Digital PDFs

XX-3AA57-86

1966

81 pages

Original

9.4MB

Document:	Hospital Computer Project
Order Number:	XX-3AA57-86
Revision:	0
Pages:	81
Original Filename:	bbn-1966-pdp1d-midas-manual.pdf

OCR Text

r
MEMORANDUM NUMBER S I X - E

PROGRAMMING

Submitted by

Bolt Bercmek and Newman Inc

SOFTWARE

STATUS REPORT

Cambridge, Massachusetts

1 May, 1966

HOSPITAL COMPUTER PROJECT*

M E M O R A N D U M

SIX

Status Report on Programming Software

Submitted by
BOLT BERANEK AND NEWMAN INC
Medical Information Technology Department

Jordan J. Barueh, Principal Investigator, ±962-1966
Paul A. Castleman^ Principal Investigator^ 1966Richard H. Bolt, Supervisor^ ±966Nancy Adams
John Barnaby
Sheldon Boilen
Richard A. Bolt
Michael Cappelletti
Jonathan Cole
Bernard Cosell
Edward Gilbert
Edith Grossman
Nancy Haggerty
Alice Hartley
Gray Hodges

Toby Jensen
Charles Jones
Larry Kershaw
Philip LaPollette
Nancy Lambert
Maureen Letu
Sanford Libman
Barbara Lieb
William Lucci
Richard Mahoney
Ernest McKinley
Charles Morgan
Andrew Munster

Nicholas Pappas
Robert Payson
Lawrence Polimeno
Sylvia Sarafian
William Slack
Sandra Sommers
Lee Stein
Peter Storkerson
Sally Teitelbaum
Frederick Webb
Walter Weiner
Steven Weiss

* This joint research project Is supported by Grant PR 00263 and
Contract PH 43-62-850 from the National Institutes of Health and
the American Hospital Association. Efforts contributed by the
Laboratory of Computer Science3 Massachusetts General Hospital^
are directed by Principal Investigator G. Octo Barnett, M.D.
The report, Memorandum Six E, gives a technical description of
programming software developed by BBN for the computer system
•being used on this project.

Report No. 1422

Bolt Beranek and Newman Inc

TABLE OP CONTENTS
INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . .
*~
._

I. The Midas Assembly System . . . . . . . . . . . . . : ±'
A. Introduction to Assembly . . . . . . . . . . . . . 2
1. The Midas Source Language . . . . . . . . . . 5
a. The Character Set . . . . . . . . . . . . 6
b. Legal Strings . . . . . . . . . . . . . . 7
1. Basic Strings . . . . . . . . . . . . 7
2. Complex Strings . . . . . . . . . . . 9
(i) Language Units . . . . . . . . . 9
(ii) Combining Operators . . . . . . . 9
2.

«.
^

The Assembly Program . . . . . . . . . . . . . 11
a. The Current-Location Counter . . . . . . . 11
b. The Symbol Table . . . . . . . . . . „ . „ 14

c. The Radix Indicator . . . . . .
14
3. Defining Symbols . . . . . . . . . . . . . . . 15
a. Address Tags . . . . . . . . . . . . . . . 15
b. Variable Names . . . . . . . . . . . . . . 16
c. Assigned Parameters . . . . . . . . . . . 17
4.

The Use of Expressions . . . . . . . . . . . . 17
a. Storage Words .
17
b.

.—
_

Constants

. 18

c. Location Assignments . . . . . . . . . . . 19
5. Source Program Format . . . . . . . . . . . . 19
B. Pseudo-Instructions . . . . . . . . . . . . . . . 22
1. OCTAL and DECIMAL . . . . . . . . . . . . . . 22
2. CHARACTER, PLEXO, and TEXT . . . . . . . . . . 22
3o CONSTANTS . . . . . . . . .
. . . 24
4. VARIABLES . . . . . . . . . . . . . . . . . . 26
5.

DIMENSION

. . . . . . . . . . . . . . . . . . 27

Report No. 1422

Bolt Beranek and Newman Inc

TABLE OP CONTENTS cont
6.

EQUALS and OPSYN

. . . . „ . „ . . . . . 27

NULL . . . . . . . . . . . . . . . . . . . . . 28

8.
9.

OFFSET
. . . . . . . . . . . . . . . 29
REPEAT . . . . . . . . . . . . . . . . . . . . 31

10. START . . . . . . . . . . . . . . . . . . . . 32
11. EXPUNGE . . . . . . . . . . . . . . . . . . . 33
12. WORD . . . . . . . . . . . . . . . . . . . . . 33
13. 0IF and IIP . . . . . . . . . . . . . . . . . 34
14. PRINT and PRINTX . . . . B . . . . . . . . . . 36
15- STOP . . . . . . . . . . . . . . . . . . . . . 37
C. Macro-Instructions
. . . . . . . . 39
1.

Macro-Definitions
a.
b.

Basic Format . . . . . . . . . . . . . . . 40
Dummy Arguments . . . . . . . . . . . . . 4l

2.
3.

Macro Calls . . . . . . . . . . . . . . . . . 43
Storage of Macro-Instructions . . „ . . . . . 44

4.
5.

Nested Macros .
. „ . 45
The Pseudo-Instructions IRP and IRPC . 0 . . . 49

Operation of the Midas Assembly System . . . . . . 53
1. Preparation of a Source-Language Program . . „ 53
2.

3.
4.
E.

. . . . . . . . . . . . . . . 39

Performing an Assembly . . . . . . . . . . . . 53
a. Initial Procedure . . . . . 0
53
b. The Control Language . . . . . . . . . . . 5^
Order of Operations . . o . . . . . . . . . . 5&
Binary Output Format . . . . . . . . . . . . . 58

Error Checking; . . . „ „ . „ „ . „ „ „ „ . „ . . .60

Report No. 1422

Bolt Beranek and Newman Inc

TABLE OF CONTENTS cont.
II.

Editor
65
A. Control Characters . . .
66
B. Paper Tape Commands . . . . . . . . . . . . . . . 6"7
C. Filing Operations . . . . . . . . . . . . . . . . &8
D. Special Characters . . . . . . . . . . . . . . . 6'9

III.

DDT
71
A. Register Examination
9
73
B. Special Registers in DDT . . . . . . . . . . . . 76
C. DDT-Output Control
.....
78
D. Special Input Symbols . . .
80
E. Symbol Definition .
Q±
F. Searching Operations
82
G. Filing Operations .
. 83
H. Loading the Program
85
I. Running the Program
87
J. Punching Operations . .
89
K. Error Comments
90
L. Miscellaneous Functions e
91

IV.

Handle t
92
A. Programming System Files
. . . . . . . 93
B. Handle Commands
9
97
1. The Control Language for Files . . . . . . . 98
2. The Control Language for Tape Manipulation, .101

Appendix A
Appendix B
Appendix C

Midas Character Set
Symbols In Permanent Midas Vocabulary
Teletype Code Conversion

*
0

.A-l
.B-l
C-l

Report No. ±422

Bolt Beranek and Newman Inc

HOSPITAL COMPUTER PROJECT
STATUS REPORT MEMORANDUM SIX E
. : PROGRAMMING SOFTWARE

INTRODUCTION

This is the fifth volume of Memorandum Six,, a series of reports
describing the Hospital Computer System developed by Bolt
Beranek and Newman in collaboration with Massachusetts General
Hospital3 under sponsorship of the National Institutes of
Health and the American Hospital Association. Other volumes
in the series describe the system hardware^ the time-sharing
Executive and Common Routines,, the hospital user programs,
and the information storage and retrieval programs.
This volume describes the on-line Programming System that is
used to compose., edit., assemble., and debug software for the
PDP-lb-45^ the central processor designed for use in the
collaborative project.
The current Programming System,, with its full range of programming aids., is the realization of goals first approached
at Bolt Beranek and Newman by the Simbug System. Simbug5 one
of the first operating time-shared systems in the generalpurpose category,, enabled several users to perform debugging
operations simultaneously. The on-line Programming System
described here accommodates a full staff of programmers and
offers each of them,, in effect, sole access to a generalpurpose computer.

Bolt Beranek and Newman Inc.

Report No. ±422

The system consists of four programs: Midas., a macroassembler; Editor, a symbolic text editor; the DDT debugging
program; and Handles a program that provides for the manipulation of files generated by the three other programs in the
system.
Chapter I contains a detailed description of the Midas
assembly language, Midas offers the programmer a wide range
of expression, permitting him to deal with individual computer
words or to manipulate large blocks of computation.
Chapter II describes Editor, the on-line text editor used for
preparing symbolic programs to be assembled by Midas.
Chapter III provides a description of DDT, the system's debugging and control program. DDT serves two purposes within
the Programming System. First, it is used to check out and
correct new programs that, for this purpose, run under its
control. Second, DDT controls the running of Midas, Editor,
and Handle, providing inter-program communication within the
Programming System.
,
The final chapter, Chapter IV, describes Handle. The discussion of Handle includes, in addition to its actual functions
and commands, a description of the file organization for the
entire Programming System.

Report No9 1422

Bolt Beranek and Newman Inc.

THE MIDAS ASSEMBLY SYSTEM

The Midas Assembly System described in this volume was adapted
by Bolt Beranek and Newman from the Midas Assembler originally
written for the PDP-1 at Massachusetts Institute of Technology.
Some features of the Massachusetts Institute of Technology's
system have been eliminated. Other features have been added
to take advantage of the file structure of the Bolt Beranek
and Newman system and to provide on-line assembly via Teletype
terminal.
Like the Massachusetts Institutes of Technology's version of
Midas,, the Bolt Beranek and Newman system incorporates the
basic features of the Macro Assembly Program. Macro was also
developed at the Massachusetts Institute of Technology^ first
for the TX-j# computer, and then for the PDP-1. The major departure that Midas makes from Macro lies in the handling of
macro-instructions. A macro-definition in a Macro source
program is partially assembled when first encountered] that
is, everything except dummy symbols is translated into machine
language and stored in the macro table. In Midas the actual
source language text of the definition is stored in the macro
table3 and a complete assembly is performed each time the macro
is used. This method significantly extends the system's macroinstruction capabilities^ permitting recursive and conditional
definitions. The advantage gained by the Midas method for
:
storing macro-definitions will become clear when macroinstructions are discussed in detail later in this section.
The following pages assume that the reader has a working
knowledge of basic PDP-1 mnemonic codes and their octal

—1—

Report No. 1422

Bolt Beranek and Newman Inc

representations. The reader who does not is referred to
Volume Six A of this Memorandum., Chapter II, Section B^
page l6.

INTRODUCTION TO ASSEMBLY

A problem that can be expressed quite briefly in words or
mathematical notation will often^ when put in the form of a
computer program written in machine language, require
hundreds of computer instructions. Since the value and location of each of these must be fully specified^ the clerical
effort involved in writing a long program may exceed that
expended in analyzing the problem in terms of individual
computer operations. Further inconveniences that arise from
handling large amounts of information in the form of numerical
code are the introduction of clerical errors into a correctly
formulated program and the need for detailed documentation so
that others may understand the coding.

Report No. ±422

Bolt Beranek and Newman Inc.

Such difficulties led to the development of translating programs that make a computer available as an aid in the preparation and documentation of the programs that will be run on
it. These translators permit the programmer to express a
computation in terms of a convenient mnemonic language (the
source language),, which the translating program is able to
translate into machine code (the object language).
Two general categories of translating programs, compilers and
assemblers, have been developed to accept symbolic input and
produce binary output. Compilers are translators that are
"problem-oriented"; that is,.the various compiling programs
are designed to interpret a symbolic language similar to the
language in which the problem would originally be stated.
Thus, a compiler designed for mathematical applications would
accept mathematical expressions, and in addition to translating the numbers and symbols into appropriate quantities,
would translate the entire expression into an appropriate
sequence of computer instructions. A PDP-1 compiler3 for
example, might translate the expression Z=X+Y into the series
—LAC X, ADD Yj DAC Z. In the same way, compilers designed
for business applications are designed to accept a language
in which it is convenient to state business problems.
Because they accommodate various users in terms with which
they are familiar, compilers permit a person with a limited
knowledge of computers to write his own programs. Assemblers,
on the other hand, are "computer-oriented," consisting principally of instructions which correspond to internal computer
instructions. The simplest assemblers also include a minimal
number of rudimentary control operations that direct the
translation process.
-3-

Report No« ±422

Bolt Beranek and Newman Inc.

The Midas Assembly Program , while offering all the basic assembly features j belongs to a class of extended assembly programs referred to as macro-assemblers. Macro-assemblers such
as Midas provide an extensive set of control operations (Midas
pseudo-instructions) which ^ in principle,, make it possible for
the assembly program to perform computations analogous to
those of any object program it can produce.
Most notable in this respect are the Midas macro-instruction
features^ which permit the programmer to define a special purpose abbreviative language to suit his own needs. Using
pseudo-instructions provided for that purpose , a user can
name a complex coding sequence and provide for varying parameters .
The name is then, in effect, an abbreviation of the sequence that
will be substituted for it at assembly time^ with desired parameters inserted in their proper contexts . A programmer might,,
for example.? take a sequence such as LAC

ADD Y, DAC
assign it the name ADDXY^ specify that Xj Y, and Z are dummy
symbols for which program symbols will be substituted as desired and then use merely the macro name ADDXY followed by
appropriate symbols throughout the rest of the source program.
The assembler will generate the predefined coding just as the
compiler generates coding appropriate to the expression it is
interpreting6

Other pseudo-instructions available in Midas

are ones that provide means for performing assembly-time list
processing,, symbol manipulation^ and loop termination as well
as for changing the course of assembly in response to certain
conditions.

Report No. ±422

Bolt Beranek and Newman Inc

Formal constraints on the construction and manipulation of
symbols are few,, so the programmer may,, within the range of
processor capabilities^ vary formats to suit particular
programs. The programmer is free to ignore any of the special
features and use Midas as a simple mnemonic code translator.
The formal rules of the Midas source language and basic processor references are described in Section I-A.
Section I-B describes the functions and formats of all system
pseudo-instructions.,
Section I-C explains the use of macro-instructions' for performing an assembly.
Section I-D provides instructions for performing an assembly.
Error conditions that are detected during an assembly and associated error messages are listed in Section I-E.
The notation used in this volume includes some special symbols
\

represents carriage return.

-»| represents a tab.

Quotation marks indicate the pressing of the Control key on
the Teletype keyboard.

The symbols < > are used to enclose

text that would normally be set off by quotation marks.

Un-

less otherwise noted,, all integers appearing in the text are
octal integers.
1.

The Midas Source Language

A Midas source program is a string of alphanumeric and operational characters. Prom this string the Midas assembly program produces the words that make up the object program and

Bolt Beranek and Newman Inc

Report No. ±422

places these words in their proper locations in memory.

order to accomplish this,, the assembler must interpret the
source program as a series of meaningful strings. In most
cases9 a string that is meaningful to the assembler represents
a word in the object program.

In other cases^ the string may

direct the assembler to produce several or no words in the
object program.
This section describes the mechanics of creating legal Midas
character strings,, the references that the assembly program
uses in associating character strings with binary values,, the
conventions that instruct Midas as to the type of value or
actual value a string is to represent^ and the overall source
program format requirements.
The source program, described above as a single string of
characters is3 more precisely, a system of arbitrary strings^
each of which consists of individual characters Juxtaposed according to formal conventions.

The construction of legal

strings is hierarchical in nature.

The lowest level constitu-

ent strings are formed from the alphabetic members of the character set.

Higher level constituent strings are'constructed

from previously defined constituent strings using those members
of the character set which function as combining operators. The
type of object a constituent string represents is indicated by
punctuation characters.
a.

The Character Set

The complete character set from which the source language is
constructed is included in Appendix A, It consists of all
characters on the Teletype keyboard net specified as illegal.

Bolt Beranek and. Newman Inc

Report No. ±422

The character set is generally divided into the following
categories:
alphanumeric characters:

The letters A-Z and the
character,, <„>(period)*,,
which may "be constituents
of symbols; and the digits
$-93 which may be constituents of symbols or integers„

combining operators:

Single characters representing fixed arithmetical or
logical operations to be
performed by Midas.

punctuation characters

These serve as string delimiters. A stringdelimiting character may
serve a variety of purposes
depending on its use.
String delimiters in general identify individual
strings and often indicate
the manner in which a string
is to be interpreted. See
Appendix A for complete
listing. Most generally used
delimiters are space,,
and carriage return.

Legal Strings
(l)

Basic Strings

The minimum character strings required to represent values in
the source program are symbols or integers, formed as follows

*The character <.> also denotes a decimal number (as in
and may also be used to represent the value of the current location.

-7-

Report No. ±422

Bolt Beranek and Newman Inc.

Integers
An integer is a string of digits (03±3...9)
that is evaluated as octal or decimal according to the radix prevailing at its appearance. The integer value is its representation as an l8.-bit binary number which
restricts integer values to 777777 if the
radix is set for octal and to 262143 for
decimal. An integer above these limits
will be evaluated modulo
Symbols
A symbol is defined as a string of characters,,
the first six of which must distinguish it
from all other symbols .
Letterss numbers3 and periods may be used as
symbol constituents^ but at least one of the
identifying six must be a letter.
Longer symbols^ useful for mnemonic or documentary purposes,, may be used^ since the Midas
processor ignores any character except a terminator in excess of six.
Symbols for macro names and pseudo-instructions are subject to
the same restrictions as symbols for numerical values.
Note that the symbols <READIN> and <READINTAPE> are both legal
symbolsj if used in the same source program^ they both will appear to the assembler as <READIN> and will be used interchangeably. If distinct symbols are desired^ care must be taken to
differentiate symbols within their first six characters.

-8-

Report No. ±422

(2)

Bolt Beranek and Newman Inc.

Complex Strings
(i) Language Units

Complex strings may be formed from basic strings by use of
characters that are provided as combining operators. Although
integers and symbols are the only basic strings in the Midas
language that are formed by purely alphanumeric concatenations^
a complex string may be bracketed and function in the same way
as a basic string in a new combination. In discussing the construction of complex strings^ language units will be called
either syllables or expressions as defined below.
Syllable
A syllable is any component string of an
expression whose value is independent of
its use in the expression. An expression
enclosed in brackets may be used as a syllable to form other expressions. In addition., the character <.>^ used to represent the current value of the location
counter modulo (212»), functions as a
syllable.
Expression
An expression is a string consisting of
one or more syllables separated by combining operators.
(ii)

Combining Operators

The characters listed below according to function^ are the
Midas combining operators. Quotation marks denote that the
Control key must be pressed while typing the character.

-9-

Bolt Beranek and Newman Inc.

Report No. ±422

Product Operators
MmM

"U"
"A"

"Q"
"R"

Additive Operators
+ or space

Function
Folded integer multiplication
Logical disjunction
(exclusive OR)
Logical union
(inclusive OR)
Logical intersection
(AND)
Quotient
Remainder
Function
Addition^ mod 21^«-1
Addition of the one's
complement

Note that <A"TMB> results in an l8.-bit quantity equal to the
sum of the unsigned magnitudes of the high and' low order
halves of the 36.-bit product produced by regular multiplication of the two ±8.-bit quantities A and B. If both A and B
are small enough, this function produces the ordinary product
When evaluating expressions,, Midas performs operations from
left to right,, all product operations preceding additive ones
If a string of consecutive additive operators occurs^ Midas
will perform only the last. A string of consecutive product
operators, however, cannot be processed.
Examples. In the following examples, various equivalent expressions are shown, and their component syllables listed.

-10-

Bolt Beranek and Newman Inc.

Report No. 1422

Expressions:
Syllables:

LIO

±00 LIO
<LIO>f<±00>

b)* Expressions:
Syllables:

LIO ADD
<LIO>,<ADD>

JSP
<JSP>

Expressions:
Syllables :

7-2"U"3

6-2

Expressions:
Syllables:

2"T"A-A

Expressions:
Syllables :

A+BMTM

A+[B"T"C]

220±00

Note that the expression [A+B]"T"C is not equivalent to those
of (e).
2.

The Assembly Program

In order to interpret symbols and integers and to assign them
to memory locationss Midas must make references to the CurrentLocation Counter, the Symbol Table,, and the Radix Indicator,,
which are described below.
a.

The Current Location Counter

The Midas Assembler assign assembled words to sequential locations,, starting from any given location. A register in the assembly program^ referred to as the Current-Location Counter,, is
indexed whenever a location is assigned^ indicating the location which will be assigned to the next word assembled. It is
*Assume that LIQ&220000, ADD=^000003 and JSP=

-11-

Report No. 1422

Bolt Beranek and Newman Inc.

±6
initially set at 110 and counts upward modulo (2 "). Conventions are provided in the source language so that the programmer may assign a numerical value to the current-location counter^
thus specifying the first location in any sequence. Using
other conventions, the programmer may direct the assembler to
increment the current location a specified number of times
although no words are assembled.
A source language string may be specified as a direct representation of the value of the location counter at any time during
the assembly and used in place of that value throughout the
source program. In addition,, a single character is provided
that may be used to represent its value in any word while it
is being assembled. Any value assumed by the current location
counter is an actual location of a storage word. Representations of its value^ however^ are equivalent to the sum of the
location plus a number referred to as the offset count.
Unless reset by the programmer^ the offset count will be $3
and representations will coincide with the locations at which
they are derived. The offset count may be set by the programmer^ however^ and an internally consistent set of values
derived relative to the locations being assigned. The
utility of offsetting the current location is discussed later
in connection with the pseudo-instruction OFFSET. It is mentioned here to emphasize the fact that a number is assigned
to a symbolic address tag in order to utilize a list position
rather than to represent it.
While the fact that program addresses and computer instruction
codes may be symbolically represented is sufficient knowledge
for writing source programs3 the fact that these represent
numbers to the assembler^ and locations and operations only

-12-

Report No. ±422

Bolt Beranek and Newman Inc.

in context of the source program^ will be useful to understanding the full range of symbolic manipulations that may be
performed. For example, if a computation to be performed requires the number Ipj2^ we might <LA¥> the number rather than
provide a register containing ±t, and if the number 1<#0 is
represented by the symbol <A>, we can <LA¥ A> whenever we
need 1$# in the accumulator whether <A> derived its value as
an address tag or as a parameter. As further illustration,,
one step in the computation might be to add the contents of
register 100^ to the value in the accumulator^ requiring a
word in the executable sequence containing the number 401p>$#.
Given the symbol <ADD> representing the numerical operation
code <k$$$$$>3 the symbol 9 which has been set to equal
<l$0j#> as a symbol address tag., and the operator space3 which
signifies addition^ the source language expressions, <ADD B>
or ^ would give the proper value. Storage of the
expression in an executable sequence is accomplished by setting the expression in the proper context. If we need to perform
a computation using the number <4^1pj2^>, we might think in terms
of creating a data word to contain it when we could just as
well use the instruction word and would particularly wish to
when computer space is limited. The source language takes
advantage of the fact that a computer word may be an instruction^ data^ or both according to its context in the program.,
in that strings^ while associated with a number in one prescribed context^ may then be generally applied.
While the advantages of contextual as well as explicit interpretation may be clear to programmers who are familiar with
the techniques of machine language coding and who will be looking for these properties in the source language,, others might
well tend to think in terms of symbols for addresses and symbols for words as quite different entities.
-13-

Report No 0 ±422

Bolt Beranek and. Newman Inc.

TheTjymTbol_ Table

Symbols3 when associated with a value,, are entered in the Midas
Symbol Tale^ which is used as a reference toy the assembler.
Mnemonic symbols for PDP-1 instruction codes are part of the
initial contents of the symbol table. (A list of these is ineluded in Appendix B e ) These and programmer constructed symbols3 which stand for numerical values, are "substitutive"
symbols 0 Two other symbol types are included in the symbol
tables pseudo-instruction names and macro names. These are
referred to as "operational" symbols j there is no single address or single l8.-toit number with which they are synonymous „
All pseudo-instruction names are included initially in the
symbol table, defined by a reference to an assembler routine.
A macro name is entered in the symbol table when a macro instruction is entered in the macro table3 and a macro name is
defined as a reference to the macro table location of .the instruction to which it is assigned.
Midas assembles a source program in two passes j that is, two
complete scannings of the source program are necessary to produce an object program. Midas was constructed this way so
that symbols may be associated with values at any point in the
source program without restricting their use prior to definition, thus permitting the programmer freedom from rigid format constraints.
c•

The Radix Indicator

Integers are interpreted as octal or decimal according to the
radix prevailing when they are encountered by Midas. A register in Midass called the current radix indicator, is initially

•14-

Report No. ±422

Bolt Beranek and Newman Inc.

set to 8 for accepting octal integers. Subsequent settings
of 8 or ±0 may be effected by source program specification.
3.

Defining Symbols

The Midas symbol table, described earlier, functions as a dictionary during the translation process. Each symbol introduced
by the programmer is inserted together with its value into the
symbol table during Pass 1. The value of a symbol is referred
to as its definition. Numerical definition of a symbol may be
accomplished by its appearance as an address tag, a variable
name, or in a parameter assignment. Symbols may also be defined as macro-instruction names or in terms of other symbols.
Macro name symbols are discussed in the section on macroinstructions. The establishment of symbol synonyms is discussed in Section II in connection with the pseudo-instructions,
EQUALS and OPSYN.
The three basic formats that Midas requires for assigning a
symbol to a numerical value are described below.
a.

Address Tags

A symbol is identified as an address tag if it is terminated
by a comma or a colon. An address tag is equated with the
value obtained by adding the value of the location counter
when scanned to the offset count. The resulting value is
IP
entered into the symbol table modulo (2 *) if terminated by
a comma, and modulo (2 *) if by a colon. A symbolic address
tag identifies a line in the source program text and is required only for lines referenced within the text.

-15-

Report No e ±422

Bolt Beranek and Newman Inc

An expression,, such as <A+3>^ terminated by a comma or colon
will be admitted for use as an address tag only if it coincides with the value of the current location counter plus the
offset count. Since relative addressing using address arithmetic is an intrinsic system capability^ an untagged word may
always3 in effect., be referenced by such an expression. If a
previously defined symbol occurs as an address tag,, the symbol
will not be redefined and will be admissible only if its value
agrees with that of the current location counter plus the offset count.
b.

Variable Names

A programmer may direct the Midas assembler to reserve a sequence of storage words for variable quantities produced by a
computation. Symbols may be substituted for these locations
before they are known. The assembly program classifies such
symbols as "undefined variables" and assigns them provisional
relative values. A string is classified as an undefined variable by the inclusion of the character, #3 in the identifying
string itself3 that is^ within the first six characters, in at
least one of its appearances. The symbol is stored in the
symbol table without the qualifying character and may be referred to with or without it. Actual values are entered for
all variables still undefined at the appearance of the pseudoinstruction VARIABLES. The following are legal variable names
<ABC#>3<#ABC>^<AB#C>. For regularity of documentation the
second form listed, however^ is usually used.

-16-

Report No. ±422

Bolt Beranek and Newman Inc.

Assj-gned Parameters

A programmer may^ with an identity declaration^ assign a value
to a symbol that is to be used as a parameter.
The format is
SYMBOL==EXPRESSION
The symbol to the left of the equals sign is entered in the
symbol table3 and the value of the expression to the'right is
entered as its definition. If no value can be obtained for the
expression,, the symbol is not defined.
4.

The Use ofExpressions

The rules for forming expressions were given earlier. The
evaluation of an expression depends on its context in the
source program as well as on the value of its component syllables. Contexts in which Midas evaluates expressions are
described below.
a.

Storage Words

An expression terminated by a tab or carriage return is a
storage word. A storage word^ when encountered by Midas,
is evaluated and assigned the memory location equal to the
value of the current location counter. The contents of a
storage word may ultimately be used as an instruction and/or
operated on by an instruction,, depending on the use of the
word in the object program.

-17-

Report No, ±422

Bolt Beranek and Newman Inc

Constants

Constant values required by a program need not be introduced
as storage words in the source program. The constant expression desired^ enclosed in parentheses^ may appear literally as
the operand of an instruction.
For example,, an instruction to subtract ±00. from the accumulator could be written as <SUB(l$#.)>. Midas will generate a
word containing the value ±00. The string <(±00.)^- is called
a constant syllable^ and the address of the word containing
<±00.> is substituted for it. Note that (±00.), (50.+50.),
or (A+25.)* where A=75. are equivalent constant syllables and
will all refer to the same location.
Constants may appear within constants to any depth, as in
(1)

LAC(LAW(FLEXO ABC))

The right parenthesis of a constant syllable may be omitted if
the constant is followed by a word terminator.
For example}
(2) LAC(LAW(FLEXO ABC *
is equivalent to (±) above.
Omission of the right parenthesis in

(3) LAC(LAW(ABC)-I .
'd
would, however, change the meaning.
-18-

Report No. ±422

Bolt Beranek and Newman Inc.

Location Assignments

A location assignment is an expression immediately followed by
a slash. When Midas encounters a location assignment,, the expression is evaluated^ and the location counter is set to this
value. If an expression that is used to assign a location contains any undefined symbols when encountered by Midas on Pass ±,
the current location becomes indefinite. This means that the
definition of address tags is inhibited until a defined location
assignment occurst and at that time the counter again becomes
definite. On Pass 2S an undefined symbol in a location assignment will cause an error message (USL). The undefined symbol
is taken as zero5 and the location remains definite. The
Midas command^ "E", discussed in Section I-D permits the programmer to arrange for Midas to type a message if the location
becomes indefinite on Pass 1.
5.

Source Program Format

Midas begins processing a source program after it encounters a
title. A title may be any string of characters terminated by
a carriage return. Initial carriage returns are ignored. The
end of the source program is indicated by the appearance of
the START pseudo-instruction.
The portion of the source program which is to be assembled^
referred to as the body,, is composed of character strings.
These strings are processed by Midas sequentially.

-19-

Report No. ±422

Bolt Beranek and Newman Inc

Comments may be entered throughout the source program. Any
string of text characterized as a comment will be ignored by
the assemblerc A comment is introduced by a slash and must
be preceded and terminated by either a tab or carriage return.

-20-

Bolt Beranek and Newman Inc.

Report No. 1422

Figure 1.

Two sample Midas programs for solving the equation:
Y = 7A+B/2, for A = 40, B = 60. The symbolic programs in columns (l) and (2) will produce the machine language program in column (3).

CD
SAMPLE PROGRAM 1
LAC A
MUL SEVEN
RIR IS
DIG SEVENA
LAC B
SAR IS
ADD SEVENA
DAC Y
HLT
40
A,
SEVEN, 7
6<2f
B,
SEVENA,

(3)

(2)
SAMPLE PROGRAM 2
LAC (40
MUL (7
RIR IS
DIG SEVENA
LAC (60
SAR IS
ADD #SEVENA
DAC #Y
HLT

CONSTANTS
VARIABLES
START

200111
11

540112

103
104

32(2114

105
106
107
110

675001
400114
240115
7604<2f0

112

111

113
114
115

200113

START 100
PRINTOUT FROM ASSEMBLY:
(2)

(1)

DEFINED SYMBOLS ALPHABETIC

A
B
Y
SEVEN
SEVENA

111

CONSTANTS AREA,

INCLUSIVE

FROM

111

113

115
112
114

DEFINED SYMBOLS

ALPHABETIC

Y
SEVENA

115
114

The strings <CONSTANTS>^ <VARIABLES>, and <START> in the sample
programs are pseudo-instructions, which will be discussed in
the following section.
-21-

Report No, ±422

Bolt Beranek and Newman Inc.

PSEUDO-INSTRUCTIONS

Pseudo-instructions are source-language expressions that serve
to direct the assembly process. A pseudo-instruction statement
consists of a pseudo-instruction symbol terminated by a delimiter and followed by arguments as required. Unless otherwise notedj a pseudo-instruction statement is terminated by a
tab or a carriage return. Pseudo-instruction symbols,, like all
other symbols^ are identified by no more than six characters.
Thus^ pseudo-instruction symbols composed of more than six
characters may always be abbreviated. For example,, DIMENSION
may be shortened in use to DIMENS. The pseudo-instruction
repertoire is described below with regard to format and function
1.

OCTAL and DECIMAL

When integers are encountered^ they are interpreted as octal
or decimal according to the value of the prevailing radix indicator. The pseudo-instructions OCTAL and DECIMAL reset the
radix indicator, which is set by Midas to OCTAL at the beginning of each pass. An integer syllable followed directly by
a period will be Interpreted as a decimal number regardless of
the current radix and will not change the radix value.
2.

CHARACTER, FLEX03 and TEXT

These three pseudo-Instructions were originally devised for
storing 6-bit Friden code characters. The translation of
7~blt Teletype code to Internal Code may result in 6 or 12. bits
It has been left to the programmer to introduce 12.-bit code
only where two 6-bit characters are permitted. A table of
characters and their Internal Code is included as Appendix C.

-22-

Report No. 1422

Bolt Beranek and Newman Inc.

The pseudo-instruction CHARACTER is used to place a character
code in the left, middle^ or right 6-bit portion of an l8.-bit
word. The symbol, CHARACTER, is followed by L, M, or R according to desired bit position, immediately followed by the character to be coded. Characters with 12.-bit codes cannot be
stored this way.
The format is

Stored as ;

CHARACTER RA
CHARACTER MB
CHARACTER LC

000041

The character strings shown above are pseudo-instruction syllables and may be used in the same manner as symbols or integers. For example,
LAC (CHARACTER LA + DTB) is equivalent to
LAC (410000 + DTB.
The pseudo-instruction FLEXO is used to specify an entire
computer word filled with character code, either three 6-bit
characters or one 12.-bit and one 6-bit.
The format is

Stored as

PLEXO ABC

414243

PLEXO A*-

417776

The pseudo-instruction TEXT is used to assemble a string of.
characters by groups of two or three, as appropriate, into
successive words in the object program.

-23-

Report No. ±422

Bolt Beranek and Newman Inc.

A TEXT statement consists of the symbol <TEXT> terminated "by a
delimiter and followed by a string of characters. The first
character in the string is used as a delimiter of the text
string itself; it is not stored as part of the text string,,
and its reappearance terminates code storage. Thus,, if given
the string <TEXT /MESSAGE/^ Midas will store character code
for the word^ MESSAGE. A 12.-"bit code. character is not a good
choice for a text delimiter. Only six-bits at either end of a
string are interpreted as delimiters^ so the remaining six
bits of a 12.-bit character would be included in the stored
text.
If the character <#> is used in the argument of CHARACTER,,
FLEXO., or TEXT.? it is handled in an exceptional way; it is
stored as end-of-mess age (internal Code 74) rather than as
its own Internal Code configuration,, 03. Consequently^ there
is no provision for entering <#> as actual text. Most hospital user programs use the code^ 7^ as a text terminator; it
may appear in no other context.
Three pseudo-instructions--CONSTANTSA VARIABLES, and DIMENSIONare provided to direct automatic storage assignment of words
for constant and variable data. Variable data words may be
generated Individually by reference or as fixed length arrays
obtained with the DIMENSION pseudo-instruction. Constant data
words are generated to accommodate literal references.
3.

CONSTANTS

The pseudo-instruction CONSTANTS effects the allocation of
constant syllables to storage words containing constant values,,

-24-

Report No. ±422

Bolt Beranek and Newman Inc

beginning at a location whose value is equal to that of the
current location counter at the appearance of CONSTANTS. When
a constant syllable is allocated,, its address is substituted
for it at all references. If different expressions enclosed
within parentheses have the same value^ they are considered to
be the same constant syllable and are associated with only one
location. The pseudo-instruction CONSTANTS may be used no more
than ten times in the same program.
Since storage space for constants is allocated on Pass 1 when
some expressions may not be definite3 the number of registers
reserved in the constants area may exceed the number Midas
needs when all references have been consolidated; thus,, a gap
of unused registers may arise between a constants area and any
subsequent portion of the object program.
The following examples show symbol prints following Pass 1
and Pass 2 for the same program.
CONSTANT AREA RESERVED, INCLUSIVE
PROM

325

337

indicates registers reserved during Pass 1.
At the completion of Pass 2, the printout
CONSTANTS AREA, INCLUSIVE
PROM

325

332

indicates those registers actually containing

-25-

constants.

Report No» ±422

Bolt Beranek and Newman Inc.

VARIABLES

The value of the current location counter at the appearance of
the pseudo-instruction VARIABLES on Pass 1 marks the beginning
of the storage area allocated to variables. All variable names
still classified as undefined are assigned locations at this
time. The relative value assigned to an undefined variable is
added to the value of the first location in the sequence and
the result obtained entered as its absolute address. Each
defined variable represents an address at which Midas assigns
a storage word whose contents are unspecified.
When a variables area has been completely allocated^ the value
of the current-location counter is that of the next location
at which a storage word will be assembled. If VARIABLES appears when the location counter is indefinite,, it is inadmissible. Thus it is wise for a programmer to use Midas command
"E" (described in Section I-D) to arrange a printout if the
location becomes indefinite on Pass 1 so that he can correct
the condition before the processing of VARIABLES.
In the current version of Midas the pseudo-instruction VARIABLES
may be used no more than ten times. If the maximum is exceeded^
the error comment TMV (too many variables) is typed. The number of defined variables, however^ is limited only by the
capacity of the symbol table.
At the occurrence of VARIABLES on Pass 2, Midas compares the
value of the current location counter with the value which was
associated with that variables area on Pass 1. A disagreement
is noted by the error message VLB (variables location disagrees)^
which indicates that subsequent symbol definitions or macroexpansions have altered the sequence of assembled words.
-26-

Report No. ±422

Bolt Beranek and Newman Inc.

DIMENSION

The pseudo-instruction DIMENSION reserves in the variables
storage area "blocks of registers, which may be referenced
relative to a single symbolic address. DIMENSION is used to
set up fixed-length arrays. The pseudo-instruction symbol and
the name and extent of any number of arrays constitute a
DIMENSION statement according to the following format:
DIMENSION NAME1(LGTH1),NAME2(LGTH2)
where the entire statement is terminated by a tab or carriage
return and requested blocks are separated from one another by
commas. The array name must be a legal symbol that has not
been previously defined. The extent must be stated as an expression whose syllable values are known when the DIMENSION
statement is encountered on Pass 1.
6.

EQUALSand OPSYN

The pseudo-instructions EQUALS and OPSYN permit a user to,establish symbol synonyms, representing the same value. The format
is
EQUALS SYNONYM,SYMBOL
or
OPSYN SYNONYM, SYMBOL
where <SYNONYM> must be a legal symbol string and the <SYMBOL>
with which it is identified must be previously defined.
<SYNONYM> is assigned the same numerical or operational value

-27-

Report No a ±422

Bolt Beranek and Newman Inc.

as <SYMBOL>, and the two are thereafter synonymous. OPSYN
effects this on Pass 1 only; EQUALS, on "both passes. The
following example illustrates the difference between the two.
Let us say, for example, that a programmer wanted to use the
macro-instruction facilities to redefine a pseudo-instruction
such that the new instruction was a function of the old. The
original pseudo-instruction would have to be represented "by a
different symbol] otherwise, its appearance in the definition
would act as a macro call, resulting in a closed loop.
For example, if one writes
EQUALS CHAR, CHARACTER
and then defines a macro-instruction, CHARACTER, in terms of
CHAR, which now calls the pseudo-instruction, on Pass 2 CHAR
will again be made equivalent to CHARACTER, which has been redefined. CHAR then no longer references the pseudo-instruction,
and the loop avoided on Pass 1 will occur anyway on Pass 2.
If one uses OPSYN, however, CHAR will be associated with
CHARACTER as desired on Pass 1 only and retain its identity
with the original pseudo-instruction on Pass 2.
7.

NULL

The NULL pseudo-instruction performs no action, but it is used
as a substitute for symbols no longer needed in a program.
Some programs are required to be compatible with various environments (different machines, data bases, etc.), and a function performed frequently in one usage may not be performed at

-28-

Report No. ±422

Bolt Beranek and Newman Inc.

all in another. For example^ a symbolic program may contain
complex macro-definitions that need not be assembled in one
instance but must be available for other processings. In
this case^ the macro names may simply be equated with NULL5
as in
EQUALS MACRO,NULL
Another case in which NULL is useful arises in connection with
macro-table storage space and the "garbage collector."* When
the amount of space available for the storage of macrodefinitions is exhausted^ the garbage collector will search the
table for definitions which no longer have a reference in the
symbol table and will recover such space by consolidating the
remaining table entries.
The symbol-table reference to a macro that has been redefined
is automatically transferred to the latest definition.
In the
case of macros that have not been redefined but are simply no
longer needed,, the reference must be suppressed in order to
notify the garbage collector of the available space.
8.

OFFSET

The pseudo-instruction OFFSET is used to set the value of the.
offset count^ whose relation to the current location counter
was described in connection with symbol definition.

*The garbage collector will soon be available for this system.
Available macro storage space is 6l^44^10 computer words (approximately l8j2T,$0^ characters).

-29-

Report No. ±422

Bolt Beranek and Newman Inc

The psuedo-instruction format is
OFFSET EXPRESSION
where the value of the expression (positive or negative) is
stored as the offset count. When the offset count is any
value other than zero., a symbol value derived as an address
tag will not equal the core location of the storage word with
which it is associated. For example^ the coding:
OFFSET 5
ABC,
LAW ±00
JMP ABC
occurring when the current location counter contains ±00 will
be assembled as :
±00,

LAW
JMP ±05

A portion of the object program that was assembled under these
conditions is not executable at the location it occupies if
storage word expressions use these symbols as referents. The
offset capability is^ however^ useful in creating a body of data
independent of its core location^ yet internally consistent.
The effect of one OFFSET declaration is terminated by the appearance of another. If a return to the normal sequence is
desired^ the programmer must set the offset count to zero.
OFFSET is used in connection with memory "renaming" and in
constructing item maps.

-30-

Report No, ±422

Bolt Beranek and Newman Inc

REPEAT

REPEAT instructs Midas to assemble a specified portion of the
source program a specified number of times and thus relieves
the programmer of the necessity for source language repetition
of a repetitive object program sequence.
The format is
REPEAT EXPRESSION,TEXT
where EXPRESSION is the count of the REPEAT, specifying the
number if iterations desired, and the TEXT is the source program section to be iterated, called the range of the REPEAT.
The count must be defined when Midas encounters the REPEAT on
Pass I; otherwise, the range is ignored and the error print
<USR> occurs.
A carriage return is used to terminate the entire instruction;
tabs are used within the range to denote storage words. Tabs
may also appear within a macro-definition or as an argument.
Brackets may be used to enclose portions of the range or the
entire range so that carriage returns may be included.
Since a REPEAT merely serves to reproduce a string, the range
may include any elements of the source language, including
other REPEATS and macro calls. An internal REPEAT, unless it
is at the end of the range, must be bracketed] otherwise its
terminating carriage return would also terminate the first
REPEAT.

-31-

Report No. ±422

Bolt Beranek and Newman Inc.

In the following example^ the statement
REPEAT 2, LAC A -| ADD B -»| DAC C
will generate for assembly the coding
LAC A
ADD B
DAC C
LAC A
ADD B
DAC C
If the count of a REPEAT is zero or negative the range is not
processed.
10.

START

The START pseudo-instruction directs Midas to stop reading characters. START must appear at the end of every source-language
file and may take as an argument an expression denoting the
starting address of the object program.
The format is
START EXPRESSION

e
At the end of Pass 2 in response to command "J", (Section I-D),
Midas appends to the binary output a word containing
<JMP EXPRESSIONS where EXPRESSION is the argument of the last
START processed.

-32-

Report No. ±422

11.

Bolt Beranek and Newman Inc.

EXPUNGE

The pseudo-instruction EXPUNGE removes symbols from the symbol
table.
The format is
EXPUNGE SYMlj,SYM2,SYMN
where the argument is a list of symbols3 separated by commas
and terminated by a tab or carriage return. Any type of symbol
may be expunged, Midas ignores undefined symbols in the list,.
If any member of the list is not a legal symbol^ Midas ignores
the rest of the list., An expunged variable will not be defined
unless it appears again with <#> after the EXPUNGE; <#> itself
may not appear in the argument list.
12.

WORD

The pseudo-instruction WORD appends l8.-bit computer words,,
specified by the argument(s) of the pseudo-instruction^ to a
block of binary output.
The format is
WORD EXPRESSION
or
WORD EXPR1.EXPR2,...EXPRN

The appended words are not necessarily part of the object
program; their values are selected to produce special binary
formats when needed. For example^ words might be appended in

-33-

Report No. ±422

Bolt Beranek and Newman Inc

order to accommodate a particular loader or to insert jump
"blocks before the end of assembly. Normal binary output format
is discussed in Section I-D.
13.

0IF and IIP

A programmer may find it useful,, particularly when handling
comples macro-instructions, to be able to test the value of an
expression and to condition part of the assembly on the result.
Such testing is effected by the pseudo-instructions $TP and
IIP, in conjunction with symbols called qualifiers,, which
represent tests available. The tests are as follows:
Qualifier
VP

Condition is true if:
the evaluated expression
is greater than or equal
to ±0

the evaluated expression
is equal to ±$

Pass 2 is being performed

the expression tested is
a defined symbol

the argument contains no
characters (usually a dummy
symbol of a macro or IRP)

Report No. ±422

Bolt Beranek and Newman Inc

The format of conditional statements is

IIP
0IF
1IF

VP
VZ
D
N

EXPRESSION
EXPRESSION
SYMBOL
SYMBOL

where the test requires an argument^ and otherwise^
The value of 1IF is 1 if the condition is true^ 0 if false;
the value of 0IF is $ if the condition is true^ 1 if false.
A conditional statement may be terminated by tab^ carriage
return or comma. A conditional value may be used as a syllable; in this case the conditional must be terminated by a
slash.
For example:
LAC (0IF VP X/+3
is equivalent to
LAC (3) or LAC (4)
while
LAC (0IF VP X+3
is equivalent to
LAC (0) or LAC (l)

-35-

Report No. ±422

Bolt Beranek and Newman Inc.

/IF and IIP are often used to obtain a zero or one as the
count of a REPEAT.
For example:

(1) END: /
REPEAT /IP VP 7777-END, PRINTX /OVERFLOWED CORE/
A Ci

The address assigned to END (colon indicating address modulo 2 ')
is subtracted from 7777. If 7777 is greater, the test is true,
and the value of /IP will be /j thus, the count of the REPEAT
will be /, and the message will not be printed.
(2) REPEAT IIP P, EXPUNGE TYO,TYI,ONE
Example (2) will on Pass 2 direct Midas to expunge the listed
symbols.
14.

PRINT and PRINTX

The pseudo-instructions PRINT and PRINTX effect an on-line
printout by Midas during assembly. These instructions are
particularly useful for obtaining information during the
processing of complex macro-instructions.
The format is
PRINT or PRINTX TEXT
where the argument may be text of the form used with the pseudoinstruction TEXT or, if used in a macro-instruction, dummy
symbols.

-36-

Report No. ±422

Bolt Beranek and Newman Inc.

PRINT will cause Midas to type a line in the same format as the
first three columns of an error listing (described in the section
on error checking). The code PNT is substituted for an error
code in the first column and is followed by the argument and
a terminal line feed.
PRINTX causes Midas to type only the argument. Since both
pseudo-instructions are effective on both passes., a repetitive
printout can be avoided only if conditioned,, using $TF or IIP,
with the qualifier P. For example, in response to
REPEAT 0IF P, PRINT /TEXT/,
Midas will print only on Pass 1.
15.

STOP

The pseudo-instruction STOP is used when the programmer wishes
to arrest the expansion of a macro-instruction, an IRP, or the
range of a REPEAT. In any other context, STOP is ignored by
Midas.
Within the range of a REPEAT, STOP will halt the expansion of
all subsequent text unless the REPEAT occurs within the body
of a macro-instruction or an IRP. In that case STOP, whether
in a REPEAT range or not, will suppress subsequent coding
until the occurrence of the next TERMINATE or ENDIRP.
STOP may be used conditionally as in the following:
REPEAT 3. [REPEAT 1IF VZ A-B,STOP
A=A-B]

-37*

Report No. 1422

Bolt Beranek and Newman Inc

The text A=A-B will appear for processing up to three times.
However, if A=2 and B=l at the start, the count of the inner
REPEAT,, which generates the STOP,, will have the value one before the second appearance3 and the expansion of the first
REPEAT will be arrested. STOP may also be supplied as an
argument for an IRP or a macro call.
The remaining pseudo-instructions—DEFINE, TERMINATE, IRP,
IRPC and ENDIRP--are described in the section on macroinstructions .

-38-

Reprot No, 1422

Bolt Beranek and Newman Inc.

MACRO-INSTRUCTIONS

A macro-instruction is any legitimate source-language text
that a programmer names and sets up so that when the name
appears in the subsequent source program,, Midas will assemble
the text. The text and macro name are established by a macrodefinition^ whose format is described below. Where the text
includes parameters that may differ with each occurrence of the
macro-instruction^ these parameters may be represented by dummy
symbols.
1.

Macro-Defiiirtions

A macro-definition is initiated by the pseudo-instruction
DEFINED delimited by any terminator. DEFINE is followed by a
macro name. A macro name must be a legal symbol, which, if
previously defined, will be redefined. The macro name is followed by a list of dummy symbols., if needed, and terminated by
a tab or carriage return. If dummy symbols that appear in the
text of the macro-instruction are not listed after the macro
name,, Midas will treat them as ordinary symbols. After a
macro name Midas interprets the first character other than
space as the first member of the argument list. The argument
list is discussed in greater detail later. Midas considers all
text following the name and argument line to be the body of the
macro-instruction. Midas stores this text until the appearance of the pseudo-instruction TERMINATED which signals the
end of the definition. The body of the macro may include any
element of the source language9 including other macrodefinitions or calls. Any dummy symbol from the list may
appear as a syllable in the body of a macro-definition.

Report No. ±422

Bolt Beranek and Newman Inc.

The basic format of a macro-definition Is illustrated "by the
following examples,
(1)
\

DEFINE
ABSOLUTE
SPA
)
CMA
•*
TERMINATE

(MACRO NAME)

The macro name,, <ABSOLUTE>5 subsequently serves as a macro call
in the source program.„ Midas will assemble the body of the
macro (<SPA> and <CMA>) into the object program at every appearance of the macro call*
(2)

DEFINE
SUM A,B,C
LAC A
ADD B
DAC #C
TERMINATE

(The character # must be the first character if it is used in
a dummy symbol string.)
The macro call <SUM XORG,XINC,,XMAX> will cause the following
sequence to be assembled.
LAC XORG
ADD XING
DAC #XMAX

-40-

Report No. ±422

Bolt Beranek and Newman Inc

Dummy Arguments

A programmer may use as many distinct symbols as desired as
dummy arguments in a macro-definition as long as each appears
in the dummy argument list. Members of the argument list are
usually separated from one another by commas.* The position
of an argument in the list is the model for the order of argu
ments supplied at a macro call.
Some syllables, although they are referenced only within the
body of the macro, will represent a different value at each
call. Such syllables may be represented by dummy symbols and
specified in the argument list as generated arguments,, for
which Midas will automatically provide a symbol. A list of
those dummy arguments for which Midas must generate symbols
is preceded by a slash and follows the list of arguments for
which the programmer must supply symbols as shown below.
DEFINE MACROSYM A,B/C,D,E
or
DEFINE MACROSYM /A,B,C
where all symbols are to be generated.
Symbols generated and inserted by Midas are of the form <...
<..A^2>, <...A$3>3 etc. If at a macro call the programmer
supplies a real argument in a list position corresponding to
that of a generated symbol,, Midas will accept the supplied

*The following delimiters are also acceptableU, n, ~, tab, =,(,)•

-41-

+, -3 space,

Bolt Beranek and Newman Inc.

Report No. ±422

symbol rather than generate one. A generated symbol may be
used to define a variable in the text; Midas will generate a
symbol of the form <#...Aj#l>.
The following examples give an idea of the use of generated
arguments.
1)

DEFINITION

CALL:

DEFINE CLEAR A.N/B
LAW A
DAP B
B,
DZM
IDX B
SAS (DZM A+N
JMP B
TERMINATE

EXPANSION
LAW TAB

DEFINITION

CALL:

DEFINE SAVEAC /A

EXPANSION

DAG #A

CLEAR TAB,

DAP,..A01
. ..A01, DZM
IDX . ..A01
SAS (DZM
JMP ...

SAVEAC

DAC #...
JSP SUBR

JSP SUBR
LAC A
TERMINATE

LAC . ..A01

If an argument is supplied at the call., as in <SAVEAC TEMP>^
the expansion will be
DAC #TEMP
JSP SUBR
LAC TEMP

-42-

Report No. 1422

Bolt Beranek and Newman Inc.

Midasa when scanning the body of a macro-definition for dummy
arguments^ compares each legal symbol in the text with the
symbols in the dummy argument list. Those symbols that correspond to any dummy symbol are stored in a special way., as
described in Section I-C-3* Storage of Macro-Instructions.
If the programmer wishes to represent only a part of a symbol
by a dummy arguemnt^ he may use an apostrophe to denote this
in the body of the macro definition.
In the pseudo-instruction <CHARACTER RA> the string <RA> satisfies the requirements for a legal symbol. Midass however^ understands its special meaning within the context of the pseudoinstruction. During the dummy symbol scan^ however^ Midas
would interpret RA as a single symbol unless an apostrophe is
used to indicate that <A> alone is a dummy symbol, as in
DEFINE MACRO A
LAC (CHARACTER R'A
The apostrophe is deleted when the macro-instruction is defined. In the case of a nested macro-clefinition^ apostrophes
are also deleted at the time of definition; that is^ when the
higher level macro is called.
2.

Macro Calls

A macro call consists of a macro name followed by a list of
arguments separated by commas. The call is terminated by a
tab or carriage return.

-43-

Report No. ±422

Bolt Beranek and Newman Inc.

The arguments of a macro call may include any character string
(including an empty string) with the following restrictions.
Since comma terminates an argument and tab or carriage return
terminates the list, these may be included only in arguments
enclosed by brackets. Brackets must be used in pairs and may
be used within other brackets. Midas will consider all but the
outermost pair to be part of the argument.
At the appearance of the macro call, Midas processes the body
of the macro (stored in the macro table) as though it had
appeared in sequence. At this time Midas substitutes for the
corresponding dummy arguments and creates the correct number
of generated arguments.
If the programmer supplies extra arguments at a macro call and
the definition specified generated arguments, the extra supplied
arguments will take the place of generated ones. If, however^
arguments are supplied in excess of the total number (supplied
and generated)9 the excess arugments are ignored. Note that
Midas will not generate a symbol when a programmer fails to
supply one that has been specified in the definition.
3.

Storage of Macro-Instructions

After the occurrence of the DEFINE pseudo-instruction^ Midas
saves the name of the following macro-definition and scans the
list of dummy arguments^ keeping count both of the total number
of arguments and the number of these arguments that are-to be
generated. While stored in the body of the macro in the macro
table, Midas scans the text for dummy symbols. When Midas encounters a symbol that matches a symbol from the dummy argument
list, the list position of the corresponding dummy argument is
stored in place of the symbol in the text and is distinguished
by a code prefix.
-44-

Reoprot No. ±422

Bolt Beranek and Newman Inc.

Macro-definitions within the "body of the macro are stored
literally and defined only when the instruction containing them
is called.
When Midas encounters the final TERMINATED the number of words
that were required to store the definition is deposited in the
first macro-table register preceding the text. The macro name
and the table location of the definition are entered in the
symbol table.
4.

Nested Macros

It is convenient when discussing nested macro-instructions to
think of DEFINES and TERMINATES as if they were parentheses,
the outermost pair constituting the highest level macrodefinition. When the programmer calls the highest level macrodefinitiorij Midas stores the second level definition in the
macro table,, and so on. Internal macro-definitions may contain
dummy arguments of higher level ones. These arguments will
be replaced by supplied arguments when the higher level definition is called. Pairs of DEFINES and TERMINATES must count
out. To ensure that they do,, the programmer may use macro
name as the argument of a TERMINATE instruction. Then if the
DEFINE associated with that TERMINATE refers to another macro
name^ the error print <MND> (macro name disagrees) will inform
the user of a "mispairing" of DEFINES and TERMINATES.
A series of examples of nested macros follows. Note in example M-I the use of apostrophe and the insertion of a supplied
argument into a nested definition.

-45-

Report No. ±422

Bolt Beranek and Newman Inc.

M-I
DEFINE FLOAT INSTR
OLD'INSTR-INSTR
DEFINE INSTR X
LAW X
JDA F«INSTR
TERMINATE INSTR
TERMINATE FLOAT

For example^ if FLOAT MUL appears,, the expansion will be
OLDMUL-MUL
DEFINE MUL X
LAW X
JDA FMUL
TERMINATE MUL

This macro-instruction may be used to change PDP-1 instructions
to subroutine calls. Their original meanings could be restored
by
DEFINE UNFLOAT INSTR
INSTR=OLD»INSTR
TERMINATE
M-2
DEFINE MACRO X,Y
LAC X
DEFINE MAC2 Y
ADD Y
TERMIN
TERMIN

-46-

Report No* ±422

Bolt Beranek and Newman Inc

The call <MACRO ONE,T¥0> will generate
LAC ONE
DEFINE MAC2 TWO
ADD TWO
TERMIN
The argument supplied for <Y> at the call of MACRO must be a
symbol3 since it will be inserted as a dummy argument in the
definition of MAC2.
M-3 It is usually safer to use rather meaningless symbols as
dummy arguments to avoid duplication of real arguments.
For example,
DEFINE MACRO X
LAC X
DEFINE MAC2 COUNT
ADD (X+3
DAC COUNT
TERMIN
TERMIN
If COUNT is also a program symbol that the programmer inadvertently supplies at the call of MACRO., the result would be
LAC COUNT
DEFINE MAC2 COUNT
ADD (COUNT+3
DAC COUNT
TERMIN
-47-

Report No. ±422

Bolt Beranek and Newman Inc

Example M-4 "below illustrates a macro-instruction that redefines itself when first called.
M-4
DEFINE INCREM
DZM X
DEFINE INCREM
LAW 10
ADD X
DAC X
TERMIN
TERMIN
At the first call of INCREM the following text is generated
DZM X
DEFINE INCREM
LAM 10
ADD X
DAC X
TERMIN
Subsequent calls will generate
LAM 10
ADD X
DAC X

-48-

Report No, ±422

Bolt Beranek and Newman Inc.

The Pseudo-Instructions IRP and IRPC

The pseudo-instruction IRP (indefinite repeat) generates
sequential iterations of text a number of times determined "by
the analysis of its arguments. A different set of arguments is
substituted at each iteration.
An IRP statement consists of the <IRP> symbol followed by a
list of arguments3 each enclosed in brackets, terminated by a
tab or carriage return. Following the argument list is the
body of the IRP that^ like the body of a macro-definition5 may
include any source language elements^ including other IRP's
and macro calls or definitions. The body of an IRP is delimited by the pseudo-instruction ENDIRP.
Each argument of the IRP is itself a list of subarguments
separated by commas or carriage returns. The first two members of a subargument list are dummy arguments,, and each may
represent a syllable jn the body of the IRP. The remaining
members of the list are the real arguments of the IRP. Upon
encountering an IRP^ Midas processes the body of the IRP repeatedly^ with different symbolic equivalents substituted for
the dummy arguments each time according to the following procedure. Midas begins by substituting the first member of the
real argument list for the first dummy symbol and the remainder
of the real argument list for the second dummy symbol. The
remainder of the real argument list is then treated as the
real argument list in subsequent processings until all lists
are exhausted.
IPRC operates exactly as does IRP but on a different type of
list. Elements of an IRP subargument list are separated by

Report No. ±422

Bolt Beranek and Newman Inc

commas and may include a text string or a "bracketed expression
Real arguments of an IRPC subargument list are not separated
by commas; each character in the string is treated as an individual member of the list.
The following examples illustrate the use of IRP and IRPC.
Notice that the second dummy symbol may be omitted if not
referenced in the body, although its position must be retained
by a comma.
1-1.

IRP [NAME,,RPA,PPA,TYO,TYI],[VALUE,,1,5,3,4]
NAME=IOT VALUE
ENDIRP

This IRP will effect symbol table entries for the listed instruction symbols and their corresponding IOT commands.
1-2.

DEFINE TYPE DIGIT
IRPC [NUM,,0123.. .9]
REPEAT IIP VZ DIGIT-NUM,PRINTX /NUM/
ENDIRP
TERMINATE

Example 1-2 shows use of an IRPC within a macro-definition.
The digit supplied at the macro call will, during the expansion process, be compared with each digit in the subargument
list until its equal is found and printed.
Example 1-3 illustrates how to use the second dummy symbol^
which represents a list.

-50-

Report No. 1422

1-3

Bolt Beranek and Newman Inc

DEFINE MACRO LIST
IRP [X,Y,LIST]
REPEAT 0IF D X,MAC2[Y]
ENDIRP
TERMINATE

The list obtained for Y in the first IRP repetition is used as
a supplied list for another macro-instruction.
Example 1-4 shows a series of nested IRP's used to define a
macro-instruction that, given the list <X1,X2,..,XN>j, will set
up a matrix of the form:
XI,X2,

X2,X3,...XN,X1

-51-

Report No. ±422

1-4.

Bolt Beranek and Newman Inc

DEFINE MATRIX LIST
LGTH = $
IRP [,,LIST]
LGTH = LGTH-KL
ENDIRP
IRP [X0,LIST2,LIST]
COUNT = 1
IRP [XNJ,LIST2]
COUNT = COUNT+1
XN
ENDIRP
IRP [X02,,LIST]
REPEAT IIP VZ COUNT-LGTH.STOP

The first IRP gets the length of the list.
The second gets the next (initially first) member.
The third processes the remainder of the list.
The fourth goes back to the beginning of the list and takes
each element until COUNT = length of text.

-52-

Report No. ±422

Bolt Beranek and Newman Inc.

OPERATION OP THE MIDAS ASSEMBLY SYSTEM
1.

Preparation of a Source-Language Program

The programmer prepares his source-language program on-line via
Teletype terminal^ using the symbolic editing program^ Editor^
to type in and edit text and to store the program in Englishfile format on a programmer's quarter-tracks.* The file is
accessible to Midas by name and version number.
2.

Performing an Assembly
a.

Initial Procedure

English files to be processed must be on the drum for access
by Midas. Editor and Handle are used to place files on the
drum; Editor will enter files from the on-line Teletype or
from paper tape; Handle,, from magnetic tape, Midas runs
under DDT control and is called in the following way. First.,
the user types C"P!S which puts the Call program under DDT
control. Call requests the file name (Midas) then asks whether
or not to start the program. If "Y" is the response to this
question^ Call brings Midas into core and starts it running.
If "N" is the response^ Call simply brings Midas into core, and
to start it,, the programmer must type l^lnG"j to continue ±t3

*The programmer's quarter-tracks are described in Section IV-A

-53-

Report No. ±422

Bolt Beranek and Newman Inc

The condition of Midas when it is first brought into core is
as follows. The current location counter is set to 11^ and
the radix indicator to 8. The macro table is empty. The
symbol table contains all pseudo-instructions and a minimal
list of PDP-i instructions (these are listed in Appendix B).
Additional instructions are available in an English file entitled, "System Midas Initial Symbols." Also available is a
p-tape that contains a complete set of map macros and system
lOT's. It is usually more convenient^ however} for a programmer to construct his own p-tape of symbols that he is likely .
to use.
b.

The Control Language

Commands to Midas are specified by control characters, such as
1 for do Pass 1 and C for continue the present pass. Some of
the commands require arguments^ such as the name of the file to
be processed. Arguments3 when required^ are typed after the
control character. Midas performs a requested function after
line feed terminates the command string. Before the line feed
is entered^ an input string may be deleted by typing any 12.-bit
character other than a carriage return. Spaces and carriage
returns are ignored in a control string.
The control characters and argument requirements are listed on
the following pages.

-54-

Bolt Beranek and Newman Inc.

Report No. ±422

Control Characters

Function

Required Arguments

Initiate Pass 1

Name of English file
followed by a.comma
and the version
number.

Initiate Pass 2

Same as for l.

Continue present
pass on additional
file.

Name and version number of additional file

Initialize symbol
and macro tables.

None

The argument of E
represents a bit
setting. Certain
bit settings inform Midas to perform a special
function during
assembly. The bit
settings and their
associated functions are listed
below.
l6--print all characters processed

An octal number

15--print an error
comment on Pass 1
if the location
goes indefinite
14--define undefined symbols as
J2f on Pass 2

Add a jump block
to assembled binary program.
Selects address
following last
START encountered

None

Halt Midas

None

-55-

Report No. ±422

Bolt Beranek and Newman Inc

Control Character

Function

Required Arguments

Set up an indexed
file of binary output in programmer's
storage area.

Name and version
number as required
for file access.

Set up indexed file
of symbol table and
macro table.

Name and version
number as required
by general filing.

Load symbol and
macro table into
Midas.

Name and version
number under which
the table is filed

A simple assembly of a symbolic program consisting of two files
would be accomplished by the following sequence of commands.
1
C
2
C

PROGRAMX,!
PROGRAMX.2
PROGRAMX,!
PROGRAMX.2

J
3.

Order jpf Operations

The commands ±3 2^ Cy and J represent functions that must be
performed in a certain order. The other commands represent
functions that the programmer may select at various times
during an assembly. These commands are discussed below with
regard to the order in which they can be useful.
S--Set up File of Symbol Table and Macro Table
The "System Midas Initial Symbols" file and available p-tapes
are processed on Pass 1 as individual files in a multiple file

-56-

Report No. ±422

Bolt Beranek and Newman Inc

assembly. Before continuing Pass 1 on his own program, the
programmer may use <S>. Having set up these files in symboltable format3 the programmer need not perform Pass 2 on them.
The file produced by an <S> command may be retrieved by command <T>.
<S> is generally also used at the end of an assembly to produce a file containing all program symbols. This file is read
into DDT to permit symbolic debugging.
I--Initialize Symbol Table and Macro Table
When assembling a very long program whose symbols may exceed
the symbol™table limit or when appending a file whose symbols
may be defined in an earlier segment, a user may wish to file
the current symbols and use to initialize the symbol table
before continuing. On Pass 2 the first table may be read in
and the table reinitialized as necessary.
The faculty for initializing the symbol and macro tables also
permits the programmer to assemble a new program without restarting Midas from DDT.
B--PileBinary Output
 is normally used once during an assembly, at the end of
Pass 2. However, if the programmer wants his binary output in
two or more sections, he may use to save a partial binary
file and thus initialize the binary output buffer. Note that
while filing and initializing are separate functions with regard to the symbol table, they are one with regard to binary
output.

-57-

Report No. ±422

Bolt Beranek and Newman Inc

H—Halt
When an assembly is completed and the binary output filed^ the
user types <H> to halt Midas. Hitting the Break key returns
control to DDT at any time.

BINARY OUTPUT FORMAT

All blocks3 with the exception of the single-word jump block,,
begin with two words that indicate position and length and
end with a checksum word. The maximum block length is 1^38
words. The number of data words in a block is derived by
subtracting the first word from the second. The checksum
word contains the sum modulo (2l8"*-l) of all other words in
the block,, including the first two.

-58-

Report No. ±422

Bolt Beranek and Newman Inc.

The first two bits of the first and second word indicate the
type of block as follows:
Bits

Type
Absolute
Relocatable
Library

The present manual will describe only Absolute Blocks.
The first word contains, in addition to the type-indicating
bits^ the address in core where the first data word is to be
stored; the second^ the address following storage of the last
word in a block.
The pseudo-instruction WORD may be used to fabricate special
formats or to insert jump blocks without stopping the assembly
When Midas encounters a WORD pseudo-instruction^ it terminates
the current block with a checksum. The arguments of WORD are
appended directly to the binary output.

-59-

Report Nos 1422

Bolt Beranek and Newman Inc.

ERROR CHECKING

If Midas encounters an error in source-language coding^ the
assembly is interrupted and a descriptive error message printed
Depending on the severity of the errora assembly may or may not
continue. The format of an error message is exemplified as
follows:

(1)

(2)

(3)

(4)

(5)

USW

1000

ALPHA+1

REPEAT

GAMMA

Column (l) contains a descriptive error code; (2), the octal
address at which the error occurred] (3)* the symbolic address
stated in terms of the last address tag seen. Column (4) contains the last pseudo-instruction symbol or macro name Midas
encountered. Column (5)* used only in errors involving symbol
definition^ contains the offending symbol.
The error codes and the conditions with which they are associated are listed on the following pages, indicating action on
or impossibility of continuation of the assembly.

-60-

Bolt Beranek and Newman Inc.

Error
Condition
Designation
Causing Error
Undefined symbols
US a

Undefined symbol (a indicates where found):

in a constant

in size of dimension
array

in OFFSET count

in argument of 0TF or IIP

in a location assignment

in storage word generated by a macro call

in argument of EQUALS
or OPSYN

in a parameter assignment

in the count of a REPEAT

in the argument of a START

in a multi-syllabic address
tag

in a storage word

UWD

Action on
Continuation
All undefined
symbols are
evaluated as
zero.

undefined symbol in argument
of a pseudo-instruction

Relocation Errors
IRa

Illegal relocation; a
identifies where the error
was found with designations
as listed for undefined symbol

Relocation

Report No. ±422

Error
Designation

Bolt Beranek and Newman Inc.

Condition
Causing Error

Action on
Continuation

Multiple Definitions

MDT

Multiply defined tag

Original definition retained

MDV

Multiply defined variable (a symbol previously defined as other
than a variable appears
with a #)

Original definition retained

MDD

Multiply defined dimension (a previously defined symbol used as an
array name)

Original definition retained

MND

Macro name disagrees
(the argument of a
terminate disagrees with
the name being defined)

First name
used

ICH

Illegal character

The character
is ignored

ILP

Illegal format

Characters
are ignored
until the
next tab or
carriage return

IPA

Improper parameter assignment. (The expression to the right of
the equals sign is
inadmissible.)

The assignment
is ignored.

VLD

Variables location disagrees. (The pseudoinstruction VARIABLES
has appeared on Pass 2
at a different location
than on Pass 1.)

Condition
ignored

Other Errors

-62-

Bolt Beranek and Newman Inc.

Report No. ±422

Error
Designation

Condition
Causing Error

Action on
Continuation

Other Errors cont.

LGI

Location gone indefinite

IRX

Illegal operation with
relocatable symbols
(e.g., logical operations )

PNT

Not an error. Result
of PRINT pseudoinstruction

If the appropriate bit is
set (by Midas
control <E>)^
LGI is printed
on Pass 1.

In the event of the following error conditions,, assembly cannot
continue.

r
r

CLD

Constants location disagrees . The pseudoinstruction CONSTANTS
has appeared on Pass 2
in a location different
from Pass 1. All constants syllables have
been assigned incorrect
values

TMC

Too many constants (The
pseudo-instruction
CONSTANTS has been used
too many times in one
program)

TMP

Too many parameters (The
storage reserved for macro<
instruction arguments has
been exceeded)

-63-

Report No. ±422

Error
Designation

Bolt Beranek and Newman Inc

Condition
Causing Error

TMV

Too many variables (The
pseudo-instruction
VARIABLES has been used
more than lj# times in
one program)

SCE

Storage capacity exceeded
(symbol table and macro
table full, too many
constant words used)

IAE

An error which cannot
be diagnosed—often due to
improper operation of Midas

-64-

Report No, ±422

Appendix A.

Bolt Beranek and Newman Inc.

Midas Character Set

I, Alphabetic
Letters (A-Z)
Digits (jZf-9)
II.

Punctuation
Character

Function(s)

1P
indicates address tag mod (2 )
separates elements of a List
terminates count of a REPEATindicates address tag mod (216.)
equates symbol to the left with
expression to the right
a)
b)
c)
d)

terminates location assignment
introduces comment
introduces list of macro-instruction
arguments to be generated
terminates a conditional

enclose a literal
expression enclosed specified for
syllable function

denotes symbol as a variable
indicates global symbol*

tab
and
carriage return

a) word terminators
b) varying meanings according to
context.

•^Relocatable programming feature

A-l

Report No. ±422

III,,

Bolt Beranek and Newman Inc

Combining Operators
Product Operators
llmU

folded integer multiplication. ';•'.•:

"X"

logical disjunction (exclusive OR)
logical union (inclusive OR)

"U"
"A"

logical intersection (AND)

"Q"

quotient

MRU

remainder

Additive Operators
A O

+ or space

Addition, m o d 2 - 1
Addition of the one's complement

IV.

Illegal
A.

General
break
rubout

A-2

Report No. ±422

Bolt Beranek and Newman Inc

(except within a macro-instruction or an IRP)

vert, tab
HT/-U

Ignored (except within a macro-instruction or an IRP)

EOT
formfeed
carriage return
t

A-3

Report No, ±422

Appendix B.
1.

Bolt Beranek and Newman Inc

Symbols in Permanent Midas Vocabulary

PDP-1 Instruction Symbols

The following list contains all instruction included on the
"MIDAS INITIAL SYMS FOR T.S." tape. Those included in the
initial vocabulary are starred.
MIDAS INITIAL SYMS FOR T.S

1S=1
2S=3

3S=7
4s=l7
5.3=37
6S=77
7S=177
8S=377
9S=777

,
CLC=65l6

* DA 0=24

HLT=7 6 4

B-l

Report No. ±422

Bolt Beranek and Newman Inc

*ISP=46
*JDA=17
*JMP=6
*JSP=6'2

LF 1=

*RCL=6630

1= 6770W
SFT=660000
500

SKP=640000
1=640400

B-2

Bolt Beranek and Newman Inc

Report No. ±422

SZM=
SZO=
*TAD=
XAI=

*XCT=
XX=HLT

Pseudo-Instructions

Symbol

Function

CHARACTER
CONSTANTS

inserts Internal Code for one character
specifies storage areas for constant words
classifies integers as decimal numbers

DECIMAL
DEFINE
DIMENSION
ENDIRP
EQUALS
EXPUNGE
IRP and IRPC
NULL
OCTAL
OFFSET

initiates macro-definition
allocates storage area for arrays
ends an indefinite repeat
establishes symbol equivalence
erases symbols from symbol table
initiates indefinite repeat
no operation
classifies integers as octal numbers

PRINTX

assigns address tags as current location
counter and sets an expression whose value
is the offset count.
same as EQUALS; Pass 1 only
generates symbolic location printout and
prints comment during assembly
prints comment during assembly

REPEAT

generates iterative source-language text

OPSYN
PRINT

B-3

Report No. ±422

START
STOP

IIP

Bolt Beranek and Newman Inc

denotes end of source program and specifies starting address
ends expansion of IRP's, macro's^ and
REPEAT»s
ends macro-definition
inserts Internal Code for character string
reserves space for variables and arrays
appends word(s) to binary output block
tests an expression; if true^ value is
zero; if false^ one.
if true,, value is one; if false^ zero.

B-4

Report No. 1422

Bolt Beranek and Newman Inc

Appendix C. Teletype Code Conversion
("X" means control-X)
INTERNAL

ASCII

CHARACTER

00
01
02

040

S]
J

03
04
05
06
07.

043
044
045
046
047

10
11

050

052

13
14
15
16
17

053
054
055
056
057

20
21
22

&
06l
062

23
24
25
26
27

063
064
065
066
067

3
4
5
6
7

070

8
9»
j
<
=
>

30
31
32
33
34
35
36
37

04i

042

%&
I

)(

051

•
/

0
1

071

072

073
074
075
076
077

C-l

Bolt Beranek and Newman Inc

Report No. 1422

INTERNAL

40
41
42
43
44
45
46
47
50
51
52
53
54
55
56
57
61
62
63
64
65
66
67
71
72
73
74
75
76
77

ASCII

CHARACTER

103
104
105
106
107

A
B
C
D
E
P
G

110
111
112

H
I
J
K
L
M
N
0 (OH)

113
114
115
116
117

120
121
122
123
124
125
126
127

Q
R
S
T
U
V
¥
X
Y
Z

130
131
132
133
175,176,033
135
015-012

EOM
CARRIAGE RETURN-LINE FEED
WARNING

C-2

Bolt Beranek and Newman Inc.

Report No. ±422

INTERNAL

7700
7701
7702
7703
7704
7705
7706
7707
7710
7711
7712
7713
7714
7715
7716
7717

ASCII:

CHARACTER

000

001
002

"B"
"C"
EOT
"E" or WRU
"F" or RU
"G" or BELL

0^3
p04
005

007

"H"
TAB
LINE PEED
"K" or VT
"L" or FORM PEED
CARRIAGE RETURN (OUTPUT ONLY)
"N"
"0"
"pn
"Q" :
"R" or TAPE ;
"S" or RDR OFF

010
011

012
013

014
015
016
017

7720
7721
7722
7723
7724
7725
7726
7727

020

7730
7731
7732
7733
7734
7735
7736
7737

030

021

022
023
024
025
026
027

Ilrj-iM

"U"

"V"

"w"
"X"

031

Mry U

032

Li
ii r ti

033
034
035
036
037

7740-7743
7744
7745
7746
7747

UNUSED
134
UNUSED
136

7750-7773
7774
7775-7777

UNUSED
177
UNUSED

SHIFT
ii I it
II A

II.

I l T II

BACKSLASH

137
RUBOUT

C-3