<\input texinfo @c -*- mode: texinfo; coding: latin-1; -*-

@c %**start of header
@setfilename yap.info
@setcontentsaftertitlepage
@settitle YAP Prolog User's Manual
@c For double-sided printing, uncomment:
@c @setchapternewpage odd
@c %**end of header

@set VERSION 4.3.21
@set EDITION 4.1.1
@set UPDATED January 2002

@c Index for C-Prolog compatible predicate
@defindex cy
@c Index for predicates not in C-Prolog
@defindex cn
@c Index for predicates sort of (almost) in C-Prolog
@defindex ca

@c Index for SICStus Prolog compatible predicate
@defindex sy
@c Index for predicates not in SICStus Prolog
@defindex sn
@c Index for predicates sort of (almost) in SICStus Prolog
@defindex sa


@setchapternewpage odd
@c @smallbook
@comment %** end of header

@ifinfo
@format
START-INFO-DIR-ENTRY
* Yap: (yap).           YAP Prolog User's Manual.
END-INFO-DIR-ENTRY
@end format
@end ifinfo

@titlepage
@title YAP User's Manual
@subtitle Version @value{VERSION}
@author V@'{@dotless{i}}tor Santos Costa,
@author Lu@'{@dotless{i}}s Damas,
@author Rog@'erio Reis, and
@author R@'uben Azevedo
@page
@vskip 2pc
Copyright @copyright{} 1989-2000 L. Damas, V. Santos Costa and Universidade
do Porto.

Permission is granted to make and distribute verbatim copies of
this manual provided the copyright notice and this permission notice
are preserved on all copies.

Permission is granted to copy and distribute modified versions of this
manual under the conditions for verbatim copying, provided that the entire
resulting derived work is distributed under the terms of a permission
notice identical to this one.

Permission is granted to copy and distribute translations of this manual
into another language, under the above conditions for modified versions.

@end titlepage

@ifinfo
@node Top, , , (dir)
@top YAP Prolog

This file documents the YAP Prolog System version @value{VERSION}, a
high-performance Prolog compiler developed at LIACC, Universidade do
Porto. YAP is based on David H. D. Warren's WAM (Warren Abstract
Machine), with several optimizations for better performance. YAP follows
the Edinburgh tradition, and is largely compatible with DEC-10 Prolog,
Quintus Prolog, and especially with C-Prolog.

This file contains the CLP(Q,R) manual as distributed by the Austrian
Research Institute for Artificial Intelligence (OFAI). Permission on
this document follows the following license:

Copyright @copyright{} 1992,1993,1994,1995 OFAI Austrian Research
Institute for Artificial Intelligence (OFAI) Schottengasse 3 A-1010
Vienna, Austria

Permission is granted to make and distribute verbatim copies of this
manual provided the copyright notice and this permission notice are
preserved on all copies.

Permission is granted to copy and distribute modified versions of this
manual under the conditions for verbatim copying, provided that the
entire resulting derived work is distributed under the terms of a
permission notice identical to this one.

Permission is granted qto copy and distribute translations of this
manual into another language, under the above conditions for modified
versions, except that this permission notice may be stated in a
translation approved by the OFAI.

This file contains a chapter on CHR. This package is distributed under
license from LMU (Ludwig-Maximilians-University), Munich, Germany:

Permission is granted to copy and distribute modified versions of this
chapter under the conditions for verbatim copying, provided that the entire
resulting derived work is distributed under the terms of a permission
notice identical to this one.

Permission is granted to copy and distribute translations of this chapter
into another language, under the above conditions for modified versions,
except that this permission notice may be stated in a translation approved
by LMU.

Copyright @copyright{} 1996-98 LMU (Ludwig-Maximilians-University)@*
Munich, Germany
@menu
* Intro:: Introduction
* Install:: Installation
* Run:: Running YAP
* Syntax:: The syntax of YAP
* Loading Programs:: Loading Prolog programs
* Modules:: Using Modules in YAP
* Builtins:: Built In Predicates
* Library:: Library Predicates
* Extensions:: Extensions to Standard YAP
* Rational Trees:: Working with Rational Trees
* Coroutining:: Changing the Execution of Goals
* Attributed Variables:: Using attributed Variables
* CLPQR:: The CLP(Q,R) System
* CHR:: The CHR System
* Logtalk:: The Logtalk Object-Oriented System
* Parallelism:: Running in Or-Parallel
* Tabling:: Storing Intermediate Solutions of programs 
* Low Level Profiling:: Profiling Abstract Machine Instructions
* Low Level Tracing:: Tracing at Abstract Machine Level
* Debugging:: Using the Debugger
* Efficiency:: Efficiency Considerations
* C-Interface:: Interfacing predicates written in C
* YapLibrary:: Using YAP as a library in other programs
* Compatibility:: Compatibility with other Prolog systems
* Predicate Index:: An item for each predicate
* Concept Index:: An item for each concept

Built In Predicates
* Control:: Controlling the execution of Prolog programs
* Undefined Procedures:: Handling calls to Undefined Procedures
* Testing Terms:: Predicates on Terms
* Comparing Terms:: Comparison of Terms
* Arithmetic:: Arithmetic in YAP
* I/O:: Input/Output with YAP
* Database:: Modifying Prolog's Database
* Sets:: Finding All Possible Solutions
* Grammars:: Grammar Rules
* Preds:: Predicate Information
* OS:: Access to Operating System Functionality
* Term Modification:: Updating Prolog Terms
* Profiling:: Profiling Prolog Execution
* Arrays:: Supporting Global and Local Arrays
* Preds:: Information on Predicates
* Misc:: Miscellaneous Predicates


Subnodes of Running
* Running Yap Interactively:: Interacting with Yap
* Running Prolog Files:: Running Prolog files as scripts

Subnodes of Syntax
* Formal Syntax:: Syntax of Terms
* Tokens:: Syntax of Prolog tokens

Subnodes of Tokens
* Numbers:: Integer and Floating-Point Numbers
* Strings:: Sequences of Characters
* Atoms:: Atomic Constants
* Variables:: Logical Variables
* Punctuation Tokens:: Tokens that separate other tokens
* Layout:: Comments and Other Layout Rules

Subnodes of Numbers
* Integers:: How Integers are read and represented
* Floats:: Floating Point Numbers

Subnodes of Loading Programs
* Compiling:: Program Loading and Updating
* Setting the Compiler:: Changing the compiler's parameters
* Saving:: Saving and Restoring Programs

Subnodes of Modules
* Module Concepts:: The Key Ideas in Modules
* Defining Modules:: How To Define a New Module
* Using Modules:: How to Use a Module
* Meta-Predicates in Modules:: How to Handle New Meta-Predicates

Subnodes of Input/Output
* Streams and Files:: Handling Streams and Files
* C-Prolog File Handling:: C-Prolog Compatible File Handling
* I/O of Terms:: Input/Output of terms
* I/O of Characters:: Input/Output of Characters
* I/O for Streams:: Input/Output using Streams
* C-Prolog to Terminal:: C-Prolog compatible Character I/O to terminal
* I/O Control:: Controlling your Input/Output
* Sockets:: Using Sockets from YAP

Subnodes of Database
* Modifying the Database:: Asserting and Retracting
* Looking at the Database:: Finding out what is in the Data Base
* Database References:: Using Data Base References
* Internal Database:: YAP's Internal Database
* BlackBoard:: Storing and Fetching Terms in the BlackBoard

Subnodes of Library
* Apply Macros:: Apply a Predicate to a list or to sub-terms.
* Association Lists:: Binary Tree Implementation of Association Lists.
* AVL Trees:: Predicates to add and lookup balanced binary  trees.
* Heaps:: Labelled binary tree where the key of each node is less
    than or equal to the keys of its children.
* Lists:: List Manipulation
* Ordered Sets:: Ordered Set Manipulation
* Pseudo Random:: Pseudo Random Numbers
* Queues:: Queue Manipulation
* Random:: Random Numbers
* RegExp:: Regular Expression Manipulation
* Splay Trees:: Splay Trees
* String I/O:: Writing To and Reading From Strings
* System:: System Utilities
* Terms:: Utilities on Terms
* Timeout:: Call With Timeout
* Trees:: Updatable Binary Trees
* UGraphs:: Unweighted Graphs

Subnodes of Debugging
* Deb Preds:: Debugging Predicates
* Deb Interaction:: Interacting with the debugger

Subnodes of Compatibility
* C-Prolog:: Compatibility with the C-Prolog interpreter
* SICStus Prolog:: Compatibility with the Quintus and SICStus Prolog systems
* ISO Prolog::  Compatibility with the ISO Prolog standard

Subnodes of Attributes
* Attribute Declarations:: Declaring New Attributes
* Attribute Manipulation:: Setting and Reading Attributes
* Attributed Unification:: Tuning the Unification Algorithm
* Displaying Attributes:: Displaying Attributes in User-Readable Form
* Projecting Attributes:: Obtaining the Attributes of Interest
* Attribute Examples:: Two Simple Examples of how to use Attributes.

Subnodes of CLP(Q,R)
* Introduction to CLPQR:: The CLP(Q,R) System
* Referencing CLPQR:: How to Reference CLP(Q,R)
* CLPQR Acknowledgments:: Acknowledgments for CLP(Q,R)
* Solver Interface:: Using the CLP(Q,R) System
* Notational Conventions:: The CLP(Q,R) Notation
* Solver Predicates:: The CLP(Q,R) Interface Predicates
* Unification:: Unification and CLP(Q,R)
* Feedback and Bindings:: Information flow in CLP(Q,R)
* Linearity and Nonlinear Residues:: Linear and Nonlinear Constraints
* How Nonlinear Residues are made to disappear:: Handling Nonlinear Residues
* Isolation Axioms:: Isolating the Variable to be Solved
* Numerical Precision and Rationals:: Reals and Rationals
* Projection and Redundancy Elimination:: Presenting Bindings for Query Variables
* Variable Ordering:: Linear Relationships between Variables
* Turning Answers into Terms:: using @code{call_residue/2}
* Projecting Inequalities:: How to project linear inequations
* Why Disequations:: Using Disequations in CLP(Q,R)
* Syntactic Sugar:: An easier syntax
* Monash Examples:: The Monash Library
* Compatibility Notes:: CLP(Q,R) and the clp(R) interpreter
* A Mixed Integer Linear Optimization Example:: MIP models
* Implementation Architecture:: CLP(Q,R) Components
* Fragments and Bits:: Final Last Words on CLP(Q,R)
* CLPQR Bugs:: Bugs in CLP(Q,R)
* CLPQR References:: References for CLP(Q,R)

Subnodes of CHR
* CHR Copyright::               
* CHR Introduction::            
* CHR Introductory Examples::   
* CHR Library::                 
* CHR Debugging::               
* CHR Programming Hints::       
* CHR Constraint Handlers::     
* CHR Backward Compatibility::  

Subnodes of C-Interface
* Manipulating Terms:: Primitives available to the C programmer
* Unifying Terms:: How to Unify Two Prolog Terms
* Manipulating Strings:: From character arrays to Lists of codes and back
* Memory Allocation:: Stealing Memory From Yap
* Controlling Streams:: Control How Yap sees Streams
* Calling Yap From C:: From C to Yap to C to Yap 
* Writing C:: Writing Predicates in C
* Loading Objects:: Loading Object Files
* Sav&Rest:: Saving and Restoring
* Yap4 Notes:: Changes in Foreign Predicates Interface

Subnodes of C-Prolog
* Major Differences with C-Prolog:: Major Differences between YAP and C-Prolog
* Fully C-Prolog Compatible:: Yap predicates fully compatible with
C-Prolog
* Not Strictly C-Prolog Compatible:: Yap predicates not strictly as C-Prolog
* Not in C-Prolog:: Yap predicates not available in C-Prolog
* Not in YAP:: C-Prolog predicates not available in YAP

Subnodes of SICStus Prolog
* Major Differences with SICStus:: Major Differences between YAP and SICStus Prolog
* Fully SICStus Compatible:: Yap predicates fully compatible with
SICStus Prolog
* Not Strictly SICStus Compatible:: Yap predicates not strictly as
SICStus Prolog
* Not in SICstus Prolog:: Yap predicates not available in SICStus Prolog


Tables
* Operators:: Predefined operators

@end menu

@end ifinfo


@node Intro, Install, , Top
@unnumbered Introduction

This document provides User information on version @value{VERSION} of
YAP (@emph{yet another prolog}). The YAP Prolog System is a
high-performance Prolog compiler developed at LIACC, Universidade do
Porto. YAP provides several important features:

@itemize @bullet
 @item Speed: YAP is widely considered one of the fastest available Prolog
systems.

 @item Functionality: it supports stream I/O, sockets, modules,
exceptions, Prolog debugger, C-interface, dynamic code, internal
database, DCGs, saved states, co-routining, arrays.

 @item We explicitly allow both commercial and non-commercial use of YAP.
@end itemize

YAP is based on the David H. D. Warren's WAM (Warren Abstract Machine),
with several optimizations for better performance. YAP follows the
Edinburgh tradition, and was originally designed to be largely
compatible with DEC-10 Prolog, Quintus Prolog, and especially with
C-Prolog.

YAP implements most of the ISO-Prolog standard. We are striving at
full compatibility, and the manual describes what is still
missing. The manual also includes a (largely incomplete) comparison
with SICStus Prolog.

The document is intended neither as an introduction to Prolog nor to the
implementation aspects of the compiler. A good introduction to
programming in Prolog is the book @cite{The Art of Prolog}, by
L. Sterling and E. Shapiro, published by "The MIT Press, Cambridge
MA". Other references should include the classical @cite{Programming in
Prolog}, by W.F. Clocksin and C.S. Mellish, published by
Springer-Verlag.

YAP 4.3 is known to build with many versions of gcc (<= gcc-2.7.2, >=
gcc-2.8.1, >= egcs-1.0.1, gcc-2.95.*) and on a variety of Unixen:
SunOS 4.1, Solaris 2.*, Irix 5.2, HP-UX 10, Dec Alpha Unix, Linux 1.2
and Linux 2.* (RedHat 4.0 thru 5.2, Debian 2.*) in both the x86 and
alpha platforms. It has been built on Windows NT 4.0 using Cygwin from
Cygnus Solutions (see README.nt) and using Visual C++ 6.0.

The overall copyright and permission notice for YAP4.3 can be found in
the Artistic file in this directory. YAP follows the Perl Artistic
license, and it is thus non-copylefted freeware.

If you have a question about this software, desire to add code, found a
bug, want to request a feature, or wonder how to get further assistance,
please send e-mail to @email{yappers@@ncc.up.pt}.  To subscribe to the
mailing list, send a request to @email{majordomo@@ncc.up.pt} with body
"subscribe yappers".

Online documentation is available for YAP at:

        @url{http://www.ncc.up.pt/~vsc/Yap/}

Recent versions of Yap, including both source and selected binaries,
can be found from this same URL.

This manual was written by V@'{@dotless{i}}tor Santos Costa,
Lu@'{@dotless{i}}s Damas, Rog@'erio Reis, and R@'uben Azevedo. The
manual is largely based on the DECsystem-10 Prolog User's Manual by
D.L. Bowen, L. Byrd, F. C. N. Pereira, L. M. Pereira, and
D. H. D. Warren. We have also used comments from the Edinburgh Prolog
library written by R. O'Keefe. We would also like to gratefully
acknowledge the contributions from Ashwin Srinivasian.

We are happy to include in YAP several excellent packages developed
under separate licenses. Our thanks to the authors for their kind
authorisation to include these packages.

The packages are, in alphabetical order:

@itemize @bullet
@item The CHR package developed at TUM by
Ludwig-Maximilians-Universitaet Muenchen (LMU) by Dr. Fruehwirth Thom
and by Dr. Christian Holzbaur. The package is distributed under license
from LMU (Ludwig-Maximilians-University), Munich, Germany:

Permission is granted to copy and distribute modified versions of this
chapter under the conditions for verbatim copying, provided that the entire
resulting derived work is distributed under the terms of a permission
notice identical to this one.

Permission is granted to copy and distribute translations of this chapter
into another language, under the above conditions for modified versions,
except that this permission notice may be stated in a translation approved
by LMU.

Copyright @copyright{} 1996-98 LMU (Ludwig-Maximilians-University)@*
Munich, Germany

@item The CLP(Q,R) package developed at OFAI Austrian Research
Institute for Artificial Intelligence by Christian Holzbaur. The package
is distributed under the OFAI license. Documentation on this package is
a chapter of this manual, which is covered by the OFAI license:

Copyright @copyright{} 1992,1993,1994,1995 OFAI Austrian Research
Institute for Artificial Intelligence (OFAI) Schottengasse 3 A-1010
Vienna, Austria

Permission is granted to make and distribute verbatim copies of this
manual provided the copyright notice and this permission notice are
preserved on all copies.

Permission is granted to copy and distribute modified versions of this
manual under the conditions for verbatim copying, provided that the
entire resulting derived work is distributed under the terms of a
permission notice identical to this one.

Permission is granted qto copy and distribute translations of this
manual into another language, under the above conditions for modified
versions, except that this permission notice may be stated in a
translation approved by the OFAI.

@item The Logtalk Object-Oriented system is developed at the University 
of Beira Interior, Portugal, by Paulo Moura.
The package is distributed under the Perl Artistic License.
Instructions about loading this package are included in this document.
The documentation on this package is distributed separately from yap.tex.

Copyright @copyright{} 1998-2001 Paulo Moura

@item The Pillow WEB library developed at Universidad Politecnica de
Madrid by the CLIP group. This package is distributed under the FSF's
LGPL. Documentation on this package is distributed separately from
yap.tex.

@item The yap2swi library implements some of the functionality of
SWI's PL interface. Please do refer to the SWI-Prolog home page:

@code{http://www.swi-prolog.org}

for more information on SWI-Prolog and for a detailed description of its
foreign interface.

@end itemize

@node Install, Run, Intro, Top
@chapter Installing YAP
@cindex installation


@menu
* Configuration Options:: Tuning the Functionality of YAP Machine
* Machine Options:: Tuning YAP for a Particular Machine and Compiler
@end menu

To compile YAP it should be sufficient to:

@enumerate 
@item @code{mkdir ARCH}.

@item @code{cd ARCH}.

@item @code{../configure ...options...}. 

Notice that by default @code{configure} gives you a vanilla
configuration. For instance, in order to use coroutining and/or CLP
you need to do

@example
../configure --enable-coroutining ...options...
@end example
Please @pxref{Configuration Options} for extra options.

@item check the Makefile for any extensions or changes you want to
make.

YAP uses @code{autoconf}. Recent versions of Yap try to follow GNU
conventions on where to place software.

@itemize @bullet
@item The main executable is placed at @code{BINDIR}. This executable is
actually a script that calls the Prolog engine, stored at @code{LIBDIR}.

@item @code{LIBDIR} is the directory where libraries are stored. YAPLIBDIR is a
subsdirectory that contains the Prolog engine and a Prolog library.

@item @code{INCLUDEDIR} is used if you want to use Yap as a library.

@item @code{INFODIR} is where to store @code{info} files. Usually
@code{/usr/local/info}, @code{/usr/info}, or @code{/usr/share/info}.
@end itemize

@item @code{make}.

@item If the compilation succeeds, try @code{./yap}.

@item If you feel satisfied with the result, do @code{make install}.

@item @code{make install-info} will create the info files in the
standard info directory.

@item @code{make html} will create documentation in html format in the
predefined directory.

In most systems you will need to be superuser in order to do @code{make
install} and @code{make info} on the standard directories.
@end enumerate

@node Configuration Options, Machine Options, ,Install
@section Tuning the Functionality of YAP 
@cindex syntax

Compiling Yap with the standard options give you a plain vanilla
Prolog. You can tune Yap to include extra functionality by calling
@code{configure} with the appropriate options:

@itemize @bullet
 @item @code{--enable-rational-trees=yes} gives you support for infinite
rational trees.

 @item @code{--enable-coroutining=yes} gives you support for coroutining,
including freezing of goals, attributed variables, and
constraints. This will also enable support for infinite rational
trees.

 @item @code{--enable-depth-limit=yes} allows depth limited evaluation, say for
implementing iterative deepening.

 @item @code{--enable-low-level-tracer=yes} allows support for tracing all calls,
retries, and backtracks in the system. This can help in debugging your
application, but results in performance loss.

 @item @code{--enable-wam-profile=yes} allows profiling of abstract machine
instructions. This is useful when developing YAP, should not be so
useful for normal users.

 @item @code{--enable-tabling=@{local,batched@}} allows one of the two
forms of tabling. This option is still experimental.

 @item @code{--enable-parallelism=@{env-copy,sba,a-cow@}} allows
or-parallelism supported by one of these three forms. This option is
still highly experimental.

 @item @code{--with-gmp[=DIR]} give a path to where one can find the
@code{GMP} library if not installed in the default path.

@end itemize

Next follow machine dependent details:

@node Machine Options, , Configuration Options,Install
@section Tuning YAP for a Particular Machine and Compiler
@cindex machine optimizations

The default options should give you best performance under
@code{GCC}. Although the system is tuned for this compiler
we have been able to compile versions of Yap under lcc in Linux,
Sun's cc compiler, IBM's xlc, SGI's cc, and Microsoft's Visual C++
6.0.

@menu
* Tuning for GCC:: Using the GNUCC ncompiler
* Compiling Under Visual C++:: Using Microsoft's Visual C++ environment
* Tuning for SGI cc:: Compiling Under SGI's @code{cc}
@end menu


@node Tuning for GCC, Compiling Under Visual C++, , Machine Options
@section Tuning YAP for @code{GCC}.

Yap has been developed to take advantage of @code{GCC} (but not to
depend on it). The major advantage of @code{GCC} is threaded code and
explicit register reservation.

YAP is set by default to compile with the best compilation flags we
know. Even so, a few specific options reduce portability.  The option 
@itemize @bullet
 @item @code{--enable-max-performance=yes} will try to support the best
available flags for a specific architecural model. Currently, the option
assumes a recent version of @code{GCC}.
 @item @code{--enable-debug-yap} compiles Yap so that it can be debugged
by tools such as @code{dbx} or @code{gdb}.
@end itemize

Here follow a few hints:

On x86 machines the flags:

@example
YAP_EXTRAS= ... -DBP_FREE=1
@end example

tells us to use the @code{%bp} register (frame-pointer) as the emulator's
program counter. This seems to be stable and is now default.

On  Sparc/Solaris2 use:

@example
YAP_EXTRAS= ...   -mno-app-regs -DOPTIMISE_ALL_REGS_FOR_SPARC=1
@end example

and YAP will get two extra registers! This trick does not work on
SunOS 4 machines.

Note that versions of GCC can be tweaked to recognise different
processors within the same instruction set, eg, 486, Pentium, and
PentiumPro for the x86; or Ultrasparc, and Supersparc for
Sparc. Unfortunately, some of these tweaks do may make Yap run slower or
not at all in other machines with the same instruction set, so they
cannot be made default.

Last, the best options also depends on the version of GCC you are using, and
it is a good idea to consult the GCC manual under the menus "Invoking
GCC"/"Submodel Options". Specifically, you should check
@code{-march=XXX} for recent versions of GCC/EGCS. In the case of
@code{GCC2.7} and other recent versions of @code{GCC} you can check:

@table @code

@item 486:
In order to take advantage of 486 specific optimisations in GCC 2.7.*:

@example
YAP_EXTRAS= ... -m486 -DBP_FREE=1
@end example

@item Pentium:
@example
YAP_EXTRAS= ... -m486 -malign-loops=2 -malign-jumps=2 \
                      -malign-functions=2
@end example

@item PentiumPro and other recent Intel and AMD machines:
PentiumPros are known not to require alignment. Check your version of
@code{GCC} for the best @code{-march} option.

@item Super and UltraSparcs:
@example
YAP_EXTRAS= ... -msupersparc
@end example

@item MIPS: if have a recent machine and you need a 64 bit wide address
space you can use the abi 64 bits or eabi option, as in:
@example
CC="gcc -mabi=64" ./configure --...
@end example
Be careful. At least for some versions of @code{GCC}, compiling with
@code{-g} seems to result in broken code.

@item WIN32: The cygwin environment is our suggested approach. The
CygWin environment is available from the URL:

@code{http://sourceware.cygnus.com}

@noindent
and mirrors. Yap should compile under cygwin 20.1 but we suggest using
the newer 1.1.1 or recent, which has a more complete implementation of
the WIN32 API and uses GCC2.95.2 instead of egcs. The compilation
steps under the cygwin shell are as follows:

@example
mkdir cyg
$YAPSRC/configure --enable-coroutining \\
                  --enable-max-performance
make
make install
@end example

By default, Yap will use the @code{--enable-cygwin=no} option to
disable the use of the cygwin dll and to enable the mingw32 subsystem
instead. Yap thus will not need the cygwin dll. It instead accesses
the system's @code{CRTDLL.DLL} @code{C} run time library supplied with
Win32 platforms through the mingw32 interface. Note that some older
WIN95 systems may not have @code{CRTDLL.DLL}, in this case it should
be sufficient to import the file from a newer WIN95 or WIN98 machine.

You should check the default installation path which is set to
@code{/PROGRA~1/Yap} in the standard Makefile. This string will usually
be expanded into @code{c:\Program Files\Yap} by Windows.

The cygwin environment does not provide @t{gm}. You can fecth a dll for
the @t{gmp} library from @url{http://www.sf.net/projects/mingwrep}.

It is also possible to configure Yap to be a part of the cygwin
environment. In this case you should use:
@example
mkdir cyg
$YAPSRC/configure --enable-coroutining \\
                  --enable-max-performance \\
                  --enable-cygwin=yes
make
make install
@end example
Yap will then compile using the cygwin library. You can use Yap from a
cygwin console, or as a standalone application as long as it can fin
@code{cygwin1.dll} in its path.

@end table

@node Compiling Under Visual C++, Tuning for SGI cc, Tuning for GCC, Machine Options
@subsection  Compiling Under Visual C++

Yap compiles cleanly under Microsoft's Visual C++ release 6.0. We next
give a step-by-step tutorial on how to compile Yap manually using this
environment.

First, it is a good idea to build Yap as a DLL:

@enumerate

@item create a project named yapdll using File.New. The project will be a
DLL project, initially empty.

Notice that either the project is named yapdll or you must replace the
preprocessor's variable @var{YAPDLL_EXPORTS} to match your project names
in the files @code{c_interface.h} and @code{c_interface.c}.

@item add all .c files in the @var{$YAPSRC/C} directory and in the
@var{$YAPSRC\OPTYap} directory to the Project's @code{Source Files} (use
FileView).

@item add all .h files in the @var{$YAPSRC/H} directory,
@var{$YAPSRC\include} directory and in the @var{$YAPSRC\OPTYap}
subdirectory to the Project's @code{Header Files}.

@item Ideally, you should now use @code{m4} to generate extra  .h from .m4 files and use
@code{configure} to create a @code{config.h}. Or, you can be lazy, and
fetch these files from @var{$YAPSRC\VC\include}.

@item You may want to go to @code{Build.Set Active Configuration} and set @code{Project
Type} to @code{Release}

@item To use Yap's own include directories you have to set the Project
option  @code{Project.Project Settings.C/C++.Preprocessor.Additional
Include Directories} to include the directories @var{$YAPSRC\H},
@var{$YAPSRC\VC\include}, @var{$YAPSRC\OPTYap} and
@var{$YAPSRC\include}.  The syntax is:

@example
$YAPSRC\H, $YAPSRC\VC\include, $YAPSRC\OPTYap, $YAPSRC\include
@end example

@item  Build: the system should generate an @code{yapdll.dll} and an @code{yapdll.lib}.

@item Copy the file @code{yapdll.dll} to your path. The file
@code{yapdll.lib} should also be copied to a location where the linker can find it.
@end enumerate

Now you are ready to create a console interface for Yap:
@enumerate
@item create a second project  say @code{wyap} with @code{File.New}. The project will be a
WIN32 console project, initially empty.

@item add @var{$YAPSRC\console\yap.c} to the @code{Source Files}.

@item add @var{$YAPSRC\VC\include\config.h} and the files in @var{$YAPSRC\include} to
the @code{Header Files}.

@item You may want to go to @code{Build.Set Active Configuration} and set
@code{Project Type} to @code{Release}.

@item you will eventually need to bootstrap the system by booting from
@code{boot.yap}, so write:

@example
        -b $YAPSRC\pl\boot.yap
@end example

        in @code{Project.Project Settings.Debug.Program Arguments}.

@item You need the sockets and yap libraries. Add

@example
ws2_32.lib yapdll.lib to
@end example

to

to @code{Project.Project Settings.Link.Object/Library Modules}

You may also need to set the @code{Link Path} so that VC++ will find @code{yapdll.lib}.

@item set @code{Project.Project Settings.C/C++.Preprocessor.Additional
Include Directories} to include the @var{$YAPSRC/VC/include} and
@var{$YAPSRC/include}.

The syntax is:

@example
$YAPSRC\VC\include, $YAPSRC\include
@end example

@item Build the system.

@item Use @code{Build.Start Debug} to boot the system, and then create the saved state with

@example
['$YAPSRC\\pl\\init'].
save_program(startup).
^Z
@end example

That's it, you've got Yap and the saved state!
@end enumerate

The $YAPSRC\VC directory has the make files to build Yap4.3.17 under VC++ 6.0.

@node Tuning for SGI cc, , Compiling Under Visual C++ ,Machine Options
@subsection  Compiling Under SGI's cc

YAP should compile under the Silicon Graphic's @code{cc} compiler,
although we advise using the GNUCC compiler, if available.

@table @code
@item 64 bit
Support for 64 bits should work by using (under Bourne shell syntax):
@example
CC="cc -64" $YAP_SRC_PATH/configure --...
@end example
@end table

@node Run, Syntax, Install, Top
@chapter Running YAP

@menu
* Running Yap Interactively:: Interacting with Yap
* Running Prolog Files:: Running Prolog files as scripts
@end menu

@cindex booting
We next describe how to invoke Yap in Unix systems.

@node Running Yap Interactively,Running Prolog Files, ,Run
@section Running Yap Interactively

Most often you will want to use Yap in interactive mode. Assuming that
YAP is in the user's search path, the top-level can be invoked under
Unix with the following command:

@example
yap [-s n] [-h n] [-a n] [-c IP_HOST port ] [filename]
@end example

@noindent
All the arguments and flags are optional and have the following meaning:
@table @code
@item -?
print a short error message.
@item -s @var{n}
allocate @var{n} K bytes for local and global stacks
@item -h @var{n}
allocate @var{n} K bytes for heap and auxiliary stacks
@item -t @var{n}
allocate @var{n} K bytes for the trail stack
@item -l @var{YAP_FILE}
compile the Prolog file @var{YAP_FILE} before entering the top-level.
@item -L @var{YAP_FILE}
compile the Prolog file @var{YAP_FILE} and then halt. This option is
useful for implementing scripts.
@item -b @var{BOOT_FILE}
boot code is in Prolog file @var{BOOT_FILE}. The filename must define
the predicate '$live'/0.
@item -c @t{IP_HOST} @t{port}
connect standard streams to host @t{IP_HOST} at port @t{port}
@item filename
restore state saved in the given file
@item --
separator for arguments to Prolog code. These arguments are visible
through the unix/1 built-in.
@end table

Note that YAP will output an error message on the following conditions:

@itemize @bullet
@item
a file name was given but the file does not exist or is not a saved
YAP state;
@item 
the necessary amount of memory could not be allocated;
@item
the allocated memory is not enough to restore the state.
@end itemize

When restoring a saved state, YAP will allocate the
same amount of memory as that in use when the state was saved, unless a
different amount is specified by flags in the command line. By default,
YAP restores the file @samp{startup} from the current directory or from
the YAP library.
@cindex environment variables

@findex YAPBINDIR
@itemize @bullet
@item
YAP usually boots from a saved state. The saved state will use the default
installation directory to search for the YAP binary unless you define
the environment variable YAPBINDIR.

@findex YAPLIBDIR
@item
YAP always tries to find saved states from the current directory
first. If it cannot it will use the environment variable YAPLIBDIR, if
defined, or search the default library directory.

@findex YAPSHAREDIR
@item
YAP will try to find library files from the YAPSHAREDIR/library
directory.
@end itemize

@node Running Prolog Files, ,Interactive Mode, Run
@section Running Prolog Files

YAP can also be used to run Prolog files as scripts, at least in
Unix-like environments. A simple example is shown next:

@example
@cartouche
#!/usr/local/bin/yap -L
#
# Hello World script file using Yap
#

:- write('Hello World'), nl.

@end cartouche
@end example

The @code{#!}  characters specify that the script should call the binary
file Yap. Notice that many systems will require the complete path to the
Yap binary. The @code{-L} flag indicates that YAP should consult the
current file when booting and then halt. The remaining arguments are
then passed to YAP. Note that YAP will skip the first lines if they
start with @code{#} (the comment sign for Unix's shell). YAP will
consult the file and execute any commands.

A slightly more sophisticated example is:

@example
@cartouche
#!/usr/bin/yap -L
#
# Hello World script file using Yap
#

:- initialization(main).

main :- write('Hello World'), nl.

@end cartouche
@end example

The @code{initialization} directive tells Yap to execute the goal main
after consulting the file. Source code is thus compiled and @code{main}
executed at the end.

By default, arguments to a script are considered arguments to YAP. As an
example, consider the following script @code{dump_args}:

@example
@cartouche
#!/usr/bin/yap -L

main( [] ).
main( [H|T] ) :-
        write( H ), nl,
        main( T ).

:- unix( argv(AllArgs) ), main( AllArgs ).

@end cartouche
@end example

If you this run this script with the arguments:
@example
./dump_args -s 10000
@end example
@noindent
the script will start an YAP process with stack size @code{10MB}, and
the list of arguments to the process will be empty.

Often one wants to run the script as any other program, and for this it
is convenient to ignore arguments to YAP. This is possible by using
@code{L --} as in the next version of @code{dump_args}:

@example
@cartouche
#!/usr/bin/yap -L --

main( [] ).
main( [H|T] ) :-
        write( H ), nl,
        main( T ).

:- unix( argv(AllArgs) ), main( AllArgs ).

@end cartouche
@end example

The @code{--} indicates the next arguments are not for YAP. Instead,
they must be sent directly to the @code{argv} builtin. Hence, running
@example
./dump_args test
@end example
@noindent
will write @code{test} on the standard output.


@node Syntax, Loading Programs, Run, Top
@chapter Syntax

We will describe the syntax of YAP at two levels. We first will
describe the syntax for Prolog terms. In a second level we describe
the @i{tokens} from which Prolog @i{terms} are
built.

@menu
* Formal Syntax:: Syntax of terms 
* Tokens:: Syntax of Prolog tokens
@end menu

@node Formal Syntax, Tokens, ,Syntax
@section Syntax of Terms
@cindex syntax

Below, we describe the syntax of YAP terms from the different
classes of tokens defined above. The formalism used will be @emph{BNF},
extended where necessary with attributes denoting integer precedence or
operator type.

@example

@code{
 term       ---->     subterm(1200)   end_of_term_marker

 subterm(N) ---->     term(M)         [M <= N]

 term(N)    ---->     op(N, fx) subterm(N-1)
             |        op(N, fy) subterm(N)
             |        subterm(N-1) op(N, xfx) subterm(N-1)
             |        subterm(N-1) op(N, xfy) subterm(N)
             |        subterm(N) op(N, yfx) subterm(N-1)
             |        subterm(N-1) op(N, xf)
             |        subterm(N) op(N, yf)

 term(0)   ---->      atom '(' arguments ')'
             |        '(' subterm(1200)  ')'
             |        '@{' subterm(1200)  '@}'
             |        list
             |        string
             |        number
             |        atom
             |        variable

 arguments ---->      subterm(999)
             |        subterm(999) ',' arguments

 list      ---->      '[]'
             |        '[' list_expr ']'

 list_expr ---->      subterm(999)
             |        subterm(999) list_tail

 list_tail ---->      ',' list_expr
             |        ',..' subterm(999)
             |        '|' subterm(999)
}
@end example

@noindent
Notes:

@itemize @bullet

@item
@i{op(N,T)} denotes an atom which has been previously declared with type
@i{T} and base precedence @i{N}.

@item
Since ',' is itself a pre-declared operator with type @i{xfy} and
precedence 1000, is @i{subterm} starts with a '(', @i{op} must be
followed by a space to avoid ambiguity with the case of a functor
followed by arguments, eg:

@example
@code{ + (a,b)        [the same as '+'(','(a,b)) of arity one]}
@end example
versus
@example
@code{ +(a,b)         [the same as '+'(a,b) of arity two]}
@end example

@item
In the first rule for term(0) no blank space should exist between
@i{atom} and '('.

@item
@cindex end of term
Each term to be read by the YAP parser must end with a single
dot, followed by a blank (in the sense mentioned in the previous
paragraph). When a name consisting of a single dot could be taken for
the end of term marker, the ambiguity should be avoided by surrounding the
dot with single quotes.

@end itemize

@node Tokens, , Formal Syntax, Syntax
@section Prolog Tokens
@cindex token

Prolog tokens are grouped into the following categories:

@menu
* Numbers:: Integer and Floating-Point Numbers
* Strings:: Sequences of Characters
* Atoms:: Atomic Constants
* Variables:: Logical Variables
* Punctuation Tokens:: Tokens that separate other tokens
* Layout:: Comments and Other Layout Rules
@end menu

@node Numbers, Strings, ,Tokens
@subsection Numbers
@cindex number

Numbers can be further subdivided into integer and floating-point numbers.

@menu
* Integers:: How Integers are read and represented
* Floats:: Floating Point Numbers
@end menu

@node Integers, Floats, ,Numbers
@subsubsection Integers
@cindex integer

Integer numbers
are described by the following regular expression:

@example
@code{
<integer> := @{<digit>+<single-quote>|0@{xXo@}@}<alpha_numeric_char>+
}
@end example
@noindent
where @{...@} stands for optionality, @i{+} optional repetition (one or
more times), @i{<digit>} denotes one of the characters 0 ... 9, @i{|}
denotes or, and @i{<single-quote>} denotes the character "'". The digits
before the @i{<single-quote>} character, when present, form the number
basis, that can go from 0, 1 and up to 36. Letters from @code{A} to
@code{Z} are used when the basis is larger than 10.

Note that if no basis is specified then base 10 is assumed. Note also
that the last digit of an integer token can not be immediately followed
by one of the characters 'e', 'E', or '.'.

Following the ISO standard, YAP also accepts directives of the
form @code{0x} to represent numbers in hexadecimal base and of the form
@code{0o} to represent numbers in octal base. For usefulness,
YAP also accepts directives of the form @code{0X} to represent
numbers in hexadecimal base.

Example:
the following tokens all denote the same integer
@example
@code{10  2'1010  3'101  8'12  16'a  36'a  0xa  0o12}
@end example

Numbers of the form @code{0'a} are used to represent character
constants. So, the following tokens denote the same integer:
@example
@code{0'd  100}
@end example

YAP (version @value{VERSION}) supports integers that can fit
the word size of the machine. This is 32 bits in most current machines,
but 64 in some others, such as the Alpha running Linux or Digital
Unix. The scanner will read larger or smaller integers erroneously.

@node Floats, , Integers,Numbers
@subsubsection Floating-point Numbers
@cindex floating-point number

Floating-point numbers are described by:

@example
@code{
   <float> := <digit>+@{<dot><digit>+@}
               <exponent-marker>@{<sign>@}<digit>+
            |<digit>+<dot><digit>+
               @{<exponent-marker>@{<sign>@}<digit>+@}
}
@end example

@noindent
where @i{<dot>} denotes the decimal-point character '.',
@i{<exponent-marker>} denotes one of 'e' or 'E', and @i{<sign>} denotes
one of '+' or '-'.

Examples:
@example
@code{10.0   10e3   10e-3   3.1415e+3}
@end example

Floating-point numbers are represented as a double in the target
machine. This is usually a 64-bit number.

@node Strings, Atoms, Numbers,Tokens
@subsection Character Strings
@cindex string

Strings are described by the following rules:
@example
  string --> '"' string_quoted_characters '"'

  string_quoted_characters --> '"' '"' string_quoted_characters
  string_quoted_characters --> '\'
                          escape_sequence string_quoted_characters
  string_quoted_characters -->
                          string_character string_quoted_characters

  escape_sequence --> 'a' | 'b' | 'r' | 'f' | 't' | 'n' | 'v'
  escape_sequence --> '\' | '"' | ''' | '`'
  escape_sequence --> at_most_3_octal_digit_seq_char '\'
  escape_sequence --> 'x' at_most_2_hexa_digit_seq_char '\'
@end example
where @code{string_character} in any character except the double quote
and escape characters.

Examples:
@example
@code{""   "a string"   "a double-quote:""" }
@end example

The first string is an empty string, the last string shows the use of
double-quoting. The implementation of YAP represents strings as
lists of integers. Since Yap4.3.0 there is no static limit on string
size.

Escape sequences can be used to include the non-printable characters
@code{a} (alert), @code{b} (backspace), @code{r} (carriage return),
@code{f} (form feed), @code{t} (horizontal tabulation), @code{n} (new
line), and @code{v} (vertical tabulation). Escape sequences also be
include the meta-characters @code{\}, @code{"}, @code{'}, and
@code{`}. Last, one can use escape sequences to include the characters
either as an octal or hexadecimal number.

The next examples demonstrates the use of escape sequences in YAP:

@example
@code{"\x0c\" "\01\" "\f" "\\" }
@end example

The first three examples return a list including only character 12 (form
feed). The last example escapes the escape character.

Escape sequences were not available in C-Prolog and in origional
versions of YAP up to 4.2.0. Escape sequences can be disable by using:
@example
@code{:- yap_flag(character_escapes,off).}
@end example


@node Atoms, Variables, Strings, Tokens
@subsection Atoms
@cindex atom

Atoms are defined by one of the following rules:
@example
   atom --> solo-character
   atom --> lower-case-letter name-character*
   atom --> symbol-character+
   atom --> single-quote  single-quote
   atom --> ''' atom_quoted_characters '''


  atom_quoted_characters --> ''' ''' atom_quoted_characters
  atom_quoted_characters --> '\' atom_sequence string_quoted_characters
  atom_quoted_characters --> character string_quoted_characters

@end example

where:
@example
   <solo-character>     denotes one of:    ! ;
   <symbol-character>   denotes one of:    # & * + - . / : < 
                                           = > ? @@ \ ^ ` ~
   <lower-case-letter>  denotes one of:    a...z
   <name-character>     denotes one of:    _ a...z A...Z 0....9
   <single-quote>       denotes:           '
@end example

and @code{string_character} denotes any character except the double quote
and escape characters. Note that escape sequences in strings and atoms
follow the same rules.

Examples:
@example
@code{a   a12x   '$a'   !   =>  '1 2'}
@end example

@c From version @code{4.1.8} onwards YAP supports the 8-bit
@c ISO-latin-1 character set. The following new symbol characters have
@c been introduced: @code{�, �, �, �, �, �, �, �, �, �, �. �, �, �, �, �,
@c �, �, �, �, �, �, �, �, �, �, �, �}. The following new
@c lower-case-characters have been introduced: @code{�, �, �, �, �, �, �,
@c �, �, �, �, �, �, �, �, �, �, �, �, �, �, �, �, �, �, �, �, �, �, �, �,
@c �, �, �, �, �, �, �}. Last, the following upper-case characters have been
@c introduced: @code{�, �, �, �, �, �, �, �, �, �, �, �, �, �, �, �, �, �,
@c �, �, �, �, �, �, �, �, �, �, �, �}.

Version @code{4.2.0} of YAP removed the previous limit of 256
characters on an atom. Size of an atom is now only limited by the space
available in the system.

@node Variables, Punctuation Tokens, Atoms, Tokens
@subsection Variables
@cindex variable

Variables are described by:
@example
   <variable-starter><variable-character>+
@end example
where
@example
  <variable-starter>   denotes one of:    _ A...Z
  <variable-character> denotes one of:    _ a...z A...Z
@end example

@cindex anonymous variable
If a variable is referred only once in a term, it needs not to be named
and one can use the character @code{_} to represent the variable. These
variables are known as anonymous variables. Note that different
occurrences of @code{_} on the same term represent @emph{different}
anonymous variables. 

@node Punctuation Tokens, Layout, Variables, Tokens
@subsection Punctuation Tokens
@cindex punctuation token

Punctuation tokens consist of one of the following characters:
@example
@center ( ) , [ ] @{ @} |
@end example

These characters are used to group terms.

@node Layout, ,Punctuation Tokens, Tokens
@subsection Layout
@cindex comment
Any characters with ASCII code less than or equal to 32 appearing before
a token are ignored.

All the text appearing in a line after the character @i{%} is taken to
be a comment and ignored (including @i{%}).  Comments can also be
inserted by using the sequence @code{/*} to start the comment and
@code{*/} to finish it. In the presence of any sequence of comments or
layout characters, the YAP parser behaves as if it had found a
single blank character. The end of a file also counts as a blank
character for this purpose.

@node Loading Programs, Modules, Syntax, Top
@chapter Loading Programs

@menu

Loading Programs
* Compiling:: Program Loading and Updating
* Setting the Compiler:: Changing the compiler's parameters
* Saving:: Saving and Restoring Programs

@end menu


@node Compiling, Setting the Compiler, , Loading Programs
@section Program loading and updating

@table @code

@item consult(@var{+F})
@findex consult/1
@snindex consult/1
@cyindex consult/1
Adds the clauses written in file @var{F} or in the list of files @var{F}
to the program.

In YAP @code{consult/1} does not remove previous clauses for
the procedures defined in @var{F}. Moreover, note that all code in YAP
is compiled.

@item reconsult(@var{+F})
@findex reconsult/1
@snindex reconsult/1
@cyindex reconsult/1
Updates the program replacing the
previous definitions for the predicates defined in @var{F}.


@item [@var{+F}]
@findex []/1
@saindex []/1
@cyindex []/1
The same as @code{consult(F)}.

@item [-@var{+F}]
@findex [-]/1
@saindex [-]/1
@cyindex [-]/1
The same as @code{reconsult(F)}

Example:

@example
?- [file1, -file2, -file3, file4].
@end example
@noindent
will consult @code{file1} @code{file4} and reconsult @code{file2} and
@code{file3}.

@item compile(@var{+F})
@findex compile/1
@syindex compile/1
@cnindex compile/1
@noindent
In YAP, the same as @code{reconsult/1}.

@item ensure_loaded(@var{+F}) [ISO]
@findex ensure_loaded/1
@syindex compile/1
@cnindex compile/1
When the files specified by @var{F} are module files,
@code{ensure_loaded/1} loads them if they have note been previously
loaded, otherwise advertises the user about the existing name clashes
and prompts about importing or not those predicates. Predicates which
are not public remain invisible.

When the files are not module files, @code{ensure_loaded/1} loads them
if they have not been loaded before, does nothing otherwise.

@var{F} must be a list containing the names of the files to load.

@item include(@var{+F}) [ISO]
@findex include/1 (directive)
@snindex compile/1 (directive)
@cnindex compile/1 (directive)
The @code{include} directive includes the text files or sequence of text
files specified by @var{F} into the file being currently consulted.

@end table

@node Setting the Compiler, Saving, Compiling, Loading Programs
@section Changing the Compiler's Behaviour

This section presents a set of built-ins predicates designed to set the 
environment for the compiler.

@table @code

@item source_mode(-@var{O},+@var{N})
@findex source_mode/2
@snindex source_mode/2
@cnindex source_mode/2
The state of source mode can either be on or off. When the source mode
is on, all clauses are kept both as compiled code and in a "hidden"
database. @var{O} is unified with the previous state and the mode is set
according to @var{N}.

@item source
@findex source/0
@snindex source/0
@cnindex source/0
After executing this goal, YAP keeps information on the source
of the predicates that will be consulted. This enables the use of
@code{listing/0}, @code{listing/1} and @code{clause/2} for those
clauses.

The same as @code{source_mode(_,on)} or as declaring all newly defined
static procedures as @code{public}.

@item no_source
@findex no_source/0
@snindex no_source/0
@cnindex no_source/0
The opposite to @code{source}.

The same as @code{source_mode(_,off)}.

@item compile_expressions
@findex compile_expressions/0
@snindex compile_expressions/0
@cnindex compile_expressions/0
After a call to this predicate, arithmetical expressions will be compiled.
(see example below). This is the default behaviour.

@item do_not_compile_expressions
@findex do_not_compile_expressions/0
@snindex do_not_compile_expressions/0
@cnindex do_not_compile_expressions/0
After a call to this predicate, arithmetical expressions will not be compiled.
@example
?- source, do_not_compile_expressions.
yes
?- [user].
| p(X) :- X is 2 * (3 + 8).
| :- end_of_file.
?- compile_expressions.
yes
?- [user].
| q(X) :- X is 2 * (3 + 8).
| :- end_of_file.
:- listing.

p(A):-
      A is 2 * (3 + 8).

q(A):-
      A is 22.
@end example

@item expand_exprs(-@var{O},+@var{N})
@findex expand_exprs/2
@snindex expand_exprs/2
@cyindex expand_exprs/2
Puts YAP in state @var{N} (@code{on} or @code{off}) and unify
@var{O} with the previous state, where @var{On} is equivalent to
@code{compile_expressions} and @code{off} is equivalent to
@code{do_not_compile_expressions}. This predicate was kept to maintain
compatibility with C-Prolog.

@item path(-@var{D})
@findex path/1
@snindex path/1
@cnindex path/1
Unifies @var{D} with the current directory search-path of YAP.
Note that this search-path is only used by YAP to find the
files for @code{consult/1}, @code{reconsult/1} and @code{restore/1} and
should not be taken for the system search path.

@item add_to_path(+@var{D})
@findex add_to_path/1
@snindex path/1
@cnindex path/1
Adds @var{D} to the end of YAP's directory search path.

@item add_to_path(+@var{D},+@var{N})
@findex add_to_path/2
@snindex path/1
@cnindex path/1
Inserts @var{D} in thgoe position, of the directory search path of
YAP, specified by @var{N}.  @var{N} must be either of
@code{first} or @code{last}.

@item remove_from_path(+@var{D})
@findex remove_from_path/1
@snindex remove_from_path/1
@cnindex remove_from_path/1
Remove @var{D} from YAP's directory search path.

@item style_check(+@var{X})
@findex style_check/1
@snindex style_check/1
@cnindex style_check/1
Turns on style checking according to the attribute specified by @var{X},
which must be one of the following:
@table @code
@item single_var
Checks single occurrences of named variables in a clause.
@item discontiguous
Checks non-contiguous clauses for the same predicate in a file.
@item multiple
Checks the presence of clauses for the same predicate in more than one
file when the predicate has not been declared as @code{multifile}
@item all
Performs style checking for all the cases mentioned above.
@end table
By default, style checking is disabled in YAP unless we are in
@code{sicstus} or @code{iso} language mode.

The @code{style_check/1} built-in is now deprecated. Please use the
@code{set_prolog_flag/1} instead.

@item no_style_check(+@var{X})
@findex no_style_check/1
@snindex style_check/1
@cnindex style_check/1
Turns off style checking according to the attribute specified by
@var{X}, which has the same meaning as in @code{style_check/1}.

The @code{no_style_check/1} built-in is now deprecated. Please use the
@code{set_prolog_flag/1} instead.

@item multifile @var{P} [ISO]
@findex multifile/1 (directive)
@syindex multifile/1 (directive)
@cnindex multifile/1 (directive)
Instructs the compiler about the declaration of a predicate @var{P} in
more than one file. It must appear in the first of the loaded files
where the predicate is declared, and before declaration of any of its
clauses.

Multifile declarations affect @code{reconsult/1} and @code{compile/1}:
when a multifile predicate is reconsulted, only the clauses from the
same file are removed.

Since Yap4.3.0 multifile procedures can be static or dynamic.

@item discontiguous(+@var{G}) [ISO]
@findex discontiguous/1 (directive)
@syindex discontiguous/1 (directive)
@cnindex discontiguous/1 (directive)

Declare that the arguments are discontiguous procedures, that is,
clauses for discontigous procedures may be separated by clauses from
other procedures.

@item initialization(+@var{G}) [ISO]
@findex initialization/1 (directive)
@snindex initialization/1 (directive)
@cnindex initialization/1 (directive)
The compiler will execute goals @var{G} after consulting the current
file.

@item library_directory(+@var{D})
@findex library_directory/1
@snindex library_directory/1
@cnindex library_directory/1
Succeeds when @var{D} is a current library directory name. Library
directories are the places where files specified in the form
@code{library(@var{File})} are searched by the predicates
@code{consult/1}, @code{reconsult/1}, @code{use_module/1} or
@code{ensure_loaded/1}.

@item file_search_path(+@var{NAME},-@var{DIRECTORY})
@findex file_search_path/2
@syindex file_search_path/2
@cnindex file_search_path/2
Allows writing file names as compound terms. The @var{NAME} and
 @var{DIRECTORY} must be atoms. The predicate may generate multiple
solutions. The predicate is originally defined as follows:

@example
file_search_path(library,A) :-
   library_directory(A).
file_search_path(system,A) :-
   prolog_flag(host_type,A).
@end example

Thus, [library(A)] will search for a file using
@var{library_directory}/1 to obtain the prefix.

@item library_directory(+@var{D})
@findex library_directory/1
@snindex library_directory/1
@cnindex library_directory/1
Succeeds when @var{D} is a current library directory name. Library
directories are the places where files specified in the form
@code{library(@var{File})} are searched by the predicates
@code{consult/1}, @code{reconsult/1}, @code{use_module/1} or
@code{ensure_loaded/1}.

@item prolog_file_name(+@var{Name},-@var{FullPath})
@findex prolog_file_name/2
@syindex prolog_file_name/1
@cnindex prolog_file_name/2
Unify @var{FullPath} with the absolute path YAP would use to consult
file @var{Name}.

@item public @var{P} [ISO]
@findex public/1 (directive)
@snindex public/1 (directive)
@cnindex public/1 (directive)
Instructs the compiler that the source of a predicate of a list of
predicates @var{P} must be kept. This source is then accessible through
the @code{clause/2} procedure and through the @code{listing} family of
built-ins.

Note that all dynamic procedures are public. The @code{source} directive
defines all new or redefined predicates to be public.

Since Yap4.3.0 multifile procedures can be static or dynamic.

@end table

@node Saving, , Setting the Compiler, Loading Programs
@section Saving and Loading Prolog States

@table @code
@item save(+@var{F})
@findex save/1
@snindex save/1
@cyindex save/1
Saves an image of the current state of YAP in file @var{F}. From
@strong{Yap4.1.3} onwards, YAP saved states are executable
files in the Unix ports.

@item save(+@var{F},-@var{OUT})
@findex save/2
@snindex save/2
@cnindex save/2
Saves an image of the current state of YAP in file @var{F}. From
@strong{Yap4.1.3} onwards, YAP saved states are executable
files in the Unix ports.

Unify @var{OUT} with 1 when saving the file and @var{OUT} with 0 when
restoring the saved state.

@item save_program(+@var{F})
@findex save_program/1
@syindex save_program/1
@cnindex save_program/1
Saves an image of the current state of the YAP database in file
@var{F}.

@item save_program(+@var{F}, :@var{G})
@findex save_program/2
@syindex save_program/2
@cnindex save_program/2
Saves an image of the current state of the YAP database in file
@var{F}, and guarantee that execution of the restored code will start by
trying goal @var{G}.

@item restore(+@var{F})
@findex restore/1
@syindex restore/1
@cnindex restore/1
Restores a previously saved state of YAP from file @var{F}.

YAP always tries to find saved states from the current directory
first. If it cannot it will use the environment variable YAPLIBDIR, if
defined, or search the default library directory.
@end table


@node Modules, Builtins, Loading Programs, Top
@chapter The Module System

Module systems are quite important for the development of large
applications. YAP implements a module system compatible with the Quintus
Prolog module system.

The YAP module system is predicate-based. This means a module consists
of a set of predicates (or procedures), such that some predicates are
public and the others are local to a module. Atoms and terms in general
are global to the system. Moreover, the module system is flat, meaning
that we do not support an hierarchy of modules. Modules can
automatically import other modules, though. For compatibility with other
module systems the YAP module system is non-strict, meaning both that
there is both a way to access predicates private to a module and that is
possible to declare predicates for a module from some other module.

YAP allows one to ignore the module system if one does not want to use
it. Last note that using the module system does not introduce any
significant overheads: only meta-calls that cross module boundaries are
slowed down by the presence of modules.

@menu

* Module Concepts:: The Key Ideas in Modules
* Defining Modules:: How To Define a New Module
* Using Modules:: How to Use a Module
* Meta-Predicates in Modules:: How to Handle New Meta-Predicates

@end menu

@node Module Concepts, Defining Modules, , Modules
@section Module Concepts

The YAP module system applies to predicates. All predicates belong to a
module. System predicates belong to the module @code{primitives}, and by
default new predicates belong to the module @code{user}. Predicates from
the module @code{primitives} are automatically visible to every module.

Every predicate must belong to a module. This module is called its
@emph{source module}.

By default, the source module for a clause occurring in a source file
with a module declaration is the declared module. For goals typed in 
a source file without module declarations, their module is the module
the file is being loaded into. If no module declarations exist, this is
the current @emph{type-in module}. The default type-in module is
@code{user}, but one can set the current module by using the built-in
@code{module/1}.

Note that in this module system one can explicitly specify the source
mode for a clause by prefixing a clause with its module, say:
@example
user:(a :- b).
@end example
@noindent
In fact, to specify the source module for a clause it is sufficient to
specify the source mode for the clause's head:
@example
user:a :- b.
@end example
@noindent

The rules for goals are similar. If a goal appears in a text file with a
module declaration, the goal's source module is the declared
module. Otherwise, it is the module the file is being loaded into or the
type-in module.

One can override this rule by prefixing a goal with the module it is
supposed to be executed into, say:
@example
nasa:launch(apollo,13).
@end example
will execute the goal @code{launch(apollo,13)} as if the current source
module was @code{nasa}.

Note that this rule breaks encapsulation and should be used with care.

@node Defining Modules, Using Modules, Module Concepts, Modules
@section Defining a New Module

A new module is defined by a @code{module} declaration:

@table @code

@item module(+@var{M},+@var{L})
@findex module/2 (directive)
@syindex module/2 (directive)
@cnindex module/2 (directive)
This predicate defines the file where it appears as a module file; it
must be the first declaration in the file.
@var{M} must be an atom specifying the module name; @var{L} must be a list
containing the module's public predicates specification, in the form
@code{[predicate_name/arity,...]}.

The public predicates of a module file can be made accessible by other
files through the predicates @code{consult/1}, @code{reconsult/1},
@code{ensure_loaded/1} or @code{use_module/2}. The non-public predicates
of a module file are not visible by other files; they can, however, be
accessed if the module name is prefixed to the file name through the
@code{:/2} operator.

@end table

The built-in @code{module/1} sets the current source module:
@table @code

@item module(+@var{M},+@var{L}, +@var{Options})
@findex module/3 (directive)
@syindex module/3 (directive)
@cnindex module/3 (directive)
Similar to @code{module/2}, this predicate defines the file where it
appears as a module file; it must be the first declaration in the file.
@var{M} must be an atom specifying the module name; @var{L} must be a
list containing the module's public predicates specification, in the
form @code{[predicate_name/arity,...]}.

The last argument @var{Options} must be a list of options, which can be:

@table @code
@item filename
 the filename for a module to import into the current module.

@item library(file)
 a library file to import into the current module.

@item hide(@var{Opt})
 if @var{Opt} is @code{false}, keep source code for current module, if
@code{true}, disable.
@end table

@item module(+@var{M})
@findex module/1
@syindex module/1
@cnindex module/1
Defines @var{M} to be the current working or type-in module. All files
which are not binded to a module are assumed to belong to the working
module (also referred to as type-in module). To compile a non-module
file into a module which is not the working one, prefix the file name
with the module name, in the form @code{@var{Module}:@var{File}}, when
loading the file.

@end table

@node Using Modules, Meta-Predicates in Modules, Defining Modules, Modules
@section Using Modules

By default, all procedures to consult a file will load the modules
defined therein. The two following declarations allow one to import a
module explicitly. They differ on whether one imports all predicate
declared in the module or not.

@table @code

@item use_module(+@var{F})
@findex use_module/1
@syindex use_module/1
@cnindex use_module/1
Loads the files specified by @var{F}, importing all their public
predicates. Predicate name clashes are resolved by asking the user about
importing or not the predicate. A warning is displayed when @var{F} is
not a module file.

@item use_module(+@var{F},+@var{L})
@findex use_module/2
@syindex use_module/2
@cnindex use_module/2
Loads the files specified by @var{F}, importing the predicates specified
in the list @var{L}. Predicate name clashes are resolved by asking the
user about importing or not the predicate. A warning is displayed when
@var{F} is not a module file.

@item use_module(?@var{M},?@var{F},+@var{L})
@findex use_module/3
@syindex use_module/3
@cnindex use_module/3
If module @var{M} has been defined, import the procedures in @var{L} to
the current module. Otherwise, load the files specified by @var{F},
importing the predicates specified in the list @var{L}. 
@end table

@node Meta-Predicates in Modules, , Using Modules, Modules
@section Meta-Predicates in Modules

The module system must know whether predicates operate on goals or
clauses. Otherwise, such predicates would call a goal in the module they
were defined, instead of calling it in the module they are currently
executing. So, for instance:
@example
:- module(example,[a/1]).

...

a(G) :- call(G)

...

@end example
The expected behaviour for this procedure is to execute goal @var{G}
within  the current module, that is, within @code{example}.
On the other hand, when executing @code{call/1} the system only knows
where @code{call/1} was defined, that is, it only knows of
@code{primitives}. A similar problem arises for @code{assert/1} and
friends.

The @code{meta_call/1} declaration informs the system that some
arguments of a procedure are goals, clauses or clauses heads, and that
these arguments must be expanded to receive the current source module:

@table @code

@item meta_predicate @var{G1},....,@var{Gn}
@findex meta_predicate/1 (directive)
@syindex meta_predicate/1 (directive)
@cnindex meta_predicate/1 (directive)
Each @var{Gi} is a mode specification. For example, a declaration for
@code{call/1} and @code{setof/3} would be of the form:

@example
:- meta_predicate call(:), setof(?,:,?).
@end example

If the argument is @code{:} or an integer, the argument is a call and
must be expanded. Otherwise, the argument should not be expanded. Note
that the system already includes declarations for all built-ins.

@end table

In the previous example, the only argument to @code{call/1} must be
expanded, resulting in the following code:

@example
:- module(example,[a/1]).

...

a(G) :- call(example:G)

...

@end example

@node Builtins, Library, Modules, Top

@chapter Built-In Predicates

@menu

Builtins, Debugging, Syntax, Top
* Control:: Controlling the Execution of Prolog Programs
* Undefined Procedures:: Handling calls to Undefined Procedures
* Testing Terms:: Predicates on Terms
* Comparing Terms:: Comparison of Terms
* Arithmetic:: Arithmetic in Yap
* I/O:: Input/Output with Yap
* Database:: Modifying Prolog's Database
* Sets:: Finding All Possible Solutions
* Grammars:: Grammar Rules
* Preds:: Predicate Information
* OS:: Access to Operating System Functionality
* Term Modification:: Updating Prolog Terms
* Profiling:: Profiling Prolog Execution
* Arrays:: Supporting Global and Local Arrays
* Preds:: Information on Predicates
* Misc:: Miscellaneous Predicates

@end menu

@node Control, Undefined Procedures, , Top
@section Control Predicates


This chapter describes the predicates for controlling the execution of
Prolog programs.

In the description of the arguments of functors the following notation
will be used:

@itemize @bullet
@item
a preceding plus sign will denote an argument as an "input argument" -
it cannot be a free variable at the time of the call; 
@item
 a preceding minus sign will denote an "output argument";
@item
an argument with no preceding symbol can be used in both ways.
@end itemize


@table @code

@item +@var{P}, +@var{Q} [ISO]
@findex ,/2
@syindex ,/2
@cyindex ,/2
Conjunction of goals (and).

@noindent
Example:
@example
 p(X) :- q(X), r(X).
@end example

@noindent
should be read as "p(@var{X}) if q(@var{X}) and r(@var{X})".

@item +@var{P} ; +@var{Q} [ISO]
@findex ;/2 
@syindex ;/2 
@cyindex ;/2 
Disjunction of goals (or).

@noindent
Example:
@example
 p(X) :- q(X); r(X).
@end example
@noindent
should be read as "p(@var{X}) if q(@var{X}) or r(@var{X})".

@item true [ISO]
@findex true/0
@syindex true/0
@cyindex true/0
Succeeds once.

@item fail [ISO]
@findex fail/0
@syindex fail/0
@cyindex fail/0
Fails always.

@item false
@findex false/0
@syindex false/0
@cnindex false/0
The same as fail

@item ! [ISO]
@findex !/0
@syindex !/0
@cyindex !/0
  Read as "cut". Cuts any choices taken in the current procedure.
When first found "cut" succeeds as a goal, but if backtracking should
later return to it, the parent goal (the one which matches the head of
the clause containing the "cut", causing the clause activation) will
fail. This is an extra-logical predicate and cannot be explained in
terms of the declarative semantics of Prolog.

example:

@example
 member(X,[X|_]).
 member(X,[_|L]) :- member(X,L).
@end example

@noindent
With the above definition

@example
 ?- member(X,[1,2,3]).
@end example

@noindent
will return each element of the list by backtracking. With the following
definition:

@example
 member(X,[X|_]) :- !.
 member(X,[_|L]) :- member(X,L).
@end example

@noindent
the same query would return only the first element of the 
list, since backtracking could not "pass through" the cut.

@item \+ +@var{P} [ISO]
@findex \+/1
@syindex \+/1
@cyindex \+/1
Goal @var{P} is not provable. The execution of this predicate fails if
and only if the goal @var{P} finitely succeeds. It is not a true logical
negation, which is impossible in standard Prolog, but
"negation-by-failure".

@noindent
This predicate might be defined as:
@example
 \+(P) :- P, !, fail.
 \+(_).
@end example
@noindent
if @var{P} did not include "cuts".

@item not +@var{P}
@findex not/1
@snindex not/1
@cyindex not/1
Goal @var{P} is not provable. The same as @code{'\+ @var{P}'}.

This predicate is kept for compatibility with C-Prolog and previous
versions of YAP. Uses of @code{not/1} should be replace by
@code{(\+)/1}, as YAP does not implement true negation.

@item  +@var{P} -> +@var{Q} [ISO]
@findex ->/2
@syindex ->/2
@cnindex ->/2
Read as "if-then-else" or "commit". This operator is similar to the
conditional operator of imperative languages and can be used alone or
with an else part as follows:

@table @code
@item +P -> +Q
"if P then Q".
@item +P -> +Q; +R
"if P then Q else R".
@end table

@noindent
These two predicates could be defined respectively in Prolog as:
@example
 (P -> Q) :- P, !, Q.
@end example
@noindent
and
@example
 (P -> Q; R) :- P, !, Q.
 (P -> Q; R) :- R.
@end example
@noindent
if there were no "cuts" in @var{P}, @var{Q} and @var{R}.

Note that the commit operator works by "cutting" any alternative
solutions of @var{P}.

Note also that you can use chains of commit operators like:
@example
    P -> Q ; R -> S ; T.
@end example
@noindent
Note that @code{(->)/2} does not affect the scope of cuts in its
arguments.

@item repeat [ISO]
@findex repeat/0
@syindex repeat/0
@cyindex repeat/0
Succeeds repeatedly.
 
In the next example, @code{repeat} is used as an efficient way to implement
a loop. The next example reads all terms in a file:

@example
 a :- repeat, read(X), write(X), nl, X=end_of_file, !.
@end example
@noindent
the loop is effectively terminated by the cut-goal, when the test-goal
@code{X=end} succeeds. While the test fails, the goals @code{read(X)},
@code{write(X)}, and @code{nl} are executed repeatedly, because
backtracking is caught by the @code{repeat} goal.

The built-in @code{repeat/1} could be defined in Prolog by:
@example
 repeat.
 repeat :- repeat.
@end example

@item call(+@var{P}) [IS0]
@findex call/1
@syindex call/1
@cyindex call/1
 If @var{P} is instantiated to an atom or a compound term, the goal
@code{call(@var{P})} is executed as if the value of @code{P} was found
instead of the call to @code{call/1}, except that any "cut" occurring in
@var{P} only cuts alternatives in the execution of @var{P}.

@item incore(+@var{P})
@findex incore/1
@syindex incore/1
@cnindex incore/1
The same as @code{call/1}.

@item call_with_args(+@var{Name},...,?@var{Ai},...)
@findex call_with_args/n
@snindex call_with_args/n
@cnindex call_with_args/n
Meta-call where @var{Name} is the name of the procedure to be called and
the @var{Ai} are the arguments. The number of arguments varies between 0
and 10.

@item +@var{P}
 The same as @code{call(@var{P})}. This feature has been kept to provide
compatibility with C-Prolog. When compiling a goal, YAP
generates a @code{call(@var{X})} whenever a variable @var{X} is found as
a goal.

@example
 a(X) :- X.
@end example
@noindent
is converted to:
@example
 a(X) :- call(X).
@end example

@item if(?@var{G},?@var{H},?@var{I}) [IS0]
@findex if/3
@syindex if/3
@cnindex if/3
Call goal @var{H} once per each solution of goal @var{H}. If goal
@var{H} has no solutions, call goal @var{I}.

The builtin @code{if/3} is similar to @code{->/3}, with the difference
that it will backtrack over the test goal. Consider the following
small data-base:

@example
a(1).        b(a).          c(x).
a(2).        b(b).          c(y).
@end example

Execution of an @code{if/3} query will proceed as follows:

@example
   ?- if(a(X),b(Y),c(Z)).

X = 1,
Y = a ? ;

X = 1,
Y = b ? ;

X = 2,
Y = a ? ;

X = 2,
Y = b ? ;

no
@end example


@noindent
The system will backtrack over the two solutions for @code{a/1} and the
two solutions for @code{b/1}, generating four solutions.

Cuts are allowed inside the first goal @var{G}, but they will only prune
over @var{G}.

If you want @var{G} to be deterministic you should use if-then-else, as
it is both more efficient and more portable.

@item once(:@var{G}) [IS0]
@findex once/1
@snindex once/1
@cnindex once/1
Execute the goal @var{G} only once. The predicate is defined by:

@example
 once(G) :- call(G), !.
@end example

@noindent
Note that cuts inside @code{once/1} can only cut the other goals inside
@code{once/1}.

@item abort
@findex abort/0
@syindex abort/0
@cyindex abort/0
Abandons the execution of the current goal and returns to top level. All
break levels (see @code{break/0} below) are terminated. It is mainly
used during debugging or after a serious execution error, to return to
the top-level.


@item break
@findex break/0
@syindex break/0
@cyindex break/0
Suspends the execution of the current goal and creates a new execution
level similar to the top level, displaying the following message:

@example
 [ Break (level <number>) ]
@end example
@noindent
telling the depth of the break level just entered. To return to the
previous level just type the end-of-file character or call the
end_of_file predicate.  This predicate is especially useful during
debugging.

@item halt [ISO]
@findex halt/0
@syindex halt/0
@cyindex halt/0
Halts Prolog, and exits to the calling application. In YAP,
@code{halt/0} returns the exit code @code{0}.

@item halt(+ @var{I}) [ISO]
@findex halt/1
@syindex halt/1
@cnindex halt/1
Halts Prolog, and exits to the calling application returning the code
given by the integer @var{I}.

@item catch(+@var{Goal},+@var{Exception},+@var{Action}) [IS0]
@findex catch/3
@snindex catch/3
@cnindex catch/3
The goal @code{catch(@var{Goal},@var{Exception},@var{Action})} tries to
execute goal @var{Goal}. If during its execution, @var{Goal} throws an
exception @var{E'} and this exception unifies with @var{Exception}, the
exception is considered to be caught and @var{Action} is executed. If
the exception @var{E'} does not unify with @var{Exception}, control
again throws the exception.

The top-level of YAP maintains a default exception handler that
is responsible to capture uncaught exceptions.

@item throw(+@var{Ball}) [ISO]
@findex throw/1
@snindex throw/1
@cnindex throw/1
The goal @code{throw(@var{Ball})} throws an exception. Execution is
stopped, and the exception is sent to the ancestor goals until reaching
a matching @code{catch/3}, or until reaching top-level.

@item garbage_collect
@findex garbage_collect/0
@syindex garbage_collect/0
@cnindex garbage_collect/0
The goal @code{garbage_collect} forces a garbage collection.

@item gc
@findex gc/0
@syindex gc/0
@cnindex gc/0
The goal @code{gc} enables garbage collection. The same as
@code{yap_flag(gc,on)}.

@item nogc
@findex nogc/0
@syindex nogc/0
@cnindex nogc/0
The goal @code{nogc} disables garbage collection. The same as
@code{yap_flag(gc,off)}.

@item grow_heap(+@var{Size})
@snindex grow_heap/1
@cnindex grow_heap/1
Increase heap size @var{Size} kilobytes.

@item grow_stack(+@var{Size})
@findex grow_stack/1
@snindex grow_stack/1
@cnindex grow_stack/1
Increase stack size @var{Size} kilobytes.

@end table

@node Undefined Procedures, Testing Terms, Control, Top
@section Handling Undefined Procedures

A predicate in a module is said to be undefined if there are no clauses
defining the predicate, and if the predicate has not been declared to be
dynamic. What YAP does when trying to execute undefined predicates can
be specified through three different ways:
@itemize @bullet
@item By setting an YAP flag, through the @code{yap_flag/2} or
@code{set_prolog_flag/2} built-ins. This solution generalises the
ISO standard.
@item By using the @code{unknown/2} built-in (this solution is
compatible with previous releases of YAP).
@item By defining clauses for the hook predicate
@code{user:unknown_predicate_handler/3}. This solution is compatible
with SICStus Prolog.
@end itemize

In more detail:
@table @code
@item unknown(-@var{O},+@var{N})
@findex unknown/2
@saindex unknown/2
@cnindex unknown/2
Specifies an handler to be called is a program tries to call an
undefined static procedure @var{P}.

The arity of @var{N} may be zero or one. If the arity is @code{0}, the
new action must be one of @code{fail}, @code{warning}, or
@code{error}. If the arity is @code{1}, @var{P} is an user-defined
handler and at run-time, the argument to the handler @var{P} will be
unified with the undefined goal. Note that @var{N} must be defined prior
to calling @code{unknown/2}, and that the single argument to @var{N} must
be unbound.

In YAP, the default action is to @code{fail} (note that in the ISO
Prolog standard the default action is @code{error}).

After defining @code{undefined/1} by:
@example
undefined(A) :- format('Undefined predicate: ~w~n'), fail.
@end example
@noindent
and executing the goal:
@example
unknown(U,undefined(X)).
@end example
@noindent
a call to a predicate for which no clauses were defined will result in
the output of a message of the form:
@example
Undefined predicate: user:xyz(A1,A2)
@end example
@noindent
followed by the failure of that call.

@item yap_flag(unknown,+SPEC)
Alternatively, one can use @code{yap_flag/2},
@code{current_prolog_flag/2}, or @code{set_prolog_flag/2}, to set this
functionality. In this case, the first argument for the built-ins should
be @code{unknown}, and the second argument should be either
@code{error}, @code{warning}, @code{fail}, or a goal.

@item user:unknown_predicate_handler(+G,+M,?NG)
@findex user:unknown_predicate_handler/3
@syindex user:unknown_predicate_handler/3
@cnindex user:unknown_predicate_handler/3
The user may also define clauses for
@code{user:unknown_predicate_handler/3} hook predicate. This
user-defined procedure is called before any system processing for the
undefined procedure, with the first argument @var{G} set to the current
goal, and the second @var{M} set to the current module.

If @code{user:unknown_predicate_handler/3} succeeds, the system will
execute @var{NG}. If  @code{user:unknown_predicate_handler/3} fails, the
system will execute default action as specified by @code{unknown/2}.
@end table

@node Testing Terms, Comparing Terms, Undefined Procedures, Top
@section Predicates on terms

@table @code

@item var(@var{T}) [ISO]
@findex var/1
@syindex var/1
@cyindex var/1
Succeeds if @var{T} is currently a free variable, otherwise fails. 

@item atom(@var{T}) [ISO]
@findex atom/1
@syindex atom/1
@cyindex atom/1
Succeeds if and only if @var{T} is currently instantiated to an  atom.

@item atomic(T) [ISO]
@findex atomic/1
@syindex atomic/1
@cyindex atomic/1
Checks whether @var{T} is an atomic symbol (atom or number).

@item compound(@var{T}) [ISO]
@findex compound/1
@syindex compound/1
@cnindex compound/1
Checks whether @var{T} is a compound term.

@item db_reference(@var{T})
@findex db_reference/1C
@syindex db_reference/1
@cyindex db_reference/1
Checks whether @var{T} is a database reference.

@item float(@var{T}) [ISO]
@findex float/1
@syindex float/1
@cnindex float/1
Checks whether @var{T} is a floating point number.

@item integer(@var{T}) [ISO]
@findex integer/1
@syindex integer/1
@cyindex integer/1
Succeeds if and only if @var{T} is currently instantiated to an  integer.

@item nonvar(@var{T}) [ISO]
@findex nonvar/1
@syindex nonvar/1
@cyindex nonvar/1
The opposite of @code{var(@var{T})}.

@item number(@var{T}) [ISO]
@findex number/1
@syindex number/1
@cyindex number/1
Checks whether @code{T} is an integer or a float.

@item primitive(@var{T})
@findex primitive/1
@syindex primitive/1
@cyindex primitive/1
Checks whether @var{T} is an atomic term or a database reference.

@item simple(@var{T})
@findex simple/1
@syindex simple/1
@cnindex simple/1
Checks whether @var{T} is unbound, an atom, or a number.

@item callable(@var{T})
@findex callable/1
@syindex callable/1
@cnindex callable/1
Checks whether @var{T} is a callable term, that is, an atom or a
compound term.

@item name(@var{A},@var{L})
@findex name/2
@syindex name/2
@cyindex name/2
The predicate holds when at least one of the arguments is ground
(otherwise, an error message will be displayed). The argument @var{A} will
be unified with an atomic symbol and @var{L} with the list of the ASCII
codes for the characters of the external representation of @var{A}.

@example
 name(yap,L).
@end example
@noindent
will return:
@example
 L = [121,97,112].
@end example
@noindent
and
@example
 name(3,L).
@end example
@noindent
will return:
@example
 L = [51].
@end example

@item atom_chars(?@var{A},?@var{L}) [ISO]
@findex atom_chars/2
@saindex atom_chars/2
@cnindex atom_chars/2
The predicate holds when at least one of the arguments is ground
(otherwise, an error message will be displayed). The argument @var{A} must
be unifiable with an atom, and the argument @var{L} with the list of the
ASCII codes for the characters of the external representation of @var{A}.

The ISO-Prolog standard dictates that @code{atom_chars/2} should unify
the second argument with a list of one-char atoms, and not the character
codes. For compatibility with previous versions of YAP, and
with other Prolog implementations, YAP unifies the second
argument with the character codes, as in @code{atom_codes/2}. Use the
@code{set_prolog_flag(to_chars_mode,iso)} to obtain ISO standard
compatibility.

@item atom_codes(?@var{A},?@var{L}) [ISO]
@findex atom_codes/2
@syindex atom_codes/2
@cnindex atom_codes/2
The predicate holds when at least one of the arguments is ground
(otherwise, an error message will be displayed). The argument @var{A} will
be unified with an atom and @var{L} with the list of the ASCII
codes for the characters of the external representation of @var{A}.

@item atom_concat(+@var{As},?@var{A})
@findex atom_concat/2
@snindex atom_concat/2
@cnindex atom_concat/2
The predicate holds when the first argument is a list of atoms, and the
second unifies with the atom obtained by concatenating all the atoms in
the first list.

@item atom_length(+@var{A},?@var{I}) [ISO]
@findex atom_length/2
@snindex atom_length/2
@cnindex atom_length/2
The predicate holds when the first argument is an atom, and the second
unifies with the number of characters forming that atom.

@item atom_concat(?@var{A1},?@var{A2},?@var{A12}) [ISO]
@findex atom_concat/3
@snindex atom_concat/3
@cnindex atom_concat/3
The predicate holds when the third argument unifies with an atom, and
the first and second unify with atoms such that their representations
concatenated are the representation for @var{A12}.

If @var{A1} and @var{A2} are unbound, the built-in will find all the atoms
that concatenated give @var{A12}.

@item number_chars(?@var{I},?@var{L})
@findex number_chars/2
@saindex number_chars/2
@cnindex number_chars/2

The predicate holds when at least one of the arguments is ground
(otherwise, an error message will be displayed). The argument @var{I} must
be unifiable with a number, and the argument @var{L} with the list of the
ASCII codes for the characters of the external representation of @var{I}.

The ISO-Prolog standard dictates that @code{number_chars/2} should unify
the second argument with a list of one-char atoms, and not the character
codes. For compatibility with previous versions of YAP, and
with other Prolog implementations, YAP unifies the second
argument with the character codes, as in @code{number_codes/2}. Use the
@code{set_prolog_flag(to_chars_mode,iso)} to obtain ISO standard
compatibility.

@item number_codes(?@var{A},?@var{L}) [ISO]
@findex number_codes/2
@syindex number_codes/2
@cnindex number_codes/2
The predicate holds when at least one of the arguments is ground
(otherwise, an error message will be displayed). The argument @var{A}
will be unified with a number and @var{L} with the list of the ASCII
codes for the characters of the external representation of @var{A}.

@item number_atom(?@var{I},?@var{L})
@findex number_atom/2
@snindex number_atom/2
@cnindex number_atom/2

The predicate holds when at least one of the arguments is ground
(otherwise, an error message will be displayed). The argument @var{I} must
be unifiable with a number, and the argument @var{L} must be unifiable
with an atom representing the number.

@item char_code(?@var{A},?@var{I}) [ISO]
@findex char_code/2
@syindex char_code/2
@cnindex char_code/2
The built-in succeeds with @var{A} bound to character represented as an
atom, and @var{I} bound to the character code represented as an
integer. At least, one of either @var{A} or @var{I} must be bound before
the call.

@item sub_atom(+@var{A},?@var{Bef}, ?@var{Size}, ?@var{After}, ?@var{At_out}) [ISO]
@findex sub_atom/5
@snindex sub_atom/5
@cnindex sub_atom/5
True when @var{A} and @var{At_out} are atoms such that the name of
@var{At_out} has size @var{Size} and is a substring of the name of
@var{A}, such that @var{Bef} is the number of characters before and
@var{After} the number of characters afterwards.

Note that @var{A} must always be known, but @var{At_out} can be unbound when
calling this built-in. If all the arguments for @code{sub_atom/5} but @var{A}
are unbound, the built-in will backtrack through all possible
substrings of @var{A}.

@item numbervars(@var{T},+@var{N1},-@var{Nn})
@findex numbervars/3
@syindex numbervars/3
@cnindex numbervars/3
Instantiates each variable in term @var{T} to a term of the form:
@code{'$VAR'(@var{I})}, with @var{I} increasing from @var{N1} to @var{Nn}.

@item ground(@var{T})
@findex ground/1
@syindex ground/1
@cnindex ground/1
Succeeds if there are no free variables in the term @var{T}.

@item arg(+@var{N},+@var{T},@var{A}) [ISO]
@findex arg/3
@syindex arg/3
@cnindex arg/3
Succeeds if the argument @var{N} of the term @var{T} unifies with
@var{A}. The arguments are numbered from 1 to the arity of the term.

The current version will generate an error if @var{T} or @var{N} are
unbound, if @var{T} is not a compound term, of if @var{N} is not a positive
integer. Note that previous versions of YAP would fail silently
under these errors.

@item functor(@var{T},@var{F},@var{N})
@findex functor/3
@syindex functor/3
@cyindex functor/3
The top functor of term @var{T} is named @var{F} and has  arity @var{N}.

When @var{T} is not instantiated, @var{F} and @var{N} must be. If
@var{N} is 0, @var{F} must be an atomic symbol, which will be unified
with @var{T}. If @var{N} is not 0, then @var{F} must be an atom and
@var{T} becomes instantiated to the most general term having functor
@var{F} and arity @var{N}. If @var{T} is instantiated to a term then
@var{F} and @var{N} are respectively unified with its top functor name
and arity.

In the current version of YAP the arity @var{N} must be an
integer. Previous versions allowed evaluable expressions, as long as the
expression would evaluate to an integer. This feature is not available
in the ISO Prolog standard.

@item @var{T} =.. @var{L} [ISO]
@findex =../2
@syindex =../2
@cyindex =../2
The list @var{L} is built with the functor and arguments of the term
@var{T}. If @var{T} is instantiated to a variable, then @var{L} must be
instantiated either to a list whose head is an atom, or to a list
consisting of just a number.

@item @var{X} = @var{Y} [ISO]
@findex =/2
@syindex =/2
@cnindex =/2
Tries to unify terms @var{X} and @var{Y}.

@item @var{X} \= @var{Y} [ISO]
@findex \=/2
@snindex \=/2
@cnindex \=/2
Succeeds if terms @var{X} and @var{Y} are not unifiable.

@item term_variables(?T,+L)
@findex term_variables/2
@syindex term_variables/2
@cnindex term_variables/2
Unify the second argument with the list of variables occurring in @var{T}.

@item term_hash(+@var{T},+@var{Depth},+@var{Max},-@var{O})
@findex term_hash/4
@syindex term_hash/4
@cnindex term_hash/4
Unify the last argument with a number, called the hash function for
term. This integer can be used to implement efficient term access
schemes. The number @var{O} is such that:
@itemize @bullet
@item The function only considers constants and functors from @var{T} up
to depth @var{Depth} are considered. The @emph{depth} of a term is defined as
follows:
@itemize @minus
@item The depth of a constant of is 1.
@item The depth of term is 1 plus the maximum depth of its sub-terms.
@end itemize

@item If a variable is found up to depth @var{Depth}, @var{O} will be
left unbound.

@item If @var{Depth} is bound to -1, the term will be fully
investigated. If the term is bound to an infinite rational tree, the
predicate will loop.

@item If the term includes floating point numbers or integers whose
absolute value is larger than 7829367, than the result is considered to
be implementation defined. Otherwise, the result is platform indepent.

@item The variable @var{Max} must be bound to a positive
integer. If bound, @var{O} will be in the range going from 0 up to, but
not including, @var{Max}.
@end itemize

@item term_hash(+@var{T},-@var{O})
@findex term_hash/2
@syindex term_hash/2
@cnindex term_hash/2
Simplified version of @code{term_hash/4} such that @var{Depth} is bound
to -1 and @var{Max} is bound to a large platform-indepent integer.

@item unify_with_occurs_check(?T1,?T2) [ISO]
@findex unify_with_occurs_check/2
@syindex unify_with_occurs_check/2
@cnindex unify_with_occurs_check/2
Obtain the most general unifier of terms @var{T1} and @var{T2}, if there
is one.

This predicate implements the full unification algorithm. An example:n
@example
unify_with_occurs_check(a(X,b,Z),a(X,A,f(B)).
@end example
@noindent
will succeed with the bindings @code{A = b} and @code{Z = f(B)}. On the
other hand:
@example
unify_with_occurs_check(a(X,b,Z),a(X,A,f(Z)).
@end example
@noindent
would fail, because @code{Z} is not unifiable with @code{f(Z)}. Note that
@code{(=)/2} would succeed for the previous examples, giving the following
bindings @code{A = b} and @code{Z = f(Z)}.


@item copy_term(?@var{TI},-@var{TF}) [ISO]
@findex copy_term/2
@syindex copy_term/2
@cnindex copy_term/2
Term @var{TF} is a variant of the original term @var{TI}, such that for
each variable @var{V} in the term @var{TI} there is a new variable @var{V'}
in term @var{TF}.

@end table

@node Comparing Terms, Arithmetic, Testing Terms, Top
@section Comparing Terms

The following predicates are used to compare and order terms, using the
standard ordering:

@itemize @bullet
@item
variables come before numbers, numbers come before atoms which in turn
come before compound terms, ie: variables @@< numbers @@< atoms @@<
compound terms.
@item
variables are roughly ordered by "age" (the "oldest" variable is put
first);
@item
floating point numbers are sorted in increasing order;
@item
Integers are sorted in increasing order;
@item
atoms are sorted in lexicographic order;
@item
compound terms are ordered first by name, then by arity of the main
functor, and finally by their arguments in left-to-right order.
@end itemize

@table @code

@item compare(@var{C},@var{X},@var{Y})
@findex compare/3
@syindex compare/3
@cyindex compare/3
As a result of comparing @var{X} and @var{Y}, @var{C} may take one of
the following values:

@itemize @bullet
@item
@code{=} if @var{X} and @var{Y} are identical;
@item
@code{<} if @var{X} precedes @var{Y} in the defined order;
@item
@code{>} if @var{Y} precedes @var{X} in the defined order;
@end itemize

@item @var{X} == @var{Y} [ISO]
@findex ==/2
@syindex ==/2
@cyindex ==/2
Succeeds if terms @var{X} and @var{Y} are strictly identical. The
difference between this predicate and @code{=/2} is that, if one of the
arguments is a free variable, it only succeeds when they have already
been unified.

@example
?- X == Y.
@end example
@noindent
fails, but,
@example
?- X = Y, X == Y.
@end example
@noindent
succeeds.
@example
?- X == 2.
@end example
@noindent
fails, but,
@example
?- X = 2, X == 2.
@end example
@noindent
succeeds.


@item @var{X} \== @var{Y} [ISO]
@findex \==/2
@syindex \==/2
@cyindex \==/2
Terms @var{X} and @var{Y} are not strictly identical.

@item @var{X} @@< @var{Y} [ISO]
@findex @@</2
@syindex @@</2
@cyindex @@</2
Term @var{X} precedes term @var{Y} in the standard order.

@item @var{X} @@=< @var{Y} [ISO]
@findex @@=</2
@syindex @@</2
@cyindex @@</2
Term @var{X} does not follow term @var{Y} in the standard order.

@item @var{X} @@> @var{Y} [ISO]
@findex @@>/2
@syindex @@>/2
@cyindex @@>/2
Term @var{X} follows term @var{Y} in the standard order.

@item @var{X} @@>= @var{Y} [ISO]
@findex @@>=/2
@syindex @@>=/2
@cyindex @@>=/2
Term @var{X} does not precede term @var{Y} in the standard order.

@item sort(+@var{L},-@var{S})
@findex sort/2
@syindex sort/2
@cyindex sort/2
Unifies @var{S} with the list obtained by sorting @var{L} and  merging
identical (in the sense of @code{==}) elements.

@item keysort(+@var{L},@var{S})
@findex keysort/2
@syindex keysort/2
@cyindex keysort/2
Assuming L is a list of the form @code{@var{Key}-@var{Value}},
@code{keysort(+@var{L},@var{S})} unifies @var{S} with the list obtained
from @var{L}, by sorting its elements according to the value of
@var{Key}.
@example
?- keysort([3-a,1-b,2-c,1-a,1-b],S).
@end example
@noindent
would return:
@example
S = [1-b,1-a,1-b,2-c,3-a]
@end example

@end table


@node Arithmetic, I/O, Comparing Terms, Top
@section Arithmetic
Arithmetic expressions in YAP may use the following operators
or @i{evaluable predicates}:

@table @code

@item +@var{X}
The value of @var{X} itself.

@item -@var{X} [ISO]
Symmetric value.

@item @var{X}+@var{Y} [ISO]
Sum.

@item @var{X}-@var{Y} [ISO]
Difference.

@item @var{X}*@var{Y} [ISO]
Product.

@item @var{X}/@var{Y} [ISO]
Quotient.

@item @var{X}//@var{Y} [ISO]
Integer quotient.

@item @var{X} mod @var{Y} [ISO]
Integer remainder.

@item @var{X} rem @var{Y}
Integer remainder, the same as @code{mod}.

@item exp(@var{X}) [ISO]
Natural exponential.

@item log(@var{X}) [ISO]
Natural logarithm.

@item log10(@var{X})
Decimal logarithm.

@item sqrt(@var{X}) [ISO]
Square root.

@item sin(@var{X}) [ISO]
Sine.

@item cos(@var{X}) [ISO]
Cosine.

@item tan(@var{X})
Tangent.

@item asin(@var{X})
Arc sine.

@item acos(@var{X})
Arc cosine.

@item atan(@var{X}) [ISO]
Arc tangent.

@item atan2(@var{X})
Four-quadrant arc tangent.

@item sinh(@var{X})
Hyperbolic sine.

@item cosh(@var{X})
Hyperbolic cosine.

@item tanh(@var{X})
Hyperbolic tangent.

@item asinh(@var{X})
Hyperbolic arc sine.

@item acosh(@var{X})
Hyperbolic arc cosine.

@item atanh(@var{X})
Hyperbolic arc tangent.

@item integer(@var{X}) [ISO]
If @var{X} evaluates to a float, the integer between the value of @var{X}
and 0 closest to the value of @var{X}, else if @var{X} evaluates to an
integer, the value of @var{X}.

@item float(@var{X}) [ISO]
If @var{X} evaluates to an integer, the corresponding float, else the float
itself.

@item float_fractional_part(@var{X}) [ISO]
The fractional part of the floating point number @var{X}, or @code{0.0}
if @var{X} is an integer. In the @code{iso} language mode,
@var{X} must be an integer.

@item float_integer_part(@var{X}) [ISO]
The float giving the integer part of the floating point number @var{X},
or @var{X} if @var{X} is an integer. In the @code{iso} language mode,
@var{X} must be an integer.

@item abs(@var{X}) [ISO]
The absolute value of @var{X}.

@item ceiling(@var{X}) [ISO]
The float that is the smallest integral value not smaller than @var{X}.

In @code{iso} language mode the argument must be a floating
point-number and the result is an integer.

@item floor(@var{X}) [ISO]
The float that is the greatest integral value not greater than @var{X}.

In @code{iso} language mode the argument must be a floating
point-number and the result is an integer.

@item round(@var{X}) [ISO]
The nearest integral value to @var{X}. If @var{X} is
equidistant to two integers, it will be rounded to the closest even
integral value.

In @code{iso} language mode the argument must be a floating
point-number, the result is an integer and it the float is equidistant
it is rounded up, that is, to the least integer greater than @var{X}.

@item sign(@var{X}) [ISO]
Return 1 if the @var{X} evaluates to a positive integer, 0 it if
evaluates to 0, and -1 if it evaluates to a negative integer. If @var{X}
evaluates to a floating-point number return 1.0 for a positive @var{X},
0.0 for 0.0, and -1.0 otherwise.

@item truncate(@var{X})
The float that is the integral value between @var{X} and 0 closest to
@var{X}.

@item max(@var{X},@var{Y})
The greater value of @var{X} and @var{Y}.

@item min(@var{X},@var{Y})
The lesser value of @var{X} and @var{Y}.

@item @var{X} ^ @var{Y}
@var{X} raised to the power of @var{Y}, (from the C-Prolog syntax).

@item exp(@var{X},@var{Y})
@var{X} raised to the power of @var{Y}, (from the Quintus Prolog syntax).

@item @var{X} ** @var{Y} [ISO]
@var{X} raised to the power of @var{Y}  (from ISO).

@item @var{X} /\ @var{Y} [ISO]
Integer bitwise conjunction.

@item @var{X} \/ @var{Y} [ISO]
Integer bitwise disjunction.

@item @var{X} # @var{Y} [ISO]
Integer bitwise exclusive disjunction.

@item @var{X} << @var{Y}
Integer bitwise left logical shift of @var{X} by @var{Y} places.

@item @var{X} >> @var{Y} [ISO]
Integer bitwise right logical shift of @var{X} by @var{Y} places.

@item \ @var{X} [ISO]
Integer bitwise negation.

@item gcd(@var{X},@var{Y})
The greatest common divisor of the two integers @var{X} and @var{Y}.

@item msb(@var{X})
The most significant bit of the integer @var{X}.

@item [@var{X}]
Evaluates to @var{X} for expression @var{X}. Useful because character
strings in Prolog are lists of character codes.

@example
X is Y*10+C-"0"
@end example
@noindent
is the same as
@example
X is Y*10+C-[48].
@end example
@noindent
which would be evaluated as:
@example
X is Y*10+C-48.
@end example

@end table

Besides numbers and the arithmetic operators described above, certain
atoms have a special meaning when present in arithmetic expressions:

@table @code
@item pi
The value of @emph{pi}, the ratio of a circle's circumrefence to its
diameter.

@item e
The base of the natural logarithms.

@item inf
Infinity according to the IEEE Floating-Point standard. Note that
evaluating this term will generate a domain error in the @code{iso}
language mode.

@item nan
Not-a-number according to the IEEE Floating-Point standard. Note that
evaluating this term will generate a domain error in the @code{iso}
language mode.

@item cputime
CPU time in seconds, since YAP was invoked.

@item heapused
Heap space used, in bytes.

@item local
Local stack in use, in bytes.

@item global
Global stack in use, in bytes.

@item random
A "random" floating point number between 0 and 1.

@end table

The primitive YAP predicates involving arithmetic expressions are:

@table @code

@item @var{X} is +@var{Y} [2]
@findex is/2
@syindex is/2
@caindex is/2
This predicate succeeds iff the result of evaluating the expression
@var{Y} unifies with @var{X}. This is the predicate normally used to
perform evaluation of arithmetic expressions:

@example
X is 2+3*4
@end example
@noindent
succeeds with @code{X = 14}.

@item +@var{X} < +@var{Y} [ISO]
@findex </2
@syindex </2
@cyindex </2
The value of the expression @var{X} is less than the value of expression
@var{Y}.

@item +@var{X} =< +@var{Y} [ISO]
@findex =</2
@syindex =</2
@cyindex =</2
The value of the expression @var{X} is less than or equal to the value
of expression @var{Y}.


@item +@var{X} > +@var{Y} [ISO]
@findex >/2
@syindex >/2
@cyindex >/2
The value of the expression @var{X} is greater than the value of
expression @var{Y}.

@item +@var{X} >= +@var{Y} [ISO]
@findex >=/2
@syindex >=/2
@cyindex >=/2
The value of the expression @var{X} is greater than or equal to the
value of expression @var{Y}.

@item +@var{X} =:= +@var{Y} [ISO]
@findex =:=/2
@syindex =:=/2
@cyindex =:=/2
The value of the expression @var{X} is equal to the value of expression
@var{Y}.

@item +@var{X} =\= +@var{Y} [ISO]
@findex =\=/2
@syindex =\=/2
@cyindex =\=/2
The value of the expression @var{X} is different from the value of
expression @var{Y}.

@item srandom(+@var{X})
@findex srandom/1
@snindex srandom/1
@cnindex srandom/1
Use the argument @var{X} as a new seed for YAP's random number
generator. The argument should be an integer, but floats are acceptable.
@end table

@noindent
@strong{Notes:}

@itemize @bullet
@item 
In contrast to previous versions of Yap, Yap4 @emph{does not} convert
automatically between integers and floats.
@item
arguments to trigonometric functions are expressed in radians.
@item
if a (non-instantiated) variable occurs in an arithmetic expression
YAP will generate an exception. If no error handler is
available, execution will be thrown back to the top-level.
@end itemize


@node I/O, Database, Arithmetic, Top
@section I/O Predicates

Some of the I/O predicates described below will in certain conditions
provide error messages and abort only if the file_errors flag is set.
If this flag is cleared the same predicates will just fail. Details on
setting and clearing this flag are given under 7.7.

@menu

Subnodes of Input/Output
* Streams and Files:: Handling Streams and Files
* C-Prolog File Handling:: C-Prolog Compatible File Handling
* I/O of Terms:: Input/Output of terms
* I/O of Characters:: Input/Output of Characters
* I/O for Streams:: Input/Output using Streams
* C-Prolog to Terminal:: C-Prolog compatible Character I/O to terminal
* I/O Control:: Controlling your Input/Output
* Sockets:: Using Sockets from Yap

@end menu

@node Streams and Files, C-Prolog File Handling, , I/O
@subsection Handling Streams and Files

@table @code

@item open(+@var{F},+@var{M},-@var{S}) [ISO]
@findex open/3
@syindex open/3
@cnindex open/3
Opens the file with name @var{F} in mode @var{M} ('read', 'write' or
'append'), returning @var{S} unified with the stream name.

At most, there are 17 streams  opened at the same time. Each stream is
either an input or an output stream but not both. There are always 3
open streams:  @code{user_input} for reading, @code{user_output} for writing
and @code{user_error} for writing. If there is no  ambiguity, the atoms
@code{user_input} and @code{user_output} may be referred to as  @code{user}.

The @code{file_errors} flag controls whether errors are reported when in
mode 'read' or 'append' the file @var{F} does not exist or is not
readable, and whether in mode 'write' or 'append' the file is not
writable.

@item open(+@var{F},+@var{M},-@var{S},+@var{Opts}) [ISO]
@findex open/4
@saindex open/4
@cnindex open/4
Opens the file with name @var{F} in mode @var{M} ('read',  'write' or
'append'), returning @var{S} unified with the stream name, and following
these options:

@table @code

@item type(+@var{T})
Specify whether the stream is a @code{text} stream (default), or a
@code{binary} stream.

@item reposition(+@var{Bool})
Specify whether it is possible to reposition the stream (@code{true}), or
not (@code{false}). By default, YAP enables repositioning for all
files, except terminal files and sockets.

@item eof_action(+@var{Action})
Specify the action to take if attempting to input characters from a
stream where we have previously found an @code{end-of-file}. The possible
actions are @code{error}, that raises an error, @code{reset}, that tries to
reset the stream and is used for @code{tty} type files, and @code{eof_code},
which generates a new @code{end-of-file} (default for non-tty files).

@item alias(+@var{Name})
Specify an alias to the stream. The alias @t{Name} must be an atom. The
alias can be used instead of the stream descriptor for every operation
concerning the stream.

The operation will fail and give an error if the alias name is already
in use. YAP allows several aliases for the same file, but only
one is returned by @code{stream_property/2}
@end table

@item close(+@var{S}) [ISO]
@findex close/1
@syindex close/1
@cyindex close/1
Closes the stream @var{S}. If @var{S} does not stand for a stream
currently opened an error is reported. The streams @code{user_input},
@code{user_output}, and @code{user_error} can never be closed.

By default, give a file name, @code{close/1} will also try to close a
corresponding open stream. This feature is not available in ISO or
SICStus languages mode and is deprecated.

@item close(+@var{S},+@var{O}) [ISO]
@findex close/2
@saindex close/2
@cnindex close/2
Closes the stream @var{S}, following options @var{O}. 

The only valid options are @code{force(true)} and @code{force(false)}.
YAP currently ignores these options.

@item absolute_file_name(+@var{Name},-@var{FullPath})
@findex absolute_file_name/2
@syindex absolute_file_name/2
@cnindex absolute_file_name/2
Give the path a full path @var{FullPath} Yap would use to consule a file
named @var{Name}.  Unify @var{FullPath} with @code{user} if the file
name is @code{user}.

@item current_stream(@var{F},@var{M},@var{S})
@findex current_stream/3
@syindex current_stream/3
@cnindex current_stream/3
Defines the relation: The stream @var{S} is opened on the file @var{F}
in mode @var{M}. It might be used to obtain all open streams (by
backtracking) or to access the stream for a file @var{F} in mode
@var{M}, or to find properties for a stream @var{S}.

@item flush_output [ISO]
@findex flush_output/0
@syindex flush_output/0
@cnindex flush_output/0
Send all data in the output buffer to current output stream.

@item flush_output(+@var{S}) [ISO]
@findex flush_output/1
@syindex flush_output/1
@cnindex flush_output/1
Send all data in the output buffer to stream @var{S}.

@item set_input(+@var{S})
@findex set_input/1
@syindex set_input/1
@cnindex set_input/1
Set stream @var{S} as the current input stream. Predicates like @code{read/1}
and @code{get/1} will start using stream @var{S}.

@item set_output(+@var{S})
@findex set_output/1
@syindex set_output/1
@cnindex set_output/1
Set stream @var{S} as the current output stream. Predicates like
@code{write/1} and @code{put/1} will start using stream @var{S}.

@item stream_select(+@var{STREAMS},+@var{TIMEOUT},-@var{READSTREAMS})
@findex stream_select/3
@syindex stream_select/3
@cnindex stream_select/3
Given a list of open @var{STREAMS} openeded in read mode and a @var{TIMEOUT}
return a list of streams who are now available for reading. 

If the @var{TIMEOUT} is instantiated to @code{off},
@code{stream_select/3} will wait indefinitely for a stream to become
open. Otherwise the timeout must be of the form @code{SECS:USECS} where
@code{SECS} is an integer gives the number of seconds to wait for a timeout
and @code{USECS} adds the number of micro-seconds.

This built-in is only defined if the system call @code{select} is
available in the system.

@item current_input(-@var{S}) [ISO]
@findex current_input/1
@syindex current_input/1
@cnindex current_input/1
Unify @var{S} with the current input stream.

@item current_output(-@var{S}) [ISO]
@findex current_output/1
@syindex current_output/1
@cnindex current_output/1
Unify @var{S} with the current output stream.

@item at_end_of_stream [ISO]
@findex at_end_of_stream/0
@syindex at_end_of_stream/0
@cnindex at_end_of_stream/0
Succeed if the current stream has stream position end-of-stream or
past-end-of-stream.

@item at_end_of_stream(+@var{S}) [ISO]
@findex at_end_of_stream/1
@syindex at_end_of_stream/1
@cnindex at_end_of_stream/1
Succeed if the stream @var{S} has stream position end-of-stream or
past-end-of-stream. Note that @var{S} must be a readable stream.

@item set_stream_position(+@var{S}, +@var{POS}) [ISO]
@findex set_stream_position/2
@syindex set_stream_position/2
@cnindex set_stream_position/2
Given a stream position @var{POS} for a stream @var{S}, set the current
stream position for @var{S} to be @var{POS}.

@item stream_property(?@var{Stream},?@var{Prop}) [ISO]
@findex stream_property/2
@snindex stream_property/2
@cnindex stream_property/2

Obtain the properties for the open streams. If the first argument is
unbound, the procedure will backtrack through all open
streams. Otherwise, the first argument must be a stream term (you may
use @code{current_stream} to obtain a current stream given a file name).

The following properties are recognised:

@table @code

@item file_name(@var{P})
An atom giving the file name for the current stream. The file names are
@code{user_input}, @code{user_output}, and @code{user_error} for the
standard streams.

@item mode(@var{P})
The mode used to open the file. It may be one of @code{append},
@code{read}, or @code{write}.

@item input
The stream is readable.

@item output
The stream is writable.

@item alias(@var{A})
ISO-Prolog primitive for stream aliases. @t{Yap} returns one of the
existing aliases for the stream.

@item position(@var{P})
A term describing the position in the stream.

@item end_of_stream(@var{E})
Whether the stream is @code{at} the end of stream, or it has found the
end of stream and is @code{past}, or whether it has @code{not} yet
reached the end of stream.

@item eof_action(@var{A})
The action to take when trying to read after reaching the end of
stream. The action may be one of @code{error}, generate an error,
@code{eof_code}, return character code @code{-1}, or @code{reset} the
stream.

@item reposition(@var{B})
Whether the stream can be repositioned or not, that is, whether it is
seekable.

@item type(@var{T})
Whether the stream is a @code{text} stream or a @code{binary} stream.

@end table

@end table

@node C-Prolog File Handling, I/O of Terms, Streams and Files, I/O
@subsection Handling Streams and Files

@table @code

@item tell(+@var{S})
@findex tell/1
@syindex tell/1
@cyindex tell/1
If @var{S} is a currently opened stream for output, it becomes the
current output stream. If @var{S} is an atom it is taken to be a
filename.  If there is no output stream currently associated with it,
then it is opened for output, and the new output stream created becomes
the current output stream. If it is not possible to open the file, an
error occurs.  If there is a single opened output stream currently
associated with the file, then it becomes the current output stream; if
there are more than one in that condition, one of them is chosen.

Whenever @var{S} is a stream not currently opened for output, an error
may be reported, depending on the state of the file_errors flag. The
predicate just fails, if @var{S} is neither a stream nor an atom.

@item telling(-@var{S})
@findex telling/1
@syindex telling/1
@cyindex telling/1
The current output stream is unified with @var{S}.

@item told
@findex told/0
@syindex told/0
@cyindex told/0
Closes the current output stream, and the user's terminal becomes again
the current output stream. It is important to remember to close streams
after having finished using them, as the maximum number of
simultaneously opened streams is 17.

@item see(+@var{S})
@findex see/1
@syindex see/1
@cyindex see/1
If @var{S} is a currently opened input stream then it is assumed to be
the current input stream. If @var{S} is an atom it is taken as a
filename. If there is no input stream currently associated with it, then
it is opened for input, and the new input stream thus created becomes
the current input stream. If it is not possible to open the file, an
error occurs.  If there is a single opened input stream currently
associated with the file, it becomes the current input stream; if there
are more than one in that condition, then one of them is chosen.

When @var{S} is a stream not currently opened for input, an error may be
reported, depending on the state of the @code{file_errors} flag. If
@var{S} is neither a stream nor an atom the predicates just fails.

@item seeing(-@var{S})
@findex seeing/1
@syindex seeing/1
@cyindex seeing/1
The current input stream is unified with @var{S}.

@item seen
@findex seen/0
@syindex seen/0
@cyindex seen/0
Closes the current input stream (see 6.7.).

@end table

@node I/O of Terms, I/O of Characters, C-Prolog File Handling, I/O
@subsection Handling Input/Output of Terms

@table @code

@item read(-@var{T}) [ISO]
@findex read/1
@syindex read/1
@cyindex read/1
Reads the next term from the current input stream, and unifies it with
@var{T}. The term must be followed by a dot ('.') and any blank-character
as previously defined. The syntax of the term must match the current
declarations for operators (see op). If the end-of-stream is reached, 
@var{T} is unified with the atom @code{end_of_file}. Further reads from of 
the same stream may cause an error failure (see @code{open/3}).

@item read_term(-@var{T},+@var{Options}) [ISO]
@findex read_term/2
@saindex read_term/2
@cnindex read_term/2
Reads term @var{T} from the current input stream with execution
controlled by the following options:

@table @code

@item  singletons(-@var{Names})
@findex singletons/1 (read_term/2 option)
Unify @var{Names} with a list of the form @var{Name=Var}, where
@var{Name} is the name of a non-anonymous singleton variable in the
original term, and @code{Var} is the variable's representation in
YAP.

@item  syntax_errors(+@var{Val})
@findex syntax_errors/1 (read_term/2 option)
Control action to be taken after syntax errors. See @code{yap_flag/2}
for detailed information.

@item  variable_names(-@var{Names})
@findex variable_names/1 (read_term/2 option)
Unify @var{Names} with a list of the form @var{Name=Var}, where @var{Name} is
the name of a non-anonymous variable in the original term, and @var{Var}
is the variable's representation in YAP.

@item  variables(-@var{Names})
@findex variables/1 (read_term/2 option)
Unify @var{Names} with a list of the variables in term @var{T}.

@end table

@item char_conversion(+@var{IN},+@var{OUT}) [ISO]
@findex char_conversion/2
@syindex char_conversion/2
@cnindex char_conversion/2
While reading terms convert unquoted occurrences of the character
@var{IN} to the character @var{OUT}. Both @var{IN} and @var{OUT} must be
bound to single characters atoms.

Character conversion only works if the flag @code{char_conversion} is
on. This is default in the @code{iso} and @code{sicstus} language
modes. As an example, character conversion can be used for instance to
convert characters from the ISO-LATIN-1 character set to ASCII.

If @var{IN} is the same character as @var{OUT}, @code{char_conversion/2}
will remove this conversion from the table.

@item current_char_conversion(?@var{IN},?@var{OUT}) [ISO]
@findex current_char_conversion/2
@syindex current_char_conversion/2
@cnindex current_char_conversion/2
If @var{IN} is unbound give all current character
translations. Otherwise, give the translation for @var{IN}, if one
exists.

@item write(@var{T}) [ISO]
@findex write/1
@syindex write/1
@cyindex write/1
The term @var{T} is written to the current output stream according to
the operator declarations in force.

@item display(+@var{T})
@findex display/1
@syindex display/1
@cyindex display/1
Displays term @var{T} on the current output stream. All Prolog terms are
written in standard parenthesized prefix notation.

@item write_canonical(+@var{T}) [ISO]
@findex display/1
@syindex display/1
@cnindex display/1
Displays term @var{T} on the current output stream. Atoms are quoted
when necessary, and operators are ignored, that is, the term is written
in standard parenthesized prefix notation.

@item write_term(+@var{T}, +@var{Opts}) [ISO]
@findex write_term/2
@syindex write_term/2
@cnindex write_term/2
Displays term @var{T} on the current output stream, according to the
following options:

@table @code
@item quoted(+@var{Bool})
If @code{true}, quote atoms if this would be necessary for the atom to
be recognised as an atom by YAP's parser. The default value is
@code{false}.

@item ignore_ops(+@var{Bool})
If @code{true}, ignore operator declarations when writing the term. The
default value is @code{false}.

@item numbervars(+@var{Bool})
If @code{true}, output terms of the form
@code{'$VAR'(N)}, where @var{N} is an integer, as a sequence of capital
letters. The default value is @code{false}.

@item portrayed(+@var{Bool})
If @code{true}, use @t{portray/1} to portray bound terms. The default
value is @code{false}.

@item max_depth(+@var{Depth})
If @code{Depth} is a positive integer, use @t{Depth} as
the maximum depth to portray a term. The default is @code{0}, that is,
unlimited depth.

@end table

@item writeq(@var{T}) [ISO]
@findex writeq/1
@syindex writeq/1
@cyindex writeq/1
 Writes the term @var{T}, quoting names to make the result acceptable to
the predicate 'read' whenever necessary.

@item print(@var{T})
@findex print/1
@syindex print/1
@cyindex print/1
Prints the term @var{T} to the current output stream using @code{write/1}
unless T is bound and a call to the user-defined  predicate
@code{portray/1} succeeds. To do pretty  printing of terms the user should
define suitable clauses for @code{portray/1} and use @code{print/1}.

@item format(+@var{T},+@var{L})
@findex format/2
@saindex format/2
@cnindex format/2
Print formatted output to the current output stream. The arguments in
list @var{L} are output according to the string or atom @var{T}.

A control sequence is introduced by a @code{w}. The following control
sequences are available in YAP:

@table @code

@item '~~'
Print a single tilde.

@item '~a'
The next argument must be an atom, that will be printed as if by @code{write}.

@item '~Nc'
The next argument must be an integer, that will be printed as a
character code. The number @var{N} is the number of times to print the
character (default 1).

@item '~Ne'
@itemx '~NE'
@itemx '~Nf'
@itemx '~Ng'
@itemx '~NG'
The next argument must be a floating point number. The float @var{F}, the number
@var{N} and the control code @code{c} will be passed to @code{printf} as:

@example
    printf("%s.Nc", F)
@end example

As an example:

@example
?- format("~8e, ~8E, ~8f, ~8g, ~8G~w",
          [3.14,3.14,3.14,3.14,3.14,3.14]).
3.140000e+00, 3.140000E+00, 3.140000, 3.14, 3.143.14
@end example

@item '~Nd'
The next argument must be an integer, and @var{N} is the number of digits
after the decimal point. If @var{N} is @code{0} no decimal points will be
printed. The default is @var{N = 0}.

@example
?- format("~2d, ~d",[15000, 15000]).
150.00, 15000
@end example

@item '~ND'
Identical to @code{'~Nd'}, except that commas are used to separate groups
of three digits.

@example
?- format("~2D, ~D",[150000, 150000]).
1,500.00, 150,000
@end example

@item '~ND'
Identical to @code{'~Nd'}, except that @code{','} is used to separate groups
of three digits.

@example
?- format("~2D, ~D",[150000, 150000]).
1,500.00, 150,000
@end example

@item '~i'
Ignore the next argument in the list of arguments:

@example
?- format('The ~i met the boregrove',[mimsy]).
The  met the boregrove
@end example

@item '~k'
Print the next argument with @code{write_canonical}:

@example
?- format("Good night ~k",a+[1,2]).
Good night +(a,[1,2])
@end example

@item '~Nn'
Print @var{N} newlines (where @var{N} defaults to 1).

@item '~NN'
Print @var{N} newlines if at the beginning of the line (where @var{N}
defaults to 1).

@item '~Nr'
The next argument must be an integer, and @var{N} is interpreted as a
radix, such that @code{2 <= N <= 36} (the default is 8).

@example
?- format("~2r, 0x~16r, ~r",
          [150000, 150000, 150000]).
100100100111110000, 0x249f0, 444760
@end example

@noindent
Note that the letters @code{a-z} denote digits larger than 9.

@item '~NR'
Similar to '~NR'. The next argument must be an integer, and @var{N} is
interpreted as a radix, such that @code{2 <= N <= 36} (the default is 8).

@example
?- format("~2r, 0x~16r, ~r",
          [150000, 150000, 150000]).
100100100111110000, 0x249F0, 444760
@end example

@noindent
The only difference is that letters @code{A-Z} denote digits larger than 9.

@item '~p'
Print the next argument with @code{print/1}:

@example
?- format("Good night ~p",a+[1,2]).
Good night a+[1,2]
@end example

@item '~q'
Print the next argument with @code{writeq/1}:

@example
?- format("Good night ~q",'Hello'+[1,2]).
Good night 'Hello'+[1,2]
@end example

@item '~Ns'
The next argument must be a list of character codes. The system then
outputs their representation as a string, where @var{N} is the maximum
number of characters for the string (@var{N} defaults to the length of the
string).

@example
?- format("The ~s are ~4s",["woods","lovely"]).
The woods are love
@end example

@item '~w'
Print the next argument with @code{writeq/1}:

@example
?- format("Good night ~w",'Hello'+[1,2]).
Good night Hello+[1,2]
@end example

@end table

The @code{format/2} built-in also allows for formatted output.  One can
specify column boundaries and fill the intermediate space by a padding
character: 

@table @code
@item '~N|'
Set a column boundary at position @var{N}, where @var{N} defaults to the
current position.

@item '~N+'
Set a column boundary at @var{N} characters past the current position, where
@var{N} defaults to @code{8}.


@item '~Nt'
Set padding for a column, where @var{N} is the fill code (default is
@key{SPC}).

@end table

The next example shows how to align columns and padding. We first show
left-alignment:

@example

@code{
   ?- format("~n*Hello~16+*~n",[]).
*Hello          *
}
@end example

Note that we reserve 16 characters for the column.

The following example shows how to do right-alignment:


@example
@code{
   ?- format("*~tHello~16+*~n",[]).
*          Hello*
}

@end example


The @code{~t} escape sequence forces filling before @code{Hello}. 

We next show how to do centering:

@example
@code{
   ?- format("*~tHello~t~16+*~n",[]).
*     Hello     *
}
@end example


The two @code{~t} escape sequence force filling both before and after
@code{Hello}. Space is then evenly divided bewteen the right and the
left sides.

@item format(+@var{S},+@var{T},+@var{L})
@findex format/3
@saindex format/3
@cnindex format/3
Print formatted output to stream @var{S}.


@end table

@node I/O of Characters, I/O for Streams, I/O of Terms, I/O
@subsection Handling Input/Output of Characters

@table @code

@item put(+@var{N})
@findex put/1
@syindex put/1
@cyindex put/1
Outputs to the current output stream the character whose ASCII code is
@var{N}. The character @var{N} must be a legal ASCII character code, an
expression yielding such a code, or a list in which case only the first
element is used.

@item put_byte(+@var{N}) [ISO]
@findex put_byte/1
@snindex put_byte/1
@cnindex put_byte/1
Outputs to the current output stream the character whose code is
@var{N}. The current output stream must be a binary stream.

@item put_char(+@var{N}) [ISO]
@findex put_char/1
@snindex put_char/1
@cnindex put_char/1
Outputs to the current output stream the character who is used to build
the representation of atom @code{A}. The current output stream must be a
text stream.

@item put_code(+@var{N}) [ISO]
@findex put_code/1
@snindex put_code/1
@cnindex put_code/1
Outputs to the current output stream the character whose ASCII code is
@var{N}. The current output stream must be a text stream. The character
@var{N} must be a legal ASCII character code, an expression yielding such
a code, or a list in which case only the first element is used.

@item get(-@var{C})
@findex get/1
@syindex get/1
@cyindex get/1
The next non-blank character from the current input stream is unified
with @var{C}. Blank characters are the ones whose ASCII codes are not
greater than 32. If there are no more non-blank characters in the
stream, @var{C} is unified with -1. If @code{end_of_stream} has already
been reached in the previous reading, this call will give an error message.

@item get0(-@var{C})
@findex get0/1
@syindex get0/1
@cyindex get0/1
The next character from the current input stream is consumed, and then
unified with @var{C}. There are no restrictions on the possible
values of the ASCII code for the character, but the character will be
internally converted by YAP.

@item get_byte(-@var{C}) [ISO]
@findex get_byte/1
@snindex get_byte/1
@cnindex get_byte/1
If @var{C} is unbound, or is a character code, and the current stream is a
binary stream, read the next byte from the current stream and unify its
code with @var{C}.

@item get_char(-@var{C}) [ISO]
@findex get_char/1
@snindex get_char/1
@cnindex get_char/1
If @var{C} is unbound, or is an atom representation of a character, and
the current stream is a text stream, read the next character from the
current stream and unify its atom representation with @var{C}.

@item get_code(-@var{C}) [ISO]
@findex get_code/1
@snindex get_code/1
@cnindex get_code/1
If @var{C} is unbound, or is the code for a character, and
the current stream is a text stream, read the next character from the
current stream and unify its code with @var{C}.

@item peek_byte(-@var{C}) [ISO]
@findex peek_byte/1
@snindex peek_byte/1
@cnindex peek_byte/1
If @var{C} is unbound, or is a character code, and the current stream is a
binary stream, read the next byte from the current stream and unify its
code with @var{C}, while leaving the current stream position unaltered.

@item peek_char(-@var{C}) [ISO]
@findex peek_char/1
@syindex peek_char/1
@cnindex peek_char/1
If @var{C} is unbound, or is an atom representation of a character, and
the current stream is a text stream, read the next character from the
current stream and unify its atom representation with @var{C}, while
leaving the current stream position unaltered.

@item peek_code(-@var{C}) [ISO]
@findex peek_code/1
@snindex peek_code/1
@cnindex peek_code/1
If @var{C} is unbound, or is the code for a character, and
the current stream is a text stream, read the next character from the
current stream and unify its code with @var{C}, while
leaving the current stream position unaltered.

@item skip(+@var{N})
@findex skip/1
@syindex skip/1
@cyindex skip/1
Skips input characters until the next occurrence of the character with
ASCII code @var{N}. The argument to this predicate can take the same forms
as those for @code{put} (see 6.11).

@item tab(+@var{N})
@findex tab/1
@syindex tab/1
@cyindex tab/1
Outputs @var{N} spaces to the current output stream.

@item nl [ISO]
@findex nl/0
@syindex nl/0
@cyindex nl/0
Outputs a new line to the current output stream.

@end table

@node I/O for Streams, C-Prolog to Terminal, I/O of Characters, I/O
@subsection Input/Output Predicates applied to Streams

@table @code

@item read(+@var{S},-@var{T}) [ISO]
@findex read/2
@syindex read/2
@cnindex read/2
Reads term @var{T} from the stream @var{S} instead of from the current input
stream.

@item read_term(+@var{S},-@var{T},+@var{Options}) [ISO]
@findex read_term/3
@saindex read_term/3
@cnindex read_term/3
Reads term @var{T} from stream @var{S} with execution controlled by the
same options as @code{read_term/2}.

@item write(+@var{S},@var{T}) [ISO]
@findex write/2
@syindex write/2
@cnindex write/2
Writes term @var{T} to stream @var{S} instead of to the current output
stream.

@item write_canonical(+@var{S},+@var{T}) [ISO]
@findex display/1
@syindex display/1
@cnindex display/1
Displays term @var{T} on the stream @var{S}. Atoms are quoted when
necessary, and operators are ignored.

@item write_term(+@var{S}, +@var{T}, +@var{Opts}) [ISO]
@findex write_term/3
@syindex write_term/3
@cnindex write_term/3
Displays term @var{T} on the current output stream, according to the same
options used by @code{write_term/3}.

@item writeq(+@var{S},@var{T}) [ISO]
@findex writeq/2
@syindex writeq/2
@cnindex writeq/2
As @code{writeq/1}, but the output is sent to the stream @var{S}.

@item display(+@var{S},@var{T})
@findex display/2
@syindex display/2
@cnindex display/2
Like @code{display/1}, but using stream @var{S} to display the term.

@item print(+@var{S},@var{T})
@findex print/2
@syindex print/2
@cnindex print/2
Prints term @var{T} to the stream @var{S} instead of to the current output
stream.

@item put(+@var{S},+@var{N})
@findex put/2
@syindex put/2
@cnindex put/2
As @code{put(N)}, but to stream @var{S}.

@item put_byte(+@var{S},+@var{N}) [ISO]
@findex put_byte/2
@snindex put_byte/2
@cnindex put_byte/2
As @code{put_byte(N)}, but to binary stream @var{S}.

@item put_char(+@var{S},+@var{A}) [ISO]
@findex put_char/2
@snindex put_char/2
@cnindex put_char/2
As @code{put_char(A)}, but to text stream @var{S}.

@item put_code(+@var{S},+@var{N}) [ISO]
@findex put_code/2
@snindex put_code/2
@cnindex put_code/2
As @code{put_code(N)}, but to text stream @var{S}.

@item get(+@var{S},-@var{C})
@findex get/2
@syindex get/2
@cnindex get/2
The same as @code{get(C)}, but from stream @var{S}.

@item get0(+@var{S},-@var{C})
@findex get0/2
@syindex get0/2
@cnindex get0/2
The same as @code{get0(C)}, but from stream @var{S}.

@item get_byte(+@var{S},-@var{C}) [ISO]
@findex get_byte/2
@snindex get_byte/2
@cnindex get_byte/2
If @var{C} is unbound, or is a character code, and the stream @var{S} is a
binary stream, read the next byte from that stream and unify its
code with @var{C}.

@item get_char(+@var{S},-@var{C}) [ISO]
@findex get_char/2
@snindex get_char/2
@cnindex get_char/2
If @var{C} is unbound, or is an atom representation of a character, and
the stream @var{S} is a text stream, read the next character from that
stream and unify its representation as an atom with @var{C}.

@item get_code(+@var{S},-@var{C}) [ISO]
@findex get_code/2
@snindex get_code/2
@cnindex get_code/2
If @var{C} is unbound, or is a character code, and the stream @var{S} is a
text stream, read the next character from that stream and unify its
code with @var{C}.

@item peek_byte(+@var{S},-@var{C}) [ISO]
@findex peek_byte/2
@snindex peek_byte/2
@cnindex peek_byte/2
If @var{C} is unbound, or is a character code, and @var{S} is a binary
stream, read the next byte from the current stream and unify its code
with @var{C}, while leaving the current stream position unaltered.

@item peek_char(+@var{S},-@var{C}) [ISO]
@findex peek_char/2
@snindex peek_char/2
@cnindex peek_char/2
If @var{C} is unbound, or is an atom representation of a character, and
the stream @var{S} is a text stream, read the next character from that
stream and unify its representation as an atom with @var{C}, while leaving
the current stream position unaltered.

@item peek_code(+@var{S},-@var{C}) [ISO]
@findex peek_code/2
@snindex peek_code/2
@cnindex peek_code/2
If @var{C} is unbound, or is an atom representation of a character, and
the stream @var{S} is a text stream, read the next character from that
stream and unify its representation as an atom with @var{C}, while leaving
the current stream position unaltered.

@item skip(+@var{S},-@var{C})
@findex skip/2
@syindex skip/2
@cnindex skip/2
Like @code{skip/1}, but using stream @var{S} instead of the current
input stream.

@item tab(+@var{S},+@var{N})
@findex tab/2
@syindex tab/2
@cnindex tab/2
The same as @code{tab/1}, but using stream @var{S}.

@item nl(+@var{S})
@findex nl/1
@syindex nl/1
@cnindex nl/1
Outputs a new line to stream @var{S}.

@end table

@node C-Prolog to Terminal, I/O Control, I/O for Streams, I/O
@subsection Compatible C-Prolog predicates for Terminal I/O

@table @code

@item ttyput(+@var{N})
@findex ttyput/1
@syindex ttyput/1
@cnindex ttyput/1
As @code{put(N)} but always to @code{user_output}.

@item ttyget(-@var{C})
@findex ttyget/1
@syindex ttyget/1
@cnindex ttyget/1
The same as @code{get(C)}, but from stream @code{user_input}.

@item ttyget0(-@var{C})
@findex ttyget0/1
@syindex ttyget0/1
@cnindex ttyget0/1
The same as @code{get0(C)}, but from stream @code{user_input}.

@item ttyskip(-@var{C})
@findex ttyskip/1
@syindex ttyskip/1
@cnindex ttyskip/1
Like @code{skip/1}, but always using stream @code{user_input}.
stream.

@item ttytab(+@var{N})
@findex ttytab/1
@syindex ttytab/1
@cnindex ttytab/1
The same as @code{tab/1}, but using stream @code{user_output}.

@item ttynl
@findex ttynl/0
@syindex ttynl/0
@cnindex ttynl/0
Outputs a new line to stream @code{user_output}.

@end table

@node I/O Control, Sockets, C-Prolog to Terminal, I/O
@subsection Controlling Input/Output

@table @code

@item exists(+@var{F})
@findex exists/1
@snindex exists/1
@cyindex exists/1
Checks if file @var{F} exists in the current directory.

@item nofileerrors
@findex nofileerrors/0
@syindex nofileerrors/0
@cyindex nofileerrors/0
Switches off the file_errors flag, so that the predicates @code{see/1},
@code{tell/1}, @code{open/3} and @code{close/1} just fail, instead of producing
an error message and aborting whenever the specified file cannot be
opened or closed.

@item fileerrors
@findex fileerrors/0
@syindex fileerrors/0
@cyindex fileerrors/0
Switches on the file_errors flag so that in certain error conditions
I/O predicates will produce an appropriated message and abort.

@item write_depth(@var{T},@var{L})
@findex write_depth/2
@snindex write_depth/2
@cnindex write_depth/2
Unifies @var{T} and L, respectively, with the values of  the maximum depth
of a term and the maximum length of a list, that will be used by
@code{write/1} or @code{write/2}. The default value for both arguments is 0,
meaning unlimited depth and length.

@example
?- write_depth(3,5).
yes
?- write(a(b(c(d(e(f(g))))))).
a(b(c(....)))
yes
?- write([1,2,3,4,5,6,7,8]).
[1,2,3,4,5,...]
yes
@end example

@item always_prompt_user
@findex always_prompt_user/0
@snindex always_prompt_user/0
@cnindex always_prompt_user/0
Force the system to prompt the user even if the @code{user_input} stream
is not a terminal. This command is useful if you want to obtain
interactive control from a pipe or a socket.

@end table

@node Sockets, , I/O Control, I/O
@subsection Using Sockets From Yap

YAP includes a SICStus Prolog compatible socket interface. This
is a low level interface that provides direct access to the major socket
system calls. These calls can be used both to open a new connection in
the network or connect to a networked server. Socket connections are
described as read/write streams, and standard I/O builtins can be used
to write on or read from sockets. The following calls are available:

@table @code

@item socket(+@var{DOMAIN},+@var{TYPE},+@var{PROTOCOL},-@var{SOCKET})
@findex socket/4
@syindex socket/4
@cnindex socket/4
Corresponds to the BSD system call @code{socket}. Create a socket for
domain @var{DOMAIN} of type @var{TYPE} and protocol
@var{PROTOCOL}. Both @var{DOMAIN} and @var{TYPE} should be atoms,
whereas @var{PROTOCOL} must be an integer. The new socket object is
accessible through a descriptor bound to the variable @var{SOCKET}.

The current implementation of YAP only accepts two socket
domains: @code{'AF_INET'} and @code{'AF_UNIX'}. Socket types depend on the
underlying operating system, but at least the following types are
supported: @code{'SOCK_STREAM'} and @code{'SOCK_DGRAM'}.

@item socket(+@var{DOMAIN},-@var{SOCKET})
@findex socket/2
@syindex socket/2
@cnindex socket/2

Call @code{socket/4} with @var{TYPE} bound to @code{'SOCK_STREAM'} and
@var{PROTOCOL} bound to @code{0}.

@item socket_close(+@var{SOCKET})
@findex socket_close/1
@syindex socket_close/1
@cnindex socket_close/1

Close socket @var{SOCKET}. Note that sockets used in
@code{socket_connet} (that is, client sockets) should not be closed with
@code{socket_close}, as they will be automatically closed when the
corresponding stream is closed with @code{close/1} or @code{close/2}.

@item socket_bind(+@var{SOCKET}, ?@var{PORT})
@findex socket_bind/2
@syindex socket_bind/2
@cnindex socket_bind/2

Interface to system call @code{bind}, as used for servers: bind socket
to a port. Port information depends on the domain:
@table @code
@item 'AF_UNIX'(+@var{FILENAME})
@item 'AF_FILE'(+@var{FILENAME})
use file name @var{FILENAME} for UNIX or local sockets.

@item 'AF_INET'(?@var{HOST},?PORT)
If @var{HOST} is bound to an atom, bind to host @var{HOST}, otherwise
if unbound bind to local host (@var{HOST} remains unbound). If port
@var{PORT} is bound to an integer, try to bind to the corresponding
port. If variable @var{PORT} is unbound allow operating systems to
choose a port number, which is unified with @var{PORT}.

@end table

@item socket_connect(+@var{SOCKET}, +@var{PORT}, -@var{STREAM})
@findex socket_connect/3
@syindex socket_connect/3
@cnindex socket_connect/3

Interface to system call @code{connect}, used for clients: connect
socket @var{SOCKET} to @var{PORT}. The connection results in the
read/write stream @var{STREAM}.

Port information depends on the domain:
@table @code
@item 'AF_UNIX'(+@var{FILENAME})
@item 'AF_FILE'(+@var{FILENAME})
connect to socket at file @var{FILENAME}.

@item 'AF_INET'(+@var{HOST},+@var{PORT})
Connect to socket at host @var{HOST} and port @var{PORT}.
@end table

@item socket_listen(+@var{SOCKET}, +@var{LENGTH})
@findex socket_listen/2
@syindex socket_listen/2
@cnindex socket_listen/2
Interface to system call @code{listen}, used for servers to indicate
willingness to wait for connections at socket @var{SOCKET}. The
integer @var{LENGTH} gives the queue limit for incoming connections,
and should be limited to @code{5} for portable applications. The socket
must be of type @code{SOCK_STREAM} or @code{SOCK_SEQPACKET}.

@item socket_accept(+@var{SOCKET}, -@var{STREAM})
@findex socket_accept/2
@syindex socket_accept/2
@cnindex socket_accept/2

@item socket_accept(+@var{SOCKET}, -@var{CLIENT}, -@var{STREAM})
@findex socket_accept/3
@syindex socket_accept/3
@cnindex socket_accept/3
Interface to system call @code{accept}, used for servers to wait for
connections at socket @var{SOCKET}. The stream descriptor @var{STREAM}
represents the resulting connection.  If the socket belongs to the
domain @code{'AF_INET'}, @var{CLIENT} unifies with an atom containing
the IP address for the client in numbers and dots notation.

@item socket_accept(+@var{SOCKET}, -@var{STREAM})
@findex socket_accept/2
@syindex socket_accept/2
@cnindex socket_accept/2
Accept a connection but do not return client information.

@item socket_buffering(+@var{SOCKET}, -@var{MODE}, -@var{OLD}, +@var{NEW})
@findex socket_buffering/4
@syindex socket_buffering/4
@cnindex socket_buffering/4
Set buffering for @var{SOCKET} in @code{read} or @code{write}
@var{MODE}. @var{OLD} is unified with the previous status, and @var{NEW}
receives the new status which may be one of @code{unbuf} or
@code{fullbuf}.

@item socket_select(+@var{SOCKETS}, -@var{NEWSTREAMS}, +@var{TIMEOUT}, +@var{STREAMS}, -@var{READSTREAMS})
@findex socket_select/5
@syindex socket_select/5
@cnindex socket_select/5
Interface to system call @code{select}, used for servers to wait for
connection requests or for data at sockets. The variable
@var{SOCKETS} is a list of form @var{KEY-SOCKET}, where @var{KEY} is
an user-defined identifier and @var{SOCKET} is a socket descriptor. The
variable @var{TIMEOUT} is either @code{off}, indicating execution will
wait until something is available, or of the form @var{SEC-USEC}, where
@var{SEC} and @var{USEC} give the seconds and microseconds before
@code{socket_select/5} returns. The variable @var{SOCKETS} is a list of
form @var{KEY-STREAM}, where @var{KEY} is an user-defined identifier
and @var{STREAM} is a stream descriptor

Execution of @code{socket_select/5} unifies @var{READSTREAMS} from
@var{STREAMS} with readable data, and @var{NEWSTREAMS} with a list of
the form @var{KEY-STREAM}, where @var{KEY} was the key for a socket
with pending data, and @var{STREAM} the stream descriptor resulting
from accepting the connection.  

@item current_host(?@var{HOSTNAME})
Unify @var{HOSTNAME} with an atom representing the fully qualified
hostname for the current host. Also succeeds if @var{HOSTNAME} is bound
to the unqualified hostname.

@item hostname_address(?@var{HOSTNAME},?@var{IP_ADDRESS})
@var{HOSTNAME} is an host name and @var{IP_ADDRESS} its IP
address in number and dots notation.


@end table

@node Database, Sets, I/O, Top
@section Using the Clausal Data Base

Predicates in YAP may be dynamic or static. By default, when
consulting or reconsulting, predicates are assumed to be static:
execution is faster and the code will probably use less space.
Static predicates impose some restrictions: in general there can be no 
addition or removal of  clauses for a procedure if it is being used in the
current execution.

Dynamic predicates allow programmers to change the Clausal Data Base with
the same flexibility as in C-Prolog. With dynamic predicates it is
always possible to add or remove clauses during execution and the
semantics will be the same as for C-Prolog. But the programmer should be
aware of the fact that asserting or retracting are still expensive operations, 
and therefore he should try to avoid them whenever possible.

@table @code

@item dynamic +@var{P}
@findex dynamic/1
@saindex dynamic/1
@cnindex dynamic/1
Declares predicate @var{P} or list of predicates [@var{P1},...,@var{Pn}]
as a dynamic predicate. @var{P} must be written in form:
@var{name/arity}.

@example
:- dynamic god/1.
@end example

@noindent 
a more convenient form can be used:

@example
:- dynamic son/3, father/2, mother/2.
@end example

or, equivalently,

@example
:- dynamic [son/3, father/2, mother/2].
@end example

@noindent
Note:

a predicate is assumed to be dynamic when 
asserted before being defined.

@item dynamic_predicate(+@var{P},+@var{Semantics})
@findex dynamic_predicate/2
@snindex dynamic_predicate/2
@cnindex dynamic_predicate/2
Declares predicate @var{P} or list of predicates [@var{P1},...,@var{Pn}]
as a dynamic predicate following either @code{logical} or
@code{immediate} semantics.

@menu

Subnodes of Database
* Modifying the Database:: Asserting and Retracting
* Looking at the Database:: Finding out what is in the Data Base
* Database References:: Using Data Base References
* Internal Database:: Yap's Internal Database
* BlackBoard:: Storing and Fetching Terms in the BlackBoard

@end menu

@end table

@node Modifying the Database, Looking at the Database, , Database
@subsection Modification of the Data Base

These predicates can be used either for static or for dynamic
predicates:

@table @code

@item assert(+@var{C})
@findex assert/1
@saindex assert/1
@caindex assert/1
 Adds clause @var{C} to the program. If the predicate is undefined,
declare it as dynamic.

 Most Prolog systems only allow asserting clauses for dynamic
predicates. This is also as specified in the ISO standard. YAP allows
asserting clauses for static predicates, as long as the predicate is not
in use and the language flag is @t{cprolog}. Note that this feature is
deprecated, if you want to assert clauses for static procedures you
should use @code{assert_static/1}.

@item asserta(+@var{C}) [ISO]
@findex asserta/1
@saindex asserta/1
@caindex asserta/1
 Adds clause @var{C} to the beginning of the program. If the predicate is
undefined, declare it as dynamic.

@item assertz(+@var{C}) [ISO]
@findex assertz/1
@saindex assertz/1
@caindex assertz/1
 Adds clause @var{C} to the end of the program. If the predicate is
undefined, declare it as dynamic.

 Most Prolog systems only allow asserting clauses for dynamic
predicates. This is also as specified in the ISO
standard. YAP allows asserting clauses for static predicates,
as long as the predicate is not in use. The current version of
YAP supports this feature, but this feature is deprecated and
support may go away in future versions.

@item abolish(+@var{PredSpec}) [ISO]
@findex abolish/1
@saindex abolish/1
@caindex abolish/1
   Deletes the predicate given by @var{PredSpec} from the database. If
@var{PredSpec} is an unbound variable, delete all predicates for the
current module. The
specification must include the name and arity, and it may include module
information. Under @t{iso} language mode this builtin will only abolish
dynamic procedures. Under other modes it will abolish any procedures, as
long as they are not currently in use. 

@item abolish(+@var{P},+@var{N})
@findex abolish/2
@saindex abolish/2
@caindex abolish/2
 Deletes the predicate with name @var{P} and arity @var{N}. It will remove
both static and dynamic predicates.

@item assert_static(:@var{C})
@findex assert_static/1
@snindex assert_static/1
@cnindex assert_static/1
 Adds clause @var{C} to a static procedure. Asserting a static clause
for a predicate while choice-points for the predicate are availabe has
undefined results.

@item asserta_static(:@var{C})
@findex asserta_static/1
@snindex asserta_static/1
@cnindex asserta_static/1
 Adds clause @var{C} to the beginning of a static procedure. 

@item assertz_static(:@var{C})
@findex assertz_static/1
@snindex assertz_static/1
@cnindex assertz_static/1
 Adds clause @var{C} to the end of a static procedure.  Asserting a
static clause for a predicate while choice-points for the predicate are
availabe has undefined results.

@end table

The following predicate can be used for dynamic predicates and for static
predicates, but only if source mode was on when they were compiled:

@table @code

@item clause(+@var{H},@var{B}) [ISO]
@findex clause/2
@saindex clause/2
@caindex clause/2
  A clause whose head matches @var{H} is searched for in the
program. Its head and body are respectively unified with @var{H} and
@var{B}. If the clause is a unit clause, @var{B} is unified with
@var{true}.

This predicate is applicable to static procedures compiled with
@code{source} active, and to all the dynamic procedures.

@end table

The following predicate can only be used for dynamic predicates:

@table @code

@item retract(+@var{C}) [ISO]
@findex retract/1
@saindex retract/1
@cnindex retract/1
Erases the first clause in the program that matches @var{C}. This
predicate may also be used for the static predicates that have been
compiled when the source mode was @code{on}. For more information on
@code{source/0} (@pxref{Setting the Compiler}).

@item retractall(+@var{G})
@findex retractall/1
@saindex retractall/1
@cnindex retractall/1
Retract all the clauses whose head matches the goal @var{G}. Goal
@var{G} must be a call to a dynamic predicate.

@end table

@node Looking at the Database, Database References, Modifying the Database, Database
@subsection Looking at the Data Base

@table @code

@item listing
@findex listing/0
@saindex listing/0
@caindex listing/0
Lists in the current output stream all the clauses for which source code
is available (these include all clauses for dynamic predicates and
clauses for static predicates compiled when source mode was @code{on}).

@item listing +@var{P}
@findex listing/1
@syindex listing/1
@caindex listing/1
Lists predicate @var{P} if its source code is available.

@item portray_clause(+@var{C})
@findex portray_clause/1
@syindex portray_clause/1
@cnindex portray_clause/1
Write clause @var{C} as if written by @code{listing/0}.

@item portray_clause(+@var{S},+@var{C})
@findex portray_clause/2
@syindex portray_clause/2
@cnindex portray_clause/2
Write clause @var{C} on stream @var{S} as if written by @code{listing/0}.

@item current_atom(@var{A})
@findex current_atom/1
@syindex current_atom/1
@cyindex current_atom/1
Checks whether @var{A} is a currently defined atom. It is used to find all
currently defined atoms by backtracking.

@item current_predicate(@var{F}) [ISO]
@findex current_predicate/1
@syindex current_predicate/1
@cyindex current_predicate/1
@var{F} is the predicate indicator for a currently defined user or
library predicate. @var{F} is of the form @var{Na/Ar}, where the atom
@var{Na} is the name of the predicate, and @var{Ar} its arity.

@item current_predicate(@var{A},@var{P})
@findex current_predicate/2
@syindex current_predicate/2
@cnindex current_predicate/2
Defines the relation: @var{P} is a currently defined predicate whose
name is the atom @var{A}.

@item system_predicate(@var{A},@var{P})
@findex system_predicate/2
@syindex system_predicate/2
@cnindex system_predicate/2
Defines the relation:  @var{P} is a built-in predicate whose name
is the atom @var{A}.

@item predicate_property(@var{P},@var{Prop})
@findex predicate_property/2
@saindex predicate_property/2
@cnindex predicate_property/2
For the predicates obeying the specification @var{P} unify @var{Prop}
with a property of @var{P}. These properties may be:
@table @code
@item built_in
true for built-in predicates,
@item dynamic
true if the predicate is dynamic
@item static
true if the predicate is static
@item meta_predicate(@var{M})
true if the predicate has a meta_predicate declaration @var{M}.
@item multifile
true if the predicate was declared to be multifile
@item imported_from(@var{Mod})
true if the predicate was imported from module @var{Mod}.
@item exported
true if the predicate is exported in the current module.
@item public
true if the predicate is public; note that all dynamic predicates are
public.
@item source
true if source for the predicate is available.
@end table

@end table

@node Database References, Internal Database, Looking at the Database, Database
@subsection Using Data Base References

Data Base references are a fast way of accessing terms. The predicates
@code{erase/1} and @code{instance/1} also apply to these references and may
sometimes be used instead of @code{retract/1} and @code{clause/2}.

@table @code

@item assert(+@var{C},-@var{R})
@findex assert/2
@saindex assert/2
@caindex assert/2
 The same as @code{assert(C)} (@pxref{Modifying the Database}) but
unifies @var{R} with the  database reference that identifies the new
clause, in a one-to-one way. Note that @code{asserta/2} only works for dynamic
predicates. If the predicate is undefined, it will automatically be
declared dynamic.

@item asserta(+@var{C},-@var{R})
@findex asserta/2
@saindex asserta/2
@caindex asserta/2
 The same as @code{asserta(C)} but unifying @var{R} with
the  database reference that identifies the new clause, in a 
one-to-one way. Note that @code{asserta/2} only works for dynamic
predicates. If the predicate is undefined, it will automatically be
declared dynamic.

@item assertz(+@var{C},-@var{R})
@findex assertz/2
@saindex assertz/2
@caindex assertz/2
 The same as @code{assertz(C)} but unifying @var{R} with
the  database reference that identifies the new clause, in a 
one-to-one way. Note that @code{asserta/2} only works for dynamic
predicates. If the predicate is undefined, it will automatically be
declared dynamic.

@item retract(+@var{C},-@var{R})
@findex retract/2
@saindex retract/2
@caindex retract/2
 Erases from the program the clause @var{C} whose 
database reference is @var{R}.


@item clause(+H,B,-@var{R})
@findex clause/3
@saindex clause/3
@caindex clause/3
 The same as @code{clause(H,B)} but @var{R} is unified with  the
reference to the clause in the database.

@end table

@node Internal Database, BlackBoard, Database References, Database
@section Internal Data Base
Some programs need global information for, eg.,  counting or collecting 
data obtained by backtracking. As a rule, to keep this information, the
internal data base should be used instead of asserting and retracting
clauses (as most novice programmers  do), .
In YAP (as in some other Prolog systems) the internal data base (i.d.b. 
for short) is faster, needs less space and provides a better insulation of 
program and data than using asserted/retracted clauses.
The i.d.b. is implemented as a set of terms, accessed by keys that 
unlikely what happens in (non-Prolog) data bases are not part of the 
term. Under each key a list of terms is kept. References are provided so that 
terms can be identified: each term in the i.d.b. has a unique reference 
(references are also available for clauses of dynamic predicates).

@table @code

@item recorda(+@var{K},@var{T},-@var{R})
@findex recorda/3
@saindex recorda/3
@cyindex recorda/3
Makes term @var{T} the first record under key @var{K} and  unifies @var{R}
with its reference.

@item recordz(+@var{K},@var{T},-@var{R})
@findex recordz/3
@saindex recordz/3
@cyindex recordz/3
Makes term @var{T} the last record under key @var{K} and unifies @var{R}
with its reference.

@item recordaifnot(+@var{K},@var{T},-@var{R})
@findex recordaifnot/3
@saindex recordaifnot/3
@cnindex recordaifnot/3
If a term equal to @var{T} up to variable renaming is stored under key
@var{K} fail. Otherwise, make term @var{T} the first record under key
@var{K} and unify @var{R} with its reference.

@item recordzifnot(+@var{K},@var{T},-@var{R})
@findex recorda/3
@saindex recorda/3
@cnindex recorda/3
If a term equal to @var{T} up to variable renaming is stored under key
@var{K} fail. Otherwise, make term @var{T} the first record under key
@var{K} and unify @var{R} with its reference.

@item recorded(+@var{K},@var{T},@var{R})
@findex recorded/3
@saindex recorded/3
@cyindex recorded/3
Searches in the internal database under the key @var{K}, a term that
unifies with @var{T} and whose reference matches @var{R}.

@item erase(+@var{R})
@findex erase/1
@saindex erase/1
@cyindex erase/1
The term referred to by @var{R} is erased from the internal database. If
reference @var{R} does not exist in the database, @code{erase} just fails.

@item erased(+@var{R})
@findex erased/1
@saindex erased/1
@cyindex erased/1
Succeeds if the object whose database reference is @var{R} has been
erased.

@item instance(+@var{R},-@var{T})
@findex instance/2
@saindex instance/2
@cyindex instance/2
If @var{R} refers to a clause or a recorded term, @var{T} is unified
with its most general instance. If @var{R} refers to an unit clause
@var{C}, then @var{T} is unified with @code{@var{C} :- true}. When
@var{R} is not a reference to an existing clause or a recorded term,
this goal fails.

@item eraseall(+@var{R})
@findex eraseall/1
@snindex eraseall/1
@cnindex eraseall/1
All terms referred to by @var{R} are erased from the internal
database. If reference @var{R} does not exist in the database,
@code{eraseall} just fails.

@item current_key(@var{A},@var{P})
@findex current_key/2
@syindex current_key/2
@cnindex current_key/2
Defines the relation: @var{P} is a currently defined database key whose
name is the atom @var{A}.

@item get_value(+@var{A},-@var{V})
@findex get_value/2
@snindex get_value/2
@cnindex get_value/2
In YAP, atoms can be associated with constants. If one such
association exists for atom @var{A}, unify the second argument with the
constant. Otherwise, unify @var{V} with @code{[]}.

This predicate is YAP specific.

@item set_value(+@var{A},+@var{C})
@findex set_value/2 
@snindex set_value/2 
@cnindex set_value/2 
Associate atom @var{A} with constant @var{C}.

The @code{set_value} and @code{get_value} built-ins give a fast alternative to
the internal data-base. This is a simple form of implementing a global
counter.
@example
       read_and_increment_counter(Value) :-
                get_value(counter, Value),
                Value1 is Value+1,
                set_value(counter, Value1).
@end example
@noindent
This predicate is YAP specific.

@item recordzifnot(+@var{K},@var{T},-@var{R})
@findex recordzifnot/3
@snindex recordzifnot/3
@cnindex recordzifnot/3
If a variant of @var{T} is stored under key @var{K} fail. Otherwise, make
term @var{T} the last record under key @var{K} and unify @var{R} with its
reference.

This predicate is YAP specific.

@item recordaifnot(+@var{K},@var{T},-@var{R})
@findex recordaifnot/3
@snindex recordaifnot/3
@cnindex recordaifnot/3
If a variant of @var{T} is stored under key @var{K} fail. Otherwise, make
term @var{T} the first record under key @var{K} and unify @var{R} with its
reference.

This predicate is YAP specific.

@end table

There is a strong analogy between the i.d.b. and the way dynamic 
predicates are stored. In fact, the main i.d.b. predicates might be 
implemented using dynamic predicates:

@example
recorda(X,T,R) :- asserta(idb(X,T),R).
recordz(X,T,R) :- assertz(idb(X,T),R).
recorded(X,T,R) :- clause(idb(X,T),R).
@end example
@noindent
 We can take advantage of this, the other way around, as it is quite 
easy to write a simple Prolog interpreter, using the i.d.b.:

@example
asserta(G) :- recorda(interpreter,G,_).
assertz(G) :- recordz(interpreter,G,_).
retract(G) :- recorded(interpreter,G,R), !, erase(R).
call(V) :- var(V), !, fail.
call((H :- B)) :- !, recorded(interpreter,(H :- B),_), call(B).
call(G) :- recorded(interpreter,G,_).
@end example
@noindent
In YAP, much attention has been given to the implementation of the 
i.d.b., especially to the problem of accelerating the access to terms kept in 
a large list under the same key. Besides using the key, YAP uses an internal 
lookup function, transparent to the user, to find only the terms that might 
unify. For instance, in a data base containing the terms

@example
b
b(a)
c(d)
e(g)
b(X)
e(h)
@end example

@noindent
stored under the key k/1, when executing the query 

@example
:- recorded(k(_),c(_),R).
@end example

@noindent
@code{recorded} would proceed directly to the third term, spending almost the 
time as if @code{a(X)} or @code{b(X)} was being searched.
 The lookup function uses the functor of the term, and its first three
arguments (when they exist). So, @code{recorded(k(_),e(h),_)} would go
directly to the last term, while @code{recorded(k(_),e(_),_)} would find
first the fourth term, and then, after backtracking, the last one.

 This mechanism may be useful to implement a sort of hierarchy, where 
the functors of the terms (and eventually the first arguments) work as 
secondary keys.

 In the YAP's i.d.b. an optimized representation is used for 
terms without free variables. This results in a faster retrieval of terms 
and better space usage. Whenever possible, avoid variables in terms in terms stored in the  i.d.b.


@node BlackBoard, , Internal Database, Database
@section The Blackboard

YAP implements a blackboard in the style of the SICStus Prolog
blackboard. The blackboard uses the same underlying mechanism as the
internal data-base but has several important differences:
@itemize @bullet
@item It is module aware, in contrast to the internal data-base.
@item Keys can only be atoms or integers, and not compound terms.
@item A single term can be stored per key.
@item An atomic update operation is provided; this is useful for
parallelism.
@end itemize


@table @code
@item bb_put(+@var{Key},?@var{Term})
@findex bb_put/2
@syindex bb_put/2
@cnindex bb_put/2
Store term table @var{Term} in the blackboard under key @var{Key}. If a
previous term was stored under key @var{Key} it is simply forgotten.

@item bb_get(+@var{Key},?@var{Term})
@findex bb_get/2
@syindex bb_get/2
@cnindex bb_get/2
Unify @var{Term} with a term stored in the blackboard under key
@var{Key}, or fail silently if no such term exists.

@item bb_delete(+@var{Key},?@var{Term})
@findex bb_delete/2
@syindex bb_delete/2
@cnindex bb_delete/2
Delete any term stored in the blackboard under key @var{Key} and unify
it with @var{Term}. Fail silently if no such term exists.

@item bb_update(+@var{Key},?@var{Term},?@var{New})
@findex bb_update/3
@syindex bb_update/3
@cnindex bb_update/3
Atomically  unify a term stored in the blackboard under key @var{Key}
with @var{Term}, and if the unification succeeds replace it by
@var{New}. Fail silently if no such term exists or if unification fails.

@end table

@node Sets, Grammars, Database, Top
@section Collecting Solutions to a Goal

When there are several solutions to a goal, if the user wants to collect all
the solutions he may be led to use the data base, because backtracking will
forget previous solutions.

YAP allows the programmer to choose from several system
predicates instead of writing his own routines.  @code{findall/3} gives you
the fastest, but crudest solution. The other built-in predicates
postprocess the result of the query in several different ways:

@table @code

@item findall(@var{T},+@var{G},-@var{L}) [ISO]
@findex findall/3
@syindex findall/3
@cyindex findall/3
Unifies @var{L} with a list that contains all the instantiations of the
term @var{T} satisfying the goal @var{G}.

With the following program:
@example
a(2,1).
a(1,1).
a(2,2).
@end example
@noindent
the answer to the query
@example
findall(X,a(X,Y),L).
@end example
@noindent
would be:
@example
X = _32
Y = _33
L = [2,1,2];
no
@end example

@item findall(@var{T},+@var{G},+@var{L},-@var{L0})
@findex findall/4
@syindex findall/4
@cnindex findall/4
Similar to @code{findall/3}, but appends all answers to list @var{L0}.

@item all(@var{T},+@var{G},-@var{L})
@findex all/3
@snindex all/3
@cnindex all/3
Similar to @code{findall(@var{T},@var{G},@var{L})} but eliminating
repeated elements. Thus, assuming the same clauses as in the above
example, the reply to the query

@example
all(X,a(X,Y),L).
@end example
@noindent
would be:

@example
X = _32
Y = _33
L = [2,1];
no
@end example

@item bagof(@var{T},+@var{G},-@var{L}) [ISO]
@findex bagof/3
@saindex bagof/3
@cyindex bagof/3
For each set of possible instances of the free variables occurring in
@var{G} but not in @var{T}, generates the list @var{L} of the instances of
@var{T} satisfying @var{G}. Again, assuming the same clauses as in the
examples above, the reply to the query

@example
bagof(X,a(X,Y),L).

would be:
X = _32
Y = 1
L = [2,1];
X = _32
Y = 2
L = [2];
no
@end example

@item setof(@var{X},+@var{P},-@var{B}) [ISO]
@findex setof/3
@saindex setof/3
@cyindex setof/3
Similar to @code{bagof(@var{T},@var{G},@var{L})} but sorting list
@var{L} and keeping only one copy of each element.  Again, assuming the
same clauses as in the examples above, the reply to the query
@example
setof(X,a(X,Y),L).
@end example
@noindent
would be:
@example
X = _32
Y = 1
L = [1,2];
X = _32
Y = 2
L = [2];
no
@end example

@end table

@node Grammars, OS, Sets, Top
@section Grammar Rules

Grammar rules in Prolog are both a convenient way to express definite
clause grammars and  an extension of the well known context-free grammars.

A grammar rule is of the form:

@example
@i{ head --> body }
@end example
@noindent
where both @i{head} and @i{body} are sequences of one or more items
linked by the standard conjunction operator ','.

@emph{Items can be:}

@itemize @bullet
@item
a @emph{non-terminal} symbol may be either a complex term or an atom.
@item
a @emph{terminal} symbol may be any Prolog symbol. Terminals are
written as Prolog lists.
@item
an @emph{empty body} is written as the empty list '[ ]'.
@item
@emph{extra conditions} may be inserted as Prolog procedure calls, by being
written inside curly brackets '@{' and '@}'.
@item
the left side of a rule consists of a nonterminal and an optional list
of terminals.
@item
alternatives may be stated in the right-hand side of the rule by using
the disjunction operator ';'.
@item
the @emph{cut} and @emph{conditional} symbol ('->') may be inserted in the 
right hand side of a grammar rule
@end itemize

Grammar related built-in predicates:

@table @code

@item expand_term(@var{T},-@var{X})
@findex expand_term/2
@syindex expand_term/2
@cyindex expand_term/2
@findex term_expansion/2
@syindex term_expansion/2
@cyindex term_expansion/2
This predicate is used by YAP for preprocessing each top level
term read when consulting a file and before asserting or executing it.
It rewrites a term @var{T} to a term @var{X} according to the following
rules: first try to use the user defined predicate
@code{term_expansion/2}. If this call fails then the translating process
for DCG rules is applied, together with the arithmetic optimizer
whenever the compilation of arithmetic expressions is in progress.

@item user:goal_expansion(+@var{G},+@var{M},-@var{NG})
@findex user:goal_expansion/3
@snindex user:goal_expansion/3
@cnindex user:goal_expansion/3
Yap now supports @code{goal_expansion/3}. This is an user-defined
procedure that is called after term expansion when compiling or
asserting goals for each sub-goal in a clause. The first argument is
bound to the goal and the second to the module under which the goal
@var{G} will execute. If @code{goal_expansion/3} succeeds the new
sub-goal @var{NG} will replace @var{G} and will be processed in the same
way. If @code{goal_expansion/3} fails the system will use the default
rules.

@item phrase(+@var{P},@var{L},@var{R})
@findex phrase/3
@syindex phrase/3
@cnindex phrase/3
This predicate succeeds when the difference list @code{@var{L}-@var{R}}
is a phrase of type @var{P}.

@item phrase(+@var{P},@var{L})
@findex phrase/2
@syindex phrase/2
@cnindex phrase/2
This predicate succeeds when @var{L} is a phrase of type @var{P}. The
same as @code{phrase(P,L,[])}.

Both this predicate and the previous are used as a convenient way to
start execution of grammar rules.

@item 'C'(@var{S1},@var{T},@var{S2})
@findex C/3
@syindex C/3
@cnindex C/3
This predicate is used by the grammar rules compiler and is defined as
@code{'C'([H|T],H,T)}.

@end table

@node OS, Term Modification, Grammars, Top
@section Access to Operating System Functionality

The following built-in predicates allow access to underlying
Operating System functionality: 

@table @code

@item cd(+@var{D})
@findex cd/1
@snindex cd/1
@cnindex cd/1
Changes the current directory (on UNIX environments).

@item environ(?@var{E},-@var{S})
@findex environ/2
@syindex environ/2
@cnindex environ/2
This backtrackable predicate unifies the first argument with an
environment variable @var{E}, and the second with its value @var{S}. It
can used to detect all environment variables.

@item getcwd(-@var{D})
@findex getcwd/1
@snindex getcwd/1
@cnindex getcwd/1
Unify the current directory, represented as an atom, with the argument
@var{D}.

@item putenv(+@var{E},+@var{S})
@findex putenv/2
@snindex putenv/2
@cnindex putenv/2
Set environment variable @var{E} to the value @var{S}. If the
environment variable @var{E} does not exist, create a new one. Both the
environment variable and the value must be atoms.

@item rename(+@var{F},+@var{G})
@findex rename/2
@snindex rename/2
@cyindex rename/2
Renames file @var{F} to @var{G}.

@item sh
@findex sh/0
@snindex sh/0
@cyindex sh/0
Creates a new shell interaction.

@item system(+@var{S})
@findex system/1
@snindex system/1
@cyindex system/1
Passes command @var{S} to the Bourne shell (on UNIX environments) or the
current command interpreter in WIN32 environments.

@item unix(+@var{S})
@findex unix/1
@snindex unix/1
@cnindex unix/1
Access to Unix-like functionality:
@table @code
@item argv/1
Return a list of arguments to the program. These are the arguments that
follow a @code{--}, as in the usual Unix convention.
@item cd/0
Change to home directory.
@item cd/1
Change to given directory. Acceptable directory names are strings or
atoms.
@item environ/2
If the first argument is an atom, unify the second argument with the
value of the corresponding environment variable.
@item getcwd/1
Unify the first argument with an atom representing the current directory.
@item putenv/2
Set environment variable @var{E} to the value @var{S}. If the
environment variable @var{E} does not exist, create a new one. Both the
environment variable and the value must be atoms.
@item shell/1
Execute command under current shell. Acceptable commands are strings or
atoms.
@item system/1
Execute command with @code{/bin/sh}. Acceptable commands are strings or
atoms.
@item shell/0
Execute a new shell.
@end table

@item alarm(+@var{Seconds},+@var{Callable},+@var{OldAlarm})
@findex alarm/3
@snindex alarm/3
@cnindex alarm/3
Arranges for YAP to be interrupted in @var{Seconds}
seconds. When interrupted, YAP will execute @var{Callable} and
then return to the previous execution. If @var{Seconds} is @code{0}, no
new alarm is scheduled. In any event, any previously set alarm is
cancelled.

The variable @var{OldAlarm} unifies with the number of seconds remaining
until any previously scheduled alarm was due to be delivered, or with
@code{0} if there was no previously scheduled alarm.

Note that execution of @var{Callable} will wait if YAP is
executing built-in predicates, such as Input/Output operations.

The next example shows how @var{alarm/3} can be used to implement a
simple clock:

@example
loop :- loop.

ticker :- write('.'), flush_output,
          get_value(tick, yes),
          alarm(1,ticker,_).

:- set_value(tick, yes), alarm(1,ticker,_), loop.
@end example

The clock, @code{ticker}, writes a dot and then checks the flag
@code{tick} to see whether it can continue ticking. If so, it calls
itself again. Note that there is no guarantee that the each dot
corresponds a second: for instance, if the YAP is waiting for
user input, @code{ticker} will wait until the user types the entry in.

The next example shows how @code{alarm/3} can be used to guarantee that
a certain procedure does not take longer than a certain amount of time:

@example
loop :- loop.

:-   catch((alarm(10, throw(ball), _),loop),
        ball,
        format('Quota exhausted.~n',[])).
@end example
In this case after @code{10} seconds our @code{loop} is interrupted,
@code{ball} is thrown,  and the handler writes @code{Quota exhausted}.
Execution then continues from the handler.

Note that in this case @code{loop/0} always executes until the alarm is
sent. Often, the code you are executing succeeds or fails before the
alarm is actually delivered. In this case, you probably want to disable
the alarm when you leave the procedure. The next procedure does exactly so:
@example
once_with_alarm(Time,Goal,DoOnAlarm) :-
   catch(execute_once_with_alarm(Time, Goal), alarm, DoOnAlarm).

execute_once_with_alarm(Time, Goal) :-
        alarm(Time, alarm, _),
        ( call(Goal) -> alarm(0, alarm, _) ; alarm(0, alarm, _)).
@end example

The procedure has three arguments: the @var{Time} before the alarm is
sent; the @var{Goal} to execute; and the goal @var{DoOnAlarm} to execute
if the alarm is sent. It uses @code{catch/3} to handle the case the
@code{alarm} is sent. Then it starts the alarm, calls the goal
@var{Goal}, and disables the alarm on success or failure.

@item on_signal(+@var{Signal},?@var{OldAction},+@var{Callable})
@findex on_signal/3
@snindex on_signal/3
@cnindex on_signal/3
Set the interrupt handler for soft interrupt @var{Signal} to be
@var{Callable}. @var{OldAction} is unified with the previous handler.

Only a subset of the software interrupts (signals) can have their
handlers maniputated through @code{on_signal/3}.
Their POSIX names, YAP names and default behaviour is given below.
The "YAP name" of the signal is the atom that is associated with
each signal, and should be used as the first argument to
@code{on_signal/3}. It is chosen so that it matches the signal's POSIX
name.

@code{on_signal/3} succeeds, unless when called with an invalid
signal name or one that is not supported on this platform. No checks
are made on the handler provided by the user.

@table @code
@item SIGHUP (Hangup)
  sig_hup in YAP; Reconsult the initialisation files
  ~/.yaprc, ~/.prologrc and ~/prolog.ini.
@item SIGUSR1 and SIGUSR2 (User signals)
  sig_usr1 and sig_usr2 in YAP; Print a message and halt.
@end table

A special case is made, where if @var{Callable} is bound to
@code{default}, then the default handler is restored for that signal.

A call in the form @code{on_signal(@var{S},@var{H},@var{H})} can be used
to retrieve a signal's current handler without changing it.

It must be noted that although a signal can be received at all times,
the handler is not executed while Yap is waiting for a query at the
prompt. The signal will be, however, registered and dealt with as soon
as the user makes a query.

Please also note, that neither POSIX Operating Systems nor Yap guarantee
that the order of delivery and handling is going to correspond with the
order of dispatch.

@end table

@node Term Modification, Profiling, OS, Top
@section Term Modification

@cindex updating terms
It is sometimes useful to change the value of instantiated
variables. Although, this is against the spirit of logic programming, it
is sometimes useful. As in other Prolog systems, YAP has
several primitives that allow updating Prolog terms. Note that these
primitives are also backtrackable.

The @code{setarg/3} primitive allows updating any argument of a Prolog
compound terms. The @code{mutable} family of predicates provides
@emph{mutable variables}. They should be used instead of @code{setarg/3},
as they allow the encapsulation of accesses to updatable
variables. Their implementation can also be more efficient for long
deterministic computations.

@table @code
@item setarg(+@var{I},+@var{S},?@var{T})
@findex setarg/3n
@snindex setarg/3n
@cnindex setarg/3n
Set the value of the @var{I}th argument of term @var{S} to term @var{T}. 

@cindex mutable variables
@item create_mutable(+@var{D},-@var{M})
@findex create_mutable/2
@syindex create_mutable/2
@cnindex create_mutable/2
Create new mutable variable @var{M} with initial value @var{D}.

@item get_mutable(?@var{D},+@var{M})
@findex get_mutable/2
@syindex get_mutable/2
@cnindex get_mutable/2
Unify the current value of mutable term @var{M} with term @var{D}.

@item is_mutable(?@var{D})
@findex is_mutable/1
@syindex is_mutable/1
@cnindex is_mutable/1
Holds if @var{D} is a mutable term.

@item set_mutable(?@var{D},+@var{M})
@findex set_mutable/2
@syindex set_mutable/2
@cnindex set_mutable/2
Unify the current value of mutable term @var{M} with term @var{D}.

@item update_mutable(+@var{D},+@var{M})
@findex update_mutable/2
@syindex update_mutable/2
@cnindex update_mutable/2
Set the current value of mutable term @var{M} to term @var{D}.
@end table

@node Profiling, Arrays, Term Modification, Top
@section Profiling Prolog Programs

@cindex profiling
Predicates compiled with YAP's flag @code{profiling} set to
@code{on}, keep information on the number of times the predicate was
called. This information can be used to detect what are the most
commonly called predicates in the program.

The YAP profiling sub-system is currently
under-development. Functionality for this sub-system will increase with
newer implementation.

@strong{Notes:}

@itemize @bullet
@item Profiling works for both static and dynamic predicates.
@item Currently only information on entries and retries to a predicate
are maintained. This may change in the future.
@item As an example, the following user-level program gives a list of
the most often called procedures in a program:
@example
list_profile :-
        % get number of calls for each profiled procedure
        findall(D-P,profile_data(P,calls,D),LP),
        % sort them
        sort(LP,SLP),
        % output so that the most often called
        % predicates will come last:
        write_profile_data(SLP).

write_profile_data([]).
write_profile_data([D-P|SLP]) :-
        % swap the two calls if you want the most often
        %  called predicates first.
        format('~w: ~w~n', [P,D]),
        write_profile_data(SLP).
@end example
@end itemize

These are  the current predicates to access and clear profiling data:

@table @code
@item profile_data(?@var{Na/Ar}, ?@var{Paramater}, -@var{Data})
@findex profile_data/3
@snindex profile_data/3
@cnindex profile_data/3
Give current profile data on @var{Parameter} for a predicate described
by the predicate indicator @var{Na/Ar}. If any of @var{Na/Ar} or
@var{Parameter} are unbound, backtrack through all profiled predicates
or stored parameters. Current parameters are:

@table @code
@item calls
Number of times a procedure was called.

@item retries
Number of times a call to the procedure was backtracked to and retried.
@end table

@item profile_reset
@findex profiled_reset/0
@snindex profiled_reset/0
@cnindex profiled_reset/0
Reset all profiling information.

@end table

@node Arrays, Preds, Profiling , Top
@section Arrays

The YAP system includes experimental support for arrays. The
support is enabled with the option @code{YAP_ARRAYS}.

There are two very distinct forms of arrays in YAP. The
@emph{dynamic arrays} are a different way to access compound terms
created during the execution. Like any other terms, any bindings to
these terms and eventually the terms themselves will be destroyed during
backtracking. Our goal in supporting dynamic arrays is twofold. First,
they provide an alternative to the standard @code{arg/3}
built-in. Second, because dynamic arrays may have name that are globally
visible, a dynamic array can be visible from any point in the
program. In more detail, the clause
@example
g(X) :- array_element(a,2,X).
@end example
will succeed as long as the programmer has used the builtin @t{array/2}
to create an array term with at least 3 elements in the current
environment, and the array was associated with the name @code{a}.  The
element @code{X} is a Prolog term, so one can bind it and any such
bindings will be undone when backtracking. Note that dynamic arrays do
not have a type: each element may be any Prolog term.

The @emph{static arrays} are an extension of the database. They provide
a compact way for manipulating data-structures formed by characters,
integers, or floats imperatively. They can also be used to provide
two-way communication between YAP and external programs through
shared memory.

In order to efficiently manage space elements in a static array must
have a type. Currently, elements of static arrays in YAP should
have one of the following predefined types:

@itemize @bullet
@item  @code{byte}: an 8-bit signed character.
@item  @code{unsigned_byte}: an 8-bit unsigned character.
@item  @code{int}: Prolog integers. Size would be the natural size for
the machine's architecture.
@item  @code{float}: Prolog floating point number. Size would be equivalent
to a double in @code{C}.
@item  @code{atom}: a Prolog atom.
@item  @code{dbref}: an internal database reference.
@item  @code{term}: a generic Prolog term. Note that this will term will
not be stored in the array itself, but instead will be stored in the
Prolog internal database.
@end itemize

Arrays may be @emph{named} or @emph{anonymous}. Most arrays will be
@emph{named}, that is associated with an atom that will be used to find
the array. Anonymous arrays do not have a name, and they are only of
interest if the @code{TERM_EXTENSIONS} compilation flag is enabled. In
this case, the unification and parser are extended to replace
occurrences of Prolog terms of the form @code{X[I]} by run-time calls to
@code{array_element/3}, so that one can use array references instead of
extra calls to @code{arg/3}. As an example:
@example
g(X,Y,Z,I,J) :- X[I] is Y[J]+Z[I].
@end example
should give the same results as:
@example
G(X,Y,Z,I,J) :-
        array_element(X,I,E1),
        array_element(Y,J,E2),  
        array_element(Z,I,E3),  
        E1 is E2+E3.
@end example

Note that the only limitation on array size are the stack size for
dynamic arrays; and, the heap size for static (not memory mapped)
arrays. Memory mapped arrays are limited by available space in the file
system and in the virtual memory space.

The following predicates manipulate arrays:

@table @code

@item array(+@var{Name}, +@var{Size})
@findex array/2
@snindex array/2
@cnindex array/2
Creates a new dynamic array. The @var{Size} must evaluate to an
integer. The @var{Name} may be either an atom (named array) or an
unbound variable (anonymous array).

Dynamic arrays work as standard compound terms, hence space for the
array is recovered automatically on backtracking.

@item static_array(+@var{Name}, +@var{Size}, +@var{Type})
@findex static_array/3
@snindex static_array/3
@cnindex static_array/3
Create a new static array with name @var{Name}. Note that the @var{Name}
must be an atom (named array). The @var{Size} must evaluate to an
integer.  The @var{Type} must be bound to one of types mentioned
previously.

@item mmapped_array(+@var{Name}, +@var{Size}, +@var{Type}, +@var{File})
@findex static_array/3
@snindex static_array/3
@cnindex static_array/3
Similar to @code{static_array/3}, but the array is memory mapped to file
@var{File}. This means that the array is initialised from the file, and
that any changes to the array will also be stored in the file. 

This built-in is only available in operating systems that support the
system call @code{mmap}. Moreover, mmapped arrays do not store generic
terms (type @code{term}).

@item close_static_array(+@var{Name})
@findex close_static_array/3
@snindex close_static_array/3
@cnindex close_static_array/3
Close an existing static array of name @var{Name}. The @var{Name} must
be an atom (named array). Space for the array will be recovered and
further accesses to the array will return an error. 

@item resize_static_array(+@var{Name}, -@var{OldSize}, +@var{NewSize})
@findex resize_static_array/3
@snindex resize_static_array/3
@cnindex resize_static_array/3
Expand or reduce a static array, The @var{Size} must evaluate to an
integer. The @var{Name} must be an atom (named array). The @var{Type}
must be bound to one of @code{int}, @code{dbref}, @code{float} or
@code{atom}.

Note that if the array is a mmapped array the size of the mmapped file
will be actually adjusted to correspond to the size of the array.

@item array_element(+@var{Name}, +@var{Index}, ?@var{Element})
@findex access_array/3
@snindex access_array/3
@cnindex access_array/3
Unify @var{Element} with @var{Name}[@var{Index}]. It works for both
static and dynamic arrays, but it is read-only for static arrays, while
it can be used to unify with an element of a dynamic array.

@item update_array(+@var{Name}, +@var{Index}, ?@var{Value}) 
@findex update_array/3
@snindex update_array/3
@cnindex update_array/3
Attribute value @var{Value} to @var{Name}[@var{Index}]. Type
restrictions must be respected for static arrays. This operation is
available for dynamic arrays if @code{MULTI_ASSIGNMENT_VARIABLES} is
enabled (true by default). Backtracking undoes @var{update_array/3} for
dynamic arrays, but not for static arrays.

Note that @code{update_array/3} actually uses @code{setarg/3} to update
elements of dynamic arrays, and @code{setarg/3} spends an extra cell for
every update. For intensive operations we suggest it may be less
expensive to unify each element of the array with a mutable terms and
to use the operations on mutable terms.

@end table

@node Preds, Misc, Arrays, Top
@section Predicate Information

Built-ins that return information on the current predicates and modules:

@table @code
@c ......... begin of 'module' documentation .........
@item current_module(@var{M})
@findex current_module/1
@syindex current_module/1
@cnindex current_module/1
Succeeds if @var{M} are current modules. A module becomes current since
some predicate defined in it is loaded.

@item current_module(@var{M},@var{F})
@findex current_module/2
@syindex current_module/2
@cnindex current_module/2
Succeeds if @var{M} are current modules associated to the file @var{F}.

@c .......... end of 'module' documentation ..........
@end table

@node Misc, , Preds, Top
@section Miscellaneous

@table @code

@item statistics/0
@findex statistics/0
@saindex statistics/0
@cyindex statistics/0
Send to the current user error stream general information on space used and time
spent by the system.
@example
?- statistics.
Heap space : 2441216
        Heap in use: 820220, max. used: 1623516
Trail space : 131068
        Trail in use: 16, max. used: 13048
Stack space : 1523712
        Global in use: 88, max. used: 658700
        Local in use: 276, max. used: 515336


90 msec. for 5 heap overflows.
90 msec. for 3 stack overflows.
0 msec. for 0 trail overflows.
800 msec. for 3 garbage collections which 
             collected 208348 bytes.
Runtime : 23.07 sec.
@end example

@item statistics(?@var{Param},-@var{Info})
@findex statistics/2
@saindex statistics/2
@cnindex statistics/2
Gives statistical information on the system parameter given by first
argument:

@table @code

@item cputime
@findex cputime (statistics/2 option)
@code{[@var{Time since Boot},@var{Time From Last Call to Cputime}]}
@* 
This gives the total cputime in milliseconds spent executing Prolog code,
garbage collection and stack shifts time included.

@item garbage_collection
@findex garbage_collection (statistics/2 option)
@code{[@var{Number of GCs},@var{Total Global Recovered},@var{Total Time
Spent}]}
@*
Number of garbage collections, amount of space recovered in kbytes, and
total time spent doing garbage collection in milliseconds. More detailed
information is available using @code{yap_flag(gc_trace,verbose)}.

@item global_stack
@findex global_stack (statistics/2 option)
@code{[@var{Global Stack Used},@var{Execution Stack Free}]}
@*
Space in kbytes currently used in the global stack, and space available for
expansion by the local and global stacks.

@item local_stack
@findex local_stack (statistics/2 option)
@code{[@var{Local Stack Used},@var{Execution Stack Free}]}
@*
Space in kbytes currently used in the local stack, and space available for
expansion by the local and global stacks.

@item heap
@findex heap (statistics/2 option)
@code{[@var{Heap Used},@var{Heap Free}]}
@*
Total space in kbytes not recoverable
in backtracking. It includes the program code, internal data base, and,
atom symbol table.

@item program
@findex program (statistics/2 option)
@code{[@var{Program Space Used},@var{Program Space Free}]}
@*
Equivalent to @code{heap}.

@item runtime
@findex runtime (statistics/2 option)
@code{[@var{Time since Boot},@var{Time From Last Call to Runtime}]}
@* 
This gives the total cputime in milliseconds spent executing Prolog
code, not including garbage collections and stack shifts. Note that
until Yap4.1.2 the @code{runtime} statistics would return time spent on
garbage collection and stack shifting.

@item stack_shifts
@findex stack_shifts (stack_shifts/3 option)
@code{[@var{Number of Heap Shifts},@var{Number of Stack
Shifts},@var{Number of Trail Shifts}]}
@*
Number of times YAP had to
expand the heap, the stacks, or the trail. More detailed information is
available using @code{yap_flag(gc_trace,verbose)}.

@item trail
@findex trail (statistics/2 option)
@code{[@var{Trail Used},@var{Trail Free}]}
@*
Space in kbytes currently being used and still available for the trail.

@item walltime
@findex walltime (statistics/2 option)
@code{[@var{Time since Boot},@var{Time From Last Call to Runtime}]}
@* 
This gives the clock time in milliseconds since starting Prolog.

@end table

@item yap_flag(?@var{Param},?@var{Value})
@findex yap_flag/2
@snindex yap_flag/2
@cnindex yap_flag/2
Set or read system properties for @var{Param}:

@table @code

@item argv
@findex argv (yap_flag/2 option)
@* Read-only flag. It unifies with a list of atoms that gives the
arguments to Yap after @code{--}.

@item bounded [ISO]
@findex bounded (yap_flag/2 option)
@*
Read-only flag telling whether integers are bounded. The value depends
on whether YAP uses the GMP library or not.

@item char_conversion [ISO]
@findex  char_conversion (yap_flag/2 option)
@*
Writable flag telling whether a character conversion table is used when
reading terms. The default value for this flag is @code{off} except in
@code{sicstus} and @code{iso} language modes, where it is @code{on}.

@item character_escapes [ISO]
@findex  character_escapes (yap_flag/2 option)
@* Writable flag telling whether a character escapes are enables,
@code{on}, or disabled, @code{off}. The default value for this flag is
@code{on}.

@c You can also use @code{cprolog} mode, which corresponds to @code{off},
@c @code{iso} mode, which corresponds to @code{on}, and @code{sicstus}
@c mode, which corresponds to the mode traditionally used in SICStus
@c Prolog. In this mode back-quoted escape sequences should not close with
@c a backquote and unrecognised escape codes do not result in error.

@item debug [ISO]
@findex debug (yap_flag/2 option)
@*
If @var{Value} is unbound, tell whether debugging is @code{on} or
@code{off}. If @var{Value} is bound to @code{on} enable debugging, and if
it is bound to @code{off} disable debugging.

@item discontiguous_warnings
@findex discontiguous_warnings (yap_flag/2 option)
@*
If @var{Value} is unbound, tell whether warnings for discontiguous
predicates are @code{on} or
@code{off}. If @var{Value} is bound to @code{on} enable these warnings,
and if it is bound to @code{off} disable them. The default for YAP is
@code{off}, unless we are in @code{sicstus} or @code{iso} mode.

@item  dollar_as_lower_case
@findex dollar_as_lower_case (yap_flag/2 option)
@*
If @code{off} (default)  consider the character '$' a control character, if
@code{on} consider '$' a lower case character.

@item double_quotes [ISO]
@findex double_quotes (yap_flag/2 option)
@*
If @var{Value} is unbound, tell whether a double quoted list of characters
token is converted to a list of atoms, @code{chars}, to a list of integers,
@code{codes}, or to a single atom, @code{atom}. If @var{Value} is bound, set to
the corresponding behaviour. The default value is @code{codes}.

@item  fast
@findex fast (yap_flag/2 option)
@*
If @code{on} allow fast machine code, if @code{off} (default) disable it. Only
available in experimental implementations.

@item  gc
@findex gc (yap_flag/2 option)
@*
If @code{on} allow garbage collection (default), if @code{off} disable it.

@item  gc_margin
@findex gc_margin (yap_flag/2 option)
@*
Set or show the minimum free stack before starting garbage
collection. The default depends on total stack size. 

@item  gc_trace
@findex gc_trace (yap_flag/2 option)
@* If @code{off} (default) do not show information on garbage collection
and stack shifts, if @code{on} inform when a garbage collection or stack
shift happened, if @code{verbose} give detailed information on garbage
collection and stack shifts. Last, if @code{very_verbose} give detailed
information on data-structures found during the garbage collection
process, namely, on choice-points.

@item  host_type
@findex host_type (yap_flag/2 option)
@* Return @code{configure} system information, including the machine-id
for which Yap was compiled and Operating System information. 

@item  index
@findex index (yap_flag/2 option)
@*
If @code{on} allow indexing (default), if @code{off} disable it.

@item  informational_messages
@findex informational_messages (yap_flag/2 option)
@*
If @code{on} allow printing of informational messages, such as the ones
that are printed when consulting. If @code{off} disable printing
these messages. It is @code{on} by default except if Yap is booted with
the @code{-L} flag.

@item integer_rounding_function [ISO]
@findex integer_rounding_function (yap_flag/2 option)
@*
Read-only flag telling the rounding function used for integers. Takes the value
@code{down} for the current version of YAP.

@item language
@findex language (yap_flag/2 option)
@* 
Choose whether YAP is closer to C-Prolog, @code{cprolog}, iso-prolog,
@code{iso} or SICStus Prolog, @code{sicstus}. The current default is
@code{cprolog}. This flag affects update semantics, leashing mode,
style_checking, handling calls to undefined procedures, how directives
are interpreted, when to use dynamic, character escapes, and how files
are consulted.

@item max_arity [ISO]
@findex max_arity (yap_flag/2 option)
@*
Read-only flag telling the maximum arity of a functor. Takes the value
@code{unbounded} for the current version of YAP.

@item max_integer [ISO]
@findex max_integer (yap_flag/2 option)
@*
Read-only flag telling the maximum integer in the
implementation. Depends on machine and Operating System
architecture, and on whether YAP uses the @code{GMP} multiprecision
library. If @code{bounded} is false, requestes for @code{max_integer}
will fail.

@item min_integer [ISO]
@findex min_integer (yap_flag/2 option)
@* Read-only flag telling the minimum integer in the
implementation. Depends on machine and Operating System architecture,
and on whether YAP uses the @code{GMP} multiprecision library. If
@code{bounded} is false, requestes for @code{min_integer} will fail.

@item n_of_integer_keys_in_bb
@findex n_of_integer_keys_in_bb (yap_flag/2 option)
@*
Read or set the size of the hash table that is used for looking up the
blackboard when the key is an integer.

@item n_of_integer_keys_in_db
@findex n_of_integer_keys_in_db (yap_flag/2 option)
@*
Read or set the size of the hash table that is used for looking up the
internal data-base when the key is an integer.

@item  profiling
@findex profiling (yap_flag/2 option)
@*
If @code{off} (default) do not compile profiling information for
procedures. If @code{on} compile predicates so that they will output
profiling information. Profiling data can be read through the
@code{profile_data/3} built-in.

@item redefine_warnings
@findex discontiguous_warnings (yap_flag/2 option)
@*
If @var{Value} is unbound, tell whether warnings for procedures defined
in several different files are @code{on} or
@code{off}. If @var{Value} is bound to @code{on} enable these warnings,
and if it is bound to @code{off} disable them. The default for YAP is
@code{off}, unless we are in @code{sicstus} or @code{iso} mode.

@item single_var_warnings
@findex single_var_warnings (yap_flag/2 option)
@*
If @var{Value} is unbound, tell whether warnings for singleton variables
are @code{on} or @code{off}. If @var{Value} is bound to @code{on} enable
these warnings, and if it is bound to @code{off} disable them. The
default for YAP is @code{off}, unless we are in @code{sicstus} or
@code{iso} mode.

@item strict_iso
@findex strict_iso (prolog_flag/2 option)
@*
 If @var{Value} is unbound, tell whether strict ISO compatibility mode
is @code{on} or @code{off}. If @var{Value} is bound to @code{on} set
language mode to @code{iso} and enable strict mode. If @var{Value} is
bound to @code{off} disable strict mode, and keep the current language
mode. The default for YAP is @code{off}.

Under strict ISO prolog mode all calls to non-ISO built-ins generate an
error. Compilation of clauses that would call non-ISO built-ins will
also generate errors. Pre-processing for grammar rules is also
disabled. Module expansion is still performed.

Arguably, ISO Prolog does not provide all the functionality required
from a modern Prolog system. Moreover, because most Prolog
implementations do not fully implement the standard and because the
standard itself gives the implementor latitude in a few important
questions, such as the unification algorithm and maximum size for
numbers there is not guarantee that prolograms compliant with this mode
will work the same way in every Prolog and in every platform. We thus
believe this mode is mostly useful when investigating how a program
depends on a Prolog's platform specific features.

@item  syntax_errors
@findex syntax_errors (yap_flag/2 option)
@*
Control action to be taken after syntax errors while executing @code{read/1},
@code{read/2}, or @code{read_term/3}:
@table @code

@item dec10
@*
Report the syntax error and retry reading the term.
 
@item fail
@*
Report the syntax error and fail (default).

@item error
@*
Report the syntax error and generate an error.

@item quiet
@*
Just fail
@end table

@item to_chars_mode
@findex to_chars_modes (yap_flag/2 option)
@* Define whether YAP should follow @code{quintus}-like
semantics for the @code{atom_chars/1} or @code{number_chars/1} built-in,
or whether it should follow the ISO standard (@code{iso} option).

@item toplevel_hook(@var{G})
@findex toplevel_hook (yap_flag/2 option)
@*
If bound, set @var{G} to a goal to be executed before entering the
top-level. If unbound show the current goal or @code{true} if none is
presented. Only the first solution is considered and the goal is not
backtracked into.

@item typein_module
@findex typein_module (yap_flag/2 option)
@*
If bound, set the current working or type-in module to the argument,
which must be an atom. If unbound, unify the argument with the current
working module.

@item unknown [ISO]
@findex unknown (yap_flag/2 option)
@*
Corresponds to calling the @code{unknown/2} built-in.

@item update_semantics
@findex update_semantics (yap_flag/2 option)
@*
Define whether YAP should follow @code{immediate} update
semantics, as in C-Prolog (default), @code{logical} update semantics,
as in Quintus Prolog, SICStus Prolog, or in the ISO standard. There is
also an intermediate mode, @code{logical_assert}, where dynamic
procedures follow logical semantics but the internal data base still
follows immediate semantics.

@item user_error
@findex user_error (yap_flag/2 option)
@*
If the second argument is bound to a stream, set @code{user_error} to
this stream. If the second argument is unbound, unify the argument with
the current @code{user_error} stream.

By default, the @code{user_error} stream is set to a stream
corresponding to the Unix @code{stderr} stream.

The next example shows how to use this flag:
@example
   ?- open( '/dev/null', append, Error,
           [alias(mauri_tripa)] ).

Error = '$stream'(3) ? ;

no
   ?- set_prolog_flag(user_error, mauri_tripa).

close(mauri_tripa).

yes
   ?- 
@end example
We execute three commands. First, we open a stream in write mode and
give it an alias, in this case @code{mauri_tripa}. Next, we set
@code{user_error} to the stream via the alias. Note that after we did so
prompts from the system were redirected to the stream
@code{mauri_tripa}. Last, we close the stream. At this point, YAP
automatically redirects the @code{user_error} alias to the original
@code{stderr}.

@item user_input
@findex user_input (yap_flag/2 option)
@*
If the second argument is bound to a stream, set @code{user_input} to
this stream. If the second argument is unbound, unify the argument with
the current @code{user_input} stream.

By default, the @code{user_input} stream is set to a stream
corresponding to the Unix @code{stdin} stream.

@item user_output
@findex user_output (yap_flag/2 option)
@*
If the second argument is bound to a stream, set @code{user_output} to
this stream. If the second argument is unbound, unify the argument with
the current @code{user_output} stream.

By default, the @code{user_output} stream is set to a stream
corresponding to the Unix @code{stdout} stream.

@item version
@findex version (yap_flag/2 option)
@*
Read-only flag that giving the current version of Yap.

@item write_strings
@findex  write_strings (yap_flag/2 option)
@* Writable flag telling whether the system should write lists of
integers that are writable character codes using the list notation. It
is @code{on} if enables or @code{off} if disabled. The default value for
this flag is @code{off}.

@end table

@item current_prolog_flag(?@var{Flag},-@var{Value}) [ISO]
@findex current_prolog_flag/2
@snindex current_prolog_flag/2
@cnindex current_prolog_flag/2

Obtain the value for a YAP Prolog flag. Equivalent to calling
@code{yap_flag/2} with the second argument unbound, and unifying the
returned second argument with @var{Value}.

@item prolog_flag(?@var{Flag},-@var{OldValue},+@var{NewValue})
@findex prolog_flag/3
@syindex prolog_flag/3
@cnindex prolog_flag/3

Obtain the value for a YAP Prolog flag and then set it to a new
value. Equivalent to first calling @code{current_prolog_flag/2} with the
second argument @var{OldValue} unbound and then calling
@code{set_prolog_flag/2} with the third argument @var{NewValue}.

@item set_prolog_flag(+@var{Flag},+@var{Value}) [ISO]
@findex set_prolog_flag/2
@snindex set_prolog_flag/2
@cnindex set_prolog_flag/2

Set the value for YAP Prolog flag @code{Flag}. Equivalent to
calling @code{yap_flag/2} with both arguments bound.

@item op(+@var{P},+@var{T},+@var{A}) [ISO]
@findex op/3
@syindex op/3
@cyindex op/3
Defines the operator @var{A} or the list of operators @var{A} with type
@var{T} (which must be one of @code{xfx}, @code{xfy},@code{yfx},
@code{xf}, @code{yf}, @code{fx} or @code{fy}) and precedence @var{P}
(see appendix iv for a list of predefined operators).

Note that if there is a preexisting operator with the same name and
type, this operator will be discarded. Also, @code{','} may not be defined
as an operator, and it is not allowed to have the same for an infix and
a posfix operator.

@item current_op(@var{P},@var{T},@var{F}) [ISO]
@findex current_op/3
@syindex current_op/3
@cnindex current_op/3
Defines the relation: @var{P} is a currently defined  operator of type
@var{T} and precedence @var{P}.

@item prompt(-@var{A},+@var{B})
@findex prompt/2
@syindex prompt/2
@cyindex prompt/2
Changes YAP input prompt from @var{A} to @var{B}.

@item initialization
@findex initialization/0
@syindex initialization/0
@cnindex initialization/0
Execute the goals defined by initialization/1. Only the first answer is
considered.

@item prolog_initialization(@var{G})
@findex prolog_initialization/1
@saindex prolog_initialization/1
@cnindex prolog_initialization/1
Add a goal to be executed on system initialization. This is compatible
with SICStus Prolog's @code{initialization/1}.

@item version
@findex version/0
@saindex version/0
@cnindex version/0
Write YAP's boot message. 

@item version(-@var{Message})
@findex version/1
@syindex version/1
@cnindex version/1
Add a message to be written when yap boots or after aborting. It is not
possible to remove messages.

@item prolog_load_context(?@var{Key}, ?@var{Value})
@findex prolog_load_context/2
@syindex prolog_load_context/2
@cnindex prolog_load_context/2
Obtain information on what is going on in the compilation process. The
following keys are available:

@table @code

@item directory
@findex directory (prolog_load_context/2 option)
@* 
Full name for the directory where YAP is currently consulting the
file.

@item file
@findex file (prolog_load_context/2 option)
@*
Full name for the file currently being consulted. Notice that included
filed are ignored.

@item module
@findex module (prolog_load_context/2 option)
@*
Current source module.

@item source
@findex file (prolog_load_context/2 option)
@*
Full name for the file currently being read in, which may be consulted,
reconsulted, or included.

@item stream
@findex file (prolog_load_context/2 option)
@*
Stream currently being read in.

@item term_position
@findex file (prolog_load_context/2 option)
@*
Stream position at the stream currently being read in.
@end table


@end table

@node Library, Extensions, Builtins, Top

@chapter Library Predicates

Library files reside in the library_directory path (set by the
@code{LIBDIR} variable in the Makefile for YAP). Currently,
most files in the library are from the Edinburgh Prolog library. 

@menu
 
Library, Extensions, Builtins, Top
* Apply Macros:: Apply a Predicate to a list or to sub-terms.
* Association Lists:: Binary Tree Implementation of Association Lists.
* AVL Trees:: Predicates to add and lookup balanced binary  trees.
* Heaps:: Labelled binary tree where the key of each node is less
    than or equal to the keys of its children.
* Lists:: List Manipulation
* Ordered Sets:: Ordered Set Manipulation
* Pseudo Random:: Pseudo Random Numbers
* Queues:: Queue Manipulation
* Random:: Random Numbers
* RegExp:: Regular Expression Manipulation
* Splay Trees:: Splay Trees
* String I/O:: Writing To and Reading From Strings
* System:: System Utilities
* Terms:: Utilities on Terms
* Timeout:: Call With Timeout
* Trees:: Updatable Binary Trees
* UGraphs:: Unweighted Graphs


@end menu
 
@node Apply Macros, Association Lists, , Library
@section Apply Macros
@cindex macros

This library provides a set of utilities to apply a predicate to all
elements of a list or to all sub-terms of a term. They allow one to
easily perform the most common do-loop constructs in Prolog. To avoid
performance degradation due to @code{apply/2}, each call creates an
equivalent Prolog program, without meta-calls, which is executed by the
Prolog engine instead. Note that if the equivalent Prolog program
already exists, it will be simply used.

The following routines are available once included with the
@code{use_module(library(apply_macros))} command.

@table @code
@item maplist(+@var{Pred}, ?@var{ListIn}, ?@var{ListOut})
@findex maplist/3
@snindex maplist/3
@cnindex maplist/3
      Creates @var{ListOut} by applying the predicate @var{Pred} to all
elements of @var{ListIn}.

@item checklist(+@var{Pred}, +@var{List})
@findex checklist/2
@snindex checklist/2
@cnindex checklist/2
      Succeeds if the predicate @var{Pred} succeeds on all elements of @var{List}.

@item selectlist(+@var{Pred}, +@var{ListIn}, ?@var{ListOut})
@findex selectlist/3
@snindex selectlist/3
@cnindex selectlist/3
      Creates @var{ListOut} of all list elements of @var{ListIn} that pass a given test

@item convlist(+@var{Pred}, +@var{ListIn}, ?@var{ListOut})
@findex convlist/3
@snindex convlist/3
@cnindex convlist/3
      A combination of maplist and selectlist: creates @var{ListOut} by
applying the predicate @var{Pred} to all list elements on which
@var{Pred} succeeds

@item sumlist(+@var{Pred}, +@var{List}, ?@var{AccIn}, ?@var{AccOut})
@findex sumlist/4
@snindex sumlist/4
@cnindex sumlist/4
      Calls @var{Pred} on all elements of List and collects a result in
@var{Accumulator}.

@item mapargs(+@var{Pred}, ?@var{TermIn}, ?@var{TermOut})
@findex mapargs/3
@snindex mapargs/3
@cnindex mapargs/3
      Creates @var{TermOut} by applying the predicate @var{Pred} to all
      arguments of @var{TermIn}

@item sumargs(+@var{Pred}, +@var{Term}, ?@var{AccIn}, ?@var{AccOut})
@findex sumargs/4
@snindex sumargs/4
@cnindex sumargs/4
  Calls the predicate @var{Pred} on all arguments of @var{Term} and
collects a  result in @var{Accumulator}

@item mapnodes(+@var{Pred}, +@var{TermIn}, ?@var{TermOut}) 
@findex mapnodes/3
@snindex mapnodes/3
@cnindex mapnodes/3
      Creates @var{TermOut} by applying the predicate @var{Pred}
      to all sub-terms of @var{TermIn} (depth-first and left-to-right order)

@item checknodes(+@var{Pred}, +@var{Term}) 
@findex checknodes/3
@snindex checknodes/3
@cnindex checknodes/3
      Succeeds if the predicate @var{Pred} succeeds on all sub-terms of
      @var{Term} (depth-first and left-to-right order)

@item sumnodes(+@var{Pred}, +@var{Term}, ?@var{AccIn}, ?@var{AccOut})
@findex sumnodes/4
@snindex sumnodes/4
@cnindex sumnodes/4
      Calls the predicate @var{Pred} on all sub-terms of @var{Term} and
collect a result in @var{Accumulator} (depth-first and left-to-right order)
@end table

Examples:

@example
%given
plus(X,Y,Z) :- Z is X + Y.
plus_if_pos(X,Y,Z) :- Y > 0, Z is X + Y.
vars(X, Y, [X|Y]) :- var(X), !.
vars(_, Y, Y).
trans(TermIn, TermOut) :-
  (compound(TermIn) ; atom(TermIn)),
  TermIn =.. [p|Args],
  TermOut =..[q|Args],
  !.
trans(X,X).

%success

maplist(plus(1), [1,2,3,4], [2,3,4,5]).
checklist(var, [X,Y,Z]).
selectlist(<(0), [-1,0,1], [1]).
convlist(plus_if_pos(1), [-1,0,1], [2]).
sumlist(plus, [1,2,3,4], 1, 11).
mapargs(number_atom,s(1,2,3), s('1','2','3')).
sumargs(vars, s(1,X,2,Y), [], [Y,X]).
mapnodes(trans, p(a,p(b,a),c), q(a,q(b,a),c)).
checknodes(\==(T), p(X,p(Y,X),Z)).
sumnodes(vars, [c(X), p(X,Y), q(Y)], [], [Y,Y,X,X]).
% another one
maplist(mapargs(number_atom),[c(1),s(1,2,3)],[c('1'),s('1','2','3')]).
@end example

@node Association Lists, AVL Trees, Apply Macros, Library
@section Association Lists
@cindex association list

The following association list manipulation predicates are available
once included with the @code{use_module(library(assoc))} command.

@table @code
@item assoc_to_list(+@var{Assoc},?@var{List})
@findex assoc_to_list/2
@syindex assoc_to_list/2
@cnindex assoc_to_list/2
Given an association list @var{Assoc} unify @var{List} with a list of
the form @var{Key-Val}, where the elements @var{Key} are in ascending
order.

@item empty_assoc(+@var{Assoc})
@findex empty_assoc/1
@syindex empty_assoc/1
@cnindex empty_assoc/1
Succeeds if association list @var{Assoc} is empty.

@item gen_assoc(+@var{Assoc},?@var{Key},?@var{Value})
@findex gen_assoc/3
@syindex gen_assoc/3
@cnindex gen_assoc/3
Given the association list @var{Assoc}, unify @var{Key} and @var{Value}
with two associated elements. It can be used to enumerate all elements
in the association list.

@item get_assoc(+@var{Key},+@var{Assoc},?@var{Value})
@findex get_assoc/3
@syindex get_assoc/3
@cnindex get_assoc/3
If @var{Key} is one of the elements in the association list @var{Assoc},
return the associated value.

@item get_assoc(+@var{Key},+@var{Assoc},?@var{Value},+@var{NAssoc},?@var{NValue})
@findex get_assoc/5
@syindex get_assoc/5
@cnindex get_assoc/5
If @var{Key} is one of the elements in the association list @var{Assoc},
return the associated value @var{Value} and a new association list
@var{NAssoc} where @var{Key} is associated with @var{NValue}.

@item list_to_assoc(+@var{List},?@var{Assoc})
@findex list_to_assoc/2
@syindex list_to_assoc/2
@cnindex list_to_assoc/2
Given a list @var{List} such that each element of @var{List} is of the
form @var{Key-Val}, and all the @var{Keys} are unique, @var{Assoc} is
the corresponding association list.

@item map_assoc(+@var{Pred},+@var{Assoc},?@var{New})
@findex map_assoc/3
@syindex map_assoc/3
@cnindex map_assoc/3
Given the binary predicate name @var{Pred} and the association list
@var{Assoc}, @var{New} in an association list with keys in @var{Assoc},
and such that if @var{Key-Val} is in @var{Assoc}, and @var{Key-Ans} is in
@var{New}, then @var{Pred}(@var{Val},@var{Ans}) holds.

@item ord_list_to_assoc(+@var{List},?@var{Assoc})
@findex ord_list_to_assoc/2
@syindex ord_list_to_assoc/2
@cnindex ord_list_to_assoc/2
Given an ordered list @var{List} such that each element of @var{List} is
of the form @var{Key-Val}, and all the @var{Keys} are unique, @var{Assoc} is
the corresponding association list.

@item put_assoc(+@var{Key},+@var{Assoc},+@var{Val},+@var{New})
@findex put_assoc/4
@syindex put_assoc/4
@cnindex put_assoc/4
The association list @var{New} includes and element of association
@var{key} with @var{Val}, and all elements of @var{Assoc} that did not
have key @var{Key}.

@end table

@node AVL Trees, Heaps, Association Lists, Library
@section AVL Trees
@cindex AVL trees

AVL trees are balanced search binary trees. They are named after their
inventors, Adelson-Velskii and Landis, and they were the first
dynamically balanced trees to be proposed. The YAP AVL tree manipulation
predicates library uses code originally written by Martin van Emdem and
published in the Logic Programming Newsletter, Autumn 1981.  A bug in
this code was fixed by Philip Vasey, in the Logic Programming
Newsletter, Summer 1982. The library currently only includes routines to
insert and lookup elements in the tree.

@table @code
@item avl_insert(+@var{Key},?@var{Value},+@var{T0},+@var{TF})
@findex avl_insert/4
@snindex avl_insert/4
@cnindex avl_insert/4
Add an element with key @var{Key} and @var{Value} to the AVL tree
@var{T0} creating a new AVL tree @var{TF}. Duplicated elements are
allowed.

@item avl_lookup(+@var{Key},-@var{Value},+@var{T})
@findex avl_lookup/3
@snindex avl_lookup/3
@cnindex avl_lookup/3
Lookup an element with key @var{Key} in the AVL tree
@var{T}, returning the value @var{Value}.

@end table

@node Heaps, Lists, AVL Trees, Library
@section Heaps
@cindex heap

A heap is a labelled binary tree where the key of each node is less than
or equal to the keys of its sons.  The point of a heap is that we can
keep on adding new elements to the heap and we can keep on taking out
the minimum element.  If there are N elements total, the total time is
O(NlgN).  If you know all the elements in advance, you are better off
doing a merge-sort, but this file is for when you want to do say a
best-first search, and have no idea when you start how many elements
there will be, let alone what they are.

The following heap manipulation routines are available once included
with the @code{use_module(library(heaps))} command. 

@table @code

@item add_to_heap(+@var{Heap},+@var{key},+@var{Datum},-@var{NewHeap})
@findex add_to_heap/4
@syindex        add_to_heap/4
@cnindex        add_to_heap/4
Inserts the new @var{Key-Datum} pair into the heap. The insertion is not
stable, that is, if you insert several pairs with the same @var{Key} it
is not defined which of them will come out first, and it is possible for
any of them to come out first depending on the  history of the heap.

@item empty_heap(?@var{Heap})
@findex empty_heap/1
@syindex        empty_heap/1
@cnindex        empty_heap/1
Succeeds if @var{Heap} is an empty heap.

@item get_from_heap(+@var{Heap},-@var{key},-@var{Datum},-@var{Heap})
@findex get_from_heap/4
@syindex        get_from_heap/4
@cnindex        get_from_heap/4
Returns the @var{Key-Datum} pair in @var{OldHeap} with the smallest
@var{Key}, and also a @var{Heap} which is the @var{OldHeap} with that
pair deleted.

@item heap_size(+@var{Heap}, -@var{Size})
@findex heap_size/2
@syindex        heap_size/2
@cnindex        heap_size/2
Reports the number of elements currently in the heap.

@item heap_to_list(+@var{Heap}, -@var{List})
@findex heap_to_list/2
@syindex        heap_to_list/2
@cnindex        heap_to_list/2
Returns the current set of @var{Key-Datum} pairs in the @var{Heap} as a
@var{List}, sorted into ascending order of @var{Keys}.

@item list_to_heap(+@var{List}, -@var{Heap})
@findex list_to_heap/2
@syindex        list_to_heap/2
@cnindex        list_to_heap/2
Takes a list of @var{Key-Datum} pairs (such as keysort could be used to sort)
and forms them into a heap.

@item min_of_heap(+@var{Heap},  -@var{Key},  -@var{Datum})
@findex min_of_heap/3
@syindex min_of_heap/3
@cnindex min_of_heap/3
Returns the Key-Datum pair at the top of the heap (which is of course
the pair with the smallest Key), but does not remove it from the heap.

@item min_of_heap(+@var{Heap},  -@var{Key1},  -@var{Datum1},
-@var{Key2},  -@var{Datum2})
@findex min_of_heap/5
@syindex min_of_heap/5
@cnindex min_of_heap/5
Returns the smallest (Key1) and second smallest (Key2) pairs in the
heap, without deleting them.
@end table

@node Lists, Ordered Sets, Heaps, Library
@section List Manipulation
@cindex list manipulation

The following list manipulation routines are available once included
with the @code{use_module(library(lists))} command. 

@table @code

@item append(?@var{Prefix},?@var{Suffix},?@var{Combined})
@findex append/3
@syindex append/3
@cnindex append/3
True when all three arguments are lists, and the members of
@var{Combined} are the members of @var{Prefix} followed by the members of @var{Suffix}.
It may be used to form @var{Combined} from a given @var{Prefix}, @var{Suffix} or to take
a given @var{Combined} apart.

@item delete(+@var{List}, ?@var{Element}, ?@var{Residue})
@findex delete/3
@syindex delete/3
@cnindex delete/3
True when @var{List} is a list, in which @var{Element} may or may not
occur, and @var{Residue} is a copy of @var{List} with all elements
identical to @var{Element} deleted.

@item is_list(+@var{List})
@findex is_list/1
@syindex is_list/1
@cnindex is_list/1
True when @var{List} is a proper list. That is, @var{List}
is bound to the empty list (nil) or a term with functor '.' and arity 2.

@item last(+@var{List},?@var{Last})
@findex last/2
@syindex last/2
@cnindex last/2
True when @var{List} is a list and @var{Last} is identical to its last element.

@item list_concat(+@var{Lists},?@var{List})
@findex list_concat/2
@snindex list_concat/2
@cnindex list_concat/2
True when @var{Lists} is a list of lists and @var{List} is the
concatenation of @var{Lists}.

@item member(?@var{Element}, ?@var{Set})
@findex member/2
@syindex member/2
@cnindex member/2
True when @var{Set} is a list, and @var{Element} occurs in it.  It may be used
to test for an element or to enumerate all the elements by backtracking.

@item memberchk(+@var{Element}, +@var{Set})
@findex memberchk/2
@syindex memberchk/2
@cnindex memberchk/2
As @code{member/2}, but may only be used to test whether a known
@var{Element} occurs in a known Set.  In return for this limited use, it
is more efficient when it is applicable.

@item nth0(?@var{N}, ?@var{List}, ?@var{Elem})
@findex nth0/2
@syindex nth0/2
@cnindex nth0/2
True when @var{Elem} is the Nth member of @var{List},
counting the first as element 0.  (That is, throw away the first
N elements and unify @var{Elem} with the next.)  It can only be used to
select a particular element given the list and index.  For that
task it is more efficient than @code{member/2}

@item nth(?@var{N}, ?@var{List}, ?@var{Elem})
@findex nth/2
@syindex nth/2
@cnindex nth/2
The same as @code{nth0/3}, except that it counts from
1, that is @code{nth(1, [H|_], H)}.

@item nth0(?@var{N}, ?@var{List}, ?@var{Elem}, ?@var{Rest})
@findex nth0/4
@syindex nth0/4
@cnindex nth0/4
Unifies @var{Elem} with the Nth element of @var{List},
counting from 0, and @var{Rest} with the other elements.  It can be used
to select the Nth element of @var{List} (yielding @var{Elem} and @var{Rest}), or to
insert @var{Elem} before the Nth (counting from 1) element of @var{Rest}, when
it yields @var{List}, e.g. @code{nth0(2, List, c, [a,b,d,e])} unifies List with
@code{[a,b,c,d,e]}.  @code{nth/4} is the same except that it counts from 1.  @code{nth0/4}
can be used to insert @var{Elem} after the Nth element of @var{Rest}.

@item nth(?@var{N}, ?@var{List}, ?@var{Elem}, ?@var{Rest})
@findex nth/4
@syindex nth/4
@cnindex nth/4
Unifies @var{Elem} with the Nth element of @var{List}, counting from 1,
and @var{Rest} with the other elements.  It can be used to select the
Nth element of @var{List} (yielding @var{Elem} and @var{Rest}), or to
insert @var{Elem} before the Nth (counting from 1) element of
@var{Rest}, when it yields @var{List}, e.g. @code{nth(1, List, c,
[a,b,d,e])} unifies List with @code{[a,b,c,d,e]}.  @code{nth/4}
can be used to insert @var{Elem} after the Nth element of @var{Rest}.

@item permutation(+@var{List},?@var{Perm})
@findex permutation/2
@syindex permutation/2
@cnindex permutation/2
True when @var{List} and @var{Perm} are permutations of each other.

@item remove_dups(+@var{List}, ?@var{Pruned})
@findex remove_dups/2
@syindex remove_dups/2
@cnindex remove_dups/2
Removes duplicated elements from @var{List}.  Beware: if the @var{List} has
non-ground elements, the result may surprise you.

@item reverse(+@var{List}, ?@var{Reversed})
@findex reverse/2
@syindex reverse/2
@cnindex reverse/2
True when @var{List} and @var{Reversed} are lists with the same elements
but in opposite orders. 
 
@item same_length(?@var{List1}, ?@var{List2})
@findex same_length/2
@syindex same_length/2
@cnindex same_length/2
True when @var{List1} and @var{List2} are both lists and have the same number
of elements.  No relation between the values of their elements is
implied.
Modes @code{same_length(-,+)} and @code{same_length(+,-)} generate either list given
the other; mode @code{same_length(-,-)} generates two lists of the same length,
in which case the arguments will be bound to lists of length 0, 1, 2, ...

@item select(?@var{Element}, ?@var{Set}, ?@var{Residue})
@findex select/3
@syindex select/3
@cnindex select/3
True when @var{Set} is a list, @var{Element} occurs in @var{Set}, and @var{Residue} is
everything in @var{Set} except @var{Element} (things stay in the same order).
 
@item sublist(?@var{Sublist}, ?@var{List})
@findex sublist/2
@syindex sublist/2
@cnindex sublist/2
True when both @code{append(_,Sublist,S)} and @code{append(S,_,List)} hold.
 
@item suffix(?@var{Suffix}, ?@var{List})
@findex suffix/2
@syindex suffix/2
@cnindex suffix/2
Holds when @code{append(_,Suffix,List)} holds.

@item sum_list(?@var{Numbers}, ?@var{Total})
@findex sum_list/2
@snindex sum_list/2
@cnindex sum_list/2
True when @var{Numbers} is a list of integers, and @var{Total} is their sum.

@item sumlist(?@var{Numbers}, ?@var{Total})
@findex sumlist/2
@syindex sumlist/2
@cnindex sumlist/2
True when @var{Numbers} is a list of integers, and @var{Total} is their
sum. The same as @code{sum_list/2}, please do use @code{sum_list/2}
instead.

@end table

@node Ordered Sets, Pseudo Random, Lists, Library
@section Ordered Sets
@cindex ordered set

The following ordered set manipulation routines are available once
included with the @code{use_module(library(ordsets))} command.  An
ordered set is represented by a list having unique and ordered
elements. Output arguments are guaranteed to be ordered sets, if the
relevant inputs are. This is a slightly patched version of Richard
O'Keefe's original library.

@table @code
@item list_to_ord_set(+@var{List}, ?@var{Set})
@findex list_to_ord_set/2
@syindex list_to_ord_set/2
@cnindex list_to_ord_set/2
Holds when @var{Set} is the ordered representation of the set
represented by the unordered representation @var{List}.

@item merge(+@var{List1}, +@var{List2}, -@var{Merged})
@findex merge/3
@syindex merge/3
@cnindex merge/3
Holds when @var{Merged} is the stable merge of the two given lists.

Notice that @code{merge/3} will not remove duplicates, so merging
ordered sets will not necessarily result in an ordered set. Use
@code{ord_union/3} instead.

@item ord_add_element(+@var{Set1}, +@var{Element}, ?@var{Set2})
@findex ord_add_element/3
@syindex ord_add_element/3
@cnindex ord_add_element/3
Insering @var{Element} in @var{Set1} returns @var{Set2}.  It should give
exactly the same result as @code{merge(Set1, [Element], Set2)}, but a
bit faster, and certainly more clearly. The same as @code{ord_insert/3}.

@item ord_del_element(+@var{Set1}, +@var{Element}, ?@var{Set2})
@findex ord_del_element/3
@syindex ord_del_element/3
@cnindex ord_del_element/3
Removing @var{Element} from @var{Set1} returns @var{Set2}.

@item ord_disjoint(+@var{Set1}, +@var{Set2})
@findex ord_disjoint/2
@syindex ord_disjoint/2
@cnindex ord_disjoint/2
Holds when the two ordered sets have no element in common.

@item ord_member(+@var{Element}, +@var{Set})
@findex ord_member/2
@syindex ord_member/2
@cnindex ord_member/2
Holds when @var{Element} is a member of @var{Set}.

@item ord_insert(+@var{Set1}, +@var{Element}, ?@var{Set2})
@findex ord_insert/3
@syindex ord_insert/3
@cnindex ord_insert/3
Insering @var{Element} in @var{Set1} returns @var{Set2}.  It should give
exactly the same result as @code{merge(Set1, [Element], Set2)}, but a
bit faster, and certainly more clearly. The same as @code{ord_add_element/3}.

@item ord_intersect(+@var{Set1}, +@var{Set2})
@findex ord_intersect/2
@syindex ord_intersect/2
@cnindex ord_intersect/2
Holds when the two ordered sets have at least one element in common.

@item ord_intersection(+@var{Set1}, +@var{Set2}, ?@var{Intersection})
@findex ord_intersect/3
@syindex ord_intersect/3
@cnindex ord_intersect/3
Holds when Intersection is the ordered representation of @var{Set1}
and @var{Set2}.

@item ord_seteq(+@var{Set1}, +@var{Set2})
@findex ord_seteq/2
@syindex ord_seteq/2
@cnindex ord_seteq/2
Holds when the two arguments represent the same set.

@item ord_setproduct(+@var{Set1}, +@var{Set2}, -@var{Set})
@findex ord_setproduct/3
@syindex ord_setproduct/3
@cnindex ord_setproduct/3
If Set1 and Set2 are ordered sets, Product will be an ordered
set of x1-x2 pairs.

@item ord_subset(+@var{Set1}, +@var{Set2})
@findex ordsubset/2
@syindex ordsubset/2
@cnindex ordsubset/2
Holds when every element of the ordered set @var{Set1} appears in the
ordered set @var{Set2}.

@item ord_subtract(+@var{Set1}, +@var{Set2}, ?@var{Difference})
@findex ord_subtract/3
@syindex ord_subtract/3
@cnindex ord_subtract/3
Holds when @var{Difference} contains all and only the elements of @var{Set1}
which are not also in @var{Set2}.

@item ord_symdiff(+@var{Set1}, +@var{Set2}, ?@var{Difference})
@findex ord_symdiff/3
@syindex ord_symdiff/3
@cnindex ord_symdiff/3
Holds when @var{Difference} is the symmetric difference of @var{Set1}
and @var{Set2}.

@item ord_union(+@var{Sets}, ?@var{Union})
@findex ord_union/2
@syindex ord_union/2
@cnindex ord_union/2
Holds when @var{Union} is the union of the lists @var{Sets}.

@item ord_union(+@var{Set1}, +@var{Set2}, ?@var{Union})
@findex ord_union/3
@syindex ord_union/3
@cnindex ord_union/3
Holds when @var{Union} is the union of @var{Set1} and @var{Set2}.

@item ord_union(+@var{Set1}, +@var{Set2}, ?@var{Union}, ?@var{Diff})
@findex ord_union/4
@syindex ord_union/4
@cnindex ord_union/4
Holds when @var{Union} is the union of @var{Set1} and @var{Set2} and
@var{Diff} is the difference.

@end table

@node Pseudo Random, Queues, Ordered Sets, Library
@section Pseudo Random Number Integer Generator
@cindex pseudo random

The following routines produce random non-negative integers in the range
0 .. 2^(w-1) -1, where w is the word size available for integers, e.g.,
32 for Intel machines and 64 for Alpha machines. Note that the numbers
generated by this random number generator are repeatable. This generator
was originally written by Allen Van Gelder and is based on Knuth Vol 2.

@table @code 
@item rannum(-@var{I})
@findex rannum/1
@snindex rannum/1
@cnindex rannum/1
Produces a random non-negative integer @var{I} whose low bits are not
all that random, so it should be scaled to a smaller range in general.
The integer @var{I} is in the range 0 .. 2^(w-1) - 1. You can use:
@example
rannum(X) :- yap_flag(max_integer,MI), rannum(R), X is R/MI.
@end example
to obtain a floating point number uniformly distributed between 0 and 1.

@item ranstart
@findex ranstart/0
@snindex ranstart/0
@cnindex ranstart/0
Initialise the random number generator using a built-in seed. The
@code{ranstart/0} built-in is always called by the system when loading
the package.

@item ranstart(+@var{Seed})
@findex ranstart/1
@snindex ranstart/1
@cnindex ranstart/1
Initialise the random number generator with user-defined @var{Seed}. The
same @var{Seed} always produces the same sequence of numbers.

@item ranunif(+@var{Range},-@var{I})
@findex ranunif/2
@snindex ranunif/2
@cnindex ranunif/2
@code{ranunif/2} produces a uniformly distributed non-negative random
integer @var{I} over a caller-specified range @var{R}.  If range is @var{R},
the result is in 0 .. @var{R}-1.

@end table

@node Queues, Random, Pseudo Random, Library
@section Queues
@cindex queue

The following queue manipulation routines are available once
included with the @code{use_module(library(queues))} command. Queues are
implemented with difference lists.

@table @code

@item make_queue(+@var{Queue})
@findex make_queue/1
@syindex make_queue/1
@cnindex make_queue/1
Creates a new empty queue. It should only be used to create a new queue.

@item join_queue(+@var{Element}, +@var{OldQueue}, -@var{NewQueue})
@findex join_queue/3
@syindex join_queue/3
@cnindex join_queue/3
Adds the new element at the end of the queue.

@item list_join_queue(+@var{List}, +@var{OldQueue}, -@var{NewQueue})
@findex list_join_queue/3
@syindex list_join_queue/3
@cnindex list_join_queue/3
Ads the new elements at the end of the queue.

@item jump_queue(+@var{Element}, +@var{OldQueue}, -@var{NewQueue})
@findex jump_queue/3
@syindex jump_queue/3
@cnindex jump_queue/3
Adds the new element at the front of the list.

@item list_jump_queue(+@var{List}, +@var{OldQueue}, +@var{NewQueue})
@findex list_jump_queue/3
@syindex list_jump_queue/3
@cnindex list_jump_queue/3
Adds all the elements of @var{List} at the front of the queue.

@item head_queue(+@var{Queue}, ?@var{Head})
@findex head_queue/2
@syindex head_queue/2
@cnindex head_queue/2
Unifies Head with the first element of the queue.

@item serve_queue(+@var{OldQueue}, +@var{Head}, -@var{NewQueue})
@findex serve_queue/3
@syindex serve_queue/3
@cnindex serve_queue/3
Removes the first element of the queue for service.

@item empty_queue(+@var{Queue})
@findex empty_queue/1
@syindex empty_queue/1
@cnindex empty_queue/1
Tests whether the queue is empty.

@item length_queue(+@var{Queue}, -@var{Length})
@findex length_queue/2
@syindex length_queue/2
@cnindex length_queue/2
Counts the number of elements currently in the queue.

@item list_to_queue(+@var{List}, -@var{Queue})
@findex list_to_queue/2
@syindex list_to_queue/2
@cnindex list_to_queue/2
Creates a new queue with the same elements as @var{List.}

@item queue_to_list(+@var{Queue}, -@var{List})
@findex queue_to_list/2
@syindex queue_to_list/2
@cnindex queue_to_list/2
Creates a new list with the same elements as @var{Queue}.

@end table


@node Random, RegExp, Queues, Library
@section Random Number Generator
@cindex queue

The following random number operations are included with the
@code{use_module(library(random))} command. Since Yap-4.3.19 Yap uses
the O'Keefe public-domain algorithm, based on the "Applied Statistics"
algorithm AS183.

@table @code

@item getrand(-@var{Key})
@findex getrand/1
@syindex getrand/1
@cnindex getrand/1
Unify @var{Key} with a term of the form @code{rand(X,Y,Z)} describing the
current state of the random number generator.

@item random(-@var{Number})
@findex random/1
@syindex random/1
@cnindex random/1
Unify @var{Number} with a floating-point number in the range @code{[0...1)}.

@item random(-@var{Number}, +@var{LOW}, +@var{HIGH})
@findex random/3
@syindex random/3
@cnindex random/3
Unify @var{Number} with a number in the range
@code{[LOW...HIGH)}. If both @var{LOW} and @var{HIGH} are
integers then @var{NUMBER} will also be an integer, otherwise
@var{NUMBER} will be a floating-point number.

@item randseq(+@var{LENGTH}, +@var{MAX}, -@var{Numbers})
@findex randseq/3
@syindex randseq/3
@cnindex randseq/3
Unify @var{Numbers} with a list of @var{LENGTH} unique random integers
in the range @code{[1�...@var{MAX})}.

@item randset(+@var{LENGTH}, +@var{MAX}, -@var{Numbers})
@findex randset/3
@syindex randset/3
@cnindex randset/3
Unify @var{Numbers} with an ordered list of @var{LENGTH} unique random
integers in the range @code{[1�...@var{MAX})}.

@item setrand(+@var{Key})
@findex setrand/1
@syindex setrand/1
@cnindex setrand/1
Use a term of the form @code{rand(X,Y,Z)} to set a new state for the
random number generator. The integer @code{X} must be in the range
@code{[1...30269)}, the integer @code{Y} must be in the range
@code{[1...30307)}, and the integer @code{Z} must be in the range
@code{[1...30323)}.

@end table

@node RegExp, Splay Trees, Random, Library
@section Regular Expressions
@cindex regular expressions

This library includes routines to determine whether a regular expression
matches part or all of a string. The routines can also return which
parts parts of the string matched the expression or subexpressions of
it. This library relies on Henry Spencer's @code{C}-package and is only
available in operating systems that support dynamic loading. The
@code{C}-code has been obtained from the sources of FreeBSD-4.0 and is
protected by copyright from Henry Spencer and from the Regents of the
University of California (see the file library/regex/COPYRIGHT for
further details).

Much of the description of regular expressions below is copied verbatim
from Henry Spencer's manual page.

A regular expression is zero or more branches, separated by ``|''.  It
matches anything that matches one of the branches.

A branch is zero or more pieces, concatenated.  It matches a match for
the first, followed by a match for the second, etc.

A piece is an atom possibly followed by ``*'', ``+'', or ``?''.  An atom
followed by ``*'' matches a sequence of 0 or more matches of the atom.
An atom followed by ``+'' matches a sequence of 1 or more matches of the
atom.  An atom followed by ``?'' matches a match of the atom, or the
null string.

An atom is a regular expression in parentheses (matching a match for the
regular expression), a range (see below), ``.''  (matching any single
character), ``^'' (matching the null string at the beginning of the
input string), ``$'' (matching the null string at the end of the input
string), a ``\'' followed by a single character (matching that
character), or a single character with no other significance (matching
that character).

A range is a sequence of characters enclosed in ``[]''.  It normally
matches any single character from the sequence.  If the sequence begins
with ``^'', it matches any single character not from the rest of the
sequence.  If two characters in the sequence are separated by ``-'',
this is shorthand for the full list of ASCII characters between them
(e.g. ``[0-9]'' matches any decimal digit).  To include a literal ``]''
in the sequence, make it the first character (following a possible
``^'').  To include a literal ``-'', make it the first or last
character.

@table @code

@item regexp(+@var{RegExp},+@var{String},+@var{Opts})
@findex regexp/3
@snindex regexp/3
@cnindex regexp/3

Match regular expression @var{RegExp} to input string @var{String}
according to options @var{Opts}. The options may be:
@itemize @bullet
@item @code{nocase}: Causes upper-case characters  in  @var{String} to
        be treated  as  lower case during the matching process.
@end itemize

@item regexp(+@var{RegExp},+@var{String},+@var{Opts},@var{SubMatchVars})
@findex regexp/4
@snindex regexp/4
@cnindex regexp/4

Match regular expression @var{RegExp} to input string @var{String}
according to options @var{Opts}. The variable @var{SubMatchVars} should
be originally a list of unbound variables all will contain a sequence of
matches, that is, the head of @var{SubMatchVars} will contain the
characters in @var{String} that matched the leftmost parenthesized
subexpression within @var{RegExp}, the next head of list will contain
the characters that matched the next parenthesized subexpression to the
right in @var{RegExp}, and so on.

The options may be:
@itemize @bullet
@item @code{nocase}: Causes upper-case characters  in  @var{String} to
        be treated  as  lower case during the matching process.
@item @code{indices}: Changes what  is  stored  in
@var{SubMatchVars}. Instead  of storing the matching characters from
@var{String}, each variable will contain a term of the form @var{IO-IF}
giving the indices in @var{String} of the first and last characters  in
the  matching range of characters.

@end itemize

In general there may be more than one way to match a regular expression
to an input string.  For example,  consider the command
@example
  regexp("(a*)b*","aabaaabb", [], [X,Y])
@end example
Considering only the rules given so far, @var{X} and @var{Y} could end up
with the values @code{"aabb"} and @code{"aa"}, @code{"aaab"} and
@code{"aaa"}, @code{"ab"} and @code{"a"}, or any of several other
combinations.  To resolve this potential ambiguity regexp chooses among
alternatives using the rule ``first then longest''.  In other words, it
considers the possible matches in order working from left to right
across the input string and the pattern, and it attempts to match longer
pieces of the input string before shorter ones.  More specifically, the
following rules apply in decreasing order of priority:


@enumerate 
@item    If a regular expression could match  two  different parts of an
input string then it will match the one that begins earliest.

@item  If a regular expression contains "|"  operators  then the leftmost matching sub-expression is chosen.

@item In *, +, and ? constructs, longer matches are chosen in preference to shorter ones.

@item In sequences of expression  components  the  components are considered from left to right.
@end enumerate

In the example from above, @code{"(a*)b*"} matches @code{"aab"}: the
@code{"(a*)"} portion of the pattern is matched first and it consumes
the leading @code{"aa"}; then the @code{"b*"} portion of the pattern
consumes the next @code{"b"}.  Or, consider the following example: 
@example
  regexp("(ab|a)(b*)c",  "abc", [], [X,Y,Z])
@end example

After this command @var{X} will be @code{"abc"}, @var{Y} will be
@code{"ab"}, and @var{Z} will be an empty string.  Rule 4 specifies that
@code{"(ab|a)"} gets first shot at the input string and Rule 2 specifies
that the @code{"ab"} sub-expression is checked before the @code{"a"}
sub-expression.  Thus the @code{"b"} has already been claimed before the
@code{"(b*)"} component is checked and @code{(b*)} must match an empty string.

@end table

@node Splay Trees, String I/O, RegExp, Library
@section Splay Trees
@cindex splay trees

Splay trees are explained in the paper "Self-adjusting Binary Search
Trees", by D.D. Sleator and R.E. Tarjan, JACM, vol. 32, No.3, July 1985,
p. 668. They are designed to support fast insertions, deletions and
removals in binary search trees without the complexity of traditional
balanced trees. The key idea is to allow the tree to become
unbalanced. To make up for this, whenever we find a node, we move it up
to the top. We use code by Vijay Saraswat originally posted to the Prolog
mailing-list.

@table @code

@item splay_access(-@var{Return},+@var{Key},?@var{Val},+@var{Tree},-@var{NewTree})
@findex splay_access/5
@snindex splay_access/5
@cnindex splay_access/5
If item @var{Key} is in tree @var{Tree}, return its @value{Val} and unify
@var{Return} with @code{true}. Otherwise unify @var{Return} with
@code{null}. The variable @var{NewTree} unifies with the new tree.

@item splay_delete(+@var{Key},?@var{Val},+@var{Tree},-@var{NewTree})
@findex splay_delete/4
@snindex splay_delete/4
@cnindex splay_delete/4
Delete item @var{Key} from tree @var{Tree}, assuming that it is present
already. The variable @var{Val} unifies with a value for key @var{Key},
and the variable @var{NewTree} unifies with the new tree. The predicate
will fail if @var{Key} is not present.

@item splay_init(-@var{NewTree})
@findex splay_init/3
@snindex splay_init/3
@cnindex splay_init/3
Initialise a new splay tree.

@item splay_insert(+@var{Key},?@var{Val},+@var{Tree},-@var{NewTree})
@findex splay_insert/4
@snindex splay_insert/4
@cnindex splay_insert/4
Insert item @var{Key} in tree @var{Tree}, assuming that it is not there
already. The variable @var{Val} unifies with a value for key @var{Key},
and the variable @var{NewTree} unifies with the new tree. In our
implementation, @var{Key} is not inserted if it is already there:
rather it is unified with the item already in the tree.

@item splay_join(+@var{LeftTree},+@var{RighTree},-@var{NewTree})
@findex splay_join/3
@snindex splay_join/3
@cnindex splay_join/3
Combine trees @var{LeftTree} and @var{RighTree} into a single
tree@var{NewTree}  containing all items from both trees. This operation
assumes that all items in @var{LeftTree} are less than all those in
@var{RighTree} and destroys both @var{LeftTree} and @var{RighTree}.

@item splay_split(+@var{Key},?@var{Val},+@var{Tree},-@var{LeftTree},-@var{RightTree})
@findex splay_split/5
@snindex splay_split/5
@cnindex splay_split/5
Construct and return two trees @var{LeftTree} and @var{RightTree}, where
@var{LeftTree} contains all items in @var{Tree} less than @var{Key}, and
@var{RightTree} contains all items in @var{Tree} greater than
@var{Key}. This operations destroys @var{Tree}.

@end table

@node String I/O, System, Splay Trees, Library
@section Reading From and Writing To Strings
@cindex string I/O

From Version 4.3.2 onwards YAP implements SICStus Prolog compatible
String I/O. The library allows users to read from and write to a memory
buffer as if it was a file. The memory buffer is built from or converted
to a string of character codes by the routines in library. Therefore, if
one wants to read from a string the string must be fully instantiated
before the library builtin opens the string for reading. These commands
are available through the @code{use_module(library(charsio))} command.

@table @code

@item format_to_chars(+@var{Form}, +@var{Args}, -@var{Result})
@findex format_to_chars/3
@syindex format_to_chars/3
@cnindex format_to_chars/3

Execute the built-in procedure @code{format/2} with form @var{Form} and
arguments @var{Args} outputting the result to the string of character
codes @var{Result}.

@item format_to_chars(+@var{Form}, +@var{Args}, -@var{Result0}, -@var{Result})
@findex format_to_chars/4
@syindex format_to_chars/4
@cnindex format_to_chars/4

Execute the built-in procedure @code{format/2} with form @var{Form} and
arguments @var{Args} outputting the result to the difference list of
character codes @var{Result-Result0}.

@item write_to_chars(+@var{Term}, -@var{Result})
@findex write_to_chars/2
@syindex write_to_chars/2
@cnindex write_to_chars/2

Execute the built-in procedure @code{write/1} with argument @var{Term}
outputting the result to the string of character codes @var{Result}.

@item write_to_chars(+@var{Term}, -@var{Result0}, -@var{Result})
@findex write_to_chars/3
@syindex write_to_chars/3
@cnindex write_to_chars/3

Execute the built-in procedure @code{write/1} with argument @var{Term}
outputting the result to the difference list of character codes
@var{Result-Result0}.

@item atom_to_chars(+@var{Atom}, -@var{Result})
@findex atom_to_chars/2
@syindex atom_to_chars/2
@cnindex atom_to_chars/2

Convert the atom @var{Atom} to the string of character codes
@var{Result}.

@item atom_to_chars(+@var{Atom}, -@var{Result0}, -@var{Result})
@findex atom_to_chars/3
@syindex atom_to_chars/3
@cnindex atom_to_chars/3

Convert the atom @var{Atom} to the difference list of character codes
@var{Result-Result0}.

@item number_to_chars(+@var{Number}, -@var{Result})
@findex number_to_chars/2
@syindex number_to_chars/2
@cnindex number_to_chars/2

Convert the number @var{Number} to the string of character codes
@var{Result}.

@item number_to_chars(+@var{Number}, -@var{Result0}, -@var{Result})
@findex number_to_chars/3
@syindex number_to_chars/3
@cnindex number_to_chars/3

Convert the atom @var{Number} to the difference list of character codes
@var{Result-Result0}.

@item read_from_chars(+@var{Chars}, -@var{Term})
@findex read_from_chars/2
@syindex read_from_chars/2
@cnindex read_from_chars/2

Parse the list of character codes @var{Chars} and return the result in
the term @var{Term}. The character codes to be read must terminate with
a dot character such that either (i) the dot character is followed by
blank characters; or (ii) the dot character is the last character in the
string.

@item open_chars_stream(+@var{Chars}, -@var{Stream})
@findex open_chars_stream/2
@syindex open_chars_stream/2
@cnindex open_chars_stream/2

Open the list of character codes @var{Chars} as a stream @var{Stream}.

@item with_output_to_chars(?@var{Goal}, -@var{Chars})
@findex with_output_to_chars/2
@syindex with_output_to_chars/2
@cnindex with_output_to_chars/2

Execute goal @var{Goal} such that its standard output will be sent to a
memory buffer. After successful execution the contents of the memory
buffer will be converted to the list of character codes @var{Chars}.

@item with_output_to_chars(?@var{Goal}, ?@var{Chars0}, -@var{Chars})
@findex with_output_to_chars/3
@syindex with_output_to_chars/3
@cnindex with_output_to_chars/3

Execute goal @var{Goal} such that its standard output will be sent to a
memory buffer. After successful execution the contents of the memory
buffer will be converted to the difference list of character codes
@var{Chars-Chars0}.

@item with_output_to_chars(?@var{Goal}, -@var{Stream}, ?@var{Chars0}, -@var{Chars})
@findex with_output_to_chars/4
@syindex with_output_to_chars/4
@cnindex with_output_to_chars/4

Execute goal @var{Goal} such that its standard output will be sent to a
memory buffer. After successful execution the contents of the memory
buffer will be converted to the difference list of character codes
@var{Chars-Chars0} and @var{Stream} receives the stream corresponding to
the memory buffer.

@end table

The implementation of the character IO operations relies on three YAP
builtins:
@table @code

@item charsio:open_mem_read_stream(+@var{String}, -@var{Stream})
Store a string in a memory buffer and output a stream that reads from this
memory buffer.

@item charsio:open_mem_write_stream(-@var{Stream})
Create a new memory buffer and output a stream that writes to  it.

@item charsio:peek_mem_write_stream(-@var{Stream}, L0, L)
Convert the memory buffer associated with stream @var{Stream} to the
difference list of character codes @var{L-L0}.

@end table
@noindent
These builtins are initialised to belong to the module @code{charsio} in
@code{init.yap}. Novel procedures for manipulating strings by explicitly
importing these built-ins.

YAP does not currently support opening a @code{charsio} stream in
@code{append} mode, or seeking in such a stream.

@node System, Terms, String I/O, Library
@section Calling The Operating System from YAP
@cindex Operating System Utilities

Yap now provides a library of system utilities compatible with the
SICStus Prolog system library. This library extends and to some point
replaces the functionality of Operating System access routines. The
library includes Unix/Linux and Win32 @code{C} code. They
are available through the @code{use_module(library(system))} command.

@table @code

@item datime(datime(-@var{Year}, -@var{Month}, -@var{DayOfTheMonth},
-@var{Hour}, -@var{Minute}, -@var{Second})
@findex  datime/1
@syindex datime/1
@cnindex datime/1
The @code{datime/1} procedure returns the current date and time, with
information on @var{Year}, @var{Month}, @var{DayOfTheMonth},
@var{Hour}, @var{Minute}, and @var{Second}. The @var{Hour} is returned
on local time. This function uses the WIN32
@code{GetLocalTime} function or the Unix @code{localtime} function.

@example
   ?- datime(X).

X = datime(2001,5,28,15,29,46) ? 
@end example

@item delete_file(+@var{File})
@findex  delete_file/1
@syindex delete_file/1
@cnindex delete_file/1
The @code{delete_file/1} procedure removes file @var{File}. If
@var{File} is a directory, remove the directory @emph{and all its
subdirectories}.

@example
   ?- delete_file(x).
@end example

@item delete_file(+@var{File},+@var{Opts})
@findex  delete_file/2
@syindex delete_file/2
@cnindex delete_file/2
The @code{delete_file/2} procedure removes file @var{File} according to
options @var{Opts}. These options are @code{directory} if one should
remove directories, @code{recursive} if one should remove directories
recursively, and @code{ignore} if errors are not to be reported.

This example is equivalent to using the @code{delete_file/1} predicate:
@example
   ?- delete_file(x, [recursive]).
@end example


@item directory_files(+@var{Dir},+@var{List})
@findex  directory_files/2
@syindex directory_files/2
@cnindex directory_files/2
Given a directory @var{Dir},  @code{directory_files/2} procedures a
listing of all files and directories in the directory:
@example
    ?- directory_files('.',L), writeq(L).
['Makefile.~1~','sys.so','Makefile','sys.o',x,..,'.']
@end example
The predicates uses the @code{dirent} family of routines in Unix
environments, and @code{findfirst} in WIN32.

@item file_exists(+@var{File})
@findex  file_exists/1
@syindex file_exists/1
@cnindex file_exists/1
The atom @var{File} corresponds to an existing file.

@item file_exists(+@var{File},+@var{Permissions})
@findex  file_exists/2
@syindex file_exists/2
@cnindex file_exists/2
The atom @var{File} corresponds to an existing file with permissions
compatible with @var{Permissions}. YAP currently only accepts for
permissions to be described as a number. The actual meaning of this
number is Operating System dependent.

@item file_property(+@var{File},?@var{Property})
@findex  file_property/2
@syindex file_property/2
@cnindex file_property/2
The atom @var{File} corresponds to an existing file, and @var{Property}
will be unified with a property of this file. The poperties are of the
form @code{type(@var{Type})}, which gives whether the file is a regular
file, a directory, a fifo file, or of unknown type;
@code{size(@var{Size})}, with gives the size for a file, and
@code{mod_time(@var{Time})}, which gives the last time a file was
modified according to some Operating System dependent
timestamp. Properties can be obtained through backtracking:

@example
   ?- file_property('Makefile',P).

P = type(regular) ? ;

P = size(2375) ? ;

P = mod_time(990826911) ? ;

no
@end example

@item make_directory(+@var{Dir})
@findex  make_directory/2
@syindex make_directory/2
@cnindex make_directory/2
Create a directory @var{Dir}. The name of the directory must be an atom.

@item rename_file(+@var{OldFile},+@var{NewFile})
@findex  rename_file/2
@syindex rename_file/2
@cnindex rename_file/2
Create file @var{OldFile} to @var{NewFile}. This predicate uses the
@code{C} builtin function @code{rename}.


@item environ(?@var{EnvVar},+@var{EnvValue})
@findex  environ/2
@syindex environ/2
@cnindex environ/2
Unify environment variable @var{EnvVar} with its value @var{EnvValue},
if there is one. This predicate is backtrackable in Unix systems, but
not currently in Win32 configurations.

@example
   ?- environ('HOME',X).

X = 'C:\\cygwin\\home\\administrator' ?
@end example

@item host_id(-@var{Id})
@findex  host_id/1
@syindex host_id/1
@cnindex host_id/1

Unify @var{Id} with an identifier of the current host. Yap uses the
@code{hostid} function when available, 

@item host_name(-@var{Name})
@findex  host_name/1
@syindex host_name/1
@cnindex host_name/1

Unify @var{Name} with a name for the current host. Yap uses the
@code{hostname} function in Unix systems when available, and the
@code{GetComputerName} function in WIN32 systems. 

@item kill(@var{Id},+@var{SIGNAL})
@findex  kill/2
@syindex kill/2
@cnindex kill/2

Send signal @var{SIGNAL} to process @var{Id}. In Unix this predicate is
a direct interface to @code{kill} so one can send signals to groups of
processes. In WIN32 the predicate is an interface to
@code{TerminateProcess}, so it kills @var{Id} indepent of @var{SIGNAL}.

@item mktemp(@var{Spec},-@var{File})
@findex  mktemp/2
@syindex mktemp/2
@cnindex mktemp/2

Direct interface to @code{mktemp}: given a @var{Spec}, that is a file
name with six @var{X} to it, create a file name @var{File}. Use
@code{tmpnam/1} instead.

@item pid(-@var{Id})
@findex  pid/1
@syindex pid/1
@cnindex pid/1

Unify @var{Id} with the process identifier for the current
process. An interface to the @t{getpid} function.

@item tmpnam(-@var{File})
@findex  tmpnam/1
@syindex tmpnam/1
@cnindex tmpnam/1

Interface with @var{tmpnam}: create an unique file and unify its name
with @var{File}.

@item
exec(+@var{Command},[+@var{InputStream},+@var{OutputStream},+@var{ErrorStream}],
-@var{Status})
@findex  exec/3
@syindex exec/3
@cnindex exec/3
Execute command @var{Command} with its streams connected to
@var{InputStream}, @var{OutputStream}, and @var{ErrorStream}. The result
for the command is returned in @var{Status}. The command is executed by
the default shell @code{bin/sh -c} in Unix.

The following example demonstrates the use of @code{exec/3} to send a
command and process its output:

@example
exec(ls,[std,pipe(S),null],P),repeat, get0(S,C), (C = -1, close(S) ! ; put(C)).
@end example

The streams may be one of standard stream, @code{std}, null stream,
@code{null}, or @code{pipe(S)}, where @var{S} is a pipe stream. Note
that it is up to the user to close the pipe.

@item working_directory(-@var{CurDir},?@var{NextDir})
@findex  working_directory/2
@syindex working_directory/2
@cnindex working_directory/2
Fetch the current directory at @var{CurDir}. If @var{NextDir} is bound
to an atom, make its value the current working directory.

@item popen(+@var{Command}, +@var{TYPE}, -@var{Stream})
@findex  popen/3
@syindex popen/3
@cnindex popen/3
Interface to the @t{popen} function. It opens a process by creating a
pipe, forking and invoking @var{Command} on the current shell. Since a
pipe is by definition unidirectional the @var{Type} argument may be
@code{read} or @code{write}, not both. The stream should be closed
using @code{close/1}, there is no need for a special @code{pclose}
command.

The following example demonstrates the use of @code{popen/3} to process
the output of a command, as @code{exec/3} would do:

@example
   ?- popen(ls,read,X),repeat, get0(X,C), (C = -1, ! ; put(C)).

X = 'C:\\cygwin\\home\\administrator' ?
@end example


The WIN32 implementation of @code{popen/3} relies on @code{exec/3}.

@item shell
@findex  shell/0
@syindex shell/0
@cnindex shell/0
Start a new shell and leave Yap in background until the shell
completes. Yap uses the shell given by the environment variable
@code{SHELL}. In WIN32 environment YAP will use @code{COMSPEC} if
@code{SHELL} is undefined.

@item shell(+@var{Command})
@findex  shell/1
@syindex shell/1
@cnindex shell/1
Execute command @var{Command} under a new shell. Yap will be in
background until the command completes. In Unix environments Yap uses
the shell given by the environment variable @code{SHELL} with the option
@code{" -c "}. In WIN32 environment YAP will use @code{COMSPEC} if
@code{SHELL} is undefined, in this case with the option @code{" /c "}.

@item shell(+@var{Command},-@var{Status})
@findex  shell/1
@syindex shell/1
@cnindex shell/1
Execute command @var{Command} under a new shell and unify @var{Status}
with the exit for the command. Yap will be in background until the
command completes. In Unix environments Yap uses the shell given by the
environment variable @code{SHELL} with the option @code{" -c "}. In
WIN32 environment YAP will use @code{COMSPEC} if @code{SHELL} is
undefined, in this case with the option @code{" /c "}.

@item sleep(+@var{Time})
@findex  sleep/1
@syindex sleep/1
@cnindex sleep/1
Block the current process for @var{Time} seconds. The number of seconds
must be a positive number, and it may an integer or a float. The Unix
implementation uses @code{usleep} if the number of seconds is below one,
and @code{sleep} if it is over a second. The WIN32 implementation uses
@code{Sleep} for both cases.

@item system
@findex  system/0
@syindex system/0
@cnindex system/0
Start a new default shell and leave Yap in background until the shell
completes. Yap uses @code{/bin/sh} in Unix systems and @code{COMSPEC} in
WIN32.

@item system(+@var{Command},-@var{Res})
@findex  system/2
@syindex system/2
@cnindex system/2
Interface to @code{system}: execute command @var{Command} and unify
@var{Res} with the result.

@item wait(+@var{PID},-@var{Status})
@findex  wait/2
@syindex wait/2
@cnindex wait/2
Wait until process @var{PID} terminates, and return its exits @var{Status}.

@end table


@node Terms, Timeout, System, Library
@section Utilities On Terms
@cindex utilities on terms

The next routines provide a set of commonly used utilities to manipulate
terms. Most of these utilities have been implemented in @code{C} for
efficiency. They are available through the
@code{use_module(library(timeout))} command.

@table @code

@item acyclic_term(?@var{Term})
@findex cyclic_term/1
@syindex cyclic_term/1
@cnindex cyclic_term/1
Succeed if the argument @var{Term} is an acyclic term.

@item cyclic_term(?@var{Term})
@findex cyclic_term/1
@syindex cyclic_term/1
@cnindex cyclic_term/1
Succeed if the argument @var{Term} is a cyclic term.

@item term_hash(+@var{Term}, ?@var{Hash})
@findex  term_hash/2
@syindex term_hash/2
@cnindex term_hash/2

If @var{Term} is ground unify @var{Hash} with a positive integer
calculated from the structure of the term. Otherwise the argument
@var{Hash} is left unbound. The range of the positive integer is from
@code{0} to, but not including, @code{33554432}.

@item term_hash(+@var{Term}, +@var{Depth}, +@var{Range}, ?@var{Hash})
@findex  term_hash/4
@syindex term_hash/4
@cnindex term_hash/4

Unify @var{Hash} with a positive integer calculated from the structure
of the term.  The range of the positive integer is from @code{0} to, but
not including, @var{Range}. If @var{Depth} is @code{-1} the whole term
is considered. Otherwise, the term is considered only up to depth
@code{1}, where the constants and the principal functor have depth
@code{1}, and an argument of a term with depth @var{I} has depth @var{I+1}. 

@item term_variables(?@var{Term}, -@var{Variables})
@findex  term_variables/2
@syindex term_variables/2
@cnindex term_variables/2

Unify @var{Variables} with a list of all variables in term @var{Term}.

@item variant(?@var{Term1}, ?@var{Term2})
@findex  variant/2
@syindex variant/2
@cnindex variant/2

Succeed if @var{Term1} and @var{Term2} are variant terms.

@item subsumes(?@var{Term1}, ?@var{Term2})
@findex  subsumes/2
@syindex subsumes/2
@cnindex subsumes/2

Succeed if @var{Term1} subsumes @var{Term2}.  Variables in term
@var{Term1} are bound so that the two terms become equal.


@item subsumes_chk(?@var{Term1}, ?@var{Term2})
@findex  subsumes_chk/2
@syindex subsumes_chk/2
@cnindex subsumes_chk/2

Succeed if @var{Term1} subsumes @var{Term2} but does not bind any
variable in @var{Term1}.

@item variable_in_term(?@var{Term},?@var{Var})
@findex variable_in_term/2
@snindex variable_in_term/2
@cnindex variable_in_term/2
Succeed if the second argument @var{Var} is a variable and occurs in
term @var{Term}.


@end table

@node Timeout, Trees, Terms, Library
@section Calls With Timeout
@cindex timeout

The @t{time_out/3} command relies on the @t{alarm/3} built-in to
implement a call with a maximum time of execution. The command is
available with the @code{use_module(library(timeout))} command.

@table @code

@item time_out(+@var{Goal}, +@var{Timeout}, -@var{Result})
@findex time_out/3
@syindex time_out/3
@cnindex time_out/3
Execute goal @var{Goal} with time limite @var{Timeout}, where
@var{Timeout} is measured in milliseconds. If the goal succeeds, unify
@var{Result} with success. If the timer expires before the goal
terminates, unify @var{Result} with @t{timeout}.

This command is implemented by activating an alarm at procedure
entry. If the timer expires before the goal completes, the alarm will
through an exception @var{timeout}.

One should note that @code{time_out/3} is not reentrant, that is, a goal
called from @code{time_out} should never itself call
@t{time_out}. Moreover, @code{time_out/3} will deactivate any previous
alarms set by @code{alarm/3} and vice-versa, hence only one of these
calls should be used in a program.

Last, even though the timer is set in milliseconds, the current
implementation relies on @t{alarm/3}, and therefore can only offer
precision on the scale of seconds.

@end table

@node Trees, UGraphs, Timeout, Library
@section Updatable Binary Trees
@cindex updatable tree

The following queue manipulation routines are available once
included with the @code{use_module(library(trees))} command.

@table @code

@item get_label(+@var{Index}, +@var{Tree}, ?@var{Label})
@findex get_label/3
@syindex get_label/3
@cnindex get_label/3
Treats the tree as an array of @var{N} elements and returns the
@var{Index}-th.

@item list_to_tree(+@var{List}, -@var{Tree})
@findex list_to_tree/2
@syindex list_to_tree/2
@cnindex list_to_tree/2
Takes a given @var{List} of @var{N} elements and constructs a binary
@var{Tree}.

@item map_tree(+@var{Pred}, +@var{OldTree}, -@var{NewTree})
@findex map_tree/3
@syindex map_tree/3
@cnindex map_tree/3
Holds when @var{OldTree} and @var{NewTree} are binary trees of the same shape
and @code{Pred(Old,New)} is true for corresponding elements of the two trees.

@item put_label(+@var{Index}, +@var{OldTree}, +@var{Label}, -@var{NewTree})
@findex put_label/4
@syindex put_label/4
@cnindex put_label/4
constructs a new tree the same shape as the old which moreover has the
same elements except that the @var{Index}-th one is @var{Label}.

@item tree_size(+@var{Tree}, -@var{Size})
@findex tree_size/2
@syindex tree_size/2
@cnindex tree_size/2
Calculates the number of elements in the @var{Tree}.

@item tree_to_list(+@var{Tree}, -@var{List})
@findex tree_to_list/2
@syindex tree_to_list/2
@cnindex tree_to_list/2
Is the converse operation to list_to_tree.

@end table

@node UGraphs, , Trees, Library
@section Unweighted Graphs
@cindex unweighted graphs

The following graph manipulation routines are based from code originally
written by Richard O'Keefe. The code was then extended to be compatible
with the SICStus Prolog ugraphs library. The routines assume directed
graphs, undirected graphs may be implemented by using two edges. Graphs
are represented in one of two ways:

@itemize @bullet
@item The P-representation of a graph is a list of (from-to) vertex
pairs, where the pairs can be in any old order.  This form is
convenient for input/output.
 
@item The S-representation of a graph is a list of (vertex-neighbours)
pairs, where the pairs are in standard order (as produced by keysort)
and the neighbours of each vertex are also in standard order (as
produced by sort).  This form is convenient for many calculations.
@end itemize

These builtins are available once included with the
@code{use_module(library(ugraphs))} command.

@table @code

@item vertices_edges_to_ugraph(+@var{Vertices}, +@var{Edges}, -@var{Graph})
@findex  vertices_edges_to_ugraph/3
@syindex vertices_edges_to_ugraph/3
@cnindex vertices_edges_to_ugraph/3
Given a graph with a set of vertices @var{Vertices} and a set of edges
@var{Edges}, @var{Graph} must unify with the corresponding
s-representation. Note that the vertices without edges will appear in
@var{Vertices} but not in @var{Edges}. Moreover, it is sufficient for a
vertice to appear in @var{Edges}.
@example
?- vertices_edges_to_ugraph([],[1-3,2-4,4-5,1-5],L).

L = [1-[3,5],2-[4],3-[],4-[5],5-[]] ? 

@end example
In this case all edges are defined implicitly. The next example shows
three unconnected edges:
@example 
?- vertices_edges_to_ugraph([6,7,8],[1-3,2-4,4-5,1-5],L).

L = [1-[3,5],2-[4],3-[],4-[5],5-[],6-[],7-[],8-[]] ? 

@end example

@item vertices(+@var{Graph}, -@var{Vertices})
@findex  vertices/2
@syindex vertices/2
@cnindex vertices/2
Unify @var{Vertices} with all vertices appearing in graph
@var{Graph}. In the next example:
@example
?- vertices([1-[3,5],2-[4],3-[],4-[5],5-[]], V).

L = [1,2,3,4,5]
@end example

@item edges(+@var{Graph}, -@var{Edges})
@findex  vertices/2
@syindex vertices/2
@cnindex vertices/2
Unify @var{Edges} with all edges appearing in graph
@var{Graph}. In the next example:
@example
?- vertices([1-[3,5],2-[4],3-[],4-[5],5-[]], V).

L = [1,2,3,4,5]
@end example

@item add_vertices(+@var{Graph}, +@var{Vertices}, -@var{NewGraph})
@findex  add_vertices/3
@syindex add_vertices/3
@cnindex add_vertices/3
Unify @var{NewGraph} with a new graph obtained by adding the list of
vertices @var{Vertices} to the graph @var{Graph}. In the next example:
@example
?- add_vertices([1-[3,5],2-[4],3-[],4-[5],
                 5-[],6-[],7-[],8-[]],
                [0,2,9,10,11],
                   NG).

NG = [0-[],1-[3,5],2-[4],3-[],4-[5],5-[],
      6-[],7-[],8-[],9-[],10-[],11-[]]
@end example

@item del_vertices(+@var{Vertices}, +@var{Graph}, -@var{NewGraph})
@findex  del_vertices/3
@syindex del_vertices/3
@cnindex del_vertices/3
Unify @var{NewGraph} with a new graph obtained by deleting the list of
vertices @var{Vertices} and all the edges that start from or go to a
vertex in @var{Vertices} to the graph @var{Graph}. In the next example:
@example
?- del_vertices([2,1],[1-[3,5],2-[4],3-[],
                 4-[5],5-[],6-[],7-[2,6],8-[]],NL).

NL = [3-[],4-[5],5-[],6-[],7-[6],8-[]]
@end example

@item add_edges(+@var{Graph}, +@var{Edges}, -@var{NewGraph})
@findex  add_edges/3
@syindex add_edges/3
@cnindex add_edges/3
Unify @var{NewGraph} with a new graph obtained by adding the list of
edges @var{Edges} to the graph @var{Graph}. In the next example:
@example
?- add_edges([1-[3,5],2-[4],3-[],4-[5],5-[],6-[],
              7-[],8-[]],[1-6,2-3,3-2,5-7,3-2,4-5],NL).

NL = [1-[3,5,6],2-[3,4],3-[2],4-[5],5-[7],6-[],7-[],8-[]]
@end example

@item sub_edges(+@var{Graph}, +@var{Edges}, -@var{NewGraph})
@findex  sub_edges/3
@syindex sub_edges/3
@cnindex sub_edges/3
Unify @var{NewGraph} with a new graph obtained by removing the list of
edges @var{Edges} from the graph @var{Graph}. Notice that no vertices
are deleted. In the next example:
@example
?- del_edges([1-[3,5],2-[4],3-[],4-[5],5-[],
              6-[],7-[],8-[]],
             [1-6,2-3,3-2,5-7,3-2,4-5,1-3],NL).

NL = [1-[5],2-[4],3-[],4-[],5-[],6-[],7-[],8-[]]
@end example

@item transpose(+@var{Graph}, -@var{NewGraph})
@findex  transpose/3
@syindex transpose/3
@cnindex transpose/3
Unify @var{NewGraph} with a new graph obtained from @var{Graph} by
replacing all edges of the form @var{V1-V2} by edges of the form
@var{V2-V1}. The cost is @code{O(|V|^2)}. In the next example:
@example
?- transpose([1-[3,5],2-[4],3-[],
              4-[5],5-[],6-[],7-[],8-[]], NL).

NL = [1-[],2-[],3-[1],4-[2],5-[1,4],6-[],7-[],8-[]]
@end example
Notice that an undirected graph is its own transpose.

@item neighbors(+@var{Vertex}, +@var{Graph}, -@var{Vertices})
@findex  neighbors/3
@syindex neighbors/3
@cnindex neighbors/3
Unify @var{Vertices} with the list of neighbors of vertex @var{Vertex}
in @var{Graph}. If the vertice is not in the graph fail. In the next
example:
@example
?- neighbors(4,[1-[3,5],2-[4],3-[],
                4-[1,2,7,5],5-[],6-[],7-[],8-[]],
             NL).

NL = [1,2,7,5]
@end example

@item neighbours(+@var{Vertex}, +@var{Graph}, -@var{Vertices})
@findex  neighbours/3
@syindex neighbours/3
@cnindex neighbours/3
Unify @var{Vertices} with the list of neighbours of vertex @var{Vertex}
in @var{Graph}. In the next example:
@example
?- neighbours(4,[1-[3,5],2-[4],3-[],
                 4-[1,2,7,5],5-[],6-[],7-[],8-[]], NL).

NL = [1,2,7,5]
@end example

@item complement(+@var{Graph}, -@var{NewGraph})
@findex  complement/2
@syindex complement/2
@cnindex complement/2
Unify @var{NewGraph} with the graph complementar to @var{Graph}.
 In the next example:
@example
?- complement([1-[3,5],2-[4],3-[],
               4-[1,2,7,5],5-[],6-[],7-[],8-[]], NL).

NL = [1-[2,4,6,7,8],2-[1,3,5,6,7,8],3-[1,2,4,5,6,7,8],
      4-[3,5,6,8],5-[1,2,3,4,6,7,8],6-[1,2,3,4,5,7,8],
      7-[1,2,3,4,5,6,8],8-[1,2,3,4,5,6,7]]
@end example

@item compose(+@var{LeftGraph}, +@var{RightGraph}, -@var{NewGraph})
@findex  compose/3
@syindex compose/3
@cnindex compose/3
Compose the graphs @var{LeftGraph} and @var{RightGraph} to form @var{NewGraph}.
 In the next example:
@example
?- compose([1-[2],2-[3]],[2-[4],3-[1,2,4]],L).

L = [1-[4],2-[1,2,4],3-[]]
@end example

@item top_sort(+@var{Graph}, +@var{Sort})
@findex  top_sort/2
@syindex top_sort/2
@cnindex top_sort/2
Generate the set of nodes @var{Sort} as a topological sorting of graph
@var{Graph}, if one is possible.
 In the next example we show how topological sorting works for a linear graph:
@example
?- top_sort([_138-[_219],_219-[_139], _139-[]],L).

L = [_138,_219,_139]
@end example

@item transitive_closure(+@var{Graph}, +@var{Closure})
@findex  transitive_closure/2
@syindex transitive_closure/2
@cnindex transitive_closure/2
Generate the graph @var{Closure} as the transitive closure of graph
@var{Graph}.
 In the next example:
@example
?- transitive_closure([1-[2,3],2-[4,5],4-[6]],L).

L = [1-[2,3,4,5,6],2-[4,5,6],4-[6]]
@end example

@item reachable(+@var{Node}, +@var{Graph}, -@var{Vertices})
@findex  reachable/3
@syindex reachable/3
@cnindex reachable/3
Unify @var{Vertices} with the set of all vertices in graph
@var{Graph} that are reachable from @var{Node}. In the next example:
@example
?- reachable(1,[1-[3,5],2-[4],3-[],4-[5],5-[]],V).

V = [1,3,5]
@end example

@end table

@node Extensions,Debugging,Library,Top 
@chapter Extensions

YAP includes several extensions that are not enabled by
default, but that can be used to extend the functionality of the
system. These options can be set at compilation time by enabling the
related compilation flag, as explained in the @code{Makefile}

@menu

Extensions to Traditional Prolog

* Rational Trees:: Working with Rational Trees
* Coroutining:: Changing the Execution of Goals
* Attributed Variables:: Using attributed Variables
* CLPQR:: The CLP(Q,R) System
* Logtalk:: The Logtalk Object-Oriented system
* Parallelism:: Running in Or-Parallel
* Tabling:: Storing Intermediate Solutions of programs 
* Low Level Profiling:: Profiling Abstract Machine Instructions
* Low Level Tracing:: Tracing at Abstract Machine Level
@end menu

@node Rational Trees, Coroutining, , Extensions
@chapter Rational Trees

Prolog unification is not a complete implementation. For efficiency
considerations, Prolog systems do not perform occur checks while
unifying terms. As an example, @code{X = a(X)} will not fail but instead
will create an infinite term of the form @code{a(a(a(a(a(...)))))}, or
@emph{rational tree}.

By default, rational trees are not supported in YAP, and these
terms can easily lead to infinite computation. For example, @code{X =
a(X), X = X} will enter an infinite loop.

The @code{RATIONAL_TREES} flag improves support for these
terms. Internal primitives are now aware that these terms can exist, and
will not enter infinite loops. Hence, the previous unification will
succeed. Another example, @code{X = a(X), ground(X)} will succeed
instead of looping. Other affected builtins include the term comparison
primitives, @code{numbervars/3}, @code{copy_term/2}, and the internal
data base routines. The support does not extend to Input/Output routines
or to @code{assert/1} YAP does not allow directly reading
rational trees, and you need to use @code{write_depth/2} to avoid
entering an infinite cycle when trying to write an infinite term.

@node Coroutining, Attributed Variables, Rational Trees, Extensions
@chapter Coroutining

Prolog uses a simple left-to-right flow of control. It is sometimes
convenient to change this control so that goals will only be executed
when conditions are fulfilled. This may result in a more "data-driven"
execution, or may be necessary to correctly implement extensions such as
negation by default.

The @code{COROUTINING} flag enables this option. Note that the support for
coroutining  will in general slow down execution.

The following declaration is supported:

@table @code
@item block/1
The argument to @code{block/1} is a condition on a goal or a conjunction
of conditions, with each element separated by commas. Each condition is
of the form @code{predname(@var{C1},...,@var{CN})}, where @var{N} is the
arity of the goal, and each @var{CI} is of the form @code{-}, if the
argument must suspend until the variable is bound, or @code{?}, otherwise.

@item wait/1
The argument to @code{wait/1} is a predicate descriptor or a conjunction
of these predicates. These predicates will suspend until their first
argument is bound.
@end table

The following primitives are supported:

@table @code
@item dif(@var{X},@var{Y})
@findex dif/2
@syindex dif/2
@cnindex dif/2
Succeed if the two arguments do not unify. A call to @code{dif/2} will
suspend if unification may still succeed or fail, and will fail if they
always unify.

@item freeze(?@var{X},:@var{G})
@findex freeze/2
@syindex freeze/2
@cnindex freeze/2
Delay execution of goal @var{G} until the variable @var{X} is bound.

@item frozen(@var{X},@var{G})
@findex frozen/2
@syindex frozen/2
@cnindex frozen/2
Unify @var{G} with a conjunction of goals suspended on variable @var{X},
or @code{true} if no goal has suspended.

@item when(+@var{C},:@var{G})
@findex when/2
@syindex when/2
@cnindex when/2
Delay execution of goal @var{G} until the conditions @var{C} are
satisfied. The conditions are of the following form:

@table @code
@item @var{C1},@var{C2}
Delay until both conditions @var{C1} and @var{C2} are satisfied.
@item @var{C1};@var{C2}
Delay until either condition @var{C1} or condition @var{C2} is satisfied.
@item ?=(@var{V1},@var{C2})
Delay until terms @var{V1} and @var{V1} have been unified.
@item nonvar(@var{V})
Delay until variable @var{V} is bound.
@item ground(@var{V})
Delay until variable @var{V} is ground.
@end table

Note that @code{when/2} will fail if the conditions fail.

@item call_residue(:@var{G},@var{L})
@findex call_residue/2
@syindex call_residue/2
@cnindex call_residue/2

Call goal @var{G}. If subgoals of @var{G} are still blocked, return
a list containing these goals and the variables they are blocked in. The
goals are then considered as unblocked. The next example shows a case
where @code{dif/2} suspends twice, once outside @code{call_residue/2},
and the other inside:

@example
?- dif(X,Y),
       call_residue((dif(X,Y),(X = f(Z) ; Y = f(Z))), L).

X = f(Z),
L = [[Y]-dif(f(Z),Y)],
dif(f(Z),Y) ? ;

Y = f(Z),
L = [[X]-dif(X,f(Z))],
dif(X,f(Z)) ? ;

no
@end example
The system only reports one invocation of @code{dif/2} as having
suspended. 

@end table


@node Attributed Variables, CLPQR, Coroutining, Extensions
@chapter Attributed Variables
@cindex attributed variables

@menu

* Attribute Declarations:: Declaring New Attributes
* Attribute Manipulation:: Setting and Reading Attributes
* Attributed Unification:: Tuning the Unification Algorithm
* Displaying Attributes:: Displaying Attributes in User-Readable Form
* Projecting Attributes:: Obtaining the Attributes of Interest
* Attribute Examples:: Two Simple Examples of how to use Attributes.
@end menu

YAP now supports the attributed variables packaged developed at OFAI by
Christian Holzbaur. Attributes are a means of declaring that an
arbitrary term is a property for a variable. These properties can be
updated during forward execution. Moreover, the unification algorithm is
aware of attributed variables and will call user defined handlers when
trying to unify these variables.

Attributed variables provide an elegant abstraction over which one can
extend Prolog systems. Their main application so far has been in
implementing constraint handlers, such as Holzbaur's CLPQR and Fruewirth
and Holzbaur's CHR, but other applications have been proposed in the
literature.


The command
@example
| ?- use_module(library(atts)).
@end example
enables the use of attributed variables. The package provides the
following functionality:
@itemize @bullet
@item Each attribute must be declared first. Attributes are described by a functor
and are declared per module. Each Prolog module declares its own sets of
attributes. Different modules may have different functors with the same
module.
@item The built-in @code{put_atts/2} adds or deletes attributes to a
variable. The variable may be unbound or may be an attributed
variable. In the latter case, YAP discards previous values for the
attributes.
@item The built-in @code{get_atts/2} can be used to check the values of
an attribute associated with a variable.
@item The unification algorit mcalls the user-defined predicate
@t{verify_attributes/3} before trying to bind an attributed
variable. Unification will resume after this call.
@item The user-defined predicate
@t{attribute_goal/2} converts from an attribute to a goal.
@item The user-defined predicate
@t{project_attributes/2} is used from a set of variables into a set of
constraints or goals. One application of @t{project_attributes/2} is in
the top-level, where it is used to output the set of
floundered constraints at the end of a query.
@end itemize

@node Attribute Declarations, Attribute Manipulation, , Attributed Variables
@section Attribute Declarations

Attributes are compound terms associated with a variable. Each attribute
has a @emph{name} which is @emph{private} to the module in which the
attribute was defined. Variables may have at most one attribute with a
name. Attribute names are defined with the following declaration:

@cindex attribute declaration
@cindex declaration, attribute
@findex attribute/1 (declaration)

@example
:- attribute @var{AttributeSpec}, ..., @var{AttributeSpec}.
@end example

@noindent
where each @var{AttributeSpec} has the form (@var{Name}/@var{Arity}).
One single such declaration is allowed per module @var{Module}.

Although the YAP module system is predicate based, attributes are local
to modules. This is implemented by rewriting all calls to the
builtins that manipulate attributes so that attribute names are
preprocessed depending on the module.  The @code{user:goal_expansion/3}
mechanism is used for this purpose.


@node Attribute Manipulation, Attributed Unification, Attribute Declarations, Attributed Variables
@section Attribute Manipulation


The  attribute manipulation predicates always work as follows:
@enumerate
@item The first argument is the unbound variable associated with
attributes,
@item The second argument is a list of attributes. Each attribute will
be a Prolog term or a constant, prefixed with the @t{+} and @t{-} unary
operators. The prefix @t{+} may be dropped for convenience.
@end enumerate

The following three procedures are available to the user. Notice that
these builtins are rewritten by the system into internal builtins, and
that the rewriting process @emph{depends} on the module on which the
builtins have been invoked.

@table @code
@item @var{Module}:get_atts(@var{-Var},@var{?ListOfAttributes})
@findex get_atts/2
@syindex get_atts/2
@cnindex get_atts/2
Unify the list @var{?ListOfAttributes} with the attributes for the unbound
variable @var{Var}. Each member of the list must be a bound term of the
form @code{+(@var{Attribute})}, @code{-(@var{Attribute})} (the @t{kbd}
prefix may be dropped). The meaning of @t{+} and @t{-} is:
@table @code
@item +(@var{Attribute})
Unifies @var{Attribute} with a corresponding attribute associated with
@var{Var}, fails otherwise.

@item -(@var{Attribute})
Succeeds if a corresponding attribute is not associated with
@var{Var}. The arguments of @var{Attribute} are ignored.
@end table

@item @var{Module}:put_atts(@var{-Var},@var{?ListOfAttributes})
@findex put_atts/2
@syindex put_atts/2
@cnindex put_atts/2
Associate with or remove attributes from a variable @var{Var}. The
attributes are given in @var{?ListOfAttributes}, and the action depends
on how they are prefixed:
@item +(@var{Attribute})
Associate @var{Var} with @var{Attribute}. A previous value for the
attribute is simply replace (like with @code{set_mutable/2}).

@item -(@var{Attribute})
Remove the attribute with the same name. If no such attribute existed,
simply succeed.
@end table

@node Attributed Unification, Displaying Attributes, Attribute Manipulation, Attributed Variables
@section Attributed Unification

The user-predicate predicate @code{verify_attributes/3} is called when
attempting to unify an attributed variable which might have attributes
in some @var{Module}.

@table @code
@item @var{Module}:verify_attributes(@var{-Var}, @var{+Value}, @var{-Goals})
@findex verify_attributes/3
@syindex verify_attributes/3
@cnindex verify_attributes/3

The predicate is called when trying to unify the attributed variable
@var{Var} with the Prolog term @var{Value}. Note that @var{Value} may be
itself an attributed variable, or may contain attributed variables.  The
goal @t{verify_attributes/3} is actually called before @var{Var} is
unified with @var{Value}.

It is up to the user to define which actions may be performed by
@t{verify_attributes/3} but the procedure is expected to return in
@var{Goals} a list of goals to be called @emph{after} @var{Var} is
unified with @var{Value}. If @t{verify_attributes/3} fails, the
unification will fail.

Notice that the @t{verify_attributes/3} may be called even if @var{Var}
has no attributes in module @t{Module}. In this case the routine should
simply succeed with @var{Goals} unified with the empty list.
@end table

@node Displaying Attributes, Projecting Attributes,Attributed Unification, Attributed Variables 
@section Displaying Attributes

Attributes are usually presented as goals. The following routines are
used by builtin predicates such as @code{call_residue/2} and by the
Prolog top-level to display attributes:

@table @code
@item @var{Module}:attribute_goal(@var{-Var}, @var{-Goal})
@findex attribute_goal/2
@syindex attribute_goal/2
@cnindex attribute_goal/2
User-defined procedure, called to convert the attributes in @var{Var} to
a @var{Goal}. Should fail when no interpretation is available.

@item @var{Module}:project_attributes(@var{-QueryVars}, @var{+AttrVars})
@findex project_attributes/2
@syindex project_attributes/2
@cnindex project_attributes/2
User-defined procedure, called to project the attributes in the query,
@var{AttrVars}, given that the set of variables in the query is
@var{QueryVars}.

@end table

@node Projecting Attributes, Attribute Examples, Displaying Attributes, Attributed Variables
@section Projecting Attributes

Constraint solvers must be able to project a set of constraints to a set
of variables. This is useful when displaying the solution to a goal, but
may also be used to manipulate computations. The user-defined
@code{project_attributes/2} is responsible for implementing this
projection.


@table @code
@item @var{Module}:project_attributes(@var{+QueryVars}, @var{+AttrVars})
@findex project_attributes/2
@syindex project_attributes/2
@cnindex project_attributes/2
Given a list of variables @var{QueryVars} and list of attributed
variables @var{AttrVars}, project all attributes in @var{AttrVars} to
@var{QueryVars}. Although projection is constraint system dependent,
typically this will involve expressing all constraints in terms of
@var{QueryVars} and considering all remaining variables as existentially
quantified.
@end table

Projection interacts with @code{attribute_goal/2} at the prolog top
level. When the query succeeds, the system first calls
@code{project_attributes/2}. The system then calls
@code{attribute_goal/2} to get a user-level representation of the
constraints. Typically, @code{attribute_goal/2} will convert from the
original constraints into a set of new constraints on the projection,
and these constraints are the ones that will have an
@code{attribute_goal/2} handler.

@node Attribute Examples, ,Projecting Attributes, Attributed Variables
@section Attribute Examples

The following two examples example is taken from the SICStus Prolog manual. It
sketches the implementation of a simple finite domain ``solver''.  Note
that an industrial strength solver would have to provide a wider range
of functionality and that it quite likely would utilize a more efficient
representation for the domains proper.  The module exports a single
predicate @code{domain(@var{-Var},@var{?Domain})} which associates
@var{Domain} (a list of terms) with @var{Var}.  A variable can be
queried for its domain by leaving @var{Domain} unbound.

We do not present here a definition for @code{project_attributes/2}.
Projecting finite domain constraints happens to be difficult.


@example
:- module(domain, [domain/2]).

:- use_module(library(atts)).
:- use_module(library(ordsets), [
        ord_intersection/3,
        ord_intersect/2,
        list_to_ord_set/2
   ]).

:- attribute dom/1.

verify_attributes(Var, Other, Goals) :-
        get_atts(Var, dom(Da)), !,          % are we involved?
        (   var(Other) ->                   % must be attributed then
            (   get_atts(Other, dom(Db)) -> %   has a domain?
                ord_intersection(Da, Db, Dc),
                Dc = [El|Els],              % at least one element
                (   Els = [] ->             % exactly one element
                    Goals = [Other=El]      % implied binding
                ;   Goals = [],
                    put_atts(Other, dom(Dc))% rescue intersection
                )
            ;   Goals = [],
                put_atts(Other, dom(Da))    % rescue the domain
            )
        ;   Goals = [],
            ord_intersect([Other], Da)      % value in domain?
        ).
verify_attributes(_, _, []).                % unification triggered
                                            % because of attributes
                                            % in other modules

attribute_goal(Var, domain(Var,Dom)) :-     % interpretation as goal
        get_atts(Var, dom(Dom)).

domain(X, Dom) :-
        var(Dom), !,
        get_atts(X, dom(Dom)).
domain(X, List) :-
        list_to_ord_set(List, Set),
        Set = [El|Els],                     % at least one element
        (   Els = [] ->                     % exactly one element
            X = El                          % implied binding
        ;   put_atts(Fresh, dom(Set)),
            X = Fresh                       % may call
                                            % verify_attributes/3
        ).
@end example

Note that the ``implied binding'' @code{Other=El} was deferred until after
the completion of @code{verify_attribute/3}.  Otherwise, there might be a
danger of recursively invoking @code{verify_attribute/3}, which might bind
@code{Var}, which is not allowed inside the scope of @code{verify_attribute/3}.
Deferring unifications into the third argument of @code{verify_attribute/3}
effectively serializes th calls to @code{verify_attribute/3}.

Assuming that the code resides in the file @file{domain.yap}, we
can use it via:

@example
| ?- use_module(domain).
@end example

Let's test it:

@example
| ?- domain(X,[5,6,7,1]), domain(Y,[3,4,5,6]), domain(Z,[1,6,7,8]).

domain(X,[1,5,6,7]),
domain(Y,[3,4,5,6]),
domain(Z,[1,6,7,8]) ? 

yes
| ?- domain(X,[5,6,7,1]), domain(Y,[3,4,5,6]), domain(Z,[1,6,7,8]), 
     X=Y.

Y = X,
domain(X,[5,6]),
domain(Z,[1,6,7,8]) ? 

yes
| ?- domain(X,[5,6,7,1]), domain(Y,[3,4,5,6]), domain(Z,[1,6,7,8]),
     X=Y, Y=Z.

X = 6,
Y = 6,
Z = 6
@end example

To demonstrate the use of the @var{Goals} argument of
@code{verify_attributes/3}, we give an implementation of
@code{freeze/2}.  We have to name it @code{myfreeze/2} in order to
avoid a name clash with the built-in predicate of the same name.

@example
:- module(myfreeze, [myfreeze/2]).

:- use_module(library(atts)).

:- attribute frozen/1.

verify_attributes(Var, Other, Goals) :-
        get_atts(Var, frozen(Fa)), !,       % are we involved?
        (   var(Other) ->                   % must be attributed then
            (   get_atts(Other, frozen(Fb)) % has a pending goal?
            ->  put_atts(Other, frozen((Fa,Fb))) % rescue conjunction
            ;   put_atts(Other, frozen(Fa)) % rescue the pending goal
            ),
            Goals = []
        ;   Goals = [Fa]
        ).
verify_attributes(_, _, []).

attribute_goal(Var, Goal) :-                % interpretation as goal
        get_atts(Var, frozen(Goal)).

myfreeze(X, Goal) :-
        put_atts(Fresh, frozen(Goal)),
        Fresh = X.
@end example

Assuming that this code lives in file @file{myfreeze.yap},
we would use it via:

@example
| ?- use_module(myfreeze).
| ?- myfreeze(X,print(bound(x,X))), X=2.

bound(x,2)                      % side effect
X = 2                           % bindings
@end example

The two solvers even work together:

@example
| ?- myfreeze(X,print(bound(x,X))), domain(X,[1,2,3]),
     domain(Y,[2,10]), X=Y.

bound(x,2)                      % side effect
X = 2,                          % bindings
Y = 2
@end example

The two example solvers interact via bindings to shared attributed
variables only.  More complicated interactions are likely to be found
in more sophisticated solvers.  The corresponding
@code{verify_attributes/3} predicates would typically refer to the
attributes from other known solvers/modules via the module prefix in
@code{@var{Module}:get_atts/2}.

@node CLPQR, CHR, Attributed Variables, Extensions
@chapter CLP(Q,R) Manual
@cindex CLPQ
@cindex CLPR

@menu
CLP(Q,R)
* Introduction to CLPQR:: The CLP(Q,R) System
* Referencing CLPQR:: How to Reference CLP(Q,R)
* CLPQR Acknowledgments:: Acknowledgments for CLP(Q,R)
* Solver Interface:: Using the CLP(Q,R) System
* Notational Conventions:: The CLP(Q,R) Notation
* Solver Predicates:: The CLP(Q,R) Interface Predicates
* Unification:: Unification and CLP(Q,R)
* Feedback and Bindings:: Information flow in CLP(Q,R)
* Linearity and Nonlinear Residues:: Linear and Nonlinear Constraints
* How Nonlinear Residues are made to disappear:: Handling Nonlinear Residues
* Isolation Axioms:: Isolating the Variable to be Solved
* Numerical Precision and Rationals:: Reals and Rationals
* Projection and Redundancy Elimination:: Presenting Bindings for Query Variables
* Variable Ordering:: Linear Relationships between Variables
* Turning Answers into Terms:: using @code{call_residue/2}
* Projecting Inequalities:: How to project linear inequations
* Why Disequations:: Using Disequations in CLP(Q,R)
* Syntactic Sugar:: An easier syntax
* Monash Examples:: The Monash Library
* Compatibility Notes:: CLP(Q,R) and the clp(R) interpreter
* A Mixed Integer Linear Optimization Example:: MIP models
* Implementation Architecture:: CLP(Q,R) Components
* Fragments and Bits:: Final Last Words on CLP(Q,R)
* CLPQR Bugs:: Bugs in CLP(Q,R)
* CLPQR References:: References for CLP(Q,R)
@end menu

This Manual documents a Prolog implementation of clp(Q,R), based on
SICStus featuring extensible unification via attributed variables.

Edition 1.3.3 December 1995

Christian Holzbaur @code{christian@@ai.univie.ac.at}

Copyright @copyright{} 1992,1993,1994,1995 OFAI Austrian Research
Institute for Artificial Intelligence (OFAI) Schottengasse 3 A-1010
Vienna, Austria

Permission is granted to make and distribute verbatim copies of this
manual provided the copyright notice and this permission notice are
preserved on all copies.

Permission is granted to copy and distribute modified versions of this
manual under the conditions for verbatim copying, provided that the
entire resulting derived work is distributed under the terms of a
permission notice identical to this one.

Permission is granted qto copy and distribute translations of this
manual into another language, under the above conditions for modified
versions, except that this permission notice may be stated in a
translation approved by the OFAI.

@node Introduction to CLPQR, Referencing CLPQR, , CLPQR
@section Introduction to CLP(Q,R)

The clp(Q,R) system described in this document is an instance of the
general Constraint Logic Programming scheme introduced by [Jaffar &
Michaylov 87].

The implementation is at least as complete as other existing clp(R)
implementations: It solves linear equations over rational or real
valued variables, covers the lazy treatment of nonlinear equations,
features a decision algorithm for linear inequalities that detects
implied equations, removes redundancies, performs projections
(quantifier elimination), allows for linear dis-equations, and
provides for linear optimization.

The full clp(Q,R) distribution, including a stand-alone manual and an
examples directory that is possibly more up to date than the version
in the SICStus Prolog distribution, is available from:
http://www.ai.univie.ac.at/clpqr/.

@node Referencing CLPQR, CLPQR Acknowledgments, Introduction to CLPQR, CLPQR
@section  Referencing CLP(Q,R)

When referring to this implementation of clp(Q,R) in publications, you
should use the following reference:

Holzbaur C.: OFAI clp(q,r) Manual, Edition 1.3.3, Austrian Research
Institute for Artificial Intelligence, Vienna, TR-95-09, 1995.

@node CLPQR Acknowledgments, Solver Interface, Referencing CLPQR, CLPQR
@section  CLP(QR) Acknowledgments
Acknowledgments

The development of this software was supported by the Austrian Fonds
zur Foerderung der Wissenschaftlichen Forschung under grant
P9426-PHY. Financial support for the Austrian Research Institute for
Artificial Intelligence is provided by the Austrian Federal Ministry
for Science and Research.

We include a collection of examples that has been distributed with the
Monash University version of clp(R) [Heintze et al. 87], and its
inclusion into this distribution was kindly permitted by Roland Yap.

@node Solver Interface, Notational Conventions, CLPQR Acknowledgments, CLPQR
@section Solver Interface

Rational numbers are not first class citizens in SICStus Prolog, so
rational arithmetics has to be emulated. Because of the emulation it
is too expensive to support arithmetics with automatic coercion
between all sorts of numbers, like you find it in CommonLisp, for
example.

You must choose whether you want to operate in the field of Q
(Rationals) or R (Reals):

@example
 ?- use_module(library(clpq)).
@end example

 or
@example
 ?- use_module(library(clpr)).
@end example

@node Notational Conventions, Solver Predicates, Solver Interface, CLPQR
@section Notational Conventions

Throughout this chapter, the prompts @code{clp(q) ?-} and @code{clp(r)
?-} are used to differentiate between clp(Q) and clp(R) in exemplary
interactions.

In general there are many ways to express the same linear
relationship. This degree of freedom is manifest in the fact that the
printed manual and an actual interaction with the current version of
clp(Q,R) may show syntactically different answer constraints, despite
the fact the same semantic relationship is being expressed. There are
means to control the presentation, see @pxref{Variable Ordering}. The
approximative nature of floating point numbers may also produce
numerical differences between the text in this manual and the actual
results of clp(R), for a given edition of the software.

@node Solver Predicates, Unification, Notational Conventions, CLPQR
@section Solver Predicates

The solver interface for both Q and R consists of the following
predicates which are exported from module(linear).

@table @code
@item @{+@var{Constraint}@}

@var{Constraint} is a term accepted by the the grammar below. The
corresponding constraint is added to the current constraint store and
checked for satisfiability. If you want to overload @{@}/1 with other
solvers, you can avoid its importation via: @code{use_module(clpq, [])}.

@example
@var{Constraint} --> @var{C}
            | @var{C} , @var{C}                 conjunction

@var{C} -->          @var{Expr} =:= @var{Expr}         equation
            | @var{Expr} = @var{Expr}           equation
            | @var{Expr} < @var{Expr}           strict inequation
            | @var{Expr} > @var{Expr}           strict inequation
            | @var{Expr} =< @var{Expr}          nonstrict inequation
            | @var{Expr} >= @var{Expr}          nonstrict inequation
            | @var{Expr} =\= @var{Expr}         disequation

@var{Expr} -->       variable             Prolog variable
            | number                floating point or integer
            | + @var{Expr}                unary plus
            | - @var{Expr}                unary minus
            | @var{Expr} + @var{Expr}           addition
            | @var{Expr} - @var{Expr}           subtraction
            | @var{Expr} * @var{Expr}           multiplication
            | @var{Expr} / @var{Expr}           division
            | abs(@var{Expr})             absolute value
            | sin(@var{Expr})             trigonometric sine
            | cos(@var{Expr})             trigonometric cosine
            | tan(@var{Expr})             trigonometric tangent
            | pow(@var{Expr},@var{Expr})        raise to the power
            | exp(@var{Expr},@var{Expr})        raise to the power
            | min(@var{Expr},@var{Expr})        minimum of the two
                                                arguments
            | max(@var{Expr},@var{Expr})        maximum of the two
                                                arguments
            | #(Const)              symbolic numerical
                                                constant
@end example

Conjunctive constraints @t{@{-C,C@}} have been made part of the syntax
in order to enable grouped submission of constraints, which could be
exploited by future versions of this software. Symbolic numerical
constants are provided for compatibility only, see @pxref{Monash Examples}.

@item entailed(+@var{Constraint})

Succeeds iff the linear @var{Constraint} is entailed by the current
constraint store. This predicate does not change the state of the
constraint store.

@example
clp(q) ?- @{A =< 4@}, entailed(A=\=5).

@{A=<4@}
yes

clp(q) ?- @{A =< 4@}, entailed(A=\=3).
no
@end example

@item inf(+@var{Expr}, -@var{Inf} )
Computes the infimum of the linear expression @var{Expr} and unifies it
with @var{Inf}. Failure indicates unboundedness.

@item sup(+@var{Expr}, -@var{Sup})

Computes the supremum of the linear expression @var{Expr} and unifies it
with @var{Sup}. Failure indicates unboundedness.

@example
clp(q) ?- @{ 2*X+Y =< 16, X+2*Y =< 11,
             X+3*Y =< 15, Z = 30*X+50*Y
          @}, sup(Z, Sup).

Sup = 310,
@{Z=30*X+50*Y@},
@{X+1/2*Y=<8@}

@{X+3*Y=<15@},
@{X+2*Y=<11@}
@end example

@item minimize(+@var{Expr})

Computes the infimum of the linear expression @var{Expr} and equates it
with the expression, i.e. as if defined as:

@example
minimize(Expr) :- inf(Expr, Expr).
@end example

@item maximize(+@var{Expr})

Computes the supremum of the linear expression @var{Expr} and equates it
with the expression.

@example
clp(q) ?- @{ 2*X+Y =< 16, X+2*Y =< 11,
             X+3*Y =< 15, Z = 30*X+50*Y
          @}, maximize(Z).

X = 7,
Y = 2,
Z = 310
@end example

@item bb_inf(+@var{Ints}, +@var{Expr}, -@var{Inf})

Computes the infimum of the linear expression @var{Expr} under the
additional constraint that all of variables in the list @var{Ints}
assume integral values at the infimum. This allows for the solution of
mixed integer linear optimization problems, see @pxref{A Mixed Integer
Linear Optimization Example}.

@item ordering(+@var{Spec})

Provides a means to control one aspect of the presentation of the answer
constraints, see @pxref{Variable Ordering}.
@end table

@node Unification, Feedback and Bindings, Solver Predicates, CLPQR
@section Unification

Equality constraints are added to the store implicitly each time
variables that have been mentioned in explicit constraints are bound -
either to another such variable or to a number.

@example
clp(r) ?- @{2*A+3*B=C/2@}, C=10.0, A=B.

A = 1.0,
B = 1.0,
C = 10.0
@end example

Is equivalent modulo rounding errors to

@example
clp(r) ?- @{2*A+3*B=C/2, C=10, A=B@}.

A = 1.0,
B = 0.9999999999999999,
C = 10.0
@end example

The shortcut bypassing the use of @{@}q/1 is allowed and makes sense
because the interpretation of this equality in Prolog and clp(R)
coincides. In general, equations involving interpreted functors,
@code{+/2} in this case, must be fed to the solver explicitly:

@example
clp(r) ?- X=3.0+1.0, X=4.0.

no
@end example

Further, variables known by clp(R) may be bound directly to floats
only. Likewise, variables known by clp(Q) may be bound directly to
rational numbers only, see @pxref{Numerical Precision and
Rationals}. Failing to do so is rewarded with an exception:

@example
clp(q) ?- @{2*A+3*B=C/2@}, C=10.0, A=B.

[ERROR: not.normalized(10.0)]
@end example

This is because 10.0 is not a rational constant. To make clp(Q) happy
you have to say:

@example
clp(q) ?- @{2*A+3*B=C/2@}, C=rat(10,1), A=B.

A = 1,
B = 1,
C = 10
@end example

If you use @code{@{@}/1}, you don't have to worry about such
details. Alternatively, you may use the automatic expansion facility,
check @pxref{Syntactic Sugar}.

@node Feedback and Bindings, Linearity and Nonlinear Residues,Unification, CLPQR
@section Feedback and Bindings

What was covered so far was how the user populates the constraint
store. The other direction of the information flow consists of the
success and failure of the above predicates and the binding of variables
to numerical values and the aliasing of variables. Example:

@example
clp(r) ?- @{A-B+C=10, C=5+5@}.

B = A,
C = 10.0
@end example

The linear constraints imply @code{A=B} and the solver consequently
exports this binding to the Prolog world, which is manifest in the fact
that the test @code{A==B} will succeed. More about answer presentation
in @pxref{Projection and Redundancy Elimination}.

@node Linearity and Nonlinear Residues, How Nonlinear Residues are made to disappear,Feedback and Bindings, CLPQR
@section Linearity and Nonlinear Residues

The clp(Q,R) system is restricted to deal with linear constraints
because the decision algorithms for general nonlinear constraints are
prohibitively expensive to run. If you need this functionality badly,
you should look into symbolic algebra packages. Although the clp(Q,R)
system cannot solve nonlinear constraints, it will collect them
faithfully in the hope that through the addition of further (linear)
constraints they might get simple enough to solve eventually. If an
answer contains constraints, you have to be aware of the fact that
success is qualified modulo the existence of a solution to the system of
residual (nonlinear) constraints:

@example
clp(r) ?- @{sin(X) = cos(X)@}.

nonlin:@{sin(X)-cos(X)=0.0@}
@end example

There are indeed infinitely many solutions to this constraint (@code{X =
0.785398 + n*Pi}), but clp(Q,R) has no direct means to find and
represent them.

The systems goes through some lengths to recognize linear expressions as
such. The method is based on a normal form for multivariate
polynomials. In addition, some simple isolation axioms, that can be used
in equality constraints, have been added. The current major limitation
of the method is that full polynomial division has not been implemented.

This is an example where the isolation axioms are sufficient to
determine the value of X.

@example
clp(r) ?- @{sin(cos(X)) = 1/2@}.

X = 1.0197267436954502
@end example

If we change the equation into an inequation, clp(Q,R) gives up:

@example
clp(r) ?- @{sin(cos(X)) < 1/2@}.

nonlin:@{sin(cos(X))-0.5!0.0@}
@end example

The following is easy again:

@example
clp(r) ?- @{sin(X+2+2)/sin(4+X) = Y@}.

Y = 1.0
@end example

And so is this:

@example
clp(r) ?- @{(X+Y)*(Y+X)/X = Y*Y/X+99@}.

@{Y=49.5-0.5*X@}
@end example

An ancient symbol manipulation benchmark consists in rising the
expression @code{X+Y+Z+1} to the 15th power:

@example
clp(q) ?- @{exp(X+Y+Z+1,15)=0@}.

nonlin:@{Z^15+Z^14*15+Z^13*105+Z^12*455+Z^11*1365+Z^10*3003+...
         ... polynomial continues for a few pages ...
         =0@}
@end example

Computing its roots is another story. 

@node How Nonlinear Residues are made to disappear, Isolation Axioms,Linearity and Nonlinear Residues, CLPQR
@section How Nonlinear Residues are made to disappear

Binding variables that appear in nonlinear residues will reduce the
complexity of the nonlinear expressions and eventually results in linear
expressions:

@example
clp(q) ?- @{exp(X+Y+1,2) = 3*X*X+Y*Y@}.

nonlin:@{Y*2-X^2*2+Y*X*2+X*2+1=0@}
@end example

Equating X and Y collapses the expression completely and even determines
the values of the two variables:

@example
clp(q) ?- @{exp(X+Y+1,2) = 3*X*X+Y*Y@}, X=Y.

X = -1/4,
Y = -1/4
@end example

@node Isolation Axioms, Numerical Precision and Rationals,How Nonlinear Residues are made to disappear, CLPQR
@section Isolation Axioms

These axioms are used to rewrite equations such that the variable to be
solved for is moved to the left hand side and the result of the
evaluation of the right hand side can be assigned to the variable. This
allows, for example, to use the exponentiation operator for the
computation of roots and logarithms, see below.

@table @code
@item @var{A} = @var{B} * @var{C}
 Residuates unless @var{B} or @var{C} is ground or @var{A} and @var{B}
or @var{C} are ground.

@item @var{A} = @var{B} / @var{C}
Residuates unless @var{C} is ground or @var{A} and @var{B} are ground. 

@item @var{X} = min(@var{Y},@var{Z})
Residuates unless @var{Y} and @var{Z} are ground.

@item @var{X} = max(@var{Y},@var{Z})
Residuates unless @var{Y} and @var{Z} are ground.

@item @var{X} = abs(@var{Y})
Residuates unless @var{Y} is ground.

@item @var{X} = pow(@var{Y},@var{Z}), @var{X} = exp(@var{Y},@var{Z})
Residuates unless any pair of two of the three variables is ground. Example:

@example
clp(r) ?- @{ 12=pow(2,X) @}.

X = 3.5849625007211565

clp(r) ?- @{ 12=pow(X,3.585) @}.

X = 1.9999854993443926

clp(r) ?- @{ X=pow(2,3.585) @}.

X = 12.000311914286545
@end example

@item  @var{X} = sin(@var{Y})
Residuates unless @var{X} or @var{Y} is ground. Example:

@example
clp(r) ?- @{ 1/2 = sin(X) @}.

X = 0.5235987755982989
@end example

@item @var{X} = cos(@var{Y})
Residuates unless @var{X} or @var{Y} is ground.

@item @var{X} = tan(@var{Y})
Residuates unless @var{X} or @var{Y} is ground.
@end table

@node Numerical Precision and Rationals, Projection and Redundancy Elimination,Isolation Axioms, CLPQR
@section Numerical Precision and Rationals

The fact that you can switch between clp(R) and clp(Q) should solve most
of your numerical problems regarding precision. Within clp(Q), floating
point constants will be coerced into rational numbers
automatically. Transcendental functions will be approximated with
rationals. The precision of the approximation is limited by the floating
point precision. These two provisions allow you to switch between clp(R)
and clp(Q) without having to change your programs.

What is to be kept in mind however is the fact that it may take quite
big rationals to accommodate the required precision. High levels of
precision are for example required if your linear program is
ill-conditioned, i.e., in a full rank system the determinant of the
coefficient matrix is close to zero. Another situation that may call for
elevated levels of precision is when a linear optimization problem
requires exceedingly many pivot steps before the optimum is reached.

If your application approximates irrational numbers, you may be out of
space particularly soon. The following program implements N steps of
Newton's approximation for the square root function at point 2.

@example
%
% from file: library('clpqr/examples/root')
%

root(N, R) :-
  root(N, 1, R).

root(0, S, R) :- !, S=R.
root(N, S, R) :-
  N1 is N-1,
  @{ S1 = S/2 + 1/S @},
  root(N1, S1, R).
@end example

It is known that this approximation converges quadratically, which means
that the number of correct digits in the decimal expansion roughly
doubles with each iteration. Therefore the numerator and denominator of
the rational approximation have to grow likewise:

@example
clp(q) ?- use+odule(library('clpqr/examples/root')).
clp(q) ?- root(3,R),print_decimal(R,70).
1.4142156862 7450980392 1568627450 9803921568 6274509803 9215686274
5098039215

R = 577/408

clp(q) ?- root(4,R),print_decimal(R,70).
1.4142135623 7468991062 6295578890 1349101165 5962211574 4044584905
0192000543

R = 665857/470832

clp(q) ?- root(5,R),print_decimal(R,70).

1.4142135623 7309504880 1689623502 5302436149 8192577619 7428498289
4986231958

R = 886731088897/627013566048

clp(q) ?- root(6,R),print_decimal(R,70).

1.4142135623 7309504880 1688724209 6980785696 7187537723 4001561013
1331132652

R = 1572584048032918633353217/1111984844349868137938112

clp(q) ?- root(7,R),print_decimal(R,70).

1.4142135623 7309504880 1688724209 6980785696 7187537694 8073176679
7379907324

R = 4946041176255201878775086487573351061418968498177 /
    3497379255757941172020851852070562919437964212608
@end example

Iterating for 8 steps produces no further change in the first 70 decimal
digits of @code{sqrt(2)}. After 15 steps the approximating rational
number has a numerator and a denominator with 12543 digits each, and the
next step runs out of memory.

Another irrational number that is easily computed is @code{e}. The
following program implements an alternating series for @code{1/e}, where
the absolute value of last term is an upper bound on the error.

@example
%
% from file: library('clpqr/examples/root')
%

e(N, E) :-
  @{ Err =:= exp(10,-(N+2)), Half =:= 1/2 @},
  inv_e_series(Half, Half, 3, Err, Inv.E),
  @{ E =:= 1/Inv_E @}.

inv_e_series(Term, S0, ., Err, Sum) :-
  @{ abs(Term) =< Err @}, !,
  S0 = Sum.

inv_e_series(Term, S0, N, Err, Sum) :-
  N1 is N+1,
  @{ Term1 =:= -Term/N, S1 =:= Term1+S0 @},
  inv_e_series(Term1, S1, N1, Err, Sum).
@end example

The computation of the rational number @var{E} that approximates
@code{e} up to at least 1000 digits in its decimal expansion requires
the evaluation of 450 terms of the series, i.e. 450 calls of
inv.e. series/5.

@example
clp(q) ?- e(1000,E).

E = 7149056228932760213666809592072842334290744221392610955845565494
    3708750229467761730471738895197792271346693089326102132000338192
    0131874187833985420922688804220167840319199699494193852403223700
    5853832741544191628747052136402176941963825543565900589161585723
    4023097417605004829991929283045372355639145644588174733401360176
    9953973706537274133283614740902771561159913069917833820285608440
    3104966899999651928637634656418969027076699082888742481392304807
    9484725489080844360397606199771786024695620205344042765860581379
    3538290451208322129898069978107971226873160872046731879753034549
    3130492167474809196348846916421782850086985668680640425192038155
    4902863298351349469211627292865440876581064873866786120098602898
    8799130098877372097360065934827751120659213470528793143805903554
    7928682131082164366007016698761961066948371407368962539467994627
    1374858249110795976398595034606994740186040425117101588480000000
    0000000000000000000000000000000000000000000000000000000000000000
    00000000000000000000000000000000000000
    /
    2629990810403002651095959155503002285441272170673105334466808931
    6863103901346024240326549035084528682487048064823380723787110941
    6809235187356318780972302796570251102928552003708556939314795678
    1978390674393498540663747334079841518303636625888963910391440709
    0887345797303470959207883316838346973393937778363411195624313553
    8835644822353659840936818391050630360633734935381528275392050975
    7271468992840907541350345459011192466892177866882264242860412188
    0652112744642450404625763019639086944558899249788084559753723892
    1643188991444945360726899532023542969572584363761073528841147012
    2634218045463494055807073778490814692996517359952229262198396182
    1838930043528583109973872348193806830382584040536394640895148751
    0766256738740729894909630785260101721285704616818889741995949666
    6303289703199393801976334974240815397920213059799071915067856758
    6716458821062645562512745336709063396510021681900076680696945309
    3660590933279867736747926648678738515702777431353845466199680991
    73361873421152165477774911660108200059
@end example

The decimal expansion itself looks like this:

@example
clp(q) ?- e(1000, E), print_decimal(E, 1000).
2.
7182818284 5904523536 0287471352 6624977572 4709369995 9574966967
6277240766 3035354759 4571382178 5251664274 2746639193 2003059921
8174135966 2904357290 0334295260 5956307381 3232862794 3490763233
8298807531 9525101901 1573834187 9307021540 8914993488 4167509244
7614606680 8226480016 8477411853 7423454424 3710753907 7744992069
5517027618 3860626133 1384583000 7520449338 2656029760 6737113200
7093287091 2744374704 7230696977 2093101416 9283681902 5515108657
4637721112 5238978442 5056953696 7707854499 6996794686 4454905987
9316368892 3009879312 7736178215 4249992295 7635148220 8269895193
6680331825 2886939849 6465105820 9392398294 8879332036 2509443117
3012381970 6841614039 7019837679 3206832823 7646480429 5311802328
7825098194 5581530175 6717361332 0698112509 9618188159 3041690351
5988885193 4580727386 6738589422 8792284998 9208680582 5749279610
4841984443 6346324496 8487560233 6248270419 7862320900 2160990235
3043699418 4914631409 3431738143 6405462531 5209618369 0888707016
7683964243 7814059271 4563549061 3031072085 1038375051 0115747704
1718986106 8739696552 1267154688 9570350354
@end example

@node Projection and Redundancy Elimination, Variable Ordering,Numerical Precision and Rationals, CLPQR
@section Projection and Redundancy Elimination

Once a derivation succeeds, the Prolog system presents the bindings for
the variables in the query. In a CLP system, the set of answer
constraints is presented in analogy. A complication in the CLP context
are variables and associated constraints that were not mentioned in the
query. A motivating example is the familiar mortgage relation:

@example
%
% from file: library('clpqr/examples/mg')
%

mg(P,T,I,B,MP):-
  @{ 
    T = 1,
    B + MP = P * (1 + I)
  @}.
mg(P,T,I,B,MP):-
  @{
    T > 1,
    P1 = P * (1 + I) - MP,
    T1 = T - 1
  @}, mg(P1, T1, I, B, MP).
@end example

A sample query yields:

@example
clp(r) ?- use_module(library('clpqr/examples/mg')).

clp(r) ?- mg(P,12,0.01,B,Mp).

@{B=1.1268250301319698*P-12.682503013196973*Mp@}
@end example

Without projection of the answer constraints onto the query variables we
would observe the following interaction:

@example
clp(r) ?- mg(P,12,0.01,B,Mp).

@{B=12.682503013196973*_A-11.682503013196971*P@},
@{Mp= -(_A)+1.01*P@},
@{_B=2.01*_A-1.01*P@}
@{_C=3.0301*_A-2.0301*P@},
@{_D=4.060401000000001*_A-3.0604009999999997*P@},
@{_E=5.101005010000001*_A-4.10100501*P@},
@{_F=6.152015060100001*_A-5.152015060099999*P@},
@{_G=7.213535210701001*_A-6.213535210700999*P@},
@{_H=8.285670562808011*_A-7.285670562808009*P@},
@{_I=9.368527268436091*_A-8.36852726843609*P@},
@{_J=10.462212541120453*_A-9.46221254112045*P@},
@{_K=11.566834666531657*_A-10.566834666531655*P@}
@end example

The variables @var{_A ... _K} are not part of the query, they originate
from the mortgage program proper. Although the latter answer is
equivalent to the former in terms of linear algebra, most users would
prefer the former.

@node Variable Ordering, Turning Answers into Terms,Projection and Redundancy Elimination, CLPQR
@section Variable Ordering

In general, there are many ways to express the same linear relationship
between variables. clp(Q,R) does not care to distinguish between them,
but the user might. The predicate ordering(+@var{Spec}) gives you some
control over the variable ordering. Suppose that instead of @var{B}, you
want @var{Mp} to be the defined variable:

@example
clp(r) ?- mg(P,12,0.01,B,Mp).

@{B=1.1268250301319698*P-12.682503013196973*Mp@}
@end example

This is achieved with:

@example
clp(r) ?- mg(P,12,0.01,B,Mp), ordering([Mp]).

@{Mp= -0.0788487886783417*B+0.08884878867834171*P@}
@end example

One could go one step further and require @var{P} to appear before (to
the left of) @var{B} in a addition:

@example
clp(r) ?- mg(P,12,0.01,B,Mp), ordering([Mp,P]).

@{Mp=0.08884878867834171*P-0.0788487886783417*B@}
@end example

@var{Spec} in ordering(+@var{Spec}) is either a list of variables with
the intended ordering, or of the form @code{@var{A}<@var{B}}. The latter
form means that @var{A} goes to the left of @var{B}. In fact,
@code{ordering([@var{A},@var{B},@var{C},@var{D}])} is shorthand for:

@example
ordering(A < B), ordering(A < C), ordering(A < D),
ordering(B < C), ordering(B < D),
ordering(C < D)
@end example

The ordering specification only affects the final presentation of the
constraints. For all other operations of clp(Q,R), the ordering is
immaterial. Note that ordering/1 acts like a constraint: you can put it
anywhere in the computation, and you can submit multiple specifications.

@example
clp(r) ?- ordering(B < Mp), mg(P,12,0.01,B,Mp).

@{B= -12.682503013196973*Mp+1.1268250301319698*P@}

yes
clp(r) ?- ordering(B < Mp), mg(P,12,0.01,B,Mp), ordering(P < Mp).

@{P=0.8874492252651537*B+11.255077473484631*Mp@}
@end example

@node Turning Answers into Terms, Projecting Inequalities,Variable Ordering, CLPQR
@section Turning Answers into Terms

In meta-programming applications one needs to get a grip on the results
computed by the clp(Q,R) solver. The SISCtus Prolog predicate
@code{call_residue/2} provides this functionality:

@example
clp(r) ?- call_residue(@{2*A+B+C=10,C-D=E,A<10@}, Constraints).

Constraints = [
               [A]-@{A<10.0@},
               [B]-@{B=10.0-2.0*A-C@},
               [D]-@{D=C-E@}
              ]
@end example

@node Projecting Inequalities, Why Disequations,Turning Answers into Terms, CLPQR
@section  Projecting Inequalities

As soon as linear inequations are involved, projection gets more
demanding complexity wise. The current clp(Q,R) version uses a
Fourier-Motzkin algorithm for the projection of linear inequalities. The
choice of a suitable algorithm is somewhat dependent on the number of
variables to be eliminated, the total number of variables, and other
factors. It is quite easy to produce problems of moderate size where the
elimination step takes some time. For example, when the dimension of the
projection is 1, you might be better off computing the supremum and the
infimum of the remaining variable instead of eliminating n-1 variables
via implicit projection.

In order to make answers as concise as possible, redundant constraints
are removed by the system as well. In the following set of inequalities,
half of them are redundant.

@example
%
% from file: library('clpqr/examples/elimination')
%

example(2, [X0,X1,X2,X3,X4]) :-
  @{
       +87*X0 +52*X1 +27*X2 -54*X3 +56*X4 =< -93,
       +33*X0 -10*X1 +61*X2 -28*X3 -29*X4 =<  63,
       -68*X0  +8*X1 +35*X2 +68*X3 +35*X4 =< -85,
       +90*X0 +60*X1 -76*X2 -53*X3 +24*X4 =< -68,
       -95*X0 -10*X1 +64*X2 +76*X3 -24*X4 =<  33,
       +43*X0 -22*X1 +67*X2 -68*X3 -92*X4 =< -97,
       +39*X0  +7*X1 +62*X2 +54*X3 -26*X4 =< -27,
       +48*X0 -13*X1  +7*X2 -61*X3 -59*X4 =<  -2,
       +49*X0 -23*X1 -31*X2 -76*X3 +27*X4 =<   3,
       -50*X0 +58*X1  -1*X2 +57*X3 +20*X4 =<   6,
       -13*X0 -63*X1 +81*X2  -3*X3 +70*X4 =<  64,
       +20*X0 +67*X1 -23*X2 -41*X3 -66*X4 =<  52,
       -81*X0 -44*X1 +19*X2 -22*X3 -73*X4 =< -17,
       -43*X0  -9*X1 +14*X2 +27*X3 +40*X4 =<  39,
       +16*X0 +83*X1 +89*X2 +25*X3 +55*X4 =<  36,
       +2*X0  +40*X1 +65*X2 +59*X3 -32*X4 =<  13,
       -65*X0 -11*X1 +10*X2 -13*X3 +91*X4 =<  49,
       +93*X0 -73*X1 +91*X2  -1*X3 +23*X4 =< -87
  @}.
@end example

Consequently, the answer consists of the system of nine non-redundant
inequalities only:

@example
clp(q) ?- use_module(library('clpqr/examples/elimination')).
clp(q) ?- example(2, [X0,X1,X2,X3,X4]).

@{X0-2/17*X1-35/68*X2-X3-35/68*X4?=5/4@},
@{X0-73/93*X1+91/93*X2-1/93*X3+23/93*X4=<-29/31@},
@{X0-29/25*X1+1/50*X2-57/50*X3-2/5*X4>=-3/25@},
@{X0+7/39*X1+62/39*X2+18/13*X3-2/3*X4=<-9/13@},
@{X0+2/19*X1-64/95*X2-4/5*X3+24/95*X4>=-33/95@},
@{X0+2/3*X1-38/45*X2-53/90*X3+4/15*X4=<-34/45@},
@{X0-23/49*X1-31/49*X2-76/49*X3+27/49*X4=<3/49@},
@{X0+44/81*X1-19/81*X2+22/81*X3+73/81*X4>=17/81@},
@{X0+9/43*X1-14/43*X2-27/43*X3-40/43*X4>=-39/43@}
@end example

The projection (the shadow) of this polyhedral set into the
@var{X0},@var{X1} space can be computed via the implicit elimination of
non-query variables:

@example
clp(q) ?- example(2, [X0,X1--.]).

@{X0+2619277/17854273*X1>=-851123/17854273@},
@{X0+6429953/16575801*X1=<-12749681/16575801@},
@{X0+19130/1213083*X1>=795400/404361@},
@{X0-1251619/3956679*X1?=21101146/3956679@},
@{X0+601502/4257189*X1>=220850/473021@}
@end example

Projection is quite a powerful concept that leads to surprisingly terse
executable specifications of nontrivial problems like the computation of
the convex hull from a set of points in an n-dimensional space: Given
the program

@example
%
% from file: library('clpqr/examples/elimination')
%
conv.hull(Points, Xs) :-
  lin_comb(Points, Lambdas, Zero, Xs),
  zero(Zero),
  polytope(Lambdas).

polytope(Xs) :-
  positive_sum(Xs, 1).

  positive_sum([], Z) :- @{Z=0@}.
  positive_sum([X--Xs], SumX) :-
    @{X >= 0, SumX = X+Sum @},
    positive_sum(Xs, Sum).

zero([]).
zero([Z--Zs]) :- @{Z=0@}, zero(Zs).

lin_comb([], [], S1, S1).
lin_comb([Ps--Rest], [K--Ks], S1, S3) :-
  lin_comb_r(Ps, K, S1, S2),
  lin_comb(Rest, Ks, S2, S3).

  lin_comb_r([], ., [], []).
  lin_comb_r([P--Ps], K, [S--Ss], [Kps--Ss1]) :-
    @{ Kps = K*P+S @},
    lin_comb_r(Ps, K, Ss, Ss1).
@end example

we can post the following query:

@example
clp(q) ?- conv.hull([ [1,1], [2,0], [3,0], [1,2], [2,2] ], [X,Y]).

@{Y=<2@},
@{X+1/2*Y=<3@},
@{X>=1@},
@{Y>=0@},
@{X+Y>=2@}
@end example

This answer is easily verified graphically:

@example
 |
2-   *   *
 |
 |
1|   *
 |
 |
0 ---|---*---*----
     1   2   3
@end example

The convex hull program directly corresponds to the mathematical
definition of the convex hull. What does the trick in operational terms
is the implicit elimination of the Lambdas from the program
formulation. Please note that this program does not limit the number of
points or the dimension of the space they are from. Please note further
that quantifier elimination is a computationally expensive operation and
therefore this program is only useful as a benchmark for the projector
and not so for the intended purpose.

@node Why Disequations, Syntactic Sugar,Projecting Inequalities, CLPQR
@section Why Disequations

A beautiful example of disequations at work is due to [Colmerauer
90]. It addresses the task of tiling a rectangle with squares of
all-different, a priori unknown sizes. Here is a translation of the
original Prolog-III program to clp(Q,R):

@example 
%
% from file: library('clpqr/examples/squares')
filled_rectangle( A, C) :-
  @{ A >= 1 @},
  distinct_squares( C),
  filled_zone( [-1,A,1], _, C, []).

distinct_squares( []).
distinct_squares( [B|C]) :-
  @{ B > 0 @},
  outof( C, B),
  distinct_squares( C).

outof( [],     _).
outof( [B1|C], B) :-
  @{ B =\= B1 @},       % *** note disequation ***
  outof( C, B).

filled_zone( [V|L], [V|L], C0, C0) :-
  @{ V >= 0 @}.
filled_zone( [V|L], L3, [B|C], C2) :-
  @{ V < 0 @},
  placed_square( B, L, L1),
  filled_zone( L1, L2, C, C1),
  @{ Vb=V+B @},
  filled_zone( [Vb,B|L2], L3, C1, C2).

placed_square( B, [H,H0,H1|L], L1) :-
  @{ B > H, H0=0, H2=H+H1 @},
  placed_square( B, [H2|L], L1).
placed_square( B, [B,V|L], [X|L]) :-
  @{ X=V-B @}.
placed_square( B, [H|L], [X,Y|L]) :-
  @{ B < H, X= -B, Y=H-B @}.
@end example

There are no tilings with less than nine squares except the trivial one
where the rectangle equals the only square. There are eight solutions
for nine squares. Six further solutions are rotations of the first two.

@example
clp(q) ?- use_module(library('clpqr/examples/squares')).
clp(q) ?- filled_rectangle(A, Squares).

A = 1,f
Squares = [1] ? ;

A = 33/32,
Squares = [15/32,9/16,1/4,7/32,1/8,7/16,1/32,5/16,9/32] ? ;

A = 69/61,
Squares = [33/61,36/61,28/61,5/61,2/61,9/61,25/61,7/61,16/61]
@end example

Depending on your hardware, the above query may take a few
minutes. Supplying the knowledge about the minimal number of squares
beforehand cuts the computation time by a factor of roughly four:

@example
clp(q) ?- length(Squares, 9), filled.rectangle(A, Squares).

A = 33/32,
Squares = [15/32,9/16,1/4,7/32,1/8,7/16,1/32,5/16,9/32] ? ;

A = 69/61,
Squares = [33/61,36/61,28/61,5/61,2/61,9/61,25/61,7/61,16/61]
@end example

@node Syntactic Sugar, Monash Examples,Why Disequations, CLPQR
@section Syntactic Sugar

There is a package that transforms programs and queries from a
eval-quote variant of clp(Q,R) into corresponding programs and queries
in a quote-eval variant. Before you use it, you need to know that in an
eval-quote language, all symbols are interpreted unless explicitly
quoted. This means that interpreted terms cannot be manipulated
syntactically directly. Meta-programming in a CLP context by definition
manipulates interpreted terms, therefore you need @code{quote/1} (just
as in LISP) and some means to put syntactical terms back to their
interpreted life: @code{@{@}/1}.

In a quote-eval language, meta-programming is (pragmatically) simpler
because everything is implicitly quoted until explicitly evaluated. On
the other hand, now object programming suffers from the dual
inconvenience.

We chose to make our version of clp(Q,R) of the quote-eval type because
this matches the intended use of the already existing boolean solver of
SICStus. In order to keep the users of the eval-quote variant happy, we
provide a source transformation package. It is activated via:

@example
| ?- use_module(library('clpqr/expand')). 
@end example

Loading the package puts you in a mode where the arithmetic functors
like @code{+/2}, @code{*/2} and all numbers (functors of arity 0) are
interpreted semantically.

@example
clp(r) ?- 2+2=X. X = 4.0
@end example

The package works by purifying programs and queries in the sense that
all references to interpreted terms are made explicit. The above query
is expanded prior to evaluation into:

@example
linear:@{2.0+2.0=X@}
@end example

The same mechanism applies when interpreted terms are nested deeper:

@example
some_predicate(10, f(A+B/2), 2*cos(A))
@end example

Expands into:

@example
linear:@{Xc=2.0*cos(A)@},
linear:@{Xb=A+B/2@},
linear:@{Xa=10.0@},
some_predicate(Xa, f(Xb), Xc)
@end example

This process also applies when files are consulted or compiled. In fact,
this is the only situation where expansion can be applied with relative
safety. To see this, consider what happens when the toplevel evaluates
the expansion, namely some calls to the clp(Q,R) solver, followed by the
call of the purified query. As we learned in @pxref{Feedback and
Bindings}, the solver may bind variables, which produces a goal with
interpreted functors in it (numbers), which leads to another stage of
expansion, and so on.

We recommend that you only turn on expansion temporarily while
consulting or compiling files needing expansion with @code{expand/0} and
@code{noexpand/0}.

@node Monash Examples,Compatibility Notes ,Syntactic Sugar, CLPQR
@section Monash Examples

This collection of examples has been distributed with the Monash
University Version of clp(R) [Heintze et al. 87], and its inclusion into
this distribution was kindly permitted by Roland Yap.

In order to execute the examples, a small compatibility package has to
be loaded first:

@example
clp(r) ?- use_module(library('clpqr/monash')).
@end example

Then, assuming you are using clp(R):

@example
clp(r) ?- expand, [library('clpqr/examples/monash/rkf45')],
          noexpand.

clp(r) ?- go.
Point    0.00000 :    0.75000    0.00000 
Point    0.50000 :    0.61969    0.47793 
Point    1.00000 :    0.29417    0.81233 
Point    1.50000 :   -0.10556    0.95809 
Point    2.00000 :   -0.49076    0.93977 
Point    2.50000 :   -0.81440    0.79929 
Point    3.00000 :   -1.05440    0.57522

Iteration finished
------------------
 439 derivative evaluations
@end example

@node Compatibility Notes, A Mixed Integer Linear Optimization Example,Monash Examples, CLPQR
@section Compatibility Notes

The Monash examples have been written for clp(R). Nevertheless, all but
rkf45 complete nicely in clp(Q). With @code{rkf45}, clp(Q) runs out of
memory. This is an instance of the problem discussed in @pxref{Numerical
Precision and Rationals}.

The Monash University clp(R) interpreter features a @code{dump/n}
predicate. It is used to print the target variables according to the
given ordering. Within this version of clp(Q,R), the corresponding
functionality is provided via @code{ordering/1}. The difference is that
@code{ordering/1} does only specify the ordering of the variables and no
printing is performed. We think Prolog has enough predicates to perform
output already. You can still run the examples referring to
@code{dump/n} from the Prolog toplevel:

@example
clp(r) ?- expand, [library('clpqr/examples/monash/mortgage')], noexpand.

% go2
%
clp(r) ?- mg(P,120,0.01,0,MP), dump([P,MP]).

@{P=69.7005220313972*MP@}

% go3
% 
clp(r) ?- mg(P,120,0.01,B,MP), dump([P,B,MP]).

@{P=0.30299477968602706*B+69.7005220313972*MP@}

% go4
%
clp(r) ?- mg(999, 3, Int, 0, 400), dump.

nonlin:@{_B-_B*Int+.A+400.0=0.0@},
nonlin:@{_A-_A*Int+400.0=0.0@},
@{_B=599.0+999.0*Int@}
@end example

@node A Mixed Integer Linear Optimization Example, Implementation Architecture,Compatibility Notes, CLPQR
@section A Mixed Integer Linear Optimization Example

In this section we are going to exercise our solver a little by the
computation of a small mixed integer optimization problem (MIP) from
miplib, a collection of MIP models, housed at Rice University. Here are
the original comments on the example:

@example
NAME:         flugpl
ROWS:         18
COLUMNS:      18
INTEGER:      11
NONZERO:      46
BEST SOLN:    1201500 (opt)
LP SOLN:      1167185.73
SOURCE:       Harvey M. Wagner              
              John W. Gregory (Cray Research)
              E. Andrew Boyd (Rice University)
APPLICATION:  airline model
COMMENTS:     no integer variables are binary 
@end example

@example
%
% from file: library('clpqr/examples/mip')
%
example(flugpl, Obj, Vs, Ints, []) :-
  Vs = [ Anm1,Anm2,Anm3,Anm4,Anm5,Anm6,
         Stm1,Stm2,Stm3,Stm4,Stm5,Stm6,
         UE1,UE2,UE3,UE4,UE5,UE6],
  Ints = [Stm6, Stm5, Stm4, Stm3, Stm2,
          Anm6, Anm5, Anm4, Anm3, Anm2, Anm1],

  Obj =    2700*Stm1 + 1500*Anm1 + 30*UE1
         + 2700*Stm2 + 1500*Anm2 + 30*UE2 
         + 2700*Stm3 + 1500*Anm3 + 30*UE3
         + 2700*Stm4 + 1500*Anm4 + 30*UE4 
         + 2700*Stm5 + 1500*Anm5 + 30*UE5
         + 2700*Stm6 + 1500*Anm6 + 30*UE6,

  allpos(Vs),
  @{  Stm1 = 60, 0.9*Stm1 +1*Anm1 -1*Stm2 = 0, 
     0.9*Stm2 +1*Anm2 -1*Stm3 = 0, 0.9*Stm3 +1*Anm3 -1*Stm4 = 0, 
     0.9*Stm4 +1*Anm4 -1*Stm5 = 0, 0.9*Stm5 +1*Anm5 -1*Stm6 = 0,
     150*Stm1 -100*Anm1 +1*UE1 >= 8000,
     150*Stm2 -100*Anm2 +1*UE2 >= 9000,
     150*Stm3 -100*Anm3 +1*UE3 >= 8000,
     150*Stm4 -100*Anm4 +1*UE4 >= 10000,
     150*Stm5 -100*Anm5 +1*UE5 >= 9000,
     150*Stm6 -100*Anm6 +1*UE6 >= 12000,
     -20*Stm1 +1*UE1 =< 0, -20*Stm2 +1*UE2 =< 0, -20*Stm3 +1*UE3 =< 0, 
     -20*Stm4 +1*UE4 =< 0, -20*Stm5 +1*UE5 =< 0, -20*Stm6 +1*UE6 =< 0,
     Anm1 =< 18, 57 =< Stm2, Stm2 =< 75, Anm2 =< 18,
     57 =< Stm3, Stm3 =< 75, Anm3 =< 18, 57 =< Stm4,
     Stm4 =< 75, Anm4 =< 18, 57 =< Stm5, Stm5 =< 75,
     Anm5 =< 18, 57 =< Stm6, Stm6 =< 75, Anm6 =< 18
   @}.

  allpos([]).
  allpos([X|Xs]) :- @{X >= 0@}, allpos(Xs).
@end example

We can first check whether the relaxed problem has indeed the quoted
infimum:

@example
clp(r) ?- example(flugpl, Obj, _, _, _), inf(Obj, Inf).

Inf = 1167185.7255923203
@end example

Computing the infimum under the additional constraints that @var{Stm6},
@var{Stm5}, @var{Stm4}, @var{Stm3}, @var{Stm2}, @var{Anm6}, @var{Anm5},
@var{Anm4}, @var{Anm3}, @var{Anm2}, @var{Anm1} assume integer values at
the infimum is computationally harder, but the query does not change
much:

@example
clp(r) ?- example(flugpl, Obj, _, Ints, _), bb_inf(Ints, Obj, Inf).

Inf = 1201500.0000000005
@end example

@node Implementation Architecture, Fragments and Bits,A Mixed Integer Linear Optimization Example, CLPQR
@section Implementation Architecture

The system consists roughly of the following components: 
@itemize @bullet
@item A polynomial normal form expression simplification mechanism.
@item A solver for linear equations [Holzbaur 92].
@item A simplex algorithm to decide linear inequalities [Holzbaur 94].
@end itemize

@node Fragments and Bits, CLPQR Bugs,Implementation Architecture, CLPQR
@section Fragments and Bits

The internal data structure for rational numbers is
@code{rat(@var{Num},@var{Den})}. @var{Den} is always positive, i.e. the
sign of the rational number is the sign of @var{Num}. Further, @var{Num}
and @var{Den} are relative prime. Note that integer @var{N} looks like
@code{rat(@var{N},1)} in this representation. You can control printing
of terms with @code{portray/1}.

Partial Evaluation

Once one has a working solver, it is obvious and attractive to run the
constraints in a clause definition at read time or compile time and
proceed with the answer constraints in place of the original
constraints. This gets you constant folding and in fact the full
algebraic power of the solver applied to the avoidance of computations
at runtime. The mechanism to realize this idea is to use
@code{call_residue/2} for the expansion of @code{@{@}/1}.

Asserting with Constraints

If you use the dynamic data base, the clauses you assert might have
constraints on the variables occurring in the clause. This should work
as follows:

@example
clp(r) ?- @{A < 10@}, assert(p(A)).

@{A < 10.0@}
yes

clp(r) ?- p(X).

@{X<10.0@}
@end example

YAP currently does not implement this feature.

@node CLPQR Bugs, CLPQR References, Fragments and Bits, CLPQR
@section CLPQR bugs

@itemize @bullet
@item The fuzzy comparison of floats is the source for all sorts of
weirdness. If a result in R surprises you, try to run the program in Q
before you send me a bug report.

@item The projector for floundered nonlinear relations keeps too many
variables. Its output is rather unreadable.

@item Disequations are not projected properly. 

@item This list is probably incomplete.
@end itemize

Please send bug reports to @code{christian@@ai.univie.ac.at}.

@node CLPQR References, ,CLPQR Bugs, CLPQR
@section CLPQR References

[Colmerauer 90] Colmerauer A.: An Introduction to Prolog III,
Communications of the ACM, 33(7), 69-90, 1990.

[Heintze et al. 87] Heintze N., Jaffar J., Michaylov S., Stuckey P., Yap
R.: The CLP(R) Programmers Manual, Monash University, Clayton, Victoria,
Australia, Department of Computer Science, 1987.

[Holzbaur 92] Holzbaur C.: A High-Level Approach to the Realization of
CLP Languages, in Proceedings of the JICSLP92 Post-Conference Workshop
on Constraint Logic Programming Systems, Washington D.C., 1992.

[Holzbaur 92] Holzbaur C.: Metastructures vs. Attributed Variables in
the Context of Extensible Unification, in Bruynooghe M. & Wirsing
M.(eds.), Programming Language Implementation and Logic Programming,
Springer, LNCS 631, pp.260- 268, 1992.

[Holzbaur 94] Holzbaur C.: A Specialized, Incremental Solved Form
Algorithm for Systems of Linear Inequalities, Austrian Research
Institute for Artificial Intelligence, Vienna, TR-94-07, 1994.

[Jaffar & Michaylov 87] Jaffar J., Michaylov S.: Methodology and
Implementation of a CLP System, in Lassez J.L.(ed.), Logic Programming -
Proceedings of the 4th International Conference - Volume 1, MIT Press,
Cambridge, MA, 1987.

@node CHR, Logtalk, CLPQR, Top
@chapter Constraint Handling Rules

@menu
* CHR Copyright::               
* CHR Introduction::            
* CHR Introductory Examples::   
* CHR Library::                 
* CHR Debugging::               
* CHR Programming Hints::       
* CHR Constraint Handlers::     
* CHR Backward Compatibility::  
@end menu

@node CHR Copyright, CHR Introduction, CHR, CHR
@unnumberedsec Copyright

This chapter is Copyright @copyright{} 1996-98 LMU

LMU (Ludwig-Maximilians-University)@*
Munich, Germany

Permission is granted to make and distribute verbatim copies of
this chapter provided the copyright notice and this permission notice
are preserved on all copies.

Permission is granted to copy and distribute modified versions of this
chapter under the conditions for verbatim copying, provided that the entire
resulting derived work is distributed under the terms of a permission
notice identical to this one.

Permission is granted to copy and distribute translations of this chapter
into another language, under the above conditions for modified versions,
except that this permission notice may be stated in a translation approved
by LMU.

@node CHR Introduction, CHR Introductory Examples, CHR Copyright, CHR
@comment  node-name,  next,  previous,  up
@section Introduction

Experience from real-life applications using constraint-based
programming has shown that typically, one is confronted with a
heterogeneous mix of different types of constraints. To be able to
express constraints as they appear in the application and to write and
combine constraint systems, a special purpose language for writing
constraint systems called @dfn{constraint handling rules} (CHR) was
developed.  CHR have been used to encode a wide range of constraint
handlers (solvers), including new domains such as terminological and
temporal reasoning.  Several CHR libraries exist in declarative
languages such as Prolog and LISP, worldwide more than 20 projects use
CHR. You can find more information about CHR at URL:
@code{http://www.pst.informatik.uni-muenchen.de/personen/fruehwir/chr-intro.html}

The high-level CHR are an excellent tool for rapid prototyping
and implementation of constraint handlers. The usual abstract formalism
to describe a constraint system, i.e.@: inference rules, rewrite rules,
sequents, formulas expressing axioms and theorems, can be written as
CHR in a straightforward way.  Starting from this executable
specification, the rules can be refined and adapted to the specifics of
the application.

The CHR library includes a compiler, which translates
CHR programs into Prolog programs on the fly, and a runtime system,
which includes a stepper for debugging.  Many constraint
handlers are provided in the example directory of the library.

CHR are essentially a committed-choice language consisting of guarded
rules that rewrite constraints into simpler ones until they are solved.
CHR define both @dfn{simplification} of and @dfn{propagation} over
constraints. Simplification replaces constraints by
simpler constraints while preserving logical equivalence (e.g.@:  
@code{X>Y,Y>X <=> fail}).  Propagation adds new constraints which are
logically redundant but may cause further simplification (e.g.@: 
@code{X>Y,Y>Z ==> X>Z}).  Repeatedly applying CHR incrementally simplifies
and finally solves constraints (e.g.@: @code{A>B,B>C,C>A}
leads to @code{fail}.

With multiple heads and propagation rules, CHR provide two features
which are essential for non-trivial constraint handling.  The
declarative reading of CHR as formulas of first order logic allows
one to reason about their correctness.  On the other hand, regarding
CHR as a rewrite system on logical formulas allows one to reason
about their termination and confluence.

In case the implementation of CHR disagrees with your expectations
based on this chapter, drop a line to the current maintainer:
@code{christian@@ai.univie.ac.at} (Christian Holzbaur).

@node CHR Introductory Examples, CHR Library, CHR Introduction, CHR
@comment  node-name,  next,  previous,  up
@section Introductory Examples

We define a CHR constraint for less-than-or-equal, @code{leq}, that can
handle variable arguments. This handler can be found in the library as
the file @code{leq.pl}.  (The code works regardless of options switched
on or off.)
@example
:- use_module(library(chr)).

handler leq.
constraints leq/2.
:- op(500, xfx, leq).

reflexivity  @@ X leq Y <=> X=Y | true.
antisymmetry @@ X leq Y , Y leq X <=> X=Y.
idempotence  @@ X leq Y \ X leq Y <=> true.
transitivity @@ X leq Y , Y leq Z ==> X leq Z.
@end example

The CHR specify how @code{leq} simplifies and propagates as a
constraint.  They implement reflexivity, idempotence, antisymmetry and
transitivity in a straightforward way.  CHR @code{reflexivity} states
that @code{X leq Y} simplifies to @code{true}, provided it is the case that
@code{X=Y}. This test forms the (optional) guard of a rule, a
precondition on the applicability of the rule.  Hence, whenever we see a
constraint of the form @code{A leq A} we can simplify it to @code{true}.

The rule @code{antisymmetry} means that if we find @code{X leq Y} as well as
@code{Y leq X} in the constraint store, we can replace it by the
logically equivalent @code{X=Y}. Note the different use of @code{X=Y} in
the two rules: In the @code{reflexivity} rule the equality is a
precondition (test) on the rule, while in the @code{antisymmetry} rule
it is enforced when the rule fires. (The reflexivity rule could also have 
been written as @code{reflexivity  @ X leq X <=> true}.)

The rules @code{reflexivity} and @code{antisymmetry} are
@dfn{simplification CHR}. In such rules, the constraints found are
removed when the rule applies and fires.  The rule @code{idempotence} is
a @dfn{simpagation CHR}, only the constraints right of @code{'\'} will
be removed. The rule  says that if we find @code{X leq Y} and another @code{X
leq Y} in the constraint store, we can remove one.

Finally, the rule @code{transitivity} states that the conjunction
@code{X leq Y, Y leq Z} implies @code{X leq Z}.  Operationally, we add
@code{X leq Z} as (redundant) constraint, without removing the
constraints @code{X leq Y, Y leq Z}. This kind of CHR is called
@dfn{propagation CHR}.

Propagation CHR are useful, as the query @code{A leq B,C leq A,B leq C}
illustrates: The first two constraints cause CHR @code{transitivity} to
fire and add @code{C leq B} to the query. This new constraint together
with @code{B leq C} matches the head of CHR @code{antisymmetry}, @code{X
leq Y, Y leq X}.  So the two constraints are replaced by
@code{B=C}. Since @code{B=C} makes @code{B} and @code{C} equivalent, CHR
@code{antisymmetry} applies to the constraints @code{A leq B, C leq A},
resulting in @code{A=B}. The query contains no more CHR constraints, the
simplification stops.  The constraint handler we built has solved
@code{A leq B, C leq A, B leq C} and produced the answer @code{A=B,
B=C}:

@example
A leq B,C leq A,B leq C. 
% C leq A, A leq B propagates C leq B by transitivity.
% C leq B, B leq C simplifies to B=C by antisymmetry.
% A leq B, C leq A simplifies to A=B by antisymmetry since B=C.
A=B,B=C.
@end example

Note that multiple heads of rules are essential in solving these
constraints.  Also note that this handler implements a (partial) order
constraint over any constraint domain, this generality is only possible
with CHR.

As another example, we can implement the sieve of Eratosthenes to
compute primes simply as (for variations see the handler
@file{primes.pl}):

@example
:- use_module(library(chr)).
handler eratosthenes.
constraints primes/1,prime/1.

primes(1) <=> true.
primes(N) <=> N>1 | M is N-1,prime(N),primes(M). % generate candidates

absorb(J) @@ prime(I) \ prime(J) <=> J mod I =:= 0 | true. 
@end example

The constraint @code{primes(N)} generates candidates for prime numbers,
@code{prime(M)}, where @code{M} is between @code{1} and @code{N}.
The candidates react with each other such that each
number absorbs multiples of itself. In the end, only prime numbers remain.

Looking at the two rules defining @code{primes/1}, note that head
matching is used in CHR, so the first rule will only apply to
@code{primes(1)}. The test @code{N>1} is a guard (precondition) on the
second rule.  A call with a free variable, like @code{primes(X)},
will delay (suspend).  The third, multi-headed rule @code{absorb(J)} 
reads as follows:
If there is a constraint @code{prime(I)} and some other constraint
@code{prime(J)} such that @code{J mod I =:= 0} holds, i.e.@: @code{J} is a
multiple of @code{I}, then keep @code{prime(I)} but remove
@code{prime(J)} and execute the body of the  rule,
@code{true}.

@node CHR Library, CHR Debugging, CHR Introductory Examples, CHR
@section CHR Library

CHR extend the Prolog syntax by a few constructs introduced in the next
sections. Technically, the extension is achieved through the
@code{user:term_expansion/2} mechanism. A file that contains a constraint
handler may also contain arbitrary Prolog code.  Constraint handling
rules can be scattered across a file. Declarations and options should
precede rules. There can only be at most one constraint handler per module.

@menu
* CHR Loading the Library::     
* CHR Declarations::            
* CHR Syntax::                  
* How CHR work::                
* CHR Pragmas::                 
* CHR Options::                 
* CHR Built-In Predicates::     
* CHR Consulting and Compiling::  
* CHR Compiler-generated Predicates::  
* CHR Operator Declarations::   
* CHR Exceptions::              
@end menu

@node CHR Loading the Library, CHR Declarations, CHR Library, CHR Library
@subsection Loading the Library

Before you can load or compile any file containing a
constraint handler (solver) written in CHR, the @code{chr} library
module has to be imported:
@example
| ?- use_module(library(chr)).
@end example
It is recommended to include the corresponding directive at the 
start of your files containing handlers:
@example
  :- use_module(library(chr)).
@end example

@node CHR Declarations, CHR Syntax, CHR Loading the Library, CHR Library
@subsection Declarations

Declarations in files containing CHR affect the compilation and thus
the behavior of the rules at runtime.

The mandatory handler declaration precedes any other CHR specific
code.
Example:
@example
handler minmax.
@end example
A handler name must be a valid Prolog @code{atom}.
Per module, only one constraint handler can be defined.

The constraints must be declared before they are used by rules.
With this mandatory declaration one lists the constraints 
the rules will later talk about. The declaration can be used more
than once per handler.
Example:
@example
constraints  leq/2, minimum/3, maximum/3.
@end example


The following
optional declaration allows for conditional rule compilation.
Only the rules mentioned get compiled. Rules are referred to by their 
names (@pxref{CHR Syntax}). The latest occurrence takes precedence if used
more than once per handler. Although it can be put anywhere in the handler
file, it makes sense, as with other declarations, to use it early.
Example:
@example
rules antisymmetry, transitivity.
@end example

To simplify the handling of operator declarations, in particular
during @code{fcompile/1}, @code{operator/3} declarations with the
same denotation as @code{op/3}, but 
taking effect during compilation and loading, are helpful.
Example:
@example
operator(700, xfx, ::).
operator(600, xfx, :).
@end example

@node CHR Syntax, How CHR work, CHR Declarations, CHR Library
@subsection Constraint Handling Rules, Syntax

A constraint handling rule has one or more heads, an optional guard, a
body and an optional name.  A @dfn{Head} is a @dfn{Constraint}. A
constraint is a callable Prolog term,
whose functor is a declared constraint.
The @dfn{Guard} is a Prolog goal.  
The @dfn{Body} of a rule is a Prolog
goal (including constraints).  
A rule can be named with
a @dfn{Name} which can be any Prolog term (including variables
from the rule). 

There are three kinds of constraint handling rules:
@example
Rule           --> [Name @@] 
                     (Simplification | Propagation | Simpagation) 
                        [pragma Pragma].

Simplification --> Heads         <=> [Guard '|'] Body
Propagation    --> Heads         ==> [Guard '|'] Body
Simpagation    --> Heads \ Heads <=> [Guard '|'] Body

Heads          --> Head | Head, Heads
Head           --> Constraint | Constraint # Id
Constraint     --> a callable term declared as constraint
Id             --> a unique variable

Guard          --> Ask | Ask & Tell
Ask            --> Goal
Tell           --> Goal
Goal           --> a callable term,
                   including conjunction and disjunction etc.

Body           --> Goal

Pragma         --> a conjunction of terms usually referring to 
                   one or more heads identified via #/2
@end example

The symbol @samp{|} separates the guard (if present) from the body of a
rule.  Since @samp{|} is read as @samp{;} (disjunction) by the  
reader, care has to be taken when using disjunction in the guard
or body of the rule. The top level disjunction will always be
interpreted as guard-body separator @samp{|}, so proper bracketing has
to be used, e.g.@: @code{a <=> (b;c) | (d;e)} instead of @code{a <=> b;c |
d;e} and @code{a <=> true | (d;e)} instead of @code{a <=> (d;e)}.

In simpagation rules, @samp{\} separates the heads of the rule into two parts.

Individual head constraints may be tagged with variables via
@samp{#}, which may be used as identifiers in pragma declarations, for
example. Constraint identifiers must be distinct variables, not
occurring elsewhere in the heads.

Guards test the applicability of a rule.  Guards come in two
parts, tell and ask, separated by @samp{&}. 
If the @samp{&} operator is not present, the whole guard is
assumed to be of the ask type.

Declaratively, a rule relates heads and body @dfn{provided the guard is
true}.  A simplification rule means that the heads are true if and only
if the body is true.  A propagation rule means that the body is true if
the heads are true.  A simpagation rule combines a simplification and a
propagation rule.  The rule @code{Heads1 \ Heads2 <=> Body} is
equivalent to the simplification rule @code{Heads1, Heads2 <=> Heads1,
Body}. However, the simpagation rule is more compact to write, more
efficient to execute and has better termination behavior than the
corresponding simplification rule, since the constraints comprising
@code{Heads1} will not be removed and inserted again.

@node How CHR work, CHR Pragmas, CHR Syntax, CHR Library
@subsection How CHR work

Each CHR constraint is associated with all rules in whose heads it
occurs by the CHR compiler. Every time a CHR constraint is executed
(called) or woken and reconsidered, it checks itself the applicability
of its associated CHR by @dfn{trying} each CHR.  By default, the rules
are tried in textual order, i.e.@: in the order they occur in the defining
file.  To try a CHR, one of its heads is matched against the constraint.
Matching succeeds if the constraint is an instance of the head.
If a CHR has more than one head, the constraint store is searched for
@dfn{partner} constraints that match the other heads.  Heads are tried
from left to right, except that in simpagation rules, the heads to be
removed are tried before the head constraints to be kept (this is done
for efficiency reasons).  If the matching succeeds, the guard is
executed. Otherwise the next rule is tried.

The guard either succeeds or fails.  A guard succeeds if the execution
of its Ask and Tell parts succeeds and in the ask part no variable that
occurs also in the heads was touched or the cause of an instantiation
error. The ask guard will fail otherwise.  A variable is @dfn{touched}
if it is unified with a term (including other variables from other
constraints) different from itself.  Tell guards, on the contrary, are
trusted and not checked for that property.  If the guard succeeds, the
rule applies.  Otherwise the next rule is tried.

If the firing CHR is a simplification rule, the matched constraints are
removed from the store and the body of the CHR is executed.  Similarly
for a firing simpagation rule, except that the constraints that matched
the heads preceding @samp{\} are kept.  If the firing CHR is a
propagation rule the body of the CHR is executed without removing any
constraints.  It is remembered that the propagation rule fired, so it
will not fire again with the same constraints if the constraint is woken
and reconsidered.  If the currently active constraint has not been
removed, the next rule is tried.

If the current constraint has not been removed and all rules have been
tried, it delays until a variable occurring in the constraint is
touched.  Delaying means that the constraint is inserted into the
constraint store. When a constraint is woken, all its rules are tried
again. (This process can be watched and inspected with the CHR debugger,
see below.)


@node CHR Pragmas, CHR Options, How CHR work, CHR Library
@subsection Pragmas

Pragmas are annotations to rules and constraints that enable the 
compiler to generate more specific, more optimized code. A pragma
can be a conjunction of the following terms:

@table @code

@item already_in_heads
The intention of simplification and simpagation rules is often
to combine the heads into a stronger version of one of them.
Depending on the strength of the guard, the new constraint may
be identical to one of the heads to removed by the rule.
This removal followed by addition is inefficient and may even cause
termination problems. If the pragma is used, this situation is
detected and the corresponding problems are avoided.
The pragma applies to all constraints removed by the rule.

@item already_in_head(Id)
Shares the intention of the previous pragma, but affects only 
the constraint indicated via @var{Id}. Note that one can use more
than one pragma per rule.

@item passive(Id)
No code will be generated for the specified constraint in the particular
head position. This means that the constraint will not see the rule, it
is passive in that rule. This changes the behavior of the CHR system,
because normally, a rule can be entered starting from each head constraint.
Usually this pragma will improve the efficiency of the
constraint handler, but care has to be taken in order not to lose
completeness.

For example, in the handler @code{leq}, any pair of constraints, say
@code{A leq B, B leq A}, that matches the head @code{X leq Y , Y leq X}
of the @code{antisymmetry} rule, will also match it when the constraints
are exchanged, @code{B leq A, A leq B}. Therefore it is enough if a
currently active constraint enters this rule in the first head only,
the second head can be declared to be passive. Similarly for the
@code{idempotence} rule.  For this rule, it is more efficient to declare
the first head passive, so that the currently active constraint will be
removed when the rule fires (instead of removing the older constraint
and redoing all the propagation with the currently active constraint).
Note that the compiler itself detects the symmetry of the two head
constraints in the simplification rule @code{antisymmetry}, thus it is
automatically declared passive and the compiler outputs @code{CHR
eliminated code for head 2 in antisymmetry}.
@example
antisymmetry @ X leq Y, Y leq X # Id <=> X=Y
           pragma passive(Id).
idempotence  @ X leq Y # Id \ X leq Y <=> true
           pragma passive(Id).
transitivity @ X leq Y # Id, Y leq Z ==> X leq Z
           pragma passive(Id).
@end example
Declaring the first head of rule @code{transitivity} passive changes the
behavior of the handler. It will propagate less depending on the order in
which the constraints arrive:
@example
?- X leq Y, Y leq Z.
X leq Y,
Y leq Z,
X leq Z ? 

?- Y leq Z, X leq Y.
Y leq Z,
X leq Y ? 

?- Y leq Z, X leq Y, Z leq X.
Y = X,
Z = X ?
@end example
The last query shows that the handler is still complete in the sense
that all circular chains of leq-relations are collapsed into equalities.

@end table

@node CHR Options, CHR Built-In Predicates, CHR Pragmas, CHR Library
@subsection Options

Options parametrise the rule compilation process.
Thus they should precede the rule definitions. Example:
@example
option(check_guard_bindings, off).
@end example

The format below lists the names of the recognized options together with
the acceptable values. The first entry in the lists is the default
value.

@table @code

@item option(debug_compile,         [off,on]).
Instruments the generated code such that the execution
of the rules may be traced (@pxref{CHR Debugging}).

@item option(check_guard_bindings,  [on,off]).
Per default, for guards of type ask the CHR runtime system makes sure
that no variables are touched or the cause of an instantiation
error. These checks may be turned off with this option, i.e.@: all guards
are treated as if they were of the tell variety. The option was
kept for backward compatibility. Tell and ask guards offer better
granularity.

@item option(already_in_store,      [off,on]).
If this option is on, the CHR runtime system checks for the presence of
an identical constraint upon the insertion into the store. If present,
the attempted insertion has no effect. Since checking for duplicates for
all constraints costs, duplicate removal specific to individual
constraints, using a few simpagation rules of the following form instead,
may be a better solution.
@example
Constraint \ Constraint <=> true.
@end example

@item option(already_in_heads,      [off,on]).
The intention of simplification and simpagation rules is often
to combine the heads into a stronger version of one of them.
Depending on the strength of the guard, the new constraint may
be identical to one of the heads removed by the rule.
This removal followed by addition is inefficient and may even cause
termination problems. If the option is enabled, this situation is
detected and the corresponding problems are avoided.
This option applies to all constraints and is provided mainly
for backward compatibility. Better grained control can be achieved
with corresponding pragmas. (@pxref{CHR Pragmas}).

@end table

The remaining options are meant for CHR implementors only:
@table @code
@item option(flatten,               [on,off]).
@item option(rule_ordering,         [canonical,heuristic]).
@item option(simpagation_scheme,    [single,multi]).
@item option(revive_scheme,         [new,old]).
@item option(dead_code_elimination, [on,off]).
@end table

@node CHR Built-In Predicates, CHR Consulting and Compiling, CHR Options, CHR Library
@subsection Built-In Predicates

This table lists the predicates made available by the CHR library.  They
are meant for advanced users, who want to tailor the CHR system 
towards their specific needs.
@table @code
@item current_handler(?Handler, ?Module)
@findex current_handler/2
Nondeterministically enumerates the defined handlers
with the module they are defined in.

@item current_constraint(?Handler, ?Constraint)
@findex current_constraint/2
Nondeterministically enumerates the defined constraints
in the form @var{Functor/Arity} and the handlers they are
defined in.

@item insert_constraint(+Constraint, -Id)
@findex insert_constraint/2
Inserts @var{Constraint} into the constraint store
without executing any rules. The constraint will be woken and reconsidered
when one of the variables in @var{Constraint} is touched. @var{Id} is
unified with an internal object representing the constraint.
This predicate only gets defined when a handler and constraints are
declared (@pxref{CHR Declarations}).

@item insert_constraint(+Constraint, -Id, ?Term)
@findex insert_constraint/3
Inserts @var{Constraint} into the constraint store without executing
any rules. The constraint will be woken and reconsidered when one of the
variables in @var{Term} is touched. @var{Id} is
unified with an internal object representing the constraint.
This predicate only gets defined when a handler and constraints are
declared (@pxref{CHR Declarations}).

@item find_constraint(?Pattern, -Id)
@findex find_constraint/2
Nondeterministically enumerates constraints from the constraint
store that match @var{Pattern}, i.e.@: which are instances of
@var{Pattern}. @var{Id} is
unified with an internal object representing the constraint.

@item find_constraint(-Var, ?Pattern, -Id)
@findex find_constraint/3
Nondeterministically enumerates constraints from the constraint
store that delay on @var{Var} and match @var{Pattern}, 
i.e.@: which are instances of
@var{Pattern}. The identifier @var{Id} can be used to
refer to the constraint later, e.g.@: for removal.

@item findall_constraints(?Pattern, ?List)
@findex findall_constraints/2
Unifies @var{List} with a list of @code{Constraint # Id} pairs
from the constraint store that match @var{Pattern}.

@item findall_constraints(-Var, ?Pattern, ?List)
@findex findall_constraints/3
Unifies @var{List} with a list of @code{Constraint # Id} pairs
from the constraint store that delay on @var{Var} and
match @var{Pattern}.

@item remove_constraint(+Id)
@findex remove_constraint/1
Removes the constraint @var{Id}, obtained with one of the previous predicates,
from the constraint store.

@item unconstrained(?Var)
@findex unconstrained/1
Succeeds if no CHR constraint delays on @var{Var}. Defined as:
@example
unconstrained(X) :- 
    find_constraint(X, _, _), !, fail.
unconstrained(_).
@end example

@item notify_constrained(?Var)
@findex notify_constrained/1
Leads to the reconsideration of the constraints associated with @var{Var}.
This mechanism allows solvers to communicate reductions on the set of
possible values of variables prior to making bindings.
@end table

@node CHR Consulting and Compiling, CHR Compiler-generated Predicates, CHR Built-In Predicates, CHR Library
@subsection Consulting and Compiling Constraint Handlers

The CHR compilation process has been made as transparent as possible.
The user deals with files containing CHR just as with files containing
ordinary Prolog predicates. Thus CHR may be consulted, compiled with
various compilation modes, and compiled to file.

@node CHR Compiler-generated Predicates, CHR Operator Declarations, CHR Consulting and Compiling, CHR Library
@subsection Compiler-generated Predicates

Besides predicates for the defined constraints,
the CHR compiler generates some support predicates in the
module containing the handler. To avoid naming conflicts,
the  following predicates must not be defined or referred to
by user code in the same module:
@table @code
@item verify_attributes/3
@item attribute_goal/2
@item attach_increment/2
@item 'attach_F/A'/2
for every defined constraint F/A.
@item 'F/A_N_M_...'/Arity
for every defined constraint F/A. N,M is are integers, Arity > A.
@end table

For the prime number example that is:
@example
        attach_increment/2
        attach_prime/1/2
        attach_primes/1/2
        attribute_goal/2
        goal_expansion/3
        prime/1
        prime/1_1/2
        prime/1_1_0/3
        prime/1_2/2
        primes/1
        primes/1_1/2
        verify_attributes/3
@end example

If an author of a handler wants to avoid naming conflicts with the
code that uses the handler, it is easy to encapsulate the handler.
The module declaration below puts the handler into module @code{primes},
which exports only selected predicates - the constraints in our example.
@example        
        :- module(primes, [primes/1,prime/1]).

        :- use_module(library(chr)).

        handler eratosthenes.
        constraints primes/1,prime/1.
        ...
@end example

@node CHR Operator Declarations, CHR Exceptions, CHR Compiler-generated Predicates, CHR Library
@subsection Operator Declarations

This table lists the operators as used by the CHR library:
@example
@group
:- op(1200, xfx, @@).
:- op(1190, xfx, pragma).
:- op(1180, xfx, [==>,<=>]).
:- op(1180, fy,  chr_spy).
:- op(1180, fy,  chr_nospy).
:- op(1150, fx,  handler).
:- op(1150, fx,  constraints).
:- op(1150, fx,  rules).
:- op(1100, xfx, '|').
:- op(1100, xfx, \ ).
:- op(1050, xfx, &).
:- op( 500, yfx, #).
@end group
@end example

@node CHR Exceptions,  , CHR Operator Declarations, CHR Library
@subsection Exceptions

The CHR runtime system reports instantiation and type errors for the predicates:
@table @code
@item find_constraint/2
@item findall_constraints/3
@item insert_constraint/2
@item remove_constraint/1
@item notify_constrained/1
@end table

The only other CHR specific runtime error is:
@table @code
@item @{CHR ERROR: registering <New>, module <Module> already hosts <Old>@}
An attempt to load a second handler New into module <Module> already 
hosting handler <Old> was made.
@end table

The following exceptional conditions are detected by the CHR
compiler:

@table @code
@item @{CHR Compiler ERROR: syntax rule <N>: <Term>@}
If the N-th <Term> in the file being loaded violates the CHR syntax (@pxref{CHR Syntax}).

@item @{CHR Compiler ERROR: too many general heads in <Name>@}
Unspecific heads in definitions like
@code{C \ C <=> true}
must not be combined with other heads in rule <Name>.

@item @{CHR Compiler ERROR: bad pragma <Pragma> in <Name>@}
The pragma <Pragma> used in rule <Name> does not qualify. Currently this only happens if
<Pragma> is unbound.

@item @{CHR Compiler ERROR: found head <F/A> in <Name>, expected one of: <F/A list>@}
Rule <Name> has a head of given F/A which is not among the defined constraints.

@item @{CHR Compiler ERROR: head identifiers in <Name> are not unique variables@}
The identifiers to refer to individual constraints (heads) via @samp{#} in
rule <Name> do not meet the indicated requirements.

@item @{CHR Compiler ERROR: no handler defined@}
CHR specific language elements, declarations or rules for example, are
used before a handler was defined. This error is usually reported a couple
of times, i.e.@: as often as there are CHR forms in the file expecting the missing
definition.

@item @{CHR Compiler ERROR: compilation failed@}
Not your fault. Send us a bug report.
@end table

@node CHR Debugging, CHR Programming Hints, CHR Library, CHR
@section Debugging CHR Programs

Use @code{option(debug_compile,on)} preceding any rules
in the file containing the handler to enable CHR debugging. 
The CHR debugging mechanism works by instrumenting the code
generated by the CHR compiler.
Basically, the CHR debugger works like the Prolog debugger. 
The main differences are: there are extra ports specific to CHR,
and the CHR debugger provides no means for the user to change
the flow of control, i.e.@: there are currently no @var{retry} and @var{fail}
options available.

@menu
* CHR Control Flow Model::      
* CHR Debugging Predicates::    
* CHR Spy-points::              
* CHR Debugging Messages::      
* CHR Debugging Options::       
@end menu

@node CHR Control Flow Model, CHR Debugging Predicates, CHR Debugging, CHR Debugging
@subsection Control Flow Model
@cindex CHR control flow model

The entities reflected by the CHR debugger are constraints
and rules. Constraints are treated like ordinary Prolog goals
with the usual ports: @code{[call,exit,redo,fail]}.
In addition, constraints may get inserted into or removed from the 
constraint store (ports: @code{insert,remove}), and stored
constraints containing variables will be woken and reconsidered
(port: @code{wake}) when variables are touched.

The execution of a constraint consists of trying to apply the rules
mentioning the constraint in their heads.  Two ports for rules reflect this
process: At a @code{try} port the active constraint matches one of the
heads of the rule, and matching constraints for the remaining heads of
the rule, if any, have been found as well.  The transition from a
@code{try} port to an @code{apply} port takes place when the guard has
been successfully evaluated, i.e.@: when the rule commits. At the
@code{apply} port, the body of the rule is just about to be executed. The
body is a Prolog goal transparent to the CHR debugger. If the
rule body contains CHR constraints, the CHR debugger will track
them again.  If the rules were consulted, the Prolog debugger can be
used to study the evaluations of the other predicates in the body.

@node CHR Debugging Predicates, CHR Spy-points, CHR Control Flow Model, CHR Debugging
@subsection CHR Debugging Predicates
@cindex CHR debugging predicates

The following predicates control the operation of the CHR debugger:

@table @code
@item chr_trace
@findex chr_trace/0
Switches the CHR debugger on and ensures that the
next time control enters a CHR port, a message will be produced and
you will be asked to interact. @refill

@noindent
At this point you have a number of options.  @xref{CHR Debugging Options}.  In
particular, you can just type @key{cr} (Return) to @dfn{creep} (or single-step)
into your program. You will
notice that the CHR debugger stops at many ports.  If this is not
what you want, the predicate @code{chr_leash} gives full control over
the ports at which you are prompted. @refill

@item chr_debug
@findex chr_debug/0
Switches the CHR debugger on and ensures that the next time
control enters a CHR port with a spy-point set, a message will be produced
and you will be asked to interact. @refill
@refill

@item chr_nodebug
@findex chr_nodebug/0
Switches the CHR debugger off.  If there are any spy-points set then they
will be kept.  @refill

@item chr_notrace
@findex chr_notrace/0
Equivalent to @code{chr_nodebug}.  @refill

@item chr_debugging
@findex chr_debugging/0
Prints onto the standard error stream information about the current
CHR debugging state.  This will show: @refill

@enumerate
@item
Whether the CHR debugger is switched on.
@item
What spy-points have been set (see below).
@item
What mode of leashing is in force (see below).
@end enumerate

@item chr_leash(@var{+Mode})
@findex chr_leash/1
The leashing mode is set to @var{Mode}. It determines the CHR ports at
which you are to be prompted when you @dfn{creep} through your program.
At unleashed ports a tracing message is still output, but program
execution does not stop to allow user interaction.  Note that the ports
of spy-points are always leashed (and cannot be unleashed).  @var{Mode}
is a list containing none, one or more of the following port names:
@refill

@table @code
@item call
Prompt when a constraint is executed for the first time.

@item exit
Prompt when the constraint is successfully processed, i.e.@: the applicable
rules have applied.

@item redo
Prompt at subsequent exits generated by non-deterministic rule bodies.

@item fail
Prompt when a constraint fails.

@item wake
Prompt when a constraint from the constraint store is woken and
reconsidered because one of its variables has been touched.

@item try
Prompt just before the guard evaluation of a rule, after constraints
matching the heads have been found.

@item  apply
Prompt upon the application of a rule, after the successful guard evaluation,
when the rule commits and fires, just before evaluating the body.

@item insert
Prompt when a constraint gets inserted into the constraint store,
i.e.@: after all rules have been tried.

@item remove
Prompt when a constraint gets removed from the constraint store, e.g.
when a simplification rule applies.

@end table

@noindent
The initial value of the CHR leashing mode is
@code{[call,exit,fail,wake,apply]}. Predefined shortcuts are:
@table @code
@item chr_leash(none), chr_leash(off)
To turn leashing off.

@item chr_leash(all)
To prompt at every port.

@item chr_leash(default)
Same as @code{chr_leash([call,exit,fail,wake,apply])}.

@item chr_leash(call)
No need to use a list if only a singular port is to be leashed.
@end table

@end table

@node CHR Spy-points, CHR Debugging Messages, CHR Debugging Predicates, CHR Debugging
@subsection CHR Spy-points
@cindex CHR spy-points

For CHR programs of any size, it is clearly impractical to creep through the
entire program.  @dfn{Spy-points} make it possible to stop the program
upon an event of interest.  Once
there, one can set further spy-points in order to catch the control flow
a bit further on, or one can start creeping.

Setting a spy-point on a constraint or a rule indicates that you wish to
see all control flow through the various ports involved, except during
skips.  When control passes through any port with a spy-point set on it,
a message is output and the user is asked to interact.  Note that the
current mode of leashing does not affect spy-points: user interaction is
requested at @emph{every} port.

Spy-points are set and removed by the following predicates,
which are declared as prefix operators:

@table @code
@item chr_spy @var{Spec}
@findex chr_spy/1
Sets spy-points on constraints and rules given by @var{Spec}, which is
is of the form:

@table @asis
@item @var{_} (variable)
denoting all constraints and rules, or:

@item @var{constraints Cs}
where @var{Cs} is one of
@table @asis

@item @var{_} (variable)
denoting all constraints

@item @var{C},...,@var{C}
denoting a list of constraints @var{C}

@item @var{Name}
denoting all constraints with this functor, regardless of arity

@item @var{Name/Arity}
denoting the constraint of that name and arity
@end table

@item @var{rules Rs}
where @var{Rs} is one of:
@table @asis

@item @var{_} (variable)
denoting all rules

@item @var{R},...,@var{R}
denoting a list of rules @var{R}

@item @var{Name}
where @var{Name} is the name of a rule in any handler.

@item @var{already_in_store}
The name of a rule implicitly defined by the system when
the option @code{already_in_store} is in effect.

@item @var{already_in_heads}
The name of a rule implicitly defined by the system when
the option @code{already_in_heads} or the corresponding 
pragmas are in effect.

@item @var{Handler:Name}
where @var{Handler} is the name of a constraint handler
and @var{Name} is the name of a rule in that handler

@end table
@end table

@noindent
Examples:

@example
| ?- chr_spy rules rule(3), transitivity, already_in_store.
| ?- chr_spy constraints prime/1.
@end example

If you set spy-points, the CHR debugger will be switched on.

@item chr_nospy @var{Spec}
@findex chr_nospy/1
Removes spy-points on constraints and rules given by @var{Spec}, where
@var{Spec} is of the form as described for @code{chr_spy @var{Spec}}.
There is no @code{chr_nospyall/0}.  To remove all CHR spy-points use
@code{chr_nospy _}.
@refill

@end table

@noindent
The options available when you arrive at a spy-point are described later.
@xref{CHR Debugging Options}.

@node CHR Debugging Messages, CHR Debugging Options, CHR Spy-points, CHR Debugging
@subsection CHR Debugging Messages
@cindex CHR debugging messages

All trace messages are output to the standard error
stream.  This allows you to trace programs while they are performing
file I/O.  The basic format is as follows:

@example
@var{S} 3 1 try eratosthenes:absorb(10) @@ prime(9)#<c4>, prime(10)#<c2> ?
@end example

@noindent
@var{S} is a spy-point indicator. It is printed as @samp{@ } if there is 
no spy-point, as @samp{r}, indicating that there is a spy-point on this
rule, or as @samp{c} if one of the involved constraints has a spy-point.

The first number indicates the current depth of the execution; i.e.@:
the number of direct @dfn{ancestors} the currently active constraint
has.
@refill

The second number indicates the head position of the currently active
constraint at rule ports.

The next item tells you which port is currently traced.

A constraint or a matching rule are printed next.
Constraints print as @code{Term#Id}, where @var{Id} is a unique
identifier pointing into the constraint store.
Rules are printed as @code{Handler:Name @@}, followed by the constraints
matching the heads.

The final @samp{?} is the prompt indicating that you should type in one of the
debug options (@pxref{CHR Debugging Options}). @refill

@node CHR Debugging Options,  , CHR Debugging Messages, CHR Debugging
@subsection CHR Debugging Options
@cindex CHR debugging options

This section describes the options available when the system prompts you
after printing out a debugging message.  Most of them you know from the
standard Prolog debugger. All the options are one letter mnemonics, some
of which can be optionally followed by a decimal integer.  They are read
from the standard input stream up to the end of the line
(Return, @key{<cr>}). Blanks will be ignored.

The only option which you really have to remember is @samp{h}.  This
provides help in the form of the following list of available options.

@example
@group
CHR debugging options:
   <cr>   creep            c      creep
    l     leap                         
    s     skip             s <i>  skip (ancestor i)
    g     ancestors                   
    &     constraints      & <i>  constraints (details)
    n     nodebug          =      debugging
    +     spy this                    
    -     nospy this       .      show rule
    <     reset printdepth < <n>  set printdepth
    a     abort            b      break
    ?     help             h      help
@end group
@end example

@table @kbd
@item c
@itemx @key{<cr>}
@dfn{creep} causes the debugger to single-step to the very next port
and print a message.  Then if the port is leashed, the
user is prompted for further interaction.  Otherwise, it continues creeping.
If leashing is off, creep is the same as @dfn{leap} (see below) except that a
complete trace is printed on the standard error stream.  @refill

@item l
@dfn{leap} causes the debugger to resume running your program, only
stopping when a spy-point is reached (or when the program terminates).
Leaping can thus be used to follow the execution at a higher level than
exhaustive tracing. @refill

@item s
@itemx s i
@dfn{skip} over the entire execution of the constraint.  That is, you
will not see anything until control comes back to this constraint (at
either the @code{exit} port or the @code{fail} port). This includes
ports with spy-points set; they will be masked out during the skip.
The command can be used with a numeric argument to skip the execution
up to and including the ancestor indicated by the argument.
Example:
@example
@group
      ...
      4   - exit    prime(8)#<c6> ? g
Ancestors:
      1   1 apply   eratosthenes:rule(2) @@ primes(10)#<c1>
      2   1 apply   eratosthenes:rule(2) @@ primes(9)#<c3>
      3   1 apply   eratosthenes:rule(2) @@ primes(8)#<c5>
      4   - call    prime(8)#<c6>

      4   - exit    prime(8)#<c6> ? s 2
      2   - exit    primes(9)#<c3> ? 
@end group
@end example

@item g
@dfn{print ancestors} provides you with a list of ancestors to the
currently active constraint, i.e.@: all constraints not yet exited that led 
to the current constraint in the derivation sequence.
The format is the same as with trace messages. 
Constraints start with @code{call} entries in the stack.
The subsequent application of a rule replaces the call entry
in the stack with an @code{apply} entry. Later the constraint
shows again as @code{redo} or @code{fail} entry.
Example:
@example
@group
      0   - call    primes(10)#<c1> ? 
      1   1 try     eratosthenes:rule(2) @@ primes(10)#<c1> ? g

Ancestors:
      1   - call    primes(10)#<c1>

      1   1 try     eratosthenes:rule(2) @@ primes(10)#<c1> ? 
      1   1 apply   eratosthenes:rule(2) @@ primes(10)#<c1> ? 
      1   - call    prime(10)#<c2> ? 
      2   - insert  prime(10)#<c2>
      2   - exit    prime(10)#<c2> ? g

Ancestors:
      1   1 apply   eratosthenes:rule(2) @@ primes(10)#<c1>
      2   - call    prime(10)#<c2>
@end group
@end example

@item &
@dfn{print constraints} prints a list of the constraints in the
constraint store. With a numeric argument, details relevant primarily to
CHR implementors are shown. @refill

@item n
@dfn{nodebug} switches the CHR debugger off. @refill

@item =
@dfn{debugging} outputs information concerning the status of the CHR debugger
as via @code{chr_debugging/0}

@item +
@dfn{spy this} sets a spy-point on the current constraint or rule.

@item -
@dfn{nospy this} removes the spy-point from the current constraint or
rule, if it exists.

@item .
@dfn{show rule}
prints the current rule instantiated by the matched constraints. Example:
@example
 8 1 apply  era:absorb(8) @@ prime(4)#<c14> \ prime(8)#<c6> ? .

 absorb(8) @@
  prime(4)#<c14> \ 
    prime(8)#<c6> <=>

    8 mod 4=:=0
    |
    true.
@end example

@item <
@itemx < n
While in the debugger, a @dfn{printdepth} is in effect for limiting
the subterm nesting level when printing rules and constraints.
The limit is initially 10.  This command, without arguments, resets the
limit to 10.  With an argument of @var{n}, the limit is set to @var{n},
treating 0 as infinity.
@refill

@item a
@dfn{abort} calls the built-in predicate @code{abort/0}.
@refill

@item b
@dfn{break} calls the built-in predicate @code{break/0}, thus putting
you at a recursive top-level.  When you end the break (entering ^D)
you will be re-prompted at the port at which you broke. The
CHR debugger is temporarily switched off as you call the break and will
be switched on again when you finish the break and go back to the old
execution.  Any changes to the CHR leashing or to spy-points during the
break will remain in effect.
@refill

@item ?
@itemx h
@dfn{help} displays the table of options given above.  @refill

@end table


@node CHR Programming Hints, CHR Constraint Handlers, CHR Debugging, CHR
@section Programming Hints

This section gives you some programming hints for CHR. For maximum
efficiency of your constraint handler, see also the previous subsections
on declarations and options.

Constraint handling rules for a given constraint system can often be
derived from its definition in formalisms such as inference rules,
rewrite rules, sequents, formulas expressing axioms and theorems.
CHR can also be found by first considering
special cases of each constraint and then looking at interactions of
pairs of constraints sharing a variable. Cases that do not occur in the
application can be ignored.

It is important to find the right @dfn{granularity} of the constraints.
Assume one wants to express that @dfn{n} variables are different from
each other.  It is more efficient to have a single constraint
@code{all_different(List_of_n_Vars)} than @dfn{n*n} inequality
constraints between each pair of different variables.  However, the
extreme case of having a single constraint modeling the whole constraint
store will usually be inefficient.

Starting from an executable specification, the rules can then be refined
and adapted to the specifics of the application.  Efficiency can be
improved by weakening the guards to perform simplification as early as
needed and by strengthening the guards to do the @dfn{just right} amount
of propagation.  Propagation rules can be expensive, because no
constraints are removed. 

The more heads a rule has, the more expensive it is.  @dfn{Rules with
several heads} are more efficient, if the heads of the rule share a
variable (which is usually the case). Then the search for a partner
constraint has to consider less candidates.  In the current
implementation, constraints are indexed by their functors, so that the
search is only performed among the constraints containing the shared
variable. Moreover, two rules with identical
(or sufficiently similar) heads can be merged into one rule so that the
search for a partner constraint is only performed once instead of twice.

As @dfn{guards} are tried frequently, they should be simple @dfn{tests}
not involving side-effects.  Head matching is more efficient than
explicitly checking equalities in the ask-part of the guard.  In the
tell part of a guard, it should be made sure that variables from the
head are never touched (e.g.@: by using @code{nonvar} or @code{ground} if
necessary). For efficiency and clarity reasons, one should also avoid
using constraints in guards.  Besides conjunctions, disjunctions are
allowed in the guard, but they should be used with care.  The use of
other control built-in predicates in the guard is
discouraged.  Negation and if-then-else in the ask part of a guard can
give wrong results, since e.g.@: failure of the negated goal may be due to
touching its variables.

@dfn{Several handlers can be used simultaneously if} they do not share
constraints with the same name. The implementation will not work
correctly if the same constraint is defined in rules of different
handlers that have been compiled separately. In such a case, the
handlers must be merged @dfn{by hand}. This means that the source code
has to be edited so that the rules for the shared constraint are
together (in one module). Changes may be necessary (like
strengthening guards) to avoid divergence or loops in the computation.


@node CHR Constraint Handlers, CHR Backward Compatibility, CHR Programming Hints, CHR
@section Constraint Handlers

The CHR library comes with plenty of constraint handlers written in CHR.
The most recent versions of these are maintained at:

@example
http://www.pst.informatik.uni-muenchen.de/~fruehwir/chr-solver.html
@end example

@table @file

@item arc.pl
classical arc-consistency over finite domains

@item bool.pl
simple Boolean constraints

@item cft.pl
feature term constraints according to the CFT theory

@item domain.pl
finite domains over arbitrary ground terms and interval domains over integers and reals, but without arithmetic functions

@item gcd.pl
elegant two-liner for the greatest common divisor

@item interval.pl
straightforward interval domains over integers and reals, with arithmetic functions

@item kl-one.pl
terminological reasoning similar to KL-ONE or feature trees

@item leq.pl
standard introductory CHR example handler for less-than-or-equal

@item list.pl
equality constraints over concatenations of lists (or strings)

@item listdom.pl
a straightforward finite enumeration list domains over integers, similar to @file{interval.pl}

@item math-elim.pl
solves linear polynomial equations and inequations using variable elimination, several variations possible

@item math-fougau.pl
solves linear polynomial equations and inequations by combining variable elimination for equations with Fourier's algorithm for inequations, several variations possible

@item math-fourier.pl 
a straightforward Fouriers algorithm to solve polynomial inequations
over the real or rational numbers

@item math-gauss.pl
a straightforward, elegant implementation of variable elimination for equations in one rule

@item minmax.pl
simple less-than and less-than-or-equal ordering constraints together with minimum and maximum constraints

@item modelgenerator.pl
example of how to use CHR for model generation in theorem proving

@item monkey.pl
classical monkey and banana problem, illustrates how CHR can be used as
a fairly efficient production rule system

@item osf.pl
constraints over order sorted feature terms according to the OSF theory

@item oztype.pl
rational trees with disequality and OZ type constraint with intersection

@item pathc.pl
the most simple example of a handler for path consistency - two rules

@item primes.pl
elegant implementations of the sieve of Eratosthenes reminiscent of the
chemical abstract machine model, also illustrates use of CHR as a
general purpose concurrent constraint language

@item scheduling.pl
simple classical constraint logic programming scheduling example on building a house

@item tarski.pl
most of Tarski's axiomatization of geometry as constraint system

@item term.pl
Prolog term manipulation built-in predicates @code{functor/3, arg/3, =../2} as constraints

@item time-pc.pl
grand generic handler for path-consistency over arbitrary constraints,
load via @file{time.pl} to get a powerful solver for temporal constraints based on Meiri's unifying framework. @file{time-rnd.pl} contains a generator for random test problems.

@item time-point.pl
quantitative temporal constraints over time points using path-consistency

@item tree.pl
equality and disequality over finite and infinite trees (terms)

@item type.pl
equalities and type constraints over finite and infinite trees (terms)

@end table

You can consult or compile a constraint handler from the CHR library using
e.g.: 
@example
?- [library('chr/examples/gcd')].
?- compile(library('chr/examples/gcd')).
@end example
If you want to learn more about the handlers, 
look at their documented source code.

In addition, there are files with example queries for some handlers,
their file name starts with @file{examples-} and the file extension
indicates the handler, e.g.@: @file{.bool}:

@example
examples-adder.bool
examples-benchmark.math
examples-deussen.bool
examples-diaz.bool
examples-fourier.math
examples-holzbaur.math
examples-lim1.math
examples-lim2.math
examples-lim3.math
examples-puzzle.bool
examples-queens.bool
examples-queens.domain
examples-stuckey.math
examples-thom.math
@end example


@node CHR Backward Compatibility,  , CHR Constraint Handlers, CHR
@section Backward Compatibility

In this section, we discuss backward compatibility with the CHR library
of Eclipse Prolog.

@enumerate

@item
The restriction on at most two heads in a rule has been abandoned.  A
rule can have as many heads as you like. Note however, that searching
for partner constraints can be expensive.

@item
By default, rules are compiled in textual order. This gives the
programmer more control over the constraint handling process. In the
Eclipse library of CHR, the compiler was optimizing the order of
rules. Therefore, when porting a handler, rules may have to
be reordered. A good heuristic is to prefer simplification to
simpagation and propagation and to prefer rules with single heads to
rules with several heads. Instead of manually rearranging an old handler
one may also use the following combination of options to get the
corresponding effect:
@example
option(rule_ordering,heuristic).
option(revive_scheme,old).
@end example

@item
For backward compatibility, the @code{already_in_store},
@code{already_in_head} and @code{guard_bindings} options
are still around, but there are CHR syntax extensions: @ref{CHR Syntax} 
and pragmas @ref{CHR Pragmas}
offering better grained control.

@item
The Eclipse library of CHR provided automatic built-in labeling through
the @code{label_with} declaration. Since it was not widely used and can
be easily simulated, built-in labeling was dropped.  The same effect can
be achieved by replacing the declaration @code{label_with Constraint if
Guard} by the simplification rule @code{chr_labeling, Constraint <=>
Guard | Constraint', chr_labeling} and by renaming the head in each
clause @code{Constraint :- Body} into @code{Constraint' :- Body} where
@code{Constraint'} is a new predicate.  Efficiency can be improved by
declaring @code{Constraint} to be passive: @code{chr_labeling,
Constraint#Id <=> Guard | Constraint', chr_labeling pragma passive(Id)}.
This translation will not work if @code{option(already_in_heads,on)}.
In that case use e.g.@: @code{chr_labeling(_), Constraint <=> Guard |
Constraint', chr_labeling(_)} to make the new call to
@code{chr_labeling} differ from the head occurrence.

@item
The set of built-in predicates for advanced CHR users is now larger
and better designed. Also the debugger has been improved. The Opium
debugging environment is not available in SICStus Prolog.
@end enumerate


@node Logtalk, Parallelism, CHR, Extensions
@chapter Logtalk
@cindex logtalk

The Logtalk object-oriented extension is available once included 
with the @code{use_module(library(logtalk))} command. Note that, 
although we load Logtalk using the @code{use_module/1} built-in 
predicate, the system is not packaged as a module not does it use 
modules in its implementation.

Logtalk documentation is included in the Logtalk directory. For the 
latest news, please see the URL @url{http://www.ci.uc.pt/logtalk/logtalk.html}.


@node Parallelism, Tabling, Logtalk, Extensions
@chapter Parallelism

@cindex parallelism
@cindex or-parallelism
There has been a sizeable amount of work on an or-parallel
implementation for YAP, called @strong{YapOr}. Most of this work has
been performed by Ricardo Rocha. In this system parallelism is exploited
implicitly by running several alternatives in or-parallel. This option
can be enabled from the @code{configure} script or by checking the
system's @code{Makefile}.

@strong{YapOr} is still a very experimental system, going through rapid
development. The following restrictions are of note:

@itemize @bullet
@item @strong{YapOr} currently only supports the Linux/X86 and SPARC/Solaris
platforms. Porting to other Unix-like platforms should be straightforward.

@item @strong{YapOr} does not support parallel updates to the
data-base.

@item @strong{YapOr} does not support opening or closing of streams during
parallel execution.

@item Garbage collection and stack shifting are not supported in
@strong{YapOr}.  

@item Built-ins that cause side-effects can only be executed when
left-most in the search-tree. There are no primitives to provide
asynchronous or cavalier execution of these built-ins, as in Aurora or
Muse.

@item YAP does not support voluntary suspension of work.
@end itemize

We expect that some of these restrictions will be removed in future
releases.

@node Tabling, Low Level Tracing, Parallelism , Extensions
@chapter Tabling

@cindex tabling
An initial cut for an implementation of tabling in the style of
XSB-Prolog is now available. Tabling was implemented by Ricardo
Rocha. To experiment with tabling use @code{-DTABLING} to
@code{YAP_EXTRAS} in the system's @code{Makefile}.

You can use the directive @code{table} to force calls for the argument
predicate to be tabled. Tabling information is stored in a trie, as for
XSB-Prolog.

@node Low Level Tracing, Low Level Profiling, Tabling, Extensions
@chapter Tracing at Low Level

It is possible to follow the flow at abstract machine level if
YAP is compiled with the flag @code{LOW_LEVEL_TRACER}. Note
that this option is of most interest to implementers, as it quickly generates
an huge amount of information.

Low level tracing can be toggled from an interrupt handler by using the
option @code{T}. There are also two builtins that activate and
deactivate low level tracing:

@table @code
@item start_low_level_trace
@findex start_low_level_trace/0
@snindex start_low_level_trace/0
@cnindex start_low_level_trace/0
Begin display of messages at procedure entry and retry.

@item stop_low_level_trace
@findex start_low_level_trace/0
@snindex start_low_level_trace/0
@cnindex start_low_level_trace/0
Stop display of messages at procedure entry and retry.
@end table

Note that this compile-time option will slow down execution.

@node Low Level Profiling, , Low Level Tracing, Extensions
@chapter Profiling the Abstract Machine

Implementors may be interested in detecting on which abstract machine
instructions are executed by a program. The @code{ANALYST} flag can give
WAM level information. Note that this option slows down execution very
substantially, and is only of interest to developers of the system
internals, or to system debuggers.

@table @code
@item reset_op_counters
@findex reset_op_counters/0
@snindex reset_op_counters/0
@cnindex reset_op_counters/0
Reinitialise all counters.

@item show_op_counters(+@var{A})
@findex show_op_counters/1
@snindex show_op_counters/1
@cnindex show_op_counters/1
Display the current value for the counters, using label @var{A}. The
label must be an atom.

@item show_ops_by_group(+@var{A})
@findex show_ops_by_group/1
@snindex show_ops_by_group/1
@cnindex show_ops_by_group/1
Display the current value for the counters, organised by groups, using
label @var{A}. The label must be an atom.

@end table

@node Debugging,Efficiency,Extensions,Top 
@chapter Debugging

@menu
* Deb Preds:: Debugging Predicates
* Deb Interaction:: Interacting with the debugger
@end menu

@node Deb Preds, Deb Interaction, , Debugging
@section Debugging Predicates

The following predicates are available to control the debugging of
programs:

@table @code
@item debug
@findex debug/0
@saindex debug/0
@cyindex debug/0
Switches the debugger on.

@item debugging
@findex debugging/0
@syindex debugging/0
@cyindex debugging/0
Outputs status information about the debugger which includes the leash
mode and the existing spy-points, when the debugger is on.

@item nodebug
@findex nodebug/0
@syindex nodebug/0
@cyindex nodebug/0
Switches the debugger off.

@item spy +@var{P}
@findex spy/1
@syindex spy/1
@cyindex spy/1
 Sets spy-points on all the predicates represented by
@var{P}. @var{P} can either be a single specification or a list of 
specifications. Each one must be of the form @var{Name/Arity} 
or @var{Name}. In the last case all predicates with the name 
@var{Name} will be spied. As in C-Prolog, system predicates and 
predicates written in C, cannot be spied.

@item nospy +@var{P}
@findex nospy/1
@syindex nospy/1
@cyindex nospy/1
 Removes spy-points from all predicates specified by @var{P}.
The possible forms for @var{P} are the same as in @code{spy P}.

@item nospyall
@findex nospyall/0
@syindex nospyall/0
@cnindex nospyall/0
Removes all existing spy-points.

@item notrace
Switches off the debugger and stops tracing.

@item leash(+@var{M})
@findex leash/1
@syindex leash/1
@cyindex leash/1
 Sets leashing mode to @var{M}.
The mode can be specified as:
@table @code
@item full
prompt on Call, Exit, Redo and Fail
@item tight
prompt on Call, Redo and Fail
@item half
prompt on Call and Redo
@item loose
prompt on Call
@item off
never prompt
@end table
@noindent
The initial leashing mode is half.


@noindent
The user may also specify directly the debugger ports 
where he wants to be prompted. If the argument for leash 
is a number @var{N}, each of lower four bits of the number is used to
control prompting at one the ports of the box model. The debugger will 
prompt according to the following conditions:

@itemize @bullet
@item
if @code{N/\ 1 =\= 0}  prompt on fail 
@item
if @code{N/\ 2 =\= 0} prompt on redo
@item
if @code{N/\ 4 =\= 0} prompt on exit
@item
if @code{N/\ 8 =\= 0} prompt on call
@end itemize
@noindent
Therefore, @code{leash(15)} is equivalent to @code{leash(full)} and
@code{leash(0)} is equivalent to @code{leash(off)}.

@noindent
Another way of using @code{leash} is to give it a list with the names of
the ports where the debugger should stop. For example,
@code{leash([call,exit,redo,fail])} is the same as @code{leash(full)} or
@code{leash(15)} and @code{leash([fail])} might be used instead of
@code{leash(1)}.

@item spy_write(+@var{Stream},Term)
@findex spy_write/2
@snindex spy_write/2
@cnindex spy_write/2
If defined by the user, this predicate will be used to print goals by
the debugger instead of @code{write/2}.

@item trace
Switches on the debugger and starts tracing.

@end table


@node Deb Interaction, , Deb Preds, Debugging
@section Interacting with the debugger

Debugging with YAP is similar to debugging with C-Prolog. Both
systems include a procedural debugger, based in the four port model. In
this model, execution is seen at the procedure level: each activation of
a procedure is seen as a box with control flowing into and out of that
box.

 In the four port model control is caught at four key points: before 
entering the procedure, after exiting the procedure (meaning successful 
evaluation of all queries activated by the procedure), after backtracking but 
before trying new alternative to the procedure and after failing the 
procedure. Each one of these points is named a port:

@smallexample
@group
           *--------------------------------------*
   Call    |                                      |    Exit
---------> +  descendant(X,Y) :- offspring(X,Y).  + --------->
           |                                      |
           |  descendant(X,Z) :-                  |
<--------- +     offspring(X,Y), descendant(Y,Z). + <---------
   Fail    |                                      |    Redo
           *--------------------------------------*
@end group
@end smallexample

@table @code

@item Call
The call port is activated before initial invocation of
procedure. Afterwards, execution will try to match the goal with the
head of existing clauses for the procedure.
@item Exit
This port is activated if the procedure succeeds.
Control will  now leave the procedure and return to its ancestor.
@item Redo
if the goal, or goals, activated after the call port
fail  then backtracking will eventually return control to this procedure
through  the redo port.
@item Fail
If all clauses for this predicate fail, then the
invocation fails,  and control will try to redo the ancestor of this
invocation.
@end table

 To start debugging, the user will usually spy the relevant procedures,
entering debug mode, and start execution of the program. When finding
the first spy-point, YAP's debugger will take control and show a
message like:

@example
* (1)  call:  quicksort([1,2,3],_38) ?
@end example

 The debugger message will be shown while creeping, or at spy-points, 
and it includes four or five fields:

@itemize @bullet
@item
 The first two characters are used to point out special states of the
debugger. If the first character is a @code{*}, execution is at a
spy-point. If the second character is a @code{>}, execution has returned
either from a skip, a fail or a redo command.
@item
 The second field is the activation number, and uniquely identifies the
activation. The number will start from 1 and will be incremented for
each activation found by the debugger.
@item
 In the third field, the debugger shows the active port.
@item
 The fourth field is the goal. The goal is written by @code{write/1}.
@end itemize

 If the active port is leashed, the debugger will prompt the user with a
@code{?}, and wait for a command. A debugger command is just a
character, followed by a return. By default, only the call and redo
entries are leashed, but the @code{leash/1} predicate can be used in
order to make the debugger stop where needed.

 There are several commands available, but the user only needs to 
remember the help command, which is @code{h}. This command shows all the 
available options, which are:
@table @code
@item c - creep
this command makes YAP continue execution and stop at the next
leashed port.
@item return - creep
the same as c
@item l - leap
YAP will continue execution until a port of a spied predicate
is found;
@item k - quasi-leap
similar to leap but faster since the computation history is
not kept; useful when leap becomes too slow.
@item s - skip
YAP will continue execution without showing any messages until
returning to the current activation. Spy-points will be  ignored in this
mode. This command is meaningless, and therefore illegal, in the fail
and exit ports.
@item t - fast-skip
similar to skip but faster since the computation history is not
kept; useful when skip becomes too slow.
@item q - quasi-leap
YAP will continue execution until a port of a spied
predicate is found or until returning to the current activation.
@item f - fail
forces YAP to fail the goal proceeding directly to the  fail port.
The command is not available in the fail port.
@item r - retry
after this command, YAP will retry the present goal,  and so go
back to the call port. Note that any side effects of the goal will  not
be undone. This command is not available at the call port.
@item a - abort
execution will be aborted, and the interpreter will return to the
top-level.
@item n - nodebug
stop debugging but continue execution. The command will clear all active
spy-points, leave debugging mode and continue execution.
@item e - exit
leave YAP.
@item h - help
show the debugger commands.
@item ! Query
execute a query. YAP will not show the result of the query.
@item b - break
break active execution and launch a break level. This is  the same as !
break.
@item + - spy this goal
start spying the active goal. The same as @code{! spy  G} where @var{G}
is the active goal.
@item - - nospy this goal
stop spying the active goal. The same as @code{! nospy G} where @var{G} is
the active goal.
@item p - print
shows the active goal using print/1
@item d - display
shows the active goal using display/1
@item <Depth - debugger write depth
sets the maximum write depth, both for composite terms and lists, that
will be used by the debugger. For more
information about @code{write_depth/2} (@pxref{I/O Control}).
@item < - full term
resets to the default of ten the debugger's maximum write depth. For
more information about @code{write_depth/2} (@pxref{I/O Control}).
@end table

The debugging information, when fast-skip @code{quasi-leap} is used, will
be lost.

@node Efficiency, C-Interface, Debugging, Top
@chapter Indexing
The indexation mechanism restricts the set of clauses to be tried in a 
procedure by using information about the status of a selected argument of 
the goal (in YAP, as in most compilers, the first argument). 
This argument 
is then used as a key, selecting a restricted set of a clauses from all the 
clauses forming the procedure.

As an example, the two clauses for concatenate:

@example
concatenate([],L,L).
concatenate([H|T],A,[H|NT]) :- concatenate(T,A,NT).
@end example

If the first argument for the goal is a list, then only the second clause 
is of interest. If the first argument is the nil atom, the system needs to 
look only for the first clause. The indexation generates instructions that 
test the value of the first argument, and then proceed to a selected clause, 
or group of clauses.

Note that if the first argument was a free variable, then both clauses 
should be tried. In general, indexation will not be useful if the first 
argument is a free variable.

When activating a predicate, a Prolog system needs to store state 
information. This information, stored in a structure known as choice point 
or fail point, is necessary when backtracking to other clauses for the 
predicate. The operations of creating and using a choice point are very 
expensive, both in the terms of space used and time spent.
Creating a choice point is not necessary if there is only a clause for 
the predicate as there are no clauses to backtrack to. With indexation, this 
situation is extended: in the example, if the first argument was the atom 
nil, then only one clause would really be of interest, and it is pointless to 
create a choice point. This feature is even more useful if the first argument 
is a list: without indexation, execution would try the first clause, creating 
a choice point. The clause would fail, the choice point would then be used to 
restore the previous state of the computation and the second clause would 
be tried. The code generated by the indexation mechanism would behave 
much more efficiently: it would test the first argument and see whether it 
is a list, and then proceed directly to the second clause.

An important side effect concerns the use of "cut". In the above 
example, some programmers would use a "cut" in the first clause just to 
inform the system that the predicate is not backtrackable and force the 
removal the choice point just created. As a result, less space is needed but 
with a great loss in expressive power: the "cut" would prevent some uses of 
the procedure, like generating lists through backtracking. Of course, with 
indexation the "cut" becomes useless: the choice point is not even created.

Indexation is also very important for predicates with a large number 
of clauses that are used like tables:

@example
logician(aristhoteles,greek).
logician(frege,german).
logician(russel,english).
logician(godel,german).
logician(whitehead,english).
@end example

An interpreter like C-Prolog, trying to answer the query:

@example
?- logician(godel,X).
@end example

@noindent
would blindly follow the standard Prolog strategy, trying first the
first clause, then the second, the third and finally finding the
relevant clause.  Also, as there are some more clauses after the
important one, a choice point has to be created, even if we know the
next clauses will certainly fail. A "cut" would be needed to prevent
some possible uses for the procedure, like generating all logicians.  In
this situation, the indexing mechanism generates instructions that
implement a search table. In this table, the value of the first argument
would be used as a key for fast search of possibly matching clauses. For
the query of the last example, the result of the search would be just
the fourth clause, and again there would be no need for a choice point.

 If the first argument is a complex term, indexation will select clauses
just by testing its main functor. However, there is an important
exception: if the first argument of a clause is a list, the algorithm
also uses the list's head if not a variable. For instance, with the
following clauses,

@example
rules([],B,B).
rules([n(N)|T],I,O) :- rules_for_noun(N,I,N), rules(T,N,O).
rules([v(V)|T],I,O) :- rules_for_verb(V,I,N), rules(T,N,O).
rules([q(Q)|T],I,O) :- rules_for_qualifier(Q,I,N), rules(T,N,O).
@end example
@noindent
if the first argument of the goal is a list, its head will be tested, and only 
the clauses matching it will be tried during execution.

Some advice on how to take a good advantage of this mechanism:

@itemize @bullet

@item
 Try to make the first argument an input argument.

@item
 Try to keep together all clauses whose first argument is not a 
variable, that will decrease the number of tests since the other clauses are 
always tried.

@item
 Try to avoid predicates having a lot of clauses with the same key. 
For instance, the procedure:

@end itemize

@example
type(n(mary),person).
type(n(john), person).
type(n(chair),object).
type(v(eat),active).
type(v(rest),passive).
@end example

@noindent
 becomes more efficient with:

@example
type(n(N),T) :- type_of_noun(N,T).
type(v(V),T) :- type_of_verb(V,T).

type_of_noun(mary,person).
type_of_noun(john,person).
type_of_noun(chair,object).

type_of_verb(eat,active).
type_of_verb(rest,passive).
@end example

@node C-Interface,YapLibrary,Efficiency,Top
@chapter C Language interface to YAP

YAP provides the user with the necessary facilities for writing
predicates in a language other than prolog. Since, under Unix systems,
most language implementations are link-able to C, we will describe here
only the YAP interface to the C language.

Before describing in full detail how to interface to C code, we will examine 
a brief example.

Assume the user requires a predicate @code{my_process_id(Id)} which succeeds
when @var{Id} unifies with the number of the process under which YAP is running.

In this case we will create a @code{my_process.c} file containing the
C-code described below.

@example
@cartouche
#include "../c_interface.h"

static int my_process_id(void) 
@{
     Term pid = MkIntTerm(getpid());
     Term out = ARG1;
     return(unify(out,pid));
@}

void init_my_predicates()
@{
     UserCPredicate("my_process_id",my_process_id,1);
@}
@end cartouche
@end example

The commands to compile the above file depend on the operating
system. Under Linux (i386 and Alpha) you should use:
@example
      gcc -c -shared -fPIC my_process.c
      ld -shared -o my_process.so my_process.o
@end example
@noindent
Under Solaris2 it is sufficient to use:
@example
      gcc  -fPIC -c my_process.c
@end example
@noindent
Under SunOS it is sufficient to use:
@example
      gcc -c my_process.c
@end example
@noindent
Under Digital Unix you need to create a @code{so} file. Use:
@example
      gcc tst.c -c -fpic
      ld my_process.o -o my_process.so -shared -expect_unresolved '*'
@end example
@noindent
and replace my @code{process.so} for my @code{process.o} in the
remainder of the example.
@noindent
And could be loaded, under YAP, by executing the following prolog goal
@example
      load_foreign_files(['my_process'],[],init_my_predicates).
@end example
Note that since Yap4.3.3 you should not give the suffix for object
files. YAP will deduce the correct suffix from the operating system it
is running under.

Yap4.3.3 now supports loading WIN/NT DLLs. Currently you must compile
YAP under cygwin to create a library yap.dll first. You can then use
this dll to create your own dlls. Have a look at the code in
library/regex to see how to create a dll under the cygwin/mingw32
environment.

After loading that file the following prolog goal 
@example
       my_process_id(N)
@end example
@noindent
would unify N with the number of the process under which Yap is running.


Having presented a full example, we will now examine in more detail the
contents of the C source code file presented above.

The include statement is used to make available to the C source code the
macros for the handling of prolog terms and also some Yap public
definitions.

The function @code{my_process_id} is the implementation, in C, of the
desired predicate.  Note that it returns an integer denoting the success
of failure of the goal and also that it has no arguments even though the
predicate being defined has one.
 In fact the arguments of a prolog predicate written in C are accessed
through macros, defined in the include file, with names @var{ARG1},
@var{ARG2}, ..., @var{ARG16} or with @var{ARG}(@var{N}) where @var{N} is
the argument number (starting with 1).  In the present case the function
uses just one local variable of type @code{ Term}, the type used for
holding Yap terms, where the integer returned by the standard unix
function @code{getpid()} is stored as an integer term (the conversion is
done by @code{MkIntTerm(Int))}. Then it calls the pre-defined routine
@code{unify(Term*, Term*)} which in turn returns an integer denoting
success or failure of the unification.

The role of the procedure @code{init_my_predicates} is to make known to
YAP, by calling @code{UserCPredicate}, the predicates being
defined in the file.  This is in fact why, in the example above,
@code{init_my_predicates} was passed as the third argument to
@code{load_foreign_files}.

The rest of this appendix describes exhaustively how to interface C to YAP.

@menu
* Manipulating Terms:: Primitives available to the C programmer
* Unifying Terms:: How to Unify Two Prolog Terms
* Manipulating Strings:: From character arrays to Lists of codes and back
* Memory Allocation:: Stealing Memory From Yap
* Controlling Streams:: Control How Yap sees Streams
* Calling Yap From C:: From C to Yap to C to Yap 
* Writing C:: Writing Predicates in C
* Loading Objects:: Loading Object Files
* Sav&Rest:: Saving and Restoring
* Yap4 Notes:: Changes in Foreign Predicates Interface
@end menu

@node Manipulating Terms, Unifying Terms, , C-Interface
@section Terms

This section provides information about the primitives available to the C
programmer for manipulating prolog terms.

Several C typedefs are included in the header file @code{yap/c_interface.h} to
describe, in a portable way, the C representation of prolog terms.
The user should write is programs using this macros to ensure portability of
code across different versions of YAP.


The  more important typedef is @var{Term} which is used to denote the type of a
prolog term.

Terms, from a point of view of the C-programmer,  can be classified as
follows
@table @i
@item    uninstantiated variables
@item    instantiated variables
@item    integers
@item    floating-point numbers
@item    database references
@item    atoms
@item    pairs (lists)
@item    compound terms
@end table

Before trying to find out the kind of a term, the C-programmer should insure
it is not an instantiated variable using the interface primitive
@example
    Term Deref(Term)
@end example
@noindent
which follows a possibly empty chain of instantiations and returns a term which
is not an instantiated variable.

Having done so, the primitive
@example
     Bool IsVarTerm(Term)
@end example
@noindent
returns true iff its argument is an uninstantiated variable. Conversely the
primitive
@example
      Bool IsGroundTerm(Term)
@end example
@noindent
returns true iff its argument is not a variable.


The user can create a new uninstantiated variable using the primitive
@example
      Term  MkVarTerm()
@end example


The following primitives can be used to discriminate among the different kinds
of non-variable terms:
@example
      Bool IsIntTerm(Term)
      Bool IsFloatTerm(Term)
      Bool IsDbRefTerm(Term)
      Bool IsAtomTerm(Term)
      Bool IsPairTerm(Term)
      Bool IsApplTerm(Term)
@end example
@noindent
@strong{Important Note:} when used on variables the primitives above
will return an unpredictable result.


The following primitives are provided for creating an integer term from an
integer and to access the value of an integer term.
@example
      Term MkIntTerm(Int)
      Int  IntOfTerm(Term)
@end example
@noindent
where @code{Int} is a typedef for the C integer type appropriate for the
machine or compiler in question (normally a 32 bit integer). Note that
the size of the allowed integers is implementation dependent but is always
greater or equal to 24 bits.


The two following primitives play a similar role for floating-point terms
@example
      Term MkFloatTerm(flt)
      flt  FloatOfTerm(Term)
@end example
@noindent
where @code{flt} is a typedef for the appropriate C floating point type.


No primitives are supplied to users for manipulating data base
references. 

A special typedef @code{Atom} is provided to describe prolog @i{atoms} and the
two following primitives can be used to manipulate atom terms
@example
      Term MkAtomTerm(Atom)
      Atom AtomOfTerm(Term)
@end example
@noindent
The two following primitives are available for associating atoms with their
names 
@example
      Atom  LookupAtom(char *)
      Atom  FullLookupAtom(char *)
      char* AtomName(Atom)
@end example
The function @code{LookupAtom} looks up an atom in the standard hash
table. The function  @code{FullLookupAtom} will also search if the atom
had been "hidden".


A @i{pair} is a Prolog term which consists of a pair of prolog terms designated
as the @i{head} and the @i{tail} of the term. The following primitives can
be used to manipulate pairs
@example
      Term  MkPairTerm(Term Head, Term Tail)
      Term  HeadOfTerm(Term)
      Term  TailOfTerm(Term)
@end example

A @i{compound} term consists of a @i{functor} and a sequence of terms with
length equal to the @i{arity} of the functor. A functor, described in C by
the typedef @code{Functor}, consists of an atom and of an integer.
The following primitives were designed to manipulate compound terms and 
functors
@example
      Term     MkApplTerm(Functor f, int n, Term[] args)
      Functor  FunctorOfTerm(Term)
      Term     ArgOfTerm(int argno,Term t)
      Functor  MkFunctor(Atom a,int arity)
      Atom     NameOfFunctor(Functor)
      Int      ArityOfFunctor(Functor)
@end example
@noindent
where @code{args} should be an array of @code{n} terms with @code{n} equal to the
arity of the functor, and @code{argno} should be greater or equal to 1 and less
or equal to the arity of the functor.

@strong{Note:} all the above primitives returning terms ensure that the
result is @i{dereferenced}, i.e. that it is not an instantiated variable.

@node Unifying Terms, Manipulating Strings, Manipulating Terms, C-Interface
@section Unification

The following routine is provided for attempting the unification of two
prolog terms
@example
      Int      unify(Term a, Term b)
@end example
@noindent
which attempts to unify the terms pointed to by @code{a} and @code{b} returning
a non-zero value if the unification succeeds and zero otherwise.

@node Manipulating Strings, Memory Allocation, Unifying Terms, C-Interface
@section Strings

The YAP C-interface now includes an utility routine to copy a string
represented as a list of a character codes to a previously allocated buffer
@example
      int StringToBuffer(Term String, char *buf, unsigned int bufsize)
@end example
@noindent
The routine copies the list of character codes @code{String} to a
previously allocated buffer @code{buf}. The string including a
terminating null character must fit in @code{bufsize} characters,
otherwise the routine will simply fail. The @code{StringToBuffer}
routine fails and generates an exception if @code{String} is not a valid
string.

The C-interface also includes utility routines to do the reverse, that
is, to copy a from a buffer to a list of character codes or to a list of
character atomsr
@example
      Term BufferToString(char *buf)
      Term BufferToAtomList(char *buf)
@end example
@noindent
The user-provided string must include a terminating null character.

@node Memory Allocation, Controlling Streams, Manipulating Strings, C-Interface
@section Memory Allocation

The next routine can be used to ask space from the Prolog data-base:
@example
      void      *AllocSpaceFromYap(int size)
@end example
@noindent
The routine returns a pointer to a buffer allocated from the code area,
or @code{NULL} if no space was available.

This Space can be released by using:
@example
      void      FreeSpaceFromYap(void *buf)
@end example
@noindent
The routine releases a buffer allocated from the code area. The system
may crash if @code{buf} is not a valid pointer to a buffer in the code
area.

@node Controlling Streams, Calling Yap From C, Memory Allocation, C-Interface
@section Controlling Yap Streams from @code{C}

The C-Interface also provides the C-application with a measure of
control over the Yap Input/Output system. The first routine allows one
to find a file number given a current stream:
@example
      int      YapStreamToFileNo(Term stream)
@end example
@noindent
This function gives the file descriptor for a currently available
stream. Note that null streams and in memory streams do not have
corresponding open streams, so the routine will return a
negative. Moreover, Yap will not be aware of any direct operations on
this stream, so information on, say, current stream position, may become
stale.

A second routine that is sometimes useful is:
@example
      void      YapCloseAllOpenStreams(void)
@end example
@noindent
This routine closes the Yap Input/Output system except for the first
three streams, that are always associated with the three standard Unix
streams. It is most useful if you are doing @code{fork()}.

The next routine allows a currently open file to become a stream. The
routine receives as arguments a file descriptor, the true file name as a
string, an atom with the yser name, and a set of flags:
@example
      void      YapOpenStream(void *FD, char *true, Term t, int flags)
@end example
@noindent
The available flags are @code{YAP_INPUT_STREAM},
@code{YAP_OUTPUT_STREAM}, @code{YAP_APPEND_STREAM},
@code{YAP_PIPE_STREAM}, @code{YAP_TTY_STREAM}, @code{YAP_POPEN_STREAM},
@code{YAP_BINARY_STREAM}, and @code{YAP_SEEKABLE_STREAM}. By default, the
stream is supposed to be at position 0.

@node Calling Yap From C, Writing C, Controlling Streams, C-Interface
@section From @code{C} back to Prolog

Newer versions of YAP allow for calling the Prolog interpreter from
@code{C}. One must first construct a goal @code{G}, and then it is
sufficient to perform:
@example
      Int      YapCallProlog(Term G)
@end example
@noindent
the result will be @code{0}, if the goal failed, or @code{1}, if the
goal succeeded. In this case, the variables in @var{G} will store the
values they have been unified with. Execution only proceeds until
finding the first solution to the goal, but you can call
@code{findall/3} or friends if you need all the solutions.

@node Writing C, Loading Objects, Calling Yap From C, C-Interface
@section Writing predicates in C

We will distinguish two kinds of predicates:
@table @i
@item @i{deterministic} predicates which either fail or succeed but are not
backtrackable, like the one in the introduction; 
@item @i{backtrackable}
predicates which can succeed more than once.
@end table

The first kind of predicates should be implemented as a C function with
no arguments which should return zero if the predicate fails and a
non-zero value otherwise. The predicate should be declared to
YAP, in the initialization routine, with a call to
@example
      void UserCPredicate(char *name, int *fn(), int arity);
@end example
@noindent
where @code{name} is the name of the predicate, @code{fn} is the C function
implementing the predicate and @code{arity} is its arity.

For the second kind of predicates we need two C functions. The first one
which is called when the predicate is first activated, and the second one
to be called on backtracking to provide (possibly) other solutions. Note
also that we normally also need to preserve some information to find out
the next solution.

In fact the role of the two functions can be better understood from the
following prolog definition
@example
       p :- start.
       p :- repeat,
                continue.
@end example
@noindent
where @code{start} and @code{continue} correspond to the two C functions
described above.


As an example we will consider implementing in C a predicate @code{n100(N)}
which, when called with an instantiated argument should succeed if that
argument is a numeral less or equal to 100, and, when called with an
uninstantiated argument, should provide, by backtracking, all the positive
integers less or equal to 100.

   To do that we first declare a structure, which can only consist
of prolog terms, containing the information to be preserved on backtracking
and a pointer variable to a structure of that type.

@example
typedef struct @{
    Term next_solution;  /* the next solution */
   @} n100_data_type;

n100_data_type *n100_data;
@end example

We now write the @code{C} function to handle the first call:

@example
static int start_n100()
@{
      Term t = ARG1;
      PRESERVE_DATA(n100_data,n100_data_type);
      if(IsVarTerm(t)) @{
          n100_data->next_solution = MkIntTerm(0);
          return(continue_n100());
       @}
      if(!IsIntTerm(t) || IntOfTerm(t)<0 || IntOfTerm(t)>100) @{
          cut_fail();
      @} else @{
          cut_succeed();
      @}
@}

@end example

The routine starts by getting the dereference value of the argument.
The call to @code{PRESERVE_DATA} is used to initialize the memory which will
hold the information to be preserved across backtracking.

If the argument of the predicate is a variable, the routine initializes the 
structure to be preserved across backtracking with the information
required to provide the next solution, and exits by calling @code{
continue_n100} to provide that solution.

If the argument was not a variable, the routine then checks if it was
an integer, and if so, if its value is positive and less than 100. In that case
it exits, denoting success, with @code{cut_succeed}, or otherwise exits with
@code{cut_fail} denoting failure.

The reason for using for using the macros @code{cut_succeed} and @code{cut_fail}
instead of just returning a non-zero value in the first case, and zero in the
second case, is that otherwise, if backtracking occurred later, the routine
@code{continue_n100} would be called to provide additional solutions.

The code required for the second function is
@example
static int continue_n100()
@{
      int n;
      Term t;
      Term sol = ARG1;
      PRESERVED_DATA(n100_data,n100_data_type);
      n = IntOfTerm(n100_data->next_solution);
      if( n == 100) @{
           t = MkIntTerm(n);
           unify(&sol,&t);
           cut_succeed();
        @}
       else @{
           unify(&sol,&(n100_data->next_solution));
           n100_data->next_solution = MkIntTerm(n+1);
           return(1);
        @}
@}
@end example

Note that again the macro @code{PRESERVED_DATA} is used at the beginning of
the function to access the data preserved from the previous solution.
Then it checks if the last solution was found and in that case exits
with @code{cut_succeed} in order to cut any further backtracking.  If this
is not the last solution then we save the value for the next solution in
the data structure and exit normally with 1 denoting success. Note also
that in any of the two cases we use the function @code{unify} to bind the
argument of the call to the value saved in @code{
n100_state->next_solution}.


Note also that the only correct way to signal failure in a backtrackable
predicate is to use the @code{cut_fail} macro.

Backtrackable predicates should be declared to YAP, in a way
similar to what happened with deterministic ones, but using instead a
call to
@example
      void UserBackCPredicate(char *name,
                 int *init(), int *cont(), int arity, int sizeof);
@end example
@noindent
where @code{name} is a string with the name of the predicate, @code{init} and
@code{cont} are the C functions used to start and continue the execution of
the predicate, and @code{arity} is the predicate arity.

@node Loading Objects, Sav&Rest, Writing C, C-Interface
@section Loading Object Files

The primitive predicate
@example
      load_foreign_files(@var{Files},@var{Libs},@var{InitRoutine})
@end example
@noindent
should be used, from inside YAP, to load object files produced by the C
compiler. The argument @var{ObjectFiles} should be a list of atoms
specifying the object files to load, @var{Libs} is a list (possibly
empty) of libraries to be passed to the unix loader (@code{ld}) and
InitRoutine is the name of the C routine (to be called after the files
are loaded) to perform the necessary declarations to YAP of the
predicates defined in the files. 

YAP will search for @var{ObjectFiles} in the current directory first. If
it cannot find them it will search for the files using the environment
variable @code{YAPLIBDIR}, if defined, or in the default library.

In a.out systems YAP by default only reserves a fixed amount of memory
for object code (64 Kbytes in the current version). Should this size
prove inadequate the flag @code{-c n} can be passed to YAP (in the
command line invoking YAP) to force the allocation of @code{n} Kbytes.

@node Sav&Rest, Yap4 Notes, Loading Objects, C-Interface
@section Saving and Restoring

@comment The primitive predicates @code{save} and @code{restore} will save and restore
@comment object code loaded with @code{load_foreign_files}. However, the values of
@comment any non-static data created by the C files loaded will not be saved nor
@comment restored.

Yap4 currently does not support @code{save} and @code{restore} for object code
loaded with @code{load_foreign_files}. We plan to support save and restore
in future releases of Yap.

@node Yap4 Notes, , Sav&Rest, C-Interface
@section Changes to the C-Interface in Yap4

Yap4 includes several changes over the previous @code{load_foreign_files}
interface. These changes were required to support the new binary code
formats, such as ELF used in Solaris2 and Linux.
@itemize @bullet
@item Access to elements in the new interface always goes through
@emph{functions}. This includes access to the argument registers,
@code{ARG1} to @code{ARG16}. This change breaks code such as
@code{unify(&ARG1,&t)}:
@example
@{
   unify(ARG1, t);
@}
@end example

@item @code{cut_fail()} and @code{cut_succeed()} are now functions.

@item The use of @code{Deref} is deprecated. All functions that return
Prolog terms, including the ones that access arguments, already
dereferenciate their arguments.

@item Space allocated with PRESERVE_DATA is ignored by garbage
collection and stack shifting. As a result, any pointers to a Prolog
stack object, including some terms, may be corrupted after garbage
collection or stack shifting. Prolog terms should instead be stored as
arguments to the backtrackable procedure.

@end itemize

@node YapLibrary, Compatibility, C-Interface, Top
@chapter Using YAP as a Library

YAP can be used as a library to be called from other
programs. To do so, you must first create the YAP library:
@example
make library
make install_library
@end example
This will install a file @code{libyap.a} in @var{LIBDIR} and the Prolog
headers in @var{INCLUDEDIR}. The library contains all the functionality
available in YAP, except the foreign function loader and for
@code{Yap}'s startup routines.

To actually use this library you must follow a five step process:

@enumerate
@item
 You must initialise the YAP environment. A single function,
@code{YapFastInit} asks for a contiguous chunk in your memory space, fills
it in with the data-base, and sets up YAP's stacks and
execution registers. You can use a saved space from a standard system by
calling @code{save_program/1}.
     
@item You then have to prepare a query to give to
YAP. A query is a Prolog term, and you just have to use the same
functions that are available in the C-interface.

@item You can then use @code{YapRunGoal(query)} to actually evaluate your
query. The argument is the query term @code{query}, and the result is 1
if the query succeeded, and 0 if it failed.

@item You can use the term destructor functions to check how
arguments were instantiated.

@item If you want extra solutions, you can use
@code{YapRestartGoal()} to obtain the next solution.

@end enumerate

The next program shows how to use this system. We assume the saved
program contains two facts for the procedure @t{b}:

@example
@cartouche
#include <stdio.h>
#include "Yap/c_interface.h"


int
main(int argc, char *argv[]) @{
  if (YapFastInit("saved_state") == YAP_BOOT_FROM_SAVED_ERROR)
    exit(1);
  if (YapRunGoal(MkAtomTerm(LookupAtom("do")))) @{
    printf("Success\n");
    while (YapRestartGoal())
      printf("Success\n");
  @}
  printf("NO\n");
@}
@end cartouche
@end example

The program first initialises YAP, calls the query for the
first time and succeeds, and then backtracks twice. The first time
backtracking succeeds, the second it fails and exits.

To compile this program it should be sufficient to do:

@example
cc -o exem -I../Yap4.3.0 test.c -lyap -lreadline -lm
@end example

You may need to adjust the libraries and library paths depending on the
Operating System and your installation of Yap.

Note that Yap4.3.0 provides the first version of the interface. The
interface may change and improve in the future.

The following C-functions are available from Yap:

@itemize @bullet
@item  YapCompileClause(@code{Term} @var{Clause})
@findex  YapCompileClause/1
Compile the Prolog term @var{Clause} and assert it as the last clause
for the corresponding procedure.

@item  @code{int} YapContinueGoal(@code{void})
@findex YapContinueGoal/0
Continue execution from the point where it stopped.

@item  @code{void} YapError(@code{char *} @var{error_description})
@findex YapError/1
Generate an YAP System Error with description given by the string
@var{error_description}.

@item  @code{void} YapExit(@code{int} @var{exit_code})
@findex YapExit/1
Exit YAP immediately. The argument @var{exit_code} gives the error code
and is supposed to be 0 after successful execution in Unix and Unix-like
systems.

@item  @code{Term} YapGetValue(@code{Atom} @var{at})
@findex  YapGetValue/1
Return the term @var{value} associated with the atom @var{at}. If no
such term exists the function will return the empty list.

@item  YapFastInit(@code{char *} @var{SavedState})
@findex  YapFastInit/1
Initialise a copy of YAP from @var{SavedState}. The copy is
monolithic and currently must be loaded at the same address where it was
saved. @code{YapFastInit} is a simpler version of @code{YapInit}.

@item  YapInit(@code{char *} @var{SavedState},  @code{int}
@var{HeapSize}, @code{int} @var{StackSize}, @code{int}
@var{TrailSize}, @code{int}  @var{NumberofWorkers}, @code{int}
@var{SchedulerLoop}, @code{int} @var{DelayedReleaseLoad}, @code{int}
@var{argc}, @code{char **}  @var{argv})
@findex  YapInit/9
Initialise YAP. In the future the arguments as a single @code{C}
structure.

If @var{SavedState} is not NULL, try to open and restore the file
@var{SavedState}. Initially YAP will search in the current directory. If
the saved state does not exist in the current directory YAP will use
either the default library directory or the directory given by the
environment variable @code{YAPLIBDIR}. Note that currently
the saved state must be loaded at the same address where it was saved.

If @var{HeapSize} is different from 0 use @var{HeapSize} as the minimum
size of the Heap (or code space). If @var{StackSize} is different from 0
use @var{HeapSize} as the minimum size for the Stacks. If
@var{TrailSize} is different from 0 use @var{TrailSize} as the minimum
size for the Trails.

The @var{NumberofWorkers}, @var{NumberofWorkers}, and
@var{DelayedReleaseLoad} are only of interest to the or-parallel system.

The argument count @var{argc} and string of arguments @var{argv}
arguments are to be passed to user programs as the arguments used to
call YAP.

@item  @code{void} YapPutValue(@code{Atom} @var{at}, @code{Term} @var{value})
@findex  YapPutValue/2
Associate the term @var{value} with the atom @var{at}. The term
@var{value} must be a constant. This functionality is used by YAP as a
simple way for controlling and communicating with the Prolog run-time.

@item  @code{Term} YapRead(@code{int (*)(void)} @var{GetC})
@findex  YapRead/1
Parse a Term  using the function @var{GetC} to input characters.

@item  @code{int} YapRunGoal(@code{Term} @var{Goal})
@findex YapRunGoal/1
Execute query @var{Goal} and return 1 if the query succeeds, and
0 otherwise.

@item  @code{int} YapRestartGoal(@code{void})
@findex YapRestartGoal/0
Look for the next solution to the current query by forcing YAP to backtrack.

@item  @code{int} YapReset(@code{void})
@findex YapReset/0
Reset execution environment (similar to the @code{abort/0}
builtin). This is useful when you want to start a new query before
asking all solutions to the previous query.

@item  @code{void} YapWrite(@code{Term} @var{t}, @code{void (*)(int)}
@var{PutC}, @code{int} @var{flags})
@findex  YapRead/1
Write a Term @var{t} using the function @var{PutC} to output
characters. The term is written according to a mask of the following
flags in the @code{flag} argument: @code{YAP_WRITE_QUOTED},
@code{YAP_WRITE_HANDLE_VARS},  and @code{YAP_WRITE_IGNORE_OPS}.

@item  @code{void} YapInitConsult(@code{int} @var{mode}, @code{char *} @var{filename})
@findex YapInitConsult/2
Enter consult mode on file @var{filename}. This mode maintains a few
data-structures internally, for instanc to know whether a predicate
before or not. It is still possible to execute goals in consult mode.

If @var{mode} is @code{TRUE} the file will be reconsulted, otherwise
just consulted. In practice, this function is most useful for
bootstraping Prolog, as otherwise one may call the Prolog predicate
@code{compile/1} or @code{consult/1} to do compilation.

Note that it is up to the user to open the file @var{filename}. The
@code{YapInitConsult} function only uses the file name for internal
bookkeeping.

@item  @code{void} YapEndConsult(@code{void})
@findex YapEndConsult/0
Finish consult mode.

@end itemize

Some observations:

@itemize @bullet
@item The system will core dump if you try to load the saved state in a
different address from where it was made. This may be a problem if
your program uses @code{mmap}. This problem will be addressed in future
versions of YAP.

@item Currently, the YAP library will pollute the name
space for your program.

@item The initial library includes the complete YAP system. In
the future we plan to split this library into several smaller libraries
(e.g., if you do not want to perform I/O).

@item You can generate your own saved states. Look at  the
@code{boot.yap} and @code{init.yap} files.

@end itemize

@node Compatibility, Operators, YapLibrary, Top
@chapter Compatibility with Other Prolog systems

YAP has been designed to be as compatible as possible with
other Prolog systems, and initially with C-Prolog. More recent work on
YAP has included features initially proposed for the Quintus
and SICStus Prolog systems.

Developments since @code{Yap4.1.6} we have striven at making
YAP compatible with the ISO-Prolog standard. 

@menu
* C-Prolog:: Compatibility with the C-Prolog interpreter
* SICStus Prolog:: Compatibility with the SICStus Prolog system
* ISO Prolog::  Compatibility with the ISO Prolog standard
@end menu

@node C-Prolog, SICStus Prolog, , Compatibility
@section Compatibility with the C-Prolog interpreter

@menu
C-Prolog Compatibility
* Major Differences with C-Prolog:: Major Differences between YAP and C-Prolog
* Fully C-Prolog Compatible:: Yap predicates fully compatible with
C-Prolog
* Not Strictly C-Prolog Compatible:: Yap predicates not strictly as C-Prolog
* Not in C-Prolog:: Yap predicates not available in C-Prolog
* Not in YAP:: C-Prolog predicates not available in YAP
@end menu

@node Major Differences with C-Prolog, Fully C-Prolog Compatible, , C-Prolog
@subsection Major Differences between YAP and C-Prolog.

YAP includes several extensions over the original C-Prolog system. Even
so, most C-Prolog programs should run under YAP without changes.

The most important difference between YAP and C-Prolog is that, being
YAP a compiler, some changes should be made if predicates such as
@code{assert}, @code{clause} and @code{retract} are used. First
predicates which will change during execution should be declared as
@code{dynamic} by using commands like:

@example
:- dynamic f/n.
@end example

@noindent where @code{f} is the predicate name and n is the arity of the
predicate. Note that  several such predicates can be declared in a
single command:
@example
 :- dynamic f/2, ..., g/1.
@end example

Primitive predicates such as @code{retract} apply only to dynamic
predicates.  Finally note that not all the C-Prolog primitive predicates
are implemented in YAP. They can easily be detected using the
@code{unknown} system predicate provided by YAP.

Last, by default YAP enables character escapes in strings. You can
disable the special interpretation for the escape character by using:
@example
@code{:- yap_flag(character_escapes,off).}
@end example
@noindent
or by using:
@example
@code{:- yap_flag(language,cprolog).}
@end example

@node Fully C-Prolog Compatible, Not Strictly C-Prolog Compatible, Major Differences with C-Prolog, C-Prolog
@subsection Yap predicates fully compatible with C-Prolog

These are the Prolog built-ins that are fully compatible in both
C-Prolog and YAP:

@printindex cy

@node Not Strictly C-Prolog Compatible, Not in C-Prolog, Fully C-Prolog Compatible, C-Prolog
@subsection Yap predicates not strictly compatible with C-Prolog

These are YAP built-ins that are also available in C-Prolog, but
that are not fully compatible:

@printindex ca

@node Not in C-Prolog, Not in YAP, Not Strictly C-Prolog Compatible, C-Prolog
@subsection Yap predicates not available in C-Prolog

These are YAP built-ins not available in C-Prolog.

@printindex cn

@node Not in YAP, , Not in C-Prolog, C-Prolog
@subsection Yap predicates not available in C-Prolog

These are C-Prolog built-ins not available in YAP:

@table @code
@item 'LC'
The following Prolog text uses lower case letters.

@item 'NOLC'
The following Prolog text uses upper case letters only.
@end table

@node SICStus Prolog, ISO Prolog, C-Prolog, Compatibility
@section Compatibility with the Quintus and SICStus Prolog systems

The Quintus Prolog system was the first Prolog compiler to use Warren's
Abstract Machine. This system was very influential in the Prolog
community. Quintus Prolog implemented compilation into an abstract
machine code, which was then emulated. Quintus Prolog also included
several new built-ins, an extensive library, and in later releases a
garbage collector. The SICStus Prolog system, developed at SICS (Swedish
Institute of Computer Science), is an emulator based Prolog system
largely compatible with Quintus Prolog. SICStus Prolog has evolved
through several versions. The current version includes several
extensions, such as an object implementation, co-routining, and
constraints.

Recent work in YAP has been influenced by work in Quintus and
SICStus Prolog. Wherever possible, we have tried to make YAP
compatible with recent versions of these systems, and specifically of
SICStus Prolog. You should use 
@example
:- yap_flag(language, sicstus).
@end example
@noindent
for maximum compatibility with SICStus Prolog.

@menu
SICStus Compatibility
* Major Differences with SICStus:: Major Differences between YAP and SICStus Prolog
* Fully SICStus Compatible:: Yap predicates fully compatible with
SICStus Prolog
* Not Strictly SICStus Compatible:: Yap predicates not strictly as
SICStus Prolog
* Not in SICstus Prolog:: Yap predicates not available in SICStus Prolog
@end menu

@node Major Differences with SICStus, Fully SICStus Compatible, , SICStus Prolog
@subsection Major Differences between YAP and SICStus Prolog.

Both YAP and SICStus Prolog obey the Edinburgh Syntax and are based on
the WAM. Even so, there are quite a few important differences:

@itemize @bullet
@item Differently from SICStus Prolog, YAP does not have a
notion of interpreted code. All code in YAP is compiled.

@item YAP does not support an intermediate byte-code
representation, so the @code{fcompile/1} and @code{load/1} built-ins are
not available in YAP.

@item YAP implements escape sequences as in the ISO standard. SICStus
Prolog implements Unix-like escape sequences.

@item YAP implements @code{initialization/1} as per the ISO
standard. Use @code{prolog_initialization/1} for the SICStus Prolog
compatible built-in.

@item YAP does not implement the tabbing primitives in
@code{format/2} and @code{format/3}.

@item Prolog flags are different in SICStus Prolog and in YAP.

@item The SICStus Prolog @code{on_exception/3} and
@code{raise_exception} built-ins correspond to the ISO builtins
@code{catch/3} and @code{throw/1}.

@item The following SICStus Prolog v3 built-ins are not (currently)
implemented in YAP (note that this is only a partial list):
@code{call_cleanup/1}, @code{file_search_path/2},
@code{stream_interrupt/3}, @code{reinitialize/0}, @code{help/0},
@code{help/1}, @code{trimcore/0}, @code{load_files/1},
@code{load_files/2}, and @code{require/1}.

      The previous list is incomplete. We also cannot guarantee full
compatibility for other built-ins (although we will try to address any
such incompatibilities). Last, SICStus Prolog is an evolving system, so
one can be expect new incompatibilities to be introduced in future
releases of SICStus Prolog.

@item YAP allows asserting and abolishing static code during
execution through the @code{assert_static/1} and @code{abolish/1}
builtins. This is not allowed in Quintus Prolog or SICStus Prolog.

@item YAP implements rational trees and co-routining but they
are not included by default in the system. You must enable these
extensions when compiling the system.

@item YAP does not currently implement constraints.

@item The socket predicates, although designed to be compatible with
SICStus Prolog, are built-ins, not library predicates, in YAP.

@item This list is incomplete.

@end itemize

The following differences only exist if the @code{language} flag is set
to @code{yap} (the default):

@itemize @bullet
@item The @code{consult/1} predicate in YAP follows C-Prolog
semantics. That is, it adds clauses to the data base, even for
preexisting procedures. This is different from @code{consult/1} in
SICStus Prolog.

@cindex update semantics
@item By default, the data-base in YAP follows "immediate update
semantics", instead of "logical update semantics", as Quintus Prolog or
SICStus Prolog do. The difference is depicted in the next example:

@example
:- dynamic a/1.

?- assert(a(1)).

?- retract(a(X)), X1 is X +1, assertz(a(X)).
@end example
With immediate semantics, new clauses or entries to the data base are
visible in backtracking. In this example, the first call to
@code{retract/1} will succeed. The call to @strong{assertz/1} will then
succeed. On backtracking, the system will retry
@code{retract/1}. Because the newly asserted goal is visible to
@code{retract/1}, it can be retracted from the data base, and
@code{retract(a(X))} will succeed again. The process will continue
generating integers for ever. Immediate semantics were used in C-Prolog.

With logical update semantics, any additions or deletions of clauses
for a goal @emph{will not affect previous activations of the
goal}. In the example, the call to @code{assertz/1} will not see the
update performed by the @code{assertz/1}, and the query will have a
single solution.

Calling @code{yap_flag(update_semantics,logical)} will switch
YAP to use logical update semantics.

@item @code{dynamic/1} is a built-in, not a directive, in YAP.

@item By default, YAP fails on undefined predicates. To follow default
SICStus Prolog use:
@example
:- yap_flag(unknown,error).
@end example

@item By default, directives in YAP can be called from the top level.

@end itemize

@node Fully SICStus Compatible, Not Strictly SICStus Compatible, Major Differences with SICStus, SICStus Prolog
@subsection Yap predicates fully compatible with SICStus Prolog

These are the Prolog built-ins that are fully compatible in both SICStus
Prolog and YAP:

@printindex sy

@node Not Strictly SICStus Compatible, Not in SICstus Prolog, Fully SICStus Compatible, SICStus Prolog
@subsection Yap predicates not strictly compatible with SICStus Prolog

These are YAP built-ins that are also available in SICStus Prolog, but
that are not fully compatible:

@printindex sa

@node Not in SICstus Prolog, , Not Strictly SICStus Compatible, SICStus Prolog
@subsection Yap predicates not available in SICStus Prolog

These are YAP built-ins not available in SICStus Prolog.

@printindex sn


@node ISO Prolog, , SICStus Prolog, Compatibility
@section Compatibility with the ISO Prolog standard

The Prolog standard was developed by ISO/IEC JTC1/SC22/WG17, the
international standardization working group for the programming language
Prolog. The book "Prolog: The Standard" by Deransart, Ed-Dbali and
Cervoni gives a complete description of this standard. Development in
YAP from YAP4.1.6 onwards have striven at making YAP
compatible with ISO Prolog. As such:

@itemize @bullet
@item   YAP now supports all of the built-ins required by the
ISO-standard, and,
@item   Error-handling is as required by the standard.
@end itemize

YAP by default is not fully ISO standard compliant. You can set the 
@code{language} flag to @code{iso} to obtain very good
compatibility. Setting this flag changes the following:

@itemize @bullet
@item By default, YAP uses "immediate update semantics" for its
database, and not "logical update semantics", as per the standard,
(@pxref{SICStus Prolog}). This affects @code{assert/1},
@code{retract/1}, and friends.

Calling @code{set_prolog_flag(update_semantics,logical)} will switch
YAP to use logical update semantics.

@item By default, YAP implements the @code{atom_chars/2}
(@pxref{Testing Terms}), and @code{number_chars/2}, (@pxref{Testing
Terms}), built-ins as per the original Quintus Prolog definition, and
not as per the ISO definition.

Calling @code{set_prolog_flag(to_chars_mode,iso)} will switch
YAP to use the ISO definition for
@code{atom_chars/2} and @code{number_chars/2}.

@item By default, YAP fails on undefined predicates. To follow the ISO
Prolog standard use:
@example
:- set_prolog_flag(unknown,error).
@end example

@item By default, YAP allows executable goals in directives. In ISO mode
most directives can only be called from top level (the exceptions are
@code{set_prolog_flag/2} and @code{op/3}).

@item Error checking for meta-calls under ISO Prolog mode is stricter
than by default.

@item The @code{strict_iso} flag automatically enables the ISO Prolog
standard. This feature should disable all features not present in the
standard.

@end itemize

The following incompatibilities between YAP and the ISO standard are
known to still exist:

@itemize @bullet

@item Currently, YAP does not handle overflow errors in integer
operations, and handles floating-point errors only in some
architectures. Otherwise, YAP follows IEEE arithmetic.

@end itemize

Please inform the authors on other incompatibilities that may still
exist.

@node Operators, Predicate Index, Compatibility, Top
@appendix Summary of Yap Predefined Operators


 The Prolog syntax caters for operators of three main kinds:

@itemize @bullet
@item
prefix;
@item
infix;
@item
postfix.
@end itemize

 Each operator has precedence in the range 1 to 1200, and this 
precedence is used to disambiguate expressions where the structure of the 
term denoted is not made explicit using brackets. The operator of higher 
precedence is the main functor.

 If there are two operators with the highest precedence, the ambiguity 
is solved analyzing the types of the operators. The possible infix types are: 
xfx, xfy, yfx.

 With an operator of type xfx both sub-expressions must have lower 
precedence than the operator itself, unless they are bracketed (which 
assigns to them zero precedence). With an operator type xfy only the  
left-hand sub-expression must have lower precedence. The opposite happens 
for yfx type.

 A prefix operator can be of type fx or fy, and a postfix operator, xf, yf. 
The meaning of the notation is analogous to the above.
@example
a + b * c
@end example
@noindent
means
@example
a + (b * c)
@end example
@noindent
as + and * have the following types and precedences:
@example
:-op(500,yfx,'+').
:-op(400,yfx,'*').
@end example

Now defining
@example
:-op(700,xfy,'++').
:-op(700,xfx,'=:=').
a ++ b =:= c
@end example
@noindent means
@example  
a ++ (b =:= c)
@end example
 

The following is the list of the declarations of the predefined operators:

@example
:-op(1200,fx,['?-', ':-']).
:-op(1200,xfx,[':-','-->']).
:-op(1150,fx,[block,dynamic,mode,public,multifile,meta_predicate,
              sequential,table,initialization]).
:-op(1100,xfy,[';','|']).
:-op(1050,xfy,->).
:-op(1000,xfy,',').
:-op(999,xfy,'.').
:-op(900,fy,['\+', not]).
:-op(900,fx,[nospy, spy]).
:-op(700,xfx,[@@>=,@@=<,@@<,@@>,<,=,>,=:=,=\=,\==,>=,=<,==,\=,=..,is]).
:-op(500,yfx,['\/','/\','+','-']).
:-op(500,fx,['+','-']).
:-op(400,yfx,['<<','>>','//','*','/']).
:-op(300,xfx,mod).
:-op(200,xfy,['^','**']).
:-op(50,xfx,same).
@end example

@node Predicate Index, Concept Index, Operators, Top
@unnumbered Predicate Index
@printindex fn

@node Concept Index, , Predicate Index, Top
@unnumbered Concept Index
@printindex cp

@contents

@bye