Digital PDFs

XX-D7B27-A4

September 1965

57 pages

Original

3.5MB

Document:	DEC-6-0-TP-FII-LM-FP-ACT00 FORTRAN II Language Programming Manual
Order Number:	XX-D7B27-A4
Revision:
Pages:	57
Original Filename:	http://bitsavers.org/pdf/dec/pdp6/DEC-6-0-TP-FII-LM-FP-ACT00_FORTRAN_II_Language_Programming_Manual_Sep65.pdf

OCR Text

DEC-6-0-TP-FIT-LM-FP-ACTOO

DIGITAL EQUIPMENT C.ORPORATION

•

MAYNARD, MASSACHUSETTS

DEC-6-0-TP-FIT-LM-FP-ACTOO

PDP-6

FORTRAN II

PROGRAMMING

DIGITAL EQUIPMENT CORPORATION

•

LANGUAGE
MANUAL

MAYNARD, MASSACHUSETTS

FOREWORD
This is a reference manual describing the specific statements and capabilities of PDP-6 FORTRAN II.

Familiarity with the basic concepts

of FORTRAN programming on the part of the reader is assumed.

III

CONTENTS

INTRODUCTION

00 . . 0

Basic Elements

0000000000

0000000000000000

The Character Set

000000

000

1-1

IF Stateme nts

1-1

•

4-3

Range •••.•.•••••••.••••••

4-3

DO Statements

Number Representation

1-1

CONTINUE Statement

1-2

PAUSE Statement

Variables.................

1-3

ST OP Statem ent

Operation Symbols

1-4

Constants

•

000

0000

Speci a I IF Statements

1-1

•

4-2

FORTRAN Words..........
0

•

000000.

4-2

4-5

•

4-5

0000000000000000000

5-1

Statement Labels

1-4

General Format Rules

1-4

Dummy Arguments

5-1

Line Formats..............

1-5

Funct ions •••••••••••••••••••••

5-1

Control Characters

1-6

External Functions

5-1

00000.00

1-6

Intrinsic Functions

5-3

END Statement

00000

•••

SUBPROGRAMS

•

Subroutines ..........••••.•...

ASSIGNMENT STATEMENTS
Arithmetic Expressions

00000000

0000000

Arithmet ic Operators.
Formation Rules

00000000000

Evaluation of an
Expression
o ••
000

.00

•• 0

000000000

The Arithmetic Statement
Assignment Rules

00.

0000.00000

Defining Subroutines

2-1

Call ing Subroutines

2-1

000000.

00000000

Return •••••••••••.•••.••••••••

5-4
5-5
5-5

2-1
INPUT-OUTPUT

0.0

Use of Parentheses

Bool ean Statements

000

2-1

5-4

2-1

Input-Output lists

2-3

.... 0

6-1

00000000.0000

6-1

00 . . . . . . . . . . . 0

Input-Output Statements

2-3

00000.00

Executabl e Statements

2-4

FORMA T Statements.

•

6-1
6-1
6-3

2-5

APPENDIXES
SPECIFICATION STATEMENTS .... o.

3-1

Array Dec larators

3-1

Dimension Statements

•

••

0000000. 0

3-1

DIAGNOSTICS

SPEC IAL PDP-6 FORTRAN II
STATEMENTS

A2-1

FUNCTION AND SUBROUTINE
LINKAGES ..

A3-1

SUMMARY OF STATEMENTS....

A4-1

FORTRAN II OPERATING
SYSTEM ••..•••.

A5-1

Common Statements

00000

Equivalence Statements
Restr ict ions

0000000

000.0.0

••• 0

•••••• 0

CONTROL STATEMENTS ..........
GO TO Statements.

00'

........ 0

3-2

3-4

4-1

00.0.

......... 0.0....

............... 0

4-1
v

00........

••••••••••

•

PDP-6 FORTRAN II COMPILER
OPERATING INSTRUCTIONS ••

•

A1-1

A6-1

CONTENTS (continued)
A P PEN 0 I XF S (c 0 n tin u e d )
A7 LIMITATIONS ON 9K
FORTRAN II COMPiLER ••••••••

3 Use of Continuation Characters ••••

1-6

4 Properties of Arithmetic
A7-1

Express ions .•..•.•••.••••...••..

2-2

5 Arithmetic Statements ••••••••••••

2-3

ILLUSTRATIONS

TABLES

Example of Subscripts •••••••••••

1-4

1 External Functions •••••••••••••••

2 FORTRAN Statement Card •••••••

1-5

Intrinsic Function ••••••••••••••••

5-2
5-4

CHAPTER 1
INTRODUCTION

BASIC ELEMENTS

labels, and commands. Of these, the first four
are most similar to the "words" used in writing

The Character Set

ordinary formulas. For example, as in ordinary
mathematics:

The symbols or characters which are meaningful

y = 2 x + 3 cos (z)

in PDP-6 FORTRAN are:
Blank

I
J
K

l'
/
,
(

constants, = and + are operators, and cos is a

function name. Statement labels are integers

Z
0

L
M

used to "name" statements within a program.

Commands are special FORTRAN words, e.g.,

DIMENSION, GO TO, ASSIGN, IF, etc.

P
Q

4
5

Each command word is an integral part of its as-

A
B
C
D
E
F
G

6
7

sociated type of statement. Therefore, com-

where x, y, and z are variables, 2 and 3 are
W
X

R
S
T
U

8
9

mands will be discussed along with their statement types.
Number Representation

Blanks, with two exceptions, are ignored and

There are two types of numbers represented in

may be used as desired to make the program

FORTRAN: integers and real*.

neater in appearance and more readable.

numbers may assume posi tive, negative, and

For

Both types of

example, READTAPE and READ TAPE are equiv-

zero va lues. Integers may, of course, assume

alent. The exceptions are FORMAT statements

only integral values and always are represented

and ASC II constants.

exactly in the machine provided their magnitude
is not greater than the largest integer containable in the machine registers.

FORTRAN Words

In the PDP-6,

this largest integer is 2 35 -1. A real number x

FORTRAN words fall into seven categories:

may take on values in the range

constants, variables, subscripted variables,

10 -38~

function names, operation symbols, statement

Ix I 10+38 and x = 0
~

*This use of the term real should not be confused with the mathematical usage. In FORTRAN, real applies
only in the limited sense described above (formerly referred to as floating point).

1-1

However, though a number is within the range,

Real Constants - Any number representable by

its representation in the machine may only be

a string of digits, from the set 0, 1, 2, •••••• 9,

approximate. The conversion to machine repre-

written with either a decimal point, exponent

sentation is accurate to approximately eight

or both. The general form is:

significant digits (27 binary bits).
Except for zero, two numbers of different types
never have the same representation in the ma-

where X., Y., and Z. are di gi ts. The plus si gns

chine. For example, if 3 is the integer three,

are optional; if no signs are given, plus is as-

and 3.0 is the type real three, the representa-

sumed. Either the stri ng Xl •••••• X n or the

tion of 3 is not at all similar to the representa-

string Y 1 •••••• Yn may be omitted if their value

tion of 3.0.

is zero. If Zl and Z2 are both zero, E ± Zl Z2

••

•

may be omitted. (However, the number must
contain either a decimal point or an E.)

Constants
Numerical

The range of real constants acceptable to PDP-6
FORTRAN II is:

In FORTRAN II, a number may be explicitly
named by writing a constant. However, there

10- 38

10+38

and zero.

are two forms for naming constants: integer and
real.

Some examples of real numbers are:

Integer Constants - Any number representable

6.023E23
1.66E-16
-.0056

by a string of digits, from the set 0, 1, 2,

-6.023E23
+72E 12
4.2

•••••• 9, written without a decimal point or
exponents

Boolean Constants
Any number representable by a string of digits
from the set 0,1,2, ••• 7.

Where X. is any digit. The plus sign is option-

•

al; if no sign appears, plus is assumed.
where X. is any octal digit. The plus sign is op-

The range of an integer constant I acceptable

•

tional; if no sign appears, plus is assumed. A

to PDP-6 FORTRAN is:

minus sign implies the l's complement of the
_2 35 +1 < I < +2 35 -1
Note: 2

number. The range of the Boolean constant B is

= 34,359,738,368.

Examples:
Some examples of integers are:

-178192

+131701

377777

-0
1-2

40404040
+15

ASCII Constants

two types of variable names identified as follows:
1. Integer variable names must begin with
one of the letters I, J, K, L, M, or N.

ASCII constants are used to convey textual information. They have the form:

2. Real variables are designated by names
beginning with any other letter.
where 0< i~ 5. The c's are any of the charac-

Some integer variable names are INDEX, KDATA,

ters from the character set. Note that there

K359, and MSIX. These are real variable names:

must be exactly i characters including blanks.

XZERO, COUNT, and FICA.

Examples:

Subscripted Variables
A = 5HPDP-6
S = 4H YES

An array is a grouping of data. A column of
figures, the elements of a vector, a list, and a

In the example above the variable on the left

matrix are all arrays. In mathematics, an ele-

must be of a real type.

ment of an array is referenced by means of a
symbol denoting the array and subscript identi-

Variables

fying the position of the element.

Variable Names
For example, the sixth element in a vector v is
A variable name is composed of from one to six

designated v 6 • Likewise, the fourth element in

characters according to the rules:

the tenth column of a matrix b is identified as

1. The only characters which may be used
in a variable name are A through Z and 0
through 9. (Blanks are ignored.)

b4 10· In general, an element of an n-dimen,
tional array m is designated by

2. The first character must be a Iphabeti c •

Some examples of acceptable variable names

•
I

•

1,2,3, ..•...•...••..• n~

In FORTRAN II, array elements are similarly

are K, P51, and EPSILO.

identified. The array is provided with a name,
subject to the same formation and type rules as

Some incorrect variable names are 6MERGE

the names of variables. All the elements in the

(first character not alphabetic), G 1 .5 (i lIega I

array have the same type. The subscripts which

character included), and EPSILON (too many

identify an element of the array are enclosed in

characters) •

parentheses and separated by commas. The two
Type Rules for Variables

elements, v 6 and b4 10 would have the following

There are two types of entities pesignated by

notation: V(6) and B(4, 10). Subscripts may be

, variable names: integers and reals. Corre-

quite diverse in form; in fact, a subscript may be

sponding to the two types of entities there are

any acceptable FORTRAN arithmetic expression

1-3

(see Chapter 2) as long as it is integer-valued

operators are given in Chapter 2. An important

and less than 2 18 (real quantities are not al-

rule about operators in FORTRAN II expressions

lowed) •

is that every operation must be explicitly represented by an operator. In particu lar, the mu Itia. X(3,3)

plication sign must never be omitted. Likewise,

b. C(1+1, J+l)

si nce superscri pt notation is not avai lab Ie , a

c. N(I(I), J(l), K(2))

symbol for exponentiation is provided.

d. Y(J/3,(K-4) +12)
Statement Labels
Figure

Example of Subscripts
It is sometimes necessary for statements in a pro-

Note that the subscripts in Figure 1, example c,

gram to refer to other statements, e. g., in trans-

are themselves subscripted. Subscripting may

ferring control, referring to format statements,

be carried to any level. Each subscripted sub-

etc. To provide for such references a statement

script, i.e., 1(1L J(lL and K(2) in Figure 1,

may be preceded by a label. A statement label

is itself treated as a subscripted variable.

may be any string of one to five decimal digits,
i.e., any integer from 1 to 99999. Leading

Function Names

zeros are ignored; th us 99, 099, 0099, and
00099 are a II consi dered to be the same labe I •

Function names are special forms of variables
consisting of a name immediately followed by

GENERAL FORMATION RULES

an argument enc losed in parentheses. The function name represents a mathematical operation

Certain general rules hold for the formation of

to be performed on the argument such as finding

all FORTRAN programs and all FORTRAN state-

the square root of a number or determining the

ments. These rules will be dealt with here.

sine or cosine of an angle. Certain basic operations such as these are provided by the FOR-

Physically a FORTRAN program is divided into a

TRAN system and are called library functions.

sequence of lines. One line may be ei ther one

A detai led discussion of functions wi II be found

punched card or all the characters punched on a

in Chapter 5.

paper tape by one line of typing on a Teletype.
Logically a FORTRAN program is divided into
Operation Symbols

statements. Therefore, some correspondence must
be established between lines and statements. The

The operation symbols may be anyone of the
following:

** or t

rules are:

+ and -. These denote

1. There can be no more than one statement
per line.

arithmetic or Boolean operations depending on
the type of statement in which they occur. The

2. If a statement is too long to fit on one
line, it may be continued onto additional

rules for the use and binding strength of these

1-4

lines. In this case, the first line is known
as the initial line and each succeeding line
is known as a continuation line.

mats are discussed below:

Card Input
Each continuation line must be identified as

Each card contains exactly one line. The line

such by a special character. Which character

is further subdivided into four fields, i.e., four

is used, and where it appears on the line, de-

sets of conti guous columns.

pends on the input medium. (See below under
Line Formats for each medium). A PDP-6 FOR-

The four fields are columns 1 to 5, column 6,

TRAN \I statement may have a maximum length

columns 7 to 72, and columns 73 to 80. This is

of 660 characters.
Line Formats

known as a fixed field format where all the information within a certain field is to receive a

The format for a line depends on the input medi-

specified interpretation. (The interpretation

um. Since punched cards and punched paper

for each field is often printed at the top of the

tape may both be used on the PDP-6, both for-

card to assist the programmers, see Figure 2).

L~O~O:U"lT ~I--------.-----------------------------I
STATEMENT
NUMBER

FORTRAN

STATEMENT

IOENTlfICAT!ON

010000000000000000000000000000000000000000000000000000000000000000000000000000000
'12 3 4 5 6 7 , • 10 11 12 13 14 15 1& 11 " " 20 21 22 23 24 2S 26 27 28 29 30 31 32 33 34 :r.; 38 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 &0 61 62 &3 &4 &5 66 67 68 68 70 71 72 73 74 75 76 17 ii 79 so

111 1 1 1 1 1 1 1 11 1 1 1 1 1 1 1 1 111 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 11 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 11 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
2122 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
133 333 33333 33333333333333 3333333333 3 333 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3333333333 333 3 3 3\3 3 3 3333 3

4144 44 414 4 4 4 4 4 4 44 44444 4 4 4 4 4 4 4 4 4 4 4 4 444 4 44 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 414 44 4 4 4 4 4
5155555555555555555555555555555555555 555555 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 55 5 5 5 555 55 5 5 sis 55555551
I:

I &:& 6 6 61616 6 6 66666 6 6 6 6 6 6666666666666666666 6 6 6 6 6 6666666666 G6 6 6 6 6 6666 6 6 GGS G6 6 616 6 6 6 6 6 6 6 I

I
717777
77 77 7 77 77 77

n 77 77 77 77 77 7 77 7 7 7 77 7 77 77 7 77 7 7 7 i 77 7 7 77 7 77 7 7 7 77 77 7 7 7 77 7 77!7 77 7 7777 i
I

818 8 8 8 8 8 8 8 8 8 8 8 888 8 8 8888888888888 8 8 888 8 8 8 8 8 B8 8 8 8 888 8 8 8 8 8 8 8 8 8 888 8 2 J 8 8 B888 6 818 8 8 8 888 8

91999999 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 999 999 9 9 9 9 9 999 9 9 9 9 9 9 ~ 9 9 9 9 9 9 9 9 9 9 9 9 999 9 999 9 9 9i9 9 2 9 9 999

112 3 4 5 & 7 , 910111213141516171619202122232425262728 29 303132 33 i4 35 38 37 38 3940 41 4243 4445 4647 48'9l(J 51525354 5G 56 5751 ~ S06162&Ht65" ,1< Ga70" 12171 'I 15 76 1778)9'0
ICC 1-'8117

Figure 2

FORTRAN Statement Card

The interpretation for each field is:

Field 2 (Column 6). This one column field
contains the continuation character; if it is
blank, or punched with a zero, the line is
an initial line. Any other character,
punched in column (", denotes a continuation line.

Field 1 (Columns 1-5). This field is further
subdivided into two overlapping subfields,
i.e., column 1 and columns 1-5. Column
1 may contain one of several special control characters (see below). Statement labels appear in columns 2-5 and may extend
to column 1 if there is no control character.

Since nine continuation lines are allowed, the

1-5

programmer may use the digits 1 to 9 to indicate

Figure 3 gives an example of the use of the back

their proper order.

slash as a continuation character.

Field 3 (Columns7-72). Th.is field contains the statement. It cannot contain more
than one statement; however I it may contain as much of a single statement as will
fit in 66 or fewer characters.

X=X +(ARG 1+2. * ARG2+2. * ARG3+\
ARG4) /6.
Figure 3

Field 4 (Columns 73-80). This field is ignored by the compi ler and may be used to
place any kind of identifying information
on the card, e.g., sequence numbers.

Use of Continuation Characters

Control Characters
Occasionally it is necessary to instruct the compi ler to interpret a statement in some specia I
way. Two control characters are provided for

Punched Tape Format

this which must appear in the first character

On punched tape, lines must be delineated by a

position of the statement's initial line. They

carri age return/I i ne feed. There are two formats

are:
for punched paper tape:
C Desi gnates a comment statement. Such
a statement is ignored by the compiler except for being retained for printing along
with the compi lation. A comment statement
may not have continuation lines. However,
this is not a serious restriction since each
line of the comment is simply preceded by
a C.

1. Card Simulated Format - Paper tape may
be punched in a card format using the tab to
space to column 7. Each line ends with a
carriage return/line feed. Continuation
characters are punched in column 6. Lines
with no statement numbers or control characters begin with a tab or six spaces. Statement numbers in columns 1-5 or control characters in column 1 may be followed by a tab
or enough spaces to reach column 7.

B Indicates that Boolean operations are to
be compiled. (See Chapter 2.)

2. Column Free Format - Lines are delineated by carriage return/I ine feed. A
continuation line is identified by a back
slash as the last character before the terminator for the previous line. Statement
labels and control characters must precede
the statement and must be followed by a
colon.

END STATEMENT
The END statement must be the last statement of
every FORTRAN program. Its function is to indicate to the compiler that nothing more connected with the preceding program is to follow.

1-6

CHAPTER

ASSIGNMENT STATEMENTS

Assignment statements are statements of the form

Formation Rules

v = e, where v is a variable name and e is some

An arithmetic expression may be:

ari thmeti c expressi on. The execution of an as1. A constant, a variable name, subscripted
variable name, or function name.

signment statement replaces the value of v with
the value of e. There are two types of assign-

2. Any entity conforming to rule 1, with a
prefixed (unary) minus sign.

ment statements: arithmetic statements and
Boolean statements.

3. Two entities conforming to rules 1 or 2
with an infixed operation symbol.
4. Two entities conforming to rules 1, 2, or
3 with an infixed operation symbol.

ARITHMETIC EXPRESSIONS
The elements of an arithmetic expression are

5. Two entities conforming to rules 1, 2, 3,
or 4 and enc losed in parentheses.

constants, variables, subscripted variables,

6. An entity conforming to rule 5 with a
prefixed unary minus.

functions, and operators. An expression may
consist of a single constant, a single variable,

7. Two entities conforming to any of the
above with an infixed operation symbol.

a function, or a string of constants, variables,
and functions connected by operators.

All variables of an arithmetic expression must be
of the same type, except for exponen ts wh i ch may

Ar i thmeti c Opera tors

be of type integers in either integer or real exArithmetic operators are symbols representing

pressions.

the common arithmetic operations as follows:
Exponentiation

Multiplication

Division

Addition

ort

Figure 4 demonstrates the properties of arithmetic
expressions. Each expression is shown with its
corresponding algebraic form.

Evaluation of an Expression

Subtraction or unary minus

Normally, a FORTRAN expression is evaluated

The symbols are compi led as arithmeti c opera-

from left to right just as an algebraic formula.

tions unless the statement has the control char-

As ina Igebra, however, there are exceptions.

acter 'B I prefixed.

Certain operations are always performed before
A unary minus is the operator which precedes a

others, regardless of order. This priority of eval-

quantity whose value is to be negated.

uation is as follows:

2-1

l. Expressions within parentheses ( )

more precisely as follows: Operations are per-

2. Unary minus

formed in order of decreasing binding strength.

3. Exponentiation

** or'

A sequence of operations of equivalent binding

4. Multiplication

strength is evaluated from left to right.

Division

5. Addition
Subtraction

Examples g and h in Figure 4 illustrate the use
of functions as variables in an arithmetic expression. Included in these examples are SINF

The term binding strength is used to refer to an

(THETA) I COSF (THETA -1 .5) I and SQRTF (Z)

operator's relative position in a table such as

corresponding to the trigonometric functions sine,

the one above in which the operations are listed

cosine, and/W.

in the order of descending binding strength.
Thus, exponentiation has a greater binding

Whenever a function is encountered, it is eval-

strength than addition, and multiplication and

uated and the result treated as a variable in the

division have equal strength.

evaluation of the expression in which the function occurs.

The left-to-right rule can now be stated a little

Algebraic Expression

FORTRAN Expression

-a

-A

2
az + bz + C

A*Z""*2+B*Z+C
(A ** 2-B- 2)/ (A+B)** 2

41Tr

4. * PI * R ** 2/3.

-32

3z -2 (z+y)

(3.* Z** 2 - 2.*(Z+Y»/4.25

4.25

a sine + 2a cos (9 -1.5)

A * SINF (THETA) +2.*A*COSF {THETA-l.5}

2vz

2.*SQRTF (Z)/3.

3
Figure 4

Properties of Arithmetic Expressions

2-2

Use of Parentheses

thesis is read. When the expression has been completely scanned, the count should be zero.

As with ordinary algebra, parentheses may be

The Ari thmeti c Statement

used to change the normal order of evaluation.
For example:

The ari thmeti c statement re lates v, a vari able
name or array element name of type integer or

A + B* C

real, to an arithmetic expression e by means of
would result in A being added to the product of

the equa I si gn (=), thus:

Band C. However, if we want, instead, the
v =e

product of A + B times C we may write:

Such a statement looks like a mathematical

(A + B) * C

equation, but it is treated differently. The equal
Also, parentheses may be embedded in parentheses.
For example:

sign is interpreted in a special sense; it does not
merely represent a relation between left and

(A * (B + C) )** D

right members, but specifies an operation to be
performed; namely, replace the value of v with

The expression is evaluated from the innermost

the va Iue of e.

expressi on outward. That is, B + Cis formed,
then the product with A, and finally the expo-

A few illustrations of the arithmetic statement

nentiation. The expression within parentheses

are given in Figure 5.

simply becomes a numerical argument for the rest
of the expression.

a. VMAX=VO+Z*TO
b. T = 2.*PI*SQRTF(1.0/G)

Expressions with many nested subexpressions can

c. PI =3. 14159

become very difficult to read making it difficult

d. THETA=OMEGAO*T+ALPHA*T**2/2.

to be sure that each left parenthesis is properly

e. MIN=M + N + 5

paired with a right parenthesis. If they are not

INDEX = INDEX +2

properly paired, an error diagnostic is printed.
Fortunately, the test used by the computer turns

Figure 5

Arithmetic Statements

out to be a very simple way of checking by hand.
The equal sign is considered to have a lower

Consider the following example:

binding strength than all of the operators. This
Z* (P* (SINF(THETA)+S) / (Z**2-(B**2+C**2)))
1 2
3
2 1 2
3
210

means that the whole of the expression on the

The procedure is this: starting with a count of

by == is performed. By this definition the state-

zero, scan the expression from left to right. In-

ment in example f of Figure 5 would mean, "add

crease the count by one for each ri ght parenthesis.

two to the current value of INDEX. The result

Decrease the count by one when a right paren-

is the new value of INDEX."

right is evaluated before the operation indicated

2-3

It should be noted that all variables occurring to

The internal arithmetic statement I V==SQRTF

the right of an equal sign must have been defined

(2. * G*Y), must be set off from the rest of the

and calculated when the expression is evaluated.

statement by parentheses. The complete state-

If the variable on the left of the equal sign was

ment in this illustration is a concise way of ex-

previously undefined, it will be defined by the

pressing the following type of mathematical

ari thmetic statement.

procedure:

Assi gnment Rules
In an arithmetic statement, the value of the ex-

Let

Q ==A/v

where

=/2;;

In the single FORTRAN statement both these

pression to the right of the equal sign replaces

equations are evaluated starting with the inner-

the value of the variable on the left, according

most statement and va I ues are assi gned to V and

to the following rules:

Q.
Another result of treating the equal sign as an

v is
type

e is
type

Rule

operator is that iust as there may be a series of
additions, a+b+c, there may be a series of re-

integer
integer
real
real

integer
real
integer
real

v == e

placements, a=b=c=cl. Since the operand to the

v = FIX(e)
v ==FLOAT(e)

left of an equal sign must be a variable, only

v =e

the ri ghtmost operand, represented by d above I
may be an arithmeti c expression. The statement

The name FIX* means truncate any fractional

is interpreted as follows: "Let the value of the

part and convert to integer representation. The

expression d replace the value of the variable c I

name FLOAT means convert to real representation.

which then replaces the value of the variable b"
and so on. ** In other words, the value of the

Internal Arithmetic Statement

rightmost expression is given to each of the varAn important result of treating the equal sign as

iables in the string to the left. A common use

an operator is that it may be used more than once

for this construction is in setting up initial values:

in an arithmetic statement. Consider the followVZERO=SZERO==AZERO== O.

ing.

P==PO=4* ITM-K
Q == A/(V=SQRTF(2.*G*Y) )

Each replacement conforms to the above rules.

* Truncate in 2's complement means find the greatest integer in X. Hence XFIXF(O .5)= 0 and
XFIXF(-O.5)=-1. The function XINTE(X) performs truncationtothe greatest integer inthe absolute
value. Hence XINTF(O.5)= Oand XINTF(-O.5)=O.
**This may seem at first to violate the left-to-right rule. However, whenever an equal sign is encountered
in scanning a statement, it cannot be executed until all operations of higher binding strength have been
performed. Thus, execution of each equal sign (replacement) is deferred unti I the expression on the right
has been evaluated. The replacements then occur in reverse order as the evaluation works back to the leftmost variable.
2-4

BOOLEAN STATEMENTS
The arithmetic statement may be used to perform
Boolean operations if a B is placed in the first
character position of the statement line. In this
case the symbols for the ari thmeti c operations
take on the following meanings:

Boolean Algebra

Resu It of Operati on (r.)
I

argl.
I

arg2.
I

0
0
1
1

0
1
0
1

0
1
1

0
0
0
1

1
0
0
1

0
1
1
0

inclusive OR

- (binary)

exclusive OR

- (unary)

l's complement

AND

equivalence

For the shifting operation the second argument

** or'"

shifting

must be an integer, i.e., the form must be:

For the unary minus:
r. = 0 if argo = 1
I

r. = 1 if argo = 0
I

arg ** ivar
For all but the shift operation, the mode of the
where arg is an arbitrary type and ivar is an in-

data is in a sense irrelevant. The Boolean op-

teger. arg is shifted right if ivar is negative,

erations are performed on the full 36 bits of the

left if ivar is positive, by ivarmod36 positions.

data word, complementing etc., ona bit by bit

That is:

basis without regard for the significance of the

r. = argo ± •
I
I Ivar mod36

bits.
Examples:
The effect of these operations may be defined

B A = B+213776347777

as follows: Let r. be the i th bi t of the 36-bit

B C = C*A-D

result. Let argl,l be the ith bit of the first ar-

B F =C**-3

gument; let arg2. be the i th bit of the second

B H =E/G

argument. Then for each operation we have:

2-5

CHAPTER 3
SPECIFICATION

STATEMENTS

There are three types of specification statements:

name. The d's, the declarator subscripts, must

dimension statements, common statements, and

be integer constants. Their presence is suffi-

equivalence statements. These are called speci -

cient to inform the compiler that an array is

fication statements since they specify structural

being declared. Their number indicates the di-

properties of the program which must be known

mensionality of the array; their magnitude, the

to the compiler if it is to set aside enough stor-

maximum value which each subscript may assume.

age for arrays, properly relate different variables,
etc.

Array Successor Function
Arrays are stored by columns in the array storage
NOTE: All specification statements
must appear before any executable statement. (Executable statements are assi gnment statements, control statements, and
the executable input-output statements.)

area. That means the following function may be
used to establish the location of a given element
relative to the beginning of the storage area:
s = i 1 + d 1 (i 2 - 1) + d 1d 2 (i 3 - 1)

Array Declarators

+ .... +d 1d 2 ... d n (in -1)-1

In an array of dimension n, each subscript is alwhere i l' i 2 ,·· .i n are the values of the n sub-

lowed to assume all integral values between 1

scripts. This function is important when an array

and some maximum, where d 1 , d 2 ,····· .d n

element must be identified by a single subscript.

represent the maximum values for each subscript.
The array will require:

NOTE: An array declarator must appear
for each array. It may appear in either
a dimension statement, or a common
statement.

words of storage. An array declarator is used
to inform the compiler of this requirement. It

DIMENSION STATEMENTS

has the form:

Dimension statements are provided explicitly to
declare arrays. They have the form:
where aname is the array name. It is subiect to

DIMENSION a 1 ,a 2 , •••••••••• a n

the same formation rules and type conventions
as ordinary variable names. All of the elements

where the a's are array dec larators. No two ar-

of the array are of the type dec lared by the

ray declarators may have the same array name.

3-1

COMMON STATEMENTS

The alternative is to list an array declarator in
the common statement rather than simply the ar-

The common statement allows the programmer to

ray name.

assign the same meaning to variable names in
different program units. Names appearing in

Considerable care must be taken to assure that

common statements are assigned storage loca-

variables which are to be identical occupy the

tions in the order in which they appear, begin-

same position in common. For example, assume

ning at location 14°8 ,

in one program we have the fo Ilowi ng statement:

Common statements have the form:

COMMON A, B, C
where B is an array name. Then I assume ina
different program unit we have:

where each v may be either a variable name,
COMMON X, Y, Z

an array name f or an array dec larator. Each
name must be distinct; e.g., a listed variable

where X, Y, and Z are all single variables.

name must not be repeated as an array name.

Then LaC (X) = LOC (A); however, C wi II not
be in the same location as Z. Rather, Y wi \I

Since the storage location of a variable is de-

equal the first element of B, and Z will equal

termined by its order in the common list, vari-

the second element of B. To have LOC (C) =

ables in different program units need not have

LaC (Z) and sti \I write the list in the above order,

the same name to have the same meaning. For

we must declare Y to be an array of the same size

example:

as B. That is, if B is a 3 x 3 array, we could declare Y to be a 3 x 3 array or a 1 x 9 array.

COMMON X, I, Z
COMMON A, J, C
may be two common statements appearing in two

EQUIVALENCE STATEMENTS

different program units. The effect of the two
statements is to assign the following locations

Equivalence statements are used to assi gn two or

to the listed variables:

more variables in a program unit to the same

LaC (X) = 1408

LaC (A) = 1408

LOC (I) = 1418

LOC (J) = 1418

LOC (Z) = 1428

LaC (C) = 1428

storage locations. Such an assignment may be
made for several reasons, perhaps the most common being to save storage. If a program contains
several variables which are never used simulta-

If an array name appears in a common statement,

neously or I more accurately, the value of one

enough sequential common locations must be set

need not be retained while another is being used;

aside to contain the array. Consequently, the

storage may be saved by having them all use the

array must be declared in a dimension statement.

same location.

3-2

The general form of the equivalence statement

LOC (G) = LOC (0) = 1408

is:

LOC (F)

= LOC (B) = 1418
LOC (A) = 1428
LOC (C) = 1438

where each e is a list of the form v 1 , v 2 ,···v n

A question arises as to what happens if an array

and n.2: 2. In turn, each v may be a variable

element appears as equivalent to a common vari-

name or an array element name. All of the en-

able. The answer is:

tities named within a single set of parentheses
are assigned to the same location in storage.

The common variable is placed in common according to its position in an equivalence statement. The array is then assigned to common locations so that the
specified element is equivalent to the
specified common variable.

The array element is identified by a single subscript, which is equal in value to the array successor function. If an array name is listed without the Iinear subscript, a subscript of 1 is
assumed.

For example, consider:

If an element in an equivalence group also ap-

COMMON A, B, C, 0

pears in a common statement, all the elements

EQUIVALENCE (E(l),C)

of the group are considered to be in common.
The result would be:

The common block is reordered so that the items
appearing in equivalence statements appear first

LOC (E(l)) = LOC (C) = 140

and in the same order as in the equivalence

LOC (E(2)) = LOC (A) = 141

statements. For example, suppose a program

LOC (E(3)) = LOC (B) = 142

had the common statement:

LOC (E(4)) = LOC (0) = 143
LOC (E(n+1)) = 140 + n

COMMON A, B, C, 0

However, the pair of statements:

This would make the following assignment:
LOC (A) = 1408

LOC (C) = 1428

COMMON A, B, C, 0

LOC (B) = 1418

LOC (0) = 1438

EQUIVALENCE (E(5), C)

would have the following effect:
If there were also an equivalence statement:

LOC (E(l)) = 140

EQUIVALENCE (0, G), (F ,B)

LOC (E(2)) ;:: 141
LOC (E(3)) ;:: 142
LOC (E(4)~ ;: 143

The result would be:

3-3

lOC (E (5» = lOC (C) = 144

Restrictions

lOC (E(6» = lOC (A) = 145

A common variable may appear in only one
equivalence statement unless its storage requirements are equal to or exceed all other variables

lOC (E(7)) = lOC (B) = 146

in the succeeding equivalence groups in which
it appears.

lOC (E(8» == LOC (D) == 147

3-4

CHAPTER

CONTROL STATEMENTS

GO TO (labell, labeI2, ••• Iabeln), iname

Normally, the executable statements of a
FORTRAN program are executed in the order

where iname is an integer variable name. The

of their appearance. However, the order of

next statement executed, following the computed

execution may be altered by any of the follow-

GO TO, wi II be the statement corresponding to

ing control statements: GO TO, IF, CALL,

the nth label in the list; n being the current val-

RETURN, CONTINUE, STOP, PAUSE, and DO.

ue of iname.
The CALL and RETURN statements are used with

Example:

subrouti nes and wi II be di scussed in Chapter 5.
GO TO (21, 4, 36, 18, lOa), INT
The control statements (except for STOP, CONIf INT = 3, the next statement to be executed is

TNuE, and RETURN) contain labels to identify

statement 36.

those statements which are to be executed following execution of the control statement itself.

Assigned GO TO - There are two forms of this

All labels in a control statement must refer to

statement, one being similar to the unconditional

executable statements.

GO TO and one simi lar to the computed GO TO.
GO TO STATEMENTS

In both cases, the effect of the statement depends
on the current value of an integer variable. The

Unconditional GO TO - This statement has the

two forms are:

form:

GO TO iname

GOTOn
and

where n is a statement label. Execution of this
statement results in the statement labeled n being

GO TO iname, (labell, labeI2, ••• Iabeln)

executed next.
The first form simply transfers the execution seExample:

quence to the statement whose label is equal to
GO TO 10

the value of iname. The second form matches
the current value of iname against the labels in

Computed GO TO - The computed GO TO is

the list. If a successful match is found, the ef-

an n-way branch, where the branch selected

fect of the statement is the same as the first form.

depends on the value of an integer variable.

If not, control passes to the next executable

The form is:

statement.

4-1

equa I condi tions may be tested by repeatin9

The statement:

label.

ASSIGN n TO iname

Examples:

assign the label n to the integer variable iname.

1. IF (K) 23,64,100
If more than one ASSIGN statement refers to the

2. IF (A**I*B+C) 23,64,23

same integer variable name, the value assigned

3. IF (K-100) 5,5,10

by the last executed statement is the current
value.

Boolean IF - The Boolean IF is a two way branch.

I t has the form:

Examples:
ASSIGN 21 TO INT

IF (exp) labell, label2
where exp is a Boolean expression. The next
statement executed after the IF is:

GO TO INT
labe 11 if exp = 0
labe 12 if exp I- 0

ASSIGN 1000 TO INT
Examples:

IF (A*672) 101,3
GO TO INT, (2,21,1000,310)

IF (-(A+B)) 4,76
IF (-(AlB)) 36,21

IF STATEMENTS
The Boolean IF must be identified by a B in colArithmetic IF - The arithmetic IF is a three way

umn 1.

branch. It has the form:
IF (exp) labell, label2, label3

NOTE: The character * may replace any
statement label in an arithmetic or Boolean
IF statement; this causes control for the
branch corresponding to the * to pass to
the next executable statement.

whereexp is any arithmetic expression of type
real or integer; and label1, label2, and label3
are three statement labels. The next statement
to be executed is:

Special IF Statements

labell if exp < 0

There are several forms of IF used to detect var-

label2 if exp = 0

ious conditions in the state of the machine.

label3 if exp > 0
Sense Lights

The three labels need not all be distinct; less

IF (SENSE LIGHT i) labell, label2

than or equal, greater than or equal, and not

4-2

where i is an integer constant and 1':::: i ,:::: 36.

to labell is executed next. Otherwise, the

If the desi gnated Ii ght is on, it is turned off

statement corresponding to label2 is executed

and the statement corresponding to labell is

next.

executed next; otherwise, control transfers to

Examples:

the statement corresponding to labe 12.
IF ACCUMULATOR OVERFLOW 14,2
IF QUOTIENT OVERFLOW 13,51

NOTE: Although the PDP-6 has no sense
lights, the FORTRAN II Compiler sets aside a register for sense light tests.

NOTE: Both ACCUMULATOR and QUOTIENT OVERFLOW tests examine the overflow flag in the PD P-6 •

The statement:
SENSE LlGHTi

DO STATEMENTS

where i is an integer constant and l~ i ~ 36
causes sense light i to be turned on. The state-

DO statements provide a convenient means for

ment SENSE LIGHT 0 causes all sense lights to

causing the repeated execution of a series of

be turned off.

statements along with incrementing an index for
each repetition.

Sense Switch
The general form of a DO statement is:
Sense switch settings may be tested by the stateDO label i =k, I, m

ment:

where label is the statement label of an execut-

IF (SENSE SWITCH i) labell, label2

able statement in the same program unit as the
Control goes to labell if sense switch i is up and

DO and sequentially following the DO. This

label2 if it is down;

statement is known as the terminal statement.

i is an integer constant in

the range ()~ i ~ 35 •

The terminal statement must not be any form of
GO TO, IF, STOP, PAUSE, or DO statement.

Overflow - There are two overflow conditions
which may be tested. The statements provided
are:

Each DO statement has an associated range which

IF ACCUMULATOR OVERFLOW labell, label2

includes the first executable statement following

IF QUOTIENT OVERFLOW labell f label2

the DO and extends to and includes the terminal
statement.

If an overflow condition is indicated, the indi-

i is an integer variable name known as the con-

cator is cleared and the statement corresponding

trol variable~ it is the index to be incremented.

4-3

k, I, and m are integer expressions.

i is ini-

When a DO is contained within a DO, i.e.,

tially set to the value of k. After execution of

nested, the range of the contained DO must be

the terminal statement, the value of i is incre-

a subset of the range of the containing DO.

mented by the value of m. If the value of i is

That is, all the statements of the inner DO, from

less than or equal to the value of I, control

the DO itself to and including its terminal state-

passes to the first executable statement following

ment, must be statements in the range of the

the DO. Otherwise control passes out of the

outer DO.

range of the DO.
The following diagram illustrates proper and improper nesting of DO's.

m may be omitted, in which case its value is
assumed to be 1 •

. - - 0 0 -------1
No statement within the range of the DO should

DO -------2

change the value of i, k, I, or m.

I~~ -------3
~------4

Notall statements in the range of the DO need
appear sequentially between the DO and the
terminal statement. For example, a statement

0-------5

-------6
~

within a DO may transfer control to a procedure,

which, when terminated, returns control to a

statement within the range of the DO. This may
be diagrammed as follows:

-------7

Allowable Nesting

0000------------1
DO ----------2
SUBPROGRAM
DO ---------3
In this case, the subprogram is considered to be
within the range of the DO.
Improper
It is often desirable to have DO's within a DO.
This might occur, for example, when a calcu-

A program branch must not occur from a state-

lation is being performed with multiply sub-

ment outside the range of a DO to a statement

scripted parameters. An innermost DO will com-

within the range of a DO. However, branches

pletely run through all the values of one sub-

are permitted out of the DO, in which case the

script for each step of an outer DO.

control variable is defined and is available.

4-4

Branches 2, 5, and 6 in the following diagram

cessary to have a conditional GO TO or IF

are permitted; branches 1, 3, and 4 are not.

statement in the range of a DO so that, if the
stated conditions are met, the control variable
is incremented and another repetition of the
entire range begins. This can be done by mak-

ing the terminal statement of the DO a CONTI N UE statement. The GO TO or IF statements
then simply transfer control to the terminal sequence of the DO.
PAUSE STATEMENT

PAUSE statements may have the form:

Finally, if two or more nested DO's share the

PAUSE n

same terminal statement, no statement outside
the range of the innermost DO may cause a

PAUSE 'MESSAGE'

where n is an octal constant and may be omitted.

branch to the terminal statement.

Execution of a PAUSE statement causes PAUSE,

Examples:

PAUSE n or PAUSE 'MESSAGE' to be typed on

DO 10011 = Nl*2,M,K*(N2*3-1)

the user Teletype. The user may type G to

DO 4012 = 1,10,4

return control to the next statement or X to ter-

DO 1013=7,J

minate the program. A carriage return must
follow the G or X.

CONTINUE STATEMENT
STOP STATEMENT

The general form of a CONTINUE statement is:
CONTINUE

The general form uf the STOP statement is:

Its execution has no effect on the program be-

STOP n

yond causing control to go to the next statement

where n is an octal constant and may be omitted.

in sequence. Nevertheless, this can be a use-

The STOP statement causes termination of a

ful statement. For example, suppose it is nec-

program.

4-5

CHAPTER 5
SUBPROGRAMS
There are two broad classes of subprograms, de-

programs, correspond most closely to what is nor-

fined by the way the subprogram is called and

mally thought of as a separate and closed sub-

whether or not it may return one or many values

routine. That is, an external function is a single

to the mai n program. Subrouti ne subprograms

and separate program. It exists only once in

must be explicitly called by a CALL statement.

memory and is separate from the main program.

They may return one, or severa I, va lues. Func-

Whenever it is called, control leaves the main

tion subprograms are implicitly called by using

program temporari Iy and goes to the external

their names in an arithmetic expression. They

function.

may return only a single explicit value. Implicit
The following rules must be observed in defining

values may be returned through arguments or

an external function:

COMMON storage.

1. The first statement must be a FUNCTION
statement. This is a statement of the form

DUMMY ARGUMENTS
All subprograms use dummy arguments in their
argument lists which accompany their definitions.

where name is the symbolic name of the
function. Name may not have a terminal
F unless it consists of three characters or
less. The a's are dummy arguments which
may be either variables or array names.

The calling argument list must match the dummy
argument list in the number, order, and type of
argument. Beyond that, there need be no cor-

Symbolic function names are constructed by
the same rules as for variables. The type
of the function name, integer or real, determi nes the type of the resu It.

respondence between the names of the arguments
in the two lists.

FUNCTIONS
2. The symbolic name of the function must
appear as a variable name within the program. Upon execution of the RETURN statement, the value of the variable is considered
the value of the function.

The two types of functions are intrinsic and external. Intrinsic functions are predefined and
are part of the FORTRAN language. External
functions are defined by subprograms external
to the program unit in which they are called.

3. The symbolic name of the function must
not appear in any non-executable statement
within this program unit except for the
FUNCTION statement.

External Functions
External functions, along with subroutine sub-

5-1

4. The symbol i c names of the dummy arguments must not appear in any COMMON
or EQUIVALENCE statements within this
program unit.

An external function to calculate the factorial
of an integer n:

5. The subprogram may not contain a SUBROUTINE statement, another FUNCTION
statement, or a reference to itself.

FUNCTION NFACT (N)
NFACT = 1
DO 10 J = 1, N
NFACT = NFACT * J
RETURN
END

6. The subprogram must be logically terminated by a RETURN or CALL statement.

Library Functions

7. The subprogram must be physically terminated by an END statement.

Certain external functions are used so commonly

External functions are called by using the sym-

that they are considered basic, and are supplied

bol ic name of the function, followed by a list

in the library. Table 1 shows these basic func-

of actua I arguments. The arguments must agree

tions.

in number, order, and type with the dummy ar-

added to the library. When library funct ions

guments. Further" if a dummy argument is an

are called, a terminal F is appended to the

array name, the corresponding actual argument

function name.

If desired, additional functions may be

must be an array name. The actual arguments
Example:

may be variable names, array elements, array
names, or expressions. The expressions may, in

ROOll = (-B+SQRTF(B**2-4.*A*C»/2.*A

turn, contain calls to other external functions.
TABLE 1
Basic
External
Functions

Definition

exponentiql
natural
logarithm

e
'

EXTERNAL FUNCTIONS

Number of
Arguments

log

Symbolic
Name

Type of
Argument
Function

EXP

real

LOG

real

sine

sine (arg. in radians)
sine (qrg. in degrees)

SIN
SIND

real
real

cosine

cos (arg. in radians)
cos (arg. in degrees)

COS
COSD

real
real

hyperbolic
tangent

tanh

TANH

real

square root

x 1/ 2

SQRT

real

arctangent

tan- 1

ATAN

real

,arctangent

tan- 1 (y/x)

ATAN2

real

5-2

Restr ict ions

TABLE I EXTERNAL FUNCTIONS (continued)
Basic
External
Functions
arcsin

Definition

Number of
Arguments

• -I

Sin

-I

Symbolic
Name

Type of
Argument
Function

ASIN

real

ACOS

real

Restrictions

arccos

cos

hyperbolic
sine

sinh

SINH

real

hyperbolic
cosine

cosh

COSH

real

common
logarithm

10910

LOGIO

real

truncation

sign of a
times largest into
~ 101

XINT
tNT

real
real

integer
real

101< 2 34
a normalized

remaindering (*note)

1 (mod O 2)

MOD

real

10/02 1<2

choose
largest
value

max
(0 1,0 2 , .a n)

n~2

XMAXO
XMAX I
MAX 0
MAXI

integer
real
integer
real

integer
integer
real
real

choose
smallest
value

min
(0 1 ,0 2 " .a n)

n~2

XMINO
XMINI
MINO
MINI

integer
real
integer
real

integer
integer
real
real

transfer
sign

sign of 02
times 10 II

XSIGN
SIGN

int/either
rea 1/e ith er

integer
real

positive
difference

max
(0 1-0 2 ,0)

XDIM
DIM

integer
real

*Note: The functions MODF(al, 02) and XMODF(al ,02) are defined as 01 -[01/02]02, where [x] is
the largest integer which does not exceed the absolute value of x, and whose sign is the same as x.

quence of program steps to evaluate the function

Intrinsic Functions

becomes part of the co II i n9 program.

Table 2 lists the intrinsic functions for PDP-6
FORTRAN. The intrinsic functions are sometimes

The function name must agree exactly with the

referred to as open subrouti nes. Th i s means each

name shown in Table 2, and it must be followed

time the function name is encountered, the se-

by a parenthesized Iist of arguments. Each ar-

5-3

gument may be any arithmetic expression which

Some examples of arithmetic expressions with

agrees in order, type, and number wi th the spec i-

intrinsic functions:

fications of Table 2.
A * Z + ABSF (C) * D
An initial X is always present in the names of

I * J + XABSF (K) * L

functions which produce results of type integer.

TABLE 2
Intrinsic
Function

Definition

INTRINSIC FUNCTIONS

No. of
Args.

Function
Name

Mode of
Args.

Mode of
Function

Restriction

XABSF
ABSF

integer
real

none
none

XMODF

integer

none

absolute
value

Ia I

remaindering (*note)

a 1 (mod a 2)

float

convert integer to real

FLOATF

integer

real

none

fix

convert real
to integer
the result is
the largest
integer ~ a

XFIXF

real

integer

none

*Note: The functions MODF(a1, a2) and XMODF(a1,a2) are defined as a1 - [a1/a2] a2, where [x] is the
largest integer which does not exceed the absolute value of x, and whose sign is the same as x.
arguments, are either variable or array
names.
2. The subroutine name must not appear in
any other statement other than the first.

SUBROUTINES

Defining Subroutines

3. No dummy argument name can appear

Subroutines are defined by writing a program

in a COMMON or EQUIVALENCE statement. However, they must appear ina
DIMENSION statement if they are array
names.

much like any other program except for the following restrictions:
1. The first statement must have the form

4. The subprogram may not contain a

SUBROUTINE name

FUNCTION statement or another SUBROUTINE statement..

or
SUBROUTINE name (a 1,a 2 ,·· .a n)

5. The subroutine must be logically terminated by a RETURN, CALL EXIT, or
CALL DUMP statement.

where the name is the symbolic name of
the subroutines and the als, the dummy

6. The subroutine must be physically terminated by an END statement.

5-4

The following example contains a subroutine, for

Calling Subroutines

multiplying two 3 x 3 matrices to form a third:

A subroutine is called by a statement of the form

SUBROUTINE MATMPY (A, B,C)
DIMENSION A(3, 3), B(3, 3), C(3, 3)
DO 100 1=1,3
DO 100 J=l,3

CALL name
or
100
where name is the symbolic name of the subroutine and the a's are the actual arguments. An

C(1,J)=A(I,1)*B(1,j)+A(I,2)*B(2,j)
X +A(I,3)*B(3,J)
RETURN
END

This subroutine could be called by the statement

argument may be a variable name, an array element, an array name, or any other expression.

CALL MATMPY (AMATl,AMAT2,PMAT)

However, the actual arguments must agree in orThe routine would then form the product of

der, number, and type wi th the dummy argument

AMATI times AMAT2 and return all the values

list in the subroutine definition.

in an array named PMAT.
None of the arguments shou Id have the same
An alternative method for calling and transmit-

name as the subroutine.

ting information is:
A subroutine may return one or more calculated
SUBROUTINE MATMPY
COMMON A(3, 3), B(3, 3), C(3, 3)
DO 100 1=1,3
DO 100 J=l, 3

values by redefining one or more of the arguments in the argument list. If this is done, the
redefined argument should not appear in this

100

Iist as an expression.

C(I, j)=A(I, 1)* B( I, J)+A(I, 2)* B(2, J)
X +A(I,3)*B(3,j)
RETURN

Example:
called by:
CALL FOFX (A(3), 5.43, SQRTF(Y), 5H ERROR)
RETURN

COMMON AMATl(3,3), AMAT2(3,3),
X PMAT(3,3)

The RE TURN statement causes control to be transferred to the next executable statement in the
calling program. RETURN statements may appear
CALL MATMPY

on Iyin subprograms.

5-5

CHAPTER 6
INPUT-OUTPUT

INPUT-OUTPUT LISTS

This list is equivalent to VAR(1), MAT(l), VAR
(2), .••• VAR(lO), MATOO).

Data transfer statements contain input-output
lists. These are ordered lists of variable names,

Implied-DO lists may be nested. In this case,

array element names, or array names (also ex-.

the range of the inner DO is exhausted for each

pressions in the case of output) each separated

increment of the outer DO.

by commas. The order of the list, from left to
right, specifies the order in which the data will

For example, suppose it is desired to read in a

appear.

matrix by row rather than column order. Then
a nested pair of implied DO's may be used as

Example:

follows:

Ir-A-,-B,-C-(-3)-,-F(-7)-,-N-U-M-,-C-0-N-S-T--'

«MAT (I,J), J=1, m), 1=1, n)

If data was being read in, the first number read
would be placed in a memory location referenced

This is equivalent to the list (MAT(l,l), MAT(l,

by A, the second by B, the third by C(3) I etc.

2), .... MAT(1 ,m), MAT(2, 1),,, •• MAT(m,n).
The appearance of an array name wi thout subscripts causes input or output of all the elements

INPUT-OUTPUT STATEMENTS

of the array, ordered by the array successor

Input-output statements are of two types: exe-

function.

cutable statements cause transmission of information from or to input-output devices. FORMAT

A DO-implied list is a list which is enclosed in

statements, referred to by executable input-

parentheses and ends in an indexing specifica-

output statements, specify the arrangement and

tion, i.e., a list of the form:

conversion of information denoted by the input(namel, name2, •••• namen, j=ml, m2, m3)

output lists.

where i, m1, m2, and m3 all have the same
meanings as in DO statements, and name 1,

Executable Statements

name2, and namen are subscripted variables.

Input statements have the form:

Example:
READ label, list

(VAR(K), MA T(K), K=1, 10)

REREAD label, list

6-1

where label is a FORMAT statement label, list

ACCEPT label, list
READ INPUT TAPE n, label, list
or RIT n, label, list

is an output list, and n is an integer expression
denoting a device number.

REREAD INPUT TAPE n, label, list
READ TAPE n, list

For the PRINT, PUNCH, TYPEOUT, WRITE OUTPUT TAPE, and WOT statements, information is

where label is a FORMAT statement label, list

written until the output list is satisfied. Each

is an input list, and n is an integer expression

item of data is converted as specified in the

denoting a device number.

FORMA T statement.
For the READ, TYPEIN, READ INPUT TAPE,
and RIT statements, information is read in begin-

The WRITE TAPE statement causes output of bi-

ning with the next item in position to be read,

nary information onto device n.

until the input list is satisfied. Each item of
Special Tape Statements -In each of the following

data is converted as specified in the FORMAT

statements n is an integer expression:

statement.

REWIND n
For the REREAD statements, the previous record
This command causes tape n to be positioned at

is rescanned with a new format.

its load point.
The READ TAPE statement causes input of binary
UNLOAD n

information from device n.

This command causes tape n to be rewound and

Examples:

unloaded.

READ 101, (A(I) ,I=M, N) ,J

BACKSPACE n

REREAD 102, (A(I) ,I=M, N) ,J
RIT N+l,5, TERM1, TERM2

This command causes tape n to backspace one

READ TAPE 4, (((BINDAT(I,J,K)
l=l,Il),J=l,Jl),K=l,Kl)

record.
SKIP RECORD n

Output statements have the form:
This command causes tape n to space at the be-

PRINT label, list

ginning of the next record.

PUNCH label, list
TYPE label, list

END FILE n

WRITE OUTPUT TAPE n, label, list
or WOT n, label, list

This command causes an end-of-fi Ie mark to be

WRITE TAPE n, list

wri tten on tape n.

6-2

FORMA T Statements

To allow ease of preparing input data several
variations are allowed on this basic form:

The general form of a FORMAT statement is:
123---m
123--p
label FORMAT(///---/f s f s ----f 5 / / / - - / )
1122
nn

zero, the f's are field descriptors or a group of

1. The decimal point may be omitted.
If so, and if the field descriptor is Fw.d,
Ew.d, or Gw.d, d digits are assumed to
follow the implicit point. If the decimal
point is present, it overrides the 'd' specification.

field descriptors enclosed in parentheses, and

2. The initial plus sign may be omitted.

the s's are ei ther commas or slashes.

3. The exponent may have the following
forms:

where m, n, or p, but not all three, may be

Basic Field Descriptors

a. A si gned integer.

The basic field descriptors have the following

b. E followed by an integer (if
positive).

forms: Fw.d, Ew.d, Gw.d, Iw, Ow, Lw/ Aw,

c. E followed by a signed integer.

wHh, h 2 ---h 3 , andwX. The lettersF, E, G,

4. The exponent may be omitted.

0, I, L, Al H, and X indicate the method of
conversion.

'w I is a non-zero integer constant

Thus, suppose the field descriptor was E10.4,

denoting the width of the field in the external

F10.4, or G 10.4. The following numbers would

character stri ng.

all be equal:

'd' is an integer constant speci-

fying the number of digits in the fractional part.
F, E, G, I, and 0 are all considered to be numerical fields. The remaining four are alpha-

bbbb427.67

bbb4276700

42.7670E+1
b+42.767+1

b+42.767El
+427670E01

numeric fields.

(decimal impl ied
by descriptor)
(decimal implied
by descriptor)

The fields designated by FIE/ and G descriptors

Note that E+1, E1, E01, E+01, +1 and +01 are

all contain real data which is to be converted

all equivalent.

internally to a real datum. For all three, the

F Conversion Output - For F conversion, the form

input fields are identical. However, the output

of the output field is:

fields may be quite different.
b b ---b ±X X ---X • Y ----V
1 2
n 12m
1
d

F,E, G Input Fields - The general form of the

input field is:

b b ---b ±X-----X • Y ----V E±Z Z
n

~jJ2

The field consists of as many blanks as necessary
to fill out its width, followed by a decimal number without exponent. Plus signs are omitted.

6-3

The following table illustrates the correspond-

Magnitude

External

ence between i nterna I and externa I numbers for

Internal

External

+4270.
-437.
+4.37
+.437
-.00437
+.0000437

b4370.0000
b-437.0000
bbbb4.3700
bbbbOA370
bbb-O.0044
bbbbb.OOOO

•

O.l~N<l

F (VV-4). d

I bbbb I

1 < N < 10

F(VV-4).(d-l) I bbbb I

10d - 2< N < 10d-l

F(VV-4). 1

I bbbb I

10d-l< N < 10d

F(VV-4). 0

I bbbb I

Otherwise

Ew.d

an F10A field descriptor:

E Conversion Output - The external field for

The following table illustrates G output conver-

conversion output is of the form:

sion for the field descriptor Gl1A.

x -----X E±Z Z
1

1 2

For this conversion, all numbers are normalized
so that the decimal point appears to the left of
the first significant digit. The number is rounded to d significant digits.

Internal Number

Conversion

External Field

43700.0
-4370.0
437.0
43.7
-4.37
.437
.0437

E11.4
F7.0bbbb
F7.1bbbb
F7.2bbbb
F7.3bbbb
F7Abbbb
Ell A

bO.4370E+05
b-4370.bbbb
bb437.0bbbb
bb43.70bbbb
b-4.370bbbb
bO.4370bbbb
bO A370E -01

The following table illustrates the correspondence
between internal and external numbers for an

I Conversion - Fields specified by the field de-

E12A field descriptor:

scriptor Iw contain decimal integers. The form
of the external field is the same for both input

Internal

External

+4370
-4.37
+.00437
+.0000437
-437.19
+437.23

bbO A370E +04
b-O A370E +0 1
bbO.4370E-02
bbO.4370E-04
b-O •4372E +03
bbO •4372E +03

and output:

b b ---b ±X X ---X
12m 1 2
n

On output the plus sign is omitted.

G Output Conversion - G conversion depends

Examples of external fields corresponding to the

on the magnitude of the internal number. For

descriptor 16:

the field descriptor Gw.d, the following table
shows the correspondence between the magnitude of the internal number and the conversion
that takes place.
6-4

bbbbb2

bbbb-2

-57296

647198

b+2763 (if output bb2763)

o Conversion - Fields specified by the field

the form:
nP

descriptor Ow contain octal integers. The form
where n, the sca Ie factor, is an integer constant

of the external field is:

or a minus sign followed by an integer constant.

b b ---b ±X X ---X
(

I 2

The scale factor precedes the three basic field
descriptors. When execution of a format statement is begun, a sca Ie fac tor of zero is assumed.

For input the sign is optional. All O-type out-

Once a scale factor is encountered in a FORMAT

put conversi ons are unsi gned.

statement, it holds for all remaining E, F, and G

Examples of conversion for 04 specification:

fields, in that FORMAT statement or until a new

External (input)

Internal

External (output)

bb27

000000000027

0027

bb-1

777777777777

7777

scale factor is encountered.
The effects of the scale factor are:
1. For F, E, and G input the scale factor
has no effect if there is an exponent in the
external field. Otherwise; for a scale of
n:
externa I number = i nterna I number x 10 n

L Conversion - L conversions are used to read
or write logical data. The form of the field
descriptor is Lw.

2. For F output wi th a scale factor of n
externa I number = i nterna I number x 10 n

For input, the external field contains a string of
3. For E output, the scale factor controls
the decimal normalization between the
number and the exponent so that:

up to w nonblank characters beginning with
either Tor F. For example,
Descriptor

Input Field

a. If n~O, there will be Inl leading
zeros followed by d significant digits
to the right of the point.

L1
L3

bbT or bTb or Fbb
bb TRUEbHONESTbb

L15

b. If n~O, there wi II be exactly n
significant digits to the left of the
decimal point and d-n+1 to the right
of the decimal point.

For output, a Tor F is written in the last character position of the field. The rest of the field

4. For G output, the scale factor is suspended un less E conversion is used.

is fi lied in with blanks.
Scale Factors - A scale factor may be used for

The following chart provides some examples of

E,F, and G conversions. A scale factor has

scale factor effects.

Internal
+4370
4370
4370

Format
2PF12.4
2PE 12.4
-2PE12.4

External-No Scale

External with Scale

4370.0
bb0.4370E+04
bbO .4370E+04

b438000.0000
bb43.700E+02
b.004370E+06

6-5

The Blank Field Descriptor

blanks in the format statement.

A blank field descriptor has the form:

The' ••• ' descriptor has the same effect as wH
except that the text is embedded between the

single quotes. When used as part of the text,
On input, w characters of the external record

the quote character appears twice in succession:

are skipped. On output, w blanks are written

DON'T is represented as 'DON liT' •

in the external record.
The Aw field descriptor causes ASCII characters
to be read into, or written from, a specified list

The ASCII Field Descriptors

element. Since up to five ASCII characters can

There are three forms of ASCII field descripl'ors,

be stored in one memory word, the following

wH,. •••• ',and Aw. The wH descriptor has the

rules apply to the Aw specification.

following effects:

1. If on input w>5, only the rightmost five
characters, including blanks, are read from
the external field.

1. On input, the next w characters are
read as ASCII text into the w character
positions following the H in the FORMAT
statement.

2. If on input w~5, all w characters wi II
be read and stored in memory left justified
with 5-w trailing blanks to fill out the
memory words.

2. On output, the w characters following
the H in the FORMAT statement are written into the record.

3. If on output w>5, the five characters
from the internal representation wi II be
written in the last five characters positions
of the external field. The leading character positions will be filled in by w-5
blanks.

For example, if we should wish to insert text in
a record, the following field descriptor might
appear in a FORMAT statement.
32HbTHISbiSbAbSAMPLEbOFbOUTPUTbTEXT

4. If on output w~5, the leftmost w characters of the internal representation wi II
be written in the external field.

Note that w, the character count, must include

::dl characters, including blanks. The following
~xample illustrates the means for altering text-

Variable Field Input

)al information in a FORMAT statement.
The PDP-6 FORTRAN allows certain relaxation
READ

100

of the input formats for use in preparing input
data. Namely, the E, F, G, I, and 0 field

100

descriptors can appear without wand d. If so,

FORMAT (5Hbbbbb)

nl!Jmbersmust be separated by an explicit delimiter.
he field read by the READ statement would

Such a delimiter may be any character which is

:tve to include exactly five characters of text

illegal as the next character in the number rep-

icluding blanks. This text would replace the

resentation.

6-6

Examples:

Carriage Control

FORMAT

External Field

The first character of each printed line is inter-

(I)
(I,E,F ,O)

-5+4/3,6-7/
9/.5E12/6.501/776/

preted as a control character for the line printer.
The following table contains the special control
characters which may be used as the first char-

Variable Field Output

acter in an Aw or wH field. Up to 120 charac-

If the E, F, G, I, and 0 descri ptors appear

ters may be printed on a line.

without wand d in a FORMAT specification for

Character

Effect

space

skip to next line
skip a line
form feed - go to top of next page
suppress skipping - will repeat line
skip 2 lines
skip to next 1/2 page
skip to next 1/3 of page
skip to next 1/6 of page
skip to next 1/10 of page
skip to next 1/20 of page
skip to next 1/30 of page

output, the field width w is set to 15. For the
E, F, and G descriptors, d is set to 7. Thus

the output FORMAT specification

+
110

FORMAT (E,F ,G,I,O)

2
3

would automatically become
10

FORMA T (E 15 .7, F 15 .7,

X G 15.7, 115, 015)

Repeat Count

Slash

The repeat count is an integer constant that
Besides being a field separator, the character /

specifies how many times a given field descrip-

closes a record and starts a new one.

tor, or group of field descriptors enclosed in
parentheses, is to be repeated. For example,

On output, a series of n+ 1 slashes wi \I produce

consider the following equivalent FORMAT

n blank records.

specifications:
Example: Output according to the FORMAT
(l2,12,E10.4,E10.4,E10.4):= (212,3E10.4)

statement.

(12,12, lHQ,F5.2, 12,1 HQ,F5.2,12, 1HQ,F5.2) ==
(12,3(12, lHQ, F5. 2)}

FORMAT (9HbMATRIXbA//5H+SINE)
wi \I produce

When a repeat count and scale factor are both

MATRIXbA
(blank line)

used, the general form is

SINE

n P rf

where f is the basic field descriptor. All basic

Terminating FORMAT Statements

field descriptors, except for wH and wX, may

When the entire format has been used and the

have a prefixed repeat count.

final right parenthesis is reached, the current

6-7

1. If the FORMAT specification contains
one or more repeat groups, beg i n with
that group with the rightmost closing
parenth eses.

record is closed and the input or output list is
examined for further entries. If no further entries are found, the data transfer is complete.
If items remain, a new record is started and the

2. If the FORMAT specification has no
repeat groups, return to the beginning
of the specification.

FORMAT statement is repeated according to the
following conditions:

6-8

APPENDIX 1
DIAGNOSTICS

The compilation process proceeds in two parts.

language into an object language, the binary

First, the compiler translates the source lan-

machine language. Therefore, there are two sets

guage into an intermediate assembler language.

of diagnostics: those printed out by the compi ler

Second, an assembler translates the intermediate

and those printed out by the assembler.

COMPILER DIAGNOSTICS

Error Message

Meaning

Steps for Correction

Parse Table Overflow

Statement too long

Break up the statement into one
or more smaller statements.

Pushdown Depth
Excessive

Same as above

Rule Storage Overflow

The program can not
be compiled since it
generated too many
rules.

Try reduc ing the number of
variables in DIMENSION statements and/or don't nest the DOs
so deeply.

If the compi ler cannot completely parse a state-

ment, it wi II pri nt three up-arrows f t f under the

ment, it will parse as much of the statement as

offending statement.

it can and generate the corresponding code. It
will indicate its failure to parse the statement

ASSEMBLY DIAGNOSTICS

completely by printing an up-arrow f under the

Assembly error flags consist of single characters

last character correctly parsed.

printed in the left margin of the assembly listing

If the compiler cannot parse any of a given state-

of the compi ler output.

Explanation

Error Flag
C

Common Error - A variable has appeared more than
once in a set of COMMON statements or is also a
subroutine argument.

Dimension Error - A variable has appeared more than
once in a set of DIMENSION statements.

A 1-1

Error Flag

Explanation
Equivalence Error - A subroutine argument has appeared
in an EQUIVALENCE statement.
Equivalence Inconsistency - More than one variable in an
equivalence group has appeared in another equivalence
group.

Storage Assignment Error in EQU IVALENCE Statement A variable in an equivalence group which has appeared
in a previous equivalence group does not account for all
the storage in the current group.

Multiple Symbols - A statement label has been used more
than once.

Undefined Symbol - A statement label which has been referenced is missing, or an op-code which is incorrect.

Table Overflow - Symbol table; COMMON, DIMENSION,
EQUIVALENCE table; or literals table has overflowed.

External Symbol Definition - A variable or array name is the
same as a library function name.

Relocation Error - Illegal arithmetic involving relocatable
symbols. *

Assignment Error - illegal use of =*.

Null Symbol - Use of illegal symbol structure such as
label in assembly which does not begin with alphabetic
character or % or. *

A constant is too large to fit in a PDP-6 machine word.

A statement is too long to be assembled correctly.

The statement labe I for the terminating statement of the current
DO loop has already been processed.

*Errors likely to occur in use of ASSEMBLE - COMPILE feature.

Al-2

APPENDIX 2
SPECIAL PDP-6 FORTRAN

STATEMENTS

DECTAPE INPUT-OUTPUT
The statements INFILE N, FILE and OUTFILE N,
FILE provide a means for referencing DECtape in

WRITE FILE1
ANAME==5HFI LE 1
OUTFI LE 3, ANAME
WRITE TAPE 3, (A(I), 1=1,100)
END FILE 3

WRITE FI LE2
BNAME=5HFI LE2
OUTFI LE 3, BNAME
WRITE TAPE 3, (B(I), 1=100)
END FILE 3

READ FILE1
INFILE 3, ANAME
READ TAPE 3, (A(I),1=1,100)

FORTRAN input-output statements. On output,
files are created by issuing the OUTF\LE statement. For example, the statement
OUTFILE 10, FILEl
causes the file name (in 7-bit ASCII) in FILEl
to be entered in the file directory on the DECtape referenced by device assignment 10. Subsequent to the OUTFILE statement, all output
statements wh ich reference devi ce 10 wi II cause
data to be written in file FILE1. The statement
END FILE 10
is necessary for closing the output, i.e. f emp-

TITLE

tying buffers and completing the entry in the
directory for fi Ie FILE 1 •

A program name may consist of up to six characters
and is declared in a TITLE statement which has the

For input, the INFILE statement causes subse-

form

quent input statements to reference data in a

TITLE NAME

particular file. The special tape statements,
except for END FILE, are ignored in DECtape

The use of the TITLE statement is optional. If not

operations.

used, the title .MAIN. is generated for main pro-

If INFILE or OUTFILE statements are not used,

grams and subprogram names for subprograms. The

the file name FORTR. DA T wi 1\ be assumed.

title is essential for initiating use of a program's

Thus one fi Ie can be created automati ca lIy on

symbols in DDT. When used, the TITLE statement

any DECtape.

must be the first statement in the program.

A2-1

ASSEMBLE-COMPILE Examples

ASSEMBLE -COMPILE
1.

The PDP-6 FORTRAN II compiler will accept

MACRO-6 code directly when inserted between
the statements ASSEMBLE and COMPI LE.

IF N IS NEGATIVE, GET RANDOM
NUMBER IN M
IF (N) 10, 20, 20
ASSEMBLE

%10 MOVE 3, [t 0142536475076]
ADD 3, RAN
ROT 3, -1
EQVB 3, RAN
MOVEM 3, M#
JRST %11

The followi ng components of MACR 0-6 code
are permissible:
1. Ail basic PDP-6 operation code mnemonics.

RAN

t 0123456707654

2. The pseudo-operations ASCII, ASCIZ,
EXP, XWD, and BLOCK.

COMPILE
DO 12 1=1, 100

3. Literals containing
a) digits 0-9
2.

b) the characters ., +1 - I E
c) the symbol
radix to 8

t 0 for changing the

d) the pseudo-operations ASCII and
SIXBIT.
4. The use of • to represent the value of the
current location counter.
5. Storage ma; be allocated by the use of
the character •
NOTE: The radix for all instructions is
'10 unless changed temporarily byt O.
The following restrictions apply to the above:
1. Terms in expressions may be combined
only with +, -, and *.
2. Only non-relocatable terms may be
combined by * •

3. Relocatable symbols may not be used
in the left half of the XWD pseudo-operation.
4. Statement labels must begin with an alphabeti c character or "%", "$" or " ." •
If a labe I starts wi th " .", the second character must be alphabetic. FORTRAN statement labels of the form XXX may be referred to by %XXX.

A2-2

REPLACE THE STATEMENTS:
DO 10 l=l,M
D010J=I,N
10 A(I,J) = B(l,J)*A(J,J)
AND EXECUTE IN ACCUMULATORS
DIMENSION A(10,20), B(10,20L
C(1O,20)
ASSEMBLE
MOVEM 2, TWOSAV#; SAVE AC 2
HRLI 2, BEGN; Set up block transfer
; in AC 2
HRRI 2, 6
BLT2,17
; Move code to AC
; 6-AC 17
JRST 6
; The following block is
; executed in accumu; lators 6-17
BEGN MOVE 2,M ; OUTER LOOP
MLUP MOVE 3,N ; INNER LOOP
MOVE 4,J
IMULI 4,10
ADD 4,1
MOVE 5, B-l1 (4) ;B(I, J)
FMPRM 5, A-I! (4) ;AO,J)=a:I,J)*A(I,J)
SOJG 4, NLUP
SOJG 3, MLUP
JRST %20
; RETURN FROM AC 17
%20
MOVE 2, TWOSAV
COMPILE
C
C
C
C
C

ASSEMBLE- CaMPI LE Format

The ASSEMBLE and COMPILE statements must

Cards

appear wi thout labe Is.

On cards, statement labels for assembly code
Punched Tape

are limited to four characters and must be punched
in columns 1-5. The instruction and comments
follow in columns 7-72 with the format
OPCODE AC, ADDR (lR)

1. Card simulated format
Tabs may be used to skip to column 7
and also to delimit fields within an
instruction.

iCOMMENT

The fields for the operation, accumulator, ad2. Column free format

dress, index register, and comments are not
fixed and may fall anywhere in columns 7-72.

Labels must be followed by a colon.

A2-3

APPENDIX 4
SUMMARY OF STATEMENTS

Output

CONTROL STATEMENTS

PRINT label, list
PUNCH label, list
TYPE label, list
WRITE OUTPUT TAPE n, label, list
WOTn, label, list
WRITE TAPE n, list

ASSIGN n TO iname
CONTINUE
DO label i = k, I, m
GO TO n
GO TO iname
GO TO iname, (labell, labeI2, ••• labeln)
GO TO (label 1, labeI2, ••• labeln), iname
IF (exp) labell, label2, label3
IF (exp) labell, label2
IF ACCUMULATOR OVERF LOW labell
label2
IF QUOTIENT OVERFLOW labell, label2
IF (SENSE LIGHT i) labell, label2
IF (SENSE SWITCH i) labell, label2
PAUSE i
PAUSE "MESSAGE"
SENSE LIGHT i
STOP

Tape Commands
BACKSPACE n
END FILE n
REWIND n
SKIP RECORD n
UNLOAD n
SPECIFICATION STATEMENTS
COMMON v l' v 2' ••• v m
DIMENSION a l , a 2 , ••• a n

INPUT -OUTPUT STATEMENTS

EQUIVALENCE (listl), (list2), ••• (\istn)

Input

SUBPROGRAM STATEMENTS

ACCEPT label, list
READ label, list
REREAD label, list
READ INPUT TAPE n, label, list
REREAD INPUT TAPE n, label, list
READ TAPE n, list
RIT n, label, list

CALL name (a l , a 2 , .•• a n)
FUNCTION name (d"

d2 , ••• d n)

RETURN
SUBROUTINE name (d l , d2 , ••• dn)

A4-1

APPENDIX 5
FORTRAN

OPERATING

1. Device Assignments

CALL PDUMP (Al, B1, F1, A2, B2,
F2,"') and CALL DUMP (A1, B1, F1,
A2, B2, F2,"') are statements which
cause portions of core to be dumped on
the device corresponding to the PRINT
statement. If no arguments are present,
the entire user core area is dumped in
octal. The argument list (A1, B1,
F1, ... ) displays the arguments for the
dump and the mode in which the dump is
to take place. Core is dumped between
the I imits Ai and Bi in the mode Fi.
Either Ai or Bi may be upper or lower
limits.

Logical device assignments for run-time
I/O are made with the use of a table called
DEVTB. in the FORTRAN library. Each entry in the table consists of a 6-bit ASCII
device name, and the numerical position
of each en try corresponds to the logi ca I
number used in FORTRAN I/O statements.
The first location of DEVTB. contains the
number of entries in the table. The last
five entries in the table are special and
correspond to the FORTRAN statements
READ, ACCEPT, PRINT, PUNCH, and
TYPE. Any entry in DEVTB. may be
changed by reassembling the table.

The modes are

Example:
DEVTB.:

O. Octal
1. Floating Point
2. Integer
3. ASCII

t D13
SIXBIT
SIXBIT
SIXBIT
SIXBIT
SIXBIT
SIXBIT
SIXBIT
SIXBIT
SIXBIT
SIXBIT
SIXBIT
SIXBIT
SIXBIT
END

.DTAO.
• TTY.
.CDR.
.LPT.
.MTAO.
.MTA1.
.DTA1.
.DTA2.
.CDR.
• TTY.
.LPT.
.PTP.
• TTY.

SYSTEM

If the final mode is missing, the core
area between A and B is dumped in
octal.
n
n
If the last two arguments Bn and Fn are
missing, an octal dump is made from An
to the end of the user's area.

;READ
;ACCEPT
;PRINT
iPUNCH
iTYPE

DUMP calls EXIT while PDUMP returns
control to the calling program when the
dump has been compl eted.
c. CHANG

With this tablet the statement WRITE OUTPUT TAPE 6 would refer to magnetic tape
unit 1. The PRINT statement would refer
directly to the line printer.

The statement A = CHANGF (X) causes
X to be changed from a sign-magnitude
negative number to a 2's complement
negative number (or vice versa). If X is
positive, CHANG has no effect on X.

2. Special Library Programs
a. EXIT

3. Error Messages

A call EXIT statement causes a run to be
terminated. All I/o devices are released from the job.

All errors which are detected by the Operating System result in terminating the run
with a CALL EXIT. The errors detected are:

b. PDUMP, DUMP

AS-l

a. Illegal character in FORMAT
statement

DECtape directory full

End of file on input

i. Tape record too short for list
spec ifi cati on

Illegal character in input string

Device not available

"'egal FORTRAN device number

k. End of file while reading binary
file
I•

f. Too many devices referenced
(15 a \lowed)

Dev i ce error or tape pari ty error

End of tape

For more detailed information about FORTRAN
1/0, see DEC-06-0-0S-FII-GM-FP-ACTOOFORTRAN \I Format and I/O Processor.

g. File name not found in a DECtape
directory

A5-2

APPENDIX 6
PDP-6 FORTRAN

II COMPILER OPERATING INSTRUCTIONS

The PDP-6 FORTRAN II Compiler contains two

The fi Ie name extensions RE Land LST are assumed

basic sections: a compiler which generates as-

for FILEl and FILE2 unless specified otherwise.

sembly code from the FORTRAN source statements and an assembler which generates relo-

Example:

catable binary programs. The 22K compiler

PTP: ,DTA3:LIST....-DTA1:S0RC.TXT

contains both parts in one program. The 9K
If FILEl is not desired, the command string should

compiler, however, prepares an intermediate

be of the form:

file for input to the assembler which is a separate program (F

o LA) •

,FILE2.-- FILE3
If FILE2 is not desired, either of the following

COMMAND STRING FOR 22K COMPILER

command strings is valid:

The command string is used to specify the input

FILE 1.-- FILE3
FILE1, + - FILE3

and output file designations for the compiler.
The 22K compiler expects up to two output fi les

If neither output file is desired, the valid com-

and one input file. The genera I form of the

mand strings are:

command string is
.-- FILE3
, ...--FILE3

FILEl, FILE2+- FILE3
FILEl will contain the relocatable binary output;

SWITCHES FOR THE 22K COMPILER

FILE2 will contain the listing of the compiled
output (source, assembly, bi nary, errors), and

Switches are letters specifying optional and extra

FILE3 is the source or input file. Each file may

functions to be performed by the com pi ler. These

have one of the following forms:

letters may appear within parentheses or after a
forward slash. Only a single letter may follow a

DEVICE:
DEVICE: FILENAME
DEVICE: FILENAME.EXTENSION

slash, whi Ie more than one letter may appear
within parentheses. The switches are as follows:

where DEVICE may be any device mnemonic
acceptable to the PDP-6 executive system,
FILENAME may be up to six letters and/or digits,
and EXTENSION may be up to three letters and/
or digits.

A6-1

Skip one file on the device (magnetic
tape only).

Do not print storage map.

Do not l.ist errors on Teletype console
if there is a listing file.

The source contai ns ru les for the compiler {for building a new compiler}.

List only source and errors (no assembly code).

The source is in column-free punched
tape format.

Rewind the device (magnetic tape
only) •

Clear the directory on the device before inserting the new file.

Switches for the 9K Compi ler: The swi tches K,
R, TI W, and Z (as described for the 22K compiler) are recognized by the 9K Compiler.

Example:
[

DTA1:FOLAIN, LPT:...- PTR: IT

This command string calls for identical output
files (input to FOLA) to be written on DECtape 1
and the line printer. The input (punched in col-

Any switches not recognized are ignored.

umn-free format) is coming from the paper tape
The switches M, N, R, 5, T may appear any-

reader.

where in the command string. K, Wand Z must
appear before the termination character of the

Command String for FOLA: The form of the FOLA

applicable file.

command string is identical to that of the 22K

For example, the command string

compiler with the exception that more than one
input file may be specified:

(W)MTAO:, (Z)DTA2: LIST + - CDR: 1M

FILE 1, FILE2 +- FILE3, FILE4, •••

calls for binary output on magnetic tape 0 (rewind first), listing on DECtape 2 (clear direc-

Switches for FOLA: The switches K, M, N, S,

tory) and source from the card reader. The

and Ware recognized by FOLA. M, N, 5, and

storage map is to be deleted from the listing.

T must appear before the termination character
for the applicable file and must appear for each

Command String for the 9K Compiler: The com-

file for wh i ch they are intended.

mand string for the 9K Compiler is similar to
that for the 22K Compiler except that there is

Example:

no binary file. However, two identical output

files are permitted (for example a listing of in-

PTP:,(K)MTA1: +- CDR:, (M)TTY:

put to the assembler F OLA). The genera I forms
In this example, the binary is to be punched on

of the command string are

paper tape, the listing is to go on magnetic tape
FILE1, FILE2+-FILE3
FILE 1 +-FILE3

1 (after skipping one file), and the input is to
come from the card reader and the teletype.

The fi Ie name extension for the output fi les are

The storage map is not to appear on the listing

assumed to be FOL unless otherwise specified.

for the input file from the teletype.

A6-2

APPENDIX 7
LIMITATIONS ON

9K FORTRAN

1. Boolean statements are not allowed.
2. Use of * in IF statements is not allowed.

A7-1

II COMPILER

~DmDDmD

EQUIPMENT
CORPORATION
MAYNARD,MASSACHUSETTS

•

8101

Cambridge, Mass.• Washington, D. C.• Parsippany, N.J .• Rochester, N.Y .• Los Angeles
Palo Alto • Chicago • Ann Arbor • Pittsburgh
Denver' Huntsville' Orlando • Carleton Place
and Toronto, Ont. • Reading, England • Paris,
France • Munich, Germany • Sydney, Australia

PRINTED IN U.S.A.

20-9/65