1\chapter{Introduction} \label{sec:intro} 2 3This document is a \jargon{reference manual}. That means that it 4documents the system, but it does not explain the basics of the Prolog 5language and it leaves many details of the syntax, semantics and 6built-in primitives undefined where SWI-Prolog follows the standards. 7This manual is intended for people that are familiar with Prolog. For 8those not familiar with Prolog, we recommend to start with a Prolog 9textbook such as \cite{Bratko:86}, \cite{Sterling:86} or 10\cite{Clocksin:87}. For more advanced Prolog usage we recommend 11\cite{Keefe:90}. 12 13 14\section{Positioning SWI-Prolog} \label{sec:swiprolog} 15 16Most implementations of the Prolog language are designed to serve a 17limited set of use cases. SWI-Prolog is no exception to this rule. 18SWI-Prolog positions itself primarily as a Prolog environment for 19`programming in the large' and use cases where it plays a central role 20in an application, i.e., where it acts as `glue' between components. At 21the same time, SWI-Prolog aims at providing a productive rapid 22prototyping environment. Its orientation towards programming in the 23large is backed up by scalability, compiler speed, program structuring 24(modules), support for multithreading to accommodate servers, Unicode 25and interfaces to a large number of document formats, protocols and 26programming languages. Prototyping is facilitated by good development 27tools, both for command line usage and for usage with graphical 28development tools. Demand loading of predicates from the library and a 29`make' facility avoids the \emph{requirement} for using declarations and 30reduces typing. 31 32SWI-Prolog is traditionally strong in education because it is free and 33portable, but also because of its compatibility with textbooks and its 34easy-to-use environment. 35 36Note that these positions do not imply that the system cannot be used 37with other scenarios. SWI-Prolog is used as an embedded language where 38it serves as a small rule subsystem in a large application. It is also 39used as a deductive database. In some cases this is the right choice 40because SWI-Prolog has features that are required in the application, 41such as threading or Unicode support. In general though, for example, 42GNU-Prolog is more suited for embedding because it is small and can 43compile to native code, XSB is better for deductive databases because it 44provides a mature implementation of \jargon{tabling} including support 45for incremental updates and \jargon{Well Founded 46Semantics}\footnote{Sponsored by Kyndi and with help from the XSB 47developers Theresa Swift and David S. Warren, SWI-Prolog now supports 48many of the XSB features.}, and ECLiPSe is better at constraint 49handling. 50 51The syntax and set of built-in predicates is based on the ISO standard 52\cite{stdprolog:98}. Most extensions follow the `Edinburgh tradition' 53(DEC10 Prolog and C-Prolog) and Quintus Prolog \cite{QUINTUS:manual}. 54The infrastructure for constraint programming is based on hProlog 55\cite{Demoen:CW350}. Some libraries are copied from the 56YAP\footnote{\url{http://www.dcc.fc.up.pt/\~{}vsc/Yap/}} system. 57Together with YAP we developed a portability framework (see 58\secref{dialect}). This framework has been filled for SICStus Prolog, 59YAP, IF/Prolog and Ciao. SWI-Prolog version~7 introduces various 60extensions to the Prolog language (see \secref{extensions}). The 61\jargon{string} data type and its supporting set of built-in predicates 62is compatible with ECLiPSe. 63 64\section{Status and releases} \label{sec:status} 65 66This manual describes version \versionshort{} of SWI-Prolog. SWI-Prolog 67is widely considered to be a robust and scalable implementation of the 68Prolog language. It is widely used in education and research. In 69addition, it is in use for $24 \times 7$ mission critical commercial 70server processes. The site \url{http://www.swi-prolog.org} is hosted 71using the SWI-Prolog HTTP server infrastructure. It receives 72approximately 2.3 million hits and serves approximately 300 Gbytes on 73manual data and downloads each month. SWI-Prolog applications range 74from student assignments to commercial applications that count more 75than one million lines of Prolog code. 76 77SWI-Prolog has two development tracks. \emph{Stable} releases have an 78even \emph{minor} version number (e.g., 6.2.1) and are released as a 79branch from the development version when the development 80version is considered stable and there is sufficient new functionality 81to justify a stable release. Stable releases often get a few patch 82updates to deal with installation issues or major flaws. A new 83\emph{Development} version is typically released every couple of weeks 84as a snapshot of the public git repository. `Extra editions' of the 85development version may be released after problems that severely hindered the 86user in their progress have been fixed. 87 88Known bugs that are not likely to be fixed soon are described as 89footnotes in this manual. 90 91 92\section{Should I be using SWI-Prolog?} \label{sec:swiorother} 93 94There are a number of reasons why it might be better to choose a 95commercial, or another free, Prolog system: 96 97\begin{itemlist} 98 \item [SWI-Prolog comes with no warranties] 99Although the developers or the community often provide a work-around or 100a fix for a bug, there is no place you can go to for guaranteed support. 101However, the full source archive is available and can be used to compile 102and debug SWI-Prolog using free tools on all major platforms. 103Users requiring more support should ensure access to knowledgeable developers. 104 105 \item [Performance is your first concern] 106Various free and commercial systems have better performance. But, 107`standard' Prolog benchmarks disregard many factors that are often 108critical to the performance of large applications. SWI-Prolog is not 109good at fast calling of simple predicates, but it is fast with dynamic 110code, meta-calling and predicates that contain large numbers of clauses 111or require more advanced clauses indexing. Many of SWI-Prolog's built-in 112predicates are written in C and have excellent performance. 113\end{itemlist} 114 115On the other hand, SWI-Prolog offers some facilities that are widely 116appreciated by users: 117 118\begin{itemlist} 119 \item [Comprehensive support of Prolog extensions] 120Many modern Prolog implementations extend the standard SLD resolution 121mechanism with which Prolog started and that is described in the ISO 122standard. SWI-Prolog offers most popular extensions. 123 124\jargon{Attributed variables} provide \jargon{Constraint Logic 125Programming} and delayed execution based on instantiation 126(\jargon{coroutining}). \jargon{Tabling} or \jargon{SGL resolution} 127provides characteristics normally associated with \jargon{bottom up 128evaluation}: better termination, better predictable performance by 129avoiding recomputation and Well Founded Semantics for negation. 130\jargon{Delimited continuations} can be used to implement high level new 131control structures and \jargon{Engines} can be used to control multiple 132Prolog goals, achieving different control structures such as massive 133numbers of cooperating agents. 134 135 \item [Nice environment] 136SWI-Prolog provides a good command line environment, including `Do What I 137Mean', autocompletion, history and a tracer that operates on single 138key strokes. The system automatically recompiles modified parts of the 139source code using the make/0 command. The system can be instructed to 140open an arbitrary editor on the right file and line based on its source 141database. It ships with various graphical tools and can be combined with the 142SWI-Prolog editor, PDT (Eclipse plugin for Prolog), VScode or GNU-Emacs. 143 144 \item [Fast compiler] 145Even very large applications can be loaded in seconds on most machines. 146If this is not enough, there is the Quick Load Format. See qcompile/1 147and qsave_program/2. 148 149 \item [Transparent compiled code] 150SWI-Prolog compiled code can be treated just as interpreted code: you 151can list it, trace it, etc. This implies you do not have to decide 152beforehand whether a module should be loaded for debugging or not, and 153the performance of debugged code is close to that of normal operation. 154 155 \item [Source level debugger] 156The source level debugger provides a good overview of your current 157location in the search tree, variable bindings, your source code and 158open choice points. Choice point inspection provides meaningful insight to 159both novices and experienced users. Avoiding unintended choice points 160often provides a huge increase in performance and a huge saving in 161memory usage. 162 163 \item [Profiling] 164SWI-Prolog offers an execution profiler with either textual output or 165graphical output. Finding and improving hotspots in a Prolog program 166may result in huge speedups. 167 168 \item [Flexibility] 169SWI-Prolog can easily be integrated with C, supporting non-determinism 170in Prolog calling C as well as C calling Prolog (see \secref{foreign}). 171It can also be \jargon{embedded} in external programs (see 172\secref{plld}). System predicates can be redefined locally to provide 173compatibility with other Prolog systems. 174 175 \item [Threads] 176Robust support for multiple threads may improve performance and is a 177key enabling factor for deploying Prolog in server applications. Threads 178also facilitates debugging and maintenance of long running processes and 179embedded Prolog engines. The native IDE tools run in a separate thread 180The \pllib{prolog_server} library provides \program{telnet} access and 181the pack \file{libssh} provides SSH login. With some restrictions 182regarding the compatibility of old and new code, code can be replaced 183while it is being executed in another thread. This allows for injecting 184debug/3 statements as well as fixing bugs without downtime. 185 186 \item [Interfaces] 187SWI-Prolog ships with many extension packages that provide robust 188interfaces to processes, encryption, TCP/IP, TIPC, ODBC, SGML/XML/HTML, 189RDF, JSON, YAML, HTTP, graphics and much more. 190\end{itemlist} 191 192 193\section{Support the SWI-Prolog project} \label{sec:sponsor} 194 195You can support the SWI-Prolog project in several ways. Academics are 196invited to cite one of the 197publications\footnote{\url{https://www.swi-prolog.org/Publications.html}} 198on SWI-Prolog. Users can help by identifying and/or fixing problems with 199the code or its 200documentation\footnote{\url{https://www.swi-prolog.org/howto/SubmitPatch.html}}. 201Users can contribute new features or, more lightweight, contribute 202packs\footnote{\url{https://www.swi-prolog.org/pack/list}}. Commercial 203users may consider contacting the 204developers\footnote{\url{mailto:info@swi-prolog.org}} to sponsor the 205development of new features or seek for opportunities to cooperate with 206the developers or other commercial users. 207 208 209\section{Implementation history} \label{sec:implhistory} 210 211 212SWI-Prolog started back in 1986 with the requirement for a Prolog that 213could handle recursive interaction with the C-language: Prolog calling C 214and C calling Prolog recursively. In those days Prolog systems were not 215very aware of their environment and we needed such a system to support 216interactive applications. Since then, SWI-Prolog's development has been 217guided by requests from the user community, especially focussing on (in 218arbitrary order) interaction with the environment, scalability, (I/O) 219performance, standard compliance, teaching and the program development 220environment. 221 222SWI-Prolog is based on a simple Prolog virtual machine called ZIP 223\cite{Bowen:83,Neumerkel:93} which defines only 7 instructions. Prolog 224can easily be compiled into this language, and the abstract machine code 225is easily decompiled back into Prolog. As it is also possible to wire a 226standard 4-port debugger in the virtual machine, there is no need for a 227distinction between compiled and interpreted code. Besides simplifying 228the design of the Prolog system itself, this approach has advantages for 229program development: the compiler is simple and fast, the user does not 230have to decide in advance whether debugging is required, and the system 231only runs slightly slower in debug mode compared to normal execution. 232The price we have to pay is some performance degradation (taking out the 233debugger from the VM interpreter improves performance by about 20\%) and 234somewhat additional memory usage to help the decompiler and debugger. 235 236SWI-Prolog extends the minimal set of instructions described in 237\cite{Bowen:83} to improve performance. While extending this set, care 238has been taken to maintain the advantages of decompilation and tracing 239of compiled code. The extensions include specialised instructions for 240unification, predicate invocation, some frequently used built-in 241predicates, arithmetic, and control (\predref{;}{2}, \predref{|}{2}), 242if-then (\predref{->}{2}) and negation-by-failure (\predref{\+}{1}). 243 244SWI-Prolog implements \jargon{attributed variables} (constraints) and 245\jargon{delimited continuations} following the design in hProlog by Bart 246Demoen. The \jargon{engine} implementation follows the design proposed 247by Paul Tarau. Tabling was implemented by Benoit Desouter based on 248delimited continuations. Tabling has been extended with \jargon{answer 249subsumption} by Fabrizio Riguzzi. The implementation of \jargon{well 250founded semantics} and \jargon{incremental tabling} follows XSB and 251has been sponsored by Kyndi and mode possible by technical support from 252notably Theresa Swift and David S. Warren. 253 254 255\section{Acknowledgements} \label{sec:acknowledge} 256 257Some small parts of the Prolog code of SWI-Prolog are modified 258versions of the corresponding Edinburgh C-Prolog code: grammar rule 259compilation and writef/2. Also some of the C-code originates from 260C-Prolog: finding the path of the currently running executable and 261some of the code underlying absolute_file_name/2. Ideas on programming 262style and techniques originate from C-Prolog and Richard O'Keefe's {\em 263thief} editor. An important source of inspiration are the programming 264techniques introduced by Anjo Anjewierden in PCE version~1 and~2. 265 266Our special thanks go to those who had the fate of using the early 267versions of this system, suggested extensions or reported bugs. Among 268them are Anjo Anjewierden, Huub Knops, Bob Wielinga, Wouter Jansweijer, 269Luc Peerdeman, Eric Nombden, Frank van Harmelen, Bert Rengel. 270 271Martin Jansche (\email{jansche@novell1.gs.uni-heidelberg.de}) has been 272so kind to reorganise the sources for version 2.1.3 of this manual. 273Horst von Brand has been so kind to fix many typos in the 2.7.14 manual. 274Thanks! Randy Sharp fixed many issues in the 6.0.x version of the 275manual. 276 277Bart Demoen and Tom Schrijvers have helped me adding coroutining, 278constraints, global variables and support for cyclic terms to the 279kernel. Tom Schrijvers has provided a first clp(fd) constraint solver, 280the CHR compiler and some of the coroutining predicates. Markus Triska 281contributed the current clp(fd) implementation as well as the clp(b) 282implementation. 283 284Tom Schrijvers and Bart Demoen initiated the implementation of 285\jargon{delimited continuations} (\secref{delcont}), which was used by 286Benoit Desouter and Tom Schrijvers to implement \jargon{tabling} 287(\secref{tabling}) as a library. Fabrizio Riguzzi added a first 288implementation for \jargon{mode directed tabling} 289(\secref{tabling-mode-directed}). 290 291The SWI-Prolog~7 extensions (\secref{extensions}) are the result of a 292long heated discussion on the mailinglist. Nicos Angelopoulos' wish for 293a smooth integration with the R language triggered the overall intend of 294these extensions to enable a smoother integration of Prolog with other 295languages. Michael Hendrix suggested and helped shaping SWI-Prolog 296\jargon{quasi quotations}. 297 298Paul Singleton has integrated Fred Dushin's Java-calls-Prolog side 299with his Prolog-calls-Java side into the current bidirectional JPL 300interface package. 301 302Richard O'Keefe is gratefully acknowledged for his efforts to educate 303beginners as well as valuable comments on proposed new developments. 304 305Scientific Software and Systems Limited, \url{www.sss.co.nz} has 306sponsored the development of the SSL library, unbounded integer and 307rational number arithmetic and many enhancements to the memory 308management of the system. 309 310Leslie de Koninck has made clp(QR) available to SWI-Prolog. 311 312Jeff Rosenwald contributed the TIPC networking library and Google's 313protocol buffer handling. 314 315Paulo Moura's great experience in maintaining Logtalk for many Prolog 316systems including SWI-Prolog has helped in many places fixing 317compatibility issues. He also worked on the MacOS port and fixed many 318typos in the 5.6.9 release of the documentation. 319 320Kyndi (\url{https://kyndi.com/}) sponsored the development of the 321\jargon{engines} interface (\chapref{engines}). The final API was 322established after discussion with the founding father of engines, Paul 323Tarau and Paulo Moura. Kyndi also sponsored JIT indexing on multiple 324arguments as well as \jargon{deep indexing}. Kyndi currently supports 325the implementation of XSB compatible tabling, including well founded 326semantics and incremental tabling. Theresa Swift, David S. Warren 327and Fabrizio Riguzzi provided input to realise advanced tabling. 328