-17

Overprinting, 6-15

Mode combinations in relational expressions,
1-13
Mode conversion, arithmetic, 2-1, 2-2
Mode dec larati on statement, 4-2
Mode specification, 4-1, 4-2, 5-6
Modes of expressions, 1-11
Modes of variables, 1-9
Multiple entries and returns, 5-9
Multiplication, 1-10

P
Paper tape reader and punch record, 6-1
Parentheses - ( ) - in expression, 1-11, 1-14
Part-word notation, 2-2
PAUSE statement, 3-1, 3-9
P control character, 6-4
PDP-9 compilers and libraries for extended
FORTRAN, C-1, C-2
PDP-15 FORTRAN facilities, C-1

Physi cal records, 6-1, 6-2, 6-3

Naming subroutines, 5-8

Precedence of logical expressions, 1-13

Negative values, 3-3

Print position (tabulating), 6-14

Nested DO statements, 3-5

PRINT statement, 6-17, 6-20

Non-executable statements, 1-2

Program
beginning, 1-2
fields, 1-3
structure, 1-2

Normal exit from DO loop, 3-5
. NOT., 1-13
Numbers, 1-5

P scale factor, 6-10

Numeric IF statement, 3-8

1-5

IN DEX (Cont)

Quotation mark (' or ") usage, 1-2, 1-8, 6-4,
6-18, 6-22

S (Cont)
Sequential access, 6-2
Sequenti 01 input-output statements, 6-19
Sharp sign (#) usage, 1-5, 6-18, 6-22

R
Range of the DO loop, 3-3
R control character, 6-4
READ statement, 6-16, 6-17, 6-19, 6-22,
6-24

Single integer variables, 6-5
Slash (I) usage, 6-3
Source program, 1-1, 3-1
Spaces as field delimiters, 6-16
Spacing, vertical, 6-15

REAL mode, 5-6

Special characters, 1-2

REAL statement, 4-1
Real constants, 1-5, 1-6
Real variable format, 6-17

Specification statements, 4-1
Statement
definition, 1-1
field, 1-3
functional categories, 1-1
function definition, 5-5
functi ons, 5-2
number field, 1-3
sequence, 1-2
syntax, 6-3

Record
changing a single, 6-20
number, 6-22
size, 6-22
Records, 6-1, 6-2, 6-3
blank, 6-3
logical, 6-1
physical, 6-1, 6-3

Statements,
arithmetic, 2-1
control, 3-1
data transmission, 6-1, 6-2
device control, 6-24
executable, 1-2
input-output, 6-17
mode declaration, 4-2
non-executab Ie, 1-2
spec ifi cati on, 4-1

Recursion - see loop
Relational
expressions, 1-12, 1-13
operators, 1-12
RETURN statement, 5-6, 5-8
REWIND statement, 6-24
R-type conversion, 6-11

Statement summary, A-1
STOP statement, 3-9

S
Sample main program, 5-3

Storage, common, 4-4
Storage

subprogram, 5-4

allocation, 4-3
sharing - see COMMON

Scale factor (P), 6-10
Science library, C-2

Storage of
array, 1-9
data, 6-1
double precision constants, 1-7
Hollerith constants, 1-7, 1-8
rea I constants, 1-6

SEEK Iibrary subroutine, 6-23
Segmented files, 6-24
Sequence of statements, 1-2

1-6

--,

INDEX (Cont)
S (Cont)

Structure of program, 1.... 2

Unary plus or minus, 1-10

Subprograms, 4-3

Unconditional GO TO statement, 3-1

BLOCK DATA, 4-7, 4-8
definition, 5-1
names, 4-1, 4-4
sample, 5-4
statements, 5-1
Subroutine calls, 5-9

Unformatted record, 6-1

v
Values, logical, 3-7
Variable type formats, 6-17

SUBROUTINE statement, 5-6, 5-9

Variables, 1-8, 1-9,4-1

Subroutines, 5-1, 5-7, 5-8

Subscripts, 1-9

dummy, 5-1
mode, 1-9
subscript, 1-9

Subtraction, 1-10

Vertical spacing, 6-15

library, 6-23

Symbol ic addressing, 2-3, 3-2
W

Symbolic address label, 2-1

WRIT E statement, 6-17, 6-24

Symbol ic name, 2-1

direct, 6-20, 6-22
sequential, 6-19, 6-20

Syntax of statement, 6-3

T
TAB, 1-3
Tabulating, 6-14
T control character, 6-4

X
X conversion control character, 6-4
X blank (carriage control), 6-14, 6-15
.XOR. (exclusive OR), 1-13, 3-7

T descriptor, 6-14
Teletypewriter record, 6-1

Temporary file, 6-20, 6-21, 6-22

Zeroes, leading, 1-5

Termination of DO loop, 3-6
Transfer of control, 3-1
Transfer of data, 6-17
Transmitting data, 6-1
.TRUE., 1-7, 1-12,3-7
Truncation, 1-5
TYPE statement, 6-17, 6-20

1-7

-.,

READER'S

COMMENTS

Digital Equipment Corporation maintains a continuous effort to improve the quality and usefulness
of its publications. To do this effectively we need user feedback -- your critical evaluation of
this manual .
Please comment on this manual's completeness, accuracy. organization, usability and readability.

Did you find errors in this manual?

If so, specify by page.

How can this manual be improved?

Other comments?

Please state your position . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ Date: _ _ _ _ _ __
Name:

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

Organization:

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

Street: _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ Department: _ _ _ _ _ _ _ _ _ _ _ _ _ __
City: _ _ _ _ _ _ _ _ _ _ _ _ State: _ _ _ _ _ _ _ _ _ _ _ Zip or Country_ _ _ _ _ __

- - -

- - - - - - - - - -

- - - -

Fold Here --------------

- - Do Not Tear - Fold Here and Staple -

- - - - -

- -

FIRST CLASS
PERMIT NO. 33
MAYNARD. MASS.
BUSINESS REPLY MAIL
NO POSTAGE STAMP NECESSARY IF MAILED IN THE UNITED STATES

Postage will be paid by:

Digital Equipment Corporation
Software Information Services
146 Main Street, Bldg. 3-5
Maynard, Massachusetts 01754

HOW TO OBTAIN SOFTWARE INFORMATION
Announcements for new and revised software, as well as programming notes,
software problems, and documentation corrections are published by Software
Information Service in the following newsletters.
Digital Software News for the PDP-8 & PDP-12
Digital Software News for the PDP-II
Digital Software News for the PDP-9/15 Family
These newsletters contain information applicable to software available from
Digitalis Program Library, Articles in Digital Software News update the
cumulative Software Performance Summary which is contained in each basic
kit of system software for new computers. To assure that the monthly Digital
Software News is sent to the appropriate software contact at your installation,
please check with the Software Specialist or Sales Engineer at your nearest
Digital office.
Questions or problems concerning Digitalis Software should be reported to
the Software Specialist. In cases where no Software Specialist is available,
please send a Software Performance Report form with details of the problem to:
Software Information Service
Digital Equipment Corporation
146 Main Street, Bldg. 3-5
Maynard, Massachusetts 01754
These forms which are provided in the software kit should be fully filled out
and accompanied by teletype output as well as listings or tapes of the user
program to facilitate a complete investigation. An answer will be sent to the
individual and appropriate topics of general interest will be printed in the
newsletter.
Orders for new and revised software and manuals, additional Software Performance Report forms, and software price lists should be directed to the
nearest Digital Field office or representative. U.S.A. customers may order
directly from the Program Library in Maynard. When ordering, include the
code number and a brief description of the software requested.
Digital Equipment Computer Users Society (DECUS) maintains a user library
and publishes a catalog of programs as well as the DECUSCOPE magazine
for its members and non-members who request it. For further information
please write to:
DECUS
Digital Equipment Corporation
146 Main Street, Bldg. 3-5
Maynard, Massachusetts 01754

- - - - - - -

- - -

- - - - - - - - - -

- - - -

Fold Here - - - - - - - - - - - - - -

- - Do Not Tear - Fold Here and Staple -

- - - - -

- -

FIRST CLASS
PERMIT NO. 33
MAYNARD. MASS.
BUSINESS REPLY MAIL
NO POSTAGE STAMP NECESSARY IF MAILED IN THE UNITED STATES

Postage will be paid by:

momoama
Digital Equipment Corporation
Software Information Services
146 Main Street, Bldg. 3-5
Maynard, Massachusetts 01754

•