gforth: gforth/doc/gforth.ds

Diff for /gforth/doc/gforth.ds between version 1.15 and 1.78

-version 1.15, Mon Jul 13 13:46:51 1998 UTC
+version 1.78, Tue Aug 22 18:15:38 2000 UTC
  Line 1
  Line 1
  Line 1
  \input texinfo   @c -*-texinfo-*-
  @comment The source is gforth.ds, from which gforth.texi is generated
+ @comment TODO: nac29jan99 - a list of things to add in the next edit:
+ @comment 1. x-ref all ambiguous or implementation-defined features?
+ @comment 2. Describe the use of Auser Avariable AConstant A, etc.
+ @comment 3. words in miscellaneous section need a home.
+ @comment 4. search for TODO for other minor and major works required.
+ @comment 5. [rats] change all @var to @i in Forth source so that info
+ @comment    file looks decent.
+ @c          Not an improvement IMO - anton
+ @c          and anyway, this should be taken up
+ @c          with Karl Berry (the texinfo guy) - anton
+ @comment .. would be useful to have a word that identified all deferred words
+ @comment should semantics stuff in intro be moved to another section
+ @c POSTPONE, COMPILE, [COMPILE], LITERAL should have their own section
  @comment %**start of header (This is for running Texinfo on a region.)
  @setfilename gforth.info
  @settitle Gforth Manual
- Line 7
+ Line 23
  Line 7
  Line 23
  @direntry
  * Gforth: (gforth).             A fast interpreter for the Forth language.
  @end direntry
+ @c The Texinfo manual also recommends doing this, but for Gforth it may
+ @c  not make much sense
+ @c @dircategory Individual utilities
+ @c @direntry
+ @c * Gforth: (gforth)Invoking Gforth.      gforth, gforth-fast, gforthmi
+ @c @end direntry
  @comment @setchapternewpage odd
+ @comment TODO this gets left in by HTML converter
  @macro progstyle {}
  Programming style note:
  @end macro
+ @macro assignment {}
+ @table @i
+ @item Assignment:
+ @end macro
+ @macro endassignment {}
+ @end table
+ @end macro
  @comment %**end of header (This is for running Texinfo on a region.)
+ @comment ----------------------------------------------------------
+ @comment macros for beautifying glossary entries
+ @comment if these are used, need to strip them out for HTML converter
+ @comment else they get repeated verbatim in HTML output.
+ @comment .. not working yet.
+ @macro GLOSS-START {}
+ @iftex
+ @ninerm
+ @end iftex
+ @end macro
+ @macro GLOSS-END {}
+ @iftex
+ @rm
+ @end iftex
+ @end macro
+ @comment ----------------------------------------------------------
  @include version.texi
- @ifinfo
+ @ifnottex
  This file documents Gforth @value{VERSION}
- Copyright @copyright{} 1995-1998 Free Software Foundation, Inc.
+ Copyright @copyright{} 1995--2000 Free Software Foundation, Inc.
       Permission is granted to make and distribute verbatim copies of
       this manual provided the copyright notice and this permission notice
- Line 43
+ Line 98
  Line 43
  Line 98
       except that the sections entitled "Distribution" and "General Public
       License" may be included in a translation approved by the author instead
       of in the original English.
- @end ifinfo
+ @end ifnottex
  @finalout
  @titlepage
- Line 52
+ Line 107
  Line 52
  Line 107
  @sp 2
  @center for version @value{VERSION}
  @sp 2
+ @center Neal Crook
  @center Anton Ertl
  @center Bernd Paysan
  @center Jens Wilke
  @sp 3
- @center This manual is permanently under construction
+ @center This manual is permanently under construction and was last updated on 15-Mar-2000
  @comment  The following two commands start the copyright page.
  @page
  @vskip 0pt plus 1filll
- Copyright @copyright{} 1995--1998 Free Software Foundation, Inc.
+ Copyright @copyright{} 1995--2000 Free Software Foundation, Inc.
  @comment !! Published by ... or You can get a copy of this manual ...
- Line 83
+ Line 139
  Line 83
  Line 139
       of in the original English.
  @end titlepage
  @node Top, License, (dir), (dir)
- @ifinfo
+ @ifnottex
  Gforth is a free implementation of ANS Forth available on many
  personal machines. This manual corresponds to version @value{VERSION}.
- @end ifinfo
+ @end ifnottex
  @menu
- * License::
+ * License::                     The GPL
  * Goals::                       About the Gforth Project
- * Other Books::                 Things you might want to read
+ * Gforth Environment::          Starting (and exiting) Gforth
- * Invoking Gforth::             Starting Gforth
+ * Tutorial::                    Hands-on Forth Tutorial
+ * Introduction::                An introduction to ANS Forth
  * Words::                       Forth words available in Gforth
+ * Error messages::              How to interpret them
  * Tools::                       Programming tools
  * ANS conformance::             Implementation-defined options etc.
+ * Standard vs Extensions::      Should I use extensions?
  * Model::                       The abstract machine of Gforth
  * Integrating Gforth::          Forth as scripting language for applications
  * Emacs and Gforth::            The Gforth Mode
  * Image Files::                 @code{.fi} files contain compiled code
  * Engine::                      The inner interpreter and the primitives
+ * Binding to System Library::
  * Cross Compiler::              The Cross Compiler
  * Bugs::                        How to report them
  * Origin::                      Authors and ancestors of Gforth
+ * Forth-related information::   Books and places to look on the WWW
  * Word Index::                  An item for each Forth word
+ * Name Index::                  Forth words, only names listed
  * Concept Index::               A menu covering many topics
-  --- The Detailed Node Listing ---
+ @detailmenu --- The Detailed Node Listing ---
+ Gforth Environment
+ * Invoking Gforth::             Getting in
+ * Leaving Gforth::              Getting out
+ * Command-line editing::
+ * Environment variables::       that affect how Gforth starts up
+ * Gforth Files::                What gets installed and where
+ * Startup speed::               When 35ms is not fast enough ...
+ Forth Tutorial
+ * Starting Gforth Tutorial::
+ * Syntax Tutorial::
+ * Crash Course Tutorial::
+ * Stack Tutorial::
+ * Arithmetics Tutorial::
+ * Stack Manipulation Tutorial::
+ * Using files for Forth code Tutorial::
+ * Comments Tutorial::
+ * Colon Definitions Tutorial::
+ * Decompilation Tutorial::
+ * Stack-Effect Comments Tutorial::
+ * Types Tutorial::
+ * Factoring Tutorial::
+ * Designing the stack effect Tutorial::
+ * Local Variables Tutorial::
+ * Conditional execution Tutorial::
+ * Flags and Comparisons Tutorial::
+ * General Loops Tutorial::
+ * Counted loops Tutorial::
+ * Recursion Tutorial::
+ * Leaving definitions or loops Tutorial::
+ * Return Stack Tutorial::
+ * Memory Tutorial::
+ * Characters and Strings Tutorial::
+ * Alignment Tutorial::
+ * Interpretation and Compilation Semantics and Immediacy Tutorial::
+ * Execution Tokens Tutorial::
+ * Exceptions Tutorial::
+ * Defining Words Tutorial::
+ * Arrays and Records Tutorial::
+ * POSTPONE Tutorial::
+ * Literal Tutorial::
+ * Advanced macros Tutorial::
+ * Compilation Tokens Tutorial::
+ * Wordlists and Search Order Tutorial::
+ An Introduction to ANS Forth
+ * Introducing the Text Interpreter::
+ * Stacks and Postfix notation::
+ * Your first definition::
+ * How does that work?::
+ * Forth is written in Forth::
+ * Review - elements of a Forth system::
+ * Where to go next::
+ * Exercises::
  Forth Words
  * Notation::
+ * Case insensitivity::
+ * Comments::
+ * Boolean Flags::
  * Arithmetic::
  * Stack Manipulation::
  * Memory::
  * Control Structures::
- * Locals::
  * Defining Words::
- * Structures::
+ * Interpretation and Compilation Semantics::
- * Object-oriented Forth::
  * Tokens for Words::
- * Wordlists::
+ * The Text Interpreter::
+ * Word Lists::
+ * Environmental Queries::
  * Files::
- * Including Files::
  * Blocks::
  * Other I/O::
+ * Locals::
+ * Structures::
+ * Object-oriented Forth::
  * Programming Tools::
  * Assembler and Code Words::
  * Threading Words::
+ * Passing Commands to the OS::
+ * Keeping track of Time::
+ * Miscellaneous Words::
  Arithmetic
  * Single precision::
- * Bitwise operations::
- * Mixed precision::             operations with single and double-cell integers
  * Double precision::            Double-cell integer arithmetic
+ * Bitwise operations::
+ * Numeric comparison::
+ * Mixed precision::             Operations with single and double-cell integers
  * Floating Point::
  Stack Manipulation
- Line 150
+ Line 278
  Line 150
  Line 278
  Memory
+ * Memory model::
+ * Dictionary allocation::
+ * Heap Allocation::
  * Memory Access::
  * Address arithmetic::
  * Memory Blocks::
  Control Structures
- * Selection::
+ * Selection::                   IF ... ELSE ... ENDIF
- * Simple Loops::
+ * Simple Loops::                BEGIN ...
- * Counted Loops::
+ * Counted Loops::               DO
  * Arbitrary control structures::
  * Calls and returns::
  * Exception Handling::
+ Defining Words
+ * CREATE::
+ * Variables::                   Variables and user variables
+ * Constants::
+ * Values::                      Initialised variables
+ * Colon Definitions::
+ * Anonymous Definitions::       Definitions without names
+ * Supplying names::             Passing definition names as strings
+ * User-defined Defining Words::
+ * Deferred words::              Allow forward references
+ * Aliases::
+ User-defined Defining Words
+ * CREATE..DOES> applications::
+ * CREATE..DOES> details::
+ * Advanced does> usage example::
+ Interpretation and Compilation Semantics
+ * Combined words::
+ Tokens for Words
+ * Execution token::             represents execution/interpretation semantics
+ * Compilation token::           represents compilation semantics
+ * Name token::                  represents named words
+ The Text Interpreter
+ * Input Sources::
+ * Number Conversion::
+ * Interpret/Compile states::
+ * Literals::
+ * Interpreter Directives::
+ Word Lists
+ * Vocabularies::
+ * Why use word lists?::
+ * Word list example::
+ Files
+ * Forth source files::
+ * General files::
+ * Search Paths::
+ Search Paths
+ * Source Search Paths::
+ * General Search Paths::
+ Other I/O
+ * Simple numeric output::       Predefined formats
+ * Formatted numeric output::    Formatted (pictured) output
+ * String Formats::              How Forth stores strings in memory
+ * Displaying characters and strings::  Other stuff
+ * Input::                       Input
  Locals
  * Gforth locals::
- Line 172
+ Line 365
  Line 172
  Line 365
  * Where are locals visible by name?::
  * How long do locals live?::
- * Programming Style::
+ * Locals programming style::
- * Implementation::
+ * Locals implementation::
- Defining Words
- * Simple Defining Words::
- * Colon Definitions::
- * User-defined Defining Words::
- * Supplying names::
- * Interpretation and Compilation Semantics::
  Structures
- Line 193
+ Line 378
  Line 193
  Line 378
  Object-oriented Forth
+ * Why object-oriented programming?::
+ * Object-Oriented Terminology::
  * Objects::
  * OOF::
  * Mini-OOF::
+ * Comparison with other object models::
- Objects
+ The @file{objects.fs} model
  * Properties of the Objects model::
- * Why object-oriented programming?::
- * Object-Oriented Terminology::
  * Basic Objects Usage::
- * The class Object::
+ * The Objects base class::
  * Creating objects::
  * Object-Oriented Programming Style::
  * Class Binding::
  * Method conveniences::
  * Classes and Scoping::
+ * Dividing classes::
  * Object Interfaces::
  * Objects Implementation::
- * Comparison with other object models::
  * Objects Glossary::
- OOF
+ The @file{oof.fs} model
  * Properties of the OOF model::
  * Basic OOF Usage::
- * The base class object::
+ * The OOF base class::
  * Class Declaration::
  * Class Implementation::
- Including Files
+ The @file{mini-oof.fs} model
- * Words for Including::
+ * Basic Mini-OOF Usage::
- * Search Path::
+ * Mini-OOF Example::
- * Changing the Search Path::
+ * Mini-OOF Implementation::
- * General Search Paths::
  Programming Tools
+ * Examining::
+ * Forgetting words::
  * Debugging::                   Simple and quick.
  * Assertions::                  Making your programs self-checking.
  * Singlestep Debugger::         Executing your program word by word.
+ Assembler and Code Words
+ * Code and ;code::
+ * Common Assembler::            Assembler Syntax
+ * Common Disassembler::
+ * 386 Assembler::               Deviations and special cases
+ * Alpha Assembler::             Deviations and special cases
+ * MIPS assembler::              Deviations and special cases
+ * Other assemblers::            How to write them
  Tools
  * ANS Report::                  Report the words used, sorted by wordset.
- Line 309
+ Line 506
  Line 309
  Line 506
  Image Files
+ * Image Licensing Issues::      Distribution terms for images.
  * Image File Background::          Why have image files?
  * Non-Relocatable Image Files::    don't always work.
  * Data-Relocatable Image Files::   are better.
  * Fully Relocatable Image Files::  better yet.
  * Stack and Dictionary Sizes::     Setting the default sizes for an image.
- * Running Image Files::            @code{gforth -i @var{file}} or @var{file}.
+ * Running Image Files::         @code{gforth -i @i{file}} or @i{file}.
  * Modifying the Startup Sequence:: and turnkey applications.
  Fully Relocatable Image Files
- Line 341
+ Line 539
  Line 341
  Line 539
  * TOS Optimization::
  * Produced code::
- System Libraries
- * Binding to System Library::
  Cross Compiler
  * Using the Cross Compiler::
  * How the Cross Compiler Works::
+ Other Forth-related information
+ * Internet resources::
+ * Books::
+ * The Forth Interest Group::
+ * Conferences::
+ @end detailmenu
  @end menu
  @node License, Goals, Top, Top
- Line 417
+ Line 619
  Line 417
  Line 619
  @iftex
  @unnumberedsec TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
  @end iftex
- @ifinfo
+ @ifnottex
  @center TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
- @end ifinfo
+ @end ifnottex
  @enumerate 0
  @item
- Line 642
+ Line 844
  Line 642
  Line 844
  @iftex
  @heading NO WARRANTY
  @end iftex
- @ifinfo
+ @ifnottex
  @center NO WARRANTY
- @end ifinfo
+ @end ifnottex
  @item
  BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
- Line 672
+ Line 874
  Line 672
  Line 874
  @iftex
  @heading END OF TERMS AND CONDITIONS
  @end iftex
- @ifinfo
+ @ifnottex
  @center END OF TERMS AND CONDITIONS
- @end ifinfo
+ @end ifnottex
  @page
  @unnumberedsec How to Apply These Terms to Your New Programs
- Line 747
+ Line 949
  Line 747
  Line 949
  @iftex
  @unnumbered Preface
  @cindex Preface
- This manual documents Gforth. The reader is expected to know
+ This manual documents Gforth. Some introductory material is provided for
- Forth. This manual is primarily a reference manual. @xref{Other Books}
+ readers who are unfamiliar with Forth or who are migrating to Gforth
- for introductory material.
+ from other Forth compilers. However, this manual is primarily a
+ reference manual.
  @end iftex
- @node    Goals, Other Books, License, Top
+ @comment TODO much more blurb here.
+ @c ******************************************************************
+ @node Goals, Gforth Environment, License, Top
  @comment node-name,     next,           previous, up
  @chapter Goals of Gforth
- @cindex Goals
+ @cindex goals of the Gforth project
  The goal of the Gforth Project is to develop a standard model for
  ANS Forth. This can be split into several subgoals:
  @itemize @bullet
  @item
- Gforth should conform to the Forth standard (ANS Forth).
+ Gforth should conform to the ANS Forth Standard.
  @item
  It should be a model, i.e. it should define all the
  implementation-dependent things.
- Line 793
+ Line 999
  Line 793
  Line 999
  appears to be quite popular. It has some similarities to and some
  differences from previous models. It has some powerful features, but not
  yet everything that we envisioned. We certainly have achieved our
- execution speed goals (@pxref{Performance}).  It is free and available
+ execution speed goals (@pxref{Performance})@footnote{However, in 1998
- on many machines.
+ the bar was raised when the major commercial Forth vendors switched to
+ native code compilers.}.  It is free and available on many machines.
- @node Other Books, Invoking Gforth, Goals, Top
+ @c ******************************************************************
- @chapter Other books on ANS Forth
+ @node Gforth Environment, Tutorial, Goals, Top
- @cindex books on Forth
+ @chapter Gforth Environment
+ @cindex Gforth environment
- As the standard is relatively new, there are not many books out yet. It
- is not recommended to learn Forth by using Gforth and a book that is
- not written for ANS Forth, as you will not know your mistakes from the
- deviations of the book.
- @cindex standard document for ANS Forth
+ Note: ultimately, the Gforth man page will be auto-generated from the
- @cindex ANS Forth document
+ material in this chapter.
- There is, of course, the standard, the definite reference if you want to
- write ANS Forth programs. It is available in printed form from the
- National Standards Institute Sales Department (Tel.: USA (212) 642-4900;
- Fax.: USA (212) 302-1286) as document @cite{X3.215-1994} for about $200. You
- can also get it from Global Engineering Documents (Tel.: USA (800)
--7179; Fax.: (303) 843-9880) for about $300.
- @cite{dpANS6}, the last draft of the standard, which was then submitted
+ @menu
- to ANSI for publication is available electronically and for free in some
+ * Invoking Gforth::             Getting in
- MS Word format, and it has been converted to HTML (this is my favourite
+ * Leaving Gforth::              Getting out
- format !!url). Some pointers to these versions can be found through
+ * Command-line editing::
- @*@url{http://www.complang.tuwien.ac.at/projects/forth.html}.
+ * Environment variables::       that affect how Gforth starts up
+ * Gforth Files::                What gets installed and where
- @cindex introductory book
+ * Startup speed::               When 35ms is not fast enough ...
- @cindex book, introductory
+ @end menu
- @cindex Woehr, Jack: @cite{Forth: The New Model}
- @cindex @cite{Forth: The new model} (book)
- @cite{Forth: The New Model} by Jack Woehr (Prentice-Hall, 1993) is an
- introductory book based on a draft version of the standard. It does not
- cover the whole standard. It also contains interesting background
- information (Jack Woehr was in the ANS Forth Technical Committee). It is
- not appropriate for complete newbies, but programmers experienced in
- other languages should find it ok.
- !!Conklin, Forth programmer's handbook
+ For related information about the creation of images see @ref{Image Files}.
- @node Invoking Gforth, Words, Other Books, Top
+ @comment ----------------------------------------------
- @chapter Invoking Gforth
+ @node Invoking Gforth, Leaving Gforth, Gforth Environment, Gforth Environment
+ @section Invoking Gforth
  @cindex invoking Gforth
  @cindex running Gforth
  @cindex command-line options
  @cindex options on the command line
  @cindex flags on the command line
- You will usually just say @code{gforth}. In many other cases the default
+ Gforth is made up of two parts; an executable ``engine'' (named
+ @file{gforth} or @file{gforth-fast}) and an image file. To start it, you
+ will usually just say @code{gforth} -- this automatically loads the
+ default image file @file{gforth.fi}. In many other cases the default
  Gforth image will be invoked like this:
  @example
- gforth [files] [-e forth-code]
+ gforth [file | -e forth-code] ...
  @end example
+ @noindent
  This interprets the contents of the files and the Forth code in the order they
  are given.
+ In addition to the @file{gforth} engine, there is also an engine called
+ @file{gforth-fast}, which is faster, but gives less informative error
+ messages (@pxref{Error messages}).
  In general, the command line looks like this:
  @example
- gforth [initialization options] [image-specific options]
+ gforth[-fast] [engine options] [image options]
  @end example
- The initialization options must come before the rest of the command
+ The engine options must come before the rest of the command
  line. They are:
  @table @code
  @cindex -i, command-line option
  @cindex --image-file, command-line option
- @item --image-file @var{file}
+ @item --image-file @i{file}
- @itemx -i @var{file}
+ @itemx -i @i{file}
- Loads the Forth image @var{file} instead of the default
+ Loads the Forth image @i{file} instead of the default
  @file{gforth.fi} (@pxref{Image Files}).
+ @cindex --appl-image, command-line option
+ @item --appl-image @i{file}
+ Loads the image @i{file} and leaves all further command-line arguments
+ to the image (instead of processing them as engine options).  This is
+ useful for building executable application images on Unix, built with
+ @code{gforthmi --application ...}.
  @cindex --path, command-line option
  @cindex -p, command-line option
- @item --path @var{path}
+ @item --path @i{path}
- @itemx -p @var{path}
+ @itemx -p @i{path}
- Uses @var{path} for searching the image file and Forth source code files
+ Uses @i{path} for searching the image file and Forth source code files
  instead of the default in the environment variable @code{GFORTHPATH} or
  the path specified at installation time (e.g.,
  @file{/usr/local/share/gforth/0.2.0:.}). A path is given as a list of
- Line 878
+ Line 1083
  Line 878
  Line 1083
  @cindex --dictionary-size, command-line option
  @cindex -m, command-line option
- @cindex @var{size} parameters for command-line options
+ @cindex @i{size} parameters for command-line options
  @cindex size of the dictionary and the stacks
- @item --dictionary-size @var{size}
+ @item --dictionary-size @i{size}
- @itemx -m @var{size}
+ @itemx -m @i{size}
- Allocate @var{size} space for the Forth dictionary space instead of
+ Allocate @i{size} space for the Forth dictionary space instead of
  using the default specified in the image (typically 256K). The
- @var{size} specification consists of an integer and a unit (e.g.,
+ @i{size} specification for this and subsequent options consists of
+ an integer and a unit (e.g.,
  @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
  size, in this case Cells), @code{k} (kilobytes), @code{M} (Megabytes),
  @code{G} (Gigabytes), and @code{T} (Terabytes). If no unit is specified,
- Line 892
+ Line 1098
  Line 892
  Line 1098
  @cindex --data-stack-size, command-line option
  @cindex -d, command-line option
- @item --data-stack-size @var{size}
+ @item --data-stack-size @i{size}
- @itemx -d @var{size}
+ @itemx -d @i{size}
- Allocate @var{size} space for the data stack instead of using the
+ Allocate @i{size} space for the data stack instead of using the
  default specified in the image (typically 16K).
  @cindex --return-stack-size, command-line option
  @cindex -r, command-line option
- @item --return-stack-size @var{size}
+ @item --return-stack-size @i{size}
- @itemx -r @var{size}
+ @itemx -r @i{size}
- Allocate @var{size} space for the return stack instead of using the
+ Allocate @i{size} space for the return stack instead of using the
  default specified in the image (typically 15K).
  @cindex --fp-stack-size, command-line option
  @cindex -f, command-line option
- @item --fp-stack-size @var{size}
+ @item --fp-stack-size @i{size}
- @itemx -f @var{size}
+ @itemx -f @i{size}
- Allocate @var{size} space for the floating point stack instead of
+ Allocate @i{size} space for the floating point stack instead of
  using the default specified in the image (typically 15.5K). In this case
  the unit specifier @code{e} refers to floating point numbers.
  @cindex --locals-stack-size, command-line option
  @cindex -l, command-line option
- @item --locals-stack-size @var{size}
+ @item --locals-stack-size @i{size}
- @itemx -l @var{size}
+ @itemx -l @i{size}
- Allocate @var{size} space for the locals stack instead of using the
+ Allocate @i{size} space for the locals stack instead of using the
  default specified in the image (typically 14.5K).
  @cindex -h, command-line option
- Line 967
+ Line 1173
  Line 967
  Line 1173
  default image @file{gforth.fi} consist of a sequence of filenames and
  @code{-e @var{forth-code}} options that are interpreted in the sequence
  in which they are given. The @code{-e @var{forth-code}} or
- @code{--evaluate @var{forth-code}} option evaluates the forth
+ @code{--evaluate @var{forth-code}} option evaluates the Forth
  code. This option takes only one argument; if you want to evaluate more
- Forth words, you have to quote them or use several @code{-e}s. To exit
+ Forth words, you have to quote them or use @code{-e} several times. To exit
  after processing the command line (instead of entering interactive mode)
  append @code{-e bye} to the command line.
  @cindex versions, invoking other versions of Gforth
  If you have several versions of Gforth installed, @code{gforth} will
- invoke the version that was installed last. @code{gforth-@var{version}}
+ invoke the version that was installed last. @code{gforth-@i{version}}
- invokes a specific version. You may want to use the option
+ invokes a specific version. If your environment contains the variable
- @code{--path}, if your environment contains the variable
+ @code{GFORTHPATH}, you may want to override it by using the
- @code{GFORTHPATH}.
+ @code{--path} option.
  Not yet implemented:
  On startup the system first executes the system initialization file
  (unless the option @code{--no-init-file} is given; note that the system
  resulting from using this option may not be ANS Forth conformant). Then
  the user initialization file @file{.gforth.fs} is executed, unless the
- option @code{--no-rc} is given; this file is first searched in @file{.},
+ option @code{--no-rc} is given; this file is searched for in @file{.},
  then in @file{~}, then in the normal path (see above).
- @node Words, Tools, Invoking Gforth, Top
- @chapter Forth Words
- @cindex Words
- @menu
- * Notation::
- * Arithmetic::
- * Stack Manipulation::
- * Memory::
- * Control Structures::
- * Locals::
- * Defining Words::
- * Structures::
- * Object-oriented Forth::
- * Tokens for Words::
- * Wordlists::
- * Files::
- * Including Files::
- * Blocks::
- * Other I/O::
- * Programming Tools::
- * Assembler and Code Words::
- * Threading Words::
- @end menu
- @node Notation, Arithmetic, Words, Words
- @section Notation
- @cindex notation of glossary entries
- @cindex format of glossary entries
- @cindex glossary notation format
- @cindex word glossary entry format
- The Forth words are described in this section in the glossary notation
+ @comment ----------------------------------------------
- that has become a de-facto standard for Forth texts, i.e.,
+ @node Leaving Gforth, Command-line editing, Invoking Gforth, Gforth Environment
+ @section Leaving Gforth
+ @cindex Gforth - leaving
+ @cindex leaving Gforth
+ You can leave Gforth by typing @code{bye} or @kbd{Ctrl-d} (at the start
+ of a line) or (if you invoked Gforth with the @code{--die-on-signal}
+ option) @kbd{Ctrl-c}. When you leave Gforth, all of your definitions and
+ data are discarded.  For ways of saving the state of the system before
+ leaving Gforth see @ref{Image Files}.
+ doc-bye
+ @comment ----------------------------------------------
+ @node Command-line editing, Environment variables, Leaving Gforth, Gforth Environment
+ @section Command-line editing
+ @cindex command-line editing
+ Gforth maintains a history file that records every line that you type to
+ the text interpreter. This file is preserved between sessions, and is
+ used to provide a command-line recall facility; if you type @kbd{Ctrl-P}
+ repeatedly you can recall successively older commands from this (or
+ previous) session(s). The full list of command-line editing facilities is:
- @format
+ @itemize @bullet
- @var{word}     @var{Stack effect}   @var{wordset}   @var{pronunciation}
+ @item
- @end format
+ @kbd{Ctrl-p} (``previous'') (or up-arrow) to recall successively older
- @var{Description}
+ commands from the history buffer.
+ @item
+ @kbd{Ctrl-n} (``next'') (or down-arrow) to recall successively newer commands
+ from the history buffer.
+ @item
+ @kbd{Ctrl-f} (or right-arrow) to move the cursor right, non-destructively.
+ @item
+ @kbd{Ctrl-b} (or left-arrow) to move the cursor left, non-destructively.
+ @item
+ @kbd{Ctrl-h} (backspace) to delete the character to the left of the cursor,
+ closing up the line.
+ @item
+ @kbd{Ctrl-k} to delete (``kill'') from the cursor to the end of the line.
+ @item
+ @kbd{Ctrl-a} to move the cursor to the start of the line.
+ @item
+ @kbd{Ctrl-e} to move the cursor to the end of the line.
+ @item
+ @key{RET} (@kbd{Ctrl-m}) or @key{LFD} (@kbd{Ctrl-j}) to submit the current
+ line.
+ @item
+ @key{TAB} to step through all possible full-word completions of the word
+ currently being typed.
+ @item
+ @kbd{Ctrl-d} on an empty line line to terminate Gforth (gracefully,
+ using @code{bye}).
+ @item
+ @kbd{Ctrl-x} (or @code{Ctrl-d} on a non-empty line) to delete the
+ character under the cursor.
+ @end itemize
- @table @var
+ When editing, displayable characters are inserted to the left of the
- @item word
+ cursor position; the line is always in ``insert'' (as opposed to
- @cindex case insensitivity
+ ``overstrike'') mode.
- The name of the word. BTW, Gforth is case insensitive, so you can
- type the words in in lower case (However, @pxref{core-idef}).
- @item Stack effect
+ @cindex history file
- @cindex stack effect
+ @cindex @file{.gforth-history}
- The stack effect is written in the notation @code{@var{before} --
+ On Unix systems, the history file is @file{~/.gforth-history} by
- @var{after}}, where @var{before} and @var{after} describe the top of
+ default@footnote{i.e. it is stored in the user's home directory.}. You
- stack entries before and after the execution of the word. The rest of
+ can find out the name and location of your history file using:
- the stack is not touched by the word. The top of stack is rightmost,
- i.e., a stack sequence is written as it is typed in. Note that Gforth
- uses a separate floating point stack, but a unified stack
- notation. Also, return stack effects are not shown in @var{stack
- effect}, but in @var{Description}. The name of a stack item describes
- the type and/or the function of the item. See below for a discussion of
- the types.
- All words have two stack effects: A compile-time stack effect and a
+ @example
- run-time stack effect. The compile-time stack-effect of most words is
+ history-file type \ Unix-class systems
- @var{ -- }. If the compile-time stack-effect of a word deviates from
- this standard behaviour, or the word does other unusual things at
- compile time, both stack effects are shown; otherwise only the run-time
- stack effect is shown.
- @cindex pronounciation of words
+ history-file type \ Other systems
- @item pronunciation
+ history-dir  type
- How the word is pronounced.
+ @end example
- @cindex wordset
+ If you enter long definitions by hand, you can use a text editor to
- @item wordset
+ paste them out of the history file into a Forth source file for reuse at
- The ANS Forth standard is divided into several wordsets. A standard
+ a later time.
- system need not support all of them. So, the fewer wordsets your program
- uses the more portable it will be in theory. However, we suspect that
- most ANS Forth systems on personal machines will feature all
- wordsets. Words that are not defined in the ANS standard have
- @code{gforth} or @code{gforth-internal} as wordset. @code{gforth}
- describes words that will work in future releases of Gforth;
- @code{gforth-internal} words are more volatile. Environmental query
- strings are also displayed like words; you can recognize them by the
- @code{environment} in the wordset field.
- @item Description
+ Gforth never trims the size of the history file, so you should do this
- A description of the behaviour of the word.
+ periodically, if necessary.
- @end table
- @cindex types of stack items
+ @comment this is all defined in history.fs
- @cindex stack item types
+ @comment NAC TODO the ctrl-D behaviour can either do a bye or a beep.. how is that option
- The type of a stack item is specified by the character(s) the name
+ @comment chosen?
- starts with:
- @table @code
- @item f
- @cindex @code{f}, stack item type
- Boolean flags, i.e. @code{false} or @code{true}.
- @item c
- @cindex @code{c}, stack item type
- Char
- @item w
- @cindex @code{w}, stack item type
- Cell, can contain an integer or an address
- @item n
- @cindex @code{n}, stack item type
- signed integer
- @item u
- @cindex @code{u}, stack item type
- unsigned integer
- @item d
- @cindex @code{d}, stack item type
- double sized signed integer
- @item ud
- @cindex @code{ud}, stack item type
- double sized unsigned integer
- @item r
- @cindex @code{r}, stack item type
- Float (on the FP stack)
- @item a_
- @cindex @code{a_}, stack item type
- Cell-aligned address
- @item c_
- @cindex @code{c_}, stack item type
- Char-aligned address (note that a Char may have two bytes in Windows NT)
- @item f_
- @cindex @code{f_}, stack item type
- Float-aligned address
- @item df_
- @cindex @code{df_}, stack item type
- Address aligned for IEEE double precision float
- @item sf_
- @cindex @code{sf_}, stack item type
- Address aligned for IEEE single precision float
- @item xt
- @cindex @code{xt}, stack item type
- Execution token, same size as Cell
- @item wid
- @cindex @code{wid}, stack item type
- Wordlist ID, same size as Cell
- @item f83name
- @cindex @code{f83name}, stack item type
- Pointer to a name structure
- @item "
- @cindex @code{"}, stack item type
- string in the input stream (not on the stack). The terminating character
- is a blank by default. If it is not a blank, it is shown in @code{<>}
- quotes.
- @end table
- @node Arithmetic, Stack Manipulation, Notation, Words
+ @comment ----------------------------------------------
- @section Arithmetic
+ @node Environment variables, Gforth Files, Command-line editing, Gforth Environment
- @cindex arithmetic words
+ @section Environment variables
+ @cindex environment variables
- @cindex division with potentially negative operands
+ Gforth uses these environment variables:
- Forth arithmetic is not checked, i.e., you will not hear about integer
- overflow on addition or multiplication, you may hear about division by
- zero if you are lucky. The operator is written after the operands, but
- the operands are still in the original order. I.e., the infix @code{2-1}
- corresponds to @code{2 1 -}. Forth offers a variety of division
- operators. If you perform division with potentially negative operands,
- you do not want to use @code{/} or @code{/mod} with its undefined
- behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
- former, @pxref{Mixed precision}).
- @menu
+ @itemize @bullet
- * Single precision::
+ @item
- * Bitwise operations::
+ @cindex @code{GFORTHHIST} -- environment variable
- * Mixed precision::             operations with single and double-cell integers
+ @code{GFORTHHIST} -- (Unix systems only) specifies the directory in which to
- * Double precision::            Double-cell integer arithmetic
+ open/create the history file, @file{.gforth-history}. Default:
- * Floating Point::
+ @code{$HOME}.
- @end menu
- @node Single precision, Bitwise operations, Arithmetic, Arithmetic
+ @item
- @subsection Single precision
+ @cindex @code{GFORTHPATH} -- environment variable
- @cindex single precision arithmetic words
+ @code{GFORTHPATH} -- specifies the path used when searching for the gforth image file and
+ for Forth source-code files.
- doc-+
+ @item
- doc--
+ @cindex @code{GFORTH} -- environment variable
- doc-*
+ @code{GFORTH} -- used by @file{gforthmi}, @xref{gforthmi}.
- doc-/
- doc-mod
- doc-/mod
- doc-negate
- doc-abs
- doc-min
- doc-max
- @node Bitwise operations, Mixed precision, Single precision, Arithmetic
+ @item
- @subsection Bitwise operations
+ @cindex @code{GFORTHD} -- environment variable
- @cindex bitwise operation words
+ @code{GFORTHD} -- used by @file{gforthmi}, @xref{gforthmi}.
- doc-and
+ @item
- doc-or
+ @cindex @code{TMP}, @code{TEMP} - environment variable
- doc-xor
+ @code{TMP}, @code{TEMP} - (non-Unix systems only) used as a potential
- doc-invert
+ location for the history file.
- doc-2*
+ @end itemize
- doc-2/
- @node Mixed precision, Double precision, Bitwise operations, Arithmetic
- @subsection Mixed precision
- @cindex mixed precision arithmetic words
- doc-m+
- doc-*/
- doc-*/mod
- doc-m*
- doc-um*
- doc-m*/
- doc-um/mod
- doc-fm/mod
- doc-sm/rem
- @node Double precision, Floating Point, Mixed precision, Arithmetic
+ @comment also POSIXELY_CORRECT LINES COLUMNS HOME but no interest in
- @subsection Double precision
+ @comment mentioning these.
- @cindex double precision arithmetic words
- @cindex double-cell numbers, input format
+ All the Gforth environment variables default to sensible values if they
- @cindex input format for double-cell numbers
+ are not set.
- The outer (aka text) interpreter converts numbers containing a dot into
- a double precision number. Note that only numbers with the dot as last
- character are standard-conforming.
- doc-d+
- doc-d-
- doc-dnegate
- doc-dabs
- doc-dmin
- doc-dmax
- @node Floating Point,  , Double precision, Arithmetic
+ @comment ----------------------------------------------
- @subsection Floating Point
+ @node Gforth Files, Startup speed, Environment variables, Gforth Environment
- @cindex floating point arithmetic words
+ @section Gforth files
+ @cindex Gforth files
- @cindex floating-point numbers, input format
+ When you install Gforth on a Unix system, it installs files in these
- @cindex input format for floating-point numbers
+ locations by default:
- The format of floating point numbers recognized by the outer (aka text)
- interpreter is: a signed decimal number, possibly containing a decimal
- point (@code{.}), followed by @code{E} or @code{e}, optionally followed
- by a signed integer (the exponent). E.g., @code{1e} is the same as
- @code{+1.0e+0}. Note that a number without @code{e} is not interpreted
- as floating-point number, but as double (if the number contains a
- @code{.}) or single precision integer. Also, conversions between string
- and floating point numbers always use base 10, irrespective of the value
- of @code{BASE} (in Gforth; for the standard this is an ambiguous
- condition). If @code{BASE} contains a value greater then 14, the
- @code{E} may be interpreted as digit and the number will be interpreted
- as integer, unless it has a signed exponent (both @code{+} and @code{-}
- are allowed as signs).
- @cindex angles in trigonometric operations
+ @itemize @bullet
- @cindex trigonometric operations
+ @item
- Angles in floating point operations are given in radians (a full circle
+ @file{/usr/local/bin/gforth}
- has 2 pi radians). Note, that Gforth has a separate floating point
+ @item
- stack, but we use the unified notation.
+ @file{/usr/local/bin/gforthmi}
+ @item
+ @file{/usr/local/man/man1/gforth.1} - man page.
+ @item
+ @file{/usr/local/info} - the Info version of this manual.
+ @item
+ @file{/usr/local/lib/gforth/<version>/...} - Gforth @file{.fi} files.
+ @item
+ @file{/usr/local/share/gforth/<version>/TAGS} - Emacs TAGS file.
+ @item
+ @file{/usr/local/share/gforth/<version>/...} - Gforth source files.
+ @item
+ @file{.../emacs/site-lisp/gforth.el} - Emacs gforth mode.
+ @end itemize
- @cindex floating-point arithmetic, pitfalls
+ You can select different places for installation by using
- Floating point numbers have a number of unpleasant surprises for the
+ @code{configure} options (listed with @code{configure --help}).
- unwary (e.g., floating point addition is not associative) and even a few
- for the wary. You should not use them unless you know what you are doing
- or you don't care that the results you get are totally bogus. If you
- want to learn about the problems of floating point numbers (and how to
- avoid them), you might start with @cite{David Goldberg, What Every
- Computer Scientist Should Know About Floating-Point Arithmetic, ACM
- Computing Surveys 23(1):5@minus{}48, March 1991}.
- doc-f+
+ @comment ----------------------------------------------
- doc-f-
+ @node Startup speed,  , Gforth Files, Gforth Environment
- doc-f*
+ @section Startup speed
- doc-f/
+ @cindex Startup speed
- doc-fnegate
+ @cindex speed, startup
- doc-fabs
- doc-fmax
+ If Gforth is used for CGI scripts or in shell scripts, its startup
- doc-fmin
+ speed may become a problem.  On a 300MHz 21064a under Linux-2.2.13 with
- doc-floor
+ glibc-2.0.7, @code{gforth -e bye} takes about 24.6ms user and 11.3ms
- doc-fround
+ system time.
- doc-f**
- doc-fsqrt
+ If startup speed is a problem, you may consider the following ways to
- doc-fexp
+ improve it; or you may consider ways to reduce the number of startups
- doc-fexpm1
+ (for example, by using Fast-CGI).
- doc-fln
- doc-flnp1
+ The first step to improve startup speed is to statically link Gforth, by
- doc-flog
+ building it with @code{XLDFLAGS=-static}.  This requires more memory for
- doc-falog
+ the code and will therefore slow down the first invocation, but
- doc-fsin
+ subsequent invocations avoid the dynamic linking overhead.  Another
- doc-fcos
+ disadvantage is that Gforth won't profit from library upgrades.  As a
- doc-fsincos
+ result, @code{gforth-static -e bye} takes about 17.1ms user and
- doc-ftan
+.2ms system time.
- doc-fasin
- doc-facos
+ The next step to improve startup speed is to use a non-relocatable image
- doc-fatan
+ (@pxref{Non-Relocatable Image Files}).  You can create this image with
- doc-fatan2
+ @code{gforth -e "savesystem gforthnr.fi bye"} and later use it with
- doc-fsinh
+ @code{gforth -i gforthnr.fi ...}.  This avoids the relocation overhead
- doc-fcosh
+ and a part of the copy-on-write overhead.  The disadvantage is that the
- doc-ftanh
+ non-relocatable image does not work if the OS gives Gforth a different
- doc-fasinh
+ address for the dictionary, for whatever reason; so you better provide a
- doc-facosh
+ fallback on a relocatable image.  @code{gforth-static -i gforthnr.fi -e
- doc-fatanh
+ bye} takes about 15.3ms user and 7.5ms system time.
+ The final step is to disable dictionary hashing in Gforth.  Gforth
+ builds the hash table on startup, which takes much of the startup
+ overhead. You can do this by commenting out the @code{include hash.fs}
+ in @file{startup.fs} and everything that requires @file{hash.fs} (at the
+ moment @file{table.fs} and @file{ekey.fs}) and then doing @code{make}.
+ The disadvantages are that functionality like @code{table} and
+ @code{ekey} is missing and that text interpretation (e.g., compiling)
+ now takes much longer. So, you should only use this method if there is
+ no significant text interpretation to perform (the script should be
+ compiled into the image, amongst other things).  @code{gforth-static -i
+ gforthnrnh.fi -e bye} takes about 2.1ms user and 6.1ms system time.
- @node Stack Manipulation, Memory, Arithmetic, Words
+ @c ******************************************************************
- @section Stack Manipulation
+ @node Tutorial, Introduction, Gforth Environment, Top
- @cindex stack manipulation words
+ @chapter Forth Tutorial
+ @cindex Tutorial
+ @cindex Forth Tutorial
+ @c Topics from nac's Introduction that could be mentioned:
+ @c press <ret> after each line
+ @c Prompt
+ @c numbers vs. words in dictionary on text interpretation
+ @c what happens on redefinition
+ @c parsing words (in particular, defining words)
+ This tutorial can be used with any ANS-compliant Forth; any
+ Gforth-specific features are marked as such and you can skip them if you
+ work with another Forth.  This tutorial does not explain all features of
+ Forth, just enough to get you started and give you some ideas about the
+ facilities available in Forth.  Read the rest of the manual and the
+ standard when you are through this.
+ The intended way to use this tutorial is that you work through it while
+ sitting in front of the console, take a look at the examples and predict
+ what they will do, then try them out; if the outcome is not as expected,
+ find out why (e.g., by trying out variations of the example), so you
+ understand what's going on.  There are also some assignments that you
+ should solve.
- @cindex floating-point stack in the standard
+ This tutorial assumes that you have programmed before and know what,
- Gforth has a data stack (aka parameter stack) for characters, cells,
+ e.g., a loop is.
- addresses, and double cells, a floating point stack for floating point
- numbers, a return stack for storing the return addresses of colon
- definitions and other data, and a locals stack for storing local
- variables. Note that while every sane Forth has a separate floating
- point stack, this is not strictly required; an ANS Forth system could
- theoretically keep floating point numbers on the data stack. As an
- additional difficulty, you don't know how many cells a floating point
- number takes. It is reportedly possible to write words in a way that
- they work also for a unified stack model, but we do not recommend trying
- it. Instead, just say that your program has an environmental dependency
- on a separate FP stack.
- @cindex return stack and locals
+ @c !! explain compat library
- @cindex locals and return stack
- Also, a Forth system is allowed to keep the local variables on the
- return stack. This is reasonable, as local variables usually eliminate
- the need to use the return stack explicitly. So, if you want to produce
- a standard complying program and if you are using local variables in a
- word, forget about return stack manipulations in that word (see the
- standard document for the exact rules).
  @menu
- * Data stack::
+ * Starting Gforth Tutorial::
- * Floating point stack::
+ * Syntax Tutorial::
- * Return stack::
+ * Crash Course Tutorial::
- * Locals stack::
+ * Stack Tutorial::
- * Stack pointer manipulation::
+ * Arithmetics Tutorial::
+ * Stack Manipulation Tutorial::
+ * Using files for Forth code Tutorial::
+ * Comments Tutorial::
+ * Colon Definitions Tutorial::
+ * Decompilation Tutorial::
+ * Stack-Effect Comments Tutorial::
+ * Types Tutorial::
+ * Factoring Tutorial::
+ * Designing the stack effect Tutorial::
+ * Local Variables Tutorial::
+ * Conditional execution Tutorial::
+ * Flags and Comparisons Tutorial::
+ * General Loops Tutorial::
+ * Counted loops Tutorial::
+ * Recursion Tutorial::
+ * Leaving definitions or loops Tutorial::
+ * Return Stack Tutorial::
+ * Memory Tutorial::
+ * Characters and Strings Tutorial::
+ * Alignment Tutorial::
+ * Interpretation and Compilation Semantics and Immediacy Tutorial::
+ * Execution Tokens Tutorial::
+ * Exceptions Tutorial::
+ * Defining Words Tutorial::
+ * Arrays and Records Tutorial::
+ * POSTPONE Tutorial::
+ * Literal Tutorial::
+ * Advanced macros Tutorial::
+ * Compilation Tokens Tutorial::
+ * Wordlists and Search Order Tutorial::
  @end menu
- @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
+ @node Starting Gforth Tutorial, Syntax Tutorial, Tutorial, Tutorial
- @subsection Data stack
+ @section Starting Gforth
- @cindex data stack manipulation words
+ @cindex starting Gforth tutorial
- @cindex stack manipulations words, data stack
+ You can start Gforth by typing its name:
- doc-drop
+ @example
- doc-nip
+ gforth
- doc-dup
+ @end example
- doc-over
- doc-tuck
- doc-swap
- doc-rot
- doc--rot
- doc-?dup
- doc-pick
- doc-roll
- doc-2drop
- doc-2nip
- doc-2dup
- doc-2over
- doc-2tuck
- doc-2swap
- doc-2rot
- @node Floating point stack, Return stack, Data stack, Stack Manipulation
+ That puts you into interactive mode; you can leave Gforth by typing
- @subsection Floating point stack
+ @code{bye}.  While in Gforth, you can edit the command line and access
- @cindex floating-point stack manipulation words
+ the command line history with cursor keys, similar to bash.
- @cindex stack manipulation words, floating-point stack
- doc-fdrop
- doc-fnip
- doc-fdup
- doc-fover
- doc-ftuck
- doc-fswap
- doc-frot
- @node Return stack, Locals stack, Floating point stack, Stack Manipulation
+ @node Syntax Tutorial, Crash Course Tutorial, Starting Gforth Tutorial, Tutorial
- @subsection Return stack
+ @section Syntax
- @cindex return stack manipulation words
+ @cindex syntax tutorial
- @cindex stack manipulation words, return stack
- doc->r
+ A @dfn{word} is a sequence of arbitrary characters (expcept white
- doc-r>
+ space).  Words are separated by white space.  E.g., each of the
- doc-r@
+ following lines contains exactly one word:
- doc-rdrop
- doc-2>r
- doc-2r>
- doc-2r@
- doc-2rdrop
- @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
+ @example
- @subsection Locals stack
+ word
+ !@@#$%^&*()
+ 1234567890
+!a
+ @end example
- @node Stack pointer manipulation,  , Locals stack, Stack Manipulation
+ A frequent beginner's error is to leave away necessary white space,
- @subsection Stack pointer manipulation
+ resulting in an error like @samp{Undefined word}; so if you see such an
- @cindex stack pointer manipulation words
+ error, check if you have put spaces wherever necessary.
- doc-sp@
+ @example
- doc-sp!
+ ." hello, world" \ correct
- doc-fp@
+ ."hello, world"  \ gives an "Undefined word" error
- doc-fp!
+ @end example
- doc-rp@
- doc-rp!
- doc-lp@
- doc-lp!
- @node Memory, Control Structures, Stack Manipulation, Words
+ Gforth and most other Forth systems ignore differences in case (they are
- @section Memory
+ case-insensitive), i.e., @samp{word} is the same as @samp{Word}.  If
- @cindex Memory words
+ your system is case-sensitive, you may have to type all the examples
+ given here in upper case.
- @menu
- * Memory Access::
- * Address arithmetic::
- * Memory Blocks::
- @end menu
- @node Memory Access, Address arithmetic, Memory, Memory
+ @node Crash Course Tutorial, Stack Tutorial, Syntax Tutorial, Tutorial
- @subsection Memory Access
+ @section Crash Course
- @cindex memory access words
- doc-@
+ Type
- doc-!
- doc-+!
- doc-c@
- doc-c!
- doc-2@
- doc-2!
- doc-f@
- doc-f!
- doc-sf@
- doc-sf!
- doc-df@
- doc-df!
- @node Address arithmetic, Memory Blocks, Memory Access, Memory
+ @example
- @subsection Address arithmetic
+0 !
- @cindex address arithmetic words
+ here execute
+ ' catch >body 20 erase abort
+ ' (quit) >body 20 erase
+ @end example
- ANS Forth does not specify the sizes of the data types. Instead, it
+ The last two examples are guaranteed to destroy parts of Gforth (and
- offers a number of words for computing sizes and doing address
+ most other systems), so you better leave Gforth afterwards (if it has
- arithmetic. Basically, address arithmetic is performed in terms of
+ not finished by itself).  On some systems you may have to kill gforth
- address units (aus); on most systems the address unit is one byte. Note
+ from outside (e.g., in Unix with @code{kill}).
- that a character may have more than one au, so @code{chars} is no noop
- (on systems where it is a noop, it compiles to nothing).
- @cindex alignment of addresses for types
+ Now that you know how to produce crashes (and that there's not much to
- ANS Forth also defines words for aligning addresses for specific
+ them), let's learn how to produce meaningful programs.
- types. Many computers require that accesses to specific data types
- must only occur at specific addresses; e.g., that cells may only be
- accessed at addresses divisible by 4. Even if a machine allows unaligned
- accesses, it can usually perform aligned accesses faster.
- For the performance-conscious: alignment operations are usually only
- necessary during the definition of a data structure, not during the
- (more frequent) accesses to it.
- ANS Forth defines no words for character-aligning addresses. This is not
+ @node Stack Tutorial, Arithmetics Tutorial, Crash Course Tutorial, Tutorial
- an oversight, but reflects the fact that addresses that are not
+ @section Stack
- char-aligned have no use in the standard and therefore will not be
+ @cindex stack tutorial
- created.
- @cindex @code{CREATE} and alignment
+ The most obvious feature of Forth is the stack.  When you type in a
- The standard guarantees that addresses returned by @code{CREATE}d words
+ number, it is pushed on the stack.  You can display the content of the
- are cell-aligned; in addition, Gforth guarantees that these addresses
+ stack with @code{.s}.
- are aligned for all purposes.
- Note that the standard defines a word @code{char}, which has nothing to
+ @example
- do with address arithmetic.
+2 .s
+.s
+ @end example
- doc-chars
+ @code{.s} displays the top-of-stack to the right, i.e., the numbers
- doc-char+
+ appear in @code{.s} output as they appeared in the input.
- doc-cells
- doc-cell+
- doc-cell
- doc-align
- doc-aligned
- doc-floats
- doc-float+
- doc-float
- doc-falign
- doc-faligned
- doc-sfloats
- doc-sfloat+
- doc-sfalign
- doc-sfaligned
- doc-dfloats
- doc-dfloat+
- doc-dfalign
- doc-dfaligned
- doc-maxalign
- doc-maxaligned
- doc-cfalign
- doc-cfaligned
- doc-address-unit-bits
- @node Memory Blocks,  , Address arithmetic, Memory
+ You can print the top of stack element with @code{.}.
- @subsection Memory Blocks
- @cindex memory block words
- doc-move
+ @example
- doc-erase
+2 3 . . .
+ @end example
- While the previous words work on address units, the rest works on
+ In general, words consume their stack arguments (@code{.s} is an
- characters.
+ exception).
- doc-cmove
+ @assignment
- doc-cmove>
+ What does the stack contain after @code{5 6 7 .}?
- doc-fill
+ @endassignment
- doc-blank
- @node Control Structures, Locals, Memory, Words
- @section Control Structures
- @cindex control structures
- Control structures in Forth cannot be used in interpret state, only in
+ @node Arithmetics Tutorial, Stack Manipulation Tutorial, Stack Tutorial, Tutorial
- compile state@footnote{More precisely, they have no interpretation
+ @section Arithmetics
- semantics (@pxref{Interpretation and Compilation Semantics})}, i.e., in
+ @cindex arithmetics tutorial
- a colon definition. We do not like this limitation, but have not seen a
- satisfying way around it yet, although many schemes have been proposed.
- @menu
+ The words @code{+}, @code{-}, @code{*}, @code{/}, and @code{mod} always
- * Selection::
+ operate on the top two stack items:
- * Simple Loops::
- * Counted Loops::
- * Arbitrary control structures::
- * Calls and returns::
- * Exception Handling::
- @end menu
- @node Selection, Simple Loops, Control Structures, Control Structures
- @subsection Selection
- @cindex selection control structures
- @cindex control structures for selection
- @cindex @code{IF} control structure
- @example
- @var{flag}
- IF
-   @var{code}
- ENDIF
- @end example
- or
  @example
- @var{flag}
+2 .s
- IF
+ + .s
-   @var{code1}
+ .
- ELSE
+1 - .
-   @var{code2}
+3 mod .
- ENDIF
  @end example
- You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
+ The operands of @code{-}, @code{/}, and @code{mod} are in the same order
- standard, and @code{ENDIF} is not, although it is quite popular. We
+ as in the corresponding infix expression (this is generally the case in
- recommend using @code{ENDIF}, because it is less confusing for people
+ Forth).
- who also know other languages (and is not prone to reinforcing negative
- prejudices against Forth in these people). Adding @code{ENDIF} to a
+ Parentheses are superfluous (and not available), because the order of
- system that only supplies @code{THEN} is simple:
+ the words unambiguously determines the order of evaluation and the
+ operands:
  @example
- : endif   POSTPONE then ; immediate
+4 + 5 * .
+4 5 * + .
  @end example
- [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
+ @assignment
- (adv.)}  has the following meanings:
+ What are the infix expressions corresponding to the Forth code above?
- @quotation
+ Write @code{6-7*8+9} in Forth notation@footnote{This notation is also
- ... 2b: following next after in order ... 3d: as a necessary consequence
+ known as Postfix or RPN (Reverse Polish Notation).}.
- (if you were there, then you saw them).
+ @endassignment
- @end quotation
- Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
- and many other programming languages has the meaning 3d.]
- Gforth also provides the words @code{?dup-if} and @code{?dup-0=-if}, so
+ To change the sign, use @code{negate}:
- you can avoid using @code{?dup}. Using these alternatives is also more
- efficient than using @code{?dup}. Definitions in plain standard Forth
- for @code{ENDIF}, @code{?DUP-IF} and @code{?DUP-0=-IF} are provided in
- @file{compat/control.fs}.
- @cindex @code{CASE} control structure
  @example
- @var{n}
+negate .
- CASE
-   @var{n1} OF @var{code1} ENDOF
-   @var{n2} OF @var{code2} ENDOF
-   @dots{}
- ENDCASE
  @end example
- Executes the first @var{codei}, where the @var{ni} is equal to
+ @assignment
- @var{n}. A default case can be added by simply writing the code after
+ Convert -(-3)*4-5 to Forth.
- the last @code{ENDOF}. It may use @var{n}, which is on top of the stack,
+ @endassignment
- but must not consume it.
- @node Simple Loops, Counted Loops, Selection, Control Structures
+ @code{/mod} performs both @code{/} and @code{mod}.
- @subsection Simple Loops
- @cindex simple loops
- @cindex loops without count
- @cindex @code{WHILE} loop
  @example
- BEGIN
+3 /mod . .
-   @var{code1}
-   @var{flag}
- WHILE
-   @var{code2}
- REPEAT
  @end example
- @var{code1} is executed and @var{flag} is computed. If it is true,
+ Reference: @ref{Arithmetic}.
- @var{code2} is executed and the loop is restarted; If @var{flag} is
- false, execution continues after the @code{REPEAT}.
+ @node Stack Manipulation Tutorial, Using files for Forth code Tutorial, Arithmetics Tutorial, Tutorial
+ @section Stack Manipulation
+ @cindex stack manipulation tutorial
+ Stack manipulation words rearrange the data on the stack.
- @cindex @code{UNTIL} loop
  @example
- BEGIN
+.s drop .s
-   @var{code}
+.s dup .s drop drop .s
-   @var{flag}
+2 .s over .s drop drop drop
- UNTIL
+2 .s swap .s drop drop
+2 3 .s rot .s drop drop drop
  @end example
- @var{code} is executed. The loop is restarted if @code{flag} is false.
+ These are the most important stack manipulation words.  There are also
+ variants that manipulate twice as many stack items:
- @cindex endless loop
- @cindex loops, endless
  @example
- BEGIN
+2 3 4 .s 2swap .s 2drop 2drop
-   @var{code}
- AGAIN
  @end example
- This is an endless loop.
+ Two more stack manipulation words are:
- @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
- @subsection Counted Loops
- @cindex counted loops
- @cindex loops, counted
- @cindex @code{DO} loops
- The basic counted loop is:
  @example
- @var{limit} @var{start}
+2 .s nip .s drop
- ?DO
+2 .s tuck .s 2drop drop
-   @var{body}
- LOOP
  @end example
- This performs one iteration for every integer, starting from @var{start}
+ @assignment
- and up to, but excluding @var{limit}. The counter, aka index, can be
+ Replace @code{nip} and @code{tuck} with combinations of other stack
- accessed with @code{i}. E.g., the loop
+ manipulation words.
  @example
-0 ?DO
+ Given:          How do you get:
-   i .
+2 3           3 2 1
- LOOP
+2 3           1 2 3 2
+2 3           1 2 3 3
+2 3           1 3 3
+2 3           2 1 3
+2 3 4         4 3 2 1
+2 3           1 2 3 1 2 3
+2 3 4         1 2 3 4 1 2
+2 3
+2 3           1 2 3 4
+2 3           1 3
  @end example
- prints
+ @endassignment
  @example
-1 2 3 4 5 6 7 8 9
+dup * .
  @end example
- The index of the innermost loop can be accessed with @code{i}, the index
- of the next loop with @code{j}, and the index of the third loop with
- @code{k}.
- doc-i
- doc-j
- doc-k
- The loop control data are kept on the return stack, so there are some
+ @assignment
- restrictions on mixing return stack accesses and counted loop
+ Write 17^3 and 17^4 in Forth, without writing @code{17} more than once.
- words. E.g., if you put values on the return stack outside the loop, you
+ Write a piece of Forth code that expects two numbers on the stack
- cannot read them inside the loop. If you put values on the return stack
+ (@var{a} and @var{b}, with @var{b} on top) and computes
- within a loop, you have to remove them before the end of the loop and
+ @code{(a-b)(a+1)}.
- before accessing the index of the loop.
+ @endassignment
- There are several variations on the counted loop:
+ Reference: @ref{Stack Manipulation}.
- @code{LEAVE} leaves the innermost counted loop immediately.
- If @var{start} is greater than @var{limit}, a @code{?DO} loop is entered
+ @node Using files for Forth code Tutorial, Comments Tutorial, Stack Manipulation Tutorial, Tutorial
- (and @code{LOOP} iterates until they become equal by wrap-around
+ @section Using files for Forth code
- arithmetic). This behaviour is usually not what you want. Therefore,
+ @cindex loading Forth code, tutorial
- Gforth offers @code{+DO} and @code{U+DO} (as replacements for
+ @cindex files containing Forth code, tutorial
- @code{?DO}), which do not enter the loop if @var{start} is greater than
- @var{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
- unsigned loop parameters.
- @code{LOOP} can be replaced with @code{@var{n} +LOOP}; this updates the
+ While working at the Forth command line is convenient for one-line
- index by @var{n} instead of by 1. The loop is terminated when the border
+ examples and short one-off code, you probably want to store your source
- between @var{limit-1} and @var{limit} is crossed. E.g.:
+ code in files for convenient editing and persistence.  You can use your
+ favourite editor (Gforth includes Emacs support, @pxref{Emacs and
+ Gforth}) to create @var{file} and use
- @code{4 0 +DO  i .  2 +LOOP}   prints @code{0 2}
+ @example
+ s" @var{file}" included
+ @end example
- @code{4 1 +DO  i .  2 +LOOP}   prints @code{1 3}
+ to load it into your Forth system.  The file name extension I use for
+ Forth files is @samp{.fs}.
- @cindex negative increment for counted loops
+ You can easily start Gforth with some files loaded like this:
- @cindex counted loops with negative increment
- The behaviour of @code{@var{n} +LOOP} is peculiar when @var{n} is negative:
- @code{-1 0 ?DO  i .  -1 +LOOP}  prints @code{0 -1}
+ @example
+ gforth @var{file1} @var{file2}
+ @end example
- @code{ 0 0 ?DO  i .  -1 +LOOP}  prints nothing
+ If an error occurs during loading these files, Gforth terminates,
+ whereas an error during @code{INCLUDED} within Gforth usually gives you
+ a Gforth command line.  Starting the Forth system every time gives you a
+ clean start every time, without interference from the results of earlier
+ tries.
- Therefore we recommend avoiding @code{@var{n} +LOOP} with negative
+ I often put all the tests in a file, then load the code and run the
- @var{n}. One alternative is @code{@var{u} -LOOP}, which reduces the
+ tests with
- index by @var{u} each iteration. The loop is terminated when the border
- between @var{limit+1} and @var{limit} is crossed. Gforth also provides
- @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
- @code{-2 0 -DO  i .  1 -LOOP}  prints @code{0 -1}
+ @example
+ gforth @var{code} @var{tests} -e bye
+ @end example
- @code{-1 0 -DO  i .  1 -LOOP}  prints @code{0}
+ (often by performing this command with @kbd{C-x C-e} in Emacs).  The
+ @code{-e bye} ensures that Gforth terminates afterwards so that I can
+ restart this command without ado.
- @code{ 0 0 -DO  i .  1 -LOOP}  prints nothing
+ The advantage of this approach is that the tests can be repeated easily
+ every time the program ist changed, making it easy to catch bugs
+ introduced by the change.
- Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
+ Reference: @ref{Forth source files}.
- @code{-LOOP} are not in the ANS Forth standard. However, an
- implementation for these words that uses only standard words is provided
- in @file{compat/loops.fs}.
- @code{?DO} can also be replaced by @code{DO}. @code{DO} always enters
- the loop, independent of the loop parameters. Do not use @code{DO}, even
- if you know that the loop is entered in any case. Such knowledge tends
- to become invalid during maintenance of a program, and then the
- @code{DO} will make trouble.
- @code{UNLOOP} is used to prepare for an abnormal loop exit, e.g., via
+ @node Comments Tutorial, Colon Definitions Tutorial, Using files for Forth code Tutorial, Tutorial
- @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
+ @section Comments
- return stack so @code{EXIT} can get to its return address.
+ @cindex comments tutorial
- @cindex @code{FOR} loops
- Another counted loop is
  @example
- @var{n}
+ \ That's a comment; it ends at the end of the line
- FOR
+ ( Another comment; it ends here: )  .s
-   @var{body}
- NEXT
  @end example
- This is the preferred loop of native code compiler writers who are too
- lazy to optimize @code{?DO} loops properly. In Gforth, this loop
- iterates @var{n+1} times; @code{i} produces values starting with @var{n}
- and ending with 0. Other Forth systems may behave differently, even if
- they support @code{FOR} loops. To avoid problems, don't use @code{FOR}
- loops.
- @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
+ @code{\} and @code{(} are ordinary Forth words and therefore have to be
- @subsection Arbitrary control structures
+ separated with white space from the following text.
- @cindex control structures, user-defined
- @cindex control-flow stack
+ @example
- ANS Forth permits and supports using control structures in a non-nested
+ \This gives an "Undefined word" error
- way. Information about incomplete control structures is stored on the
+ @end example
- control-flow stack. This stack may be implemented on the Forth data
- stack, and this is what we have done in Gforth.
- @cindex @code{orig}, control-flow stack item
+ The first @code{)} ends a comment started with @code{(}, so you cannot
- @cindex @code{dest}, control-flow stack item
+ nest @code{(}-comments; and you cannot comment out text containing a
- An @i{orig} entry represents an unresolved forward branch, a @i{dest}
+ @code{)} with @code{( ... )}@footnote{therefore it's a good idea to
- entry represents a backward branch target. A few words are the basis for
+ avoid @code{)} in word names.}.
- building any control structure possible (except control structures that
- need storage, like calls, coroutines, and backtracking).
- doc-if
+ I use @code{\}-comments for descriptive text and for commenting out code
- doc-ahead
+ of one or more line; I use @code{(}-comments for describing the stack
- doc-then
+ effect, the stack contents, or for commenting out sub-line pieces of
- doc-begin
+ code.
- doc-until
- doc-again
- doc-cs-pick
- doc-cs-roll
- On many systems control-flow stack items take one word, in Gforth they
+ The Emacs mode @file{gforth.el} (@pxref{Emacs and Gforth}) supports
- currently take three (this may change in the future). Therefore it is a
+ these uses by commenting out a region with @kbd{C-x \}, uncommenting a
- really good idea to manipulate the control flow stack with
+ region with @kbd{C-u C-x \}, and filling a @code{\}-commented region
- @code{cs-pick} and @code{cs-roll}, not with data stack manipulation
+ with @kbd{M-q}.
- words.
- Some standard control structure words are built from these words:
+ Reference: @ref{Comments}.
- doc-else
- doc-while
- doc-repeat
- Gforth adds some more control-structure words:
+ @node Colon Definitions Tutorial, Decompilation Tutorial, Comments Tutorial, Tutorial
+ @section Colon Definitions
+ @cindex colon definitions, tutorial
+ @cindex definitions, tutorial
+ @cindex procedures, tutorial
+ @cindex functions, tutorial
- doc-endif
+ are similar to procedures and functions in other programming languages.
- doc-?dup-if
- doc-?dup-0=-if
- Counted loop words constitute a separate group of words:
+ @example
+ : squared ( n -- n^2 )
+    dup * ;
+squared .
+squared .
+ @end example
- doc-?do
+ @code{:} starts the colon definition; its name is @code{squared}.  The
- doc-+do
+ following comment describes its stack effect.  The words @code{dup *}
- doc-u+do
+ are not executed, but compiled into the definition.  @code{;} ends the
- doc--do
+ colon definition.
- doc-u-do
- doc-do
- doc-for
- doc-loop
- doc-+loop
- doc--loop
- doc-next
- doc-leave
- doc-?leave
- doc-unloop
- doc-done
- The standard does not allow using @code{cs-pick} and @code{cs-roll} on
+ The newly-defined word can be used like any other word, including using
- @i{do-sys}. Our system allows it, but it's your job to ensure that for
+ it in other definitions:
- every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
- through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
- fall-through path). Also, you have to ensure that all @code{LEAVE}s are
- resolved (by using one of the loop-ending words or @code{DONE}).
- Another group of control structure words are
+ @example
+ : cubed ( n -- n^3 )
+    dup squared * ;
+ -5 cubed .
+ : fourth-power ( n -- n^4 )
+    squared squared ;
+fourth-power .
+ @end example
- doc-case
+ @assignment
- doc-endcase
+ Write colon definitions for @code{nip}, @code{tuck}, @code{negate}, and
- doc-of
+ @code{/mod} in terms of other Forth words, and check if they work (hint:
- doc-endof
+ test your tests on the originals first).  Don't let the
+ @samp{redefined}-Messages spook you, they are just warnings.
+ @endassignment
- @i{case-sys} and @i{of-sys} cannot be processed using @code{cs-pick} and
+ Reference: @ref{Colon Definitions}.
- @code{cs-roll}.
- @subsubsection Programming Style
- In order to ensure readability we recommend that you do not create
+ @node Decompilation Tutorial, Stack-Effect Comments Tutorial, Colon Definitions Tutorial, Tutorial
- arbitrary control structures directly, but define new control structure
+ @section Decompilation
- words for the control structure you want and use these words in your
+ @cindex decompilation tutorial
- program.
+ @cindex see tutorial
- E.g., instead of writing
+ You can decompile colon definitions with @code{see}:
  @example
- begin
+ see squared
-   ...
+ see cubed
- if [ 1 cs-roll ]
-   ...
- again then
  @end example
- we recommend defining control structure words, e.g.,
+ In Gforth @code{see} shows you a reconstruction of the source code from
+ the executable code.  Informations that were present in the source, but
+ not in the executable code, are lost (e.g., comments).
- @example
+ You can also decompile the predefined words:
- : while ( dest -- orig dest )
-  POSTPONE if
-cs-roll ; immediate
- : repeat ( orig dest -- )
+ @example
-  POSTPONE again
+ see .
-  POSTPONE then ; immediate
+ see +
  @end example
- and then using these to create the control structure:
+ @node Stack-Effect Comments Tutorial, Types Tutorial, Decompilation Tutorial, Tutorial
+ @section Stack-Effect Comments
+ @cindex stack-effect comments, tutorial
+ @cindex --, tutorial
+ By convention the comment after the name of a definition describes the
+ stack effect: The part in from of the @samp{--} describes the state of
+ the stack before the execution of the definition, i.e., the parameters
+ that are passed into the colon definition; the part behind the @samp{--}
+ is the state of the stack after the execution of the definition, i.e.,
+ the results of the definition.  The stack comment only shows the top
+ stack items that the definition accesses and/or changes.
+ You should put a correct stack effect on every definition, even if it is
+ just @code{( -- )}.  You should also add some descriptive comment to
+ more complicated words (I usually do this in the lines following
+ @code{:}).  If you don't do this, your code becomes unreadable (because
+ you have to work through every definition before you can undertsand
+ any).
+ @assignment
+ The stack effect of @code{swap} can be written like this: @code{x1 x2 --
+ x2 x1}.  Describe the stack effect of @code{-}, @code{drop}, @code{dup},
+ @code{over}, @code{rot}, @code{nip}, and @code{tuck}.  Hint: When you
+ are done, you can compare your stack effects to those in this manual
+ (@pxref{Word Index}).
+ @endassignment
+ Sometimes programmers put comments at various places in colon
+ definitions that describe the contents of the stack at that place (stack
+ comments); i.e., they are like the first part of a stack-effect
+ comment. E.g.,
  @example
- begin
+ : cubed ( n -- n^3 )
-   ...
+    dup squared  ( n n^2 ) * ;
- while
-   ...
- repeat
  @end example
- That's much easier to read, isn't it? Of course, @code{REPEAT} and
+ In this case the stack comment is pretty superfluous, because the word
- @code{WHILE} are predefined, so in this example it would not be
+ is simple enough.  If you think it would be a good idea to add such a
- necessary to define them.
+ comment to increase readability, you should also consider factoring the
+ word into several simpler words (@pxref{Factoring Tutorial,,
+ Factoring}), which typically eliminates the need for the stack comment;
+ however, if you decide not to refactor it, then having such a comment is
+ better than not having it.
- @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
+ The names of the stack items in stack-effect and stack comments in the
- @subsection Calls and returns
+ standard, in this manual, and in many programs specify the type through
- @cindex calling a definition
+ a type prefix, similar to Fortran and Hungarian notation.  The most
- @cindex returning from a definition
+ frequent prefixes are:
- @cindex recursive definitions
+ @table @code
- A definition can be called simply be writing the name of the definition
+ @item n
- to be called. Note that normally a definition is invisible during its
+ signed integer
- definition. If you want to write a directly recursive definition, you
+ @item u
- can use @code{recursive} to make the current definition visible.
+ unsigned integer
+ @item c
+ character
+ @item f
+ Boolean flags, i.e. @code{false} or @code{true}.
+ @item a-addr,a-
+ Cell-aligned address
+ @item c-addr,c-
+ Char-aligned address (note that a Char may have two bytes in Windows NT)
+ @item xt
+ Execution token, same size as Cell
+ @item w,x
+ Cell, can contain an integer or an address.  It usually takes 32, 64 or
+bits (depending on your platform and Forth system). A cell is more
+ commonly known as machine word, but the term @emph{word} already means
+ something different in Forth.
+ @item d
+ signed double-cell integer
+ @item ud
+ unsigned double-cell integer
+ @item r
+ Float (on the FP stack)
+ @end table
- doc-recursive
+ You can find a more complete list in @ref{Notation}.
- Another way to perform a recursive call is
+ @assignment
+ Write stack-effect comments for all definitions you have written up to
+ now.
+ @endassignment
+ @node Types Tutorial, Factoring Tutorial, Stack-Effect Comments Tutorial, Tutorial
+ @section Types
+ @cindex types tutorial
+ In Forth the names of the operations are not overloaded; so similar
+ operations on different types need different names; e.g., @code{+} adds
+ integers, and you have to use @code{f+} to add floating-point numbers.
+ The following prefixes are often used for related operations on
+ different types:
- doc-recurse
+ @table @code
+ @item (none)
+ signed integer
+ @item u
+ unsigned integer
+ @item c
+ character
+ @item d
+ signed double-cell integer
+ @item ud, du
+ unsigned double-cell integer
+ @item 2
+ two cells (not-necessarily double-cell numbers)
+ @item m, um
+ mixed single-cell and double-cell operations
+ @item f
+ floating-point (note that in stack comments @samp{f} represents flags,
+ and @samp{r} represents FP numbers).
+ @end table
- @quotation
+ If there are no differences between the signed and the unsigned variant
- @progstyle
+ (e.g., for @code{+}), there is only the prefix-less variant.
- I prefer using @code{recursive} to @code{recurse}, because calling the
- definition by name is more descriptive (if the name is well-chosen) than
- the somewhat cryptic @code{recurse}.  E.g., in a quicksort
- implementation, it is much better to read (and think) ``now sort the
- partitions'' than to read ``now do a recursive call''.
- @end quotation
- For mutual recursion, use @code{defer}red words, like this:
+ Forth does not perform type checking, neither at compile time, nor at
+ run time.  If you use the wrong oeration, the data are interpreted
+ incorrectly:
  @example
- defer foo
+ -1 u.
- : bar ( ... -- ... )
-  ... foo ... ;
- :noname ( ... -- ... )
-  ... bar ... ;
- IS foo
  @end example
- When the end of the definition is reached, it returns. An earlier return
+ If you have only experience with type-checked languages until now, and
- can be forced using
+ have heard how important type-checking is, don't panic!  In my
+ experience (and that of other Forthers), type errors in Forth code are
+ usually easy to find (once you get used to it), the increased vigilance
+ of the programmer tends to catch some harder errors in addition to most
+ type errors, and you never have to work around the type system, so in
+ most situations the lack of type-checking seems to be a win (projects to
+ add type checking to Forth have not caught on).
- doc-exit
- Don't forget to clean up the return stack and @code{UNLOOP} any
+ @node Factoring Tutorial, Designing the stack effect Tutorial, Types Tutorial, Tutorial
- outstanding @code{?DO}...@code{LOOP}s before @code{EXIT}ing. The
+ @section Factoring
- primitive compiled by @code{EXIT} is
+ @cindex factoring tutorial
- doc-;s
+ If you try to write longer definitions, you will soon find it hard to
+ keep track of the stack contents.  Therefore, good Forth programmers
+ tend to write only short definitions (e.g., three lines).  The art of
+ finding meaningful short definitions is known as factoring (as in
+ factoring polynomials).
- @node Exception Handling,  , Calls and returns, Control Structures
+ Well-factored programs offer additional advantages: smaller, more
- @subsection Exception Handling
+ general words, are easier to test and debug and can be reused more and
- @cindex Exceptions
+ better than larger, specialized words.
- doc-catch
+ So, if you run into difficulties with stack management, when writing
- doc-throw
+ code, try to define meaningful factors for the word, and define the word
+ in terms of those.  Even if a factor contains only two words, it is
+ often helpful.
- @node Locals, Defining Words, Control Structures, Words
+ Good factoring is not easy, and it takes some practice to get the knack
- @section Locals
+ for it; but even experienced Forth programmers often don't find the
- @cindex locals
+ right solution right away, but only when rewriting the program.  So, if
+ you don't come up with a good solution immediately, keep trying, don't
+ despair.
- Local variables can make Forth programming more enjoyable and Forth
+ @c example !!
- programs easier to read. Unfortunately, the locals of ANS Forth are
- laden with restrictions. Therefore, we provide not only the ANS Forth
- locals wordset, but also our own, more powerful locals wordset (we
- implemented the ANS Forth locals wordset through our locals wordset).
- The ideas in this section have also been published in the paper
- @cite{Automatic Scoping of Local Variables} by M. Anton Ertl, presented
- at EuroForth '94; it is available at
- @*@url{http://www.complang.tuwien.ac.at/papers/ertl94l.ps.gz}.
- @menu
+ @node Designing the stack effect Tutorial, Local Variables Tutorial, Factoring Tutorial, Tutorial
- * Gforth locals::
+ @section Designing the stack effect
- * ANS Forth locals::
+ @cindex Stack effect design, tutorial
- @end menu
+ @cindex design of stack effects, tutorial
- @node Gforth locals, ANS Forth locals, Locals, Locals
+ In other languages you can use an arbitrary order of parameters for a
- @subsection Gforth locals
+ function; and since there is only one result, you don't have to deal with
- @cindex Gforth locals
+ the order of results, either.
- @cindex locals, Gforth style
- Locals can be defined with
+ In Forth (and other stack-based languages, e.g., Postscript) the
+ parameter and result order of a definition is important and should be
+ designed well.  The general guideline is to design the stack effect such
+ that the word is simple to use in most cases, even if that complicates
+ the implementation of the word.  Some concrete rules are:
+ @itemize @bullet
+ @item
+ Words consume all of their parameters (e.g., @code{.}).
+ @item
+ If there is a convention on the order of parameters (e.g., from
+ mathematics or another programming language), stick with it (e.g.,
+ @code{-}).
+ @item
+ If one parameter usually requires only a short computation (e.g., it is
+ a constant), pass it on the top of the stack.  Conversely, parameters
+ that usually require a long sequence of code to compute should be passed
+ as the bottom (i.e., first) parameter.  This makes the code easier to
+ read, because reader does not need to keep track of the bottom item
+ through a long sequence of code (or, alternatively, through stack
+ manipulations). E.g., @code{!} (store, @pxref{Memory}) expects the
+ address on top of the stack because it is usually simpler to compute
+ than the stored value (often the address is just a variable).
+ @item
+ Similarly, results that are usually consumed quickly should be returned
+ on the top of stack, whereas a result that is often used in long
+ computations should be passed as bottom result.  E.g., the file words
+ like @code{open-file} return the error code on the top of stack, because
+ it is usually consumed quickly by @code{throw}; moreover, the error code
+ has to be checked before doing anything with the other results.
+ @end itemize
+ These rules are just general guidelines, don't lose sight of the overall
+ goal to make the words easy to use.  E.g., if the convention rule
+ conflicts with the computation-length rule, you might decide in favour
+ of the convention if the word will be used rarely, and in favour of the
+ computation-length rule if the word will be used frequently (because
+ with frequent use the cost of breaking the computation-length rule would
+ be quite high, and frequent use makes it easier to remember an
+ unconventional order).
+ @c example !! structure package
+ @node Local Variables Tutorial, Conditional execution Tutorial, Designing the stack effect Tutorial, Tutorial
+ @section Local Variables
+ @cindex local variables, tutorial
+ You can define local variables (@emph{locals}) in a colon definition:
  @example
- @{ local1 local2 ... -- comment @}
+ : swap @{ a b -- b a @}
+   b a ;
+2 swap .s 2drop
  @end example
- or
+ (If your Forth system does not support this syntax, include
+ @file{compat/anslocals.fs} first).
+ In this example @code{@{ a b -- b a @}} is the locals definition; it
+ takes two cells from the stack, puts the top of stack in @code{b} and
+ the next stack element in @code{a}.  @code{--} starts a comment ending
+ with @code{@}}.  After the locals definition, using the name of the
+ local will push its value on the stack.  You can leave the comment
+ part (@code{-- b a}) away:
  @example
- @{ local1 local2 ... @}
+ : swap ( x1 x2 -- x2 x1 )
+   @{ a b @} b a ;
  @end example
- E.g.,
+ In Gforth you can have several locals definitions, anywhere in a colon
+ definition; in contrast, in a standard program you can have only one
+ locals definition per colon definition, and that locals definition must
+ be outside any controll structure.
+ With locals you can write slightly longer definitions without running
+ into stack trouble.  However, I recommend trying to write colon
+ definitions without locals for exercise purposes to help you gain the
+ essential factoring skills.
+ @assignment
+ Rewrite your definitions until now with locals
+ @endassignment
+ Reference: @ref{Locals}.
+ @node Conditional execution Tutorial, Flags and Comparisons Tutorial, Local Variables Tutorial, Tutorial
+ @section Conditional execution
+ @cindex conditionals, tutorial
+ @cindex if, tutorial
+ In Forth you can use control structures only inside colon definitions.
+ An @code{if}-structure looks like this:
  @example
- : max @{ n1 n2 -- n3 @}
+ : abs ( n1 -- +n2 )
-  n1 n2 > if
+     dup 0 < if
-    n1
+         negate
-  else
-    n2
   endif ;
+abs .
+ -5 abs .
  @end example
- The similarity of locals definitions with stack comments is intended. A
+ @code{if} takes a flag from the stack.  If the flag is non-zero (true),
- locals definition often replaces the stack comment of a word. The order
+ the following code is performed, otherwise execution continues after the
- of the locals corresponds to the order in a stack comment and everything
+ @code{endif} (or @code{else}).  @code{<} compares the top two stack
- after the @code{--} is really a comment.
+ elements and prioduces a flag:
- This similarity has one disadvantage: It is too easy to confuse locals
+ @example
- declarations with stack comments, causing bugs and making them hard to
+2 < .
- find. However, this problem can be avoided by appropriate coding
+1 < .
- conventions: Do not use both notations in the same program. If you do,
+1 < .
- they should be distinguished using additional means, e.g. by position.
+ @end example
- @cindex types of locals
+ Actually the standard name for @code{endif} is @code{then}.  This
- @cindex locals types
+ tutorial presents the examples using @code{endif}, because this is often
- The name of the local may be preceded by a type specifier, e.g.,
+ less confusing for people familiar with other programming languages
- @code{F:} for a floating point value:
+ where @code{then} has a different meaning.  If your system does not have
+ @code{endif}, define it with
  @example
- : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
+ : endif postpone then ; immediate
- \ complex multiplication
-  Ar Br f* Ai Bi f* f-
-  Ar Bi f* Ai Br f* f+ ;
  @end example
- @cindex flavours of locals
+ You can optionally use an @code{else}-part:
- @cindex locals flavours
- @cindex value-flavoured locals
- @cindex variable-flavoured locals
- Gforth currently supports cells (@code{W:}, @code{W^}), doubles
- (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
- (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
- with @code{W:}, @code{D:} etc.) produces its value and can be changed
- with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
- produces its address (which becomes invalid when the variable's scope is
- left). E.g., the standard word @code{emit} can be defined in terms of
- @code{type} like this:
  @example
- : emit @{ C^ char* -- @}
+ : min ( n1 n2 -- n )
-     char* 1 type ;
+dup < if
+     drop
+   else
+     nip
+   endif ;
+3 min .
+2 min .
  @end example
- @cindex default type of locals
+ @assignment
- @cindex locals, default type
+ Write @code{min} without @code{else}-part (hint: what's the definition
- A local without type specifier is a @code{W:} local. Both flavours of
+ of @code{nip}?).
- locals are initialized with values from the data or FP stack.
+ @endassignment
- Currently there is no way to define locals with user-defined data
+ Reference: @ref{Selection}.
- structures, but we are working on it.
- Gforth allows defining locals everywhere in a colon definition. This
- poses the following questions:
- @menu
+ @node Flags and Comparisons Tutorial, General Loops Tutorial, Conditional execution Tutorial, Tutorial
- * Where are locals visible by name?::
+ @section Flags and Comparisons
- * How long do locals live?::
+ @cindex flags tutorial
- * Programming Style::
+ @cindex comparison tutorial
- * Implementation::
- @end menu
- @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
+ In a false-flag all bits are clear (0 when interpreted as integer).  In
- @subsubsection Where are locals visible by name?
+ a canonical true-flag all bits are set (-1 as a twos-complement signed
- @cindex locals visibility
+ integer); in many contexts (e.g., @code{if}) any non-zero value is
- @cindex visibility of locals
+ treated as true flag.
- @cindex scope of locals
- Basically, the answer is that locals are visible where you would expect
+ @example
- it in block-structured languages, and sometimes a little longer. If you
+ false .
- want to restrict the scope of a local, enclose its definition in
+ true .
- @code{SCOPE}...@code{ENDSCOPE}.
+ true hex u. decimal
+ @end example
- doc-scope
+ Comparison words produce canonical flags:
- doc-endscope
- These words behave like control structure words, so you can use them
+ @example
- with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
+1 = .
- arbitrary ways.
+0= .
+1 < .
+0 < .
+ -1 1 u< . \ type error, u< interprets -1 as large unsigned number
+ -1 1 < .
+ @end example
- If you want a more exact answer to the visibility question, here's the
+ Gforth supports all combinations of the prefixes @code{0 u d d0 du f f0}
- basic principle: A local is visible in all places that can only be
+ (or none) and the comparisons @code{= <> < > <= >=}.  Only a part of
- reached through the definition of the local@footnote{In compiler
+ these combinations are standard (for details see the standard,
- construction terminology, all places dominated by the definition of the
+ @ref{Numeric comparison}, @ref{Floating Point} or @ref{Word Index}).
- local.}. In other words, it is not visible in places that can be reached
- without going through the definition of the local. E.g., locals defined
- in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
- defined in @code{BEGIN}...@code{UNTIL} are visible after the
- @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
- The reasoning behind this solution is: We want to have the locals
+ You can use @code{and or xor invert} can be used as operations on
- visible as long as it is meaningful. The user can always make the
+ canonical flags.  Actually they are bitwise operations:
- visibility shorter by using explicit scoping. In a place that can
- only be reached through the definition of a local, the meaning of a
- local name is clear. In other places it is not: How is the local
- initialized at the control flow path that does not contain the
- definition? Which local is meant, if the same name is defined twice in
- two independent control flow paths?
- This should be enough detail for nearly all users, so you can skip the
+ @example
- rest of this section. If you really must know all the gory details and
+2 and .
- options, read on.
+2 or .
+3 xor .
+invert .
+ @end example
- In order to implement this rule, the compiler has to know which places
+ You can convert a zero/non-zero flag into a canonical flag with
- are unreachable. It knows this automatically after @code{AHEAD},
+ @code{0<>} (and complement it on the way with @code{0=}).
- @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
- most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
- compiler that the control flow never reaches that place. If
- @code{UNREACHABLE} is not used where it could, the only consequence is
- that the visibility of some locals is more limited than the rule above
- says. If @code{UNREACHABLE} is used where it should not (i.e., if you
- lie to the compiler), buggy code will be produced.
- doc-unreachable
+ @example
+0= .
+0<> .
+ @end example
+ You can use the all-bits-set feature of canonical flags and the bitwise
+ operation of the Boolean operations to avoid @code{if}s:
- Another problem with this rule is that at @code{BEGIN}, the compiler
- does not know which locals will be visible on the incoming
- back-edge. All problems discussed in the following are due to this
- ignorance of the compiler (we discuss the problems using @code{BEGIN}
- loops as examples; the discussion also applies to @code{?DO} and other
- loops). Perhaps the most insidious example is:
  @example
- AHEAD
+ : foo ( n1 -- n2 )
- BEGIN
+= if
-   x
- [ 1 CS-ROLL ] THEN
+   else
-   @{ x @}
-   ...
+   endif ;
- UNTIL
+foo .
+foo .
+ : foo ( n1 -- n2 )
+= 14 and ;
+foo .
+foo .
  @end example
- This should be legal according to the visibility rule. The use of
+ @assignment
- @code{x} can only be reached through the definition; but that appears
+ Write @code{min} without @code{if}.
- textually below the use.
+ @endassignment
- From this example it is clear that the visibility rules cannot be fully
+ For reference, see @ref{Boolean Flags}, @ref{Numeric comparison}, and
- implemented without major headaches. Our implementation treats common
+ @ref{Bitwise operations}.
- cases as advertised and the exceptions are treated in a safe way: The
- compiler makes a reasonable guess about the locals visible after a
- @code{BEGIN}; if it is too pessimistic, the
+ @node General Loops Tutorial, Counted loops Tutorial, Flags and Comparisons Tutorial, Tutorial
- user will get a spurious error about the local not being defined; if the
+ @section General Loops
- compiler is too optimistic, it will notice this later and issue a
+ @cindex loops, indefinite, tutorial
- warning. In the case above the compiler would complain about @code{x}
- being undefined at its use. You can see from the obscure examples in
+ The endless loop is the most simple one:
- this section that it takes quite unusual control structures to get the
- compiler into trouble, and even then it will often do fine.
- If the @code{BEGIN} is reachable from above, the most optimistic guess
- is that all locals visible before the @code{BEGIN} will also be
- visible after the @code{BEGIN}. This guess is valid for all loops that
- are entered only through the @code{BEGIN}, in particular, for normal
- @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
- @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
- compiler. When the branch to the @code{BEGIN} is finally generated by
- @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
- warns the user if it was too optimistic:
  @example
- IF
+ : endless ( -- )
-   @{ x @}
+begin
- BEGIN
+     dup . 1+
-   \ x ?
+   again ;
- [ 1 cs-roll ] THEN
+ endless
-   ...
- UNTIL
  @end example
- Here, @code{x} lives only until the @code{BEGIN}, but the compiler
+ Terminate this loop by pressing @kbd{Ctrl-C} (in Gforth).  @code{begin}
- optimistically assumes that it lives until the @code{THEN}. It notices
+ does nothing at run-time, @code{again} jumps back to @code{begin}.
- this difference when it compiles the @code{UNTIL} and issues a
- warning. The user can avoid the warning, and make sure that @code{x}
+ A loop with one exit at any place looks like this:
- is not used in the wrong area by using explicit scoping:
  @example
- IF
+ : log2 ( +n1 -- n2 )
-   SCOPE
+ \ logarithmus dualis of n1>0, rounded down to the next integer
-   @{ x @}
+   assert( dup 0> )
-   ENDSCOPE
+/ 0 begin
- BEGIN
+     over 0> while
- [ 1 cs-roll ] THEN
++ swap 2/ swap
-   ...
+   repeat
- UNTIL
+   nip ;
+log2 .
+log2 .
  @end example
- Since the guess is optimistic, there will be no spurious error messages
+ At run-time @code{while} consumes a flag; if it is 0, execution
- about undefined locals.
+ continues behind the @code{repeat}; if the flag is non-zero, execution
+ continues behind the @code{while}.  @code{Repeat} jumps back to
+ @code{begin}, just like @code{again}.
- If the @code{BEGIN} is not reachable from above (e.g., after
+ In Forth there are many combinations/abbreviations, like @code{1+}.
- @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
+ However, @code{2/} is not one of them; it shifts it's argument right by
- optimistic guess, as the locals visible after the @code{BEGIN} may be
+ one bit (arithmetic shift right):
- defined later. Therefore, the compiler assumes that no locals are
- visible after the @code{BEGIN}. However, the user can use
- @code{ASSUME-LIVE} to make the compiler assume that the same locals are
- visible at the BEGIN as at the point where the top control-flow stack
- item was created.
- doc-assume-live
+ @example
+ -5 2 / .
+ -5 2/ .
+ @end example
+ @code{assert(} is no standard word, but you can get it on systems other
+ then Gforth by including @file{compat/assert.fs}.  You can see what it
+ does by trying
- E.g.,
  @example
- @{ x @}
+log2 .
- AHEAD
- ASSUME-LIVE
- BEGIN
-   x
- [ 1 CS-ROLL ] THEN
-   ...
- UNTIL
  @end example
- Other cases where the locals are defined before the @code{BEGIN} can be
+ Here's a loop with an exit at the end:
- handled by inserting an appropriate @code{CS-ROLL} before the
- @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
- behind the @code{ASSUME-LIVE}).
- Cases where locals are defined after the @code{BEGIN} (but should be
- visible immediately after the @code{BEGIN}) can only be handled by
- rearranging the loop. E.g., the ``most insidious'' example above can be
- arranged into:
  @example
- BEGIN
+ : log2 ( +n1 -- n2 )
-   @{ x @}
+ \ logarithmus dualis of n1>0, rounded down to the next integer
-   ... 0=
+   assert( dup 0 > )
- WHILE
+   -1 begin
-   x
++ swap 2/ swap
- REPEAT
+     over 0 <=
+   until
+   nip ;
  @end example
- @node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
+ @code{Until} consumes a flag; if it is non-zero, execution continues at
- @subsubsection How long do locals live?
+ the @code{begin}, otherwise after the @code{until}.
- @cindex locals lifetime
- @cindex lifetime of locals
- The right answer for the lifetime question would be: A local lives at
+ @assignment
- least as long as it can be accessed. For a value-flavoured local this
+ Write a definition for computing the greatest common divisor.
- means: until the end of its visibility. However, a variable-flavoured
+ @endassignment
- local could be accessed through its address far beyond its visibility
- scope. Ultimately, this would mean that such locals would have to be
- garbage collected. Since this entails un-Forth-like implementation
- complexities, I adopted the same cowardly solution as some other
- languages (e.g., C): The local lives only as long as it is visible;
- afterwards its address is invalid (and programs that access it
- afterwards are erroneous).
- @node Programming Style, Implementation, How long do locals live?, Gforth locals
+ Reference: @ref{Simple Loops}.
- @subsubsection Programming Style
- @cindex locals programming style
- @cindex programming style, locals
- The freedom to define locals anywhere has the potential to change
- programming styles dramatically. In particular, the need to use the
- return stack for intermediate storage vanishes. Moreover, all stack
- manipulations (except @code{PICK}s and @code{ROLL}s with run-time
- determined arguments) can be eliminated: If the stack items are in the
- wrong order, just write a locals definition for all of them; then
- write the items in the order you want.
- This seems a little far-fetched and eliminating stack manipulations is
- unlikely to become a conscious programming objective. Still, the number
- of stack manipulations will be reduced dramatically if local variables
- are used liberally (e.g., compare @code{max} in @ref{Gforth locals} with
- a traditional implementation of @code{max}).
- This shows one potential benefit of locals: making Forth programs more
- readable. Of course, this benefit will only be realized if the
- programmers continue to honour the principle of factoring instead of
- using the added latitude to make the words longer.
- @cindex single-assignment style for locals
- Using @code{TO} can and should be avoided.  Without @code{TO},
- every value-flavoured local has only a single assignment and many
- advantages of functional languages apply to Forth. I.e., programs are
- easier to analyse, to optimize and to read: It is clear from the
- definition what the local stands for, it does not turn into something
- different later.
- E.g., a definition using @code{TO} might look like this:
+ @node Counted loops Tutorial, Recursion Tutorial, General Loops Tutorial, Tutorial
- @example
+ @section Counted loops
- : strcmp @{ addr1 u1 addr2 u2 -- n @}
+ @cindex loops, counted, tutorial
-  u1 u2 min 0
-  ?do
-    addr1 c@@ addr2 c@@ -
-    ?dup-if
-      unloop exit
-    then
-    addr1 char+ TO addr1
-    addr2 char+ TO addr2
-  loop
-  u1 u2 - ;
- @end example
- Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
- every loop iteration. @code{strcmp} is a typical example of the
- readability problems of using @code{TO}. When you start reading
- @code{strcmp}, you think that @code{addr1} refers to the start of the
- string. Only near the end of the loop you realize that it is something
- else.
- This can be avoided by defining two locals at the start of the loop that
- are initialized with the right value for the current iteration.
  @example
- : strcmp @{ addr1 u1 addr2 u2 -- n @}
+ : ^ ( n1 u -- n )
-  addr1 addr2
+ \ n = the uth power of u1
-  u1 u2 min 0
+swap 0 u+do
-  ?do @{ s1 s2 @}
+     over *
-    s1 c@@ s2 c@@ -
-    ?dup-if
-      unloop exit
-    then
-    s1 char+ s2 char+
   loop
-drop
+   nip ;
-  u1 u2 - ;
+2 ^ .
+3 ^ .
  @end example
- Here it is clear from the start that @code{s1} has a different value
- in every loop iteration.
- @node Implementation,  , Programming Style, Gforth locals
+ @code{U+do} (from @file{compat/loops.fs}, if your Forth system doesn't
- @subsubsection Implementation
+ have it) takes two numbers of the stack @code{( u3 u4 -- )}, and then
- @cindex locals implementation
+ performs the code between @code{u+do} and @code{loop} for @code{u3-u4}
- @cindex implementation of locals
+ times (or not at all, if @code{u3-u4<0}).
- @cindex locals stack
+ You can see the stack effect design rules at work in the stack effect of
- Gforth uses an extra locals stack. The most compelling reason for
+ the loop start words: Since the start value of the loop is more
- this is that the return stack is not float-aligned; using an extra stack
+ frequently constant than the end value, the start value is passed on
- also eliminates the problems and restrictions of using the return stack
+ the top-of-stack.
- as locals stack. Like the other stacks, the locals stack grows toward
- lower addresses. A few primitives allow an efficient implementation:
- doc-@local#
+ You can access the counter of a counted loop with @code{i}:
- doc-f@local#
- doc-laddr#
- doc-lp+!#
- doc-lp!
- doc->l
- doc-f>l
- In addition to these primitives, some specializations of these
+ @example
- primitives for commonly occurring inline arguments are provided for
+ : fac ( u -- u! )
- efficiency reasons, e.g., @code{@@local0} as specialization of
+swap 1+ 1 u+do
- @code{@@local#} for the inline argument 0. The following compiling words
+     i *
- compile the right specialized version, or the general version, as
+   loop ;
- appropriate:
+fac .
+fac .
+ @end example
- doc-compile-@local
+ There is also @code{+do}, which expects signed numbers (important for
- doc-compile-f@local
+ deciding whether to enter the loop).
- doc-compile-lp+!
- Combinations of conditional branches and @code{lp+!#} like
+ @assignment
- @code{?branch-lp+!#} (the locals pointer is only changed if the branch
+ Write a definition for computing the nth Fibonacci number.
- is taken) are provided for efficiency and correctness in loops.
+ @endassignment
- A special area in the dictionary space is reserved for keeping the
+ You can also use increments other than 1:
- local variable names. @code{@{} switches the dictionary pointer to this
- area and @code{@}} switches it back and generates the locals
- initializing code. @code{W:} etc.@ are normal defining words. This
- special area is cleared at the start of every colon definition.
- @cindex wordlist for defining locals
+ @example
- A special feature of Gforth's dictionary is used to implement the
+ : up2 ( n1 n2 -- )
- definition of locals without type specifiers: every wordlist (aka
+   +do
- vocabulary) has its own methods for searching
+     i .
- etc. (@pxref{Wordlists}). For the present purpose we defined a wordlist
++loop ;
- with a special search method: When it is searched for a word, it
+0 up2
- actually creates that word using @code{W:}. @code{@{} changes the search
- order to first search the wordlist containing @code{@}}, @code{W:} etc.,
- and then the wordlist for defining locals without type specifiers.
- The lifetime rules support a stack discipline within a colon
+ : down2 ( n1 n2 -- )
- definition: The lifetime of a local is either nested with other locals
+   -do
- lifetimes or it does not overlap them.
+     i .
+-loop ;
+10 down2
+ @end example
- At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
+ Reference: @ref{Counted Loops}.
- pointer manipulation is generated. Between control structure words
- locals definitions can push locals onto the locals stack. @code{AGAIN}
- is the simplest of the other three control flow words. It has to
- restore the locals stack depth of the corresponding @code{BEGIN}
- before branching. The code looks like this:
- @format
- @code{lp+!#} current-locals-size @minus{} dest-locals-size
- @code{branch} <begin>
- @end format
- @code{UNTIL} is a little more complicated: If it branches back, it
- must adjust the stack just like @code{AGAIN}. But if it falls through,
- the locals stack must not be changed. The compiler generates the
- following code:
- @format
- @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
- @end format
- The locals stack pointer is only adjusted if the branch is taken.
- @code{THEN} can produce somewhat inefficient code:
+ @node Recursion Tutorial, Leaving definitions or loops Tutorial, Counted loops Tutorial, Tutorial
- @format
+ @section Recursion
- @code{lp+!#} current-locals-size @minus{} orig-locals-size
+ @cindex recursion tutorial
- <orig target>:
- @code{lp+!#} orig-locals-size @minus{} new-locals-size
- @end format
- The second @code{lp+!#} adjusts the locals stack pointer from the
- level at the @var{orig} point to the level after the @code{THEN}. The
- first @code{lp+!#} adjusts the locals stack pointer from the current
- level to the level at the orig point, so the complete effect is an
- adjustment from the current level to the right level after the
- @code{THEN}.
- @cindex locals information on the control-flow stack
+ Usually the name of a definition is not visible in the definition; but
- @cindex control-flow stack items, locals information
+ earlier definitions are usually visible:
- In a conventional Forth implementation a dest control-flow stack entry
- is just the target address and an orig entry is just the address to be
- patched. Our locals implementation adds a wordlist to every orig or dest
- item. It is the list of locals visible (or assumed visible) at the point
- described by the entry. Our implementation also adds a tag to identify
- the kind of entry, in particular to differentiate between live and dead
- (reachable and unreachable) orig entries.
- A few unusual operations have to be performed on locals wordlists:
+ @example
+0 / . \ "Floating-point unidentified fault" in Gforth on most platforms
+ : / ( n1 n2 -- n )
+   dup 0= if
+     -10 throw \ report division by zero
+   endif
+   /           \ old version
+ ;
+0 /
+ @end example
- doc-common-list
+ For recursive definitions you can use @code{recursive} (non-standard) or
- doc-sub-list?
+ @code{recurse}:
- doc-list-size
- Several features of our locals wordlist implementation make these
+ @example
- operations easy to implement: The locals wordlists are organised as
+ : fac1 ( n -- n! ) recursive
- linked lists; the tails of these lists are shared, if the lists
+  dup 0> if
- contain some of the same locals; and the address of a name is greater
+    dup 1- fac1 *
- than the address of the names behind it in the list.
+  else
+    drop 1
+  endif ;
+fac1 .
- Another important implementation detail is the variable
+ : fac2 ( n -- n! )
- @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
+  dup 0> if
- determine if they can be reached directly or only through the branch
+    dup 1- recurse *
- that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
+  else
- @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
+    drop 1
- definition, by @code{BEGIN} and usually by @code{THEN}.
+  endif ;
+fac2 .
+ @end example
- Counted loops are similar to other loops in most respects, but
+ @assignment
- @code{LEAVE} requires special attention: It performs basically the same
+ Write a recursive definition for computing the nth Fibonacci number.
- service as @code{AHEAD}, but it does not create a control-flow stack
+ @endassignment
- entry. Therefore the information has to be stored elsewhere;
- traditionally, the information was stored in the target fields of the
- branches created by the @code{LEAVE}s, by organizing these fields into a
- linked list. Unfortunately, this clever trick does not provide enough
- space for storing our extended control flow information. Therefore, we
- introduce another stack, the leave stack. It contains the control-flow
- stack entries for all unresolved @code{LEAVE}s.
- Local names are kept until the end of the colon definition, even if
+ Reference (including indirect recursion): @xref{Calls and returns}.
- they are no longer visible in any control-flow path. In a few cases
- this may lead to increased space needs for the locals name area, but
- usually less than reclaiming this space would cost in code size.
- @node ANS Forth locals,  , Gforth locals, Locals
+ @node Leaving definitions or loops Tutorial, Return Stack Tutorial, Recursion Tutorial, Tutorial
- @subsection ANS Forth locals
+ @section Leaving definitions or loops
- @cindex locals, ANS Forth style
+ @cindex leaving definitions, tutorial
+ @cindex leaving loops, tutorial
- The ANS Forth locals wordset does not define a syntax for locals, but
+ @code{EXIT} exits the current definition right away.  For every counted
- words that make it possible to define various syntaxes. One of the
+ loop that is left in this way, an @code{UNLOOP} has to be performed
- possible syntaxes is a subset of the syntax we used in the Gforth locals
+ before the @code{EXIT}:
- wordset, i.e.:
+ @c !! real examples
  @example
- @{ local1 local2 ... -- comment @}
+ : ...
- @end example
+  ... u+do
- or
+    ... if
- @example
+      ... unloop exit
- @{ local1 local2 ... @}
+    endif
+    ...
+  loop
+  ... ;
  @end example
- The order of the locals corresponds to the order in a stack comment. The
+ @code{LEAVE} leaves the innermost counted loop right away:
- restrictions are:
- @itemize @bullet
+ @example
- @item
+ : ...
- Locals can only be cell-sized values (no type specifiers are allowed).
+  ... u+do
- @item
+    ... if
- Locals can be defined only outside control structures.
+      ... leave
- @item
+    endif
- Locals can interfere with explicit usage of the return stack. For the
+    ...
- exact (and long) rules, see the standard. If you don't use return stack
+  loop
- accessing words in a definition using locals, you will be all right. The
+  ... ;
- purpose of this rule is to make locals implementation on the return
+ @end example
- stack easier.
- @item
- The whole definition must be in one line.
- @end itemize
- Locals defined in this way behave like @code{VALUE}s (@xref{Simple
+ @c !! example
- Defining Words}). I.e., they are initialized from the stack. Using their
- name produces their value. Their value can be changed using @code{TO}.
- Since this syntax is supported by Gforth directly, you need not do
+ Reference: @ref{Calls and returns}, @ref{Counted Loops}.
- anything to use it. If you want to port a program using this syntax to
- another ANS Forth system, use @file{compat/anslocal.fs} to implement the
- syntax on the other system.
- Note that a syntax shown in the standard, section A.13 looks
- similar, but is quite different in having the order of locals
- reversed. Beware!
- The ANS Forth locals wordset itself consists of the following word
+ @node Return Stack Tutorial, Memory Tutorial, Leaving definitions or loops Tutorial, Tutorial
+ @section Return Stack
+ @cindex return stack tutorial
- doc-(local)
+ In addition to the data stack Forth also has a second stack, the return
+ stack; most Forth systems store the return addresses of procedure calls
+ there (thus its name).  Programmers can also use this stack:
- The ANS Forth locals extension wordset defines a syntax, but it is so
+ @example
- awful that we strongly recommend not to use it. We have implemented this
+ : foo ( n1 n2 -- )
- syntax to make porting to Gforth easy, but do not document it here. The
+  .s
- problem with this syntax is that the locals are defined in an order
+  >r .s
- reversed with respect to the standard stack comment notation, making
+  r@@ .
- programs harder to read, and easier to misread and miswrite. The only
+  >r .s
- merit of this syntax is that it is easy to implement using the ANS Forth
+  r@@ .
- locals wordset.
+  r> .
+  r@@ .
+  r> . ;
+2 foo
+ @end example
- @node Defining Words, Structures, Locals, Words
+ @code{>r} takes an element from the data stack and pushes it onto the
- @section Defining Words
+ return stack; conversely, @code{r>} moves an elementm from the return to
- @cindex defining words
+ the data stack; @code{r@@} pushes a copy of the top of the return stack
+ on the return stack.
- @menu
+ Forth programmers usually use the return stack for storing data
- * Simple Defining Words::
+ temporarily, if using the data stack alone would be too complex, and
- * Colon Definitions::
+ factoring and locals are not an option:
- * User-defined Defining Words::
- * Supplying names::
- * Interpretation and Compilation Semantics::
- @end menu
- @node Simple Defining Words, Colon Definitions, Defining Words, Defining Words
+ @example
- @subsection Simple Defining Words
+ : 2swap ( x1 x2 x3 x4 -- x3 x4 x1 x2 )
- @cindex simple defining words
+  rot >r rot r> ;
- @cindex defining words, simple
+ @end example
- doc-constant
+ The return address of the definition and the loop control parameters of
- doc-2constant
+ counted loops usually reside on the return stack, so you have to take
- doc-fconstant
+ all items, that you have pushed on the return stack in a colon
- doc-variable
+ definition or counted loop, from the return stack before the definition
- doc-2variable
+ or loop ends.  You cannot access items that you pushed on the return
- doc-fvariable
+ stack outside some definition or loop within the definition of loop.
- doc-create
- doc-user
- doc-value
- doc-to
- doc-defer
- doc-is
- @node Colon Definitions, User-defined Defining Words, Simple Defining Words, Defining Words
+ If you miscount the return stack items, this usually ends in a crash:
- @subsection Colon Definitions
- @cindex colon definitions
  @example
- : name ( ... -- ... )
+ : crash ( n -- )
-     word1 word2 word3 ;
+   >r ;
+crash
  @end example
- creates a word called @code{name}, that, upon execution, executes
+ You cannot mix using locals and using the return stack (according to the
- @code{word1 word2 word3}. @code{name} is a @dfn{(colon) definition}.
+ standard; Gforth has no problem).  However, they solve the same
+ problems, so this shouldn't be an issue.
- The explanation above is somewhat superficial. @xref{Interpretation and
+ @assignment
- Compilation Semantics} for an in-depth discussion of some of the issues
+ Can you rewrite any of the definitions you wrote until now in a better
- involved.
+ way using the return stack?
+ @endassignment
- doc-:
+ Reference: @ref{Return stack}.
- doc-;
- @node User-defined Defining Words, Supplying names, Colon Definitions, Defining Words
- @subsection User-defined Defining Words
- @cindex user-defined defining words
- @cindex defining words, user-defined
- You can create new defining words simply by wrapping defining-time code
+ @node Memory Tutorial, Characters and Strings Tutorial, Return Stack Tutorial, Tutorial
- around existing defining words and putting the sequence in a colon
+ @section Memory
- definition.
+ @cindex memory access/allocation tutorial
- @cindex @code{CREATE} ... @code{DOES>}
+ You can create a global variable @code{v} with
- If you want the words defined with your defining words to behave
- differently from words defined with standard defining words, you can
- write your defining word like this:
  @example
- : def-word ( "name" -- )
+ variable v ( -- addr )
-     Create @var{code1}
+ @end example
- DOES> ( ... -- ... )
-     @var{code2} ;
- def-word name
+ @code{v} pushes the address of a cell in memory on the stack.  This cell
+ was reserved by @code{variable}.  You can use @code{!} (store) to store
+ values into this cell and @code{@@} (fetch) to load the value from the
+ stack into memory:
+ @example
+ v .
+v ! .s
+ v @@ .
  @end example
- Technically, this fragment defines a defining word @code{def-word}, and
+ You can see a raw dump of memory with @code{dump}:
- a word @code{name}; when you execute @code{name}, the address of the
- body of @code{name} is put on the data stack and @var{code2} is executed
- (the address of the body of @code{name} is the address @code{HERE}
- returns immediately after the @code{CREATE}).
- In other words, if you make the following definitions:
+ @example
+ v 1 cells .s dump
+ @end example
+ @code{Cells ( n1 -- n2 )} gives you the number of bytes (or, more
+ generally, address units (aus)) that @code{n1 cells} occupy.  You can
+ also reserve more memory:
  @example
- : def-word1 ( "name" -- )
+ create v2 20 cells allot
-     Create @var{code1} ;
+ v2 20 cells dump
+ @end example
- : action1 ( ... -- ... )
+ creates a word @code{v2} and reserves 20 uninitialized cells; the
-     @var{code2} ;
+ address pushed by @code{v2} points to the start of these 20 cells.  You
+ can use address arithmetic to access these cells:
+ @example
+v2 5 cells + !
+ v2 20 cells dump
+ @end example
+ You can reserve and initialize memory with @code{,}:
- def-word name1
+ @example
+ create v3
+, 4 , 3 , 2 , 1 ,
+ v3 @@ .
+ v3 cell+ @@ .
+ v3 2 cells + @@ .
+ v3 5 cells dump
  @end example
- Using @code{name1 action1} is equivalent to using @code{name}.
+ @assignment
+ Write a definition @code{vsum ( addr u -- n )} that computes the sum of
+ @code{u} cells, with the first of these cells at @code{addr}, the next
+ one at @code{addr cell+} etc.
+ @endassignment
- E.g., you can implement @code{Constant} in this way:
+ You can also reserve memory without creating a new word:
  @example
- : constant ( w "name" -- )
+ here 10 cells allot .
-     create ,
+ here .
- DOES> ( -- w )
-     @@ ;
  @end example
- When you create a constant with @code{5 constant five}, first a new word
+ @code{Here} pushes the start address of the memory area.  You should
- @code{five} is created, then the value 5 is laid down in the body of
+ store it somewhere, or you will have a hard time finding the memory area
- @code{five} with @code{,}. When @code{five} is invoked, the address of
+ again.
- the body is put on the stack, and @code{@@} retrieves the value 5.
- @cindex stack effect of @code{DOES>}-parts
+ @code{Allot} manages dictionary memory.  The dictionary memory contains
- @cindex @code{DOES>}-parts, stack effect
+ the system's data structures for words etc. on Gforth and most other
- In the example above the stack comment after the @code{DOES>} specifies
+ Forth systems.  It is managed like a stack: You can free the memory that
- the stack effect of the defined words, not the stack effect of the
+ you have just @code{allot}ed with
- following code (the following code expects the address of the body on
- the top of stack, which is not reflected in the stack comment). This is
- the convention that I use and recommend (it clashes a bit with using
- locals declarations for stack effect specification, though).
- @subsubsection Applications of @code{CREATE..DOES>}
+ @example
- @cindex @code{CREATE} ... @code{DOES>}, applications
+ -10 cells allot
+ here .
+ @end example
- You may wonder how to use this feature. Here are some usage patterns:
+ Note that you cannot do this if you have created a new word in the
+ meantime (because then your @code{allot}ed memory is no longer on the
+ top of the dictionary ``stack'').
+ Alternatively, you can use @code{allocate} and @code{free} which allow
+ freeing memory in any order:
- @cindex factoring similar colon definitions
- When you see a sequence of code occurring several times, and you can
- identify a meaning, you will factor it out as a colon definition. When
- you see similar colon definitions, you can factor them using
- @code{CREATE..DOES>}. E.g., an assembler usually defines several words
- that look very similar:
  @example
- : ori, ( reg-target reg-source n -- )
+cells allocate throw .s
-asm-reg-reg-imm ;
+cells allocate throw .s
- : andi, ( reg-target reg-source n -- )
+ swap
-asm-reg-reg-imm ;
+ free throw
+ free throw
  @end example
- This could be factored with:
+ The @code{throw}s deal with errors (e.g., out of memory).
+ And there is also a
+ @uref{http://www.complang.tuwien.ac.at/forth/garbage-collection.zip,
+ garbage collector}, which eliminates the need to @code{free} memory
+ explicitly.
+ Reference: @ref{Memory}.
+ @node Characters and Strings Tutorial, Alignment Tutorial, Memory Tutorial, Tutorial
+ @section Characters and Strings
+ @cindex strings tutorial
+ @cindex characters tutorial
+ On the stack characters take up a cell, like numbers.  In memory they
+ have their own size (one 8-bit byte on most systems), and therefore
+ require their own words for memory access:
  @example
- : reg-reg-imm ( op-code -- )
+ create v4
-     create ,
+c, 97 c, 108 c, 108 c, 111 c,
- DOES> ( reg-target reg-source n -- )
+ v4 4 chars + c@@ .
-     @@ asm-reg-reg-imm ;
+ v4 5 chars dump
+ @end example
-reg-reg-imm ori,
+ The preferred representation of strings on the stack is @code{addr
-reg-reg-imm andi,
+ u-count}, where @code{addr} is the address of the first character and
+ @code{u-count} is the number of characters in the string.
+ @example
+ v4 5 type
  @end example
- @cindex currying
+ You get a string constant with
- Another view of @code{CREATE..DOES>} is to consider it as a crude way to
- supply a part of the parameters for a word (known as @dfn{currying} in
- the functional language community). E.g., @code{+} needs two
- parameters. Creating versions of @code{+} with one parameter fixed can
- be done like this:
  @example
- : curry+ ( n1 -- )
+ s" hello, world" .s
-     create ,
+ type
- DOES> ( n2 -- n1+n2 )
+ @end example
-     @@ + ;
-curry+ 3+
+ Make sure you have a space between @code{s"} and the string; @code{s"}
- -2 curry+ 2-
+ is a normal Forth word and must be delimited with white space (try what
+ happens when you remove the space).
+ However, this interpretive use of @code{s"} is quite restricted: the
+ string exists only until the next call of @code{s"} (some Forth systems
+ keep more than one of these strings, but usually they still have a
+ limited lifetime).
+ @example
+ s" hello," s" world" .s
+ type
+ type
  @end example
- @subsubsection The gory details of @code{CREATE..DOES>}
+ You can also use @code{s"} in a definition, and the resulting
- @cindex @code{CREATE} ... @code{DOES>}, details
+ strings then live forever (well, for as long as the definition):
- doc-does>
+ @example
+ : foo s" hello," s" world" ;
+ foo .s
+ type
+ type
+ @end example
+ @assignment
+ @code{Emit ( c -- )} types @code{c} as character (not a number).
+ Implement @code{type ( addr u -- )}.
+ @endassignment
+ Reference: @ref{Memory Blocks}.
+ @node Alignment Tutorial, Interpretation and Compilation Semantics and Immediacy Tutorial, Characters and Strings Tutorial, Tutorial
+ @section Alignment
+ @cindex alignment tutorial
+ @cindex memory alignment tutorial
+ On many processors cells have to be aligned in memory, if you want to
+ access them with @code{@@} and @code{!} (and even if the processor does
+ not require alignment, access to aligned cells is faster).
+ @code{Create} aligns @code{here} (i.e., the place where the next
+ allocation will occur, and that the @code{create}d word points to).
+ Likewise, the memory produced by @code{allocate} starts at an aligned
+ address.  Adding a number of @code{cells} to an aligned address produces
+ another aligned address.
+ However, address arithmetic involving @code{char+} and @code{chars} can
+ create an address that is not cell-aligned.  @code{Aligned ( addr --
+ a-addr )} produces the next aligned address:
- @cindex @code{DOES>} in a separate definition
- This means that you need not use @code{CREATE} and @code{DOES>} in the
- same definition; E.g., you can put the @code{DOES>}-part in a separate
- definition. This allows us to, e.g., select among different DOES>-parts:
  @example
- : does1
+ v3 char+ aligned .s @@ .
- DOES> ( ... -- ... )
+ v3 char+ .s @@ .
-     ... ;
+ @end example
- : does2
+ Similarly, @code{align} advances @code{here} to the next aligned
- DOES> ( ... -- ... )
+ address:
-     ... ;
- : def-word ( ... -- ... )
+ @example
-     create ...
+ create v5 97 c,
-     IF
+ here .
-        does1
+ align here .
-     ELSE
+,
-        does2
-     ENDIF ;
  @end example
- @cindex @code{DOES>} in interpretation state
+ Note that you should use aligned addresses even if your processor does
- In a standard program you can apply a @code{DOES>}-part only if the last
+ not require them, if you want your program to be portable.
- word was defined with @code{CREATE}. In Gforth, the @code{DOES>}-part
- will override the behaviour of the last word defined in any case. In a
+ Reference: @ref{Address arithmetic}.
- standard program, you can use @code{DOES>} only in a colon
- definition. In Gforth, you can also use it in interpretation state, in a
- kind of one-shot mode:
+ @node Interpretation and Compilation Semantics and Immediacy Tutorial, Execution Tokens Tutorial, Alignment Tutorial, Tutorial
+ @section Interpretation and Compilation Semantics and Immediacy
+ @cindex semantics tutorial
+ @cindex interpretation semantics tutorial
+ @cindex compilation semantics tutorial
+ @cindex immediate, tutorial
+ When a word is compiled, it behaves differently from being interpreted.
+ E.g., consider @code{+}:
  @example
- CREATE name ( ... -- ... )
+2 + .
-   @var{initialization}
+ : foo + ;
- DOES>
-   @var{code} ;
  @end example
- This is equivalent to the standard
+ These two behaviours are known as compilation and interpretation
+ semantics.  For normal words (e.g., @code{+}), the compilation semantics
+ is to append the interpretation semantics to the currently defined word
+ (@code{foo} in the example above).  I.e., when @code{foo} is executed
+ later, the interpretation semantics of @code{+} (i.e., adding two
+ numbers) will be performed.
+ However, there are words with non-default compilation semantics, e.g.,
+ the control-flow words like @code{if}.  You can use @code{immediate} to
+ change the compilation semantics of the last defined word to be equal to
+ the interpretation semantics:
  @example
- :noname
+ : [FOO] ( -- )
- DOES>
+. ; immediate
-     @var{code} ;
- CREATE name EXECUTE ( ... -- ... )
+ [FOO]
-     @var{initialization}
+ : bar ( -- )
+   [FOO] ;
+ bar
+ see bar
  @end example
- You can get the address of the body of a word with
+ Two conventions to mark words with non-default compilation semnatics are
+ names with brackets (more frequently used) and to write them all in
+ upper case (less frequently used).
- doc->body
+ In Gforth (and many other systems) you can also remove the
+ interpretation semantics with @code{compile-only} (the compilation
+ semantics is derived from the original interpretation semantics):
- @node Supplying names, Interpretation and Compilation Semantics, User-defined Defining Words, Defining Words
+ @example
- @subsection Supplying names for the defined words
+ : flip ( -- )
- @cindex names for defined words
+. ; compile-only \ but not immediate
- @cindex defining words, name parameter
+ flip
- @cindex defining words, name given in a string
+ : flop ( -- )
- By default, defining words take the names for the defined words from the
+  flip ;
- input stream. Sometimes you want to supply the name from a string. You
+ flop
- can do this with
+ @end example
- doc-nextname
+ In this example the interpretation semantics of @code{flop} is equal to
+ the original interpretation semantics of @code{flip}.
- E.g.,
+ The text interpreter has two states: in interpret state, it performs the
+ interpretation semantics of words it encounters; in compile state, it
+ performs the compilation semantics of these words.
+ Among other things, @code{:} switches into compile state, and @code{;}
+ switches back to interpret state.  They contain the factors @code{]}
+ (switch to compile state) and @code{[} (switch to interpret state), that
+ do nothing but switch the state.
  @example
- s" foo" nextname create
+ : xxx ( -- )
+   [ 5 . ]
+ ;
+ xxx
+ see xxx
  @end example
- is equivalent to
+ These brackets are also the source of the naming convention mentioned
+ above.
+ Reference: @ref{Interpretation and Compilation Semantics}.
+ @node Execution Tokens Tutorial, Exceptions Tutorial, Interpretation and Compilation Semantics and Immediacy Tutorial, Tutorial
+ @section Execution Tokens
+ @cindex execution tokens tutorial
+ @cindex XT tutorial
+ @code{' word} gives you the execution token (XT) of a word.  The XT is a
+ cell representing the interpretation semantics of a word.  You can
+ execute this semantics with @code{execute}:
  @example
- create foo
+ ' + .s
+2 rot execute .
  @end example
- @cindex defining words without name
+ The XT is similar to a function pointer in C.  However, parameter
- Sometimes you want to define a word without a name. You can do this with
+ passing through the stack makes it a little more flexible:
- doc-noname
+ @example
+ : map-array ( ... addr u xt -- ... )
+ \ executes xt ( ... x -- ... ) for every element of the array starting
+ \ at addr and containing u elements
+   @{ xt @}
+   cells over + swap ?do
+     i @@ xt execute
+cells +loop ;
- @cindex execution token of last defined word
+ create a 3 , 4 , 2 , -1 , 4 ,
- To make any use of the newly defined word, you need its execution
+ a 5 ' . map-array .s
- token. You can get it with
+a 5 ' + map-array .
+ s" max-n" environment? drop .s
+ a 5 ' min map-array .
+ @end example
- doc-lastxt
+ You can use map-array with the XTs of words that consume one element
+ more than they produce.  In theory you can also use it with other XTs,
+ but the stack effect then depends on the size of the array, which is
+ hard to understand.
+ Since XTs are cell-sized, you can store them in memory and manipulate
+ them on the stack like other cells.  You can also compile the XT into a
+ word with @code{compile,}:
- E.g., you can initialize a deferred word with an anonymous colon
- definition:
  @example
- Defer deferred
+ : foo1 ( n1 n2 -- n )
- noname : ( ... -- ... )
+    [ ' + compile, ] ;
-   ... ;
+ see foo
- lastxt IS deferred
  @end example
- @code{lastxt} also works when the last word was not defined as
+ This is non-standard, because @code{compile,} has no compilation
- @code{noname}.
+ semantics in the standard, but it works in good Forth systems.  For the
+ broken ones, use
- The standard has also recognized the need for anonymous words and
+ @example
- provides
+ : [compile,] compile, ; immediate
- doc-:noname
+ : foo1 ( n1 n2 -- n )
+    [ ' + ] [compile,] ;
+ see foo
+ @end example
+ @code{'} is a word with default compilation semantics; it parses the
+ next word when its interpretation semantics are executed, not during
+ compilation:
- This leaves the execution token for the word on the stack after the
- closing @code{;}. You can rewrite the last example with @code{:noname}:
  @example
- Defer deferred
+ : foo ( -- xt )
- :noname ( ... -- ... )
+   ' ;
-   ... ;
+ see foo
- IS deferred
+ : bar ( ... "word" -- ... )
+   ' execute ;
+ see bar
+2 bar + .
  @end example
- @node Interpretation and Compilation Semantics,  , Supplying names, Defining Words
+ You often want to parse a word during compilation and compile its XT so
- @subsection Interpretation and Compilation Semantics
+ it will be pushed on the stack at run-time.  @code{[']} does this:
- @cindex semantics, interpretation and compilation
+ @example
+ : xt-+ ( -- xt )
+   ['] + ;
+ see xt-+
+2 xt-+ execute .
+ @end example
+ Many programmers tend to see @code{'} and the word it parses as one
+ unit, and expect it to behave like @code{[']} when compiled, and are
+ confused by the actual behaviour.  If you are, just remember that the
+ Forth system just takes @code{'} as one unit and has no idea that it is
+ a parsing word (attempts to convenience programmers in this issue have
+ usually resulted in even worse pitfalls, see
+ @uref{http://www.complang.tuwien.ac.at/papers/ertl98.ps.gz,
+ @code{State}-smartness---Why it is evil and How to Exorcise it}).
+ Note that the state of the interpreter does not come into play when
+ creating and executing XTs.  I.e., even when you execute @code{'} in
+ compile state, it still gives you the interpretation semantics.  And
+ whatever that state is, @code{execute} performs the semantics
+ represented by the XT (i.e., for XTs produced with @code{'} the
+ interpretation semantics).
+ Reference: @ref{Tokens for Words}.
+ @node Exceptions Tutorial, Defining Words Tutorial, Execution Tokens Tutorial, Tutorial
+ @section Exceptions
+ @cindex exceptions tutorial
+ @code{throw ( n -- )} causes an exception unless n is zero.
+ @example
+throw .s
+throw .s
+ @end example
+ @code{catch ( ... xt -- ... n )} behaves similar to @code{execute}, but
+ it catches exceptions and pushes the number of the exception on the
+ stack (or 0, if the xt executed without exception).  If there was an
+ exception, the stacks have the same depth as when entering @code{catch}:
+ @example
+ .s
+0 ' / catch .s
+2 ' / catch .s
+ @end example
+ @assignment
+ Try the same with @code{execute} instead of @code{catch}.
+ @endassignment
+ @code{Throw} always jumps to the dynamically next enclosing
+ @code{catch}, even if it has to leave several call levels to achieve
+ this:
+ @example
+ : foo 100 throw ;
+ : foo1 foo ." after foo" ;
+ : bar ['] foo1 catch ;
+ bar .
+ @end example
+ It is often important to restore a value upon leaving a definition, even
+ if the definition is left through an exception.  You can ensure this
+ like this:
+ @example
+ : ...
+    save-x
+    ['] word-changing-x catch ( ... n )
+    restore-x
+    ( ... n ) throw ;
+ @end example
+ Gforth provides an alternative syntax in addition to @code{catch}:
+ @code{try ... recover ... endtry}.  If the code between @code{try} and
+ @code{recover} has an exception, the stack depths are restored, the
+ exception number is pushed on the stack, and the code between
+ @code{recover} and @code{endtry} is performed.  E.g., the definition for
+ @code{catch} is
+ @example
+ : catch ( x1 .. xn xt -- y1 .. ym 0 / z1 .. zn error ) \ exception
+   try
+     execute 0
+   recover
+     nip
+   endtry ;
+ @end example
+ The equivalent to the restoration code above is
+ @example
+ : ...
+   save-x
+   try
+     word-changing-x
+   end-try
+   restore-x
+   throw ;
+ @end example
+ As you can see, the @code{recover} part is optional.
+ Reference: @ref{Exception Handling}.
+ @node Defining Words Tutorial, Arrays and Records Tutorial, Exceptions Tutorial, Tutorial
+ @section Defining Words
+ @cindex defining words tutorial
+ @cindex does> tutorial
+ @cindex create...does> tutorial
+ @c before semantics?
+ @code{:}, @code{create}, and @code{variable} are definition words: They
+ define other words.  @code{Constant} is another definition word:
+ @example
+constant foo
+ foo .
+ @end example
+ You can also use the prefixes @code{2} (double-cell) and @code{f}
+ (floating point) with @code{variable} and @code{constant}.
+ You can also define your own defining words.  E.g.:
+ @example
+ : variable ( "name" -- )
+   create 0 , ;
+ @end example
+ You can also define defining words that create words that do something
+ other than just producing their address:
+ @example
+ : constant ( n "name" -- )
+   create ,
+ does> ( -- n )
+   ( addr ) @@ ;
+constant foo
+ foo .
+ @end example
+ The definition of @code{constant} above ends at the @code{does>}; i.e.,
+ @code{does>} replaces @code{;}, but it also does something else: It
+ changes the last defined word such that it pushes the address of the
+ body of the word and then performs the code after the @code{does>}
+ whenever it is called.
+ In the example above, @code{constant} uses @code{,} to store 5 into the
+ body of @code{foo}.  When @code{foo} executes, it pushes the address of
+ the body onto the stack, then (in the code after the @code{does>})
+ fetches the 5 from there.
+ The stack comment near the @code{does>} reflects the stack effect of the
+ defined word, not the stack effect of the code after the @code{does>}
+ (the difference is that the code expects the address of the body that
+ the stack comment does not show).
+ You can use these definition words to do factoring in cases that involve
+ (other) definition words.  E.g., a field offset is always added to an
+ address.  Instead of defining
+ @example
+cells constant offset-field1
+ @end example
+ and using this like
+ @example
+ ( addr ) offset-field1 +
+ @end example
+ you can define a definition word
+ @example
+ : simple-field ( n "name" -- )
+   create ,
+ does> ( n1 -- n1+n )
+   ( addr ) @@ + ;
+ @end example
+ Definition and use of field offsets now look like this:
+ @example
+cells simple-field field1
+ create mystruct 4 cells allot
+ mystruct .s field1 .s drop
+ @end example
+ If you want to do something with the word without performing the code
+ after the @code{does>}, you can access the body of a @code{create}d word
+ with @code{>body ( xt -- addr )}:
+ @example
+ : value ( n "name" -- )
+   create ,
+ does> ( -- n1 )
+   @@ ;
+ : to ( n "name" -- )
+   ' >body ! ;
+value foo
+ foo .
+to foo
+ foo .
+ @end example
+ @assignment
+ Define @code{defer ( "name" -- )}, which creates a word that stores an
+ XT (at the start the XT of @code{abort}), and upon execution
+ @code{execute}s the XT.  Define @code{is ( xt "name" -- )} that stores
+ @code{xt} into @code{name}, a word defined with @code{defer}.  Indirect
+ recursion is one application of @code{defer}.
+ @endassignment
+ Reference: @ref{User-defined Defining Words}.
+ @node Arrays and Records Tutorial, POSTPONE Tutorial, Defining Words Tutorial, Tutorial
+ @section Arrays and Records
+ @cindex arrays tutorial
+ @cindex records tutorial
+ @cindex structs tutorial
+ Forth has no standard words for defining data structures such as arrays
+ and records (structs in C terminology), but you can build them yourself
+ based on address arithmetic.  You can also define words for defining
+ arrays and records (@pxref{Defining Words Tutorial,, Defining Words}).
+ One of the first projects a Forth newcomer sets out upon when learning
+ about defining words is an array defining word (possibly for
+ n-dimensional arrays).  Go ahead and do it, I did it, too; you will
+ learn something from it.  However, don't be disappointed when you later
+ learn that you have little use for these words (inappropriate use would
+ be even worse).  I have not yet found a set of useful array words yet;
+ the needs are just too diverse, and named, global arrays (the result of
+ naive use of defining words) are often not flexible enough (e.g.,
+ consider how to pass them as parameters).  Another such project is a set
+ of words to help dealing with strings.
+ On the other hand, there is a useful set of record words, and it has
+ been defined in @file{compat/struct.fs}; these words are predefined in
+ Gforth.  They are explained in depth elsewhere in this manual (see
+ @pxref{Structures}).  The @code{simple-field} example above is
+ simplified variant of fields in this package.
+ @node POSTPONE Tutorial, Literal Tutorial, Arrays and Records Tutorial, Tutorial
+ @section @code{POSTPONE}
+ @cindex postpone tutorial
+ You can compile the compilation semantics (instead of compiling the
+ interpretation semantics) of a word with @code{POSTPONE}:
+ @example
+ : MY-+ ( Compilation: -- ; Run-time of compiled code: n1 n2 -- n )
+  POSTPONE + ; immediate
+ : foo ( n1 n2 -- n )
+  MY-+ ;
+2 foo .
+ see foo
+ @end example
+ During the definition of @code{foo} the text interpreter performs the
+ compilation semantics of @code{MY-+}, which performs the compilation
+ semantics of @code{+}, i.e., it compiles @code{+} into @code{foo}.
+ This example also displays separate stack comments for the compilation
+ semantics and for the stack effect of the compiled code.  For words with
+ default compilation semantics these stack effects are usually not
+ displayed; the stack effect of the compilation semantics is always
+ @code{( -- )} for these words, the stack effect for the compiled code is
+ the stack effect of the interpretation semantics.
+ Note that the state of the interpreter does not come into play when
+ performing the compilation semantics in this way.  You can also perform
+ it interpretively, e.g.:
+ @example
+ : foo2 ( n1 n2 -- n )
+  [ MY-+ ] ;
+2 foo .
+ see foo
+ @end example
+ However, there are some broken Forth systems where this does not always
+ work, and therefore this practice was been declared non-standard in
+.
+ @c !! repair.fs
+ Here is another example for using @code{POSTPONE}:
+ @example
+ : MY-- ( Compilation: -- ; Run-time of compiled code: n1 n2 -- n )
+  POSTPONE negate POSTPONE + ; immediate compile-only
+ : bar ( n1 n2 -- n )
+   MY-- ;
+1 bar .
+ see bar
+ @end example
+ You can define @code{ENDIF} in this way:
+ @example
+ : ENDIF ( Compilation: orig -- )
+   POSTPONE then ; immediate
+ @end example
+ @assignment
+ Write @code{MY-2DUP} that has compilation semantics equivalent to
+ @code{2dup}, but compiles @code{over over}.
+ @endassignment
+ @c !! @xref{Macros} for reference
+ @node Literal Tutorial, Advanced macros Tutorial, POSTPONE Tutorial, Tutorial
+ @section @code{Literal}
+ @cindex literal tutorial
+ You cannot @code{POSTPONE} numbers:
+ @example
+ : [FOO] POSTPONE 500 ; immediate
+ @end example
+ Instead, you can use @code{LITERAL (compilation: n --; run-time: -- n )}:
+ @example
+ : [FOO] ( compilation: --; run-time: -- n )
+POSTPONE literal ; immediate
+ : flip [FOO] ;
+ flip .
+ see flip
+ @end example
+ @code{LITERAL} consumes a number at compile-time (when it's compilation
+ semantics are executed) and pushes it at run-time (when the code it
+ compiled is executed).  A frequent use of @code{LITERAL} is to compile a
+ number computed at compile time into the current word:
+ @example
+ : bar ( -- n )
+   [ 2 2 + ] literal ;
+ see bar
+ @end example
+ @assignment
+ Write @code{]L} which allows writing the example above as @code{: bar (
+ -- n ) [ 2 2 + ]L ;}
+ @endassignment
+ @c !! @xref{Macros} for reference
+ @node Advanced macros Tutorial, Compilation Tokens Tutorial, Literal Tutorial, Tutorial
+ @section Advanced macros
+ @cindex macros, advanced tutorial
+ @cindex run-time code generation, tutorial
+ Reconsider @code{map-array} from @ref{Execution Tokens Tutorial,,
+ Execution Tokens}.  It frequently performs @code{execute}, a relatively
+ expensive operation in some Forth implementations.  You can use
+ @code{compile,} and @code{POSTPONE} to eliminate these @code{execute}s
+ and produce a word that contains the word to be performed directly:
+ @c use ]] ... [[
+ @example
+ : compile-map-array ( compilation: xt -- ; run-time: ... addr u -- ... )
+ \ at run-time, execute xt ( ... x -- ... ) for each element of the
+ \ array beginning at addr and containing u elements
+   @{ xt @}
+   POSTPONE cells POSTPONE over POSTPONE + POSTPONE swap POSTPONE ?do
+     POSTPONE i POSTPONE @@ xt compile,
+cells POSTPONE literal POSTPONE +loop ;
+ : sum-array ( addr u -- n )
+rot rot [ ' + compile-map-array ] ;
+ see sum-array
+ a 5 sum-array .
+ @end example
+ You can use the full power of Forth for generating the code; here's an
+ example where the code is generated in a loop:
+ @example
+ : compile-vmul-step ( compilation: n --; run-time: n1 addr1 -- n2 addr2 )
+ \ n2=n1+(addr1)*n, addr2=addr1+cell
+   POSTPONE tuck POSTPONE @@
+   POSTPONE literal POSTPONE * POSTPONE +
+   POSTPONE swap POSTPONE cell+ ;
+ : compile-vmul ( compilation: addr1 u -- ; run-time: addr2 -- n )
+ \ n=v1*v2 (inner product), where the v_i are represented as addr_i u
+postpone literal postpone swap
+   [ ' compile-vmul-step compile-map-array ]
+   postpone drop ;
+ see compile-vmul
+ : a-vmul ( addr -- n )
+ \ n=a*v, where v is a vector that's as long as a and starts at addr
+  [ a 5 compile-vmul ] ;
+ see a-vmul
+ a a-vmul .
+ @end example
+ This example uses @code{compile-map-array} to show off, but you could
+ also use @code{map-array} instead (try it now!).
+ You can use this technique for efficient multiplication of large
+ matrices.  In matrix multiplication, you multiply every line of one
+ matrix with every column of the other matrix.  You can generate the code
+ for one line once, and use it for every column.  The only downside of
+ this technique is that it is cumbersome to recover the memory consumed
+ by the generated code when you are done (and in more complicated cases
+ it is not possible portably).
+ @c !! @xref{Macros} for reference
+ @node Compilation Tokens Tutorial, Wordlists and Search Order Tutorial, Advanced macros Tutorial, Tutorial
+ @section Compilation Tokens
+ @cindex compilation tokens, tutorial
+ @cindex CT, tutorial
+ This section is Gforth-specific.  You can skip it.
+ @code{' word compile,} compiles the interpretation semantics.  For words
+ with default compilation semantics this is the same as performing the
+ compilation semantics.  To represent the compilation semantics of other
+ words (e.g., words like @code{if} that have no interpretation
+ semantics), Gforth has the concept of a compilation token (CT,
+ consisting of two cells), and words @code{comp'} and @code{[comp']}.
+ You can perform the compilation semantics represented by a CT with
+ @code{execute}:
+ @example
+ : foo2 ( n1 n2 -- n )
+    [ comp' + execute ] ;
+ see foo
+ @end example
+ You can compile the compilation semantics represented by a CT with
+ @code{postpone,}:
+ @example
+ : foo3 ( -- )
+   [ comp' + postpone, ] ;
+ see foo3
+ @end example
+ @code{[ comp' word postpone, ]} is equivalent to @code{POSTPONE word}.
+ @code{comp'} is particularly useful for words that have no
+ interpretation semantics:
+ @example
+ ' if
+ comp' if .s 2drop
+ @end example
+ Reference: @ref{Tokens for Words}.
+ @node Wordlists and Search Order Tutorial,  , Compilation Tokens Tutorial, Tutorial
+ @section Wordlists and Search Order
+ @cindex wordlists tutorial
+ @cindex search order, tutorial
+ The dictionary is not just a memory area that allows you to allocate
+ memory with @code{allot}, it also contains the Forth words, arranged in
+ several wordlists.  When searching for a word in a wordlist,
+ conceptually you start searching at the youngest and proceed towards
+ older words (in reality most systems nowadays use hash-tables); i.e., if
+ you define a word with the same name as an older word, the new word
+ shadows the older word.
+ Which wordlists are searched in which order is determined by the search
+ order.  You can display the search order with @code{order}.  It displays
+ first the search order, starting with the wordlist searched first, then
+ it displays the wordlist that will contain newly defined words.
+ You can create a new, empty wordlist with @code{wordlist ( -- wid )}:
+ @example
+ wordlist constant mywords
+ @end example
+ @code{Set-current ( wid -- )} sets the wordlist that will contain newly
+ defined words (the @emph{current} wordlist):
+ @example
+ mywords set-current
+ order
+ @end example
+ Gforth does not display a name for the wordlist in @code{mywords}
+ because this wordlist was created anonymously with @code{wordlist}.
+ You can get the current wordlist with @code{get-current ( -- wid)}.  If
+ you want to put something into a specific wordlist without overall
+ effect on the current wordlist, this typically looks like this:
+ @example
+ get-current mywords set-current ( wid )
+ create someword
+ ( wid ) set-current
+ @end example
+ You can write the search order with @code{set-order ( wid1 .. widn n --
+ )} and read it with @code{get-order ( -- wid1 .. widn n )}.  The first
+ searched wordlist is topmost.
+ @example
+ get-order mywords swap 1+ set-order
+ order
+ @end example
+ Yes, the order of wordlists in the output of @code{order} is reversed
+ from stack comments and the output of @code{.s} and thus unintuitive.
+ @assignment
+ Define @code{>order ( wid -- )} with adds @code{wid} as first searched
+ wordlist to the search order.  Define @code{previous ( -- )}, which
+ removes the first searched wordlist from the search order.  Experiment
+ with boundary conditions (you will see some crashes or situations that
+ are hard or impossible to leave).
+ @endassignment
+ The search order is a powerful foundation for providing features similar
+ to Modula-2 modules and C++ namespaces.  However, trying to modularize
+ programs in this way has disadvantages for debugging and reuse/factoring
+ that overcome the advantages in my experience (I don't do huge projects,
+ though).  These disadvantages are not so clear in other
+ languages/programming environments, because these langauges are not so
+ strong in debugging and reuse.
+ @c !! example
+ Reference: @ref{Word Lists}.
+ @c ******************************************************************
+ @node Introduction, Words, Tutorial, Top
+ @comment node-name,     next,           previous, up
+ @chapter An Introduction to ANS Forth
+ @cindex Forth - an introduction
+ The primary purpose of this manual is to document Gforth. However, since
+ Forth is not a widely-known language and there is a lack of up-to-date
+ teaching material, it seems worthwhile to provide some introductory
+ material.  For other sources of Forth-related
+ information, see @ref{Forth-related information}.
+ The examples in this section should work on any ANS Forth; the
+ output shown was produced using Gforth. Each example attempts to
+ reproduce the exact output that Gforth produces. If you try out the
+ examples (and you should), what you should type is shown @kbd{like this}
+ and Gforth's response is shown @code{like this}. The single exception is
+ that, where the example shows @key{RET} it means that you should
+ press the ``carriage return'' key. Unfortunately, some output formats for
+ this manual cannot show the difference between @kbd{this} and
+ @code{this} which will make trying out the examples harder (but not
+ impossible).
+ Forth is an unusual language. It provides an interactive development
+ environment which includes both an interpreter and compiler. Forth
+ programming style encourages you to break a problem down into many
+ @cindex factoring
+ small fragments (@dfn{factoring}), and then to develop and test each
+ fragment interactively. Forth advocates assert that breaking the
+ edit-compile-test cycle used by conventional programming languages can
+ lead to great productivity improvements.
+ @menu
+ * Introducing the Text Interpreter::
+ * Stacks and Postfix notation::
+ * Your first definition::
+ * How does that work?::
+ * Forth is written in Forth::
+ * Review - elements of a Forth system::
+ * Where to go next::
+ * Exercises::
+ @end menu
+ @comment ----------------------------------------------
+ @node Introducing the Text Interpreter, Stacks and Postfix notation, Introduction, Introduction
+ @section Introducing the Text Interpreter
+ @cindex text interpreter
+ @cindex outer interpreter
+ @c IMO this is too detailed and the pace is too slow for
+ @c an introduction.  If you know German, take a look at
+ @c http://www.complang.tuwien.ac.at/anton/lvas/skriptum-stack.html
+ @c to see how I do it - anton
+ @c nac-> Where I have accepted your comments 100% and modified the text
+ @c accordingly, I have deleted your comments. Elsewhere I have added a
+ @c response like this to attempt to rationalise what I have done. Of
+ @c course, this is a very clumsy mechanism for something that would be
+ @c done far more efficiently over a beer. Please delete any dialogue
+ @c you consider closed.
+ When you invoke the Forth image, you will see a startup banner printed
+ and nothing else (if you have Gforth installed on your system, try
+ invoking it now, by typing @kbd{gforth@key{RET}}). Forth is now running
+ its command line interpreter, which is called the @dfn{Text Interpreter}
+ (also known as the @dfn{Outer Interpreter}).  (You will learn a lot
+ about the text interpreter as you read through this chapter, for more
+ detail @pxref{The Text Interpreter}).
+ Although it's not obvious, Forth is actually waiting for your
+ input. Type a number and press the @key{RET} key:
+ @example
+ @kbd{45@key{RET}}  ok
+ @end example
+ Rather than give you a prompt to invite you to input something, the text
+ interpreter prints a status message @i{after} it has processed a line
+ of input. The status message in this case (``@code{ ok}'' followed by
+ carriage-return) indicates that the text interpreter was able to process
+ all of your input successfully. Now type something illegal:
+ @example
+ @kbd{qwer341@key{RET}}
+ :1: Undefined word
+ qwer341
+ ^^^^^^^
+ $400D2BA8 Bounce
+ $400DBDA8 no.extensions
+ @end example
+ The exact text, other than the ``Undefined word'' may differ slightly on
+ your system, but the effect is the same; when the text interpreter
+ detects an error, it discards any remaining text on a line, resets
+ certain internal state and prints an error message. For a detailed description of error messages see @ref{Error
+ messages}.
+ The text interpreter waits for you to press carriage-return, and then
+ processes your input line. Starting at the beginning of the line, it
+ breaks the line into groups of characters separated by spaces. For each
+ group of characters in turn, it makes two attempts to do something:
+ @itemize @bullet
+ @item
+ @cindex name dictionary
+ It tries to treat it as a command. It does this by searching a @dfn{name
+ dictionary}. If the group of characters matches an entry in the name
+ dictionary, the name dictionary provides the text interpreter with
+ information that allows the text interpreter perform some actions. In
+ Forth jargon, we say that the group
+ @cindex word
+ @cindex definition
+ @cindex execution token
+ @cindex xt
+ of characters names a @dfn{word}, that the dictionary search returns an
+ @dfn{execution token (xt)} corresponding to the @dfn{definition} of the
+ word, and that the text interpreter executes the xt. Often, the terms
+ @dfn{word} and @dfn{definition} are used interchangeably.
+ @item
+ If the text interpreter fails to find a match in the name dictionary, it
+ tries to treat the group of characters as a number in the current number
+ base (when you start up Forth, the current number base is base 10). If
+ the group of characters legitimately represents a number, the text
+ interpreter pushes the number onto a stack (we'll learn more about that
+ in the next section).
+ @end itemize
+ If the text interpreter is unable to do either of these things with any
+ group of characters, it discards the group of characters and the rest of
+ the line, then prints an error message. If the text interpreter reaches
+ the end of the line without error, it prints the status message ``@code{ ok}''
+ followed by carriage-return.
+ This is the simplest command we can give to the text interpreter:
+ @example
+ @key{RET}  ok
+ @end example
+ The text interpreter did everything we asked it to do (nothing) without
+ an error, so it said that everything is ``@code{ ok}''. Try a slightly longer
+ command:
+ @example
+ @kbd{12 dup fred dup@key{RET}}
+ :1: Undefined word
+dup fred dup
+        ^^^^
+ $400D2BA8 Bounce
+ $400DBDA8 no.extensions
+ @end example
+ When you press the carriage-return key, the text interpreter starts to
+ work its way along the line:
+ @itemize @bullet
+ @item
+ When it gets to the space after the @code{2}, it takes the group of
+ characters @code{12} and looks them up in the name
+ dictionary@footnote{We can't tell if it found them or not, but assume
+ for now that it did not}. There is no match for this group of characters
+ in the name dictionary, so it tries to treat them as a number. It is
+ able to do this successfully, so it puts the number, 12, ``on the stack''
+ (whatever that means).
+ @item
+ The text interpreter resumes scanning the line and gets the next group
+ of characters, @code{dup}. It looks it up in the name dictionary and
+ (you'll have to take my word for this) finds it, and executes the word
+ @code{dup} (whatever that means).
+ @item
+ Once again, the text interpreter resumes scanning the line and gets the
+ group of characters @code{fred}. It looks them up in the name
+ dictionary, but can't find them. It tries to treat them as a number, but
+ they don't represent any legal number.
+ @end itemize
+ At this point, the text interpreter gives up and prints an error
+ message. The error message shows exactly how far the text interpreter
+ got in processing the line. In particular, it shows that the text
+ interpreter made no attempt to do anything with the final character
+ group, @code{dup}, even though we have good reason to believe that the
+ text interpreter would have no problem looking that word up and
+ executing it a second time.
+ @comment ----------------------------------------------
+ @node Stacks and Postfix notation, Your first definition, Introducing the Text Interpreter, Introduction
+ @section Stacks, postfix notation and parameter passing
+ @cindex text interpreter
+ @cindex outer interpreter
+ In procedural programming languages (like C and Pascal), the
+ building-block of programs is the @dfn{function} or @dfn{procedure}. These
+ functions or procedures are called with @dfn{explicit parameters}. For
+ example, in C we might write:
+ @example
+ total = total + new_volume(length,height,depth);
+ @end example
+ @noindent
+ where new_volume is a function-call to another piece of code, and total,
+ length, height and depth are all variables. length, height and depth are
+ parameters to the function-call.
+ In Forth, the equivalent of the function or procedure is the
+ @dfn{definition} and parameters are implicitly passed between
+ definitions using a shared stack that is visible to the
+ programmer. Although Forth does support variables, the existence of the
+ stack means that they are used far less often than in most other
+ programming languages. When the text interpreter encounters a number, it
+ will place (@dfn{push}) it on the stack. There are several stacks (the
+ actual number is implementation-dependent ...) and the particular stack
+ used for any operation is implied unambiguously by the operation being
+ performed. The stack used for all integer operations is called the @dfn{data
+ stack} and, since this is the stack used most commonly, references to
+ ``the data stack'' are often abbreviated to ``the stack''.
+ The stacks have a last-in, first-out (LIFO) organisation. If you type:
+ @example
+ @kbd{1 2 3@key{RET}}  ok
+ @end example
+ Then this instructs the text interpreter to placed three numbers on the
+ (data) stack. An analogy for the behaviour of the stack is to take a
+ pack of playing cards and deal out the ace (1), 2 and 3 into a pile on
+ the table. The 3 was the last card onto the pile (``last-in'') and if
+ you take a card off the pile then, unless you're prepared to fiddle a
+ bit, the card that you take off will be the 3 (``first-out''). The
+ number that will be first-out of the stack is called the @dfn{top of
+ stack}, which
+ @cindex TOS definition
+ is often abbreviated to @dfn{TOS}.
+ To understand how parameters are passed in Forth, consider the
+ behaviour of the definition @code{+} (pronounced ``plus''). You will not
+ be surprised to learn that this definition performs addition. More
+ precisely, it adds two number together and produces a result. Where does
+ it get the two numbers from? It takes the top two numbers off the
+ stack. Where does it place the result? On the stack. You can act-out the
+ behaviour of @code{+} with your playing cards like this:
+ @itemize @bullet
+ @item
+ Pick up two cards from the stack on the table
+ @item
+ Stare at them intently and ask yourself ``what @i{is} the sum of these two
+ numbers''
+ @item
+ Decide that the answer is 5
+ @item
+ Shuffle the two cards back into the pack and find a 5
+ @item
+ Put a 5 on the remaining ace that's on the table.
+ @end itemize
+ If you don't have a pack of cards handy but you do have Forth running,
+ you can use the definition @code{.s} to show the current state of the stack,
+ without affecting the stack. Type:
+ @example
+ @kbd{clearstack 1 2 3@key{RET}} ok
+ @kbd{.s@key{RET}} <3> 1 2 3  ok
+ @end example
+ The text interpreter looks up the word @code{clearstack} and executes
+ it; it tidies up the stack and removes any entries that may have been
+ left on it by earlier examples. The text interpreter pushes each of the
+ three numbers in turn onto the stack. Finally, the text interpreter
+ looks up the word @code{.s} and executes it. The effect of executing
+ @code{.s} is to print the ``<3>'' (the total number of items on the stack)
+ followed by a list of all the items on the stack; the item on the far
+ right-hand side is the TOS.
+ You can now type:
+ @example
+ @kbd{+ .s@key{RET}} <2> 1 5  ok
+ @end example
+ @noindent
+ which is correct; there are now 2 items on the stack and the result of
+ the addition is 5.
+ If you're playing with cards, try doing a second addition: pick up the
+ two cards, work out that their sum is 6, shuffle them into the pack,
+ look for a 6 and place that on the table. You now have just one item on
+ the stack. What happens if you try to do a third addition? Pick up the
+ first card, pick up the second card -- ah! There is no second card. This
+ is called a @dfn{stack underflow} and consitutes an error. If you try to
+ do the same thing with Forth it will report an error (probably a Stack
+ Underflow or an Invalid Memory Address error).
+ The opposite situation to a stack underflow is a @dfn{stack overflow},
+ which simply accepts that there is a finite amount of storage space
+ reserved for the stack. To stretch the playing card analogy, if you had
+ enough packs of cards and you piled the cards up on the table, you would
+ eventually be unable to add another card; you'd hit the ceiling. Gforth
+ allows you to set the maximum size of the stacks. In general, the only
+ time that you will get a stack overflow is because a definition has a
+ bug in it and is generating data on the stack uncontrollably.
+ There's one final use for the playing card analogy. If you model your
+ stack using a pack of playing cards, the maximum number of items on
+ your stack will be 52 (I assume you didn't use the Joker). The maximum
+ @i{value} of any item on the stack is 13 (the King). In fact, the only
+ possible numbers are positive integer numbers 1 through 13; you can't
+ have (for example) 0 or 27 or 3.52 or -2. If you change the way you
+ think about some of the cards, you can accommodate different
+ numbers. For example, you could think of the Jack as representing 0,
+ the Queen as representing -1 and the King as representing -2. Your
+ @i{range} remains unchanged (you can still only represent a total of 13
+ numbers) but the numbers that you can represent are -2 through 10.
+ In that analogy, the limit was the amount of information that a single
+ stack entry could hold, and Forth has a similar limit. In Forth, the
+ size of a stack entry is called a @dfn{cell}. The actual size of a cell is
+ implementation dependent and affects the maximum value that a stack
+ entry can hold. A Standard Forth provides a cell size of at least
+-bits, and most desktop systems use a cell size of 32-bits.
+ Forth does not do any type checking for you, so you are free to
+ manipulate and combine stack items in any way you wish. A convenient way
+ of treating stack items is as 2's complement signed integers, and that
+ is what Standard words like @code{+} do. Therefore you can type:
+ @example
+ @kbd{-5 12 + .s@key{RET}} <1> 7  ok
+ @end example
+ If you use numbers and definitions like @code{+} in order to turn Forth
+ into a great big pocket calculator, you will realise that it's rather
+ different from a normal calculator. Rather than typing 2 + 3 = you had
+ to type 2 3 + (ignore the fact that you had to use @code{.s} to see the
+ result). The terminology used to describe this difference is to say that
+ your calculator uses @dfn{Infix Notation} (parameters and operators are
+ mixed) whilst Forth uses @dfn{Postfix Notation} (parameters and
+ operators are separate), also called @dfn{Reverse Polish Notation}.
+ Whilst postfix notation might look confusing to begin with, it has
+ several important advantages:
+ @itemize @bullet
+ @item
+ it is unambiguous
+ @item
+ it is more concise
+ @item
+ it fits naturally with a stack-based system
+ @end itemize
+ To examine these claims in more detail, consider these sums:
+ @example
++ 5 * 4 =
+* 5 + 6 =
+ @end example
+ If you're just learning maths or your maths is very rusty, you will
+ probably come up with the answer 44 for the first and 26 for the
+ second. If you are a bit of a whizz at maths you will remember the
+ @i{convention} that multiplication takes precendence over addition, and
+ you'd come up with the answer 26 both times. To explain the answer 26
+ to someone who got the answer 44, you'd probably rewrite the first sum
+ like this:
+ @example
++ (5 * 4) =
+ @end example
+ If what you really wanted was to perform the addition before the
+ multiplication, you would have to use parentheses to force it.
+ If you did the first two sums on a pocket calculator you would probably
+ get the right answers, unless you were very cautious and entered them using
+ these keystroke sequences:
++ 5 = * 4 =
+* 5 = + 6 =
+ Postfix notation is unambiguous because the order that the operators
+ are applied is always explicit; that also means that parentheses are
+ never required. The operators are @i{active} (the act of quoting the
+ operator makes the operation occur) which removes the need for ``=''.
+ The sum 6 + 5 * 4 can be written (in postfix notation) in two
+ equivalent ways:
+ @example
+5 4 * +      or:
+4 * 6 +
+ @end example
+ An important thing that you should notice about this notation is that
+ the @i{order} of the numbers does not change; if you want to subtract
+from 10 you type @code{10 2 -}.
+ The reason that Forth uses postfix notation is very simple to explain: it
+ makes the implementation extremely simple, and it follows naturally from
+ using the stack as a mechanism for passing parameters. Another way of
+ thinking about this is to realise that all Forth definitions are
+ @i{active}; they execute as they are encountered by the text
+ interpreter. The result of this is that the syntax of Forth is trivially
+ simple.
+ @comment ----------------------------------------------
+ @node Your first definition, How does that work?, Stacks and Postfix notation, Introduction
+ @section Your first Forth definition
+ @cindex first definition
+ Until now, the examples we've seen have been trivial; we've just been
+ using Forth as a bigger-than-pocket calculator. Also, each calculation
+ we've shown has been a ``one-off'' -- to repeat it we'd need to type it in
+ again@footnote{That's not quite true. If you press the up-arrow key on
+ your keyboard you should be able to scroll back to any earlier command,
+ edit it and re-enter it.} In this section we'll see how to add new
+ words to Forth's vocabulary.
+ The easiest way to create a new word is to use a @dfn{colon
+ definition}. We'll define a few and try them out before worrying too
+ much about how they work. Try typing in these examples; be careful to
+ copy the spaces accurately:
+ @example
+ : add-two 2 + . ;
+ : greet ." Hello and welcome" ;
+ : demo 5 add-two ;
+ @end example
+ @noindent
+ Now try them out:
+ @example
+ @kbd{greet@key{RET}} Hello and welcome  ok
+ @kbd{greet greet@key{RET}} Hello and welcomeHello and welcome  ok
+ @kbd{4 add-two@key{RET}} 6  ok
+ @kbd{demo@key{RET}} 7  ok
+ @kbd{9 greet demo add-two@key{RET}} Hello and welcome7 11  ok
+ @end example
+ The first new thing that we've introduced here is the pair of words
+ @code{:} and @code{;}. These are used to start and terminate a new
+ definition, respectively. The first word after the @code{:} is the name
+ for the new definition.
+ As you can see from the examples, a definition is built up of words that
+ have already been defined; Forth makes no distinction between
+ definitions that existed when you started the system up, and those that
+ you define yourself.
+ The examples also introduce the words @code{.} (dot), @code{."}
+ (dot-quote) and @code{dup} (dewp). Dot takes the value from the top of
+ the stack and displays it. It's like @code{.s} except that it only
+ displays the top item of the stack and it is destructive; after it has
+ executed, the number is no longer on the stack. There is always one
+ space printed after the number, and no spaces before it. Dot-quote
+ defines a string (a sequence of characters) that will be printed when
+ the word is executed. The string can contain any printable characters
+ except @code{"}. A @code{"} has a special function; it is not a Forth
+ word but it acts as a delimiter (the way that delimiters work is
+ described in the next section). Finally, @code{dup} duplicates the value
+ at the top of the stack. Try typing @code{5 dup .s} to see what it does.
+ We already know that the text interpreter searches through the
+ dictionary to locate names. If you've followed the examples earlier, you
+ will already have a definition called @code{add-two}. Lets try modifying
+ it by typing in a new definition:
+ @example
+ @kbd{: add-two dup . ." + 2 =" 2 + . ;@key{RET}} redefined add-two  ok
+ @end example
+ Forth recognised that we were defining a word that already exists, and
+ printed a message to warn us of that fact. Let's try out the new
+ definition:
+ @example
+ @kbd{9 add-two@key{RET}} 9 + 2 =11  ok
+ @end example
+ @noindent
+ All that we've actually done here, though, is to create a new
+ definition, with a particular name. The fact that there was already a
+ definition with the same name did not make any difference to the way
+ that the new definition was created (except that Forth printed a warning
+ message). The old definition of add-two still exists (try @code{demo}
+ again to see that this is true). Any new definition will use the new
+ definition of @code{add-two}, but old definitions continue to use the
+ version that already existed at the time that they were @code{compiled}.
+ Before you go on to the next section, try defining and redefining some
+ words of your own.
+ @comment ----------------------------------------------
+ @node How does that work?, Forth is written in Forth, Your first definition, Introduction
+ @section How does that work?
+ @cindex parsing words
+ @c That's pretty deep (IMO way too deep) for an introduction. - anton
+ @c Is it a good idea to talk about the interpretation semantics of a
+ @c number? We don't have an xt to go along with it. - anton
+ @c Now that I have eliminated execution semantics, I wonder if it would not
+ @c be better to keep them (or add run-time semantics), to make it easier to
+ @c explain what compilation semantics usually does. - anton
+ @c nac-> I removed the term ``default compilation sematics'' from the
+ @c introductory chapter. Removing ``execution semantics'' was making
+ @c everything simpler to explain, then I think the use of this term made
+ @c everything more complex again. I replaced it with ``default
+ @c semantics'' (which is used elsewhere in the manual) by which I mean
+ @c ``a definition that has neither the immediate nor the compile-only
+ @c flag set''. I reworded big chunks of the ``how does that work''
+ @c section (and, unusually for me, I think I even made it shorter!).  See
+ @c what you think -- I know I have not addressed your primary concern
+ @c that it is too heavy-going for an introduction. From what I understood
+ @c of your course notes it looks as though they might be a good framework.
+ @c Things that I've tried to capture here are some things that came as a
+ @c great revelation here when I first understood them. Also, I like the
+ @c fact that a very simple code example shows up almost all of the issues
+ @c that you need to understand to see how Forth works. That's unique and
+ @c worthwhile to emphasise.
+ Now we're going to take another look at the definition of @code{add-two}
+ from the previous section. From our knowledge of the way that the text
+ interpreter works, we would have expected this result when we tried to
+ define @code{add-two}:
+ @example
+ @kbd{: add-two 2 + . ;@key{RET}}
+   ^^^^^^^
+ Error: Undefined word
+ @end example
+ The reason that this didn't happen is bound up in the way that @code{:}
+ works. The word @code{:} does two special things. The first special
+ thing that it does prevents the text interpreter from ever seeing the
+ characters @code{add-two}. The text interpreter uses a variable called
+ @cindex modifying >IN
+ @code{>IN} (pronounced ``to-in'') to keep track of where it is in the
+ input line. When it encounters the word @code{:} it behaves in exactly
+ the same way as it does for any other word; it looks it up in the name
+ dictionary, finds its xt and executes it. When @code{:} executes, it
+ looks at the input buffer, finds the word @code{add-two} and advances the
+ value of @code{>IN} to point past it. It then does some other stuff
+ associated with creating the new definition (including creating an entry
+ for @code{add-two} in the name dictionary). When the execution of @code{:}
+ completes, control returns to the text interpreter, which is oblivious
+ to the fact that it has been tricked into ignoring part of the input
+ line.
+ @cindex parsing words
+ Words like @code{:} -- words that advance the value of @code{>IN} and so
+ prevent the text interpreter from acting on the whole of the input line
+ -- are called @dfn{parsing words}.
+ @cindex @code{state} - effect on the text interpreter
+ @cindex text interpreter - effect of state
+ The second special thing that @code{:} does is change the value of a
+ variable called @code{state}, which affects the way that the text
+ interpreter behaves. When Gforth starts up, @code{state} has the value
+, and the text interpreter is said to be @dfn{interpreting}. During a
+ colon definition (started with @code{:}), @code{state} is set to -1 and
+ the text interpreter is said to be @dfn{compiling}.
+ In this example, the text interpreter is compiling when it processes the
+ string ``@code{2 + . ;}''. It still breaks the string down into
+ character sequences in the same way. However, instead of pushing the
+ number @code{2} onto the stack, it lays down (@dfn{compiles}) some magic
+ into the definition of @code{add-two} that will make the number @code{2} get
+ pushed onto the stack when @code{add-two} is @dfn{executed}. Similarly,
+ the behaviours of @code{+} and @code{.} are also compiled into the
+ definition.
+ One category of words don't get compiled. These so-called @dfn{immediate
+ words} get executed (performed @i{now}) regardless of whether the text
+ interpreter is interpreting or compiling. The word @code{;} is an
+ immediate word. Rather than being compiled into the definition, it
+ executes. Its effect is to terminate the current definition, which
+ includes changing the value of @code{state} back to 0.
+ When you execute @code{add-two}, it has a @dfn{run-time effect} that is
+ exactly the same as if you had typed @code{2 + . @key{RET}} outside of a
+ definition.
+ In Forth, every word or number can be described in terms of two
+ properties:
+ @itemize @bullet
+ @item
+ @cindex interpretation semantics
+ Its @dfn{interpretation semantics} describe how it will behave when the
+ text interpreter encounters it in @dfn{interpret} state. The
+ interpretation semantics of a word are represented by an @dfn{execution
+ token}.
+ @item
+ @cindex compilation semantics
+ Its @dfn{compilation semantics} describe how it will behave when the
+ text interpreter encounters it in @dfn{compile} state. The compilation
+ semantics of a word are represented in an implementation-dependent way;
+ Gforth uses a @dfn{compilation token}.
+ @end itemize
+ @noindent
+ Numbers are always treated in a fixed way:
+ @itemize @bullet
+ @item
+ When the number is @dfn{interpreted}, its behaviour is to push the
+ number onto the stack.
+ @item
+ When the number is @dfn{compiled}, a piece of code is appended to the
+ current definition that pushes the number when it runs. (In other words,
+ the compilation semantics of a number are to postpone its interpretation
+ semantics until the run-time of the definition that it is being compiled
+ into.)
+ @end itemize
+ Words don't behave in such a regular way, but most have @i{default
+ semantics} which means that they behave like this:
+ @itemize @bullet
+ @item
+ The @dfn{interpretation semantics} of the word are to do something useful.
+ @item
+ The @dfn{compilation semantics} of the word are to append its
+ @dfn{interpretation semantics} to the current definition (so that its
+ run-time behaviour is to do something useful).
+ @end itemize
+ @cindex immediate words
+ The actual behaviour of any particular word can be controlled by using
+ the words @code{immediate} and @code{compile-only} when the word is
+ defined. These words set flags in the name dictionary entry of the most
+ recently defined word, and these flags are retrieved by the text
+ interpreter when it finds the word in the name dictionary.
+ A word that is marked as @dfn{immediate} has compilation semantics that
+ are identical to its interpretation semantics. In other words, it
+ behaves like this:
+ @itemize @bullet
+ @item
+ The @dfn{interpretation semantics} of the word are to do something useful.
+ @item
+ The @dfn{compilation semantics} of the word are to do something useful
+ (and actually the same thing); i.e., it is executed during compilation.
+ @end itemize
+ Marking a word as @dfn{compile-only} prohibits the text interpreter from
+ performing the interpretation semantics of the word directly; an attempt
+ to do so will generate an error. It is never necessary to use
+ @code{compile-only} (and it is not even part of ANS Forth, though it is
+ provided by many implementations) but it is good etiquette to apply it
+ to a word that will not behave correctly (and might have unexpected
+ side-effects) in interpret state. For example, it is only legal to use
+ the conditional word @code{IF} within a definition. If you forget this
+ and try to use it elsewhere, the fact that (in Gforth) it is marked as
+ @code{compile-only} allows the text interpreter to generate a helpful
+ error message rather than subjecting you to the consequences of your
+ folly.
+ This example shows the difference between an immediate and a
+ non-immediate word:
+ @example
+ : show-state state @@ . ;
+ : show-state-now show-state ; immediate
+ : word1 show-state ;
+ : word2 show-state-now ;
+ @end example
+ The word @code{immediate} after the definition of @code{show-state-now}
+ makes that word an immediate word. These definitions introduce a new
+ word: @code{@@} (pronounced ``fetch''). This word fetches the value of a
+ variable, and leaves it on the stack. Therefore, the behaviour of
+ @code{show-state} is to print a number that represents the current value
+ of @code{state}.
+ When you execute @code{word1}, it prints the number 0, indicating that
+ the system is interpreting. When the text interpreter compiled the
+ definition of @code{word1}, it encountered @code{show-state} whose
+ compilation semantics are to append its interpretation semantics to the
+ current definition. When you execute @code{word1}, it performs the
+ interpretation semantics of @code{show-state}.  At the time that @code{word1}
+ (and therefore @code{show-state}) are executed, the system is
+ interpreting.
+ When you pressed @key{RET} after entering the definition of @code{word2},
+ you should have seen the number -1 printed, followed by ``@code{
+ ok}''. When the text interpreter compiled the definition of
+ @code{word2}, it encountered @code{show-state-now}, an immediate word,
+ whose compilation semantics are therefore to perform its interpretation
+ semantics. It is executed straight away (even before the text
+ interpreter has moved on to process another group of characters; the
+ @code{;} in this example). The effect of executing it are to display the
+ value of @code{state} @i{at the time that the definition of}
+ @code{word2} @i{is being defined}. Printing -1 demonstrates that the
+ system is compiling at this time. If you execute @code{word2} it does
+ nothing at all.
+ @cindex @code{."}, how it works
+ Before leaving the subject of immediate words, consider the behaviour of
+ @code{."} in the definition of @code{greet}, in the previous
+ section. This word is both a parsing word and an immediate word. Notice
+ that there is a space between @code{."} and the start of the text
+ @code{Hello and welcome}, but that there is no space between the last
+ letter of @code{welcome} and the @code{"} character. The reason for this
+ is that @code{."} is a Forth word; it must have a space after it so that
+ the text interpreter can identify it. The @code{"} is not a Forth word;
+ it is a @dfn{delimiter}. The examples earlier show that, when the string
+ is displayed, there is neither a space before the @code{H} nor after the
+ @code{e}. Since @code{."} is an immediate word, it executes at the time
+ that @code{greet} is defined. When it executes, its behaviour is to
+ search forward in the input line looking for the delimiter. When it
+ finds the delimiter, it updates @code{>IN} to point past the
+ delimiter. It also compiles some magic code into the definition of
+ @code{greet}; the xt of a run-time routine that prints a text string. It
+ compiles the string @code{Hello and welcome} into memory so that it is
+ available to be printed later. When the text interpreter gains control,
+ the next word it finds in the input stream is @code{;} and so it
+ terminates the definition of @code{greet}.
+ @comment ----------------------------------------------
+ @node Forth is written in Forth, Review - elements of a Forth system, How does that work?, Introduction
+ @section Forth is written in Forth
+ @cindex structure of Forth programs
+ When you start up a Forth compiler, a large number of definitions
+ already exist. In Forth, you develop a new application using bottom-up
+ programming techniques to create new definitions that are defined in
+ terms of existing definitions. As you create each definition you can
+ test and debug it interactively.
+ If you have tried out the examples in this section, you will probably
+ have typed them in by hand; when you leave Gforth, your definitions will
+ be lost. You can avoid this by using a text editor to enter Forth source
+ code into a file, and then loading code from the file using
+ @code{include} (@pxref{Forth source files}). A Forth source file is
+ processed by the text interpreter, just as though you had typed it in by
+ hand@footnote{Actually, there are some subtle differences -- see
+ @ref{The Text Interpreter}.}.
+ Gforth also supports the traditional Forth alternative to using text
+ files for program entry (@pxref{Blocks}).
+ In common with many, if not most, Forth compilers, most of Gforth is
+ actually written in Forth. All of the @file{.fs} files in the
+ installation directory@footnote{For example,
+ @file{/usr/local/share/gforth...}} are Forth source files, which you can
+ study to see examples of Forth programming.
+ Gforth maintains a history file that records every line that you type to
+ the text interpreter. This file is preserved between sessions, and is
+ used to provide a command-line recall facility. If you enter long
+ definitions by hand, you can use a text editor to paste them out of the
+ history file into a Forth source file for reuse at a later time
+ (for more information @pxref{Command-line editing}).
+ @comment ----------------------------------------------
+ @node Review - elements of a Forth system, Where to go next, Forth is written in Forth, Introduction
+ @section Review - elements of a Forth system
+ @cindex elements of a Forth system
+ To summarise this chapter:
+ @itemize @bullet
+ @item
+ Forth programs use @dfn{factoring} to break a problem down into small
+ fragments called @dfn{words} or @dfn{definitions}.
+ @item
+ Forth program development is an interactive process.
+ @item
+ The main command loop that accepts input, and controls both
+ interpretation and compilation, is called the @dfn{text interpreter}
+ (also known as the @dfn{outer interpreter}).
+ @item
+ Forth has a very simple syntax, consisting of words and numbers
+ separated by spaces or carriage-return characters. Any additional syntax
+ is imposed by @dfn{parsing words}.
+ @item
+ Forth uses a stack to pass parameters between words. As a result, it
+ uses postfix notation.
+ @item
+ To use a word that has previously been defined, the text interpreter
+ searches for the word in the @dfn{name dictionary}.
+ @item
+ Words have @dfn{interpretation semantics} and @dfn{compilation semantics}.
+ @item
+ The text interpreter uses the value of @code{state} to select between
+ the use of the @dfn{interpretation semantics} and the  @dfn{compilation
+ semantics} of a word that it encounters.
+ @item
+ The relationship between the @dfn{interpretation semantics} and
+ @dfn{compilation semantics} for a word
+ depend upon the way in which the word was defined (for example, whether
+ it is an @dfn{immediate} word).
+ @item
+ Forth definitions can be implemented in Forth (called @dfn{high-level
+ definitions}) or in some other way (usually a lower-level language and
+ as a result often called @dfn{low-level definitions}, @dfn{code
+ definitions} or @dfn{primitives}).
+ @item
+ Many Forth systems are implemented mainly in Forth.
+ @end itemize
+ @comment ----------------------------------------------
+ @node Where to go next, Exercises, Review - elements of a Forth system, Introduction
+ @section Where To Go Next
+ @cindex where to go next
+ Amazing as it may seem, if you have read (and understood) this far, you
+ know almost all the fundamentals about the inner workings of a Forth
+ system. You certainly know enough to be able to read and understand the
+ rest of this manual and the ANS Forth document, to learn more about the
+ facilities that Forth in general and Gforth in particular provide. Even
+ scarier, you know almost enough to implement your own Forth system.
+ However, that's not a good idea just yet... better to try writing some
+ programs in Gforth.
+ Forth has such a rich vocabulary that it can be hard to know where to
+ start in learning it. This section suggests a few sets of words that are
+ enough to write small but useful programs. Use the word index in this
+ document to learn more about each word, then try it out and try to write
+ small definitions using it. Start by experimenting with these words:
+ @itemize @bullet
+ @item
+ Arithmetic: @code{+ - * / /MOD */ ABS INVERT}
+ @item
+ Comparison: @code{MIN MAX =}
+ @item
+ Logic: @code{AND OR XOR NOT}
+ @item
+ Stack manipulation: @code{DUP DROP SWAP OVER}
+ @item
+ Loops and decisions: @code{IF ELSE ENDIF ?DO I LOOP}
+ @item
+ Input/Output: @code{. ." EMIT CR KEY}
+ @item
+ Defining words: @code{: ; CREATE}
+ @item
+ Memory allocation words: @code{ALLOT ,}
+ @item
+ Tools: @code{SEE WORDS .S MARKER}
+ @end itemize
+ When you have mastered those, go on to:
+ @itemize @bullet
+ @item
+ More defining words: @code{VARIABLE CONSTANT VALUE TO CREATE DOES>}
+ @item
+ Memory access: @code{@@ !}
+ @end itemize
+ When you have mastered these, there's nothing for it but to read through
+ the whole of this manual and find out what you've missed.
+ @comment ----------------------------------------------
+ @node Exercises,  , Where to go next, Introduction
+ @section Exercises
+ @cindex exercises
+ TODO: provide a set of programming excercises linked into the stuff done
+ already and into other sections of the manual. Provide solutions to all
+ the exercises in a .fs file in the distribution.
+ @c Get some inspiration from Starting Forth and Kelly&Spies.
+ @c excercises:
+ @c 1. take inches and convert to feet and inches.
+ @c 2. take temperature and convert from fahrenheight to celcius;
+ @c    may need to care about symmetric vs floored??
+ @c 3. take input line and do character substitution
+ @c    to encipher or decipher
+ @c 4. as above but work on a file for in and out
+ @c 5. take input line and convert to pig-latin
+ @c
+ @c thing of sets of things to exercise then come up with
+ @c problems that need those things.
+ @c ******************************************************************
+ @node Words, Error messages, Introduction, Top
+ @chapter Forth Words
+ @cindex words
+ @menu
+ * Notation::
+ * Case insensitivity::
+ * Comments::
+ * Boolean Flags::
+ * Arithmetic::
+ * Stack Manipulation::
+ * Memory::
+ * Control Structures::
+ * Defining Words::
+ * Interpretation and Compilation Semantics::
+ * Tokens for Words::
+ * The Text Interpreter::
+ * Word Lists::
+ * Environmental Queries::
+ * Files::
+ * Blocks::
+ * Other I/O::
+ * Locals::
+ * Structures::
+ * Object-oriented Forth::
+ * Programming Tools::
+ * Assembler and Code Words::
+ * Threading Words::
+ * Passing Commands to the OS::
+ * Keeping track of Time::
+ * Miscellaneous Words::
+ @end menu
+ @node Notation, Case insensitivity, Words, Words
+ @section Notation
+ @cindex notation of glossary entries
+ @cindex format of glossary entries
+ @cindex glossary notation format
+ @cindex word glossary entry format
+ The Forth words are described in this section in the glossary notation
+ that has become a de-facto standard for Forth texts:
+ @format
+ @i{word}     @i{Stack effect}   @i{wordset}   @i{pronunciation}
+ @end format
+ @i{Description}
+ @table @var
+ @item word
+ The name of the word.
+ @item Stack effect
+ @cindex stack effect
+ The stack effect is written in the notation @code{@i{before} --
+ @i{after}}, where @i{before} and @i{after} describe the top of
+ stack entries before and after the execution of the word. The rest of
+ the stack is not touched by the word. The top of stack is rightmost,
+ i.e., a stack sequence is written as it is typed in. Note that Gforth
+ uses a separate floating point stack, but a unified stack
+ notation. Also, return stack effects are not shown in @i{stack
+ effect}, but in @i{Description}. The name of a stack item describes
+ the type and/or the function of the item. See below for a discussion of
+ the types.
+ All words have two stack effects: A compile-time stack effect and a
+ run-time stack effect. The compile-time stack-effect of most words is
+ @i{ -- }. If the compile-time stack-effect of a word deviates from
+ this standard behaviour, or the word does other unusual things at
+ compile time, both stack effects are shown; otherwise only the run-time
+ stack effect is shown.
+ @cindex pronounciation of words
+ @item pronunciation
+ How the word is pronounced.
+ @cindex wordset
+ @cindex environment wordset
+ @item wordset
+ The ANS Forth standard is divided into several word sets. A standard
+ system need not support all of them. Therefore, in theory, the fewer
+ word sets your program uses the more portable it will be. However, we
+ suspect that most ANS Forth systems on personal machines will feature
+ all word sets. Words that are not defined in ANS Forth have
+ @code{gforth} or @code{gforth-internal} as word set. @code{gforth}
+ describes words that will work in future releases of Gforth;
+ @code{gforth-internal} words are more volatile. Environmental query
+ strings are also displayed like words; you can recognize them by the
+ @code{environment} in the word set field.
+ @item Description
+ A description of the behaviour of the word.
+ @end table
+ @cindex types of stack items
+ @cindex stack item types
+ The type of a stack item is specified by the character(s) the name
+ starts with:
+ @table @code
+ @item f
+ @cindex @code{f}, stack item type
+ Boolean flags, i.e. @code{false} or @code{true}.
+ @item c
+ @cindex @code{c}, stack item type
+ Char
+ @item w
+ @cindex @code{w}, stack item type
+ Cell, can contain an integer or an address
+ @item n
+ @cindex @code{n}, stack item type
+ signed integer
+ @item u
+ @cindex @code{u}, stack item type
+ unsigned integer
+ @item d
+ @cindex @code{d}, stack item type
+ double sized signed integer
+ @item ud
+ @cindex @code{ud}, stack item type
+ double sized unsigned integer
+ @item r
+ @cindex @code{r}, stack item type
+ Float (on the FP stack)
+ @item a-
+ @cindex @code{a_}, stack item type
+ Cell-aligned address
+ @item c-
+ @cindex @code{c_}, stack item type
+ Char-aligned address (note that a Char may have two bytes in Windows NT)
+ @item f-
+ @cindex @code{f_}, stack item type
+ Float-aligned address
+ @item df-
+ @cindex @code{df_}, stack item type
+ Address aligned for IEEE double precision float
+ @item sf-
+ @cindex @code{sf_}, stack item type
+ Address aligned for IEEE single precision float
+ @item xt
+ @cindex @code{xt}, stack item type
+ Execution token, same size as Cell
+ @item wid
+ @cindex @code{wid}, stack item type
+ Word list ID, same size as Cell
+ @item ior, wior
+ @cindex ior type description
+ @cindex wior type description
+ I/O result code, cell-sized.  In Gforth, you can @code{throw} iors.
+ @item f83name
+ @cindex @code{f83name}, stack item type
+ Pointer to a name structure
+ @item "
+ @cindex @code{"}, stack item type
+ string in the input stream (not on the stack). The terminating character
+ is a blank by default. If it is not a blank, it is shown in @code{<>}
+ quotes.
+ @end table
+ @comment ----------------------------------------------
+ @node Case insensitivity, Comments, Notation, Words
+ @section Case insensitivity
+ @cindex case sensitivity
+ @cindex upper and lower case
+ Gforth is case-insensitive; you can enter definitions and invoke
+ Standard words using upper, lower or mixed case (however,
+ @pxref{core-idef, Implementation-defined options, Implementation-defined
+ options}).
+ ANS Forth only @i{requires} implementations to recognise Standard words
+ when they are typed entirely in upper case. Therefore, a Standard
+ program must use upper case for all Standard words. You can use whatever
+ case you like for words that you define, but in a Standard program you
+ have to use the words in the same case that you defined them.
+ Gforth supports case sensitivity through @code{table}s (case-sensitive
+ wordlists, @pxref{Word Lists}).
+ Two people have asked how to convert Gforth to be case-sensitive; while
+ we think this is a bad idea, you can change all wordlists into tables
+ like this:
+ @example
+ ' table-find forth-wordlist wordlist-map @ !
+ @end example
+ Note that you now have to type the predefined words in the same case
+ that we defined them, which are varying.  You may want to convert them
+ to your favourite case before doing this operation (I won't explain how,
+ because if you are even contemplating doing this, you'd better have
+ enough knowledge of Forth systems to know this already).
+ @node Comments, Boolean Flags, Case insensitivity, Words
+ @section Comments
+ @cindex comments
+ Forth supports two styles of comment; the traditional @i{in-line} comment,
+ @code{(} and its modern cousin, the @i{comment to end of line}; @code{\}.
+ doc-(
+ doc-\
+ doc-\G
+ @node Boolean Flags, Arithmetic, Comments, Words
+ @section Boolean Flags
+ @cindex Boolean flags
+ A Boolean flag is cell-sized. A cell with all bits clear represents the
+ flag @code{false} and a flag with all bits set represents the flag
+ @code{true}. Words that check a flag (for example, @code{IF}) will treat
+ a cell that has @i{any} bit set as @code{true}.
+ @c on and off to Memory?
+ @c true and false to "Bitwise operations" or "Numeric comparison"?
+ doc-true
+ doc-false
+ doc-on
+ doc-off
+ @node Arithmetic, Stack Manipulation, Boolean Flags, Words
+ @section Arithmetic
+ @cindex arithmetic words
+ @cindex division with potentially negative operands
+ Forth arithmetic is not checked, i.e., you will not hear about integer
+ overflow on addition or multiplication, you may hear about division by
+ zero if you are lucky. The operator is written after the operands, but
+ the operands are still in the original order. I.e., the infix @code{2-1}
+ corresponds to @code{2 1 -}. Forth offers a variety of division
+ operators. If you perform division with potentially negative operands,
+ you do not want to use @code{/} or @code{/mod} with its undefined
+ behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
+ former, @pxref{Mixed precision}).
+ @comment TODO discuss the different division forms and the std approach
+ @menu
+ * Single precision::
+ * Double precision::            Double-cell integer arithmetic
+ * Bitwise operations::
+ * Numeric comparison::
+ * Mixed precision::             Operations with single and double-cell integers
+ * Floating Point::
+ @end menu
+ @node Single precision, Double precision, Arithmetic, Arithmetic
+ @subsection Single precision
+ @cindex single precision arithmetic words
+ @c !! cell undefined
+ By default, numbers in Forth are single-precision integers that are one
+ cell in size. They can be signed or unsigned, depending upon how you
+ treat them. For the rules used by the text interpreter for recognising
+ single-precision integers see @ref{Number Conversion}.
+ These words are all defined for signed operands, but some of them also
+ work for unsigned numbers: @code{+}, @code{1+}, @code{-}, @code{1-},
+ @code{*}.
+ doc-+
+ doc-1+
+ doc--
+ doc-1-
+ doc-*
+ doc-/
+ doc-mod
+ doc-/mod
+ doc-negate
+ doc-abs
+ doc-min
+ doc-max
+ doc-floored
+ @node Double precision, Bitwise operations, Single precision, Arithmetic
+ @subsection Double precision
+ @cindex double precision arithmetic words
+ For the rules used by the text interpreter for
+ recognising double-precision integers, see @ref{Number Conversion}.
+ A double precision number is represented by a cell pair, with the most
+ significant cell at the TOS. It is trivial to convert an unsigned single
+ to a double: simply push a @code{0} onto the TOS. Since numbers are
+ represented by Gforth using 2's complement arithmetic, converting a
+ signed single to a (signed) double requires sign-extension across the
+ most significant cell. This can be achieved using @code{s>d}. The moral
+ of the story is that you cannot convert a number without knowing whether
+ it represents an unsigned or a signed number.
+ These words are all defined for signed operands, but some of them also
+ work for unsigned numbers: @code{d+}, @code{d-}.
+ doc-s>d
+ doc-d>s
+ doc-d+
+ doc-d-
+ doc-dnegate
+ doc-dabs
+ doc-dmin
+ doc-dmax
+ @node Bitwise operations, Numeric comparison, Double precision, Arithmetic
+ @subsection Bitwise operations
+ @cindex bitwise operation words
+ doc-and
+ doc-or
+ doc-xor
+ doc-invert
+ doc-lshift
+ doc-rshift
+ doc-2*
+ doc-d2*
+ doc-2/
+ doc-d2/
+ @node Numeric comparison, Mixed precision, Bitwise operations, Arithmetic
+ @subsection Numeric comparison
+ @cindex numeric comparison words
+ Note that the words that compare for equality (@code{= <> 0= 0<> d= d<>
+ d0= d0<>}) work for for both signed and unsigned numbers.
+ doc-<
+ doc-<=
+ doc-<>
+ doc-=
+ doc->
+ doc->=
+ doc-0<
+ doc-0<=
+ doc-0<>
+ doc-0=
+ doc-0>
+ doc-0>=
+ doc-u<
+ doc-u<=
+ @c u<> and u= exist but are the same as <> and =
+ @c doc-u<>
+ @c doc-u=
+ doc-u>
+ doc-u>=
+ doc-within
+ doc-d<
+ doc-d<=
+ doc-d<>
+ doc-d=
+ doc-d>
+ doc-d>=
+ doc-d0<
+ doc-d0<=
+ doc-d0<>
+ doc-d0=
+ doc-d0>
+ doc-d0>=
+ doc-du<
+ doc-du<=
+ @c du<> and du= exist but are the same as d<> and d=
+ @c doc-du<>
+ @c doc-du=
+ doc-du>
+ doc-du>=
+ @node Mixed precision, Floating Point, Numeric comparison, Arithmetic
+ @subsection Mixed precision
+ @cindex mixed precision arithmetic words
+ doc-m+
+ doc-*/
+ doc-*/mod
+ doc-m*
+ doc-um*
+ doc-m*/
+ doc-um/mod
+ doc-fm/mod
+ doc-sm/rem
+ @node Floating Point,  , Mixed precision, Arithmetic
+ @subsection Floating Point
+ @cindex floating point arithmetic words
+ For the rules used by the text interpreter for
+ recognising floating-point numbers see @ref{Number Conversion}.
+ Gforth has a separate floating point stack, but the documentation uses
+ the unified notation.@footnote{It's easy to generate the separate
+ notation from that by just separating the floating-point numbers out:
+ e.g. @code{( n r1 u r2 -- r3 )} becomes @code{( n u -- ) ( F: r1 r2 --
+ r3 )}.}
+ @cindex floating-point arithmetic, pitfalls
+ Floating point numbers have a number of unpleasant surprises for the
+ unwary (e.g., floating point addition is not associative) and even a few
+ for the wary. You should not use them unless you know what you are doing
+ or you don't care that the results you get are totally bogus. If you
+ want to learn about the problems of floating point numbers (and how to
+ avoid them), you might start with @cite{David Goldberg,
+ @uref{http://www.validgh.com/goldberg/paper.ps,What Every Computer
+ Scientist Should Know About Floating-Point Arithmetic}, ACM Computing
+ Surveys 23(1):5@minus{}48, March 1991}.
+ doc-d>f
+ doc-f>d
+ doc-f+
+ doc-f-
+ doc-f*
+ doc-f/
+ doc-fnegate
+ doc-fabs
+ doc-fmax
+ doc-fmin
+ doc-floor
+ doc-fround
+ doc-f**
+ doc-fsqrt
+ doc-fexp
+ doc-fexpm1
+ doc-fln
+ doc-flnp1
+ doc-flog
+ doc-falog
+ doc-f2*
+ doc-f2/
+ doc-1/f
+ doc-precision
+ doc-set-precision
+ @cindex angles in trigonometric operations
+ @cindex trigonometric operations
+ Angles in floating point operations are given in radians (a full circle
+ has 2 pi radians).
+ doc-fsin
+ doc-fcos
+ doc-fsincos
+ doc-ftan
+ doc-fasin
+ doc-facos
+ doc-fatan
+ doc-fatan2
+ doc-fsinh
+ doc-fcosh
+ doc-ftanh
+ doc-fasinh
+ doc-facosh
+ doc-fatanh
+ doc-pi
+ @cindex equality of floats
+ @cindex floating-point comparisons
+ One particular problem with floating-point arithmetic is that comparison
+ for equality often fails when you would expect it to succeed.  For this
+ reason approximate equality is often preferred (but you still have to
+ know what you are doing).  Also note that IEEE NaNs may compare
+ differently from what you might expect.  The comparison words are:
+ doc-f~rel
+ doc-f~abs
+ doc-f~
+ doc-f=
+ doc-f<>
+ doc-f<
+ doc-f<=
+ doc-f>
+ doc-f>=
+ doc-f0<
+ doc-f0<=
+ doc-f0<>
+ doc-f0=
+ doc-f0>
+ doc-f0>=
+ @node Stack Manipulation, Memory, Arithmetic, Words
+ @section Stack Manipulation
+ @cindex stack manipulation words
+ @cindex floating-point stack in the standard
+ Gforth maintains a number of separate stacks:
+ @cindex data stack
+ @cindex parameter stack
+ @itemize @bullet
+ @item
+ A data stack (also known as the @dfn{parameter stack}) -- for
+ characters, cells, addresses, and double cells.
+ @cindex floating-point stack
+ @item
+ A floating point stack -- for holding floating point (FP) numbers.
+ @cindex return stack
+ @item
+ A return stack -- for holding the return addresses of colon
+ definitions and other (non-FP) data.
+ @cindex locals stack
+ @item
+ A locals stack -- for holding local variables.
+ @end itemize
+ @menu
+ * Data stack::
+ * Floating point stack::
+ * Return stack::
+ * Locals stack::
+ * Stack pointer manipulation::
+ @end menu
+ @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
+ @subsection Data stack
+ @cindex data stack manipulation words
+ @cindex stack manipulations words, data stack
+ doc-drop
+ doc-nip
+ doc-dup
+ doc-over
+ doc-tuck
+ doc-swap
+ doc-pick
+ doc-rot
+ doc--rot
+ doc-?dup
+ doc-roll
+ doc-2drop
+ doc-2nip
+ doc-2dup
+ doc-2over
+ doc-2tuck
+ doc-2swap
+ doc-2rot
+ @node Floating point stack, Return stack, Data stack, Stack Manipulation
+ @subsection Floating point stack
+ @cindex floating-point stack manipulation words
+ @cindex stack manipulation words, floating-point stack
+ Whilst every sane Forth has a separate floating-point stack, it is not
+ strictly required; an ANS Forth system could theoretically keep
+ floating-point numbers on the data stack. As an additional difficulty,
+ you don't know how many cells a floating-point number takes. It is
+ reportedly possible to write words in a way that they work also for a
+ unified stack model, but we do not recommend trying it. Instead, just
+ say that your program has an environmental dependency on a separate
+ floating-point stack.
+ doc-floating-stack
+ doc-fdrop
+ doc-fnip
+ doc-fdup
+ doc-fover
+ doc-ftuck
+ doc-fswap
+ doc-fpick
+ doc-frot
+ @node Return stack, Locals stack, Floating point stack, Stack Manipulation
+ @subsection Return stack
+ @cindex return stack manipulation words
+ @cindex stack manipulation words, return stack
+ @cindex return stack and locals
+ @cindex locals and return stack
+ A Forth system is allowed to keep local variables on the
+ return stack. This is reasonable, as local variables usually eliminate
+ the need to use the return stack explicitly. So, if you want to produce
+ a standard compliant program and you are using local variables in a
+ word, forget about return stack manipulations in that word (refer to the
+ standard document for the exact rules).
+ doc->r
+ doc-r>
+ doc-r@
+ doc-rdrop
+ doc-2>r
+ doc-2r>
+ doc-2r@
+ doc-2rdrop
+ @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
+ @subsection Locals stack
+ Gforth uses an extra locals stack.  It is described, along with the
+ reasons for its existence, in @ref{Locals implementation}.
+ @node Stack pointer manipulation,  , Locals stack, Stack Manipulation
+ @subsection Stack pointer manipulation
+ @cindex stack pointer manipulation words
+ @c removed s0 r0 l0 -- they are obsolete aliases for sp0 rp0 lp0
+ doc-sp0
+ doc-sp@
+ doc-sp!
+ doc-fp0
+ doc-fp@
+ doc-fp!
+ doc-rp0
+ doc-rp@
+ doc-rp!
+ doc-lp0
+ doc-lp@
+ doc-lp!
+ @node Memory, Control Structures, Stack Manipulation, Words
+ @section Memory
+ @cindex memory words
+ @menu
+ * Memory model::
+ * Dictionary allocation::
+ * Heap Allocation::
+ * Memory Access::
+ * Address arithmetic::
+ * Memory Blocks::
+ @end menu
+ In addition to the standard Forth memory allocation words, there is also
+ a @uref{http://www.complang.tuwien.ac.at/forth/garbage-collection.zip,
+ garbage collector}.
+ @node Memory model, Dictionary allocation, Memory, Memory
+ @subsection ANS Forth and Gforth memory models
+ @c The ANS Forth description is a mess (e.g., is the heap part of
+ @c the dictionary?), so let's not stick to closely with it.
+ ANS Forth considers a Forth system as consisting of several address
+ spaces, of which only @dfn{data space} is managed and accessible with
+ the memory words.  Memory not necessarily in data space includes the
+ stacks, the code (called code space) and the headers (called name
+ space). In Gforth everything is in data space, but the code for the
+ primitives is usually read-only.
+ Data space is divided into a number of areas: The (data space portion of
+ the) dictionary@footnote{Sometimes, the term @dfn{dictionary} is used to
+ refer to the search data structure embodied in word lists and headers,
+ because it is used for looking up names, just as you would in a
+ conventional dictionary.}, the heap, and a number of system-allocated
+ buffers.
+ @cindex address arithmetic restrictions, ANS vs. Gforth
+ @cindex contiguous regions, ANS vs. Gforth
+ In ANS Forth data space is also divided into contiguous regions.  You
+ can only use address arithmetic within a contiguous region, not between
+ them.  Usually each allocation gives you one contiguous region, but the
+ dictionary allocation words have additional rules (@pxref{Dictionary
+ allocation}).
+ Gforth provides one big address space, and address arithmetic can be
+ performed between any addresses. However, in the dictionary headers or
+ code are interleaved with data, so almost the only contiguous data space
+ regions there are those described by ANS Forth as contiguous; but you
+ can be sure that the dictionary is allocated towards increasing
+ addresses even between contiguous regions.  The memory order of
+ allocations in the heap is platform-dependent (and possibly different
+ from one run to the next).
+ @node Dictionary allocation, Heap Allocation, Memory model, Memory
+ @subsection Dictionary allocation
+ @cindex reserving data space
+ @cindex data space - reserving some
+ Dictionary allocation is a stack-oriented allocation scheme, i.e., if
+ you want to deallocate X, you also deallocate everything
+ allocated after X.
+ @cindex contiguous regions in dictionary allocation
+ The allocations using the words below are contiguous and grow the region
+ towards increasing addresses.  Other words that allocate dictionary
+ memory of any kind (i.e., defining words including @code{:noname}) end
+ the contiguous region and start a new one.
+ In ANS Forth only @code{create}d words are guaranteed to produce an
+ address that is the start of the following contiguous region.  In
+ particular, the cell allocated by @code{variable} is not guaranteed to
+ be contiguous with following @code{allot}ed memory.
+ You can deallocate memory by using @code{allot} with a negative argument
+ (with some restrictions, see @code{allot}). For larger deallocations use
+ @code{marker}.
+ doc-here
+ doc-unused
+ doc-allot
+ doc-c,
+ doc-f,
+ doc-,
+ doc-2,
+ Memory accesses have to be aligned (@pxref{Address arithmetic}). So of
+ course you should allocate memory in an aligned way, too. I.e., before
+ allocating allocating a cell, @code{here} must be cell-aligned, etc.
+ The words below align @code{here} if it is not already.  Basically it is
+ only already aligned for a type, if the last allocation was a multiple
+ of the size of this type and if @code{here} was aligned for this type
+ before.
+ After freshly @code{create}ing a word, @code{here} is @code{align}ed in
+ ANS Forth (@code{maxalign}ed in Gforth).
+ doc-align
+ doc-falign
+ doc-sfalign
+ doc-dfalign
+ doc-maxalign
+ doc-cfalign
+ @node Heap Allocation, Memory Access, Dictionary allocation, Memory
+ @subsection Heap allocation
+ @cindex heap allocation
+ @cindex dynamic allocation of memory
+ @cindex memory-allocation word set
+ @cindex contiguous regions and heap allocation
+ Heap allocation supports deallocation of allocated memory in any
+ order. Dictionary allocation is not affected by it (i.e., it does not
+ end a contiguous region). In Gforth, these words are implemented using
+ the standard C library calls malloc(), free() and resize().
+ The memory region produced by one invocation of @code{allocate} or
+ @code{resize} is internally contiguous.  There is no contiguity between
+ such a region and any other region (including others allocated from the
+ heap).
+ doc-allocate
+ doc-free
+ doc-resize
+ @node Memory Access, Address arithmetic, Heap Allocation, Memory
+ @subsection Memory Access
+ @cindex memory access words
+ doc-@
+ doc-!
+ doc-+!
+ doc-c@
+ doc-c!
+ doc-2@
+ doc-2!
+ doc-f@
+ doc-f!
+ doc-sf@
+ doc-sf!
+ doc-df@
+ doc-df!
+ @node Address arithmetic, Memory Blocks, Memory Access, Memory
+ @subsection Address arithmetic
+ @cindex address arithmetic words
+ Address arithmetic is the foundation on which you can build data
+ structures like arrays, records (@pxref{Structures}) and objects
+ (@pxref{Object-oriented Forth}).
+ @cindex address unit
+ @cindex au (address unit)
+ ANS Forth does not specify the sizes of the data types. Instead, it
+ offers a number of words for computing sizes and doing address
+ arithmetic. Address arithmetic is performed in terms of address units
+ (aus); on most systems the address unit is one byte. Note that a
+ character may have more than one au, so @code{chars} is no noop (on
+ platforms where it is a noop, it compiles to nothing).
+ The basic address arithmetic words are @code{+} and @code{-}.  E.g., if
+ you have the address of a cell, perform @code{1 cells +}, and you will
+ have the address of the next cell.
+ @cindex contiguous regions and address arithmetic
+ In ANS Forth you can perform address arithmetic only within a contiguous
+ region, i.e., if you have an address into one region, you can only add
+ and subtract such that the result is still within the region; you can
+ only subtract or compare addresses from within the same contiguous
+ region.  Reasons: several contiguous regions can be arranged in memory
+ in any way; on segmented systems addresses may have unusual
+ representations, such that address arithmetic only works within a
+ region.  Gforth provides a few more guarantees (linear address space,
+ dictionary grows upwards), but in general I have found it easy to stay
+ within contiguous regions (exception: computing and comparing to the
+ address just beyond the end of an array).
+ @cindex alignment of addresses for types
+ ANS Forth also defines words for aligning addresses for specific
+ types. Many computers require that accesses to specific data types
+ must only occur at specific addresses; e.g., that cells may only be
+ accessed at addresses divisible by 4. Even if a machine allows unaligned
+ accesses, it can usually perform aligned accesses faster.
+ For the performance-conscious: alignment operations are usually only
+ necessary during the definition of a data structure, not during the
+ (more frequent) accesses to it.
+ ANS Forth defines no words for character-aligning addresses. This is not
+ an oversight, but reflects the fact that addresses that are not
+ char-aligned have no use in the standard and therefore will not be
+ created.
+ @cindex @code{CREATE} and alignment
+ ANS Forth guarantees that addresses returned by @code{CREATE}d words
+ are cell-aligned; in addition, Gforth guarantees that these addresses
+ are aligned for all purposes.
+ Note that the ANS Forth word @code{char} has nothing to do with address
+ arithmetic.
+ doc-chars
+ doc-char+
+ doc-cells
+ doc-cell+
+ doc-cell
+ doc-aligned
+ doc-floats
+ doc-float+
+ doc-float
+ doc-faligned
+ doc-sfloats
+ doc-sfloat+
+ doc-sfaligned
+ doc-dfloats
+ doc-dfloat+
+ doc-dfaligned
+ doc-maxaligned
+ doc-cfaligned
+ doc-address-unit-bits
+ @node Memory Blocks,  , Address arithmetic, Memory
+ @subsection Memory Blocks
+ @cindex memory block words
+ @cindex character strings - moving and copying
+ Memory blocks often represent character strings; For ways of storing
+ character strings in memory see @ref{String Formats}.  For other
+ string-processing words see @ref{Displaying characters and strings}.
+ A few of these words work on address unit blocks.  In that case, you
+ usually have to insert @code{CHARS} before the word when working on
+ character strings.  Most words work on character blocks, and expect a
+ char-aligned address.
+ When copying characters between overlapping memory regions, use
+ @code{chars move} or choose carefully between @code{cmove} and
+ @code{cmove>}.
+ doc-move
+ doc-erase
+ doc-cmove
+ doc-cmove>
+ doc-fill
+ doc-blank
+ doc-compare
+ doc-search
+ doc--trailing
+ doc-/string
+ @comment TODO examples
+ @node Control Structures, Defining Words, Memory, Words
+ @section Control Structures
+ @cindex control structures
+ Control structures in Forth cannot be used interpretively, only in a
+ colon definition@footnote{To be precise, they have no interpretation
+ semantics (@pxref{Interpretation and Compilation Semantics}).}. We do
+ not like this limitation, but have not seen a satisfying way around it
+ yet, although many schemes have been proposed.
+ @menu
+ * Selection::                   IF ... ELSE ... ENDIF
+ * Simple Loops::                BEGIN ...
+ * Counted Loops::               DO
+ * Arbitrary control structures::
+ * Calls and returns::
+ * Exception Handling::
+ @end menu
+ @node Selection, Simple Loops, Control Structures, Control Structures
+ @subsection Selection
+ @cindex selection control structures
+ @cindex control structures for selection
+ @cindex @code{IF} control structure
+ @example
+ @i{flag}
+ IF
+   @i{code}
+ ENDIF
+ @end example
+ @noindent
+ If @i{flag} is non-zero (as far as @code{IF} etc. are concerned, a cell
+ with any bit set represents truth) @i{code} is executed.
+ @example
+ @i{flag}
+ IF
+   @i{code1}
+ ELSE
+   @i{code2}
+ ENDIF
+ @end example
+ If @var{flag} is true, @i{code1} is executed, otherwise @i{code2} is
+ executed.
+ You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
+ standard, and @code{ENDIF} is not, although it is quite popular. We
+ recommend using @code{ENDIF}, because it is less confusing for people
+ who also know other languages (and is not prone to reinforcing negative
+ prejudices against Forth in these people). Adding @code{ENDIF} to a
+ system that only supplies @code{THEN} is simple:
+ @example
+ : ENDIF   POSTPONE THEN ; immediate
+ @end example
+ [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
+ (adv.)}  has the following meanings:
+ @quotation
+ ... 2b: following next after in order ... 3d: as a necessary consequence
+ (if you were there, then you saw them).
+ @end quotation
+ Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
+ and many other programming languages has the meaning 3d.]
+ Gforth also provides the words @code{?DUP-IF} and @code{?DUP-0=-IF}, so
+ you can avoid using @code{?dup}. Using these alternatives is also more
+ efficient than using @code{?dup}. Definitions in ANS Forth
+ for @code{ENDIF}, @code{?DUP-IF} and @code{?DUP-0=-IF} are provided in
+ @file{compat/control.fs}.
+ @cindex @code{CASE} control structure
+ @example
+ @i{n}
+ CASE
+   @i{n1} OF @i{code1} ENDOF
+   @i{n2} OF @i{code2} ENDOF
+   @dots{}
+   ( n ) @i{default-code} ( n )
+ ENDCASE
+ @end example
+ Executes the first @i{codei}, where the @i{ni} is equal to @i{n}.  If no
+ @i{ni} matches, the optional @i{default-code} is executed. The optional
+ default case can be added by simply writing the code after the last
+ @code{ENDOF}. It may use @i{n}, which is on top of the stack, but must
+ not consume it.
+ @progstyle
+ To keep the code understandable, you should ensure that on all paths
+ through a selection construct the stack is changed in the same way
+ (wrt. number and types of stack items consumed and pushed).
+ @node Simple Loops, Counted Loops, Selection, Control Structures
+ @subsection Simple Loops
+ @cindex simple loops
+ @cindex loops without count
+ @cindex @code{WHILE} loop
+ @example
+ BEGIN
+   @i{code1}
+   @i{flag}
+ WHILE
+   @i{code2}
+ REPEAT
+ @end example
+ @i{code1} is executed and @i{flag} is computed. If it is true,
+ @i{code2} is executed and the loop is restarted; If @i{flag} is
+ false, execution continues after the @code{REPEAT}.
+ @cindex @code{UNTIL} loop
+ @example
+ BEGIN
+   @i{code}
+   @i{flag}
+ UNTIL
+ @end example
+ @i{code} is executed. The loop is restarted if @code{flag} is false.
+ @progstyle
+ To keep the code understandable, a complete iteration of the loop should
+ not change the number and types of the items on the stacks.
+ @cindex endless loop
+ @cindex loops, endless
+ @example
+ BEGIN
+   @i{code}
+ AGAIN
+ @end example
+ This is an endless loop.
+ @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
+ @subsection Counted Loops
+ @cindex counted loops
+ @cindex loops, counted
+ @cindex @code{DO} loops
+ The basic counted loop is:
+ @example
+ @i{limit} @i{start}
+ ?DO
+   @i{body}
+ LOOP
+ @end example
+ This performs one iteration for every integer, starting from @i{start}
+ and up to, but excluding @i{limit}. The counter, or @i{index}, can be
+ accessed with @code{i}. For example, the loop:
+ @example
+0 ?DO
+   i .
+ LOOP
+ @end example
+ @noindent
+ prints @code{0 1 2 3 4 5 6 7 8 9}
+ The index of the innermost loop can be accessed with @code{i}, the index
+ of the next loop with @code{j}, and the index of the third loop with
+ @code{k}.
+ doc-i
+ doc-j
+ doc-k
+ The loop control data are kept on the return stack, so there are some
+ restrictions on mixing return stack accesses and counted loop words. In
+ particuler, if you put values on the return stack outside the loop, you
+ cannot read them inside the loop@footnote{well, not in a way that is
+ portable.}. If you put values on the return stack within a loop, you
+ have to remove them before the end of the loop and before accessing the
+ index of the loop.
+ There are several variations on the counted loop:
+ @itemize @bullet
+ @item
+ @code{LEAVE} leaves the innermost counted loop immediately; execution
+ continues after the associated @code{LOOP} or @code{NEXT}. For example:
+ @example
+0 ?DO  i DUP . 3 = IF LEAVE THEN LOOP
+ @end example
+ prints @code{0 1 2 3}
+ @item
+ @code{UNLOOP} prepares for an abnormal loop exit, e.g., via
+ @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
+ return stack so @code{EXIT} can get to its return address. For example:
+ @example
+ : demo 10 0 ?DO i DUP . 3 = IF UNLOOP EXIT THEN LOOP ." Done" ;
+ @end example
+ prints @code{0 1 2 3}
+ @item
+ If @i{start} is greater than @i{limit}, a @code{?DO} loop is entered
+ (and @code{LOOP} iterates until they become equal by wrap-around
+ arithmetic). This behaviour is usually not what you want. Therefore,
+ Gforth offers @code{+DO} and @code{U+DO} (as replacements for
+ @code{?DO}), which do not enter the loop if @i{start} is greater than
+ @i{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
+ unsigned loop parameters.
+ @item
+ @code{?DO} can be replaced by @code{DO}. @code{DO} always enters
+ the loop, independent of the loop parameters. Do not use @code{DO}, even
+ if you know that the loop is entered in any case. Such knowledge tends
+ to become invalid during maintenance of a program, and then the
+ @code{DO} will make trouble.
+ @item
+ @code{LOOP} can be replaced with @code{@i{n} +LOOP}; this updates the
+ index by @i{n} instead of by 1. The loop is terminated when the border
+ between @i{limit-1} and @i{limit} is crossed. E.g.:
+ @example
+0 +DO  i .  2 +LOOP
+ @end example
+ @noindent
+ prints @code{0 2}
+ @example
+1 +DO  i .  2 +LOOP
+ @end example
+ @noindent
+ prints @code{1 3}
+ @item
+ @cindex negative increment for counted loops
+ @cindex counted loops with negative increment
+ The behaviour of @code{@i{n} +LOOP} is peculiar when @i{n} is negative:
+ @example
+ -1 0 ?DO  i .  -1 +LOOP
+ @end example
+ @noindent
+ prints @code{0 -1}
+ @example
+0 ?DO  i .  -1 +LOOP
+ @end example
+ prints nothing.
+ Therefore we recommend avoiding @code{@i{n} +LOOP} with negative
+ @i{n}. One alternative is @code{@i{u} -LOOP}, which reduces the
+ index by @i{u} each iteration. The loop is terminated when the border
+ between @i{limit+1} and @i{limit} is crossed. Gforth also provides
+ @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
+ @example
+ -2 0 -DO  i .  1 -LOOP
+ @end example
+ @noindent
+ prints @code{0 -1}
+ @example
+ -1 0 -DO  i .  1 -LOOP
+ @end example
+ @noindent
+ prints @code{0}
+ @example
+0 -DO  i .  1 -LOOP
+ @end example
+ @noindent
+ prints nothing.
+ @end itemize
+ Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
+ @code{-LOOP} are not defined in ANS Forth. However, an implementation
+ for these words that uses only standard words is provided in
+ @file{compat/loops.fs}.
+ @cindex @code{FOR} loops
+ Another counted loop is:
+ @example
+ @i{n}
+ FOR
+   @i{body}
+ NEXT
+ @end example
+ This is the preferred loop of native code compiler writers who are too
+ lazy to optimize @code{?DO} loops properly. This loop structure is not
+ defined in ANS Forth. In Gforth, this loop iterates @i{n+1} times;
+ @code{i} produces values starting with @i{n} and ending with 0. Other
+ Forth systems may behave differently, even if they support @code{FOR}
+ loops. To avoid problems, don't use @code{FOR} loops.
+ @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
+ @subsection Arbitrary control structures
+ @cindex control structures, user-defined
+ @cindex control-flow stack
+ ANS Forth permits and supports using control structures in a non-nested
+ way. Information about incomplete control structures is stored on the
+ control-flow stack. This stack may be implemented on the Forth data
+ stack, and this is what we have done in Gforth.
+ @cindex @code{orig}, control-flow stack item
+ @cindex @code{dest}, control-flow stack item
+ An @i{orig} entry represents an unresolved forward branch, a @i{dest}
+ entry represents a backward branch target. A few words are the basis for
+ building any control structure possible (except control structures that
+ need storage, like calls, coroutines, and backtracking).
+ doc-if
+ doc-ahead
+ doc-then
+ doc-begin
+ doc-until
+ doc-again
+ doc-cs-pick
+ doc-cs-roll
+ The Standard words @code{CS-PICK} and @code{CS-ROLL} allow you to
+ manipulate the control-flow stack in a portable way. Without them, you
+ would need to know how many stack items are occupied by a control-flow
+ entry (many systems use one cell. In Gforth they currently take three,
+ but this may change in the future).
+ Some standard control structure words are built from these words:
+ doc-else
+ doc-while
+ doc-repeat
+ @noindent
+ Gforth adds some more control-structure words:
+ doc-endif
+ doc-?dup-if
+ doc-?dup-0=-if
+ @noindent
+ Counted loop words constitute a separate group of words:
+ doc-?do
+ doc-+do
+ doc-u+do
+ doc--do
+ doc-u-do
+ doc-do
+ doc-for
+ doc-loop
+ doc-+loop
+ doc--loop
+ doc-next
+ doc-leave
+ doc-?leave
+ doc-unloop
+ doc-done
+ The standard does not allow using @code{CS-PICK} and @code{CS-ROLL} on
+ @i{do-sys}. Gforth allows it, but it's your job to ensure that for
+ every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
+ through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
+ fall-through path). Also, you have to ensure that all @code{LEAVE}s are
+ resolved (by using one of the loop-ending words or @code{DONE}).
+ @noindent
+ Another group of control structure words are:
+ doc-case
+ doc-endcase
+ doc-of
+ doc-endof
+ @i{case-sys} and @i{of-sys} cannot be processed using @code{CS-PICK} and
+ @code{CS-ROLL}.
+ @subsubsection Programming Style
+ @cindex control structures programming style
+ @cindex programming style, arbitrary control structures
+ In order to ensure readability we recommend that you do not create
+ arbitrary control structures directly, but define new control structure
+ words for the control structure you want and use these words in your
+ program. For example, instead of writing:
+ @example
+ BEGIN
+   ...
+ IF [ 1 CS-ROLL ]
+   ...
+ AGAIN THEN
+ @end example
+ @noindent
+ we recommend defining control structure words, e.g.,
+ @example
+ : WHILE ( DEST -- ORIG DEST )
+  POSTPONE IF
+CS-ROLL ; immediate
+ : REPEAT ( orig dest -- )
+  POSTPONE AGAIN
+  POSTPONE THEN ; immediate
+ @end example
+ @noindent
+ and then using these to create the control structure:
+ @example
+ BEGIN
+   ...
+ WHILE
+   ...
+ REPEAT
+ @end example
+ That's much easier to read, isn't it? Of course, @code{REPEAT} and
+ @code{WHILE} are predefined, so in this example it would not be
+ necessary to define them.
+ @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
+ @subsection Calls and returns
+ @cindex calling a definition
+ @cindex returning from a definition
+ @cindex recursive definitions
+ A definition can be called simply be writing the name of the definition
+ to be called. Normally a definition is invisible during its own
+ definition. If you want to write a directly recursive definition, you
+ can use @code{recursive} to make the current definition visible, or
+ @code{recurse} to call the current definition directly.
+ doc-recursive
+ doc-recurse
+ @comment TODO add example of the two recursion methods
+ @quotation
+ @progstyle
+ I prefer using @code{recursive} to @code{recurse}, because calling the
+ definition by name is more descriptive (if the name is well-chosen) than
+ the somewhat cryptic @code{recurse}.  E.g., in a quicksort
+ implementation, it is much better to read (and think) ``now sort the
+ partitions'' than to read ``now do a recursive call''.
+ @end quotation
+ For mutual recursion, use @code{Defer}red words, like this:
+ @example
+ Defer foo
+ : bar ( ... -- ... )
+  ... foo ... ;
+ :noname ( ... -- ... )
+  ... bar ... ;
+ IS foo
+ @end example
+ Deferred words are discussed in more detail in @ref{Deferred words}.
+ The current definition returns control to the calling definition when
+ the end of the definition is reached or @code{EXIT} is encountered.
+ doc-exit
+ doc-;s
+ @node Exception Handling,  , Calls and returns, Control Structures
+ @subsection Exception Handling
+ @cindex exceptions
+ @c quit is a very bad idea for error handling,
+ @c because it does not translate into a THROW
+ @c it also does not belong into this chapter
+ If a word detects an error condition that it cannot handle, it can
+ @code{throw} an exception.  In the simplest case, this will terminate
+ your program, and report an appropriate error.
+ doc-throw
+ @code{Throw} consumes a cell-sized error number on the stack. There are
+ some predefined error numbers in ANS Forth (see @file{errors.fs}).  In
+ Gforth (and most other systems) you can use the iors produced by various
+ words as error numbers (e.g., a typical use of @code{allocate} is
+ @code{allocate throw}).  Gforth also provides the word @code{exception}
+ to define your own error numbers (with decent error reporting); an ANS
+ Forth version of this word (but without the error messages) is available
+ in @code{compat/except.fs}.  And finally, you can use your own error
+ numbers (anything outside the range -4095..0), but won't get nice error
+ messages, only numbers.  For example, try:
+ @example
+ -10 throw                    \ ANS defined
+ -267 throw                   \ system defined
+ s" my error" exception throw \ user defined
+throw                      \ arbitrary number
+ @end example
+ doc---exception-exception
+ A common idiom to @code{THROW} a specific error if a flag is true is
+ this:
+ @example
+ @code{( flag ) 0<> @i{errno} and throw}
+ @end example
+ Your program can provide exception handlers to catch exceptions.  An
+ exception handler can be used to correct the problem, or to clean up
+ some data structures and just throw the exception to the next exception
+ handler.  Note that @code{throw} jumps to the dynamically innermost
+ exception handler.  The system's exception handler is outermost, and just
+ prints an error and restarts command-line interpretation (or, in batch
+ mode (i.e., while processing the shell command line), leaves Gforth).
+ The ANS Forth way to catch exceptions is @code{catch}:
+ doc-catch
+ The most common use of exception handlers is to clean up the state when
+ an error happens.  E.g.,
+ @example
+ base @ >r hex \ actually the hex should be inside foo, or we h
+ ['] foo catch ( nerror|0 )
+ r> base !
+ ( nerror|0 ) throw \ pass it on
+ @end example
+ A use of @code{catch} for handling the error @code{myerror} might look
+ like this:
+ @example
+ ['] foo catch
+ CASE
+   myerror OF ... ( do something about it ) ENDOF
+   dup throw \ default: pass other errors on, do nothing on non-errors
+ ENDCASE
+ @end example
+ Having to wrap the code into a separate word is often cumbersome,
+ therefore Gforth provides an alternative syntax:
+ @example
+ TRY
+   @i{code1}
+ RECOVER     \ optional
+   @i{code2} \ optional
+ ENDTRY
+ @end example
+ This performs @i{Code1}.  If @i{code1} completes normally, execution
+ continues after the @code{endtry}.  If @i{Code1} throws, the stacks are
+ reset to the state during @code{try}, the throw value is pushed on the
+ data stack, and execution constinues at @i{code2}, and finally falls
+ through the @code{endtry} into the following code. If there is no
+ @code{recover} clause, this works like an empty recover clause.
+ doc-try
+ doc-recover
+ doc-endtry
+ The cleanup example from above in this syntax:
+ @example
+ base @ >r TRY
+   hex foo \ now the hex is placed correctly
+      \ value for throw
+ ENDTRY
+ r> base ! throw
+ @end example
+ And here's the error handling example:
+ @example
+ TRY
+   foo
+ RECOVER
+   CASE
+     myerror OF ... ( do something about it ) ENDOF
+     throw \ pass other errors on
+   ENDCASE
+ ENDTRY
+ @end example
+ @progstyle
+ As usual, you should ensure that the stack depth is statically known at
+ the end: either after the @code{throw} for passing on errors, or after
+ the @code{ENDTRY} (or, if you use @code{catch}, after the end of the
+ selection construct for handling the error).
+ There are two alternatives to @code{throw}: @code{Abort"} is conditional
+ and you can provide an error message.  @code{Abort} just produces an
+ ``Aborted'' error.
+ The problem with these words is that exception handlers cannot
+ differentiate between different @code{abort"}s; they just look like
+ @code{-2 throw} to them (the error message cannot be accessed by
+ standard programs).  Similar @code{abort} looks like @code{-1 throw} to
+ exception handlers.
+ doc-abort"
+ doc-abort
+ @c -------------------------------------------------------------
+ @node Defining Words, Interpretation and Compilation Semantics, Control Structures, Words
+ @section Defining Words
+ @cindex defining words
+ Defining words are used to extend Forth by creating new entries in the dictionary.
+ @menu
+ * CREATE::
+ * Variables::                   Variables and user variables
+ * Constants::
+ * Values::                      Initialised variables
+ * Colon Definitions::
+ * Anonymous Definitions::       Definitions without names
+ * Supplying names::             Passing definition names as strings
+ * User-defined Defining Words::
+ * Deferred words::              Allow forward references
+ * Aliases::
+ @end menu
+ @node CREATE, Variables, Defining Words, Defining Words
+ @subsection @code{CREATE}
+ @cindex simple defining words
+ @cindex defining words, simple
+ Defining words are used to create new entries in the dictionary. The
+ simplest defining word is @code{CREATE}. @code{CREATE} is used like
+ this:
+ @example
+ CREATE new-word1
+ @end example
+ @code{CREATE} is a parsing word, i.e., it takes an argument from the
+ input stream (@code{new-word1} in our example).  It generates a
+ dictionary entry for @code{new-word1}. When @code{new-word1} is
+ executed, all that it does is leave an address on the stack. The address
+ represents the value of the data space pointer (@code{HERE}) at the time
+ that @code{new-word1} was defined. Therefore, @code{CREATE} is a way of
+ associating a name with the address of a region of memory.
+ doc-create
+ Note that in ANS Forth guarantees only for @code{create} that its body
+ is in dictionary data space (i.e., where @code{here}, @code{allot}
+ etc. work, @pxref{Dictionary allocation}).  Also, in ANS Forth only
+ @code{create}d words can be modified with @code{does>}
+ (@pxref{User-defined Defining Words}).  And in ANS Forth @code{>body}
+ can only be applied to @code{create}d words.
+ By extending this example to reserve some memory in data space, we end
+ up with something like a @i{variable}. Here are two different ways to do
+ it:
+ @example
+ CREATE new-word2 1 cells allot  \ reserve 1 cell - initial value undefined
+ CREATE new-word3 4 ,            \ reserve 1 cell and initialise it (to 4)
+ @end example
+ The variable can be examined and modified using @code{@@} (``fetch'') and
+ @code{!} (``store'') like this:
+ @example
+ new-word2 @@ .      \ get address, fetch from it and display
+new-word2 !   \ new value, get address, store to it
+ @end example
+ @cindex arrays
+ A similar mechanism can be used to create arrays. For example, an
+-character text input buffer:
+ @example
+ CREATE text-buf 80 chars allot
+ text-buf 0 chars c@@ \ the 1st character (offset 0)
+ text-buf 3 chars c@@ \ the 4th character (offset 3)
+ @end example
+ You can build arbitrarily complex data structures by allocating
+ appropriate areas of memory. For further discussions of this, and to
+ learn about some Gforth tools that make it easier,
+ @xref{Structures}.
+ @node Variables, Constants, CREATE, Defining Words
+ @subsection Variables
+ @cindex variables
+ The previous section showed how a sequence of commands could be used to
+ generate a variable.  As a final refinement, the whole code sequence can
+ be wrapped up in a defining word (pre-empting the subject of the next
+ section), making it easier to create new variables:
+ @example
+ : myvariableX ( "name" -- a-addr ) CREATE 1 cells allot ;
+ : myvariable0 ( "name" -- a-addr ) CREATE 0 , ;
+ myvariableX foo \ variable foo starts off with an unknown value
+ myvariable0 joe \ whilst joe is initialised to 0
+3 * foo !   \ set foo to 135
+joe !     \ set joe to 1234
+joe +!       \ increment joe by 3.. to 1237
+ @end example
+ Not surprisingly, there is no need to define @code{myvariable}, since
+ Forth already has a definition @code{Variable}. ANS Forth does not
+ guarantee that a @code{Variable} is initialised when it is created
+ (i.e., it may behave like @code{myvariableX}). In contrast, Gforth's
+ @code{Variable} initialises the variable to 0 (i.e., it behaves exactly
+ like @code{myvariable0}). Forth also provides @code{2Variable} and
+ @code{fvariable} for double and floating-point variables, respectively
+ -- they are initialised to 0. and 0e in Gforth. If you use a @code{Variable} to
+ store a boolean, you can use @code{on} and @code{off} to toggle its
+ state.
+ doc-variable
+ doc-2variable
+ doc-fvariable
+ @cindex user variables
+ @cindex user space
+ The defining word @code{User} behaves in the same way as @code{Variable}.
+ The difference is that it reserves space in @i{user (data) space} rather
+ than normal data space. In a Forth system that has a multi-tasker, each
+ task has its own set of user variables.
+ doc-user
+ @c doc-udp
+ @c doc-uallot
+ @comment TODO is that stuff about user variables strictly correct? Is it
+ @comment just terminal tasks that have user variables?
+ @comment should document tasker.fs (with some examples) elsewhere
+ @comment in this manual, then expand on user space and user variables.
+ @node Constants, Values, Variables, Defining Words
+ @subsection Constants
+ @cindex constants
+ @code{Constant} allows you to declare a fixed value and refer to it by
+ name. For example:
+ @example
+Constant INCHES-PER-FOOT
+E+08 fconstant SPEED-O-LIGHT
+ @end example
+ A @code{Variable} can be both read and written, so its run-time
+ behaviour is to supply an address through which its current value can be
+ manipulated. In contrast, the value of a @code{Constant} cannot be
+ changed once it has been declared@footnote{Well, often it can be -- but
+ not in a Standard, portable way. It's safer to use a @code{Value} (read
+ on).} so it's not necessary to supply the address -- it is more
+ efficient to return the value of the constant directly. That's exactly
+ what happens; the run-time effect of a constant is to put its value on
+ the top of the stack (You can find one
+ way of implementing @code{Constant} in @ref{User-defined Defining Words}).
+ Forth also provides @code{2Constant} and @code{fconstant} for defining
+ double and floating-point constants, respectively.
+ doc-constant
+ doc-2constant
+ doc-fconstant
+ @c that's too deep, and it's not necessarily true for all ANS Forths. - anton
+ @c nac-> How could that not be true in an ANS Forth? You can't define a
+ @c constant, use it and then delete the definition of the constant..
+ @c anton->An ANS Forth system can compile a constant to a literal; On
+ @c decompilation you would see only the number, just as if it had been used
+ @c in the first place.  The word will stay, of course, but it will only be
+ @c used by the text interpreter (no run-time duties, except when it is
+ @c POSTPONEd or somesuch).
+ @c nac:
+ @c I agree that it's rather deep, but IMO it is an important difference
+ @c relative to other programming languages.. often it's annoying: it
+ @c certainly changes my programming style relative to C.
+ @c anton: In what way?
+ Constants in Forth behave differently from their equivalents in other
+ programming languages. In other languages, a constant (such as an EQU in
+ assembler or a #define in C) only exists at compile-time; in the
+ executable program the constant has been translated into an absolute
+ number and, unless you are using a symbolic debugger, it's impossible to
+ know what abstract thing that number represents. In Forth a constant has
+ an entry in the header space and remains there after the code that uses
+ it has been defined. In fact, it must remain in the dictionary since it
+ has run-time duties to perform. For example:
+ @example
+Constant INCHES-PER-FOOT
+ : FEET-TO-INCHES ( n1 -- n2 ) INCHES-PER-FOOT * ;
+ @end example
+ @cindex in-lining of constants
+ When @code{FEET-TO-INCHES} is executed, it will in turn execute the xt
+ associated with the constant @code{INCHES-PER-FOOT}. If you use
+ @code{see} to decompile the definition of @code{FEET-TO-INCHES}, you can
+ see that it makes a call to @code{INCHES-PER-FOOT}. Some Forth compilers
+ attempt to optimise constants by in-lining them where they are used. You
+ can force Gforth to in-line a constant like this:
+ @example
+ : FEET-TO-INCHES ( n1 -- n2 ) [ INCHES-PER-FOOT ] LITERAL * ;
+ @end example
+ If you use @code{see} to decompile @i{this} version of
+ @code{FEET-TO-INCHES}, you can see that @code{INCHES-PER-FOOT} is no
+ longer present. To understand how this works, read
+ @ref{Interpret/Compile states}, and @ref{Literals}.
+ In-lining constants in this way might improve execution time
+ fractionally, and can ensure that a constant is now only referenced at
+ compile-time. However, the definition of the constant still remains in
+ the dictionary. Some Forth compilers provide a mechanism for controlling
+ a second dictionary for holding transient words such that this second
+ dictionary can be deleted later in order to recover memory
+ space. However, there is no standard way of doing this.
+ @node Values, Colon Definitions, Constants, Defining Words
+ @subsection Values
+ @cindex values
+ A @code{Value} behaves like a @code{Constant}, but it can be changed.
+ @code{TO} is a parsing word that changes a @code{Values}.  In Gforth
+ (not in ANS Forth) you can access (and change) a @code{value} also with
+ @code{>body}.
+ Here are some
+ examples:
+ @example
+Value APPLES     \ Define APPLES with an initial value of 12
+TO APPLES        \ Change the value of APPLES. TO is a parsing word
+' APPLES >body +! \ Increment APPLES.  Non-standard usage.
+ APPLES              \ puts 35 on the top of the stack.
+ @end example
+ doc-value
+ doc-to
+ @node Colon Definitions, Anonymous Definitions, Values, Defining Words
+ @subsection Colon Definitions
+ @cindex colon definitions
+ @example
+ : name ( ... -- ... )
+     word1 word2 word3 ;
+ @end example
+ @noindent
+ Creates a word called @code{name} that, upon execution, executes
+ @code{word1 word2 word3}. @code{name} is a @dfn{(colon) definition}.
+ The explanation above is somewhat superficial. For simple examples of
+ colon definitions see @ref{Your first definition}.  For an in-depth
+ discussion of some of the issues involved, @xref{Interpretation and
+ Compilation Semantics}.
+ doc-:
+ doc-;
+ @node Anonymous Definitions, Supplying names, Colon Definitions, Defining Words
+ @subsection Anonymous Definitions
+ @cindex colon definitions
+ @cindex defining words without name
+ Sometimes you want to define an @dfn{anonymous word}; a word without a
+ name. You can do this with:
+ doc-:noname
+ This leaves the execution token for the word on the stack after the
+ closing @code{;}. Here's an example in which a deferred word is
+ initialised with an @code{xt} from an anonymous colon definition:
+ @example
+ Defer deferred
+ :noname ( ... -- ... )
+   ... ;
+ IS deferred
+ @end example
+ @noindent
+ Gforth provides an alternative way of doing this, using two separate
+ words:
+ doc-noname
+ @cindex execution token of last defined word
+ doc-lastxt
+ @noindent
+ The previous example can be rewritten using @code{noname} and
+ @code{lastxt}:
+ @example
+ Defer deferred
+ noname : ( ... -- ... )
+   ... ;
+ lastxt IS deferred
+ @end example
+ @noindent
+ @code{noname} works with any defining word, not just @code{:}.
+ @code{lastxt} also works when the last word was not defined as
+ @code{noname}.  It does not work for combined words, though.  It also has
+ the useful property that is is valid as soon as the header for a
+ definition has been built. Thus:
+ @example
+ lastxt . : foo [ lastxt . ] ; ' foo .
+ @end example
+ @noindent
+ prints 3 numbers; the last two are the same.
+ @node Supplying names, User-defined Defining Words, Anonymous Definitions, Defining Words
+ @subsection Supplying the name of a defined word
+ @cindex names for defined words
+ @cindex defining words, name given in a string
+ By default, a defining word takes the name for the defined word from the
+ input stream. Sometimes you want to supply the name from a string. You
+ can do this with:
+ doc-nextname
+ For example:
+ @example
+ s" foo" nextname create
+ @end example
+ @noindent
+ is equivalent to:
+ @example
+ create foo
+ @end example
+ @noindent
+ @code{nextname} works with any defining word.
+ @node User-defined Defining Words, Deferred words, Supplying names, Defining Words
+ @subsection User-defined Defining Words
+ @cindex user-defined defining words
+ @cindex defining words, user-defined
+ You can create a new defining word by wrapping defining-time code around
+ an existing defining word and putting the sequence in a colon
+ definition.
+ @c anton: This example is very complex and leads in a quite different
+ @c direction from the CREATE-DOES> stuff that follows.  It should probably
+ @c be done elsewhere, or as a subsubsection of this subsection (or as a
+ @c subsection of Defining Words)
+ For example, suppose that you have a word @code{stats} that
+ gathers statistics about colon definitions given the @i{xt} of the
+ definition, and you want every colon definition in your application to
+ make a call to @code{stats}. You can define and use a new version of
+ @code{:} like this:
+ @example
+ : stats ( xt -- ) DUP ." (Gathering statistics for " . ." )"
+   ... ;  \ other code
+ : my: : lastxt postpone literal ['] stats compile, ;
+ my: foo + - ;
+ @end example
+ When @code{foo} is defined using @code{my:} these steps occur:
+ @itemize @bullet
+ @item
+ @code{my:} is executed.
+ @item
+ The @code{:} within the definition (the one between @code{my:} and
+ @code{lastxt}) is executed, and does just what it always does; it parses
+ the input stream for a name, builds a dictionary header for the name
+ @code{foo} and switches @code{state} from interpret to compile.
+ @item
+ The word @code{lastxt} is executed. It puts the @i{xt} for the word that is
+ being defined -- @code{foo} -- onto the stack.
+ @item
+ The code that was produced by @code{postpone literal} is executed; this
+ causes the value on the stack to be compiled as a literal in the code
+ area of @code{foo}.
+ @item
+ The code @code{['] stats} compiles a literal into the definition of
+ @code{my:}. When @code{compile,} is executed, that literal -- the
+ execution token for @code{stats} -- is layed down in the code area of
+ @code{foo} , following the literal@footnote{Strictly speaking, the
+ mechanism that @code{compile,} uses to convert an @i{xt} into something
+ in the code area is implementation-dependent. A threaded implementation
+ might spit out the execution token directly whilst another
+ implementation might spit out a native code sequence.}.
+ @item
+ At this point, the execution of @code{my:} is complete, and control
+ returns to the text interpreter. The text interpreter is in compile
+ state, so subsequent text @code{+ -} is compiled into the definition of
+ @code{foo} and the @code{;} terminates the definition as always.
+ @end itemize
+ You can use @code{see} to decompile a word that was defined using
+ @code{my:} and see how it is different from a normal @code{:}
+ definition. For example:
+ @example
+ : bar + - ;  \ like foo but using : rather than my:
+ see bar
+ : bar
+   + - ;
+ see foo
+ : foo
+   107645672 stats + - ;
+ \ use ' stats . to show that 107645672 is the xt for stats
+ @end example
+ You can use techniques like this to make new defining words in terms of
+ @i{any} existing defining word.
+ @cindex defining defining words
+ @cindex @code{CREATE} ... @code{DOES>}
+ If you want the words defined with your defining words to behave
+ differently from words defined with standard defining words, you can
+ write your defining word like this:
+ @example
+ : def-word ( "name" -- )
+     CREATE @i{code1}
+ DOES> ( ... -- ... )
+     @i{code2} ;
+ def-word name
+ @end example
+ @cindex child words
+ This fragment defines a @dfn{defining word} @code{def-word} and then
+ executes it.  When @code{def-word} executes, it @code{CREATE}s a new
+ word, @code{name}, and executes the code @i{code1}. The code @i{code2}
+ is not executed at this time. The word @code{name} is sometimes called a
+ @dfn{child} of @code{def-word}.
+ When you execute @code{name}, the address of the body of @code{name} is
+ put on the data stack and @i{code2} is executed (the address of the body
+ of @code{name} is the address @code{HERE} returns immediately after the
+ @code{CREATE}, i.e., the address a @code{create}d word returns by
+ default).
+ @c anton:
+ @c www.dictionary.com says:
+ @c at�a�vism: 1.The reappearance of a characteristic in an organism after
+ @c several generations of absence, usually caused by the chance
+ @c recombination of genes.  2.An individual or a part that exhibits
+ @c atavism. Also called throwback.  3.The return of a trait or recurrence
+ @c of previous behavior after a period of absence.
+ @c
+ @c Doesn't seem to fit.
+ @c @cindex atavism in child words
+ You can use @code{def-word} to define a set of child words that behave
+ similarly; they all have a common run-time behaviour determined by
+ @i{code2}. Typically, the @i{code1} sequence builds a data area in the
+ body of the child word. The structure of the data is common to all
+ children of @code{def-word}, but the data values are specific -- and
+ private -- to each child word. When a child word is executed, the
+ address of its private data area is passed as a parameter on TOS to be
+ used and manipulated@footnote{It is legitimate both to read and write to
+ this data area.} by @i{code2}.
+ The two fragments of code that make up the defining words act (are
+ executed) at two completely separate times:
+ @itemize @bullet
+ @item
+ At @i{define time}, the defining word executes @i{code1} to generate a
+ child word
+ @item
+ At @i{child execution time}, when a child word is invoked, @i{code2}
+ is executed, using parameters (data) that are private and specific to
+ the child word.
+ @end itemize
+ Another way of understanding the behaviour of @code{def-word} and
+ @code{name} is to say that, if you make the following definitions:
+ @example
+ : def-word1 ( "name" -- )
+     CREATE @i{code1} ;
+ : action1 ( ... -- ... )
+     @i{code2} ;
+ def-word1 name1
+ @end example
+ @noindent
+ Then using @code{name1 action1} is equivalent to using @code{name}.
+ The classic example is that you can define @code{CONSTANT} in this way:
+ @example
+ : CONSTANT ( w "name" -- )
+     CREATE ,
+ DOES> ( -- w )
+     @@ ;
+ @end example
+ @comment There is a beautiful description of how this works and what
+ @comment it does in the Forthwrite 100th edition.. as well as an elegant
+ @comment commentary on the Counting Fruits problem.
+ When you create a constant with @code{5 CONSTANT five}, a set of
+ define-time actions take place; first a new word @code{five} is created,
+ then the value 5 is laid down in the body of @code{five} with
+ @code{,}. When @code{five} is executed, the address of the body is put on
+ the stack, and @code{@@} retrieves the value 5. The word @code{five} has
+ no code of its own; it simply contains a data field and a pointer to the
+ code that follows @code{DOES>} in its defining word. That makes words
+ created in this way very compact.
+ The final example in this section is intended to remind you that space
+ reserved in @code{CREATE}d words is @i{data} space and therefore can be
+ both read and written by a Standard program@footnote{Exercise: use this
+ example as a starting point for your own implementation of @code{Value}
+ and @code{TO} -- if you get stuck, investigate the behaviour of @code{'} and
+ @code{[']}.}:
+ @example
+ : foo ( "name" -- )
+     CREATE -1 ,
+ DOES> ( -- )
+     @@ . ;
+ foo first-word
+ foo second-word
+' first-word >BODY !
+ @end example
+ If @code{first-word} had been a @code{CREATE}d word, we could simply
+ have executed it to get the address of its data field. However, since it
+ was defined to have @code{DOES>} actions, its execution semantics are to
+ perform those @code{DOES>} actions. To get the address of its data field
+ it's necessary to use @code{'} to get its xt, then @code{>BODY} to
+ translate the xt into the address of the data field.  When you execute
+ @code{first-word}, it will display @code{123}. When you execute
+ @code{second-word} it will display @code{-1}.
+ @cindex stack effect of @code{DOES>}-parts
+ @cindex @code{DOES>}-parts, stack effect
+ In the examples above the stack comment after the @code{DOES>} specifies
+ the stack effect of the defined words, not the stack effect of the
+ following code (the following code expects the address of the body on
+ the top of stack, which is not reflected in the stack comment). This is
+ the convention that I use and recommend (it clashes a bit with using
+ locals declarations for stack effect specification, though).
+ @menu
+ * CREATE..DOES> applications::
+ * CREATE..DOES> details::
+ * Advanced does> usage example::
+ @end menu
+ @node CREATE..DOES> applications, CREATE..DOES> details, User-defined Defining Words, User-defined Defining Words
+ @subsubsection Applications of @code{CREATE..DOES>}
+ @cindex @code{CREATE} ... @code{DOES>}, applications
+ You may wonder how to use this feature. Here are some usage patterns:
+ @cindex factoring similar colon definitions
+ When you see a sequence of code occurring several times, and you can
+ identify a meaning, you will factor it out as a colon definition. When
+ you see similar colon definitions, you can factor them using
+ @code{CREATE..DOES>}. E.g., an assembler usually defines several words
+ that look very similar:
+ @example
+ : ori, ( reg-target reg-source n -- )
+asm-reg-reg-imm ;
+ : andi, ( reg-target reg-source n -- )
+asm-reg-reg-imm ;
+ @end example
+ @noindent
+ This could be factored with:
+ @example
+ : reg-reg-imm ( op-code -- )
+     CREATE ,
+ DOES> ( reg-target reg-source n -- )
+     @@ asm-reg-reg-imm ;
+reg-reg-imm ori,
+reg-reg-imm andi,
+ @end example
+ @cindex currying
+ Another view of @code{CREATE..DOES>} is to consider it as a crude way to
+ supply a part of the parameters for a word (known as @dfn{currying} in
+ the functional language community). E.g., @code{+} needs two
+ parameters. Creating versions of @code{+} with one parameter fixed can
+ be done like this:
+ @example
+ : curry+ ( n1 -- )
+     CREATE ,
+ DOES> ( n2 -- n1+n2 )
+     @@ + ;
+curry+ 3+
+ -2 curry+ 2-
+ @end example
+ @node CREATE..DOES> details, Advanced does> usage example, CREATE..DOES> applications, User-defined Defining Words
+ @subsubsection The gory details of @code{CREATE..DOES>}
+ @cindex @code{CREATE} ... @code{DOES>}, details
+ doc-does>
+ @cindex @code{DOES>} in a separate definition
+ This means that you need not use @code{CREATE} and @code{DOES>} in the
+ same definition; you can put the @code{DOES>}-part in a separate
+ definition. This allows us to, e.g., select among different @code{DOES>}-parts:
+ @example
+ : does1
+ DOES> ( ... -- ... )
+     ... ;
+ : does2
+ DOES> ( ... -- ... )
+     ... ;
+ : def-word ( ... -- ... )
+     create ...
+     IF
+        does1
+     ELSE
+        does2
+     ENDIF ;
+ @end example
+ In this example, the selection of whether to use @code{does1} or
+ @code{does2} is made at definition-time; at the time that the child word is
+ @code{CREATE}d.
+ @cindex @code{DOES>} in interpretation state
+ In a standard program you can apply a @code{DOES>}-part only if the last
+ word was defined with @code{CREATE}. In Gforth, the @code{DOES>}-part
+ will override the behaviour of the last word defined in any case. In a
+ standard program, you can use @code{DOES>} only in a colon
+ definition. In Gforth, you can also use it in interpretation state, in a
+ kind of one-shot mode; for example:
+ @example
+ CREATE name ( ... -- ... )
+   @i{initialization}
+ DOES>
+   @i{code} ;
+ @end example
+ @noindent
+ is equivalent to the standard:
+ @example
+ :noname
+ DOES>
+     @i{code} ;
+ CREATE name EXECUTE ( ... -- ... )
+     @i{initialization}
+ @end example
+ doc->body
+ @node Advanced does> usage example,  , CREATE..DOES> details, User-defined Defining Words
+ @subsubsection Advanced does> usage example
+ The MIPS disassembler (@file{arch/mips/disasm.fs}) contains many words
+ for disassembling instructions, that follow a very repetetive scheme:
+ @example
+ :noname @var{disasm-operands} s" @var{inst-name}" type ;
+ @var{entry-num} cells @var{table} + !
+ @end example
+ Of course, this inspires the idea to factor out the commonalities to
+ allow a definition like
+ @example
+ @var{disasm-operands} @var{entry-num} @var{table} define-inst @var{inst-name}
+ @end example
+ The parameters @var{disasm-operands} and @var{table} are usually
+ correlated.  Moreover, before I wrote the disassembler, there already
+ existed code that defines instructions like this:
+ @example
+ @var{entry-num} @var{inst-format} @var{inst-name}
+ @end example
+ This code comes from the assembler and resides in
+ @file{arch/mips/insts.fs}.
+ So I had to define the @var{inst-format} words that performed the scheme
+ above when executed.  At first I chose to use run-time code-generation:
+ @example
+ : @var{inst-format} ( entry-num "name" -- ; compiled code: addr w -- )
+   :noname Postpone @var{disasm-operands}
+   name Postpone sliteral Postpone type Postpone ;
+   swap cells @var{table} + ! ;
+ @end example
+ Note that this supplies the other two parameters of the scheme above.
+ An alternative would have been to write this using
+ @code{create}/@code{does>}:
+ @example
+ : @var{inst-format} ( entry-num "name" -- )
+   here name string, ( entry-num c-addr ) \ parse and save "name"
+   noname create , ( entry-num )
+   lastxt swap cells @var{table} + !
+ does> ( addr w -- )
+   \ disassemble instruction w at addr
+   @@ >r
+   @var{disasm-operands}
+   r> count type ;
+ @end example
+ Somehow the first solution is simpler, mainly because it's simpler to
+ shift a string from definition-time to use-time with @code{sliteral}
+ than with @code{string,} and friends.
+ I wrote a lot of words following this scheme and soon thought about
+ factoring out the commonalities among them.  Note that this uses a
+ two-level defining word, i.e., a word that defines ordinary defining
+ words.
+ This time a solution involving @code{postpone} and friends seemed more
+ difficult (try it as an exercise), so I decided to use a
+ @code{create}/@code{does>} word; since I was already at it, I also used
+ @code{create}/@code{does>} for the lower level (try using
+ @code{postpone} etc. as an exercise), resulting in the following
+ definition:
+ @example
+ : define-format ( disasm-xt table-xt -- )
+     \ define an instruction format that uses disasm-xt for
+     \ disassembling and enters the defined instructions into table
+     \ table-xt
+     create 2,
+ does> ( u "inst" -- )
+     \ defines an anonymous word for disassembling instruction inst,
+     \ and enters it as u-th entry into table-xt
+@@ swap here name string, ( u table-xt disasm-xt c-addr ) \ remember string
+     noname create 2,      \ define anonymous word
+     execute lastxt swap ! \ enter xt of defined word into table-xt
+ does> ( addr w -- )
+     \ disassemble instruction w at addr
+@@ >r ( addr w disasm-xt R: c-addr )
+     execute ( R: c-addr ) \ disassemble operands
+     r> count type ; \ print name
+ @end example
+ Note that the tables here (in contrast to above) do the @code{cells +}
+ by themselves (that's why you have to pass an xt).  This word is used in
+ the following way:
+ @example
+ ' @var{disasm-operands} ' @var{table} define-format @var{inst-format}
+ @end example
+ As shown above, the defined instruction format is then used like this:
+ @example
+ @var{entry-num} @var{inst-format} @var{inst-name}
+ @end example
+ In terms of currying, this kind of two-level defining word provides the
+ parameters in three stages: first @var{disasm-operands} and @var{table},
+ then @var{entry-num} and @var{inst-name}, finally @code{addr w}, i.e.,
+ the instruction to be disassembled.
+ Of course this did not quite fit all the instruction format names used
+ in @file{insts.fs}, so I had to define a few wrappers that conditioned
+ the parameters into the right form.
+ If you have trouble following this section, don't worry.  First, this is
+ involved and takes time (and probably some playing around) to
+ understand; second, this is the first two-level
+ @code{create}/@code{does>} word I have written in seventeen years of
+ Forth; and if I did not have @file{insts.fs} to start with, I may well
+ have elected to use just a one-level defining word (with some repeating
+ of parameters when using the defining word). So it is not necessary to
+ understand this, but it may improve your understanding of Forth.
+ @node Deferred words, Aliases, User-defined Defining Words, Defining Words
+ @subsection Deferred words
+ @cindex deferred words
+ The defining word @code{Defer} allows you to define a word by name
+ without defining its behaviour; the definition of its behaviour is
+ deferred. Here are two situation where this can be useful:
+ @itemize @bullet
+ @item
+ Where you want to allow the behaviour of a word to be altered later, and
+ for all precompiled references to the word to change when its behaviour
+ is changed.
+ @item
+ For mutual recursion; @xref{Calls and returns}.
+ @end itemize
+ In the following example, @code{foo} always invokes the version of
+ @code{greet} that prints ``@code{Good morning}'' whilst @code{bar}
+ always invokes the version that prints ``@code{Hello}''. There is no way
+ of getting @code{foo} to use the later version without re-ordering the
+ source code and recompiling it.
+ @example
+ : greet ." Good morning" ;
+ : foo ... greet ... ;
+ : greet ." Hello" ;
+ : bar ... greet ... ;
+ @end example
+ This problem can be solved by defining @code{greet} as a @code{Defer}red
+ word. The behaviour of a @code{Defer}red word can be defined and
+ redefined at any time by using @code{IS} to associate the xt of a
+ previously-defined word with it. The previous example becomes:
+ @example
+ Defer greet ( -- )
+ : foo ... greet ... ;
+ : bar ... greet ... ;
+ : greet1 ( -- ) ." Good morning" ;
+ : greet2 ( -- ) ." Hello" ;
+ ' greet2 <IS> greet  \ make greet behave like greet2
+ @end example
+ @progstyle
+ You should write a stack comment for every deferred word, and put only
+ XTs into deferred words that conform to this stack effect.  Otherwise
+ it's too difficult to use the deferred word.
+ A deferred word can be used to improve the statistics-gathering example
+ from @ref{User-defined Defining Words}; rather than edit the
+ application's source code to change every @code{:} to a @code{my:}, do
+ this:
+ @example
+ : real: : ;     \ retain access to the original
+ defer :         \ redefine as a deferred word
+ ' my: <IS> :      \ use special version of :
+ \
+ \ load application here
+ \
+ ' real: <IS> :    \ go back to the original
+ @end example
+ One thing to note is that @code{<IS>} consumes its name when it is
+ executed.  If you want to specify the name at compile time, use
+ @code{[IS]}:
+ @example
+ : set-greet ( xt -- )
+   [IS] greet ;
+ ' greet1 set-greet
+ @end example
+ A deferred word can only inherit execution semantics from the xt
+ (because that is all that an xt can represent -- for more discussion of
+ this @pxref{Tokens for Words}); by default it will have default
+ interpretation and compilation semantics deriving from this execution
+ semantics.  However, you can change the interpretation and compilation
+ semantics of the deferred word in the usual ways:
+ @example
+ : bar .... ; compile-only
+ Defer fred immediate
+ Defer jim
+ ' bar <IS> jim  \ jim has default semantics
+ ' bar <IS> fred \ fred is immediate
+ @end example
+ doc-defer
+ doc-<is>
+ doc-[is]
+ doc-is
+ @comment TODO document these: what's defers [is]
+ doc-what's
+ doc-defers
+ @c Use @code{words-deferred} to see a list of deferred words.
+ Definitions in ANS Forth for @code{defer}, @code{<is>} and @code{[is]}
+ are provided in @file{compat/defer.fs}.
+ @node Aliases,  , Deferred words, Defining Words
+ @subsection Aliases
+ @cindex aliases
+ The defining word @code{Alias} allows you to define a word by name that
+ has the same behaviour as some other word. Here are two situation where
+ this can be useful:
+ @itemize @bullet
+ @item
+ When you want access to a word's definition from a different word list
+ (for an example of this, see the definition of the @code{Root} word list
+ in the Gforth source).
+ @item
+ When you want to create a synonym; a definition that can be known by
+ either of two names (for example, @code{THEN} and @code{ENDIF} are
+ aliases).
+ @end itemize
+ Like deferred words, an alias has default compilation and interpretation
+ semantics at the beginning (not the modifications of the other word),
+ but you can change them in the usual ways (@code{immediate},
+ @code{compile-only}). For example:
+ @example
+ : foo ... ; immediate
+ ' foo Alias bar \ bar is not an immediate word
+ ' foo Alias fooby immediate \ fooby is an immediate word
+ @end example
+ Words that are aliases have the same xt, different headers in the
+ dictionary, and consequently different name tokens (@pxref{Tokens for
+ Words}) and possibly different immediate flags.  An alias can only have
+ default or immediate compilation semantics; you can define aliases for
+ combined words with @code{interpret/compile:} -- see @ref{Combined words}.
+ doc-alias
+ @node Interpretation and Compilation Semantics, Tokens for Words, Defining Words, Words
+ @section Interpretation and Compilation Semantics
+ @cindex semantics, interpretation and compilation
+ @c !! state and ' are used without explanation
+ @c example for immediate/compile-only? or is the tutorial enough
+ @cindex interpretation semantics
+ The @dfn{interpretation semantics} of a (named) word are what the text
+ interpreter does when it encounters the word in interpret state. It also
+ appears in some other contexts, e.g., the execution token returned by
+ @code{' @i{word}} identifies the interpretation semantics of @i{word}
+ (in other words, @code{' @i{word} execute} is equivalent to
+ interpret-state text interpretation of @code{@i{word}}).
+ @cindex compilation semantics
+ The @dfn{compilation semantics} of a (named) word are what the text
+ interpreter does when it encounters the word in compile state. It also
+ appears in other contexts, e.g, @code{POSTPONE @i{word}}
+ compiles@footnote{In standard terminology, ``appends to the current
+ definition''.} the compilation semantics of @i{word}.
+ @cindex execution semantics
+ The standard also talks about @dfn{execution semantics}. They are used
+ only for defining the interpretation and compilation semantics of many
+ words. By default, the interpretation semantics of a word are to
+ @code{execute} its execution semantics, and the compilation semantics of
+ a word are to @code{compile,} its execution semantics.@footnote{In
+ standard terminology: The default interpretation semantics are its
+ execution semantics; the default compilation semantics are to append its
+ execution semantics to the execution semantics of the current
+ definition.}
+ Unnamed words (@pxref{Anonymous Definitions}) cannot be encountered by
+ the text interpreter, ticked, or @code{postpone}d, so they have no
+ interpretation or compilation semantics.  Their behaviour is represented
+ by their XT (@pxref{Tokens for Words}), and we call it execution
+ semantics, too.
+ @comment TODO expand, make it co-operate with new sections on text interpreter.
+ @cindex immediate words
+ @cindex compile-only words
+ You can change the semantics of the most-recently defined word:
+ doc-immediate
+ doc-compile-only
+ doc-restrict
+ Note that ticking (@code{'}) a compile-only word gives an error
+ (``Interpreting a compile-only word'').
+ @menu
+ * Combined words::
+ @end menu
+ @node Combined words,  , Interpretation and Compilation Semantics, Interpretation and Compilation Semantics
+ @subsection Combined Words
+ @cindex combined words
+ Gforth allows you to define @dfn{combined words} -- words that have an
+ arbitrary combination of interpretation and compilation semantics.
+ doc-interpret/compile:
+ This feature was introduced for implementing @code{TO} and @code{S"}. I
+ recommend that you do not define such words, as cute as they may be:
+ they make it hard to get at both parts of the word in some contexts.
+ E.g., assume you want to get an execution token for the compilation
+ part. Instead, define two words, one that embodies the interpretation
+ part, and one that embodies the compilation part.  Once you have done
+ that, you can define a combined word with @code{interpret/compile:} for
+ the convenience of your users.
+ You might try to use this feature to provide an optimizing
+ implementation of the default compilation semantics of a word. For
+ example, by defining:
+ @example
+ :noname
+    foo bar ;
+ :noname
+    POSTPONE foo POSTPONE bar ;
+ interpret/compile: opti-foobar
+ @end example
+ @noindent
+ as an optimizing version of:
+ @example
+ : foobar
+     foo bar ;
+ @end example
+ Unfortunately, this does not work correctly with @code{[compile]},
+ because @code{[compile]} assumes that the compilation semantics of all
+ @code{interpret/compile:} words are non-default. I.e., @code{[compile]
+ opti-foobar} would compile compilation semantics, whereas
+ @code{[compile] foobar} would compile interpretation semantics.
+ @cindex state-smart words (are a bad idea)
+ Some people try to use @dfn{state-smart} words to emulate the feature provided
+ by @code{interpret/compile:} (words are state-smart if they check
+ @code{STATE} during execution). E.g., they would try to code
+ @code{foobar} like this:
+ @example
+ : foobar
+   STATE @@
+   IF ( compilation state )
+     POSTPONE foo POSTPONE bar
+   ELSE
+     foo bar
+   ENDIF ; immediate
+ @end example
+ Although this works if @code{foobar} is only processed by the text
+ interpreter, it does not work in other contexts (like @code{'} or
+ @code{POSTPONE}). E.g., @code{' foobar} will produce an execution token
+ for a state-smart word, not for the interpretation semantics of the
+ original @code{foobar}; when you execute this execution token (directly
+ with @code{EXECUTE} or indirectly through @code{COMPILE,}) in compile
+ state, the result will not be what you expected (i.e., it will not
+ perform @code{foo bar}). State-smart words are a bad idea. Simply don't
+ write them@footnote{For a more detailed discussion of this topic, see
+ M. Anton Ertl,
+ @cite{@uref{http://www.complang.tuwien.ac.at/papers/ertl98.ps.gz,@code{State}-smartness---Why
+ it is Evil and How to Exorcise it}}, EuroForth '98.}!
+ @cindex defining words with arbitrary semantics combinations
+ It is also possible to write defining words that define words with
+ arbitrary combinations of interpretation and compilation semantics. In
+ general, they look like this:
+ @example
+ : def-word
+     create-interpret/compile
+     @i{code1}
+ interpretation>
+     @i{code2}
+ <interpretation
+ compilation>
+     @i{code3}
+ <compilation ;
+ @end example
+ For a @i{word} defined with @code{def-word}, the interpretation
+ semantics are to push the address of the body of @i{word} and perform
+ @i{code2}, and the compilation semantics are to push the address of
+ the body of @i{word} and perform @i{code3}. E.g., @code{constant}
+ can also be defined like this (except that the defined constants don't
+ behave correctly when @code{[compile]}d):
+ @example
+ : constant ( n "name" -- )
+     create-interpret/compile
+     ,
+ interpretation> ( -- n )
+     @@
+ <interpretation
+ compilation> ( compilation. -- ; run-time. -- n )
+     @@ postpone literal
+ <compilation ;
+ @end example
+ doc-create-interpret/compile
+ doc-interpretation>
+ doc-<interpretation
+ doc-compilation>
+ doc-<compilation
+ Words defined with @code{interpret/compile:} and
+ @code{create-interpret/compile} have an extended header structure that
+ differs from other words; however, unless you try to access them with
+ plain address arithmetic, you should not notice this. Words for
+ accessing the header structure usually know how to deal with this; e.g.,
+ @code{'} @i{word} @code{>body} also gives you the body of a word created
+ with @code{create-interpret/compile}.
+ doc-postpone
+ @comment TODO -- expand glossary text for POSTPONE
+ @c -------------------------------------------------------------
+ @node Tokens for Words, The Text Interpreter, Interpretation and Compilation Semantics, Words
+ @section Tokens for Words
+ @cindex tokens for words
+ This section describes the creation and use of tokens that represent
+ words.
+ @menu
+ * Execution token::             represents execution/interpretation semantics
+ * Compilation token::           represents compilation semantics
+ * Name token::                  represents named words
+ @end menu
+ @node Execution token, Compilation token, Tokens for Words, Tokens for Words
+ @subsection Execution token
+ @cindex xt
+ @cindex execution token
+ An @dfn{execution token} (@i{XT}) represents some behaviour of a word.
+ You can use @code{execute} to invoke this behaviour.
+ @cindex tick (')
+ You can use @code{'} to get an execution token that represents the
+ interpretation semantics of a named word:
+ @example
+' .
+ execute