File:  [gforth] / gforth / Attic / gforth.ds
Revision 1.18: download - view: text, annotated - select for diffs
Sat Oct 7 17:38:14 1995 UTC (25 years ago) by anton
Branches: MAIN
CVS tags: HEAD
added code.fs (code, ;code, end-code, assembler)
renamed dostruc to dofield
made index and doc-entries nicer
Only words containing 'e' or 'E' are converted to FP numbers.
added many wordset comments
added flush-icache primitive and FLUSH_ICACHE macro
added +DO, U+DO, -DO, U-DO and -LOOP
added code address labels (`docol:' etc.)
fixed sparc cache_flush

    1: \input texinfo   @c -*-texinfo-*-
    2: @comment The source is gforth.ds, from which gforth.texi is generated
    3: @comment %**start of header (This is for running Texinfo on a region.)
    4: @setfilename gforth.info
    5: @settitle Gforth Manual
    6: @comment @setchapternewpage odd
    7: @comment %**end of header (This is for running Texinfo on a region.)
    8: 
    9: @ifinfo
   10: This file documents Gforth 0.1
   11: 
   12: Copyright @copyright{} 1994 Gforth Development Group
   13: 
   14:      Permission is granted to make and distribute verbatim copies of
   15:      this manual provided the copyright notice and this permission notice
   16:      are preserved on all copies.
   17:      
   18: @ignore
   19:      Permission is granted to process this file through TeX and print the
   20:      results, provided the printed document carries a copying permission
   21:      notice identical to this one except for the removal of this paragraph
   22:      (this paragraph not being relevant to the printed manual).
   23:      
   24: @end ignore
   25:      Permission is granted to copy and distribute modified versions of this
   26:      manual under the conditions for verbatim copying, provided also that the
   27:      sections entitled "Distribution" and "General Public License" are
   28:      included exactly as in the original, and provided that the entire
   29:      resulting derived work is distributed under the terms of a permission
   30:      notice identical to this one.
   31:      
   32:      Permission is granted to copy and distribute translations of this manual
   33:      into another language, under the above conditions for modified versions,
   34:      except that the sections entitled "Distribution" and "General Public
   35:      License" may be included in a translation approved by the author instead
   36:      of in the original English.
   37: @end ifinfo
   38: 
   39: @titlepage
   40: @sp 10
   41: @center @titlefont{Gforth Manual}
   42: @sp 2
   43: @center for version 0.1
   44: @sp 2
   45: @center Anton Ertl
   46: @sp 3
   47: @center This manual is under construction
   48: 
   49: @comment  The following two commands start the copyright page.
   50: @page
   51: @vskip 0pt plus 1filll
   52: Copyright @copyright{} 1994 Gforth Development Group
   53: 
   54: @comment !! Published by ... or You can get a copy of this manual ...
   55: 
   56:      Permission is granted to make and distribute verbatim copies of
   57:      this manual provided the copyright notice and this permission notice
   58:      are preserved on all copies.
   59:      
   60:      Permission is granted to copy and distribute modified versions of this
   61:      manual under the conditions for verbatim copying, provided also that the
   62:      sections entitled "Distribution" and "General Public License" are
   63:      included exactly as in the original, and provided that the entire
   64:      resulting derived work is distributed under the terms of a permission
   65:      notice identical to this one.
   66:      
   67:      Permission is granted to copy and distribute translations of this manual
   68:      into another language, under the above conditions for modified versions,
   69:      except that the sections entitled "Distribution" and "General Public
   70:      License" may be included in a translation approved by the author instead
   71:      of in the original English.
   72: @end titlepage
   73: 
   74: 
   75: @node Top, License, (dir), (dir)
   76: @ifinfo
   77: Gforth is a free implementation of ANS Forth available on many
   78: personal machines. This manual corresponds to version 0.0.
   79: @end ifinfo
   80: 
   81: @menu
   82: * License::                     
   83: * Goals::                       About the Gforth Project
   84: * Other Books::                 Things you might want to read
   85: * Invocation::                  Starting Gforth
   86: * Words::                       Forth words available in Gforth
   87: * ANS conformance::             Implementation-defined options etc.
   88: * Model::                       The abstract machine of Gforth
   89: * Emacs and Gforth::            The Gforth Mode
   90: * Internals::                   Implementation details
   91: * Bugs::                        How to report them
   92: * Pedigree::                    Ancestors of Gforth
   93: * Word Index::                  An item for each Forth word
   94: * Node Index::                  An item for each node
   95: @end menu
   96: 
   97: @node License, Goals, Top, Top
   98: @unnumbered License
   99: !! Insert GPL here
  100: 
  101: @iftex
  102: @unnumbered Preface
  103: This manual documents Gforth. The reader is expected to know
  104: Forth. This manual is primarily a reference manual. @xref{Other Books}
  105: for introductory material.
  106: @end iftex
  107: 
  108: @node    Goals, Other Books, License, Top
  109: @comment node-name,     next,           previous, up
  110: @chapter Goals of Gforth
  111: @cindex Goals
  112: The goal of the Gforth Project is to develop a standard model for
  113: ANSI Forth. This can be split into several subgoals:
  114: 
  115: @itemize @bullet
  116: @item
  117: Gforth should conform to the ANSI Forth standard.
  118: @item
  119: It should be a model, i.e. it should define all the
  120: implementation-dependent things.
  121: @item
  122: It should become standard, i.e. widely accepted and used. This goal
  123: is the most difficult one.
  124: @end itemize
  125: 
  126: To achieve these goals Gforth should be
  127: @itemize @bullet
  128: @item
  129: Similar to previous models (fig-Forth, F83)
  130: @item
  131: Powerful. It should provide for all the things that are considered
  132: necessary today and even some that are not yet considered necessary.
  133: @item
  134: Efficient. It should not get the reputation of being exceptionally
  135: slow.
  136: @item
  137: Free.
  138: @item
  139: Available on many machines/easy to port.
  140: @end itemize
  141: 
  142: Have we achieved these goals? Gforth conforms to the ANS Forth
  143: standard. It may be considered a model, but we have not yet documented
  144: which parts of the model are stable and which parts we are likely to
  145: change. It certainly has not yet become a de facto standard. It has some
  146: similarities and some differences to previous models. It has some
  147: powerful features, but not yet everything that we envisioned. We
  148: certainly have achieved our execution speed goals (@pxref{Performance}).
  149: It is free and available on many machines.
  150: 
  151: @node Other Books, Invocation, Goals, Top
  152: @chapter Other books on ANS Forth
  153: 
  154: As the standard is relatively new, there are not many books out yet. It
  155: is not recommended to learn Forth by using Gforth and a book that is
  156: not written for ANS Forth, as you will not know your mistakes from the
  157: deviations of the book.
  158: 
  159: There is, of course, the standard, the definite reference if you want to
  160: write ANS Forth programs. It will be available in printed form from
  161: Global Engineering Documents !! somtime in spring or summer 1994. If you
  162: are lucky, you can still get dpANS6 (the draft that was approved as
  163: standard) by aftp from ftp.uu.net:/vendor/minerva/x3j14.
  164: 
  165: @cite{Forth: The new model} by Jack Woehr (!! Publisher) is an
  166: introductory book based on a draft version of the standard. It does not
  167: cover the whole standard. It also contains interesting background
  168: information (Jack Woehr was in the ANS Forth Technical Committe). It is
  169: not appropriate for complete newbies, but programmers experienced in
  170: other languages should find it ok.
  171: 
  172: @node Invocation, Words, Other Books, Top
  173: @chapter Invocation
  174: 
  175: You will usually just say @code{gforth}. In many other cases the default
  176: Gforth image will be invoked like this:
  177: 
  178: @example
  179: gforth [files] [-e forth-code]
  180: @end example
  181: 
  182: executing the contents of the files and the Forth code in the order they
  183: are given.
  184: 
  185: In general, the command line looks like this:
  186: 
  187: @example
  188: gforth [initialization options] [image-specific options]
  189: @end example
  190: 
  191: The initialization options must come before the rest of the command
  192: line. They are:
  193: 
  194: @table @code
  195: @item --image-file @var{file}
  196: Loads the Forth image @var{file} instead of the default
  197: @file{gforth.fi}.
  198: 
  199: @item --path @var{path}
  200: Uses @var{path} for searching the image file and Forth source code
  201: files instead of the default in the environment variable
  202: @code{GFORTHPATH} or the path specified at installation time (typically
  203: @file{/usr/local/lib/gforth:.}). A path is given as a @code{:}-separated
  204: list.
  205: 
  206: @item --dictionary-size @var{size}
  207: @item -m @var{size}
  208: Allocate @var{size} space for the Forth dictionary space instead of
  209: using the default specified in the image (typically 256K). The
  210: @var{size} specification consists of an integer and a unit (e.g.,
  211: @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
  212: size, in this case Cells), @code{k} (kilobytes), and @code{M}
  213: (Megabytes). If no unit is specified, @code{e} is used.
  214: 
  215: @item --data-stack-size @var{size}
  216: @item -d @var{size}
  217: Allocate @var{size} space for the data stack instead of using the
  218: default specified in the image (typically 16K).
  219: 
  220: @item --return-stack-size @var{size}
  221: @item -r @var{size}
  222: Allocate @var{size} space for the return stack instead of using the
  223: default specified in the image (typically 16K).
  224: 
  225: @item --fp-stack-size @var{size}
  226: @item -f @var{size}
  227: Allocate @var{size} space for the floating point stack instead of
  228: using the default specified in the image (typically 16K). In this case
  229: the unit specifier @code{e} refers to floating point numbers.
  230: 
  231: @item --locals-stack-size @var{size}
  232: @item -l @var{size}
  233: Allocate @var{size} space for the locals stack instead of using the
  234: default specified in the image (typically 16K).
  235: 
  236: @end table
  237: 
  238: As explained above, the image-specific command-line arguments for the
  239: default image @file{gforth.fi} consist of a sequence of filenames and
  240: @code{-e @var{forth-code}} options that are interpreted in the seqence
  241: in which they are given. The @code{-e @var{forth-code}} or
  242: @code{--evaluate @var{forth-code}} option evaluates the forth
  243: code. This option takes only one argument; if you want to evaluate more
  244: Forth words, you have to quote them or use several @code{-e}s. To exit
  245: after processing the command line (instead of entering interactive mode)
  246: append @code{-e bye} to the command line.
  247: 
  248: Not yet implemented:
  249: On startup the system first executes the system initialization file
  250: (unless the option @code{--no-init-file} is given; note that the system
  251: resulting from using this option may not be ANS Forth conformant). Then
  252: the user initialization file @file{.gforth.fs} is executed, unless the
  253: option @code{--no-rc} is given; this file is first searched in @file{.},
  254: then in @file{~}, then in the normal path (see above).
  255: 
  256: @node Words, ANS conformance, Invocation, Top
  257: @chapter Forth Words
  258: 
  259: @menu
  260: * Notation::                    
  261: * Arithmetic::                  
  262: * Stack Manipulation::          
  263: * Memory access::               
  264: * Control Structures::          
  265: * Locals::                      
  266: * Defining Words::              
  267: * Wordlists::                   
  268: * Files::                       
  269: * Blocks::                      
  270: * Other I/O::                   
  271: * Programming Tools::           
  272: * Assembler and Code words::    
  273: * Threading Words::             
  274: @end menu
  275: 
  276: @node Notation, Arithmetic, Words, Words
  277: @section Notation
  278: 
  279: The Forth words are described in this section in the glossary notation
  280: that has become a de-facto standard for Forth texts, i.e.
  281: 
  282: @format
  283: @var{word}     @var{Stack effect}   @var{wordset}   @var{pronunciation}
  284: @end format
  285: @var{Description}
  286: 
  287: @table @var
  288: @item word
  289: The name of the word. BTW, Gforth is case insensitive, so you can
  290: type the words in in lower case (However, @pxref{core-idef}).
  291: 
  292: @item Stack effect
  293: The stack effect is written in the notation @code{@var{before} --
  294: @var{after}}, where @var{before} and @var{after} describe the top of
  295: stack entries before and after the execution of the word. The rest of
  296: the stack is not touched by the word. The top of stack is rightmost,
  297: i.e., a stack sequence is written as it is typed in. Note that Gforth
  298: uses a separate floating point stack, but a unified stack
  299: notation. Also, return stack effects are not shown in @var{stack
  300: effect}, but in @var{Description}. The name of a stack item describes
  301: the type and/or the function of the item. See below for a discussion of
  302: the types.
  303: 
  304: @item pronunciation
  305: How the word is pronounced
  306: 
  307: @item wordset
  308: The ANS Forth standard is divided into several wordsets. A standard
  309: system need not support all of them. So, the fewer wordsets your program
  310: uses the more portable it will be in theory. However, we suspect that
  311: most ANS Forth systems on personal machines will feature all
  312: wordsets. Words that are not defined in the ANS standard have
  313: @code{gforth} as wordset.
  314: 
  315: @item Description
  316: A description of the behaviour of the word.
  317: @end table
  318: 
  319: The type of a stack item is specified by the character(s) the name
  320: starts with:
  321: 
  322: @table @code
  323: @item f
  324: Bool, i.e. @code{false} or @code{true}.
  325: @item c
  326: Char
  327: @item w
  328: Cell, can contain an integer or an address
  329: @item n
  330: signed integer
  331: @item u
  332: unsigned integer
  333: @item d
  334: double sized signed integer
  335: @item ud
  336: double sized unsigned integer
  337: @item r
  338: Float
  339: @item a_
  340: Cell-aligned address
  341: @item c_
  342: Char-aligned address (note that a Char is two bytes in Windows NT)
  343: @item f_
  344: Float-aligned address
  345: @item df_
  346: Address aligned for IEEE double precision float
  347: @item sf_
  348: Address aligned for IEEE single precision float
  349: @item xt
  350: Execution token, same size as Cell
  351: @item wid
  352: Wordlist ID, same size as Cell
  353: @item f83name
  354: Pointer to a name structure
  355: @end table
  356: 
  357: @node Arithmetic, Stack Manipulation, Notation, Words
  358: @section Arithmetic
  359: Forth arithmetic is not checked, i.e., you will not hear about integer
  360: overflow on addition or multiplication, you may hear about division by
  361: zero if you are lucky. The operator is written after the operands, but
  362: the operands are still in the original order. I.e., the infix @code{2-1}
  363: corresponds to @code{2 1 -}. Forth offers a variety of division
  364: operators. If you perform division with potentially negative operands,
  365: you do not want to use @code{/} or @code{/mod} with its undefined
  366: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
  367: former, @pxref{Mixed precision}).
  368: 
  369: @menu
  370: * Single precision::            
  371: * Bitwise operations::          
  372: * Mixed precision::             operations with single and double-cell integers
  373: * Double precision::            Double-cell integer arithmetic
  374: * Floating Point::              
  375: @end menu
  376: 
  377: @node Single precision, Bitwise operations, Arithmetic, Arithmetic
  378: @subsection Single precision
  379: doc-+
  380: doc--
  381: doc-*
  382: doc-/
  383: doc-mod
  384: doc-/mod
  385: doc-negate
  386: doc-abs
  387: doc-min
  388: doc-max
  389: 
  390: @node Bitwise operations, Mixed precision, Single precision, Arithmetic
  391: @subsection Bitwise operations
  392: doc-and
  393: doc-or
  394: doc-xor
  395: doc-invert
  396: doc-2*
  397: doc-2/
  398: 
  399: @node Mixed precision, Double precision, Bitwise operations, Arithmetic
  400: @subsection Mixed precision
  401: doc-m+
  402: doc-*/
  403: doc-*/mod
  404: doc-m*
  405: doc-um*
  406: doc-m*/
  407: doc-um/mod
  408: doc-fm/mod
  409: doc-sm/rem
  410: 
  411: @node Double precision, Floating Point, Mixed precision, Arithmetic
  412: @subsection Double precision
  413: 
  414: The outer (aka text) interpreter converts numbers containing a dot into
  415: a double precision number. Note that only numbers with the dot as last
  416: character are standard-conforming.
  417: 
  418: doc-d+
  419: doc-d-
  420: doc-dnegate
  421: doc-dabs
  422: doc-dmin
  423: doc-dmax
  424: 
  425: @node Floating Point,  , Double precision, Arithmetic
  426: @subsection Floating Point
  427: 
  428: The format of floating point numbers recognized by the outer (aka text)
  429: interpreter is: a signed decimal number, possibly containing a decimal
  430: point (@code{.}), followed by @code{E} or @code{e}, optionally followed
  431: by a signed integer (the exponent). E.g., @code{1e} ist the same as
  432: @code{+1.0e+1}. Note that a number without @code{e}
  433: is not interpreted as floating-point number, but as double (if the
  434: number contains a @code{.}) or single precision integer. Also,
  435: conversions between string and floating point numbers always use base
  436: 10, irrespective of the value of @code{BASE}. If @code{BASE} contains a
  437: value greater then 14, the @code{E} may be interpreted as digit and the
  438: number will be interpreted as integer, unless it has a signed exponent
  439: (both @code{+} and @code{-} are allowed as signs).
  440: 
  441: Angles in floating point operations are given in radians (a full circle
  442: has 2 pi radians). Note, that Gforth has a separate floating point
  443: stack, but we use the unified notation.
  444: 
  445: Floating point numbers have a number of unpleasant surprises for the
  446: unwary (e.g., floating point addition is not associative) and even a few
  447: for the wary. You should not use them unless you know what you are doing
  448: or you don't care that the results you get are totally bogus. If you
  449: want to learn about the problems of floating point numbers (and how to
  450: avoid them), you might start with @cite{David Goldberg, What Every
  451: Computer Scientist Should Know About Floating-Point Arithmetic, ACM
  452: Computing Surveys 23(1):5@minus{}48, March 1991}.
  453: 
  454: doc-f+
  455: doc-f-
  456: doc-f*
  457: doc-f/
  458: doc-fnegate
  459: doc-fabs
  460: doc-fmax
  461: doc-fmin
  462: doc-floor
  463: doc-fround
  464: doc-f**
  465: doc-fsqrt
  466: doc-fexp
  467: doc-fexpm1
  468: doc-fln
  469: doc-flnp1
  470: doc-flog
  471: doc-falog
  472: doc-fsin
  473: doc-fcos
  474: doc-fsincos
  475: doc-ftan
  476: doc-fasin
  477: doc-facos
  478: doc-fatan
  479: doc-fatan2
  480: doc-fsinh
  481: doc-fcosh
  482: doc-ftanh
  483: doc-fasinh
  484: doc-facosh
  485: doc-fatanh
  486: 
  487: @node Stack Manipulation, Memory access, Arithmetic, Words
  488: @section Stack Manipulation
  489: 
  490: Gforth has a data stack (aka parameter stack) for characters, cells,
  491: addresses, and double cells, a floating point stack for floating point
  492: numbers, a return stack for storing the return addresses of colon
  493: definitions and other data, and a locals stack for storing local
  494: variables. Note that while every sane Forth has a separate floating
  495: point stack, this is not strictly required; an ANS Forth system could
  496: theoretically keep floating point numbers on the data stack. As an
  497: additional difficulty, you don't know how many cells a floating point
  498: number takes. It is reportedly possible to write words in a way that
  499: they work also for a unified stack model, but we do not recommend trying
  500: it. Instead, just say that your program has an environmental dependency
  501: on a separate FP stack.
  502: 
  503: Also, a Forth system is allowed to keep the local variables on the
  504: return stack. This is reasonable, as local variables usually eliminate
  505: the need to use the return stack explicitly. So, if you want to produce
  506: a standard complying program and if you are using local variables in a
  507: word, forget about return stack manipulations in that word (see the
  508: standard document for the exact rules).
  509: 
  510: @menu
  511: * Data stack::                  
  512: * Floating point stack::        
  513: * Return stack::                
  514: * Locals stack::                
  515: * Stack pointer manipulation::  
  516: @end menu
  517: 
  518: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
  519: @subsection Data stack
  520: doc-drop
  521: doc-nip
  522: doc-dup
  523: doc-over
  524: doc-tuck
  525: doc-swap
  526: doc-rot
  527: doc--rot
  528: doc-?dup
  529: doc-pick
  530: doc-roll
  531: doc-2drop
  532: doc-2nip
  533: doc-2dup
  534: doc-2over
  535: doc-2tuck
  536: doc-2swap
  537: doc-2rot
  538: 
  539: @node Floating point stack, Return stack, Data stack, Stack Manipulation
  540: @subsection Floating point stack
  541: doc-fdrop
  542: doc-fnip
  543: doc-fdup
  544: doc-fover
  545: doc-ftuck
  546: doc-fswap
  547: doc-frot
  548: 
  549: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
  550: @subsection Return stack
  551: doc->r
  552: doc-r>
  553: doc-r@
  554: doc-rdrop
  555: doc-2>r
  556: doc-2r>
  557: doc-2r@
  558: doc-2rdrop
  559: 
  560: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
  561: @subsection Locals stack
  562: 
  563: @node Stack pointer manipulation,  , Locals stack, Stack Manipulation
  564: @subsection Stack pointer manipulation
  565: doc-sp@
  566: doc-sp!
  567: doc-fp@
  568: doc-fp!
  569: doc-rp@
  570: doc-rp!
  571: doc-lp@
  572: doc-lp!
  573: 
  574: @node Memory access, Control Structures, Stack Manipulation, Words
  575: @section Memory access
  576: 
  577: @menu
  578: * Stack-Memory transfers::      
  579: * Address arithmetic::          
  580: * Memory block access::         
  581: @end menu
  582: 
  583: @node Stack-Memory transfers, Address arithmetic, Memory access, Memory access
  584: @subsection Stack-Memory transfers
  585: 
  586: doc-@
  587: doc-!
  588: doc-+!
  589: doc-c@
  590: doc-c!
  591: doc-2@
  592: doc-2!
  593: doc-f@
  594: doc-f!
  595: doc-sf@
  596: doc-sf!
  597: doc-df@
  598: doc-df!
  599: 
  600: @node Address arithmetic, Memory block access, Stack-Memory transfers, Memory access
  601: @subsection Address arithmetic
  602: 
  603: ANS Forth does not specify the sizes of the data types. Instead, it
  604: offers a number of words for computing sizes and doing address
  605: arithmetic. Basically, address arithmetic is performed in terms of
  606: address units (aus); on most systems the address unit is one byte. Note
  607: that a character may have more than one au, so @code{chars} is no noop
  608: (on systems where it is a noop, it compiles to nothing).
  609: 
  610: ANS Forth also defines words for aligning addresses for specific
  611: addresses. Many computers require that accesses to specific data types
  612: must only occur at specific addresses; e.g., that cells may only be
  613: accessed at addresses divisible by 4. Even if a machine allows unaligned
  614: accesses, it can usually perform aligned accesses faster. 
  615: 
  616: For the performance-conscious: alignment operations are usually only
  617: necessary during the definition of a data structure, not during the
  618: (more frequent) accesses to it.
  619: 
  620: ANS Forth defines no words for character-aligning addresses. This is not
  621: an oversight, but reflects the fact that addresses that are not
  622: char-aligned have no use in the standard and therefore will not be
  623: created.
  624: 
  625: The standard guarantees that addresses returned by @code{CREATE}d words
  626: are cell-aligned; in addition, Gforth guarantees that these addresses
  627: are aligned for all purposes.
  628: 
  629: Note that the standard defines a word @code{char}, which has nothing to
  630: do with address arithmetic.
  631: 
  632: doc-chars
  633: doc-char+
  634: doc-cells
  635: doc-cell+
  636: doc-align
  637: doc-aligned
  638: doc-floats
  639: doc-float+
  640: doc-falign
  641: doc-faligned
  642: doc-sfloats
  643: doc-sfloat+
  644: doc-sfalign
  645: doc-sfaligned
  646: doc-dfloats
  647: doc-dfloat+
  648: doc-dfalign
  649: doc-dfaligned
  650: doc-maxalign
  651: doc-maxaligned
  652: doc-cfalign
  653: doc-cfaligned
  654: doc-address-unit-bits
  655: 
  656: @node Memory block access,  , Address arithmetic, Memory access
  657: @subsection Memory block access
  658: 
  659: doc-move
  660: doc-erase
  661: 
  662: While the previous words work on address units, the rest works on
  663: characters.
  664: 
  665: doc-cmove
  666: doc-cmove>
  667: doc-fill
  668: doc-blank
  669: 
  670: @node Control Structures, Locals, Memory access, Words
  671: @section Control Structures
  672: 
  673: Control structures in Forth cannot be used in interpret state, only in
  674: compile state, i.e., in a colon definition. We do not like this
  675: limitation, but have not seen a satisfying way around it yet, although
  676: many schemes have been proposed.
  677: 
  678: @menu
  679: * Selection::                   
  680: * Simple Loops::                
  681: * Counted Loops::               
  682: * Arbitrary control structures::  
  683: * Calls and returns::           
  684: * Exception Handling::          
  685: @end menu
  686: 
  687: @node Selection, Simple Loops, Control Structures, Control Structures
  688: @subsection Selection
  689: 
  690: @example
  691: @var{flag}
  692: IF
  693:   @var{code}
  694: ENDIF
  695: @end example
  696: or
  697: @example
  698: @var{flag}
  699: IF
  700:   @var{code1}
  701: ELSE
  702:   @var{code2}
  703: ENDIF
  704: @end example
  705: 
  706: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
  707: standard, and @code{ENDIF} is not, although it is quite popular. We
  708: recommend using @code{ENDIF}, because it is less confusing for people
  709: who also know other languages (and is not prone to reinforcing negative
  710: prejudices against Forth in these people). Adding @code{ENDIF} to a
  711: system that only supplies @code{THEN} is simple:
  712: @example
  713: : endif   POSTPONE then ; immediate
  714: @end example
  715: 
  716: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
  717: (adv.)}  has the following meanings:
  718: @quotation
  719: ... 2b: following next after in order ... 3d: as a necessary consequence
  720: (if you were there, then you saw them).
  721: @end quotation
  722: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
  723: and many other programming languages has the meaning 3d.]
  724: 
  725: We also provide the words @code{?dup-if} and @code{?dup-0=-if}, so you
  726: can avoid using @code{?dup}.
  727: 
  728: @example
  729: @var{n}
  730: CASE
  731:   @var{n1} OF @var{code1} ENDOF
  732:   @var{n2} OF @var{code2} ENDOF
  733:   @dots{}
  734: ENDCASE
  735: @end example
  736: 
  737: Executes the first @var{codei}, where the @var{ni} is equal to
  738: @var{n}. A default case can be added by simply writing the code after
  739: the last @code{ENDOF}. It may use @var{n}, which is on top of the stack,
  740: but must not consume it.
  741: 
  742: @node Simple Loops, Counted Loops, Selection, Control Structures
  743: @subsection Simple Loops
  744: 
  745: @example
  746: BEGIN
  747:   @var{code1}
  748:   @var{flag}
  749: WHILE
  750:   @var{code2}
  751: REPEAT
  752: @end example
  753: 
  754: @var{code1} is executed and @var{flag} is computed. If it is true,
  755: @var{code2} is executed and the loop is restarted; If @var{flag} is false, execution continues after the @code{REPEAT}.
  756: 
  757: @example
  758: BEGIN
  759:   @var{code}
  760:   @var{flag}
  761: UNTIL
  762: @end example
  763: 
  764: @var{code} is executed. The loop is restarted if @code{flag} is false.
  765: 
  766: @example
  767: BEGIN
  768:   @var{code}
  769: AGAIN
  770: @end example
  771: 
  772: This is an endless loop.
  773: 
  774: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
  775: @subsection Counted Loops
  776: 
  777: The basic counted loop is:
  778: @example
  779: @var{limit} @var{start}
  780: ?DO
  781:   @var{body}
  782: LOOP
  783: @end example
  784: 
  785: This performs one iteration for every integer, starting from @var{start}
  786: and up to, but excluding @var{limit}. The counter, aka index, can be
  787: accessed with @code{i}. E.g., the loop
  788: @example
  789: 10 0 ?DO
  790:   i .
  791: LOOP
  792: @end example
  793: prints
  794: @example
  795: 0 1 2 3 4 5 6 7 8 9
  796: @end example
  797: The index of the innermost loop can be accessed with @code{i}, the index
  798: of the next loop with @code{j}, and the index of the third loop with
  799: @code{k}.
  800: 
  801: The loop control data are kept on the return stack, so there are some
  802: restrictions on mixing return stack accesses and counted loop
  803: words. E.g., if you put values on the return stack outside the loop, you
  804: cannot read them inside the loop. If you put values on the return stack
  805: within a loop, you have to remove them before the end of the loop and
  806: before accessing the index of the loop.
  807: 
  808: There are several variations on the counted loop:
  809: 
  810: @code{LEAVE} leaves the innermost counted loop immediately.
  811: 
  812: If @var{start} is greater than @var{limit}, a @code{?DO} loop is entered
  813: (and @code{LOOP} iterates until they become equal by wrap-around
  814: arithmetic). This behaviour is usually not what you want. Therefore,
  815: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
  816: @code{?DO}), which do not enter the loop if @var{start} is greater than
  817: @var{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
  818: unsigned loop parameters. These words can be implemented easily on
  819: standard systems, so using them does not make your programs hard to
  820: port; e.g.:
  821: @example
  822: : +DO ( compile-time: -- do-sys; run-time: n1 n2 -- )
  823:     POSTPONE over POSTPONE min POSTPONE ?DO ; immediate
  824: @end example
  825: 
  826: @code{LOOP} can be replaced with @code{@var{n} +LOOP}; this updates the
  827: index by @var{n} instead of by 1. The loop is terminated when the border
  828: between @var{limit-1} and @var{limit} is crossed. E.g.:
  829: 
  830: @code{4 0 +DO  i .  2 +LOOP}   prints @code{0 2}
  831: 
  832: @code{4 1 +DO  i .  2 +LOOP}   prints @code{1 3}
  833: 
  834: The behaviour of @code{@var{n} +LOOP} is peculiar when @var{n} is negative:
  835: 
  836: @code{-1 0 ?DO  i .  -1 +LOOP}  prints @code{0 -1}
  837: 
  838: @code{ 0 0 ?DO  i .  -1 +LOOP}  prints nothing
  839: 
  840: Therefore we recommend avoiding @code{@var{n} +LOOP} with negative
  841: @var{n}. One alternative is @code{@var{u} -LOOP}, which reduces the
  842: index by @var{u} each iteration. The loop is terminated when the border
  843: between @var{limit+1} and @var{limit} is crossed. Gforth also provides
  844: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
  845: 
  846: @code{-2 0 -DO  i .  1 -LOOP}  prints @code{0 -1}
  847: 
  848: @code{-1 0 -DO  i .  1 -LOOP}  prints @code{0}
  849: 
  850: @code{ 0 0 -DO  i .  1 -LOOP}  prints nothing
  851: 
  852: Another alternative is @code{@var{n} S+LOOP}, where the negative
  853: case behaves symmetrical to the positive case:
  854: 
  855: @code{-2 0 -DO  i .  -1 S+LOOP}  prints @code{0 -1}
  856: 
  857: The loop is terminated when the border between @var{limit@minus{}sgn(n)}
  858: and @var{limit} is crossed. Unfortunately, neither @code{-LOOP} nor
  859: @code{S+LOOP} are part of the ANS Forth standard, and they are not easy
  860: to implement using standard words. If you want to write standard
  861: programs, just avoid counting down.
  862: 
  863: @code{?DO} can also be replaced by @code{DO}. @code{DO} always enters
  864: the loop, independent of the loop parameters. Do not use @code{DO}, even
  865: if you know that the loop is entered in any case. Such knowledge tends
  866: to become invalid during maintenance of a program, and then the
  867: @code{DO} will make trouble.
  868: 
  869: @code{UNLOOP} is used to prepare for an abnormal loop exit, e.g., via
  870: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
  871: return stack so @code{EXIT} can get to its return address.
  872: 
  873: Another counted loop is
  874: @example
  875: @var{n}
  876: FOR
  877:   @var{body}
  878: NEXT
  879: @end example
  880: This is the preferred loop of native code compiler writers who are too
  881: lazy to optimize @code{?DO} loops properly. In Gforth, this loop
  882: iterates @var{n+1} times; @code{i} produces values starting with @var{n}
  883: and ending with 0. Other Forth systems may behave differently, even if
  884: they support @code{FOR} loops.
  885: 
  886: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
  887: @subsection Arbitrary control structures
  888: 
  889: ANS Forth permits and supports using control structures in a non-nested
  890: way. Information about incomplete control structures is stored on the
  891: control-flow stack. This stack may be implemented on the Forth data
  892: stack, and this is what we have done in Gforth.
  893: 
  894: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
  895: entry represents a backward branch target. A few words are the basis for
  896: building any control structure possible (except control structures that
  897: need storage, like calls, coroutines, and backtracking).
  898: 
  899: doc-if
  900: doc-ahead
  901: doc-then
  902: doc-begin
  903: doc-until
  904: doc-again
  905: doc-cs-pick
  906: doc-cs-roll
  907: 
  908: On many systems control-flow stack items take one word, in Gforth they
  909: currently take three (this may change in the future). Therefore it is a
  910: really good idea to manipulate the control flow stack with
  911: @code{cs-pick} and @code{cs-roll}, not with data stack manipulation
  912: words.
  913: 
  914: Some standard control structure words are built from these words:
  915: 
  916: doc-else
  917: doc-while
  918: doc-repeat
  919: 
  920: Counted loop words constitute a separate group of words:
  921: 
  922: doc-?do
  923: doc-+do
  924: doc-u+do
  925: doc--do
  926: doc-u-do
  927: doc-do
  928: doc-for
  929: doc-loop
  930: doc-s+loop
  931: doc-+loop
  932: doc--loop
  933: doc-next
  934: doc-leave
  935: doc-?leave
  936: doc-unloop
  937: doc-done
  938: 
  939: The standard does not allow using @code{cs-pick} and @code{cs-roll} on
  940: @i{do-sys}. Our system allows it, but it's your job to ensure that for
  941: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
  942: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
  943: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
  944: resolved (by using one of the loop-ending words or @code{DONE}).
  945: 
  946: Another group of control structure words are
  947: 
  948: doc-case
  949: doc-endcase
  950: doc-of
  951: doc-endof
  952: 
  953: @i{case-sys} and @i{of-sys} cannot be processed using @code{cs-pick} and
  954: @code{cs-roll}.
  955: 
  956: @subsubsection Programming Style
  957: 
  958: In order to ensure readability we recommend that you do not create
  959: arbitrary control structures directly, but define new control structure
  960: words for the control structure you want and use these words in your
  961: program.
  962: 
  963: E.g., instead of writing
  964: 
  965: @example
  966: begin
  967:   ...
  968: if [ 1 cs-roll ]
  969:   ...
  970: again then
  971: @end example
  972: 
  973: we recommend defining control structure words, e.g.,
  974: 
  975: @example
  976: : while ( dest -- orig dest )
  977:  POSTPONE if
  978:  1 cs-roll ; immediate
  979: 
  980: : repeat ( orig dest -- )
  981:  POSTPONE again
  982:  POSTPONE then ; immediate
  983: @end example
  984: 
  985: and then using these to create the control structure:
  986: 
  987: @example
  988: begin
  989:   ...
  990: while
  991:   ...
  992: repeat
  993: @end example
  994: 
  995: That's much easier to read, isn't it? Of course, @code{BEGIN} and
  996: @code{WHILE} are predefined, so in this example it would not be
  997: necessary to define them.
  998: 
  999: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
 1000: @subsection Calls and returns
 1001: 
 1002: A definition can be called simply be writing the name of the
 1003: definition. When the end of the definition is reached, it returns. An
 1004: earlier return can be forced using
 1005: 
 1006: doc-exit
 1007: 
 1008: Don't forget to clean up the return stack and @code{UNLOOP} any
 1009: outstanding @code{?DO}...@code{LOOP}s before @code{EXIT}ing. The
 1010: primitive compiled by @code{EXIT} is
 1011: 
 1012: doc-;s
 1013: 
 1014: @node Exception Handling,  , Calls and returns, Control Structures
 1015: @subsection Exception Handling
 1016: 
 1017: doc-catch
 1018: doc-throw
 1019: 
 1020: @node Locals, Defining Words, Control Structures, Words
 1021: @section Locals
 1022: 
 1023: Local variables can make Forth programming more enjoyable and Forth
 1024: programs easier to read. Unfortunately, the locals of ANS Forth are
 1025: laden with restrictions. Therefore, we provide not only the ANS Forth
 1026: locals wordset, but also our own, more powerful locals wordset (we
 1027: implemented the ANS Forth locals wordset through our locals wordset).
 1028: 
 1029: @menu
 1030: * Gforth locals::               
 1031: * ANS Forth locals::            
 1032: @end menu
 1033: 
 1034: @node Gforth locals, ANS Forth locals, Locals, Locals
 1035: @subsection Gforth locals
 1036: 
 1037: Locals can be defined with
 1038: 
 1039: @example
 1040: @{ local1 local2 ... -- comment @}
 1041: @end example
 1042: or
 1043: @example
 1044: @{ local1 local2 ... @}
 1045: @end example
 1046: 
 1047: E.g.,
 1048: @example
 1049: : max @{ n1 n2 -- n3 @}
 1050:  n1 n2 > if
 1051:    n1
 1052:  else
 1053:    n2
 1054:  endif ;
 1055: @end example
 1056: 
 1057: The similarity of locals definitions with stack comments is intended. A
 1058: locals definition often replaces the stack comment of a word. The order
 1059: of the locals corresponds to the order in a stack comment and everything
 1060: after the @code{--} is really a comment.
 1061: 
 1062: This similarity has one disadvantage: It is too easy to confuse locals
 1063: declarations with stack comments, causing bugs and making them hard to
 1064: find. However, this problem can be avoided by appropriate coding
 1065: conventions: Do not use both notations in the same program. If you do,
 1066: they should be distinguished using additional means, e.g. by position.
 1067: 
 1068: The name of the local may be preceded by a type specifier, e.g.,
 1069: @code{F:} for a floating point value:
 1070: 
 1071: @example
 1072: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
 1073: \ complex multiplication
 1074:  Ar Br f* Ai Bi f* f-
 1075:  Ar Bi f* Ai Br f* f+ ;
 1076: @end example
 1077: 
 1078: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
 1079: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
 1080: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
 1081: with @code{W:}, @code{D:} etc.) produces its value and can be changed
 1082: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
 1083: produces its address (which becomes invalid when the variable's scope is
 1084: left). E.g., the standard word @code{emit} can be defined in therms of
 1085: @code{type} like this:
 1086: 
 1087: @example
 1088: : emit @{ C^ char* -- @}
 1089:     char* 1 type ;
 1090: @end example
 1091: 
 1092: A local without type specifier is a @code{W:} local. Both flavours of
 1093: locals are initialized with values from the data or FP stack.
 1094: 
 1095: Currently there is no way to define locals with user-defined data
 1096: structures, but we are working on it.
 1097: 
 1098: Gforth allows defining locals everywhere in a colon definition. This
 1099: poses the following questions:
 1100: 
 1101: @menu
 1102: * Where are locals visible by name?::  
 1103: * How long do locals live?::    
 1104: * Programming Style::           
 1105: * Implementation::              
 1106: @end menu
 1107: 
 1108: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
 1109: @subsubsection Where are locals visible by name?
 1110: 
 1111: Basically, the answer is that locals are visible where you would expect
 1112: it in block-structured languages, and sometimes a little longer. If you
 1113: want to restrict the scope of a local, enclose its definition in
 1114: @code{SCOPE}...@code{ENDSCOPE}.
 1115: 
 1116: doc-scope
 1117: doc-endscope
 1118: 
 1119: These words behave like control structure words, so you can use them
 1120: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
 1121: arbitrary ways.
 1122: 
 1123: If you want a more exact answer to the visibility question, here's the
 1124: basic principle: A local is visible in all places that can only be
 1125: reached through the definition of the local@footnote{In compiler
 1126: construction terminology, all places dominated by the definition of the
 1127: local.}. In other words, it is not visible in places that can be reached
 1128: without going through the definition of the local. E.g., locals defined
 1129: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
 1130: defined in @code{BEGIN}...@code{UNTIL} are visible after the
 1131: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
 1132: 
 1133: The reasoning behind this solution is: We want to have the locals
 1134: visible as long as it is meaningful. The user can always make the
 1135: visibility shorter by using explicit scoping. In a place that can
 1136: only be reached through the definition of a local, the meaning of a
 1137: local name is clear. In other places it is not: How is the local
 1138: initialized at the control flow path that does not contain the
 1139: definition? Which local is meant, if the same name is defined twice in
 1140: two independent control flow paths?
 1141: 
 1142: This should be enough detail for nearly all users, so you can skip the
 1143: rest of this section. If you relly must know all the gory details and
 1144: options, read on.
 1145: 
 1146: In order to implement this rule, the compiler has to know which places
 1147: are unreachable. It knows this automatically after @code{AHEAD},
 1148: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
 1149: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
 1150: compiler that the control flow never reaches that place. If
 1151: @code{UNREACHABLE} is not used where it could, the only consequence is
 1152: that the visibility of some locals is more limited than the rule above
 1153: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
 1154: lie to the compiler), buggy code will be produced.
 1155: 
 1156: Another problem with this rule is that at @code{BEGIN}, the compiler
 1157: does not know which locals will be visible on the incoming
 1158: back-edge. All problems discussed in the following are due to this
 1159: ignorance of the compiler (we discuss the problems using @code{BEGIN}
 1160: loops as examples; the discussion also applies to @code{?DO} and other
 1161: loops). Perhaps the most insidious example is:
 1162: @example
 1163: AHEAD
 1164: BEGIN
 1165:   x
 1166: [ 1 CS-ROLL ] THEN
 1167:   @{ x @}
 1168:   ...
 1169: UNTIL
 1170: @end example
 1171: 
 1172: This should be legal according to the visibility rule. The use of
 1173: @code{x} can only be reached through the definition; but that appears
 1174: textually below the use.
 1175: 
 1176: From this example it is clear that the visibility rules cannot be fully
 1177: implemented without major headaches. Our implementation treats common
 1178: cases as advertised and the exceptions are treated in a safe way: The
 1179: compiler makes a reasonable guess about the locals visible after a
 1180: @code{BEGIN}; if it is too pessimistic, the
 1181: user will get a spurious error about the local not being defined; if the
 1182: compiler is too optimistic, it will notice this later and issue a
 1183: warning. In the case above the compiler would complain about @code{x}
 1184: being undefined at its use. You can see from the obscure examples in
 1185: this section that it takes quite unusual control structures to get the
 1186: compiler into trouble, and even then it will often do fine.
 1187: 
 1188: If the @code{BEGIN} is reachable from above, the most optimistic guess
 1189: is that all locals visible before the @code{BEGIN} will also be
 1190: visible after the @code{BEGIN}. This guess is valid for all loops that
 1191: are entered only through the @code{BEGIN}, in particular, for normal
 1192: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
 1193: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
 1194: compiler. When the branch to the @code{BEGIN} is finally generated by
 1195: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
 1196: warns the user if it was too optimisitic:
 1197: @example
 1198: IF
 1199:   @{ x @}
 1200: BEGIN
 1201:   \ x ? 
 1202: [ 1 cs-roll ] THEN
 1203:   ...
 1204: UNTIL
 1205: @end example
 1206: 
 1207: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
 1208: optimistically assumes that it lives until the @code{THEN}. It notices
 1209: this difference when it compiles the @code{UNTIL} and issues a
 1210: warning. The user can avoid the warning, and make sure that @code{x}
 1211: is not used in the wrong area by using explicit scoping:
 1212: @example
 1213: IF
 1214:   SCOPE
 1215:   @{ x @}
 1216:   ENDSCOPE
 1217: BEGIN
 1218: [ 1 cs-roll ] THEN
 1219:   ...
 1220: UNTIL
 1221: @end example
 1222: 
 1223: Since the guess is optimistic, there will be no spurious error messages
 1224: about undefined locals.
 1225: 
 1226: If the @code{BEGIN} is not reachable from above (e.g., after
 1227: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
 1228: optimistic guess, as the locals visible after the @code{BEGIN} may be
 1229: defined later. Therefore, the compiler assumes that no locals are
 1230: visible after the @code{BEGIN}. However, the user can use
 1231: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
 1232: visible at the BEGIN as at the point where the top control-flow stack
 1233: item was created.
 1234: 
 1235: doc-assume-live
 1236: 
 1237: E.g.,
 1238: @example
 1239: @{ x @}
 1240: AHEAD
 1241: ASSUME-LIVE
 1242: BEGIN
 1243:   x
 1244: [ 1 CS-ROLL ] THEN
 1245:   ...
 1246: UNTIL
 1247: @end example
 1248: 
 1249: Other cases where the locals are defined before the @code{BEGIN} can be
 1250: handled by inserting an appropriate @code{CS-ROLL} before the
 1251: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
 1252: behind the @code{ASSUME-LIVE}).
 1253: 
 1254: Cases where locals are defined after the @code{BEGIN} (but should be
 1255: visible immediately after the @code{BEGIN}) can only be handled by
 1256: rearranging the loop. E.g., the ``most insidious'' example above can be
 1257: arranged into:
 1258: @example
 1259: BEGIN
 1260:   @{ x @}
 1261:   ... 0=
 1262: WHILE
 1263:   x
 1264: REPEAT
 1265: @end example
 1266: 
 1267: @node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
 1268: @subsubsection How long do locals live?
 1269: 
 1270: The right answer for the lifetime question would be: A local lives at
 1271: least as long as it can be accessed. For a value-flavoured local this
 1272: means: until the end of its visibility. However, a variable-flavoured
 1273: local could be accessed through its address far beyond its visibility
 1274: scope. Ultimately, this would mean that such locals would have to be
 1275: garbage collected. Since this entails un-Forth-like implementation
 1276: complexities, I adopted the same cowardly solution as some other
 1277: languages (e.g., C): The local lives only as long as it is visible;
 1278: afterwards its address is invalid (and programs that access it
 1279: afterwards are erroneous).
 1280: 
 1281: @node Programming Style, Implementation, How long do locals live?, Gforth locals
 1282: @subsubsection Programming Style
 1283: 
 1284: The freedom to define locals anywhere has the potential to change
 1285: programming styles dramatically. In particular, the need to use the
 1286: return stack for intermediate storage vanishes. Moreover, all stack
 1287: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
 1288: determined arguments) can be eliminated: If the stack items are in the
 1289: wrong order, just write a locals definition for all of them; then
 1290: write the items in the order you want.
 1291: 
 1292: This seems a little far-fetched and eliminating stack manipulations is
 1293: unlikely to become a conscious programming objective. Still, the number
 1294: of stack manipulations will be reduced dramatically if local variables
 1295: are used liberally (e.g., compare @code{max} in @ref{Gforth locals} with
 1296: a traditional implementation of @code{max}).
 1297: 
 1298: This shows one potential benefit of locals: making Forth programs more
 1299: readable. Of course, this benefit will only be realized if the
 1300: programmers continue to honour the principle of factoring instead of
 1301: using the added latitude to make the words longer.
 1302: 
 1303: Using @code{TO} can and should be avoided.  Without @code{TO},
 1304: every value-flavoured local has only a single assignment and many
 1305: advantages of functional languages apply to Forth. I.e., programs are
 1306: easier to analyse, to optimize and to read: It is clear from the
 1307: definition what the local stands for, it does not turn into something
 1308: different later.
 1309: 
 1310: E.g., a definition using @code{TO} might look like this:
 1311: @example
 1312: : strcmp @{ addr1 u1 addr2 u2 -- n @}
 1313:  u1 u2 min 0
 1314:  ?do
 1315:    addr1 c@ addr2 c@ - ?dup
 1316:    if
 1317:      unloop exit
 1318:    then
 1319:    addr1 char+ TO addr1
 1320:    addr2 char+ TO addr2
 1321:  loop
 1322:  u1 u2 - ;
 1323: @end example
 1324: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
 1325: every loop iteration. @code{strcmp} is a typical example of the
 1326: readability problems of using @code{TO}. When you start reading
 1327: @code{strcmp}, you think that @code{addr1} refers to the start of the
 1328: string. Only near the end of the loop you realize that it is something
 1329: else.
 1330: 
 1331: This can be avoided by defining two locals at the start of the loop that
 1332: are initialized with the right value for the current iteration.
 1333: @example
 1334: : strcmp @{ addr1 u1 addr2 u2 -- n @}
 1335:  addr1 addr2
 1336:  u1 u2 min 0 
 1337:  ?do @{ s1 s2 @}
 1338:    s1 c@ s2 c@ - ?dup 
 1339:    if
 1340:      unloop exit
 1341:    then
 1342:    s1 char+ s2 char+
 1343:  loop
 1344:  2drop
 1345:  u1 u2 - ;
 1346: @end example
 1347: Here it is clear from the start that @code{s1} has a different value
 1348: in every loop iteration.
 1349: 
 1350: @node Implementation,  , Programming Style, Gforth locals
 1351: @subsubsection Implementation
 1352: 
 1353: Gforth uses an extra locals stack. The most compelling reason for
 1354: this is that the return stack is not float-aligned; using an extra stack
 1355: also eliminates the problems and restrictions of using the return stack
 1356: as locals stack. Like the other stacks, the locals stack grows toward
 1357: lower addresses. A few primitives allow an efficient implementation:
 1358: 
 1359: doc-@local#
 1360: doc-f@local#
 1361: doc-laddr#
 1362: doc-lp+!#
 1363: doc-lp!
 1364: doc->l
 1365: doc-f>l
 1366: 
 1367: In addition to these primitives, some specializations of these
 1368: primitives for commonly occurring inline arguments are provided for
 1369: efficiency reasons, e.g., @code{@@local0} as specialization of
 1370: @code{@@local#} for the inline argument 0. The following compiling words
 1371: compile the right specialized version, or the general version, as
 1372: appropriate:
 1373: 
 1374: doc-compile-@local
 1375: doc-compile-f@local
 1376: doc-compile-lp+!
 1377: 
 1378: Combinations of conditional branches and @code{lp+!#} like
 1379: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
 1380: is taken) are provided for efficiency and correctness in loops.
 1381: 
 1382: A special area in the dictionary space is reserved for keeping the
 1383: local variable names. @code{@{} switches the dictionary pointer to this
 1384: area and @code{@}} switches it back and generates the locals
 1385: initializing code. @code{W:} etc.@ are normal defining words. This
 1386: special area is cleared at the start of every colon definition.
 1387: 
 1388: A special feature of Gforth's dictionary is used to implement the
 1389: definition of locals without type specifiers: every wordlist (aka
 1390: vocabulary) has its own methods for searching
 1391: etc. (@pxref{Wordlists}). For the present purpose we defined a wordlist
 1392: with a special search method: When it is searched for a word, it
 1393: actually creates that word using @code{W:}. @code{@{} changes the search
 1394: order to first search the wordlist containing @code{@}}, @code{W:} etc.,
 1395: and then the wordlist for defining locals without type specifiers.
 1396: 
 1397: The lifetime rules support a stack discipline within a colon
 1398: definition: The lifetime of a local is either nested with other locals
 1399: lifetimes or it does not overlap them.
 1400: 
 1401: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
 1402: pointer manipulation is generated. Between control structure words
 1403: locals definitions can push locals onto the locals stack. @code{AGAIN}
 1404: is the simplest of the other three control flow words. It has to
 1405: restore the locals stack depth of the corresponding @code{BEGIN}
 1406: before branching. The code looks like this:
 1407: @format
 1408: @code{lp+!#} current-locals-size @minus{} dest-locals-size
 1409: @code{branch} <begin>
 1410: @end format
 1411: 
 1412: @code{UNTIL} is a little more complicated: If it branches back, it
 1413: must adjust the stack just like @code{AGAIN}. But if it falls through,
 1414: the locals stack must not be changed. The compiler generates the
 1415: following code:
 1416: @format
 1417: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
 1418: @end format
 1419: The locals stack pointer is only adjusted if the branch is taken.
 1420: 
 1421: @code{THEN} can produce somewhat inefficient code:
 1422: @format
 1423: @code{lp+!#} current-locals-size @minus{} orig-locals-size
 1424: <orig target>:
 1425: @code{lp+!#} orig-locals-size @minus{} new-locals-size
 1426: @end format
 1427: The second @code{lp+!#} adjusts the locals stack pointer from the
 1428: level at the @var{orig} point to the level after the @code{THEN}. The
 1429: first @code{lp+!#} adjusts the locals stack pointer from the current
 1430: level to the level at the orig point, so the complete effect is an
 1431: adjustment from the current level to the right level after the
 1432: @code{THEN}.
 1433: 
 1434: In a conventional Forth implementation a dest control-flow stack entry
 1435: is just the target address and an orig entry is just the address to be
 1436: patched. Our locals implementation adds a wordlist to every orig or dest
 1437: item. It is the list of locals visible (or assumed visible) at the point
 1438: described by the entry. Our implementation also adds a tag to identify
 1439: the kind of entry, in particular to differentiate between live and dead
 1440: (reachable and unreachable) orig entries.
 1441: 
 1442: A few unusual operations have to be performed on locals wordlists:
 1443: 
 1444: doc-common-list
 1445: doc-sub-list?
 1446: doc-list-size
 1447: 
 1448: Several features of our locals wordlist implementation make these
 1449: operations easy to implement: The locals wordlists are organised as
 1450: linked lists; the tails of these lists are shared, if the lists
 1451: contain some of the same locals; and the address of a name is greater
 1452: than the address of the names behind it in the list.
 1453: 
 1454: Another important implementation detail is the variable
 1455: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
 1456: determine if they can be reached directly or only through the branch
 1457: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
 1458: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
 1459: definition, by @code{BEGIN} and usually by @code{THEN}.
 1460: 
 1461: Counted loops are similar to other loops in most respects, but
 1462: @code{LEAVE} requires special attention: It performs basically the same
 1463: service as @code{AHEAD}, but it does not create a control-flow stack
 1464: entry. Therefore the information has to be stored elsewhere;
 1465: traditionally, the information was stored in the target fields of the
 1466: branches created by the @code{LEAVE}s, by organizing these fields into a
 1467: linked list. Unfortunately, this clever trick does not provide enough
 1468: space for storing our extended control flow information. Therefore, we
 1469: introduce another stack, the leave stack. It contains the control-flow
 1470: stack entries for all unresolved @code{LEAVE}s.
 1471: 
 1472: Local names are kept until the end of the colon definition, even if
 1473: they are no longer visible in any control-flow path. In a few cases
 1474: this may lead to increased space needs for the locals name area, but
 1475: usually less than reclaiming this space would cost in code size.
 1476: 
 1477: 
 1478: @node ANS Forth locals,  , Gforth locals, Locals
 1479: @subsection ANS Forth locals
 1480: 
 1481: The ANS Forth locals wordset does not define a syntax for locals, but
 1482: words that make it possible to define various syntaxes. One of the
 1483: possible syntaxes is a subset of the syntax we used in the Gforth locals
 1484: wordset, i.e.:
 1485: 
 1486: @example
 1487: @{ local1 local2 ... -- comment @}
 1488: @end example
 1489: or
 1490: @example
 1491: @{ local1 local2 ... @}
 1492: @end example
 1493: 
 1494: The order of the locals corresponds to the order in a stack comment. The
 1495: restrictions are:
 1496: 
 1497: @itemize @bullet
 1498: @item
 1499: Locals can only be cell-sized values (no type specifiers are allowed).
 1500: @item
 1501: Locals can be defined only outside control structures.
 1502: @item
 1503: Locals can interfere with explicit usage of the return stack. For the
 1504: exact (and long) rules, see the standard. If you don't use return stack
 1505: accessing words in a definition using locals, you will be all right. The
 1506: purpose of this rule is to make locals implementation on the return
 1507: stack easier.
 1508: @item
 1509: The whole definition must be in one line.
 1510: @end itemize
 1511: 
 1512: Locals defined in this way behave like @code{VALUE}s
 1513: (@xref{Values}). I.e., they are initialized from the stack. Using their
 1514: name produces their value. Their value can be changed using @code{TO}.
 1515: 
 1516: Since this syntax is supported by Gforth directly, you need not do
 1517: anything to use it. If you want to port a program using this syntax to
 1518: another ANS Forth system, use @file{anslocal.fs} to implement the syntax
 1519: on the other system.
 1520: 
 1521: Note that a syntax shown in the standard, section A.13 looks
 1522: similar, but is quite different in having the order of locals
 1523: reversed. Beware!
 1524: 
 1525: The ANS Forth locals wordset itself consists of the following word
 1526: 
 1527: doc-(local)
 1528: 
 1529: The ANS Forth locals extension wordset defines a syntax, but it is so
 1530: awful that we strongly recommend not to use it. We have implemented this
 1531: syntax to make porting to Gforth easy, but do not document it here. The
 1532: problem with this syntax is that the locals are defined in an order
 1533: reversed with respect to the standard stack comment notation, making
 1534: programs harder to read, and easier to misread and miswrite. The only
 1535: merit of this syntax is that it is easy to implement using the ANS Forth
 1536: locals wordset.
 1537: 
 1538: @node Defining Words, Wordlists, Locals, Words
 1539: @section Defining Words
 1540: 
 1541: @menu
 1542: * Values::                      
 1543: @end menu
 1544: 
 1545: @node Values,  , Defining Words, Defining Words
 1546: @subsection Values
 1547: 
 1548: @node Wordlists, Files, Defining Words, Words
 1549: @section Wordlists
 1550: 
 1551: @node Files, Blocks, Wordlists, Words
 1552: @section Files
 1553: 
 1554: @node Blocks, Other I/O, Files, Words
 1555: @section Blocks
 1556: 
 1557: @node Other I/O, Programming Tools, Blocks, Words
 1558: @section Other I/O
 1559: 
 1560: @node Programming Tools, Assembler and Code words, Other I/O, Words
 1561: @section Programming Tools
 1562: 
 1563: @menu
 1564: * Debugging::                   Simple and quick.
 1565: * Assertions::                  Making your programs self-checking.
 1566: @end menu
 1567: 
 1568: @node Debugging, Assertions, Programming Tools, Programming Tools
 1569: @subsection Debugging
 1570: 
 1571: The simple debugging aids provided in @file{debugging.fs}
 1572: are meant to support a different style of debugging than the
 1573: tracing/stepping debuggers used in languages with long turn-around
 1574: times.
 1575: 
 1576: A much better (faster) way in fast-compilig languages is to add
 1577: printing code at well-selected places, let the program run, look at
 1578: the output, see where things went wrong, add more printing code, etc.,
 1579: until the bug is found.
 1580: 
 1581: The word @code{~~} is easy to insert. It just prints debugging
 1582: information (by default the source location and the stack contents). It
 1583: is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
 1584: query-replace them with nothing). The deferred words
 1585: @code{printdebugdata} and @code{printdebugline} control the output of
 1586: @code{~~}. The default source location output format works well with
 1587: Emacs' compilation mode, so you can step through the program at the
 1588: source level using @kbd{C-x `} (the advantage over a stepping debugger
 1589: is that you can step in any direction and you know where the crash has
 1590: happened or where the strange data has occurred).
 1591: 
 1592: Note that the default actions clobber the contents of the pictured
 1593: numeric output string, so you should not use @code{~~}, e.g., between
 1594: @code{<#} and @code{#>}.
 1595: 
 1596: doc-~~
 1597: doc-printdebugdata
 1598: doc-printdebugline
 1599: 
 1600: @node Assertions,  , Debugging, Programming Tools
 1601: @subsection Assertions
 1602: 
 1603: It is a good idea to make your programs self-checking, in particular, if
 1604: you use an assumption (e.g., that a certain field of a data structure is
 1605: never zero) that may become wrong during maintenance. Gforth supports
 1606: assertions for this purpose. They are used like this:
 1607: 
 1608: @example
 1609: assert( @var{flag} )
 1610: @end example
 1611: 
 1612: The code between @code{assert(} and @code{)} should compute a flag, that
 1613: should be true if everything is alright and false otherwise. It should
 1614: not change anything else on the stack. The overall stack effect of the
 1615: assertion is @code{( -- )}. E.g.
 1616: 
 1617: @example
 1618: assert( 1 1 + 2 = ) \ what we learn in school
 1619: assert( dup 0<> ) \ assert that the top of stack is not zero
 1620: assert( false ) \ this code should not be reached
 1621: @end example
 1622: 
 1623: The need for assertions is different at different times. During
 1624: debugging, we want more checking, in production we sometimes care more
 1625: for speed. Therefore, assertions can be turned off, i.e., the assertion
 1626: becomes a comment. Depending on the importance of an assertion and the
 1627: time it takes to check it, you may want to turn off some assertions and
 1628: keep others turned on. Gforth provides several levels of assertions for
 1629: this purpose:
 1630: 
 1631: doc-assert0(
 1632: doc-assert1(
 1633: doc-assert2(
 1634: doc-assert3(
 1635: doc-assert(
 1636: doc-)
 1637: 
 1638: @code{Assert(} is the same as @code{assert1(}. The variable
 1639: @code{assert-level} specifies the highest assertions that are turned
 1640: on. I.e., at the default @code{assert-level} of one, @code{assert0(} and
 1641: @code{assert1(} assertions perform checking, while @code{assert2(} and
 1642: @code{assert3(} assertions are treated as comments.
 1643: 
 1644: Note that the @code{assert-level} is evaluated at compile-time, not at
 1645: run-time. I.e., you cannot turn assertions on or off at run-time, you
 1646: have to set the @code{assert-level} appropriately before compiling a
 1647: piece of code. You can compile several pieces of code at several
 1648: @code{assert-level}s (e.g., a trusted library at level 1 and newly
 1649: written code at level 3).
 1650: 
 1651: doc-assert-level
 1652: 
 1653: If an assertion fails, a message compatible with Emacs' compilation mode
 1654: is produced and the execution is aborted (currently with @code{ABORT"}.
 1655: If there is interest, we will introduce a special throw code. But if you
 1656: intend to @code{catch} a specific condition, using @code{throw} is
 1657: probably more appropriate than an assertion).
 1658: 
 1659: @node Assembler and Code words, Threading Words, Programming Tools, Words
 1660: @section Assembler and Code words
 1661: 
 1662: Gforth provides some words for defining primitives (words written in
 1663: machine code), and for defining the the machine-code equivalent of
 1664: @code{DOES>}-based defining words. However, the machine-independent
 1665: nature of Gforth poses a few problems: First of all. Gforth runs on
 1666: several architectures, so it can provide no standard assembler. What's
 1667: worse is that the register allocation not only depends on the processor,
 1668: but also on the gcc version and options used.
 1669: 
 1670: The words Gforth offers encapsulate some system dependences (e.g., the
 1671: header structure), so a system-independent assembler may be used in
 1672: Gforth. If you do not have an assembler, you can compile machine code
 1673: directly with @code{,} and @code{c,}.
 1674: 
 1675: doc-assembler
 1676: doc-code
 1677: doc-end-code
 1678: doc-;code
 1679: doc-flush-icache
 1680: 
 1681: If @code{flush-icache} does not work correctly, @code{code} words
 1682: etc. will not work (reliably), either.
 1683: 
 1684: These words are rarely used. Therefore they reside in @code{code.fs},
 1685: which is usually not loaded (except @code{flush-icache}, which is always
 1686: present). You can load it with @code{require code.fs}.
 1687: 
 1688: Another option for implementing normal and defining words efficiently
 1689: is: adding the wanted functionality to the source of Gforth. For normal
 1690: words you just have to edit @file{primitives}, defining words (for fast
 1691: defined words) probably require changes in @file{engine.c},
 1692: @file{kernal.fs}, @file{prims2x.fs}, and possibly @file{cross.fs}.
 1693: 
 1694: 
 1695: @node Threading Words,  , Assembler and Code words, Words
 1696: @section Threading Words
 1697: 
 1698: These words provide access to code addresses and other threading stuff
 1699: in Gforth (and, possibly, other interpretive Forths). It more or less
 1700: abstracts away the differences between direct and indirect threading
 1701: (and, for direct threading, the machine dependences). However, at
 1702: present this wordset is still inclomplete. It is also pretty low-level;
 1703: some day it will hopefully be made unnecessary by an internals words set
 1704: that abstracts implementation details away completely.
 1705: 
 1706: doc->code-address
 1707: doc->does-code
 1708: doc-code-address!
 1709: doc-does-code!
 1710: doc-does-handler!
 1711: doc-/does-handler
 1712: 
 1713: The code addresses produced by various defining words are produced by
 1714: the following words:
 1715: 
 1716: doc-docol:
 1717: doc-docon:
 1718: doc-dovar:
 1719: doc-douser:
 1720: doc-dodefer:
 1721: doc-dofield:
 1722: 
 1723: Currently there is no installation-independent way for recogizing words
 1724: defined by a @code{CREATE}...@code{DOES>} word; however, once you know
 1725: that a word is defined by a @code{CREATE}...@code{DOES>} word, you can
 1726: use @code{>DOES-CODE}.
 1727: 
 1728: @node ANS conformance, Model, Words, Top
 1729: @chapter ANS conformance
 1730: 
 1731: To the best of our knowledge, Gforth is an
 1732: 
 1733: ANS Forth System
 1734: @itemize
 1735: @item providing the Core Extensions word set
 1736: @item providing the Block word set
 1737: @item providing the Block Extensions word set
 1738: @item providing the Double-Number word set
 1739: @item providing the Double-Number Extensions word set
 1740: @item providing the Exception word set
 1741: @item providing the Exception Extensions word set
 1742: @item providing the Facility word set
 1743: @item providing @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
 1744: @item providing the File Access word set
 1745: @item providing the File Access Extensions word set
 1746: @item providing the Floating-Point word set
 1747: @item providing the Floating-Point Extensions word set
 1748: @item providing the Locals word set
 1749: @item providing the Locals Extensions word set
 1750: @item providing the Memory-Allocation word set
 1751: @item providing the Memory-Allocation Extensions word set (that one's easy)
 1752: @item providing the Programming-Tools word set
 1753: @item providing @code{;code}, @code{AHEAD}, @code{ASSEMBLER}, @code{BYE}, @code{CODE}, @code{CS-PICK}, @code{CS-ROLL}, @code{STATE}, @code{[ELSE]}, @code{[IF]}, @code{[THEN]} from the Programming-Tools Extensions word set
 1754: @item providing the Search-Order word set
 1755: @item providing the Search-Order Extensions word set
 1756: @item providing the String word set
 1757: @item providing the String Extensions word set (another easy one)
 1758: @end itemize
 1759: 
 1760: In addition, ANS Forth systems are required to document certain
 1761: implementation choices. This chapter tries to meet these
 1762: requirements. In many cases it gives a way to ask the system for the
 1763: information instead of providing the information directly, in
 1764: particular, if the information depends on the processor, the operating
 1765: system or the installation options chosen, or if they are likely to
 1766: change during the maintenance of Gforth.
 1767: 
 1768: @comment The framework for the rest has been taken from pfe.
 1769: 
 1770: @menu
 1771: * The Core Words::              
 1772: * The optional Block word set::  
 1773: * The optional Double Number word set::  
 1774: * The optional Exception word set::  
 1775: * The optional Facility word set::  
 1776: * The optional File-Access word set::  
 1777: * The optional Floating-Point word set::  
 1778: * The optional Locals word set::  
 1779: * The optional Memory-Allocation word set::  
 1780: * The optional Programming-Tools word set::  
 1781: * The optional Search-Order word set::  
 1782: @end menu
 1783: 
 1784: 
 1785: @c =====================================================================
 1786: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
 1787: @comment  node-name,  next,  previous,  up
 1788: @section The Core Words
 1789: @c =====================================================================
 1790: 
 1791: @menu
 1792: * core-idef::                   Implementation Defined Options                   
 1793: * core-ambcond::                Ambiguous Conditions                
 1794: * core-other::                  Other System Documentation                  
 1795: @end menu
 1796: 
 1797: @c ---------------------------------------------------------------------
 1798: @node core-idef, core-ambcond, The Core Words, The Core Words
 1799: @subsection Implementation Defined Options
 1800: @c ---------------------------------------------------------------------
 1801: 
 1802: @table @i
 1803: 
 1804: @item (Cell) aligned addresses:
 1805: processor-dependent. Gforth's alignment words perform natural alignment
 1806: (e.g., an address aligned for a datum of size 8 is divisible by
 1807: 8). Unaligned accesses usually result in a @code{-23 THROW}.
 1808: 
 1809: @item @code{EMIT} and non-graphic characters:
 1810: The character is output using the C library function (actually, macro)
 1811: @code{putchar}.
 1812: 
 1813: @item character editing of @code{ACCEPT} and @code{EXPECT}:
 1814: This is modeled on the GNU readline library (@pxref{Readline
 1815: Interaction, , Command Line Editing, readline, The GNU Readline
 1816: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
 1817: producing a full word completion every time you type it (instead of
 1818: producing the common prefix of all completions).
 1819: 
 1820: @item character set:
 1821: The character set of your computer and display device. Gforth is
 1822: 8-bit-clean (but some other component in your system may make trouble).
 1823: 
 1824: @item Character-aligned address requirements:
 1825: installation-dependent. Currently a character is represented by a C
 1826: @code{unsigned char}; in the future we might switch to @code{wchar_t}
 1827: (Comments on that requested).
 1828: 
 1829: @item character-set extensions and matching of names:
 1830: Any character except the ASCII NUL charcter can be used in a
 1831: name. Matching is case-insensitive. The matching is performed using the
 1832: C function @code{strncasecmp}, whose function is probably influenced by
 1833: the locale. E.g., the @code{C} locale does not know about accents and
 1834: umlauts, so they are matched case-sensitively in that locale. For
 1835: portability reasons it is best to write programs such that they work in
 1836: the @code{C} locale. Then one can use libraries written by a Polish
 1837: programmer (who might use words containing ISO Latin-2 encoded
 1838: characters) and by a French programmer (ISO Latin-1) in the same program
 1839: (of course, @code{WORDS} will produce funny results for some of the
 1840: words (which ones, depends on the font you are using)). Also, the locale
 1841: you prefer may not be available in other operating systems. Hopefully,
 1842: Unicode will solve these problems one day.
 1843: 
 1844: @item conditions under which control characters match a space delimiter:
 1845: If @code{WORD} is called with the space character as a delimiter, all
 1846: white-space characters (as identified by the C macro @code{isspace()})
 1847: are delimiters. @code{PARSE}, on the other hand, treats space like other
 1848: delimiters. @code{PARSE-WORD} treats space like @code{WORD}, but behaves
 1849: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
 1850: interpreter (aka text interpreter) by default, treats all white-space
 1851: characters as delimiters.
 1852: 
 1853: @item format of the control flow stack:
 1854: The data stack is used as control flow stack. The size of a control flow
 1855: stack item in cells is given by the constant @code{cs-item-size}. At the
 1856: time of this writing, an item consists of a (pointer to a) locals list
 1857: (third), an address in the code (second), and a tag for identifying the
 1858: item (TOS). The following tags are used: @code{defstart},
 1859: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
 1860: @code{scopestart}.
 1861: 
 1862: @item conversion of digits > 35
 1863: The characters @code{[\]^_'} are the digits with the decimal value
 1864: 36@minus{}41. There is no way to input many of the larger digits.
 1865: 
 1866: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
 1867: The cursor is moved to the end of the entered string. If the input is
 1868: terminated using the @kbd{Return} key, a space is typed.
 1869: 
 1870: @item exception abort sequence of @code{ABORT"}:
 1871: The error string is stored into the variable @code{"error} and a
 1872: @code{-2 throw} is performed.
 1873: 
 1874: @item input line terminator:
 1875: For interactive input, @kbd{C-m} and @kbd{C-j} terminate lines. One of
 1876: these characters is typically produced when you type the @kbd{Enter} or
 1877: @kbd{Return} key.
 1878: 
 1879: @item maximum size of a counted string:
 1880: @code{s" /counted-string" environment? drop .}. Currently 255 characters
 1881: on all ports, but this may change.
 1882: 
 1883: @item maximum size of a parsed string:
 1884: Given by the constant @code{/line}. Currently 255 characters.
 1885: 
 1886: @item maximum size of a definition name, in characters:
 1887: 31
 1888: 
 1889: @item maximum string length for @code{ENVIRONMENT?}, in characters:
 1890: 31
 1891: 
 1892: @item method of selecting the user input device:
 1893: The user input device is the standard input. There is currently no way to
 1894: change it from within Gforth. However, the input can typically be
 1895: redirected in the command line that starts Gforth.
 1896: 
 1897: @item method of selecting the user output device:
 1898: The user output device is the standard output. It cannot be redirected
 1899: from within Gforth, but typically from the command line that starts
 1900: Gforth. Gforth uses buffered output, so output on a terminal does not
 1901: become visible before the next newline or buffer overflow. Output on
 1902: non-terminals is invisible until the buffer overflows.
 1903: 
 1904: @item methods of dictionary compilation:
 1905: What are we expected to document here?
 1906: 
 1907: @item number of bits in one address unit:
 1908: @code{s" address-units-bits" environment? drop .}. 8 in all current
 1909: ports.
 1910: 
 1911: @item number representation and arithmetic:
 1912: Processor-dependent. Binary two's complement on all current ports.
 1913: 
 1914: @item ranges for integer types:
 1915: Installation-dependent. Make environmental queries for @code{MAX-N},
 1916: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
 1917: unsigned (and positive) types is 0. The lower bound for signed types on
 1918: two's complement and one's complement machines machines can be computed
 1919: by adding 1 to the upper bound.
 1920: 
 1921: @item read-only data space regions:
 1922: The whole Forth data space is writable.
 1923: 
 1924: @item size of buffer at @code{WORD}:
 1925: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
 1926: shared with the pictured numeric output string. If overwriting
 1927: @code{PAD} is acceptable, it is as large as the remaining dictionary
 1928: space, although only as much can be sensibly used as fits in a counted
 1929: string.
 1930: 
 1931: @item size of one cell in address units:
 1932: @code{1 cells .}.
 1933: 
 1934: @item size of one character in address units:
 1935: @code{1 chars .}. 1 on all current ports.
 1936: 
 1937: @item size of the keyboard terminal buffer:
 1938: Varies. You can determine the size at a specific time using @code{lp@
 1939: tib - .}. It is shared with the locals stack and TIBs of files that
 1940: include the current file. You can change the amount of space for TIBs
 1941: and locals stack at Gforth startup with the command line option
 1942: @code{-l}.
 1943: 
 1944: @item size of the pictured numeric output buffer:
 1945: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
 1946: shared with @code{WORD}.
 1947: 
 1948: @item size of the scratch area returned by @code{PAD}:
 1949: The remainder of dictionary space. You can even use the unused part of
 1950: the data stack space. The current size can be computed with @code{sp@
 1951: pad - .}.
 1952: 
 1953: @item system case-sensitivity characteristics:
 1954: Dictionary searches are case insensitive. However, as explained above
 1955: under @i{character-set extensions}, the matching for non-ASCII
 1956: characters is determined by the locale you are using. In the default
 1957: @code{C} locale all non-ASCII characters are matched case-sensitively.
 1958: 
 1959: @item system prompt:
 1960: @code{ ok} in interpret state, @code{ compiled} in compile state.
 1961: 
 1962: @item division rounding:
 1963: installation dependent. @code{s" floored" environment? drop .}. We leave
 1964: the choice to gcc (what to use for @code{/}) and to you (whether to use
 1965: @code{fm/mod}, @code{sm/rem} or simply @code{/}).
 1966: 
 1967: @item values of @code{STATE} when true:
 1968: -1.
 1969: 
 1970: @item values returned after arithmetic overflow:
 1971: On two's complement machines, arithmetic is performed modulo
 1972: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
 1973: arithmetic (with appropriate mapping for signed types). Division by zero
 1974: typically results in a @code{-55 throw} (floatingpoint unidentified
 1975: fault), although a @code{-10 throw} (divide by zero) would be more
 1976: appropriate.
 1977: 
 1978: @item whether the current definition can be found after @t{DOES>}:
 1979: No.
 1980: 
 1981: @end table
 1982: 
 1983: @c ---------------------------------------------------------------------
 1984: @node core-ambcond, core-other, core-idef, The Core Words
 1985: @subsection Ambiguous conditions
 1986: @c ---------------------------------------------------------------------
 1987: 
 1988: @table @i
 1989: 
 1990: @item a name is neither a word nor a number:
 1991: @code{-13 throw} (Undefined word)
 1992: 
 1993: @item a definition name exceeds the maximum length allowed:
 1994: @code{-19 throw} (Word name too long)
 1995: 
 1996: @item addressing a region not inside the various data spaces of the forth system:
 1997: The stacks, code space and name space are accessible. Machine code space is
 1998: typically readable. Accessing other addresses gives results dependent on
 1999: the operating system. On decent systems: @code{-9 throw} (Invalid memory
 2000: address).
 2001: 
 2002: @item argument type incompatible with parameter:
 2003: This is usually not caught. Some words perform checks, e.g., the control
 2004: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
 2005: mismatch).
 2006: 
 2007: @item attempting to obtain the execution token of a word with undefined execution semantics:
 2008: You get an execution token representing the compilation semantics
 2009: instead.
 2010: 
 2011: @item dividing by zero:
 2012: typically results in a @code{-55 throw} (floating point unidentified
 2013: fault), although a @code{-10 throw} (divide by zero) would be more
 2014: appropriate.
 2015: 
 2016: @item insufficient data stack or return stack space:
 2017: Not checked. This typically results in mysterious illegal memory
 2018: accesses, producing @code{-9 throw} (Invalid memory address) or
 2019: @code{-23 throw} (Address alignment exception).
 2020: 
 2021: @item insufficient space for loop control parameters:
 2022: like other return stack overflows.
 2023: 
 2024: @item insufficient space in the dictionary:
 2025: Not checked. Similar results as stack overflows. However, typically the
 2026: error appears at a different place when one inserts or removes code.
 2027: 
 2028: @item interpreting a word with undefined interpretation semantics:
 2029: For some words, we defined interpretation semantics. For the others:
 2030: @code{-14 throw} (Interpreting a compile-only word). Note that this is
 2031: checked only by the outer (aka text) interpreter; if the word is
 2032: @code{execute}d in some other way, it will typically perform it's
 2033: compilation semantics even in interpret state. (We could change @code{'}
 2034: and relatives not to give the xt of such words, but we think that would
 2035: be too restrictive).
 2036: 
 2037: @item modifying the contents of the input buffer or a string literal:
 2038: These are located in writable memory and can be modified.
 2039: 
 2040: @item overflow of the pictured numeric output string:
 2041: Not checked.
 2042: 
 2043: @item parsed string overflow:
 2044: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
 2045: 
 2046: @item producing a result out of range:
 2047: On two's complement machines, arithmetic is performed modulo
 2048: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
 2049: arithmetic (with appropriate mapping for signed types). Division by zero
 2050: typically results in a @code{-55 throw} (floatingpoint unidentified
 2051: fault), although a @code{-10 throw} (divide by zero) would be more
 2052: appropriate. @code{convert} and @code{>number} currently overflow
 2053: silently.
 2054: 
 2055: @item reading from an empty data or return stack:
 2056: The data stack is checked by the outer (aka text) interpreter after
 2057: every word executed. If it has underflowed, a @code{-4 throw} (Stack
 2058: underflow) is performed. Apart from that, the stacks are not checked and
 2059: underflows can result in similar behaviour as overflows (of adjacent
 2060: stacks).
 2061: 
 2062: @item unexepected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
 2063: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
 2064: use zero-length string as a name). Words like @code{'} probably will not
 2065: find what they search. Note that it is possible to create zero-length
 2066: names with @code{nextname} (should it not?).
 2067: 
 2068: @item @code{>IN} greater than input buffer:
 2069: The next invocation of a parsing word returns a string wih length 0.
 2070: 
 2071: @item @code{RECURSE} appears after @code{DOES>}:
 2072: Compiles a recursive call to the defining word not to the defined word.
 2073: 
 2074: @item argument input source different than current input source for @code{RESTORE-INPUT}:
 2075: !!???If the argument input source is a valid input source then it gets
 2076: restored. Otherwise causes @code{-12 THROW} which unless caught issues
 2077: the message "argument type mismatch" and aborts.
 2078: 
 2079: @item data space containing definitions gets de-allocated:
 2080: Deallocation with @code{allot} is not checked. This typically resuls in
 2081: memory access faults or execution of illegal instructions.
 2082: 
 2083: @item data space read/write with incorrect alignment:
 2084: Processor-dependent. Typically results in a @code{-23 throw} (Address
 2085: alignment exception). Under Linux on a 486 or later processor with
 2086: alignment turned on, incorrect alignment results in a @code{-9 throw}
 2087: (Invalid memory address). There are reportedly some processors with
 2088: alignment restrictions that do not report them.
 2089: 
 2090: @item data space pointer not properly aligned, @code{,}, @code{C,}:
 2091: Like other alignment errors.
 2092: 
 2093: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
 2094: Not checked. May cause an illegal memory access.
 2095: 
 2096: @item loop control parameters not available:
 2097: Not checked. The counted loop words simply assume that the top of return
 2098: stack items are loop control parameters and behave accordingly.
 2099: 
 2100: @item most recent definition does not have a name (@code{IMMEDIATE}):
 2101: @code{abort" last word was headerless"}.
 2102: 
 2103: @item name not defined by @code{VALUE} used by @code{TO}:
 2104: @code{-32 throw} (Invalid name argument)
 2105: 
 2106: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
 2107: @code{-13 throw} (Undefined word)
 2108: 
 2109: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
 2110: Gforth behaves as if they were of the same type. I.e., you can predict
 2111: the behaviour by interpreting all parameters as, e.g., signed.
 2112: 
 2113: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
 2114: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} is equivalent to
 2115: @code{TO}.
 2116: 
 2117: @item String longer than a counted string returned by @code{WORD}:
 2118: Not checked. The string will be ok, but the count will, of course,
 2119: contain only the least significant bits of the length.
 2120: 
 2121: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
 2122: Processor-dependent. Typical behaviours are returning 0 and using only
 2123: the low bits of the shift count.
 2124: 
 2125: @item word not defined via @code{CREATE}:
 2126: @code{>BODY} produces the PFA of the word no matter how it was defined.
 2127: 
 2128: @code{DOES>} changes the execution semantics of the last defined word no
 2129: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
 2130: @code{CREATE , DOES>}.
 2131: 
 2132: @item words improperly used outside @code{<#} and @code{#>}:
 2133: Not checked. As usual, you can expect memory faults.
 2134: 
 2135: @end table
 2136: 
 2137: 
 2138: @c ---------------------------------------------------------------------
 2139: @node core-other,  , core-ambcond, The Core Words
 2140: @subsection Other system documentation
 2141: @c ---------------------------------------------------------------------
 2142: 
 2143: @table @i
 2144: 
 2145: @item nonstandard words using @code{PAD}:
 2146: None.
 2147: 
 2148: @item operator's terminal facilities available:
 2149: !!??
 2150: 
 2151: @item program data space available:
 2152: @code{sp@ here - .} gives the space remaining for dictionary and data
 2153: stack together.
 2154: 
 2155: @item return stack space available:
 2156: !!??
 2157: 
 2158: @item stack space available:
 2159: @code{sp@ here - .} gives the space remaining for dictionary and data
 2160: stack together.
 2161: 
 2162: @item system dictionary space required, in address units:
 2163: Type @code{here forthstart - .} after startup. At the time of this
 2164: writing, this gives 70108 (bytes) on a 32-bit system.
 2165: @end table
 2166: 
 2167: 
 2168: @c =====================================================================
 2169: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
 2170: @section The optional Block word set
 2171: @c =====================================================================
 2172: 
 2173: @menu
 2174: * block-idef::                  Implementation Defined Options                  
 2175: * block-ambcond::               Ambiguous Conditions               
 2176: * block-other::                 Other System Documentation                 
 2177: @end menu
 2178: 
 2179: 
 2180: @c ---------------------------------------------------------------------
 2181: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
 2182: @subsection Implementation Defined Options
 2183: @c ---------------------------------------------------------------------
 2184: 
 2185: @table @i
 2186: 
 2187: @item the format for display by @code{LIST}:
 2188: First the screen number is displayed, then 16 lines of 64 characters,
 2189: each line preceded by the line number.
 2190: 
 2191: @item the length of a line affected by @code{\}:
 2192: 64 characters.
 2193: @end table
 2194: 
 2195: 
 2196: @c ---------------------------------------------------------------------
 2197: @node block-ambcond, block-other, block-idef, The optional Block word set
 2198: @subsection Ambiguous conditions
 2199: @c ---------------------------------------------------------------------
 2200: 
 2201: @table @i
 2202: 
 2203: @item correct block read was not possible:
 2204: Typically results in a @code{throw} of some OS-derived value (between
 2205: -512 and -2048). If the blocks file was just not long enough, blanks are
 2206: supplied for the missing portion.
 2207: 
 2208: @item I/O exception in block transfer:
 2209: Typically results in a @code{throw} of some OS-derived value (between
 2210: -512 and -2048).
 2211: 
 2212: @item invalid block number:
 2213: @code{-35 throw} (Invalid block number)
 2214: 
 2215: @item a program directly alters the contents of @code{BLK}:
 2216: The input stream is switched to that other block, at the same
 2217: position. If the storing to @code{BLK} happens when interpreting
 2218: non-block input, the system will get quite confused when the block ends.
 2219: 
 2220: @item no current block buffer for @code{UPDATE}:
 2221: @code{UPDATE} has no effect.
 2222: 
 2223: @end table
 2224: 
 2225: 
 2226: @c ---------------------------------------------------------------------
 2227: @node block-other,  , block-ambcond, The optional Block word set
 2228: @subsection Other system documentation
 2229: @c ---------------------------------------------------------------------
 2230: 
 2231: @table @i
 2232: 
 2233: @item any restrictions a multiprogramming system places on the use of buffer addresses:
 2234: No restrictions (yet).
 2235: 
 2236: @item the number of blocks available for source and data:
 2237: depends on your disk space.
 2238: 
 2239: @end table
 2240: 
 2241: 
 2242: @c =====================================================================
 2243: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
 2244: @section The optional Double Number word set
 2245: @c =====================================================================
 2246: 
 2247: @menu
 2248: * double-ambcond::              Ambiguous Conditions              
 2249: @end menu
 2250: 
 2251: 
 2252: @c ---------------------------------------------------------------------
 2253: @node double-ambcond,  , The optional Double Number word set, The optional Double Number word set
 2254: @subsection Ambiguous conditions
 2255: @c ---------------------------------------------------------------------
 2256: 
 2257: @table @i
 2258: 
 2259: @item @var{d} outside of range of @var{n} in @code{D>S}:
 2260: The least significant cell of @var{d} is produced.
 2261: 
 2262: @end table
 2263: 
 2264: 
 2265: @c =====================================================================
 2266: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
 2267: @section The optional Exception word set
 2268: @c =====================================================================
 2269: 
 2270: @menu
 2271: * exception-idef::              Implementation Defined Options              
 2272: @end menu
 2273: 
 2274: 
 2275: @c ---------------------------------------------------------------------
 2276: @node exception-idef,  , The optional Exception word set, The optional Exception word set
 2277: @subsection Implementation Defined Options
 2278: @c ---------------------------------------------------------------------
 2279: 
 2280: @table @i
 2281: @item @code{THROW}-codes used in the system:
 2282: The codes -256@minus{}-511 are used for reporting signals (see
 2283: @file{errore.fs}). The codes -512@minus{}-2047 are used for OS errors
 2284: (for file and memory allocation operations). The mapping from OS error
 2285: numbers to throw code is -512@minus{}@var{errno}. One side effect of
 2286: this mapping is that undefined OS errors produce a message with a
 2287: strange number; e.g., @code{-1000 THROW} results in @code{Unknown error
 2288: 488} on my system.
 2289: @end table
 2290: 
 2291: @c =====================================================================
 2292: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
 2293: @section The optional Facility word set
 2294: @c =====================================================================
 2295: 
 2296: @menu
 2297: * facility-idef::               Implementation Defined Options               
 2298: * facility-ambcond::            Ambiguous Conditions            
 2299: @end menu
 2300: 
 2301: 
 2302: @c ---------------------------------------------------------------------
 2303: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
 2304: @subsection Implementation Defined Options
 2305: @c ---------------------------------------------------------------------
 2306: 
 2307: @table @i
 2308: 
 2309: @item encoding of keyboard events (@code{EKEY}):
 2310: Not yet implemeted.
 2311: 
 2312: @item duration of a system clock tick
 2313: System dependent. With respect to @code{MS}, the time is specified in
 2314: microseconds. How well the OS and the hardware implement this, is
 2315: another question.
 2316: 
 2317: @item repeatability to be expected from the execution of @code{MS}:
 2318: System dependent. On Unix, a lot depends on load. If the system is
 2319: lightly loaded, and the delay is short enough that Gforth does not get
 2320: swapped out, the performance should be acceptable. Under MS-DOS and
 2321: other single-tasking systems, it should be good.
 2322: 
 2323: @end table
 2324: 
 2325: 
 2326: @c ---------------------------------------------------------------------
 2327: @node facility-ambcond,  , facility-idef, The optional Facility word set
 2328: @subsection Ambiguous conditions
 2329: @c ---------------------------------------------------------------------
 2330: 
 2331: @table @i
 2332: 
 2333: @item @code{AT-XY} can't be performed on user output device:
 2334: Largely terminal dependant. No range checks are done on the arguments.
 2335: No errors are reported. You may see some garbage appearing, you may see
 2336: simply nothing happen.
 2337: 
 2338: @end table
 2339: 
 2340: 
 2341: @c =====================================================================
 2342: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
 2343: @section The optional File-Access word set
 2344: @c =====================================================================
 2345: 
 2346: @menu
 2347: * file-idef::                   Implementation Defined Options                   
 2348: * file-ambcond::                Ambiguous Conditions                
 2349: @end menu
 2350: 
 2351: 
 2352: @c ---------------------------------------------------------------------
 2353: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
 2354: @subsection Implementation Defined Options
 2355: @c ---------------------------------------------------------------------
 2356: 
 2357: @table @i
 2358: 
 2359: @item File access methods used:
 2360: @code{R/O}, @code{R/W} and @code{BIN} work as you would
 2361: expect. @code{W/O} translates into the C file opening mode @code{w} (or
 2362: @code{wb}): The file is cleared, if it exists, and created, if it does
 2363: not (both with @code{open-file} and @code{create-file}).  Under Unix
 2364: @code{create-file} creates a file with 666 permissions modified by your
 2365: umask.
 2366: 
 2367: @item file exceptions:
 2368: The file words do not raise exceptions (except, perhaps, memory access
 2369: faults when you pass illegal addresses or file-ids).
 2370: 
 2371: @item file line terminator:
 2372: System-dependent. Gforth uses C's newline character as line
 2373: terminator. What the actual character code(s) of this are is
 2374: system-dependent.
 2375: 
 2376: @item file name format
 2377: System dependent. Gforth just uses the file name format of your OS.
 2378: 
 2379: @item information returned by @code{FILE-STATUS}:
 2380: @code{FILE-STATUS} returns the most powerful file access mode allowed
 2381: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
 2382: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
 2383: along with the retured mode.
 2384: 
 2385: @item input file state after an exception when including source:
 2386: All files that are left via the exception are closed.
 2387: 
 2388: @item @var{ior} values and meaning:
 2389: The @var{ior}s returned by the file and memory allocation words are
 2390: intended as throw codes. They typically are in the range
 2391: -512@minus{}-2047 of OS errors.  The mapping from OS error numbers to
 2392: @var{ior}s is -512@minus{}@var{errno}.
 2393: 
 2394: @item maximum depth of file input nesting:
 2395: limited by the amount of return stack, locals/TIB stack, and the number
 2396: of open files available. This should not give you troubles.
 2397: 
 2398: @item maximum size of input line:
 2399: @code{/line}. Currently 255.
 2400: 
 2401: @item methods of mapping block ranges to files:
 2402: Currently, the block words automatically access the file
 2403: @file{blocks.fb} in the currend working directory. More sophisticated
 2404: methods could be implemented if there is demand (and a volunteer).
 2405: 
 2406: @item number of string buffers provided by @code{S"}:
 2407: 1
 2408: 
 2409: @item size of string buffer used by @code{S"}:
 2410: @code{/line}. currently 255.
 2411: 
 2412: @end table
 2413: 
 2414: @c ---------------------------------------------------------------------
 2415: @node file-ambcond,  , file-idef, The optional File-Access word set
 2416: @subsection Ambiguous conditions
 2417: @c ---------------------------------------------------------------------
 2418: 
 2419: @table @i
 2420: 
 2421: @item attempting to position a file outside it's boundaries:
 2422: @code{REPOSITION-FILE} is performed as usual: Afterwards,
 2423: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
 2424: 
 2425: @item attempting to read from file positions not yet written:
 2426: End-of-file, i.e., zero characters are read and no error is reported.
 2427: 
 2428: @item @var{file-id} is invalid (@code{INCLUDE-FILE}):
 2429: An appropriate exception may be thrown, but a memory fault or other
 2430: problem is more probable.
 2431: 
 2432: @item I/O exception reading or closing @var{file-id} (@code{include-file}, @code{included}):
 2433: The @var{ior} produced by the operation, that discovered the problem, is
 2434: thrown.
 2435: 
 2436: @item named file cannot be opened (@code{included}):
 2437: The @var{ior} produced by @code{open-file} is thrown.
 2438: 
 2439: @item requesting an unmapped block number:
 2440: There are no unmapped legal block numbers. On some operating systems,
 2441: writing a block with a large number may overflow the file system and
 2442: have an error message as consequence.
 2443: 
 2444: @item using @code{source-id} when @code{blk} is non-zero:
 2445: @code{source-id} performs its function. Typically it will give the id of
 2446: the source which loaded the block. (Better ideas?)
 2447: 
 2448: @end table
 2449: 
 2450: 
 2451: @c =====================================================================
 2452: @node  The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
 2453: @section The optional Floating-Point word set
 2454: @c =====================================================================
 2455: 
 2456: @menu
 2457: * floating-idef::               Implementation Defined Options
 2458: * floating-ambcond::            Ambiguous Conditions            
 2459: @end menu
 2460: 
 2461: 
 2462: @c ---------------------------------------------------------------------
 2463: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
 2464: @subsection Implementation Defined Options
 2465: @c ---------------------------------------------------------------------
 2466: 
 2467: @table @i
 2468: 
 2469: @item format and range of floating point numbers:
 2470: System-dependent; the @code{double} type of C.
 2471: 
 2472: @item results of @code{REPRESENT} when @var{float} is out of range:
 2473: System dependent; @code{REPRESENT} is implemented using the C library
 2474: function @code{ecvt()} and inherits its behaviour in this respect.
 2475: 
 2476: @item rounding or truncation of floating-point numbers:
 2477: What's the question?!!
 2478: 
 2479: @item size of floating-point stack:
 2480: @code{s" FLOATING-STACK" environment? drop .}. Can be changed at startup
 2481: with the command-line option @code{-f}.
 2482: 
 2483: @item width of floating-point stack:
 2484: @code{1 floats}.
 2485: 
 2486: @end table
 2487: 
 2488: 
 2489: @c ---------------------------------------------------------------------
 2490: @node floating-ambcond,  , floating-idef, The optional Floating-Point word set
 2491: @subsection Ambiguous conditions
 2492: @c ---------------------------------------------------------------------
 2493: 
 2494: @table @i
 2495: 
 2496: @item @code{df@@} or @code{df!} used with an address that is not double-float  aligned:
 2497: System-dependent. Typically results in an alignment fault like other
 2498: alignment violations.
 2499: 
 2500: @item @code{f@@} or @code{f!} used with an address that is not float  aligned:
 2501: System-dependent. Typically results in an alignment fault like other
 2502: alignment violations.
 2503: 
 2504: @item Floating-point result out of range:
 2505: System-dependent. Can result in a @code{-55 THROW} (Floating-point
 2506: unidentified fault), or can produce a special value representing, e.g.,
 2507: Infinity.
 2508: 
 2509: @item @code{sf@@} or @code{sf!} used with an address that is not single-float  aligned:
 2510: System-dependent. Typically results in an alignment fault like other
 2511: alignment violations.
 2512: 
 2513: @item BASE is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
 2514: The floating-point number is converted into decimal nonetheless.
 2515: 
 2516: @item Both arguments are equal to zero (@code{FATAN2}):
 2517: System-dependent. @code{FATAN2} is implemented using the C library
 2518: function @code{atan2()}.
 2519: 
 2520: @item Using ftan on an argument @var{r1} where cos(@var{r1}) is zero:
 2521: System-dependent. Anyway, typically the cos of @var{r1} will not be zero
 2522: because of small errors and the tan will be a very large (or very small)
 2523: but finite number.
 2524: 
 2525: @item @var{d} cannot be presented precisely as a float in @code{D>F}:
 2526: The result is rounded to the nearest float.
 2527: 
 2528: @item dividing by zero:
 2529: @code{-55 throw} (Floating-point unidentified fault)
 2530: 
 2531: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
 2532: System dependent. On IEEE-FP based systems the number is converted into
 2533: an infinity.
 2534: 
 2535: @item @var{float}<1 (@code{facosh}):
 2536: @code{-55 throw} (Floating-point unidentified fault)
 2537: 
 2538: @item @var{float}=<-1 (@code{flnp1}):
 2539: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
 2540: negative infinity is typically produced for @var{float}=-1.
 2541: 
 2542: @item @var{float}=<0 (@code{fln}, @code{flog}):
 2543: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
 2544: negative infinity is typically produced for @var{float}=0.
 2545: 
 2546: @item @var{float}<0 (@code{fasinh}, @code{fsqrt}):
 2547: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
 2548: produces values for these inputs on my Linux box (Bug in the C library?)
 2549: 
 2550: @item |@var{float}|>1 (@code{facos}, @code{fasin}, @code{fatanh}):
 2551: @code{-55 throw} (Floating-point unidentified fault).
 2552: 
 2553: @item integer part of float cannot be represented by @var{d} in @code{f>d}:
 2554: @code{-55 throw} (Floating-point unidentified fault).
 2555: 
 2556: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
 2557: This does not happen.
 2558: @end table
 2559: 
 2560: 
 2561: 
 2562: @c =====================================================================
 2563: @node  The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
 2564: @section The optional Locals word set
 2565: @c =====================================================================
 2566: 
 2567: @menu
 2568: * locals-idef::                 Implementation Defined Options                 
 2569: * locals-ambcond::              Ambiguous Conditions              
 2570: @end menu
 2571: 
 2572: 
 2573: @c ---------------------------------------------------------------------
 2574: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
 2575: @subsection Implementation Defined Options
 2576: @c ---------------------------------------------------------------------
 2577: 
 2578: @table @i
 2579: 
 2580: @item maximum number of locals in a definition:
 2581: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
 2582: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
 2583: characters. The number of locals in a definition is bounded by the size
 2584: of locals-buffer, which contains the names of the locals.
 2585: 
 2586: @end table
 2587: 
 2588: 
 2589: @c ---------------------------------------------------------------------
 2590: @node locals-ambcond,  , locals-idef, The optional Locals word set
 2591: @subsection Ambiguous conditions
 2592: @c ---------------------------------------------------------------------
 2593: 
 2594: @table @i
 2595: 
 2596: @item executing a named local in interpretation state:
 2597: @code{-14 throw} (Interpreting a compile-only word).
 2598: 
 2599: @item @var{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
 2600: @code{-32 throw} (Invalid name argument)
 2601: 
 2602: @end table
 2603: 
 2604: 
 2605: @c =====================================================================
 2606: @node  The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
 2607: @section The optional Memory-Allocation word set
 2608: @c =====================================================================
 2609: 
 2610: @menu
 2611: * memory-idef::                 Implementation Defined Options                 
 2612: @end menu
 2613: 
 2614: 
 2615: @c ---------------------------------------------------------------------
 2616: @node memory-idef,  , The optional Memory-Allocation word set, The optional Memory-Allocation word set
 2617: @subsection Implementation Defined Options
 2618: @c ---------------------------------------------------------------------
 2619: 
 2620: @table @i
 2621: 
 2622: @item values and meaning of @var{ior}:
 2623: The @var{ior}s returned by the file and memory allocation words are
 2624: intended as throw codes. They typically are in the range
 2625: -512@minus{}-2047 of OS errors.  The mapping from OS error numbers to
 2626: @var{ior}s is -512@minus{}@var{errno}.
 2627: 
 2628: @end table
 2629: 
 2630: @c =====================================================================
 2631: @node  The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
 2632: @section The optional Programming-Tools word set
 2633: @c =====================================================================
 2634: 
 2635: @menu
 2636: * programming-idef::            Implementation Defined Options            
 2637: * programming-ambcond::         Ambiguous Conditions         
 2638: @end menu
 2639: 
 2640: 
 2641: @c ---------------------------------------------------------------------
 2642: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
 2643: @subsection Implementation Defined Options
 2644: @c ---------------------------------------------------------------------
 2645: 
 2646: @table @i
 2647: 
 2648: @item ending sequence for input following @code{;code} and @code{code}:
 2649: Not implemented (yet).
 2650: 
 2651: @item manner of processing input following @code{;code} and @code{code}:
 2652: Not implemented (yet).
 2653: 
 2654: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
 2655: Not implemented (yet). If they were implemented, they would use the
 2656: search order wordset.
 2657: 
 2658: @item source and format of display by @code{SEE}:
 2659: The source for @code{see} is the intermediate code used by the inner
 2660: interpreter.  The current @code{see} tries to output Forth source code
 2661: as well as possible.
 2662: 
 2663: @end table
 2664: 
 2665: @c ---------------------------------------------------------------------
 2666: @node programming-ambcond,  , programming-idef, The optional Programming-Tools word set
 2667: @subsection Ambiguous conditions
 2668: @c ---------------------------------------------------------------------
 2669: 
 2670: @table @i
 2671: 
 2672: @item deleting the compilation wordlist (@code{FORGET}):
 2673: Not implemented (yet).
 2674: 
 2675: @item fewer than @var{u}+1 items on the control flow stack (@code{CS-PICK}, @code{CS-ROLL}):
 2676: This typically results in an @code{abort"} with a descriptive error
 2677: message (may change into a @code{-22 throw} (Control structure mismatch)
 2678: in the future). You may also get a memory access error. If you are
 2679: unlucky, this ambiguous condition is not caught.
 2680: 
 2681: @item @var{name} can't be found (@code{forget}):
 2682: Not implemented (yet).
 2683: 
 2684: @item @var{name} not defined via @code{CREATE}:
 2685: @code{;code} is not implemented (yet). If it were, it would behave like
 2686: @code{DOES>} in this respect, i.e., change the execution semantics of
 2687: the last defined word no matter how it was defined.
 2688: 
 2689: @item @code{POSTPONE} applied to @code{[IF]}:
 2690: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
 2691: equivalent to @code{[IF]}.
 2692: 
 2693: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
 2694: Continue in the same state of conditional compilation in the next outer
 2695: input source. Currently there is no warning to the user about this.
 2696: 
 2697: @item removing a needed definition (@code{FORGET}):
 2698: Not implemented (yet).
 2699: 
 2700: @end table
 2701: 
 2702: 
 2703: @c =====================================================================
 2704: @node  The optional Search-Order word set,  , The optional Programming-Tools word set, ANS conformance
 2705: @section The optional Search-Order word set
 2706: @c =====================================================================
 2707: 
 2708: @menu
 2709: * search-idef::                 Implementation Defined Options                 
 2710: * search-ambcond::              Ambiguous Conditions              
 2711: @end menu
 2712: 
 2713: 
 2714: @c ---------------------------------------------------------------------
 2715: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
 2716: @subsection Implementation Defined Options
 2717: @c ---------------------------------------------------------------------
 2718: 
 2719: @table @i
 2720: 
 2721: @item maximum number of word lists in search order:
 2722: @code{s" wordlists" environment? drop .}. Currently 16.
 2723: 
 2724: @item minimum search order:
 2725: @code{root root}.
 2726: 
 2727: @end table
 2728: 
 2729: @c ---------------------------------------------------------------------
 2730: @node search-ambcond,  , search-idef, The optional Search-Order word set
 2731: @subsection Ambiguous conditions
 2732: @c ---------------------------------------------------------------------
 2733: 
 2734: @table @i
 2735: 
 2736: @item changing the compilation wordlist (during compilation):
 2737: The definition is put into the wordlist that is the compilation wordlist
 2738: when @code{REVEAL} is executed (by @code{;}, @code{DOES>},
 2739: @code{RECURSIVE}, etc.).
 2740: 
 2741: @item search order empty (@code{previous}):
 2742: @code{abort" Vocstack empty"}.
 2743: 
 2744: @item too many word lists in search order (@code{also}):
 2745: @code{abort" Vocstack full"}.
 2746: 
 2747: @end table
 2748: 
 2749: 
 2750: @node Model, Emacs and Gforth, ANS conformance, Top
 2751: @chapter Model
 2752: 
 2753: @node Emacs and Gforth, Internals, Model, Top
 2754: @chapter Emacs and Gforth
 2755: 
 2756: Gforth comes with @file{gforth.el}, an improved version of
 2757: @file{forth.el} by Goran Rydqvist (icluded in the TILE package). The
 2758: improvements are a better (but still not perfect) handling of
 2759: indentation. I have also added comment paragraph filling (@kbd{M-q}),
 2760: commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) regions and
 2761: removing debugging tracers (@kbd{C-x ~}, @pxref{Debugging}). I left the
 2762: stuff I do not use alone, even though some of it only makes sense for
 2763: TILE. To get a description of these features, enter Forth mode and type
 2764: @kbd{C-h m}.
 2765: 
 2766: In addition, Gforth supports Emacs quite well: The source code locations
 2767: given in error messages, debugging output (from @code{~~}) and failed
 2768: assertion messages are in the right format for Emacs' compilation mode
 2769: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
 2770: Manual}) so the source location corresponding to an error or other
 2771: message is only a few keystrokes away (@kbd{C-x `} for the next error,
 2772: @kbd{C-c C-c} for the error under the cursor).
 2773: 
 2774: Also, if you @code{include} @file{etags.fs}, a new @file{TAGS} file
 2775: (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) will be produced that
 2776: contains the definitions of all words defined afterwards. You can then
 2777: find the source for a word using @kbd{M-.}. Note that emacs can use
 2778: several tags files at the same time (e.g., one for the Gforth sources
 2779: and one for your program).
 2780: 
 2781: To get all these benefits, add the following lines to your @file{.emacs}
 2782: file:
 2783: 
 2784: @example
 2785: (autoload 'forth-mode "gforth.el")
 2786: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
 2787: @end example
 2788: 
 2789: @node Internals, Bugs, Emacs and Gforth, Top
 2790: @chapter Internals
 2791: 
 2792: Reading this section is not necessary for programming with Gforth. It
 2793: should be helpful for finding your way in the Gforth sources.
 2794: 
 2795: @menu
 2796: * Portability::                 
 2797: * Threading::                   
 2798: * Primitives::                  
 2799: * System Architecture::         
 2800: * Performance::                 
 2801: @end menu
 2802: 
 2803: @node Portability, Threading, Internals, Internals
 2804: @section Portability
 2805: 
 2806: One of the main goals of the effort is availability across a wide range
 2807: of personal machines. fig-Forth, and, to a lesser extent, F83, achieved
 2808: this goal by manually coding the engine in assembly language for several
 2809: then-popular processors. This approach is very labor-intensive and the
 2810: results are short-lived due to progress in computer architecture.
 2811: 
 2812: Others have avoided this problem by coding in C, e.g., Mitch Bradley
 2813: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
 2814: particularly popular for UNIX-based Forths due to the large variety of
 2815: architectures of UNIX machines. Unfortunately an implementation in C
 2816: does not mix well with the goals of efficiency and with using
 2817: traditional techniques: Indirect or direct threading cannot be expressed
 2818: in C, and switch threading, the fastest technique available in C, is
 2819: significantly slower. Another problem with C is that it's very
 2820: cumbersome to express double integer arithmetic.
 2821: 
 2822: Fortunately, there is a portable language that does not have these
 2823: limitations: GNU C, the version of C processed by the GNU C compiler
 2824: (@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
 2825: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
 2826: Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
 2827: threading possible, its @code{long long} type (@pxref{Long Long, ,
 2828: Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forths
 2829: double numbers. GNU C is available for free on all important (and many
 2830: unimportant) UNIX machines, VMS, 80386s running MS-DOS, the Amiga, and
 2831: the Atari ST, so a Forth written in GNU C can run on all these
 2832: machines.
 2833: 
 2834: Writing in a portable language has the reputation of producing code that
 2835: is slower than assembly. For our Forth engine we repeatedly looked at
 2836: the code produced by the compiler and eliminated most compiler-induced
 2837: inefficiencies by appropriate changes in the source-code.
 2838: 
 2839: However, register allocation cannot be portably influenced by the
 2840: programmer, leading to some inefficiencies on register-starved
 2841: machines. We use explicit register declarations (@pxref{Explicit Reg
 2842: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
 2843: improve the speed on some machines. They are turned on by using the
 2844: @code{gcc} switch @code{-DFORCE_REG}. Unfortunately, this feature not
 2845: only depends on the machine, but also on the compiler version: On some
 2846: machines some compiler versions produce incorrect code when certain
 2847: explicit register declarations are used. So by default
 2848: @code{-DFORCE_REG} is not used.
 2849: 
 2850: @node Threading, Primitives, Portability, Internals
 2851: @section Threading
 2852: 
 2853: GNU C's labels as values extension (available since @code{gcc-2.0},
 2854: @pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
 2855: makes it possible to take the address of @var{label} by writing
 2856: @code{&&@var{label}}.  This address can then be used in a statement like
 2857: @code{goto *@var{address}}. I.e., @code{goto *&&x} is the same as
 2858: @code{goto x}.
 2859: 
 2860: With this feature an indirect threaded NEXT looks like:
 2861: @example
 2862: cfa = *ip++;
 2863: ca = *cfa;
 2864: goto *ca;
 2865: @end example
 2866: For those unfamiliar with the names: @code{ip} is the Forth instruction
 2867: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
 2868: execution token and points to the code field of the next word to be
 2869: executed; The @code{ca} (code address) fetched from there points to some
 2870: executable code, e.g., a primitive or the colon definition handler
 2871: @code{docol}.
 2872: 
 2873: Direct threading is even simpler:
 2874: @example
 2875: ca = *ip++;
 2876: goto *ca;
 2877: @end example
 2878: 
 2879: Of course we have packaged the whole thing neatly in macros called
 2880: @code{NEXT} and @code{NEXT1} (the part of NEXT after fetching the cfa).
 2881: 
 2882: @menu
 2883: * Scheduling::                  
 2884: * Direct or Indirect Threaded?::  
 2885: * DOES>::                       
 2886: @end menu
 2887: 
 2888: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
 2889: @subsection Scheduling
 2890: 
 2891: There is a little complication: Pipelined and superscalar processors,
 2892: i.e., RISC and some modern CISC machines can process independent
 2893: instructions while waiting for the results of an instruction. The
 2894: compiler usually reorders (schedules) the instructions in a way that
 2895: achieves good usage of these delay slots. However, on our first tries
 2896: the compiler did not do well on scheduling primitives. E.g., for
 2897: @code{+} implemented as
 2898: @example
 2899: n=sp[0]+sp[1];
 2900: sp++;
 2901: sp[0]=n;
 2902: NEXT;
 2903: @end example
 2904: the NEXT comes strictly after the other code, i.e., there is nearly no
 2905: scheduling. After a little thought the problem becomes clear: The
 2906: compiler cannot know that sp and ip point to different addresses (and
 2907: the version of @code{gcc} we used would not know it even if it was
 2908: possible), so it could not move the load of the cfa above the store to
 2909: the TOS. Indeed the pointers could be the same, if code on or very near
 2910: the top of stack were executed. In the interest of speed we chose to
 2911: forbid this probably unused ``feature'' and helped the compiler in
 2912: scheduling: NEXT is divided into the loading part (@code{NEXT_P1}) and
 2913: the goto part (@code{NEXT_P2}). @code{+} now looks like:
 2914: @example
 2915: n=sp[0]+sp[1];
 2916: sp++;
 2917: NEXT_P1;
 2918: sp[0]=n;
 2919: NEXT_P2;
 2920: @end example
 2921: This can be scheduled optimally by the compiler.
 2922: 
 2923: This division can be turned off with the switch @code{-DCISC_NEXT}. This
 2924: switch is on by default on machines that do not profit from scheduling
 2925: (e.g., the 80386), in order to preserve registers.
 2926: 
 2927: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
 2928: @subsection Direct or Indirect Threaded?
 2929: 
 2930: Both! After packaging the nasty details in macro definitions we
 2931: realized that we could switch between direct and indirect threading by
 2932: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
 2933: defining a few machine-specific macros for the direct-threading case.
 2934: On the Forth level we also offer access words that hide the
 2935: differences between the threading methods (@pxref{Threading Words}).
 2936: 
 2937: Indirect threading is implemented completely
 2938: machine-independently. Direct threading needs routines for creating
 2939: jumps to the executable code (e.g. to docol or dodoes). These routines
 2940: are inherently machine-dependent, but they do not amount to many source
 2941: lines. I.e., even porting direct threading to a new machine is a small
 2942: effort.
 2943: 
 2944: @node DOES>,  , Direct or Indirect Threaded?, Threading
 2945: @subsection DOES>
 2946: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
 2947: the chunk of code executed by every word defined by a
 2948: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
 2949: the Forth code to be executed, i.e. the code after the @code{DOES>} (the
 2950: DOES-code)? There are two solutions:
 2951: 
 2952: In fig-Forth the code field points directly to the dodoes and the
 2953: DOES-code address is stored in the cell after the code address
 2954: (i.e. at cfa cell+). It may seem that this solution is illegal in the
 2955: Forth-79 and all later standards, because in fig-Forth this address
 2956: lies in the body (which is illegal in these standards). However, by
 2957: making the code field larger for all words this solution becomes legal
 2958: again. We use this approach for the indirect threaded version. Leaving
 2959: a cell unused in most words is a bit wasteful, but on the machines we
 2960: are targetting this is hardly a problem. The other reason for having a
 2961: code field size of two cells is to avoid having different image files
 2962: for direct and indirect threaded systems (@pxref{System Architecture}).
 2963: 
 2964: The other approach is that the code field points or jumps to the cell
 2965: after @code{DOES}. In this variant there is a jump to @code{dodoes} at
 2966: this address. @code{dodoes} can then get the DOES-code address by
 2967: computing the code address, i.e., the address of the jump to dodoes,
 2968: and add the length of that jump field. A variant of this is to have a
 2969: call to @code{dodoes} after the @code{DOES>}; then the return address
 2970: (which can be found in the return register on RISCs) is the DOES-code
 2971: address. Since the two cells available in the code field are usually
 2972: used up by the jump to the code address in direct threading, we use
 2973: this approach for direct threading. We did not want to add another
 2974: cell to the code field.
 2975: 
 2976: @node Primitives, System Architecture, Threading, Internals
 2977: @section Primitives
 2978: 
 2979: @menu
 2980: * Automatic Generation::        
 2981: * TOS Optimization::            
 2982: * Produced code::               
 2983: @end menu
 2984: 
 2985: @node Automatic Generation, TOS Optimization, Primitives, Primitives
 2986: @subsection Automatic Generation
 2987: 
 2988: Since the primitives are implemented in a portable language, there is no
 2989: longer any need to minimize the number of primitives. On the contrary,
 2990: having many primitives is an advantage: speed. In order to reduce the
 2991: number of errors in primitives and to make programming them easier, we
 2992: provide a tool, the primitive generator (@file{prims2x.fs}), that
 2993: automatically generates most (and sometimes all) of the C code for a
 2994: primitive from the stack effect notation.  The source for a primitive
 2995: has the following form:
 2996: 
 2997: @format
 2998: @var{Forth-name}	@var{stack-effect}	@var{category}	[@var{pronounc.}]
 2999: [@code{""}@var{glossary entry}@code{""}]
 3000: @var{C code}
 3001: [@code{:}
 3002: @var{Forth code}]
 3003: @end format
 3004: 
 3005: The items in brackets are optional. The category and glossary fields
 3006: are there for generating the documentation, the Forth code is there
 3007: for manual implementations on machines without GNU C. E.g., the source
 3008: for the primitive @code{+} is:
 3009: @example
 3010: +    n1 n2 -- n    core    plus
 3011: n = n1+n2;
 3012: @end example
 3013: 
 3014: This looks like a specification, but in fact @code{n = n1+n2} is C
 3015: code. Our primitive generation tool extracts a lot of information from
 3016: the stack effect notations@footnote{We use a one-stack notation, even
 3017: though we have separate data and floating-point stacks; The separate
 3018: notation can be generated easily from the unified notation.}: The number
 3019: of items popped from and pushed on the stack, their type, and by what
 3020: name they are referred to in the C code. It then generates a C code
 3021: prelude and postlude for each primitive. The final C code for @code{+}
 3022: looks like this:
 3023: 
 3024: @example
 3025: I_plus:	/* + ( n1 n2 -- n ) */  /* label, stack effect */
 3026: /*  */                          /* documentation */
 3027: @{
 3028: DEF_CA                          /* definition of variable ca (indirect threading) */
 3029: Cell n1;                        /* definitions of variables */
 3030: Cell n2;
 3031: Cell n;
 3032: n1 = (Cell) sp[1];              /* input */
 3033: n2 = (Cell) TOS;
 3034: sp += 1;                        /* stack adjustment */
 3035: NAME("+")                       /* debugging output (with -DDEBUG) */
 3036: @{
 3037: n = n1+n2;                      /* C code taken from the source */
 3038: @}
 3039: NEXT_P1;                        /* NEXT part 1 */
 3040: TOS = (Cell)n;                  /* output */
 3041: NEXT_P2;                        /* NEXT part 2 */
 3042: @}
 3043: @end example
 3044: 
 3045: This looks long and inefficient, but the GNU C compiler optimizes quite
 3046: well and produces optimal code for @code{+} on, e.g., the R3000 and the
 3047: HP RISC machines: Defining the @code{n}s does not produce any code, and
 3048: using them as intermediate storage also adds no cost.
 3049: 
 3050: There are also other optimizations, that are not illustrated by this
 3051: example: Assignments between simple variables are usually for free (copy
 3052: propagation). If one of the stack items is not used by the primitive
 3053: (e.g.  in @code{drop}), the compiler eliminates the load from the stack
 3054: (dead code elimination). On the other hand, there are some things that
 3055: the compiler does not do, therefore they are performed by
 3056: @file{prims2x.fs}: The compiler does not optimize code away that stores
 3057: a stack item to the place where it just came from (e.g., @code{over}).
 3058: 
 3059: While programming a primitive is usually easy, there are a few cases
 3060: where the programmer has to take the actions of the generator into
 3061: account, most notably @code{?dup}, but also words that do not (always)
 3062: fall through to NEXT.
 3063: 
 3064: @node TOS Optimization, Produced code, Automatic Generation, Primitives
 3065: @subsection TOS Optimization
 3066: 
 3067: An important optimization for stack machine emulators, e.g., Forth
 3068: engines, is keeping  one or more of the top stack items in
 3069: registers.  If a word has the stack effect @var{in1}...@var{inx} @code{--}
 3070: @var{out1}...@var{outy}, keeping the top @var{n} items in registers
 3071: @itemize
 3072: @item
 3073: is better than keeping @var{n-1} items, if @var{x>=n} and @var{y>=n},
 3074: due to fewer loads from and stores to the stack.
 3075: @item is slower than keeping @var{n-1} items, if @var{x<>y} and @var{x<n} and
 3076: @var{y<n}, due to additional moves between registers.
 3077: @end itemize
 3078: 
 3079: In particular, keeping one item in a register is never a disadvantage,
 3080: if there are enough registers. Keeping two items in registers is a
 3081: disadvantage for frequent words like @code{?branch}, constants,
 3082: variables, literals and @code{i}. Therefore our generator only produces
 3083: code that keeps zero or one items in registers. The generated C code
 3084: covers both cases; the selection between these alternatives is made at
 3085: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
 3086: code for @code{+} is just a simple variable name in the one-item case,
 3087: otherwise it is a macro that expands into @code{sp[0]}. Note that the
 3088: GNU C compiler tries to keep simple variables like @code{TOS} in
 3089: registers, and it usually succeeds, if there are enough registers.
 3090: 
 3091: The primitive generator performs the TOS optimization for the
 3092: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
 3093: operations the benefit of this optimization is even larger:
 3094: floating-point operations take quite long on most processors, but can be
 3095: performed in parallel with other operations as long as their results are
 3096: not used. If the FP-TOS is kept in a register, this works. If
 3097: it is kept on the stack, i.e., in memory, the store into memory has to
 3098: wait for the result of the floating-point operation, lengthening the
 3099: execution time of the primitive considerably.
 3100: 
 3101: The TOS optimization makes the automatic generation of primitives a
 3102: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
 3103: @code{TOS} is not sufficient. There are some special cases to
 3104: consider:
 3105: @itemize
 3106: @item In the case of @code{dup ( w -- w w )} the generator must not
 3107: eliminate the store to the original location of the item on the stack,
 3108: if the TOS optimization is turned on.
 3109: @item Primitives with stack effects of the form @code{--}
 3110: @var{out1}...@var{outy} must store the TOS to the stack at the start.
 3111: Likewise, primitives with the stack effect @var{in1}...@var{inx} @code{--}
 3112: must load the TOS from the stack at the end. But for the null stack
 3113: effect @code{--} no stores or loads should be generated.
 3114: @end itemize
 3115: 
 3116: @node Produced code,  , TOS Optimization, Primitives
 3117: @subsection Produced code
 3118: 
 3119: To see what assembly code is produced for the primitives on your machine
 3120: with your compiler and your flag settings, type @code{make engine.s} and
 3121: look at the resulting file @file{engine.s}.
 3122: 
 3123: @node System Architecture, Performance, Primitives, Internals
 3124: @section System Architecture
 3125: 
 3126: Our Forth system consists not only of primitives, but also of
 3127: definitions written in Forth. Since the Forth compiler itself belongs
 3128: to those definitions, it is not possible to start the system with the
 3129: primitives and the Forth source alone. Therefore we provide the Forth
 3130: code as an image file in nearly executable form. At the start of the
 3131: system a C routine loads the image file into memory, sets up the
 3132: memory (stacks etc.) according to information in the image file, and
 3133: starts executing Forth code.
 3134: 
 3135: The image file format is a compromise between the goals of making it
 3136: easy to generate image files and making them portable. The easiest way
 3137: to generate an image file is to just generate a memory dump. However,
 3138: this kind of image file cannot be used on a different machine, or on
 3139: the next version of the engine on the same machine, it even might not
 3140: work with the same engine compiled by a different version of the C
 3141: compiler. We would like to have as few versions of the image file as
 3142: possible, because we do not want to distribute many versions of the
 3143: same image file, and to make it easy for the users to use their image
 3144: files on many machines. We currently need to create a different image
 3145: file for machines with different cell sizes and different byte order
 3146: (little- or big-endian)@footnote{We are considering adding information to the
 3147: image file that enables the loader to change the byte order.}.
 3148: 
 3149: Forth code that is going to end up in a portable image file has to
 3150: comply to some restrictions: addresses have to be stored in memory with
 3151: special words (@code{A!}, @code{A,}, etc.) in order to make the code
 3152: relocatable. Cells, floats, etc., have to be stored at the natural
 3153: alignment boundaries@footnote{E.g., store floats (8 bytes) at an address
 3154: dividable by~8. This happens automatically in our system when you use
 3155: the ANS Forth alignment words.}, in order to avoid alignment faults on
 3156: machines with stricter alignment. The image file is produced by a
 3157: metacompiler (@file{cross.fs}).
 3158: 
 3159: So, unlike the image file of Mitch Bradleys @code{cforth}, our image
 3160: file is not directly executable, but has to undergo some manipulations
 3161: during loading. Address relocation is performed at image load-time, not
 3162: at run-time. The loader also has to replace tokens standing for
 3163: primitive calls with the appropriate code-field addresses (or code
 3164: addresses in the case of direct threading).
 3165: 
 3166: @node  Performance,  , System Architecture, Internals
 3167: @section Performance
 3168: 
 3169: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
 3170: impossible to write a significantly faster engine.
 3171: 
 3172: On register-starved machines like the 386 architecture processors
 3173: improvements are possible, because @code{gcc} does not utilize the
 3174: registers as well as a human, even with explicit register declarations;
 3175: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
 3176: and hand-tuned it for the 486; this system is 1.19 times faster on the
 3177: Sieve benchmark on a 486DX2/66 than Gforth compiled with
 3178: @code{gcc-2.6.3} with @code{-DFORCE_REG}.
 3179: 
 3180: However, this potential advantage of assembly language implementations
 3181: is not necessarily realized in complete Forth systems: We compared
 3182: Gforth (compiled with @code{gcc-2.6.3} and @code{-DFORCE_REG}) with
 3183: Win32Forth 1.2093 and LMI's NT Forth (Beta, May 1994), two systems
 3184: written in assembly, and with two systems written in C: PFE-0.9.11
 3185: (compiled with @code{gcc-2.6.3} with the default configuration for
 3186: Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS}) and ThisForth Beta
 3187: (compiled with gcc-2.6.3 -O3 -fomit-frame-pointer). We benchmarked
 3188: Gforth, PFE and ThisForth on a 486DX2/66 under Linux. Kenneth O'Heskin
 3189: kindly provided the results for Win32Forth and NT Forth on a 486DX2/66
 3190: with similar memory performance under Windows NT.
 3191:  
 3192: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
 3193: matrix multiplication come from the Stanford integer benchmarks and have
 3194: been translated into Forth by Martin Fraeman; we used the versions
 3195: included in the TILE Forth package; and a recursive Fibonacci number
 3196: computation for benchmark calling performance. The following table shows
 3197: the time taken for the benchmarks scaled by the time taken by Gforth (in
 3198: other words, it shows the speedup factor that Gforth achieved over the
 3199: other systems).
 3200: 
 3201: @example
 3202: relative             Win32-        NT               This-
 3203:   time     Gforth     Forth     Forth       PFE     Forth
 3204: sieve        1.00      1.30      1.07      1.67      2.98
 3205: bubble       1.00      1.30      1.40      1.66
 3206: matmul       1.00      1.40      1.29      2.24
 3207: fib          1.00      1.44      1.26      1.82      2.82
 3208: @end example
 3209: 
 3210: You may find the good performance of Gforth compared with the systems
 3211: written in assembly language quite surprising. One important reason for
 3212: the disappointing performance of these systems is probably that they are
 3213: not written optimally for the 486 (e.g., they use the @code{lods}
 3214: instruction). In addition, Win32Forth uses a comfortable, but costly
 3215: method for relocating the Forth image: like @code{cforth}, it computes
 3216: the actual addresses at run time, resulting in two address computations
 3217: per NEXT (@pxref{System Architecture}).
 3218: 
 3219: The speedup of Gforth over PFE and ThisForth can be easily explained
 3220: with the self-imposed restriction to standard C (although the measured
 3221: implementation of PFE uses a GNU C extension: global register
 3222: variables), which makes efficient threading impossible.  Moreover,
 3223: current C compilers have a hard time optimizing other aspects of the
 3224: ThisForth source.
 3225: 
 3226: Note that the performance of Gforth on 386 architecture processors
 3227: varies widely with the version of @code{gcc} used. E.g., @code{gcc-2.5.8}
 3228: failed to allocate any of the virtual machine registers into real
 3229: machine registers by itself and would not work correctly with explicit
 3230: register declarations, giving a 1.3 times slower engine (on a 486DX2/66
 3231: running the Sieve) than the one measured above.
 3232: 
 3233: @node Bugs, Pedigree, Internals, Top
 3234: @chapter Bugs
 3235: 
 3236: Known bugs are described in the file BUGS in the Gforth distribution.
 3237: 
 3238: If you find a bug, please send a bug report to !!. A bug report should
 3239: describe the Gforth version used (it is announced at the start of an
 3240: interactive Gforth session), the machine and operating system (on Unix
 3241: systems you can use @code{uname -a} to produce this information), the
 3242: installation options (!! a way to find them out), and a complete list of
 3243: changes you (or your installer) have made to the Gforth sources (if
 3244: any); it should contain a program (or a sequence of keyboard commands)
 3245: that reproduces the bug and a description of what you think constitutes
 3246: the buggy behaviour.
 3247: 
 3248: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
 3249: to Report Bugs, gcc.info, GNU C Manual}.
 3250: 
 3251: 
 3252: @node Pedigree, Word Index, Bugs, Top
 3253: @chapter Pedigree
 3254: 
 3255: Gforth descends from BigForth (1993) and fig-Forth. Gforth and PFE (by
 3256: Dirk Zoller) will cross-fertilize each other. Of course, a significant part of the design of Gforth was prescribed by ANS Forth.
 3257: 
 3258: Bernd Paysan wrote BigForth, a child of VolksForth.
 3259: 
 3260: VolksForth descends from F83. !! Authors? When?
 3261: 
 3262: Laxen and Perry wrote F83 as a model implementation of the
 3263: Forth-83 standard. !! Pedigree? When?
 3264: 
 3265: A team led by Bill Ragsdale implemented fig-Forth on many processors in
 3266: 1979. Dean Sanderson and Bill Ragsdale developed the original
 3267: implementation of fig-Forth based on microForth.
 3268: 
 3269: !! microForth pedigree
 3270: 
 3271: A part of the information in this section comes from @cite{The Evolution
 3272: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
 3273: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
 3274: Notices 28(3), 1993.  You can find more historical and genealogical
 3275: information about Forth there.
 3276: 
 3277: @node Word Index, Node Index, Pedigree, Top
 3278: @chapter Word Index
 3279: 
 3280: This index is as incomplete as the manual. Each word is listed with
 3281: stack effect and wordset.
 3282: 
 3283: @printindex fn
 3284: 
 3285: @node Node Index,  , Word Index, Top
 3286: @chapter Node Index
 3287: 
 3288: This index is even less complete than the manual.
 3289: 
 3290: @contents
 3291: @bye
 3292: 

FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>