Dec 282017
Pd version of prolog lang.
File PROLOG19.ZIP from The Programmer’s Corner in
Category Miscellaneous Language Source Code
Pd version of prolog lang.
File Name File Size Zip Size Zip Type
ALESS.PRO 256 148 deflated
ANS.PRO 512 250 deflated
BENCH.PRO 768 357 deflated
BICYCLE.PRO 1280 467 deflated
CONSOLE.PRO 640 256 deflated
DOCTOR.PRO 1920 673 deflated
FAMILY.PRO 768 321 deflated
GENSYM.PRO 1280 477 deflated
GRRULES.PRO 1024 459 deflated
HEMAT.PRO 3968 1182 deflated
MATRIX.PRO 1024 444 deflated
MEMBER.PRO 256 151 deflated
MEXP.PRO 1664 659 deflated
NUMBER.PRO 256 144 deflated
PARTS.PRO 1664 570 deflated
PDPROLOG.DOC 86784 24687 deflated
PDPROLOG.EXE 136064 44514 deflated
PLAYTEXT.PRO 384 171 deflated
PROLIB.PRO 384 174 deflated
ROCK.PRO 4736 1703 deflated
RSENT.PRO 1152 359 deflated
SIEVE.PRO 640 302 deflated
SUCCESS.PRO 256 68 deflated
WRCLAUSE.PRO 384 138 deflated

Download File PROLOG19.ZIP Here

Contents of the PDPROLOG.DOC file


1570 Arran Way
Dresher, Pa.
December 14,1985
Dear PD user:

Here's version 1.8 of PD PROLOG. It's better, of course. And
now I must ask a favor of you.

The AAAI fifth national conference on artificial
intelligence is on August 11 to 15th, 1986 in Philadelphia. Since
I live next to Philly, I'd love to exhibit there. The problem is
that the rate is $2200 for a 10' x 10' booth, with a $600 rate
for publishers.

I am appealing to the coordinator of the conference, Ms.
Claudia Mazzetti, for a publishers rate break with the
justification of my contribution to the public domain. I have
sent to Ms. Mazzetti a financial statement which shows clearly
that I cannot afford the standard rate. Ms. Mazzetti was not
particularly impressed with my request, but perhaps some
substantiation of my contribution will change her mind.

I therefore request that you write Ms. Mazzetti before
February at the below address, and inform her of what I've done
for mankind, the Byte readership, etc:

Ms. Claudia Mazzetti
Director, AAAI-86
445 Burgess Drive
Menlo Park, CA 94025-3496

Sincerely yours,

Bob Morein
Author, A.D.A. PROLOG

A.D.A PROLOG Documentation Version 1.90
for the Educational and Public Domain Versions

January 5, 1985

Copyright Robert Morein and Automata Design Associates

1570 Arran Way
Dresher, Pa. 19025



Release 1.90 of PD PROLOG now accesses all the memory in the
machine. The space for programs and stack in a PC with 640k of
memory is about 470 kbytes. This means that it is equal to our
commercial products in basic power, though without virtual memory
and the predicate extensions.

There is a very interesting natural language parsing system
by Lou Schumacher of Future Dimensions in directory "ATN". Lou's
system, known as an augmented transition network parser, is very
well documented, since the paper he wrote on the subject is also
included. You'll need Volkswriter to print it out, but it's
readable as is. The system requires ED PROLOG to run, since it
makes use of the grammar rule syntax. I challenge a user to write
a grammar rule translator to make Lou's program run with PD

Another project worth doing would be a simple editor written
in PROLOG. It would use the "read" predicate, and display
selected clauses for modification or deletion.

Simon Blackwell's "PIE" Truth Maintenance System is
presented in revised, debugged, and enlarged form. This system is
found in the directory "expert" and augments the strictly
deductive capabilities of raw Prolog with additional forms of
reasoning. PIE has a syntax that resembles colloquial English.
Wait till you see the backwards quote marks!

The predicates "batch" and "nobatch" are introduced to allow
execution of batch files without confusing messages and prompts
appearing on the screen. I've put one in Simon's "KOPS" file.

Copyright Notice

The public domain PD PROLOG system has been contributed to
the public domain for unrestricted use with one exception: the
object code may not be disassembled or modified. Electronic
bulletin boards and SIG groups are urged to aid in giving this
software the widest possible distribution.

This documentation may be reproduced freely, but it may not
be included in any other documentation without the permission of
the author.


We are pleased to present the third major version of PD PROLOG,
version 1.7. Version 1.7 continues to refine "problems" and adds
the entertaining feature of IBM PC video screen support. The
memory requirements are somewhat greater than the original, since
it is uses the large memory model. It compensates in
thoroughness. The memory requirement is about 210K bytes of TPA,
and it will benefit from any available memory.

We hope that you'll get some fun out of this PROLOG. It will
afford you exposure to THE fifth generation language at the cost
only of some intellectual effort. The motive is perfectly
explicable: We want you to think of Automata Design Associates
for fifth generation software. It also gives us a nice warm

The minimum memory requirement is 200 k of transient program
area, plus whatever space is needed to execute programs from
within PROLOG. DOS or MSDOS 2.0 are required. The program does
not require IBM PC compatibility to run, although the screen
access routines do require compatibility.

Products by Automata Design Associates

Automata Design Associates specializes in software for
artificial intelligence and robotic applications. A PROLOG
language system is available in various configurations. A LISP
interpreter will be introduced in March of 1985.

There are six versions of PROLOG available from Automata
Design Associates. The systems run under the MSDOS or PCDOS. All
of them access all available memory up to 1 megabyte.

.Public Domain PROLOG

This serves to further the general awareness of the public about
PROLOG. It also is an excellent adjunct to anyone learning the
language. Most of the core PROLOG described by Clocksin and
Mellish in the book Programming In PROLOG is implemented. A
complete IBM PC video screen support library is included in this
and all other A.D.A. prologs. Trace predicates are not. This
version is available from us for $10.00 postage paid.

.Educational PROLOG

At extremely modest cost this affords an educational institution
or individual a PROLOG system which provides the maximum
available programming area available within the 8086 small
programming model. Tracing, a debugging aid, allows monitoring
a program as it runs. User settable spy points selectively allow
this. Exhaustive tracing is also available. I/O redirection
gives some file ability.

An "exec" function allows the execution of a program or
editor from within PROLOG, thus encouraging an interactive

An "interrupt" menu is added, permitting the control of
tracing, toggling the printer, and screen printing.

Definite clause grammar support is now included.

The cost of Educational PROLOG is $29.95.


A small increment in price adds full random access file
capability. Character and structure I/O are allowed.

The "asserta and "assertz" predicates are expanded and work
with a clause indexing ability. One can assert clauses anywhere
in the database under precise pattern matching control.

A tree structured lexical scoping system and floating point
arithmetic are other enhancements.

The cost of FSM PROLOG is $49.95

.VMI PROLOG -- Virtual Memory (Replaces type VMS)

At reasonable cost the addition of virtual memory gives an
expansion of capabilities of an order of magnitude.

The database on disk is treated transparently. No special
provisions need be made by the user. Virtual and resident
databases may be mixed. A unique updating algorithim preserves
the format of the database as typed by the user while making only
those changes necessary to make it equivalent to the database in
central memory.

The cost of VMI PROLOG is $99.95

.VML PROLOG Large model Virtual Memory System

A.D.A. PROLOG is a remarkable fifth generation developement
tool for the implementation of intelligent strategies and
optimized control. It is both the kernel for applications of
virtually unlimited scope and a sophisticated developement tool
that multiplies the productivity of the programmer many times.

With a cost/performance ratio exceeding that of any other
product and a compatibility insured by compliance to the
Edinburgh syntax, performance is enhanced by numerous extensions,
many of them invisible to the user.

A quick overview of some of the features discloses:

1) Invisible compilation to a semantic network
preserves the flexibility of the interpreted mode and
the speed of a compiler.

The programmer can compile and recompile any portion
of a module at any time. The edit/compile/test cycle
is short and free of strain. An interface is provided
to an editor of choice.

2) Floating point arithmetic with a full complement
of input and output methods, transcendental and
conversion functions.

3) Virtual memory. Module size and number are
unrestricted, with a total capacity of several
hundred megabytes. Resident and virtual modules may
be co-resident. Compilation is incremental. The cache
algorithim is sophisticated. Changes made in the
database can be updated to disk by a single command.

4) A powerful exec function and acceptance of stream
input make integration into applications practical.

5) Global polymorphic variables retain information
that would otherwise require the "assertion" of

6) A quoted variable class, borrowed from LISP,
permits referencing variables as objects as well as
by value.

7) Multidimensional arrays, dynamically created and

destroyed, efficiently store numeric and nonnumeric
structures. Arrays are ideal for representing spatial
and ordinal relationships.
8) Debugging facilities let you see your program run
without any additional generation steps.

9) Totally invisible and incremental garbage

collection. There is NEVER any wait for this

10) A tree structured, dynamically configurable
lexical scoping system. The work of many programmers
can be coupled together in the form of libraries and
nested domains.

Each lexically scoped domain is a hidden space which
communicates with the parent domain via exports and
imports. Domains can be linked together under program
control to achieve any desired configuration.

11) The Grammar Rule Notation is an integral feature.

12) Keyword redefinition makes porting code easy.

The cost of this system is $200 for the MSDOS version.

.VMA PROLOG Large model Virtual Memory System

This system has additional forms of virtual memory. It was
intended for deep reasoning problems. Contact us for more

Upgrade Policy

Half the cost of any A.D.A. PROLOG system may be credited to
the purchase of a higher level version. The full cost of VMS
prolog may be applied to the purchase of VMI or VML PROLOG.
Updates to a particular level product vary from $15.00 to $35.00.

Run-time Packages

Software developers wishing to integrate an A.D.A. product
into their system should inquire about specialized run-time
packages available at reasonable cost.

Technical Information

Technical information may be obtained at (215) - 646- 4894

Perhaps we can answer the following questions in advance:

There is no support for: APPLE II, Atari, Commodore, or
CPM 80 . Other machines from these manufactures may be supported
in the future.

The MSDOS products are available on 5" and 8" diskettes.

To Place Your Order:

You may place your order at the following number:

(215)-646-4894 - day and night.


The software may be returned within 30 days of purchase for
a full refund. This applies whether or not the software has been
used. We do ask that manuals, disks and packaging be returned in
excellent condition.


You will need an IBM PC compatible machine with a minimum of
256 kbytes of memory. ED PROLOG benefits from all available
memory up to the 1 megabyte INTEL limit.

To determine the amount of TPA your machine has, run the
"chkdsk" program which is supplied with DOS. The last line reads:

XXXXXX bytes free

where XXXXXX is a six digit number.

If this number is greater than 200000, ED PROLOG will have
reduced workspace. If it's over 245000, the amount of memory is
optimal. If this is not the case, there are two possibilities:

1) The machine doesn't have enough memory.

2) Something else is removing memory from TPA, such as a co-
resident program, a ramdisk, a large dos driver, or a large
number of file or disk buffers.

If you're short of memory, make sure that no other programs,
ramdisks, or drivers besides DOS are running in the machine. You
may find it helps to eliminate (by renaming) the config.sys file
when you intend to run ED PROLOG.

How to run the Demonstration Programs
without Knowing What You're Doing

We strongly advise that you purchase the book Programming in
PROLOG by Clocksin and Mellish, publisher Springer Verlag, 1981.
For the impatient we give some advice. Type the demonstration
program you wish to run. There must be at least one entry point
within the program.

Note: Please understand that these are demonstrations programs.
Regarding user interface, they are poorly written. You will
probably have to read Clocksin and Mellish to appreciate that the
following examples of input are not equivalent: "yes." , "yes" .

The animals program - "animal"

Most of the examples require C & M for comprehension. The
program "animals", however, can be appreciated by anyone. It is a
traditional example of an expert system.

We had hoped to include the animals program on disk, but we
have found to our dismay that the version which we used is
allegedly copyrighted by the implementors of PROLOG 86. Don't be
surprised - even "happy birthday" is copyrighted. We will simply
point out that the November '84 issue of Robotics Age included a
version of the animals game, which you can, at the risk of
copyright infringement, type in. There is only one change that
need be made. The "concat" function used in that program has
arguments of the form:

concat( [atom1, atom2,...], result ).

In order to make the concat definition more closely resemble that
of "name", which is described by Clocksin and Mellish, the
argments have been reversed:

concat( result, [atom1, atom2,...] )

Assuming that you have typed in the program and made the change
just noted, the following steps are required to run it:

Run the prolog.exe file. The prompt "?-" will appear. Type
"consult( 'animals' ).". Here indicates you are to type
a carriage return. The PROLOG system will load "animals" and
compile it into an internal form. When the "?-" prompt appears
PROLOG is ready to run the "animals" guessing game. The object of
the program is to deduce the animal you are thinking of. To start
it off type "help.". PROLOG will respond by asking a
Because of the way the animals program is written, you must

respond in a rigid format. You may type "yes", "no", or
Eventually the program will terminate with either a guess as
to what animal you are thinking of, or a remark that the animal
is not within its domain of knowledge. The program has learned,
however. You may run the program again to see what effect
additional knowledge has on the program's behavior.

The program fragment "console" shows how you may improve the
console input routines of any of these programs.

The Hematology Diagnosis Program - "hemat"

Although the logical structure is not as sophisticated as
that of "animals", it is interesting for several reasons:

1) The program evaluates numerical data to arrive at a

2) Although inaccurate, it demonstrates that useful question
answering systems are not difficult to write in PROLOG.

3) There are some mistakes in the program, which only
slightly impede its usefulness.

This program uses structure input. Terminate all your
answers with a period, as in "y.", or "no.".
The starting point is "signs.". PROLOG will prompt you
for signs of anemia. The program attempts to diagnose two
varieties of a hemolytic anemia.
The program could use a good working over by a hematologist
and we would be delighted to collaborate.

Prime Number Generator - "sieve"

This program demonstrates that anything can be programed in
PROLOG if one tries hard enough. Asking the question
"primes( 50, L )." causes a list of prime numbers less than

50 to be printed out. "Sieve" is heavily recursive and quickly
exhausts the stack space of the small model interpreters.


This is an example of the use of the definite clause
grammer notation. PD PROLOG does not have this translation
facility, but ED PROLOG and all of our other versions do. It is
possible to perform the translations by hand if you have
thoroughly read C & M. Then you would have the pleasure of

?-sentence( X, [every,man,loves,a,woman], [] ).

and having the meaning elucidated as a statment in the predicate

Special Offer # 1

For some inexplicable reason, demonstration programs are
hard to come by. We are too busy writing PROLOG fill this gap. We
will reward the contribution of "cute" sample programs with the

1) A free copy of type VMI virtual memory PROLOG

2) The sample program will be published as an intact file
together with whatever comments or advertisments the author
may see fit to include, on our distribution disks.

3) Exceptional contributions may merit a copy of type VML
large model virtual memory PROLOG which now incorporates a
UNIX1 style tree structured domain system.

Special Offer # 2

If you are a hardware manufacturer and would like a PROLOG
language for your system, the solution is simple. Just send us
one of your machines! Provided your system implements a "C"
compiler, it will be ported in no time flat.

1. Trademark of AT & T.

Writing Programs For ED PROLOG

You do not type in programs at the "?-" prompt. There is no
built-in editor. The command "consult( user )" is accepted but
does not cause PROLOG to enter an editing mode. We feel that the
most universally acceptable editing method is for the user to use
a text editor of choice, which can be invoked from within PROLOG
by the "exec" predicate.

Use Wordstar or your customary editor to write a program.
Then run PD PROLOG and use the consult function to load the

In all cases except PD PROLOG, you can run your editor
without leaving PROLOG by use of the "exec" predicate.

Running the Interpreter

COMMANDS: Give commands in lower case.

Invoke PROLOG.EXE. After the "?-" prompt appears,
type "consult( )", where is the
desired database.
To exit, type "exitsys."

Don't do it except with the "assert" predicates. This is the
most frequently misunderstood aspect of A.D.A. Prolog. If
you want to enter a bunch of facts, put them in a file and
"consult" them using the "consult" predicate.

At the prompt, type ".", where
is a question as described by Clocksin and
Mellish. Be sure to terminate the question with a period.
The question may be up to 500 characters long.

The structure may be up to 500 characters in length. Be sure
to terminate with a period.

If a solution has been provided, the PROLOG interpreter will
ask "More? (Y/N):". Only if a "y" is typed will the
interpreter perform a search.

Simply type the escape key. The interpreter will
respond with "Interrrupted.", and return to the command

Type "consult()." The file name must have the
extension ".PRO". It is not necessary to include the
extension in the argument of consult. The file name as given
must not be the same as a predicate name in the file or any
file which will be loaded.

When the system displays the prompt "?-", type "trace.".
The display will likely move too rapidly for you to read. To
stop the display, type Control S. To restart the display,
type Control S. To turn the trace display off, type

"notrace" at the prompt "?-". The interrupt menu
contains additional options, such as sending all trace
output to a file, as well as display at the console.

See the section entitled "The Interrupt Menu" for a
description of the flexible options. Basically, one types
ESC to terminate a program, while Control V or Control I
interrupt a program.

The predicate "recon" is identical to the Edinburgh
predicate "reconsult."

Type "forget()."

Type "exitsys."

The system is totally interactive; any commands the operator
gives are and must be valid program statements. Statements must
terminate with a period. All commands which take a file name
also accept a path name. Any name which is not a valid PROLOG
atom (refer to C & M) must be enclosed in single quotes. Thus one
could say

consult( expert )

but one would need single quotes with

consult( 'b:\samples\subtype\expert' ).

To exit the system, type "exitsys."

Atoms may contain MSDOS pathnames if they are enclosed by single
quotes, ie., '\b:\samples\animal' .

You may consult more than one file at a time. However, all names
are public and name conflicts must be avoided. The order in which
modules are loaded may, in cases of poor program design, affect
the behavior.

Command Line Arguments

ED and PD PROLOG accept one command line argument, which is
the name of a "stream" which replaces the console for input. The
"stream" in MSDOS is a pipe or file which supplies input until
end-of-file is reached. Control then reverts back to the console.
To avoid noisy parser error messages when end-of-file is reached,
the last command in the file should be "see( user )." See Simon
Blackwell's PIE program for an example of this.

A Reference of Note

With minor exceptions, the syntax is a superset of that
described by Clocksin and Mellish in the book Programming in
Prolog by W.F. Clocksin and C.S. Mellish, published by Springer
Verlag in Berlin, Heidelberg, New York, and Tokyo. We shall refer
to this book as "C & M".

There are very few syntactical differences, mostly
unrecognized and/or minor.
When an operator is declared using the "op" statement, the
operator must be enclosed in single quotes in the "op" statement
itself, if it would not otherwise be a legal Edinburgh functor.
Subsequently, however, the parser will recognize it for what it
is, except in the "unop" statement, where it must again be
enclosed in single quotes.

Variable numbers of functor paramaters are supported.

A goal may be represented by a variable, which is less
restrictive than the C & M requirement that all goals be
functors. The variable must be instantiated to a functor when
that goal is pursued.

Rules which appear inside other expressions must be enclosed
in parenthesis if the "," operator is to be recognized as a
logical connective.

All infix operators described by C & M, and user defined
infix, prefix, and postfix operators with variable associativity
and precedence are supported exactly as in C & M.

The Built In Predicate Library

Available Operators in PD and ED PROLOG

Column 1 gives the function symbol.

Column 2 gives the precedence. The range of precedence is 1 to 255.
A zero in the precedence column indicates the symbol is parsed as
a functor, and precedence is meaningless in this case.

Column 3 gives the associativity.
A zero in the associativity column indicates the symbol is parsed
as a functor, and associativity is meaningless in this case.

Column 4 indicates which version the function is available in.
Unless otherwise noted, the function is available in all versions.
Nonstandard predicates are indicated by "NS".

op/pred precedence associativity availability

"!" 0 0
"|" 0 0
"=" 40, XFX
"==" 40, XFX
"\\=" 40, XFX
"\\==" 40, XFX
"/" 21, YFX
"@=" 40, XFX
">=" 40, XFX
"=<" 40, XFX
">" 40, XFX
"<" 40, XFX
"-" 31, YFX
"*" 21, YFX
"+" 31, YFX
"=.." 40, XFX
"-->" 255, YFY (not in PD PROLOG)
"?-" 255, FY

"arg" 0, 0,
"asserta" 0, 0,
"assertz" 0, 0,
"atom" 0, 0,
"atomic" 0, 0,
"batch" 0, 0
"clause" 0, 0,
"clearops" 0, 0,
"cls" 0, 0, NS
"concat" 0, 0,
"consult" 8, FX,
"crtgmode" 0, 0, NS
"crtset" 0, 0, NS
"curset" 0, 0, NS
"curwh" 0, 0, NS
"debugging 0, 0,
"dir" 0, 0,

"display" 0, 0,
"dotcolor" 0, 0, NS
"drawchar" 0, 0, NS
"drawdot" 0, 0, NS
"drawline" 0, 0, NS
"exec" 0, 0,
"exitsys" 0, 0, NS
"forget" 0, 0, NS
"functor" 0, 0,
"get0" 8, FX,
"integer" 0, 0,
"is" 40, XFX,
"listing" 0, 0,
"memleft" 0, 0, NS
"mod" 11, XFX,
"name" 0, 0,
"nl" 0, 0,
"nodebug" 0, 0, (not in PD PROLOG)
"nonvar" 0, 0,
"nospy" 50, FX, (not in PD PROLOG)
"not" 60 FX
"notrace" 0, 0, (not in PD PROLOG)
"op" 0, 0,
"popoff" 0, 0, NS
"popoffd" 0, 0, NS
"popon" 0, 0, NS
"popond" 0, 0, NS
"print" 0, 0,
"prtscr" 0, 0, NS
"put" 0, 0,
"ratom" 0, 0,
"read" 0, 0,
"recon" 0, 0, (Note: this is "reconsult")
"repeat" 0, 0,
"retract" 0, 0
"rnum" 0, 0,
"see" 0, 0,
"seeing" 0, 0,
"seen" 0, 0,
"skip" 0, 0, (not in PD PROLOG)
"spy" 50, FX,
"tab" 0, 0,
"tell" 0, 0,
"telling" 0, 0,
"told" 0, 0,
"trace" 0, 0, (not in PD PROLOG)
"true" 0, 0,
"unop" 0, 0,
"var" 0, 0,
"write" 0, 0,
Description of the Modifications.

I/O Redirection

I/O redirection is a feature described by Clocksin and Mellish.
The predicates "see", "seeing", "seen", "tell", "telling", and
"told" are used to select the streams used for input and output.

The predicates "seen" and "told" require as arguments the

name of the stream that is to be closed. This enables the system
to remember the indices of several streams and switch back and
forth between them.

The predicate "batch", when inserted at the beginning of a
stream file, has the following properties:

1) The normal prompt, "?-", and advisory messages do not
appear at the screen.

2) It is self cancelling if the input stream is reassigned
to the console.

3) It may also be cancelled by the predicate "batch".

call( )

The predicate as defined in C & M is obsolete. The purpose
was to permit a goal search where the goal name was a variable
instantiated to some functor name. A.D.A. permits writing of
goals with such names, so the mediation of the "call" clause is
no longer necessary.

The "call" predicate may be trivially implemented for
compatibility with the PROLOG definition

call( X ) :- X.


The function clause( X, Y ) has the new optional form
clause( X, Y, I ). If the third variable is written, it is
instantiated to the current address of a clause in memory. The
only use of the result is with succeeding assertfa and assertfz


"Debugging" prints a list of the current spypoints. After
each name a sequence of numbers may appear, indicating the number
of arguments that is a condition of the trace. The word "all"
appears if the number of arguments is not a condition of the

op( , , )

Defines the user definable grammar of a functor. The
definition conforms to that in C & M. We mention here a minor but
important point. If is not a valid PROLOG atom it must
be enclosed in single quotes when declared in the "op"
declaration. It is not necessary or legal to do this when the
functor is actually being used as an operator. In version 1.6, a
declared or built-in operator can be used either as an operator
or as a functor. For example,

+(2,3) = 2 + 3.

is a true statement.

Declared operators are annotated in the directory display
with their precedence and associativity.

Output predicates


These functions have been modified to accept multiple
arguments in the form:

print( , , ,... )

Thus, "put( a, b, c )" would result in the display of "abc".
The names of some PROLOG atoms that may occur are not
accepted by the PROLOG scanner unless surrounded by single
quotes. This only applies when such an atom is read in, not when
it is internally generated. Nevertheless, this presents us with a
problem: We would like to be capable of writing valid PROLOG
terms to a file. In some cases, it is necessary to add the single
quotes. In other cases, such as human oriented output, they are
an irritant. The modified definitions of the following predicates
are an attempt at a solution:

Operator precedence is ignored, all functors are printed
prefix and single quotes are printed if needed or they were
supplied if and when the atom was originally input.

Operator precedence is taken into account and operators are
printed according to precedence. Single quotes are printed
under the same conditions as for "display."

Operator precedence is taken into account and operators are
printed according to precedence. Single quotes are never
displayed. This is a human oriented form of output and
should never be used for writing of terms for the PROLOG


The functions "get0" and "read" have been extended to
support input from a stream other than the one currently selected
by "see". To direct output to a file or other stream, an optional
argument is used. For example, "get0( char, )" or
"get0( char, user )" would cause input to come from
or the console. If the file has not already been opened, "get0"
will fail.

Atoms enclosed by single quotest, eg. '\nthis is a new line'
can contain the escape sequences

'\n', '\r', '\t' and '\''.

If these atoms are printed by "display" or "write" they are
printed just as they are. If they are printed by the "print"
clause they are translated as follows:

'\n' results in the printing of a carriage return and a line
'\r' results in the printing of a carriage return only.
'\t' results in the printing of a tab character.
'\'' allows the printing of a single quote within a quoted atom.

The "portray" feature is not presently implemented.

Description of the New Predicates


Nullify the operator status of every operator in the

concat( ( | ), )

A list of functors or operators is concatenated into one
string, which becomes the name of a new atom to which
or must match or be instantiated.

dir( option )

Provide an alphabetized listing to the console of atoms,
constants, or open files. Without options, simply type
"dir.". Options are:

dir( p ) - list clause names only.
dir( c ) - list consulted files only.

Consulted files are prefixed by "S:".


Exit to the operating system.

forget( )
Make a database unavailable for use and reclaim the storage
it occupied.

ratom( , )-
Read an atom from the input stream, to which matches
or is instantiated. is optional. If is not
given, the input stream defaults to the standar input.
Input is terminated by a CR or LF, which are not included in
the stream.

Arithmetic Capabilities

Integer arithmetic is supported. Numbers are 32 bit signed
quantities. The following arithmetic operators are supported:

"+", "-", "*", "/", <, <=, >, >=, mod.

Arithmetic operators must never be used as goals, although they
may be part of structures. It is legal to write:

X = a + b

which results in the instantiation of X to the struture (a + b).
But the following is not legal:

alpha( X, Y ) :- X + Y, beta( Y ).

Evaluation of an arithemtic expression is mediated by the "is"
and inequality predicates. For instance, the following would be

alpha( X, Y, Z ) :- Z is X + Y.

beta( X, Y ) :- X + 2 < Y + 3.

Memory Metric Predicate

The purpose of this predicate is to give the prolog system
a sense of how much memory remains so that expensive search
strategies can be controlled. It is not possible to exactly
quantify how much memory remains. At the lowest level, there are
two types of memory - the stack and the heap. The stack expands
down from high memory, while the heap tends to expand at
unpredictable intervals upwards. If the stack and heap meet, the
prolog system must abort the search and return to the prompt.
Judicious use of the memory metric predicates reduces the
probability of this happening.

The stack is easy to quantify because it expands downwards
in a predictable way with recursion. The symbol space is a
"heap". For those interested, the structure of the heap is
determined by the C compiler under which Prolog was compiled.
There is a function internal to Prolog known as the allocator
searches the heap for enough contiguous memory to create a new
symbol. The heap resembles a piece of Swiss cheese; the holes
represent symbols and already allocated memory while the remained
is searched by the allocator for a piece of contiguous memory
large enough to satisfy a request. If one cannot be found, the
uppermost bound of the heap is expanded upwards, and that bound
is the number which we measure for an estimate of remaining

The sqace between the top of the heap, and the top of the
stack, which we call "gap", serves as a rough guide to how much
memory remains. The demands of the system are not entirely
predictable, however. For example, the creation of a new symbol
larger than "gap" would cause an abort. The user must use the
numbers supplied by these functions as a heuristic guide,
tempered by experience, to minimize the possibility of an
unexpected abort.

"Gap" is measured in 256 byte pages.

memleft( X )

If the argument is an integer, this is satisfied if the size of
"gap" is greater than "X".

If the argument is a variable, it is instantiated to the amount
of "gap" remaining.

IBM PC Video Display Predicates

A high level method is provided for drawing and displaying on the
screen of IBM PC and compatible computers.


Clear the screen and position the cursor at the upper left hand

crtgmode( X )

Matches the argument to the mode byte of the display which is
defined as follows:

mode meaning

0 40 x 25 BW (default)
1 40 x 25 COLOR
2 80 x 25 BW
3 80 x 25 COLOR
4 320 x 200 COLOR
5 320 x 200 BW
6 640 x 200 BW
7 80 x 25 monochrome display card

crtset( X )

This sets the mode of the display. The argument must be one of
the modes given above.

curset( , , )

Sets the cursor to the given row, column, and page. The arguments
must be integers.

curwh( , )

Reports the current position of the cursor. The argument must be
an integer or variable. The format is:

1) page zero is assumed.
2) The row is in the range 0 to 79, left to right.
3) The column is in the range 0 to 24, bottom to top.
dotcolor( , , )

The argument is matched to the color of the specified
dot. The monitor must be in graphics mode.

drawchar( , )

Put a character at the current cursor position with the specified
attribute. The arguments and must be
integers. Consult the IBM technical reference manual regarding

drawdot( , , )

Put a dot at the specified position. The monitor must be in the
graphics mode. The arguments must be integer. The argument
is mapped to integers by default in the following manner:

drawline( , , , , )

Draw a line on the between the coordinate pairs. The monitor must
be in the graphics mode and the arguments are integer.


Print the screen as it currently appears. Be sure that the
printer is on line and ready before invoking this predicate,
since otherwise, the system may lock up or abort.

The integer argument referred to in the above predicates
is represented as follows:


0 background background
1 green cyan
2 red magenta
3 brown white

To change the palette and the background, see the IBM Technical
Reference Bios listings for more information.

Trace Files

(type ED only)

You can now dump your trace to disk, instead of (groan)
wasting reams of printer paper. This option is described in the
next section.

The Interrupt Menu

(type ED only)

This menu has been modified. It was formerly called the
ESCAPE menu, but the meaning of the ESCAPE key has been
redefined. It is no longer necessary to display the menu to
perform one of the menu functions. This reduces the amount of
display which is lost by scrolling off the screen.

Version 1.9 offers the ability to attempt a goal from the
interrupt menu, while not disturbing the ongoing computation.

At any time while searching, printing, or accepting keyboard
input, you can break to this menu. It is generally not possible
to do this during disk access, since control passes to the
operating system at this time. Two keys cause this break to

^V: The menu is displayed and a command is accepted at the
prompt "INTERRUPT>". After a command, the menu is
redisplayed until the user selects a command which
causes an exit.

^I: The menu is not displayed. Command is accepted at the
prompt "INTERRUPT>" until the user selects a command
which causes an exit.

ESC: Typing this key causes a termination of the PROLOG
search and control returns to the user command level
with a prompt of "?-". Notice that previously, the ESC
key invoked this menu.

As the resulting menu indicates, the following functions are

A: Abort the search and return to the prompt.

O Open a trace file. The user is prompted for the file
name. The file receives all trace output. If a file is
already opened it is closed with all output preserved.

C Close the trace file if one is open. Otherwise there is
no effect.

^C: Immediately exit PROLOG without closing files. This is
not advised.

^P: Typing P toggles the printer. If the printer is
on, all input and output will also be routed to the

S: If the machine in use is an IBM PC compatible machine,
the currently displayed screen will be printed. If the
machine is not an IBM PC compatible, do not use this

T: If trace is in use, most of the trace output can be
temporarily turned off by use of this function, which is
a toggle.

G: Satisfy a goal. The system prompt will appear. Do not be
confused into thinking that the system has reverted to
the prompt. After the user has entered the goal in the

usual manner, satisfaction of the goal is immediately
attempted. This need not affect the running computation
in any way, although it is certainly possible to do so.
After the goal is attempted, control returns to the
INTERRUPT menu. The purpose is to facilitate detailed
analysis of a running system.

R: Entering another ESC causes a return to the current
activity (keyboard input or search) with no residual
effect from the interruption.

Conserving memory space

Success popping is controlled by the predicates "popond",
"popoffd", "popon", and "popoff". Success popping is means of
reclaiming storage which is used on backtracking to reconstruct
how a particular goal was satisfied. If it is obvious that there
is no alternative solution to a goal this PROLOG system is smart
enough to reclaim that storage.

In this system, succees popping is an expensive operation.
Therefore, there is a tradeoff of memory versus time. On the
other hand, discrete use of success popping can actually speed up
a program by recreating structures in a more accessible form.

The definitions of the control predicates is given in this
manual and their use is totally optional. The modulation of
success popping has no effect on program logic (read solution.)

The "cut" can save substantial time and computational
overhead as well as storage. Although the execution of the cut
costs time, you can design your program to use cuts in critical
places to avoid unnecessary backtracking. Thus the execution
speed of the program can actually increase.

Anyone who has read Clocksin and Mellish is aware, of
course, that the "cut" has a powerful logical impact which is not
always desirable.


See the below definition.


The inference engine does complete success popping for goals
which appear after "popon". Consider this example:

goal :- a, popon, b, c, popoff, d.

If no alternative solutions exist for b, then success popping
will reclaim storage by removing unnecessary records describing
how "b" was satisfied. If the Prolog system cannot rule out
possible additional solutions, success popping will never occur,
regardless of your use of "popon".
Since goal "d" occurs after "popoff", success popping will
never occur.


If no "popon" or "popoff" declarations occur in a clause, the
default action is determined by "popoffd" and "popond". If
"popoffd" has been invoked, the default is that success popping
will not occur.


The inverse of "popoffd". Turns on default success popping.
printf( , ,,... )

Prolog Tutorial


Probably you have heard of the language PROLOG within the
last year or so. You probably wondered the following things:

1) What does the name stand for? Names of computer languages are
almost always acronyms.

2) What is it good for?

3) Why now?

4) Can I get a copy to play with?

Congratulations! You obviously know the answer to the fourth
question. We now respond to the other three.

1) The name stands for "programming in logic." This we shall
elaborate on in depth later on.

2) PROLOG is good for writing question answering systems. It is
also good for writing programs that perform complicated
strategies that compute the best or worst way to accomplish a
task, or avoid an undesirable result.

3) PROLOG was virtually unknown in this country until researchers
in Japan announced that it was to be the core language of that
country's fifth generation computer project. This is the project
with which Japan hopes to achieve a domainant position in the
world information industry of the 1990's.

PROLOG is one of the most unusual computer languages ever
invented. It cannot be compared to FORTRAN, PASCAL, "C", or
BASIC. The facilities complement, rather than replace those of
conventional languages. Although it has great potential for
database work, it has nothing in common with the database
languages used on microcomputers.

Perhaps the best point to make is that while conventional
languages are prescriptive, PROLOG is descriptive. A statement in
a conventional language might read:

if( car_wheels = TRUE ) then
(some sort of procedure)
X = X + 1;

A statment in PROLOG could just be a statment of fact about cars
and wheels. There are many relationships that hold. For instance,

has( car, wheels ).

has( car, quant(wheels, four) ).

round( wheels ).

Each of these statments is an independent fact relating cars,
wheels, and the characteristics of wheels. Because they are
independent, they can be put into a PROLOG program by programmers
working separately. The man who is a specialist on car bodies can
say his thing, the wheel specialist can have his say, and the
participants can work with relative independence. And this brings
to light a major advantage of PROLOG:


With conventional programming languages projects can still be
"chunked", or divided between programmers. But efficiency on a
team project drops drastically below that of the individual
programmer wrapped up in his own trance. As the number of
participants grows the need for communication grows
geometrically. The time spent communicating can exceed that spent
Although PROLOG does not eliminate the need for
task coordination, the problem is considerably simplified. It
also provides the ability to answer questions in a "ready to eat
form." Consider your favorite BASIC interpreter. Based upon the
statements about cars and wheels previously given, could you ask
it the following question?

has( car, X ), round( X ).

Does a car have anything which is round? The question
instructs the PROLOG interpreter to consider all the objects that
it knows are possessed by a car and find those which are round.
Perhaps you are beginning to guess that PROLOG has the abilities
of a smart database searcher. It can not only find the facts but
selectively find them and interpret them.

Consider the problem of a fault tree, as exemplified by this
abbreviated one:

{Car won't start}
[Engine turns over](No) --> [Battery voltage](no)-\
(Yes) v
| {Check battery}
[Smell gasoline](yes) --> {Try full throttle cranking}
| (failure)
/--------/ |

(details omitted)

The fault tree is easily programmed in BASIC. Later we shall
show that PROLOG supplies a superior replacement for the fault
tree. Though the tree is capable of diagnosing only the problem
for which it was designed, PROLOG dynamically constructs the
appropriate tree from facts and rules you have provided. PROLOG
is not limited to answering one specific question. Given enough
information, it will attempt to find all deductive solutions to
any problem.


I. Rules and Facts

This is where you should start if you know nothing about
PROLOG. Let us consider a simple statment in PROLOG, such as:

1) has( car, wheels ).

This statement is a "fact. The word "has" in this statment is
known either as a functor or predicate. It is a name for the
relationship within the parenthesis. It implies that a car has
wheels. But the order of the words inside the bracket is
arbitrary and established by you. You could just as easily say:

2) has( wheels, car ).

and if you wrote this way consistently, all would be well. The
words has, wheels, and car are all PROLOG atoms. "Wheels" and
"car" are constants.

A database of facts can illustrate the data retrieval
capabilities of PROLOG. For instance:

3) has( car, wheels ).
has( car, frame ).
has( car, windshield ).
has( car, engine ).

You could then ask PROLOG the question:

4) has( car, Part ).

The capital "P" of Part means that Part is a variable. PROLOG
will make Part equal to whatever constant is required to make the
question match one of the facts in the database. Thus PROLOG will

Part = wheels.


If you type "y" the next answer will appear:

Part = frame.


If you continue, PROLOG will produce the answers Part = windshield
and Part = engine. Finally, you will see:



indicating that PROLOG has exhausted the database. Incidentally,
when a variable is set equal to a constant or other variable,

it is said to be instantiated to that object.

Notice that PROLOG searches the database forwards and in
this case, from the beginning. The forward search path is built
into PROLOG and cannot be changed. An author of a program written
in a prescriptive language is quite conscious of the order of
execution of his program, while in PROLOG it is not directly
under his control.

The other major element is the rule which is a fact which is
conditionally true. In logic this is called a Horn clause:

5) has( X, wheels ) :- iscar( X ).

The fact iscar( car ) and the above rule are equivalent to

6) has( car, wheels).

The symbol :- is the "rule sign." The expression on the left of
:-is the "head" and on the right is the body. The variable X has
scope of the rule, which means that it has meaning only within
the rule. For instance, we could have two rules in the database
using identically named variables.

7) has( X, transportation ) :-
has( X, car ), has( license, X ).

8) has( X, elephant ) :- istrainer( X ), hasjob( X ).

The variables X in the two expressions are completely distinct
and have nothing to do with each other.

The comma between has( X, car ) and has( license, X ) means "and"
or logical conjuction. The rule will not be true unless both the
clauses has(X, car) and has( license, X ) are true.

On the other hand if there is a rule

9) has( license, X ) :- passedexam( X ).

consider what PROLOG will do in response to the question:

10) has( harry, transportation ).

(Notice that harry has not been capitalized because we do not
want it taken as a variable. We could, however, say 'Harry'
enclosed in single quotes.)

It will scan the database and use (7), in which X will be
instantiated to harry. The rule generates two new questions:

11) has( harry, car ).

12) has( license, harry ).

Assuming that harry has a car, the first clause of (7) is
satisfied and the database is scanned for a match to (12). PROLOG

picks up rule (9) in which X is instantiated to harry and the
question is now posed:

13) passedexam( harry ).

If there is a fact:

passedexam( harry ).

in the database then all is well and harry has transportation.
If there is not, then PROLOG will succinctly tell you:


But suppose Harry has money and can hire a chauffer as any good
programmer can. That could be made part of the program in the
following way.

The rule which PROLOG tried to use was:

14) has( X, transportation ) :-
has( X, car ), has( license, X ).

At any point following it there could be included another rule:

15) has( X, transportation ) :- has( X, money ).

or simply the bald fact:

16) has( harry, transportation ).

These additional rules or facts would be used in two
circumstances. If at any point a rule does not yield a solution,
PROLOG will scan forward from that rule to find another
applicable one. This process is known as "backtracking search"
and can be quite time consuming.

If in response to the "More?" prompt you answer "y" PROLOG will
search for an additional distinct solution. It will attempt to
find an alternate rule or fact for the last rule or fact used. If
that fails, it will back up to the antecedent rule and try to

find an alternate antecedent. And it will continue to back up
until it arrives at the question you asked, at which point it
will say:


"Antecedent" to a rule means that it gave rise to its' use. For
example, (7) is the antecedent of (9) in the context of the
question (16).

II. Grammar

It is a boring subject, but it must be discussed. All PROLOG
statements are composed of valid terms, possibly a rule sign (":-
"), commas representing conjunction ("and"), and a period at the

very end.
A term is a structure, constant, variable, or number.

What is a structure? It is a kind of grouping:

1) Structures consist of a functor, and a set of objects or
structures in parenthesis.

2) Objects are constants, variables, numbers, or lists,
which we have not discussed yet.

3) A constant or functor must be a string of letters and
numbers, beginning with a lower case letter, unless
you choose to enclose it in single quotes ( 'howdy
pardner' ), in which case you are freed from these
4) A variable must be a string of letters and numbers
beginning with a capital letter.

5) A functor may optionally have arguments enclosed in
parenthesis , as in: hascar( X ) or hascar.

An example: "has( X, transportation )." is a structure.

III. Input / Output

You now know enough to write simple databases and
interrogate them profitably. But before we examine more
sophisticated examples, it will be necessary to add input and
output to the language. There are built in functions which appear
as rules which are satisfied once. Thus the statment:

write( 'Hello world.' ).

can be included on the right side of a rule:

greetings( X ) :- ishuman( X ), write( 'Hello world.' ). You
can also write "write( X )" where X is some variable. Note that
'Hello world.' is not enclosed in double quotes. Single quotes,
which denote a constant, are required. Double quotes would denote
a list, which is another thing entirely.

Provided that a match to "ishuman" can be found, the builtin
function "write" is executed and the message printed to the
The builtin read( X ) reads a "structure" that you input
from the keyboard. More formally, we have

read( or )
write( or )

If you write read( Input ), then the variable "keyboard" will be
assigned to whatever is typed at the keyboard, provided that the
input is a valid PROLOG structure. The builtin "read" will fail
if instead of Keyboard we wrote read( baloney ), where "baloney"
is a constant, and the user at the keyboard did not type exactly

When you input a structure in response to a "read" statement, be
sure to end it with a period and an .

There is a convenient way of putting the cursor on a new
line. This is the builtin "nl". For example:

showme :- write( 'line 1' ), nl, write( 'line 2' ).

would result in:

line 1
line 2

There is also a primitive form of input/output for single
characters. It will be discussed later.

IV. A Fault Tree Example

Consider the "won't start" fault tree for an automobile:

{Car won't start}
[Engine turns over](No) --> [Battery voltage](no)-\
(Yes) v
| {Check battery}
[Smell gasoline](yes) --> {Try full throttle cranking}
| (failure)
/--------/ |
| /------------------------/
| |
| |
| [Check for fuel line leaks](yes)-->{Replace fuel line}
| (no)
| |
| |
| [Check for defective carburator](yes)--\
| (no) v
| {Repair carburator}
[Is spark present](no)-->[Do points open and close](no)-\
| (yes) v
/----/ | {Adjust points}
| /------------------------/
| |
| [Pull distributor wire, observe spark](blue)--\
| (orange) v
| | {Check plug wires & cap}
| |
| [Measure voltage on coil primary](not 12V)--\
| (12V) v
| | {Check wiring, ballast resistor}
| |
| [Check condenser with ohmmeter](conducts)--\
| (no conduction) v
| | {Replace condenser}
| |
| [Open and close points](voltage not 0 - 12)--\
| (voltage swings 0 - 12) v
| | {Fix primary circuit}
| |
| {Consider hidden fault, swap components]
\-------{Call a tow truck!!}

A PROLOG program to implement this is simple. Each statment
represents a decision point fragment of the tree. The PROLOG
interpreter dynamically assembles the tree as it attempts a

'car wont start' :- write( 'Is the battery voltage low?' ),
affirm, nl,
write( 'Check battery' ).

'car wont start' :- write( 'Smell gasoline?' ),
affirm, nl,
'fuel system'.

'fuel system' :- write( 'Try full throttle cranking' ).

'fuel system' :- write( 'Are there fuel line leaks?' ),
affirm, nl,
write( 'Replace fuel line.' ).

'fuel system' :- write( 'Check carburator' ).

'car wont start' :- write( 'Is spark present?' ),
not( affirm ), nl,
'no spark'.

'no spark' :- write( 'Do points open and close?' ),
not( affirm ), nl,
write( 'Adjust or replace points.' ).

'no spark' :- write( 'Is the spark off the coil good?' ),
write( 'Check plug wires and cap.' ).

'no spark' :- write( 'What is the voltage on the primary
of the coil: ' ),
read( Volts ),
Volts < 10,
write('Check wiring and ballast resistor.').

'no spark' :- write( 'Does the capacitor leak?' ),
write( 'Replace the capacitor.' ).

'no spark' :- not( 'primary circuit' ).

'primary circuit'
:- write( 'Open the points. Voltage across
coil?:'), nl,
read( Openvolts ), Openvolts < 1,
write( 'Close the points. Voltage across
read( Closevolts ), Closevolts > 10, nl,
write( 'Primary circuit is OK.' ).

'no spark' :- write( 'Consider a hidden fault. Swap
cap, rotor,points,capacitor.' ).

'Car wont start' :- write( 'Get a tow truck!!' ).

--End program--

The above is a simple example of an expert system. A
sophisticated system would tell you exactly the method by which
it has reached a conclusion. It would communicate by a "shell"
program written in PROLOG which would accept a wider range of
input than the "valid structure" required by the PROLOG
interpreter directly.

V. Lists

Consider a shopping list given you by your wife. It is a
piece of paper with items written on it in an order that probably
symbolizes their importance. At the top it may say EGGS!!!,
followed by carrots, hamburger, and finally a flea collar for the
dog, if you can find one. In PROLOG such a list would be written:

1) [eggs, carrots, hamburger, fleacollar]

The order of a list is important so that eggs and carrots cannot
be reversed and PROLOG be uncaring.

Let us put the list in a structure:

shopping( [eggs, carrots, hamburger, fleacollar] ).

Then if you wished to isolate the head of the list you could ask
the question:

shopping( [ Mostimportant | Rest ] ).

and PROLOG would respond:

Mostimportant = eggs,
Rest = [carrots, hamburger, fleacollar].

The vertical bar "|" is crucial here. It is the string extraction
operator, which performs a combination of the CDR and CAR
functions of LISP. When it appears in the context [X|Y] it can
separate the head of the list from the rest, or tail.

You may have gained the impression that PROLOG is a rather
static language capable of answering simple questions, but it is
far more powerful than that. The string extraction operator is
the key. It permits PROLOG to whittle a complex expression down
to the bare remainder. If the rules you have given it permit it
to whittle the remainder down to nothing, then success is
achieved. An example of this is the definition of "append."

Let us suppose you have not yet done yesterday's shopping,
let alone today's. You pull it out of your wallet and sootch tape
it to the list your wife just gave you. Yesterday's list was:

[tomatoes, onions, ketchup]

Combined with [eggs, carrots, hamburger, fleacollar] we obtain


To take one list and to attach it to the tail of another list is
to "append" the first to the second. The PROLOG definition of
append is:

Rule1: append( [], L, L ).

Rule2: append( [X|List1], List2, [X|List3] ) :-

append( List1, List2, List3 ].

The general scheme is this:

The definition consists of one rule and one fact. The rule will
be used over and over again until what little is left matches the
fact. The [] stands for empty list, which is like a bag without
anything in it. This is an example of a recursive definition.
Suppose we ask:

append( [a,b,c], [d,e,f], Whatgives ).

1. Rule 2 is invoked with arguments ( [a,b,c], [d,e,f], Whatgives ).
2. Rule 2 is invoked again with arguments:
( [b,c], [d,e,f], List3 ).
3. Rule 2 is invoked again with arguments:
( [b], [d,e,f], List3 ).
4. The arguments are now ([], [d,e,f], List3 ). Rule 1 now
matches. End.

How does this cause a list to be constructed? The key is to watch
the third argument. Supplied by the user, it is named
"Whatgives". The inference engine matches it to [X|List3] in rule
2. Now lets trace this as rule two is successivly invoked:

Rule2: [X|List3] (List1 = [b,c])
| \
| \
| \
v \
Rule2: a [X'|List3'] (List1' = [c])
| \
| \
| \
v \
Rule2: b [X''|List3''] (List1'' = [], ie., empty set.)
| \
| \
| \
Rule1: c L ( in Rule1 = [d,e,f] )


L in rule 1 is [d,e,f] for the following reason: Notice that rule
2 never alters List2. It supplies it to whatever rule it invokes.
So L in rule 1 is the original List2, or [a,b,c].

This example would not have worked if the order of rules one
and two were reversed. The PROLOG inference engine always
attempts to use the the first rule encountered. You could imagine
it as always reading your program from the top down in attempting
to find an appropriate rule. Since rule 2 would always satisfy,
an unpleasant thing would have happened if the order were

reversed. The program would loop forever.

I hope that this tiny introduction to PROLOG whets your
appetite. You should now purchase the book

Programming In Prolog
W.F. Clocksin and C.S. Mellish
Springer - Verlag
Berlin,Heidelberg,New York. 1981,1984

 December 28, 2017  Add comments

Leave a Reply