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: @dircategory GNU programming tools
7: @direntry
8: * Gforth: (gforth). A fast interpreter for the Forth language.
9: @end direntry
10: @comment @setchapternewpage odd
11: @macro progstyle {TEXT}
12: @quotation
13: {Programming style note:} \TEXT\
14: @end quotation
15: @end macro
16: @comment %**end of header (This is for running Texinfo on a region.)
17:
18: @ifinfo
19: This file documents Gforth 0.4
20:
21: Copyright @copyright{} 1995-1997 Free Software Foundation, Inc.
22:
23: Permission is granted to make and distribute verbatim copies of
24: this manual provided the copyright notice and this permission notice
25: are preserved on all copies.
26:
27: @ignore
28: Permission is granted to process this file through TeX and print the
29: results, provided the printed document carries a copying permission
30: notice identical to this one except for the removal of this paragraph
31: (this paragraph not being relevant to the printed manual).
32:
33: @end ignore
34: Permission is granted to copy and distribute modified versions of this
35: manual under the conditions for verbatim copying, provided also that the
36: sections entitled "Distribution" and "General Public License" are
37: included exactly as in the original, and provided that the entire
38: resulting derived work is distributed under the terms of a permission
39: notice identical to this one.
40:
41: Permission is granted to copy and distribute translations of this manual
42: into another language, under the above conditions for modified versions,
43: except that the sections entitled "Distribution" and "General Public
44: License" may be included in a translation approved by the author instead
45: of in the original English.
46: @end ifinfo
47:
48: @finalout
49: @titlepage
50: @sp 10
51: @center @titlefont{Gforth Manual}
52: @sp 2
53: @center for version 0.4
54: @sp 2
55: @center Anton Ertl
56: @center Bernd Paysan
57: @center Jens Wilke
58: @sp 3
59: @center This manual is permanently under construction
60:
61: @comment The following two commands start the copyright page.
62: @page
63: @vskip 0pt plus 1filll
64: Copyright @copyright{} 1995--1997 Free Software Foundation, Inc.
65:
66: @comment !! Published by ... or You can get a copy of this manual ...
67:
68: Permission is granted to make and distribute verbatim copies of
69: this manual provided the copyright notice and this permission notice
70: are preserved on all copies.
71:
72: Permission is granted to copy and distribute modified versions of this
73: manual under the conditions for verbatim copying, provided also that the
74: sections entitled "Distribution" and "General Public License" are
75: included exactly as in the original, and provided that the entire
76: resulting derived work is distributed under the terms of a permission
77: notice identical to this one.
78:
79: Permission is granted to copy and distribute translations of this manual
80: into another language, under the above conditions for modified versions,
81: except that the sections entitled "Distribution" and "General Public
82: License" may be included in a translation approved by the author instead
83: of in the original English.
84: @end titlepage
85:
86:
87: @node Top, License, (dir), (dir)
88: @ifinfo
89: Gforth is a free implementation of ANS Forth available on many
90: personal machines. This manual corresponds to version 0.4.
91: @end ifinfo
92:
93: @menu
94: * License::
95: * Goals:: About the Gforth Project
96: * Other Books:: Things you might want to read
97: * Invoking Gforth:: Starting Gforth
98: * Words:: Forth words available in Gforth
99: * Tools:: Programming tools
100: * ANS conformance:: Implementation-defined options etc.
101: * Model:: The abstract machine of Gforth
102: * Integrating Gforth:: Forth as scripting language for applications
103: * Emacs and Gforth:: The Gforth Mode
104: * Image Files:: @code{.fi} files contain compiled code
105: * Engine:: The inner interpreter and the primitives
106: * Bugs:: How to report them
107: * Origin:: Authors and ancestors of Gforth
108: * Word Index:: An item for each Forth word
109: * Concept Index:: A menu covering many topics
110: @end menu
111:
112: @node License, Goals, Top, Top
113: @unnumbered GNU GENERAL PUBLIC LICENSE
114: @center Version 2, June 1991
115:
116: @display
117: Copyright @copyright{} 1989, 1991 Free Software Foundation, Inc.
118: 675 Mass Ave, Cambridge, MA 02139, USA
119:
120: Everyone is permitted to copy and distribute verbatim copies
121: of this license document, but changing it is not allowed.
122: @end display
123:
124: @unnumberedsec Preamble
125:
126: The licenses for most software are designed to take away your
127: freedom to share and change it. By contrast, the GNU General Public
128: License is intended to guarantee your freedom to share and change free
129: software---to make sure the software is free for all its users. This
130: General Public License applies to most of the Free Software
131: Foundation's software and to any other program whose authors commit to
132: using it. (Some other Free Software Foundation software is covered by
133: the GNU Library General Public License instead.) You can apply it to
134: your programs, too.
135:
136: When we speak of free software, we are referring to freedom, not
137: price. Our General Public Licenses are designed to make sure that you
138: have the freedom to distribute copies of free software (and charge for
139: this service if you wish), that you receive source code or can get it
140: if you want it, that you can change the software or use pieces of it
141: in new free programs; and that you know you can do these things.
142:
143: To protect your rights, we need to make restrictions that forbid
144: anyone to deny you these rights or to ask you to surrender the rights.
145: These restrictions translate to certain responsibilities for you if you
146: distribute copies of the software, or if you modify it.
147:
148: For example, if you distribute copies of such a program, whether
149: gratis or for a fee, you must give the recipients all the rights that
150: you have. You must make sure that they, too, receive or can get the
151: source code. And you must show them these terms so they know their
152: rights.
153:
154: We protect your rights with two steps: (1) copyright the software, and
155: (2) offer you this license which gives you legal permission to copy,
156: distribute and/or modify the software.
157:
158: Also, for each author's protection and ours, we want to make certain
159: that everyone understands that there is no warranty for this free
160: software. If the software is modified by someone else and passed on, we
161: want its recipients to know that what they have is not the original, so
162: that any problems introduced by others will not reflect on the original
163: authors' reputations.
164:
165: Finally, any free program is threatened constantly by software
166: patents. We wish to avoid the danger that redistributors of a free
167: program will individually obtain patent licenses, in effect making the
168: program proprietary. To prevent this, we have made it clear that any
169: patent must be licensed for everyone's free use or not licensed at all.
170:
171: The precise terms and conditions for copying, distribution and
172: modification follow.
173:
174: @iftex
175: @unnumberedsec TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
176: @end iftex
177: @ifinfo
178: @center TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
179: @end ifinfo
180:
181: @enumerate 0
182: @item
183: This License applies to any program or other work which contains
184: a notice placed by the copyright holder saying it may be distributed
185: under the terms of this General Public License. The ``Program'', below,
186: refers to any such program or work, and a ``work based on the Program''
187: means either the Program or any derivative work under copyright law:
188: that is to say, a work containing the Program or a portion of it,
189: either verbatim or with modifications and/or translated into another
190: language. (Hereinafter, translation is included without limitation in
191: the term ``modification''.) Each licensee is addressed as ``you''.
192:
193: Activities other than copying, distribution and modification are not
194: covered by this License; they are outside its scope. The act of
195: running the Program is not restricted, and the output from the Program
196: is covered only if its contents constitute a work based on the
197: Program (independent of having been made by running the Program).
198: Whether that is true depends on what the Program does.
199:
200: @item
201: You may copy and distribute verbatim copies of the Program's
202: source code as you receive it, in any medium, provided that you
203: conspicuously and appropriately publish on each copy an appropriate
204: copyright notice and disclaimer of warranty; keep intact all the
205: notices that refer to this License and to the absence of any warranty;
206: and give any other recipients of the Program a copy of this License
207: along with the Program.
208:
209: You may charge a fee for the physical act of transferring a copy, and
210: you may at your option offer warranty protection in exchange for a fee.
211:
212: @item
213: You may modify your copy or copies of the Program or any portion
214: of it, thus forming a work based on the Program, and copy and
215: distribute such modifications or work under the terms of Section 1
216: above, provided that you also meet all of these conditions:
217:
218: @enumerate a
219: @item
220: You must cause the modified files to carry prominent notices
221: stating that you changed the files and the date of any change.
222:
223: @item
224: You must cause any work that you distribute or publish, that in
225: whole or in part contains or is derived from the Program or any
226: part thereof, to be licensed as a whole at no charge to all third
227: parties under the terms of this License.
228:
229: @item
230: If the modified program normally reads commands interactively
231: when run, you must cause it, when started running for such
232: interactive use in the most ordinary way, to print or display an
233: announcement including an appropriate copyright notice and a
234: notice that there is no warranty (or else, saying that you provide
235: a warranty) and that users may redistribute the program under
236: these conditions, and telling the user how to view a copy of this
237: License. (Exception: if the Program itself is interactive but
238: does not normally print such an announcement, your work based on
239: the Program is not required to print an announcement.)
240: @end enumerate
241:
242: These requirements apply to the modified work as a whole. If
243: identifiable sections of that work are not derived from the Program,
244: and can be reasonably considered independent and separate works in
245: themselves, then this License, and its terms, do not apply to those
246: sections when you distribute them as separate works. But when you
247: distribute the same sections as part of a whole which is a work based
248: on the Program, the distribution of the whole must be on the terms of
249: this License, whose permissions for other licensees extend to the
250: entire whole, and thus to each and every part regardless of who wrote it.
251:
252: Thus, it is not the intent of this section to claim rights or contest
253: your rights to work written entirely by you; rather, the intent is to
254: exercise the right to control the distribution of derivative or
255: collective works based on the Program.
256:
257: In addition, mere aggregation of another work not based on the Program
258: with the Program (or with a work based on the Program) on a volume of
259: a storage or distribution medium does not bring the other work under
260: the scope of this License.
261:
262: @item
263: You may copy and distribute the Program (or a work based on it,
264: under Section 2) in object code or executable form under the terms of
265: Sections 1 and 2 above provided that you also do one of the following:
266:
267: @enumerate a
268: @item
269: Accompany it with the complete corresponding machine-readable
270: source code, which must be distributed under the terms of Sections
271: 1 and 2 above on a medium customarily used for software interchange; or,
272:
273: @item
274: Accompany it with a written offer, valid for at least three
275: years, to give any third party, for a charge no more than your
276: cost of physically performing source distribution, a complete
277: machine-readable copy of the corresponding source code, to be
278: distributed under the terms of Sections 1 and 2 above on a medium
279: customarily used for software interchange; or,
280:
281: @item
282: Accompany it with the information you received as to the offer
283: to distribute corresponding source code. (This alternative is
284: allowed only for noncommercial distribution and only if you
285: received the program in object code or executable form with such
286: an offer, in accord with Subsection b above.)
287: @end enumerate
288:
289: The source code for a work means the preferred form of the work for
290: making modifications to it. For an executable work, complete source
291: code means all the source code for all modules it contains, plus any
292: associated interface definition files, plus the scripts used to
293: control compilation and installation of the executable. However, as a
294: special exception, the source code distributed need not include
295: anything that is normally distributed (in either source or binary
296: form) with the major components (compiler, kernel, and so on) of the
297: operating system on which the executable runs, unless that component
298: itself accompanies the executable.
299:
300: If distribution of executable or object code is made by offering
301: access to copy from a designated place, then offering equivalent
302: access to copy the source code from the same place counts as
303: distribution of the source code, even though third parties are not
304: compelled to copy the source along with the object code.
305:
306: @item
307: You may not copy, modify, sublicense, or distribute the Program
308: except as expressly provided under this License. Any attempt
309: otherwise to copy, modify, sublicense or distribute the Program is
310: void, and will automatically terminate your rights under this License.
311: However, parties who have received copies, or rights, from you under
312: this License will not have their licenses terminated so long as such
313: parties remain in full compliance.
314:
315: @item
316: You are not required to accept this License, since you have not
317: signed it. However, nothing else grants you permission to modify or
318: distribute the Program or its derivative works. These actions are
319: prohibited by law if you do not accept this License. Therefore, by
320: modifying or distributing the Program (or any work based on the
321: Program), you indicate your acceptance of this License to do so, and
322: all its terms and conditions for copying, distributing or modifying
323: the Program or works based on it.
324:
325: @item
326: Each time you redistribute the Program (or any work based on the
327: Program), the recipient automatically receives a license from the
328: original licensor to copy, distribute or modify the Program subject to
329: these terms and conditions. You may not impose any further
330: restrictions on the recipients' exercise of the rights granted herein.
331: You are not responsible for enforcing compliance by third parties to
332: this License.
333:
334: @item
335: If, as a consequence of a court judgment or allegation of patent
336: infringement or for any other reason (not limited to patent issues),
337: conditions are imposed on you (whether by court order, agreement or
338: otherwise) that contradict the conditions of this License, they do not
339: excuse you from the conditions of this License. If you cannot
340: distribute so as to satisfy simultaneously your obligations under this
341: License and any other pertinent obligations, then as a consequence you
342: may not distribute the Program at all. For example, if a patent
343: license would not permit royalty-free redistribution of the Program by
344: all those who receive copies directly or indirectly through you, then
345: the only way you could satisfy both it and this License would be to
346: refrain entirely from distribution of the Program.
347:
348: If any portion of this section is held invalid or unenforceable under
349: any particular circumstance, the balance of the section is intended to
350: apply and the section as a whole is intended to apply in other
351: circumstances.
352:
353: It is not the purpose of this section to induce you to infringe any
354: patents or other property right claims or to contest validity of any
355: such claims; this section has the sole purpose of protecting the
356: integrity of the free software distribution system, which is
357: implemented by public license practices. Many people have made
358: generous contributions to the wide range of software distributed
359: through that system in reliance on consistent application of that
360: system; it is up to the author/donor to decide if he or she is willing
361: to distribute software through any other system and a licensee cannot
362: impose that choice.
363:
364: This section is intended to make thoroughly clear what is believed to
365: be a consequence of the rest of this License.
366:
367: @item
368: If the distribution and/or use of the Program is restricted in
369: certain countries either by patents or by copyrighted interfaces, the
370: original copyright holder who places the Program under this License
371: may add an explicit geographical distribution limitation excluding
372: those countries, so that distribution is permitted only in or among
373: countries not thus excluded. In such case, this License incorporates
374: the limitation as if written in the body of this License.
375:
376: @item
377: The Free Software Foundation may publish revised and/or new versions
378: of the General Public License from time to time. Such new versions will
379: be similar in spirit to the present version, but may differ in detail to
380: address new problems or concerns.
381:
382: Each version is given a distinguishing version number. If the Program
383: specifies a version number of this License which applies to it and ``any
384: later version'', you have the option of following the terms and conditions
385: either of that version or of any later version published by the Free
386: Software Foundation. If the Program does not specify a version number of
387: this License, you may choose any version ever published by the Free Software
388: Foundation.
389:
390: @item
391: If you wish to incorporate parts of the Program into other free
392: programs whose distribution conditions are different, write to the author
393: to ask for permission. For software which is copyrighted by the Free
394: Software Foundation, write to the Free Software Foundation; we sometimes
395: make exceptions for this. Our decision will be guided by the two goals
396: of preserving the free status of all derivatives of our free software and
397: of promoting the sharing and reuse of software generally.
398:
399: @iftex
400: @heading NO WARRANTY
401: @end iftex
402: @ifinfo
403: @center NO WARRANTY
404: @end ifinfo
405:
406: @item
407: BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
408: FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN
409: OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
410: PROVIDE THE PROGRAM ``AS IS'' WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED
411: OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
412: MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS
413: TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE
414: PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING,
415: REPAIR OR CORRECTION.
416:
417: @item
418: IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
419: WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
420: REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
421: INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
422: OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED
423: TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY
424: YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER
425: PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
426: POSSIBILITY OF SUCH DAMAGES.
427: @end enumerate
428:
429: @iftex
430: @heading END OF TERMS AND CONDITIONS
431: @end iftex
432: @ifinfo
433: @center END OF TERMS AND CONDITIONS
434: @end ifinfo
435:
436: @page
437: @unnumberedsec How to Apply These Terms to Your New Programs
438:
439: If you develop a new program, and you want it to be of the greatest
440: possible use to the public, the best way to achieve this is to make it
441: free software which everyone can redistribute and change under these terms.
442:
443: To do so, attach the following notices to the program. It is safest
444: to attach them to the start of each source file to most effectively
445: convey the exclusion of warranty; and each file should have at least
446: the ``copyright'' line and a pointer to where the full notice is found.
447:
448: @smallexample
449: @var{one line to give the program's name and a brief idea of what it does.}
450: Copyright (C) 19@var{yy} @var{name of author}
451:
452: This program is free software; you can redistribute it and/or modify
453: it under the terms of the GNU General Public License as published by
454: the Free Software Foundation; either version 2 of the License, or
455: (at your option) any later version.
456:
457: This program is distributed in the hope that it will be useful,
458: but WITHOUT ANY WARRANTY; without even the implied warranty of
459: MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
460: GNU General Public License for more details.
461:
462: You should have received a copy of the GNU General Public License
463: along with this program; if not, write to the Free Software
464: Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
465: @end smallexample
466:
467: Also add information on how to contact you by electronic and paper mail.
468:
469: If the program is interactive, make it output a short notice like this
470: when it starts in an interactive mode:
471:
472: @smallexample
473: Gnomovision version 69, Copyright (C) 19@var{yy} @var{name of author}
474: Gnomovision comes with ABSOLUTELY NO WARRANTY; for details
475: type `show w'.
476: This is free software, and you are welcome to redistribute it
477: under certain conditions; type `show c' for details.
478: @end smallexample
479:
480: The hypothetical commands @samp{show w} and @samp{show c} should show
481: the appropriate parts of the General Public License. Of course, the
482: commands you use may be called something other than @samp{show w} and
483: @samp{show c}; they could even be mouse-clicks or menu items---whatever
484: suits your program.
485:
486: You should also get your employer (if you work as a programmer) or your
487: school, if any, to sign a ``copyright disclaimer'' for the program, if
488: necessary. Here is a sample; alter the names:
489:
490: @smallexample
491: Yoyodyne, Inc., hereby disclaims all copyright interest in the program
492: `Gnomovision' (which makes passes at compilers) written by James Hacker.
493:
494: @var{signature of Ty Coon}, 1 April 1989
495: Ty Coon, President of Vice
496: @end smallexample
497:
498: This General Public License does not permit incorporating your program into
499: proprietary programs. If your program is a subroutine library, you may
500: consider it more useful to permit linking proprietary applications with the
501: library. If this is what you want to do, use the GNU Library General
502: Public License instead of this License.
503:
504: @iftex
505: @unnumbered Preface
506: @cindex Preface
507: This manual documents Gforth. The reader is expected to know
508: Forth. This manual is primarily a reference manual. @xref{Other Books}
509: for introductory material.
510: @end iftex
511:
512: @node Goals, Other Books, License, Top
513: @comment node-name, next, previous, up
514: @chapter Goals of Gforth
515: @cindex Goals
516: The goal of the Gforth Project is to develop a standard model for
517: ANS Forth. This can be split into several subgoals:
518:
519: @itemize @bullet
520: @item
521: Gforth should conform to the Forth standard (ANS Forth).
522: @item
523: It should be a model, i.e. it should define all the
524: implementation-dependent things.
525: @item
526: It should become standard, i.e. widely accepted and used. This goal
527: is the most difficult one.
528: @end itemize
529:
530: To achieve these goals Gforth should be
531: @itemize @bullet
532: @item
533: Similar to previous models (fig-Forth, F83)
534: @item
535: Powerful. It should provide for all the things that are considered
536: necessary today and even some that are not yet considered necessary.
537: @item
538: Efficient. It should not get the reputation of being exceptionally
539: slow.
540: @item
541: Free.
542: @item
543: Available on many machines/easy to port.
544: @end itemize
545:
546: Have we achieved these goals? Gforth conforms to the ANS Forth
547: standard. It may be considered a model, but we have not yet documented
548: which parts of the model are stable and which parts we are likely to
549: change. It certainly has not yet become a de facto standard. It has some
550: similarities and some differences to previous models. It has some
551: powerful features, but not yet everything that we envisioned. We
552: certainly have achieved our execution speed goals (@pxref{Performance}).
553: It is free and available on many machines.
554:
555: @node Other Books, Invoking Gforth, Goals, Top
556: @chapter Other books on ANS Forth
557: @cindex books on Forth
558:
559: As the standard is relatively new, there are not many books out yet. It
560: is not recommended to learn Forth by using Gforth and a book that is
561: not written for ANS Forth, as you will not know your mistakes from the
562: deviations of the book.
563:
564: @cindex standard document for ANS Forth
565: @cindex ANS Forth document
566: There is, of course, the standard, the definite reference if you want to
567: write ANS Forth programs. It is available in printed form from the
568: National Standards Institute Sales Department (Tel.: USA (212) 642-4900;
569: Fax.: USA (212) 302-1286) as document @cite{X3.215-1994} for about $200. You
570: can also get it from Global Engineering Documents (Tel.: USA (800)
571: 854-7179; Fax.: (303) 843-9880) for about $300.
572:
573: @cite{dpANS6}, the last draft of the standard, which was then submitted to ANSI
574: for publication is available electronically and for free in some MS Word
575: format, and it has been converted to HTML. Some pointers to these
576: versions can be found through
577: @*@url{http://www.complang.tuwien.ac.at/projects/forth.html}.
578:
579: @cindex introductory book
580: @cindex book, introductory
581: @cindex Woehr, Jack: @cite{Forth: The New Model}
582: @cindex @cite{Forth: The new model} (book)
583: @cite{Forth: The New Model} by Jack Woehr (Prentice-Hall, 1993) is an
584: introductory book based on a draft version of the standard. It does not
585: cover the whole standard. It also contains interesting background
586: information (Jack Woehr was in the ANS Forth Technical Committee). It is
587: not appropriate for complete newbies, but programmers experienced in
588: other languages should find it ok.
589:
590: @node Invoking Gforth, Words, Other Books, Top
591: @chapter Invoking Gforth
592: @cindex invoking Gforth
593: @cindex running Gforth
594: @cindex command-line options
595: @cindex options on the command line
596: @cindex flags on the command line
597:
598: You will usually just say @code{gforth}. In many other cases the default
599: Gforth image will be invoked like this:
600:
601: @example
602: gforth [files] [-e forth-code]
603: @end example
604:
605: executing the contents of the files and the Forth code in the order they
606: are given.
607:
608: In general, the command line looks like this:
609:
610: @example
611: gforth [initialization options] [image-specific options]
612: @end example
613:
614: The initialization options must come before the rest of the command
615: line. They are:
616:
617: @table @code
618: @cindex -i, command-line option
619: @cindex --image-file, command-line option
620: @item --image-file @var{file}
621: @itemx -i @var{file}
622: Loads the Forth image @var{file} instead of the default
623: @file{gforth.fi} (@pxref{Image Files}).
624:
625: @cindex --path, command-line option
626: @cindex -p, command-line option
627: @item --path @var{path}
628: @itemx -p @var{path}
629: Uses @var{path} for searching the image file and Forth source code files
630: instead of the default in the environment variable @code{GFORTHPATH} or
631: the path specified at installation time (e.g.,
632: @file{/usr/local/share/gforth/0.2.0:.}). A path is given as a list of
633: directories, separated by @samp{:} (on Unix) or @samp{;} (on other OSs).
634:
635: @cindex --dictionary-size, command-line option
636: @cindex -m, command-line option
637: @cindex @var{size} parameters for command-line options
638: @cindex size of the dictionary and the stacks
639: @item --dictionary-size @var{size}
640: @itemx -m @var{size}
641: Allocate @var{size} space for the Forth dictionary space instead of
642: using the default specified in the image (typically 256K). The
643: @var{size} specification consists of an integer and a unit (e.g.,
644: @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
645: size, in this case Cells), @code{k} (kilobytes), and @code{M}
646: (Megabytes). If no unit is specified, @code{e} is used.
647:
648: @cindex --data-stack-size, command-line option
649: @cindex -d, command-line option
650: @item --data-stack-size @var{size}
651: @itemx -d @var{size}
652: Allocate @var{size} space for the data stack instead of using the
653: default specified in the image (typically 16K).
654:
655: @cindex --return-stack-size, command-line option
656: @cindex -r, command-line option
657: @item --return-stack-size @var{size}
658: @itemx -r @var{size}
659: Allocate @var{size} space for the return stack instead of using the
660: default specified in the image (typically 15K).
661:
662: @cindex --fp-stack-size, command-line option
663: @cindex -f, command-line option
664: @item --fp-stack-size @var{size}
665: @itemx -f @var{size}
666: Allocate @var{size} space for the floating point stack instead of
667: using the default specified in the image (typically 15.5K). In this case
668: the unit specifier @code{e} refers to floating point numbers.
669:
670: @cindex --locals-stack-size, command-line option
671: @cindex -l, command-line option
672: @item --locals-stack-size @var{size}
673: @itemx -l @var{size}
674: Allocate @var{size} space for the locals stack instead of using the
675: default specified in the image (typically 14.5K).
676:
677: @cindex -h, command-line option
678: @cindex --help, command-line option
679: @item --help
680: @itemx -h
681: Print a message about the command-line options
682:
683: @cindex -v, command-line option
684: @cindex --version, command-line option
685: @item --version
686: @itemx -v
687: Print version and exit
688:
689: @cindex --debug, command-line option
690: @item --debug
691: Print some information useful for debugging on startup.
692:
693: @cindex --offset-image, command-line option
694: @item --offset-image
695: Start the dictionary at a slightly different position than would be used
696: otherwise (useful for creating data-relocatable images,
697: @pxref{Data-Relocatable Image Files}).
698:
699: @cindex --no-offset-im, command-line option
700: @item --no-offset-im
701: Start the dictionary at the normal position.
702:
703: @cindex --clear-dictionary, command-line option
704: @item --clear-dictionary
705: Initialize all bytes in the dictionary to 0 before loading the image
706: (@pxref{Data-Relocatable Image Files}).
707:
708: @cindex --die-on-signal, command-line-option
709: @item --die-on-signal
710: Normally Gforth handles most signals (e.g., the user interrupt SIGINT,
711: or the segmentation violation SIGSEGV) by translating it into a Forth
712: @code{THROW}. With this option, Gforth exits if it receives such a
713: signal. This option is useful when the engine and/or the image might be
714: severely broken (such that it causes another signal before recovering
715: from the first); this option avoids endless loops in such cases.
716: @end table
717:
718: @cindex loading files at startup
719: @cindex executing code on startup
720: @cindex batch processing with Gforth
721: As explained above, the image-specific command-line arguments for the
722: default image @file{gforth.fi} consist of a sequence of filenames and
723: @code{-e @var{forth-code}} options that are interpreted in the sequence
724: in which they are given. The @code{-e @var{forth-code}} or
725: @code{--evaluate @var{forth-code}} option evaluates the forth
726: code. This option takes only one argument; if you want to evaluate more
727: Forth words, you have to quote them or use several @code{-e}s. To exit
728: after processing the command line (instead of entering interactive mode)
729: append @code{-e bye} to the command line.
730:
731: @cindex versions, invoking other versions of Gforth
732: If you have several versions of Gforth installed, @code{gforth} will
733: invoke the version that was installed last. @code{gforth-@var{version}}
734: invokes a specific version. You may want to use the option
735: @code{--path}, if your environment contains the variable
736: @code{GFORTHPATH}.
737:
738: Not yet implemented:
739: On startup the system first executes the system initialization file
740: (unless the option @code{--no-init-file} is given; note that the system
741: resulting from using this option may not be ANS Forth conformant). Then
742: the user initialization file @file{.gforth.fs} is executed, unless the
743: option @code{--no-rc} is given; this file is first searched in @file{.},
744: then in @file{~}, then in the normal path (see above).
745:
746: @node Words, Tools, Invoking Gforth, Top
747: @chapter Forth Words
748: @cindex Words
749:
750: @menu
751: * Notation::
752: * Arithmetic::
753: * Stack Manipulation::
754: * Memory::
755: * Control Structures::
756: * Locals::
757: * Defining Words::
758: * Structures::
759: * Objects::
760: * Object Oriented Forth::
761: * Mini-OOF::
762: * Tokens for Words::
763: * Wordlists::
764: * Files::
765: * Blocks::
766: * Other I/O::
767: * Programming Tools::
768: * Assembler and Code Words::
769: * Threading Words::
770: * Including Files::
771: @end menu
772:
773: @node Notation, Arithmetic, Words, Words
774: @section Notation
775: @cindex notation of glossary entries
776: @cindex format of glossary entries
777: @cindex glossary notation format
778: @cindex word glossary entry format
779:
780: The Forth words are described in this section in the glossary notation
781: that has become a de-facto standard for Forth texts, i.e.,
782:
783: @format
784: @var{word} @var{Stack effect} @var{wordset} @var{pronunciation}
785: @end format
786: @var{Description}
787:
788: @table @var
789: @item word
790: @cindex case insensitivity
791: The name of the word. BTW, Gforth is case insensitive, so you can
792: type the words in in lower case (However, @pxref{core-idef}).
793:
794: @item Stack effect
795: @cindex stack effect
796: The stack effect is written in the notation @code{@var{before} --
797: @var{after}}, where @var{before} and @var{after} describe the top of
798: stack entries before and after the execution of the word. The rest of
799: the stack is not touched by the word. The top of stack is rightmost,
800: i.e., a stack sequence is written as it is typed in. Note that Gforth
801: uses a separate floating point stack, but a unified stack
802: notation. Also, return stack effects are not shown in @var{stack
803: effect}, but in @var{Description}. The name of a stack item describes
804: the type and/or the function of the item. See below for a discussion of
805: the types.
806:
807: All words have two stack effects: A compile-time stack effect and a
808: run-time stack effect. The compile-time stack-effect of most words is
809: @var{ -- }. If the compile-time stack-effect of a word deviates from
810: this standard behaviour, or the word does other unusual things at
811: compile time, both stack effects are shown; otherwise only the run-time
812: stack effect is shown.
813:
814: @cindex pronounciation of words
815: @item pronunciation
816: How the word is pronounced.
817:
818: @cindex wordset
819: @item wordset
820: The ANS Forth standard is divided into several wordsets. A standard
821: system need not support all of them. So, the fewer wordsets your program
822: uses the more portable it will be in theory. However, we suspect that
823: most ANS Forth systems on personal machines will feature all
824: wordsets. Words that are not defined in the ANS standard have
825: @code{gforth} or @code{gforth-internal} as wordset. @code{gforth}
826: describes words that will work in future releases of Gforth;
827: @code{gforth-internal} words are more volatile. Environmental query
828: strings are also displayed like words; you can recognize them by the
829: @code{environment} in the wordset field.
830:
831: @item Description
832: A description of the behaviour of the word.
833: @end table
834:
835: @cindex types of stack items
836: @cindex stack item types
837: The type of a stack item is specified by the character(s) the name
838: starts with:
839:
840: @table @code
841: @item f
842: @cindex @code{f}, stack item type
843: Boolean flags, i.e. @code{false} or @code{true}.
844: @item c
845: @cindex @code{c}, stack item type
846: Char
847: @item w
848: @cindex @code{w}, stack item type
849: Cell, can contain an integer or an address
850: @item n
851: @cindex @code{n}, stack item type
852: signed integer
853: @item u
854: @cindex @code{u}, stack item type
855: unsigned integer
856: @item d
857: @cindex @code{d}, stack item type
858: double sized signed integer
859: @item ud
860: @cindex @code{ud}, stack item type
861: double sized unsigned integer
862: @item r
863: @cindex @code{r}, stack item type
864: Float (on the FP stack)
865: @item a_
866: @cindex @code{a_}, stack item type
867: Cell-aligned address
868: @item c_
869: @cindex @code{c_}, stack item type
870: Char-aligned address (note that a Char may have two bytes in Windows NT)
871: @item f_
872: @cindex @code{f_}, stack item type
873: Float-aligned address
874: @item df_
875: @cindex @code{df_}, stack item type
876: Address aligned for IEEE double precision float
877: @item sf_
878: @cindex @code{sf_}, stack item type
879: Address aligned for IEEE single precision float
880: @item xt
881: @cindex @code{xt}, stack item type
882: Execution token, same size as Cell
883: @item wid
884: @cindex @code{wid}, stack item type
885: Wordlist ID, same size as Cell
886: @item f83name
887: @cindex @code{f83name}, stack item type
888: Pointer to a name structure
889: @item "
890: @cindex @code{"}, stack item type
891: string in the input stream (not the stack). The terminating character is
892: a blank by default. If it is not a blank, it is shown in @code{<>}
893: quotes.
894: @end table
895:
896: @node Arithmetic, Stack Manipulation, Notation, Words
897: @section Arithmetic
898: @cindex arithmetic words
899:
900: @cindex division with potentially negative operands
901: Forth arithmetic is not checked, i.e., you will not hear about integer
902: overflow on addition or multiplication, you may hear about division by
903: zero if you are lucky. The operator is written after the operands, but
904: the operands are still in the original order. I.e., the infix @code{2-1}
905: corresponds to @code{2 1 -}. Forth offers a variety of division
906: operators. If you perform division with potentially negative operands,
907: you do not want to use @code{/} or @code{/mod} with its undefined
908: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
909: former, @pxref{Mixed precision}).
910:
911: @menu
912: * Single precision::
913: * Bitwise operations::
914: * Mixed precision:: operations with single and double-cell integers
915: * Double precision:: Double-cell integer arithmetic
916: * Floating Point::
917: @end menu
918:
919: @node Single precision, Bitwise operations, Arithmetic, Arithmetic
920: @subsection Single precision
921: @cindex single precision arithmetic words
922:
923: doc-+
924: doc--
925: doc-*
926: doc-/
927: doc-mod
928: doc-/mod
929: doc-negate
930: doc-abs
931: doc-min
932: doc-max
933:
934: @node Bitwise operations, Mixed precision, Single precision, Arithmetic
935: @subsection Bitwise operations
936: @cindex bitwise operation words
937:
938: doc-and
939: doc-or
940: doc-xor
941: doc-invert
942: doc-2*
943: doc-2/
944:
945: @node Mixed precision, Double precision, Bitwise operations, Arithmetic
946: @subsection Mixed precision
947: @cindex mixed precision arithmetic words
948:
949: doc-m+
950: doc-*/
951: doc-*/mod
952: doc-m*
953: doc-um*
954: doc-m*/
955: doc-um/mod
956: doc-fm/mod
957: doc-sm/rem
958:
959: @node Double precision, Floating Point, Mixed precision, Arithmetic
960: @subsection Double precision
961: @cindex double precision arithmetic words
962:
963: @cindex double-cell numbers, input format
964: @cindex input format for double-cell numbers
965: The outer (aka text) interpreter converts numbers containing a dot into
966: a double precision number. Note that only numbers with the dot as last
967: character are standard-conforming.
968:
969: doc-d+
970: doc-d-
971: doc-dnegate
972: doc-dabs
973: doc-dmin
974: doc-dmax
975:
976: @node Floating Point, , Double precision, Arithmetic
977: @subsection Floating Point
978: @cindex floating point arithmetic words
979:
980: @cindex floating-point numbers, input format
981: @cindex input format for floating-point numbers
982: The format of floating point numbers recognized by the outer (aka text)
983: interpreter is: a signed decimal number, possibly containing a decimal
984: point (@code{.}), followed by @code{E} or @code{e}, optionally followed
985: by a signed integer (the exponent). E.g., @code{1e} is the same as
986: @code{+1.0e+0}. Note that a number without @code{e}
987: is not interpreted as floating-point number, but as double (if the
988: number contains a @code{.}) or single precision integer. Also,
989: conversions between string and floating point numbers always use base
990: 10, irrespective of the value of @code{BASE}. If @code{BASE} contains a
991: value greater then 14, the @code{E} may be interpreted as digit and the
992: number will be interpreted as integer, unless it has a signed exponent
993: (both @code{+} and @code{-} are allowed as signs).
994:
995: @cindex angles in trigonometric operations
996: @cindex trigonometric operations
997: Angles in floating point operations are given in radians (a full circle
998: has 2 pi radians). Note, that Gforth has a separate floating point
999: stack, but we use the unified notation.
1000:
1001: @cindex floating-point arithmetic, pitfalls
1002: Floating point numbers have a number of unpleasant surprises for the
1003: unwary (e.g., floating point addition is not associative) and even a few
1004: for the wary. You should not use them unless you know what you are doing
1005: or you don't care that the results you get are totally bogus. If you
1006: want to learn about the problems of floating point numbers (and how to
1007: avoid them), you might start with @cite{David Goldberg, What Every
1008: Computer Scientist Should Know About Floating-Point Arithmetic, ACM
1009: Computing Surveys 23(1):5@minus{}48, March 1991}.
1010:
1011: doc-f+
1012: doc-f-
1013: doc-f*
1014: doc-f/
1015: doc-fnegate
1016: doc-fabs
1017: doc-fmax
1018: doc-fmin
1019: doc-floor
1020: doc-fround
1021: doc-f**
1022: doc-fsqrt
1023: doc-fexp
1024: doc-fexpm1
1025: doc-fln
1026: doc-flnp1
1027: doc-flog
1028: doc-falog
1029: doc-fsin
1030: doc-fcos
1031: doc-fsincos
1032: doc-ftan
1033: doc-fasin
1034: doc-facos
1035: doc-fatan
1036: doc-fatan2
1037: doc-fsinh
1038: doc-fcosh
1039: doc-ftanh
1040: doc-fasinh
1041: doc-facosh
1042: doc-fatanh
1043:
1044: @node Stack Manipulation, Memory, Arithmetic, Words
1045: @section Stack Manipulation
1046: @cindex stack manipulation words
1047:
1048: @cindex floating-point stack in the standard
1049: Gforth has a data stack (aka parameter stack) for characters, cells,
1050: addresses, and double cells, a floating point stack for floating point
1051: numbers, a return stack for storing the return addresses of colon
1052: definitions and other data, and a locals stack for storing local
1053: variables. Note that while every sane Forth has a separate floating
1054: point stack, this is not strictly required; an ANS Forth system could
1055: theoretically keep floating point numbers on the data stack. As an
1056: additional difficulty, you don't know how many cells a floating point
1057: number takes. It is reportedly possible to write words in a way that
1058: they work also for a unified stack model, but we do not recommend trying
1059: it. Instead, just say that your program has an environmental dependency
1060: on a separate FP stack.
1061:
1062: @cindex return stack and locals
1063: @cindex locals and return stack
1064: Also, a Forth system is allowed to keep the local variables on the
1065: return stack. This is reasonable, as local variables usually eliminate
1066: the need to use the return stack explicitly. So, if you want to produce
1067: a standard complying program and if you are using local variables in a
1068: word, forget about return stack manipulations in that word (see the
1069: standard document for the exact rules).
1070:
1071: @menu
1072: * Data stack::
1073: * Floating point stack::
1074: * Return stack::
1075: * Locals stack::
1076: * Stack pointer manipulation::
1077: @end menu
1078:
1079: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
1080: @subsection Data stack
1081: @cindex data stack manipulation words
1082: @cindex stack manipulations words, data stack
1083:
1084: doc-drop
1085: doc-nip
1086: doc-dup
1087: doc-over
1088: doc-tuck
1089: doc-swap
1090: doc-rot
1091: doc--rot
1092: doc-?dup
1093: doc-pick
1094: doc-roll
1095: doc-2drop
1096: doc-2nip
1097: doc-2dup
1098: doc-2over
1099: doc-2tuck
1100: doc-2swap
1101: doc-2rot
1102:
1103: @node Floating point stack, Return stack, Data stack, Stack Manipulation
1104: @subsection Floating point stack
1105: @cindex floating-point stack manipulation words
1106: @cindex stack manipulation words, floating-point stack
1107:
1108: doc-fdrop
1109: doc-fnip
1110: doc-fdup
1111: doc-fover
1112: doc-ftuck
1113: doc-fswap
1114: doc-frot
1115:
1116: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
1117: @subsection Return stack
1118: @cindex return stack manipulation words
1119: @cindex stack manipulation words, return stack
1120:
1121: doc->r
1122: doc-r>
1123: doc-r@
1124: doc-rdrop
1125: doc-2>r
1126: doc-2r>
1127: doc-2r@
1128: doc-2rdrop
1129:
1130: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
1131: @subsection Locals stack
1132:
1133: @node Stack pointer manipulation, , Locals stack, Stack Manipulation
1134: @subsection Stack pointer manipulation
1135: @cindex stack pointer manipulation words
1136:
1137: doc-sp@
1138: doc-sp!
1139: doc-fp@
1140: doc-fp!
1141: doc-rp@
1142: doc-rp!
1143: doc-lp@
1144: doc-lp!
1145:
1146: @node Memory, Control Structures, Stack Manipulation, Words
1147: @section Memory
1148: @cindex Memory words
1149:
1150: @menu
1151: * Memory Access::
1152: * Address arithmetic::
1153: * Memory Blocks::
1154: @end menu
1155:
1156: @node Memory Access, Address arithmetic, Memory, Memory
1157: @subsection Memory Access
1158: @cindex memory access words
1159:
1160: doc-@
1161: doc-!
1162: doc-+!
1163: doc-c@
1164: doc-c!
1165: doc-2@
1166: doc-2!
1167: doc-f@
1168: doc-f!
1169: doc-sf@
1170: doc-sf!
1171: doc-df@
1172: doc-df!
1173:
1174: @node Address arithmetic, Memory Blocks, Memory Access, Memory
1175: @subsection Address arithmetic
1176: @cindex address arithmetic words
1177:
1178: ANS Forth does not specify the sizes of the data types. Instead, it
1179: offers a number of words for computing sizes and doing address
1180: arithmetic. Basically, address arithmetic is performed in terms of
1181: address units (aus); on most systems the address unit is one byte. Note
1182: that a character may have more than one au, so @code{chars} is no noop
1183: (on systems where it is a noop, it compiles to nothing).
1184:
1185: @cindex alignment of addresses for types
1186: ANS Forth also defines words for aligning addresses for specific
1187: types. Many computers require that accesses to specific data types
1188: must only occur at specific addresses; e.g., that cells may only be
1189: accessed at addresses divisible by 4. Even if a machine allows unaligned
1190: accesses, it can usually perform aligned accesses faster.
1191:
1192: For the performance-conscious: alignment operations are usually only
1193: necessary during the definition of a data structure, not during the
1194: (more frequent) accesses to it.
1195:
1196: ANS Forth defines no words for character-aligning addresses. This is not
1197: an oversight, but reflects the fact that addresses that are not
1198: char-aligned have no use in the standard and therefore will not be
1199: created.
1200:
1201: @cindex @code{CREATE} and alignment
1202: The standard guarantees that addresses returned by @code{CREATE}d words
1203: are cell-aligned; in addition, Gforth guarantees that these addresses
1204: are aligned for all purposes.
1205:
1206: Note that the standard defines a word @code{char}, which has nothing to
1207: do with address arithmetic.
1208:
1209: doc-chars
1210: doc-char+
1211: doc-cells
1212: doc-cell+
1213: doc-cell
1214: doc-align
1215: doc-aligned
1216: doc-floats
1217: doc-float+
1218: doc-float
1219: doc-falign
1220: doc-faligned
1221: doc-sfloats
1222: doc-sfloat+
1223: doc-sfalign
1224: doc-sfaligned
1225: doc-dfloats
1226: doc-dfloat+
1227: doc-dfalign
1228: doc-dfaligned
1229: doc-maxalign
1230: doc-maxaligned
1231: doc-cfalign
1232: doc-cfaligned
1233: doc-address-unit-bits
1234:
1235: @node Memory Blocks, , Address arithmetic, Memory
1236: @subsection Memory Blocks
1237: @cindex memory block words
1238:
1239: doc-move
1240: doc-erase
1241:
1242: While the previous words work on address units, the rest works on
1243: characters.
1244:
1245: doc-cmove
1246: doc-cmove>
1247: doc-fill
1248: doc-blank
1249:
1250: @node Control Structures, Locals, Memory, Words
1251: @section Control Structures
1252: @cindex control structures
1253:
1254: Control structures in Forth cannot be used in interpret state, only in
1255: compile state@footnote{More precisely, they have no interpretation
1256: semantics (@pxref{Interpretation and Compilation Semantics})}, i.e., in
1257: a colon definition. We do not like this limitation, but have not seen a
1258: satisfying way around it yet, although many schemes have been proposed.
1259:
1260: @menu
1261: * Selection::
1262: * Simple Loops::
1263: * Counted Loops::
1264: * Arbitrary control structures::
1265: * Calls and returns::
1266: * Exception Handling::
1267: @end menu
1268:
1269: @node Selection, Simple Loops, Control Structures, Control Structures
1270: @subsection Selection
1271: @cindex selection control structures
1272: @cindex control structures for selection
1273:
1274: @cindex @code{IF} control structure
1275: @example
1276: @var{flag}
1277: IF
1278: @var{code}
1279: ENDIF
1280: @end example
1281: or
1282: @example
1283: @var{flag}
1284: IF
1285: @var{code1}
1286: ELSE
1287: @var{code2}
1288: ENDIF
1289: @end example
1290:
1291: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
1292: standard, and @code{ENDIF} is not, although it is quite popular. We
1293: recommend using @code{ENDIF}, because it is less confusing for people
1294: who also know other languages (and is not prone to reinforcing negative
1295: prejudices against Forth in these people). Adding @code{ENDIF} to a
1296: system that only supplies @code{THEN} is simple:
1297: @example
1298: : endif POSTPONE then ; immediate
1299: @end example
1300:
1301: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
1302: (adv.)} has the following meanings:
1303: @quotation
1304: ... 2b: following next after in order ... 3d: as a necessary consequence
1305: (if you were there, then you saw them).
1306: @end quotation
1307: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
1308: and many other programming languages has the meaning 3d.]
1309:
1310: Gforth also provides the words @code{?dup-if} and @code{?dup-0=-if}, so
1311: you can avoid using @code{?dup}. Using these alternatives is also more
1312: efficient than using @code{?dup}. Definitions in plain standard Forth
1313: for @code{ENDIF}, @code{?DUP-IF} and @code{?DUP-0=-IF} are provided in
1314: @file{compat/control.fs}.
1315:
1316: @cindex @code{CASE} control structure
1317: @example
1318: @var{n}
1319: CASE
1320: @var{n1} OF @var{code1} ENDOF
1321: @var{n2} OF @var{code2} ENDOF
1322: @dots{}
1323: ENDCASE
1324: @end example
1325:
1326: Executes the first @var{codei}, where the @var{ni} is equal to
1327: @var{n}. A default case can be added by simply writing the code after
1328: the last @code{ENDOF}. It may use @var{n}, which is on top of the stack,
1329: but must not consume it.
1330:
1331: @node Simple Loops, Counted Loops, Selection, Control Structures
1332: @subsection Simple Loops
1333: @cindex simple loops
1334: @cindex loops without count
1335:
1336: @cindex @code{WHILE} loop
1337: @example
1338: BEGIN
1339: @var{code1}
1340: @var{flag}
1341: WHILE
1342: @var{code2}
1343: REPEAT
1344: @end example
1345:
1346: @var{code1} is executed and @var{flag} is computed. If it is true,
1347: @var{code2} is executed and the loop is restarted; If @var{flag} is
1348: false, execution continues after the @code{REPEAT}.
1349:
1350: @cindex @code{UNTIL} loop
1351: @example
1352: BEGIN
1353: @var{code}
1354: @var{flag}
1355: UNTIL
1356: @end example
1357:
1358: @var{code} is executed. The loop is restarted if @code{flag} is false.
1359:
1360: @cindex endless loop
1361: @cindex loops, endless
1362: @example
1363: BEGIN
1364: @var{code}
1365: AGAIN
1366: @end example
1367:
1368: This is an endless loop.
1369:
1370: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
1371: @subsection Counted Loops
1372: @cindex counted loops
1373: @cindex loops, counted
1374: @cindex @code{DO} loops
1375:
1376: The basic counted loop is:
1377: @example
1378: @var{limit} @var{start}
1379: ?DO
1380: @var{body}
1381: LOOP
1382: @end example
1383:
1384: This performs one iteration for every integer, starting from @var{start}
1385: and up to, but excluding @var{limit}. The counter, aka index, can be
1386: accessed with @code{i}. E.g., the loop
1387: @example
1388: 10 0 ?DO
1389: i .
1390: LOOP
1391: @end example
1392: prints
1393: @example
1394: 0 1 2 3 4 5 6 7 8 9
1395: @end example
1396: The index of the innermost loop can be accessed with @code{i}, the index
1397: of the next loop with @code{j}, and the index of the third loop with
1398: @code{k}.
1399:
1400: doc-i
1401: doc-j
1402: doc-k
1403:
1404: The loop control data are kept on the return stack, so there are some
1405: restrictions on mixing return stack accesses and counted loop
1406: words. E.g., if you put values on the return stack outside the loop, you
1407: cannot read them inside the loop. If you put values on the return stack
1408: within a loop, you have to remove them before the end of the loop and
1409: before accessing the index of the loop.
1410:
1411: There are several variations on the counted loop:
1412:
1413: @code{LEAVE} leaves the innermost counted loop immediately.
1414:
1415: If @var{start} is greater than @var{limit}, a @code{?DO} loop is entered
1416: (and @code{LOOP} iterates until they become equal by wrap-around
1417: arithmetic). This behaviour is usually not what you want. Therefore,
1418: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
1419: @code{?DO}), which do not enter the loop if @var{start} is greater than
1420: @var{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
1421: unsigned loop parameters.
1422:
1423: @code{LOOP} can be replaced with @code{@var{n} +LOOP}; this updates the
1424: index by @var{n} instead of by 1. The loop is terminated when the border
1425: between @var{limit-1} and @var{limit} is crossed. E.g.:
1426:
1427: @code{4 0 +DO i . 2 +LOOP} prints @code{0 2}
1428:
1429: @code{4 1 +DO i . 2 +LOOP} prints @code{1 3}
1430:
1431: @cindex negative increment for counted loops
1432: @cindex counted loops with negative increment
1433: The behaviour of @code{@var{n} +LOOP} is peculiar when @var{n} is negative:
1434:
1435: @code{-1 0 ?DO i . -1 +LOOP} prints @code{0 -1}
1436:
1437: @code{ 0 0 ?DO i . -1 +LOOP} prints nothing
1438:
1439: Therefore we recommend avoiding @code{@var{n} +LOOP} with negative
1440: @var{n}. One alternative is @code{@var{u} -LOOP}, which reduces the
1441: index by @var{u} each iteration. The loop is terminated when the border
1442: between @var{limit+1} and @var{limit} is crossed. Gforth also provides
1443: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
1444:
1445: @code{-2 0 -DO i . 1 -LOOP} prints @code{0 -1}
1446:
1447: @code{-1 0 -DO i . 1 -LOOP} prints @code{0}
1448:
1449: @code{ 0 0 -DO i . 1 -LOOP} prints nothing
1450:
1451: Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
1452: @code{-LOOP} are not in the ANS Forth standard. However, an
1453: implementation for these words that uses only standard words is provided
1454: in @file{compat/loops.fs}.
1455:
1456: @code{?DO} can also be replaced by @code{DO}. @code{DO} always enters
1457: the loop, independent of the loop parameters. Do not use @code{DO}, even
1458: if you know that the loop is entered in any case. Such knowledge tends
1459: to become invalid during maintenance of a program, and then the
1460: @code{DO} will make trouble.
1461:
1462: @code{UNLOOP} is used to prepare for an abnormal loop exit, e.g., via
1463: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
1464: return stack so @code{EXIT} can get to its return address.
1465:
1466: @cindex @code{FOR} loops
1467: Another counted loop is
1468: @example
1469: @var{n}
1470: FOR
1471: @var{body}
1472: NEXT
1473: @end example
1474: This is the preferred loop of native code compiler writers who are too
1475: lazy to optimize @code{?DO} loops properly. In Gforth, this loop
1476: iterates @var{n+1} times; @code{i} produces values starting with @var{n}
1477: and ending with 0. Other Forth systems may behave differently, even if
1478: they support @code{FOR} loops. To avoid problems, don't use @code{FOR}
1479: loops.
1480:
1481: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
1482: @subsection Arbitrary control structures
1483: @cindex control structures, user-defined
1484:
1485: @cindex control-flow stack
1486: ANS Forth permits and supports using control structures in a non-nested
1487: way. Information about incomplete control structures is stored on the
1488: control-flow stack. This stack may be implemented on the Forth data
1489: stack, and this is what we have done in Gforth.
1490:
1491: @cindex @code{orig}, control-flow stack item
1492: @cindex @code{dest}, control-flow stack item
1493: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
1494: entry represents a backward branch target. A few words are the basis for
1495: building any control structure possible (except control structures that
1496: need storage, like calls, coroutines, and backtracking).
1497:
1498: doc-if
1499: doc-ahead
1500: doc-then
1501: doc-begin
1502: doc-until
1503: doc-again
1504: doc-cs-pick
1505: doc-cs-roll
1506:
1507: On many systems control-flow stack items take one word, in Gforth they
1508: currently take three (this may change in the future). Therefore it is a
1509: really good idea to manipulate the control flow stack with
1510: @code{cs-pick} and @code{cs-roll}, not with data stack manipulation
1511: words.
1512:
1513: Some standard control structure words are built from these words:
1514:
1515: doc-else
1516: doc-while
1517: doc-repeat
1518:
1519: Gforth adds some more control-structure words:
1520:
1521: doc-endif
1522: doc-?dup-if
1523: doc-?dup-0=-if
1524:
1525: Counted loop words constitute a separate group of words:
1526:
1527: doc-?do
1528: doc-+do
1529: doc-u+do
1530: doc--do
1531: doc-u-do
1532: doc-do
1533: doc-for
1534: doc-loop
1535: doc-+loop
1536: doc--loop
1537: doc-next
1538: doc-leave
1539: doc-?leave
1540: doc-unloop
1541: doc-done
1542:
1543: The standard does not allow using @code{cs-pick} and @code{cs-roll} on
1544: @i{do-sys}. Our system allows it, but it's your job to ensure that for
1545: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
1546: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
1547: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
1548: resolved (by using one of the loop-ending words or @code{DONE}).
1549:
1550: Another group of control structure words are
1551:
1552: doc-case
1553: doc-endcase
1554: doc-of
1555: doc-endof
1556:
1557: @i{case-sys} and @i{of-sys} cannot be processed using @code{cs-pick} and
1558: @code{cs-roll}.
1559:
1560: @subsubsection Programming Style
1561:
1562: In order to ensure readability we recommend that you do not create
1563: arbitrary control structures directly, but define new control structure
1564: words for the control structure you want and use these words in your
1565: program.
1566:
1567: E.g., instead of writing
1568:
1569: @example
1570: begin
1571: ...
1572: if [ 1 cs-roll ]
1573: ...
1574: again then
1575: @end example
1576:
1577: we recommend defining control structure words, e.g.,
1578:
1579: @example
1580: : while ( dest -- orig dest )
1581: POSTPONE if
1582: 1 cs-roll ; immediate
1583:
1584: : repeat ( orig dest -- )
1585: POSTPONE again
1586: POSTPONE then ; immediate
1587: @end example
1588:
1589: and then using these to create the control structure:
1590:
1591: @example
1592: begin
1593: ...
1594: while
1595: ...
1596: repeat
1597: @end example
1598:
1599: That's much easier to read, isn't it? Of course, @code{REPEAT} and
1600: @code{WHILE} are predefined, so in this example it would not be
1601: necessary to define them.
1602:
1603: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
1604: @subsection Calls and returns
1605: @cindex calling a definition
1606: @cindex returning from a definition
1607:
1608: @cindex recursive definitions
1609: A definition can be called simply be writing the name of the definition
1610: to be called. Note that normally a definition is invisible during its
1611: definition. If you want to write a directly recursive definition, you
1612: can use @code{recursive} to make the current definition visible.
1613:
1614: doc-recursive
1615:
1616: Another way to perform a recursive call is
1617:
1618: doc-recurse
1619:
1620: @c @progstyle{
1621: I prefer using @code{recursive} to @code{recurse}, because
1622: calling the definition by name is more descriptive (if the name is
1623: well-chosen) than the somewhat cryptic @code{recurse}. E.g., in a
1624: quicksort implementation, it is much better to read (and think) ``now
1625: sort the partitions'' than to read ``now do a recursive call''.
1626:
1627: For mutual recursion, use @code{defer}red words, like this:
1628:
1629: @example
1630: defer foo
1631:
1632: : bar ( ... -- ... )
1633: ... foo ... ;
1634:
1635: :noname ( ... -- ... )
1636: ... bar ... ;
1637: IS foo
1638: @end example
1639:
1640: When the end of the definition is reached, it returns. An earlier return
1641: can be forced using
1642:
1643: doc-exit
1644:
1645: Don't forget to clean up the return stack and @code{UNLOOP} any
1646: outstanding @code{?DO}...@code{LOOP}s before @code{EXIT}ing. The
1647: primitive compiled by @code{EXIT} is
1648:
1649: doc-;s
1650:
1651: @node Exception Handling, , Calls and returns, Control Structures
1652: @subsection Exception Handling
1653: @cindex Exceptions
1654:
1655: doc-catch
1656: doc-throw
1657:
1658: @node Locals, Defining Words, Control Structures, Words
1659: @section Locals
1660: @cindex locals
1661:
1662: Local variables can make Forth programming more enjoyable and Forth
1663: programs easier to read. Unfortunately, the locals of ANS Forth are
1664: laden with restrictions. Therefore, we provide not only the ANS Forth
1665: locals wordset, but also our own, more powerful locals wordset (we
1666: implemented the ANS Forth locals wordset through our locals wordset).
1667:
1668: The ideas in this section have also been published in the paper
1669: @cite{Automatic Scoping of Local Variables} by M. Anton Ertl, presented
1670: at EuroForth '94; it is available at
1671: @*@url{http://www.complang.tuwien.ac.at/papers/ertl94l.ps.gz}.
1672:
1673: @menu
1674: * Gforth locals::
1675: * ANS Forth locals::
1676: @end menu
1677:
1678: @node Gforth locals, ANS Forth locals, Locals, Locals
1679: @subsection Gforth locals
1680: @cindex Gforth locals
1681: @cindex locals, Gforth style
1682:
1683: Locals can be defined with
1684:
1685: @example
1686: @{ local1 local2 ... -- comment @}
1687: @end example
1688: or
1689: @example
1690: @{ local1 local2 ... @}
1691: @end example
1692:
1693: E.g.,
1694: @example
1695: : max @{ n1 n2 -- n3 @}
1696: n1 n2 > if
1697: n1
1698: else
1699: n2
1700: endif ;
1701: @end example
1702:
1703: The similarity of locals definitions with stack comments is intended. A
1704: locals definition often replaces the stack comment of a word. The order
1705: of the locals corresponds to the order in a stack comment and everything
1706: after the @code{--} is really a comment.
1707:
1708: This similarity has one disadvantage: It is too easy to confuse locals
1709: declarations with stack comments, causing bugs and making them hard to
1710: find. However, this problem can be avoided by appropriate coding
1711: conventions: Do not use both notations in the same program. If you do,
1712: they should be distinguished using additional means, e.g. by position.
1713:
1714: @cindex types of locals
1715: @cindex locals types
1716: The name of the local may be preceded by a type specifier, e.g.,
1717: @code{F:} for a floating point value:
1718:
1719: @example
1720: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
1721: \ complex multiplication
1722: Ar Br f* Ai Bi f* f-
1723: Ar Bi f* Ai Br f* f+ ;
1724: @end example
1725:
1726: @cindex flavours of locals
1727: @cindex locals flavours
1728: @cindex value-flavoured locals
1729: @cindex variable-flavoured locals
1730: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
1731: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
1732: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
1733: with @code{W:}, @code{D:} etc.) produces its value and can be changed
1734: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
1735: produces its address (which becomes invalid when the variable's scope is
1736: left). E.g., the standard word @code{emit} can be defined in terms of
1737: @code{type} like this:
1738:
1739: @example
1740: : emit @{ C^ char* -- @}
1741: char* 1 type ;
1742: @end example
1743:
1744: @cindex default type of locals
1745: @cindex locals, default type
1746: A local without type specifier is a @code{W:} local. Both flavours of
1747: locals are initialized with values from the data or FP stack.
1748:
1749: Currently there is no way to define locals with user-defined data
1750: structures, but we are working on it.
1751:
1752: Gforth allows defining locals everywhere in a colon definition. This
1753: poses the following questions:
1754:
1755: @menu
1756: * Where are locals visible by name?::
1757: * How long do locals live?::
1758: * Programming Style::
1759: * Implementation::
1760: @end menu
1761:
1762: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
1763: @subsubsection Where are locals visible by name?
1764: @cindex locals visibility
1765: @cindex visibility of locals
1766: @cindex scope of locals
1767:
1768: Basically, the answer is that locals are visible where you would expect
1769: it in block-structured languages, and sometimes a little longer. If you
1770: want to restrict the scope of a local, enclose its definition in
1771: @code{SCOPE}...@code{ENDSCOPE}.
1772:
1773: doc-scope
1774: doc-endscope
1775:
1776: These words behave like control structure words, so you can use them
1777: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
1778: arbitrary ways.
1779:
1780: If you want a more exact answer to the visibility question, here's the
1781: basic principle: A local is visible in all places that can only be
1782: reached through the definition of the local@footnote{In compiler
1783: construction terminology, all places dominated by the definition of the
1784: local.}. In other words, it is not visible in places that can be reached
1785: without going through the definition of the local. E.g., locals defined
1786: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
1787: defined in @code{BEGIN}...@code{UNTIL} are visible after the
1788: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
1789:
1790: The reasoning behind this solution is: We want to have the locals
1791: visible as long as it is meaningful. The user can always make the
1792: visibility shorter by using explicit scoping. In a place that can
1793: only be reached through the definition of a local, the meaning of a
1794: local name is clear. In other places it is not: How is the local
1795: initialized at the control flow path that does not contain the
1796: definition? Which local is meant, if the same name is defined twice in
1797: two independent control flow paths?
1798:
1799: This should be enough detail for nearly all users, so you can skip the
1800: rest of this section. If you really must know all the gory details and
1801: options, read on.
1802:
1803: In order to implement this rule, the compiler has to know which places
1804: are unreachable. It knows this automatically after @code{AHEAD},
1805: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
1806: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
1807: compiler that the control flow never reaches that place. If
1808: @code{UNREACHABLE} is not used where it could, the only consequence is
1809: that the visibility of some locals is more limited than the rule above
1810: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
1811: lie to the compiler), buggy code will be produced.
1812:
1813: doc-unreachable
1814:
1815: Another problem with this rule is that at @code{BEGIN}, the compiler
1816: does not know which locals will be visible on the incoming
1817: back-edge. All problems discussed in the following are due to this
1818: ignorance of the compiler (we discuss the problems using @code{BEGIN}
1819: loops as examples; the discussion also applies to @code{?DO} and other
1820: loops). Perhaps the most insidious example is:
1821: @example
1822: AHEAD
1823: BEGIN
1824: x
1825: [ 1 CS-ROLL ] THEN
1826: @{ x @}
1827: ...
1828: UNTIL
1829: @end example
1830:
1831: This should be legal according to the visibility rule. The use of
1832: @code{x} can only be reached through the definition; but that appears
1833: textually below the use.
1834:
1835: From this example it is clear that the visibility rules cannot be fully
1836: implemented without major headaches. Our implementation treats common
1837: cases as advertised and the exceptions are treated in a safe way: The
1838: compiler makes a reasonable guess about the locals visible after a
1839: @code{BEGIN}; if it is too pessimistic, the
1840: user will get a spurious error about the local not being defined; if the
1841: compiler is too optimistic, it will notice this later and issue a
1842: warning. In the case above the compiler would complain about @code{x}
1843: being undefined at its use. You can see from the obscure examples in
1844: this section that it takes quite unusual control structures to get the
1845: compiler into trouble, and even then it will often do fine.
1846:
1847: If the @code{BEGIN} is reachable from above, the most optimistic guess
1848: is that all locals visible before the @code{BEGIN} will also be
1849: visible after the @code{BEGIN}. This guess is valid for all loops that
1850: are entered only through the @code{BEGIN}, in particular, for normal
1851: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
1852: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
1853: compiler. When the branch to the @code{BEGIN} is finally generated by
1854: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
1855: warns the user if it was too optimistic:
1856: @example
1857: IF
1858: @{ x @}
1859: BEGIN
1860: \ x ?
1861: [ 1 cs-roll ] THEN
1862: ...
1863: UNTIL
1864: @end example
1865:
1866: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
1867: optimistically assumes that it lives until the @code{THEN}. It notices
1868: this difference when it compiles the @code{UNTIL} and issues a
1869: warning. The user can avoid the warning, and make sure that @code{x}
1870: is not used in the wrong area by using explicit scoping:
1871: @example
1872: IF
1873: SCOPE
1874: @{ x @}
1875: ENDSCOPE
1876: BEGIN
1877: [ 1 cs-roll ] THEN
1878: ...
1879: UNTIL
1880: @end example
1881:
1882: Since the guess is optimistic, there will be no spurious error messages
1883: about undefined locals.
1884:
1885: If the @code{BEGIN} is not reachable from above (e.g., after
1886: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
1887: optimistic guess, as the locals visible after the @code{BEGIN} may be
1888: defined later. Therefore, the compiler assumes that no locals are
1889: visible after the @code{BEGIN}. However, the user can use
1890: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
1891: visible at the BEGIN as at the point where the top control-flow stack
1892: item was created.
1893:
1894: doc-assume-live
1895:
1896: E.g.,
1897: @example
1898: @{ x @}
1899: AHEAD
1900: ASSUME-LIVE
1901: BEGIN
1902: x
1903: [ 1 CS-ROLL ] THEN
1904: ...
1905: UNTIL
1906: @end example
1907:
1908: Other cases where the locals are defined before the @code{BEGIN} can be
1909: handled by inserting an appropriate @code{CS-ROLL} before the
1910: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
1911: behind the @code{ASSUME-LIVE}).
1912:
1913: Cases where locals are defined after the @code{BEGIN} (but should be
1914: visible immediately after the @code{BEGIN}) can only be handled by
1915: rearranging the loop. E.g., the ``most insidious'' example above can be
1916: arranged into:
1917: @example
1918: BEGIN
1919: @{ x @}
1920: ... 0=
1921: WHILE
1922: x
1923: REPEAT
1924: @end example
1925:
1926: @node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
1927: @subsubsection How long do locals live?
1928: @cindex locals lifetime
1929: @cindex lifetime of locals
1930:
1931: The right answer for the lifetime question would be: A local lives at
1932: least as long as it can be accessed. For a value-flavoured local this
1933: means: until the end of its visibility. However, a variable-flavoured
1934: local could be accessed through its address far beyond its visibility
1935: scope. Ultimately, this would mean that such locals would have to be
1936: garbage collected. Since this entails un-Forth-like implementation
1937: complexities, I adopted the same cowardly solution as some other
1938: languages (e.g., C): The local lives only as long as it is visible;
1939: afterwards its address is invalid (and programs that access it
1940: afterwards are erroneous).
1941:
1942: @node Programming Style, Implementation, How long do locals live?, Gforth locals
1943: @subsubsection Programming Style
1944: @cindex locals programming style
1945: @cindex programming style, locals
1946:
1947: The freedom to define locals anywhere has the potential to change
1948: programming styles dramatically. In particular, the need to use the
1949: return stack for intermediate storage vanishes. Moreover, all stack
1950: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
1951: determined arguments) can be eliminated: If the stack items are in the
1952: wrong order, just write a locals definition for all of them; then
1953: write the items in the order you want.
1954:
1955: This seems a little far-fetched and eliminating stack manipulations is
1956: unlikely to become a conscious programming objective. Still, the number
1957: of stack manipulations will be reduced dramatically if local variables
1958: are used liberally (e.g., compare @code{max} in @ref{Gforth locals} with
1959: a traditional implementation of @code{max}).
1960:
1961: This shows one potential benefit of locals: making Forth programs more
1962: readable. Of course, this benefit will only be realized if the
1963: programmers continue to honour the principle of factoring instead of
1964: using the added latitude to make the words longer.
1965:
1966: @cindex single-assignment style for locals
1967: Using @code{TO} can and should be avoided. Without @code{TO},
1968: every value-flavoured local has only a single assignment and many
1969: advantages of functional languages apply to Forth. I.e., programs are
1970: easier to analyse, to optimize and to read: It is clear from the
1971: definition what the local stands for, it does not turn into something
1972: different later.
1973:
1974: E.g., a definition using @code{TO} might look like this:
1975: @example
1976: : strcmp @{ addr1 u1 addr2 u2 -- n @}
1977: u1 u2 min 0
1978: ?do
1979: addr1 c@@ addr2 c@@ -
1980: ?dup-if
1981: unloop exit
1982: then
1983: addr1 char+ TO addr1
1984: addr2 char+ TO addr2
1985: loop
1986: u1 u2 - ;
1987: @end example
1988: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
1989: every loop iteration. @code{strcmp} is a typical example of the
1990: readability problems of using @code{TO}. When you start reading
1991: @code{strcmp}, you think that @code{addr1} refers to the start of the
1992: string. Only near the end of the loop you realize that it is something
1993: else.
1994:
1995: This can be avoided by defining two locals at the start of the loop that
1996: are initialized with the right value for the current iteration.
1997: @example
1998: : strcmp @{ addr1 u1 addr2 u2 -- n @}
1999: addr1 addr2
2000: u1 u2 min 0
2001: ?do @{ s1 s2 @}
2002: s1 c@@ s2 c@@ -
2003: ?dup-if
2004: unloop exit
2005: then
2006: s1 char+ s2 char+
2007: loop
2008: 2drop
2009: u1 u2 - ;
2010: @end example
2011: Here it is clear from the start that @code{s1} has a different value
2012: in every loop iteration.
2013:
2014: @node Implementation, , Programming Style, Gforth locals
2015: @subsubsection Implementation
2016: @cindex locals implementation
2017: @cindex implementation of locals
2018:
2019: @cindex locals stack
2020: Gforth uses an extra locals stack. The most compelling reason for
2021: this is that the return stack is not float-aligned; using an extra stack
2022: also eliminates the problems and restrictions of using the return stack
2023: as locals stack. Like the other stacks, the locals stack grows toward
2024: lower addresses. A few primitives allow an efficient implementation:
2025:
2026: doc-@local#
2027: doc-f@local#
2028: doc-laddr#
2029: doc-lp+!#
2030: doc-lp!
2031: doc->l
2032: doc-f>l
2033:
2034: In addition to these primitives, some specializations of these
2035: primitives for commonly occurring inline arguments are provided for
2036: efficiency reasons, e.g., @code{@@local0} as specialization of
2037: @code{@@local#} for the inline argument 0. The following compiling words
2038: compile the right specialized version, or the general version, as
2039: appropriate:
2040:
2041: doc-compile-@local
2042: doc-compile-f@local
2043: doc-compile-lp+!
2044:
2045: Combinations of conditional branches and @code{lp+!#} like
2046: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
2047: is taken) are provided for efficiency and correctness in loops.
2048:
2049: A special area in the dictionary space is reserved for keeping the
2050: local variable names. @code{@{} switches the dictionary pointer to this
2051: area and @code{@}} switches it back and generates the locals
2052: initializing code. @code{W:} etc.@ are normal defining words. This
2053: special area is cleared at the start of every colon definition.
2054:
2055: @cindex wordlist for defining locals
2056: A special feature of Gforth's dictionary is used to implement the
2057: definition of locals without type specifiers: every wordlist (aka
2058: vocabulary) has its own methods for searching
2059: etc. (@pxref{Wordlists}). For the present purpose we defined a wordlist
2060: with a special search method: When it is searched for a word, it
2061: actually creates that word using @code{W:}. @code{@{} changes the search
2062: order to first search the wordlist containing @code{@}}, @code{W:} etc.,
2063: and then the wordlist for defining locals without type specifiers.
2064:
2065: The lifetime rules support a stack discipline within a colon
2066: definition: The lifetime of a local is either nested with other locals
2067: lifetimes or it does not overlap them.
2068:
2069: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
2070: pointer manipulation is generated. Between control structure words
2071: locals definitions can push locals onto the locals stack. @code{AGAIN}
2072: is the simplest of the other three control flow words. It has to
2073: restore the locals stack depth of the corresponding @code{BEGIN}
2074: before branching. The code looks like this:
2075: @format
2076: @code{lp+!#} current-locals-size @minus{} dest-locals-size
2077: @code{branch} <begin>
2078: @end format
2079:
2080: @code{UNTIL} is a little more complicated: If it branches back, it
2081: must adjust the stack just like @code{AGAIN}. But if it falls through,
2082: the locals stack must not be changed. The compiler generates the
2083: following code:
2084: @format
2085: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
2086: @end format
2087: The locals stack pointer is only adjusted if the branch is taken.
2088:
2089: @code{THEN} can produce somewhat inefficient code:
2090: @format
2091: @code{lp+!#} current-locals-size @minus{} orig-locals-size
2092: <orig target>:
2093: @code{lp+!#} orig-locals-size @minus{} new-locals-size
2094: @end format
2095: The second @code{lp+!#} adjusts the locals stack pointer from the
2096: level at the @var{orig} point to the level after the @code{THEN}. The
2097: first @code{lp+!#} adjusts the locals stack pointer from the current
2098: level to the level at the orig point, so the complete effect is an
2099: adjustment from the current level to the right level after the
2100: @code{THEN}.
2101:
2102: @cindex locals information on the control-flow stack
2103: @cindex control-flow stack items, locals information
2104: In a conventional Forth implementation a dest control-flow stack entry
2105: is just the target address and an orig entry is just the address to be
2106: patched. Our locals implementation adds a wordlist to every orig or dest
2107: item. It is the list of locals visible (or assumed visible) at the point
2108: described by the entry. Our implementation also adds a tag to identify
2109: the kind of entry, in particular to differentiate between live and dead
2110: (reachable and unreachable) orig entries.
2111:
2112: A few unusual operations have to be performed on locals wordlists:
2113:
2114: doc-common-list
2115: doc-sub-list?
2116: doc-list-size
2117:
2118: Several features of our locals wordlist implementation make these
2119: operations easy to implement: The locals wordlists are organised as
2120: linked lists; the tails of these lists are shared, if the lists
2121: contain some of the same locals; and the address of a name is greater
2122: than the address of the names behind it in the list.
2123:
2124: Another important implementation detail is the variable
2125: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
2126: determine if they can be reached directly or only through the branch
2127: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
2128: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
2129: definition, by @code{BEGIN} and usually by @code{THEN}.
2130:
2131: Counted loops are similar to other loops in most respects, but
2132: @code{LEAVE} requires special attention: It performs basically the same
2133: service as @code{AHEAD}, but it does not create a control-flow stack
2134: entry. Therefore the information has to be stored elsewhere;
2135: traditionally, the information was stored in the target fields of the
2136: branches created by the @code{LEAVE}s, by organizing these fields into a
2137: linked list. Unfortunately, this clever trick does not provide enough
2138: space for storing our extended control flow information. Therefore, we
2139: introduce another stack, the leave stack. It contains the control-flow
2140: stack entries for all unresolved @code{LEAVE}s.
2141:
2142: Local names are kept until the end of the colon definition, even if
2143: they are no longer visible in any control-flow path. In a few cases
2144: this may lead to increased space needs for the locals name area, but
2145: usually less than reclaiming this space would cost in code size.
2146:
2147:
2148: @node ANS Forth locals, , Gforth locals, Locals
2149: @subsection ANS Forth locals
2150: @cindex locals, ANS Forth style
2151:
2152: The ANS Forth locals wordset does not define a syntax for locals, but
2153: words that make it possible to define various syntaxes. One of the
2154: possible syntaxes is a subset of the syntax we used in the Gforth locals
2155: wordset, i.e.:
2156:
2157: @example
2158: @{ local1 local2 ... -- comment @}
2159: @end example
2160: or
2161: @example
2162: @{ local1 local2 ... @}
2163: @end example
2164:
2165: The order of the locals corresponds to the order in a stack comment. The
2166: restrictions are:
2167:
2168: @itemize @bullet
2169: @item
2170: Locals can only be cell-sized values (no type specifiers are allowed).
2171: @item
2172: Locals can be defined only outside control structures.
2173: @item
2174: Locals can interfere with explicit usage of the return stack. For the
2175: exact (and long) rules, see the standard. If you don't use return stack
2176: accessing words in a definition using locals, you will be all right. The
2177: purpose of this rule is to make locals implementation on the return
2178: stack easier.
2179: @item
2180: The whole definition must be in one line.
2181: @end itemize
2182:
2183: Locals defined in this way behave like @code{VALUE}s (@xref{Simple
2184: Defining Words}). I.e., they are initialized from the stack. Using their
2185: name produces their value. Their value can be changed using @code{TO}.
2186:
2187: Since this syntax is supported by Gforth directly, you need not do
2188: anything to use it. If you want to port a program using this syntax to
2189: another ANS Forth system, use @file{compat/anslocal.fs} to implement the
2190: syntax on the other system.
2191:
2192: Note that a syntax shown in the standard, section A.13 looks
2193: similar, but is quite different in having the order of locals
2194: reversed. Beware!
2195:
2196: The ANS Forth locals wordset itself consists of the following word
2197:
2198: doc-(local)
2199:
2200: The ANS Forth locals extension wordset defines a syntax, but it is so
2201: awful that we strongly recommend not to use it. We have implemented this
2202: syntax to make porting to Gforth easy, but do not document it here. The
2203: problem with this syntax is that the locals are defined in an order
2204: reversed with respect to the standard stack comment notation, making
2205: programs harder to read, and easier to misread and miswrite. The only
2206: merit of this syntax is that it is easy to implement using the ANS Forth
2207: locals wordset.
2208:
2209: @node Defining Words, Structures, Locals, Words
2210: @section Defining Words
2211: @cindex defining words
2212:
2213: @menu
2214: * Simple Defining Words::
2215: * Colon Definitions::
2216: * User-defined Defining Words::
2217: * Supplying names::
2218: * Interpretation and Compilation Semantics::
2219: @end menu
2220:
2221: @node Simple Defining Words, Colon Definitions, Defining Words, Defining Words
2222: @subsection Simple Defining Words
2223: @cindex simple defining words
2224: @cindex defining words, simple
2225:
2226: doc-constant
2227: doc-2constant
2228: doc-fconstant
2229: doc-variable
2230: doc-2variable
2231: doc-fvariable
2232: doc-create
2233: doc-user
2234: doc-value
2235: doc-to
2236: doc-defer
2237: doc-is
2238:
2239: @node Colon Definitions, User-defined Defining Words, Simple Defining Words, Defining Words
2240: @subsection Colon Definitions
2241: @cindex colon definitions
2242:
2243: @example
2244: : name ( ... -- ... )
2245: word1 word2 word3 ;
2246: @end example
2247:
2248: creates a word called @code{name}, that, upon execution, executes
2249: @code{word1 word2 word3}. @code{name} is a @dfn{(colon) definition}.
2250:
2251: The explanation above is somewhat superficial. @xref{Interpretation and
2252: Compilation Semantics} for an in-depth discussion of some of the issues
2253: involved.
2254:
2255: doc-:
2256: doc-;
2257:
2258: @node User-defined Defining Words, Supplying names, Colon Definitions, Defining Words
2259: @subsection User-defined Defining Words
2260: @cindex user-defined defining words
2261: @cindex defining words, user-defined
2262:
2263: You can create new defining words simply by wrapping defining-time code
2264: around existing defining words and putting the sequence in a colon
2265: definition.
2266:
2267: @cindex @code{CREATE} ... @code{DOES>}
2268: If you want the words defined with your defining words to behave
2269: differently from words defined with standard defining words, you can
2270: write your defining word like this:
2271:
2272: @example
2273: : def-word ( "name" -- )
2274: Create @var{code1}
2275: DOES> ( ... -- ... )
2276: @var{code2} ;
2277:
2278: def-word name
2279: @end example
2280:
2281: Technically, this fragment defines a defining word @code{def-word}, and
2282: a word @code{name}; when you execute @code{name}, the address of the
2283: body of @code{name} is put on the data stack and @var{code2} is executed
2284: (the address of the body of @code{name} is the address @code{HERE}
2285: returns immediately after the @code{CREATE}).
2286:
2287: In other words, if you make the following definitions:
2288:
2289: @example
2290: : def-word1 ( "name" -- )
2291: Create @var{code1} ;
2292:
2293: : action1 ( ... -- ... )
2294: @var{code2} ;
2295:
2296: def-word name1
2297: @end example
2298:
2299: Using @code{name1 action1} is equivalent to using @code{name}.
2300:
2301: E.g., you can implement @code{Constant} in this way:
2302:
2303: @example
2304: : constant ( w "name" -- )
2305: create ,
2306: DOES> ( -- w )
2307: @@ ;
2308: @end example
2309:
2310: When you create a constant with @code{5 constant five}, first a new word
2311: @code{five} is created, then the value 5 is laid down in the body of
2312: @code{five} with @code{,}. When @code{five} is invoked, the address of
2313: the body is put on the stack, and @code{@@} retrieves the value 5.
2314:
2315: @cindex stack effect of @code{DOES>}-parts
2316: @cindex @code{DOES>}-parts, stack effect
2317: In the example above the stack comment after the @code{DOES>} specifies
2318: the stack effect of the defined words, not the stack effect of the
2319: following code (the following code expects the address of the body on
2320: the top of stack, which is not reflected in the stack comment). This is
2321: the convention that I use and recommend (it clashes a bit with using
2322: locals declarations for stack effect specification, though).
2323:
2324: @subsubsection Applications of @code{CREATE..DOES>}
2325: @cindex @code{CREATE} ... @code{DOES>}, applications
2326:
2327: You may wonder how to use this feature. Here are some usage patterns:
2328:
2329: @cindex factoring similar colon definitions
2330: When you see a sequence of code occurring several times, and you can
2331: identify a meaning, you will factor it out as a colon definition. When
2332: you see similar colon definitions, you can factor them using
2333: @code{CREATE..DOES>}. E.g., an assembler usually defines several words
2334: that look very similar:
2335: @example
2336: : ori, ( reg-target reg-source n -- )
2337: 0 asm-reg-reg-imm ;
2338: : andi, ( reg-target reg-source n -- )
2339: 1 asm-reg-reg-imm ;
2340: @end example
2341:
2342: This could be factored with:
2343: @example
2344: : reg-reg-imm ( op-code -- )
2345: create ,
2346: DOES> ( reg-target reg-source n -- )
2347: @@ asm-reg-reg-imm ;
2348:
2349: 0 reg-reg-imm ori,
2350: 1 reg-reg-imm andi,
2351: @end example
2352:
2353: @cindex currying
2354: Another view of @code{CREATE..DOES>} is to consider it as a crude way to
2355: supply a part of the parameters for a word (known as @dfn{currying} in
2356: the functional language community). E.g., @code{+} needs two
2357: parameters. Creating versions of @code{+} with one parameter fixed can
2358: be done like this:
2359: @example
2360: : curry+ ( n1 -- )
2361: create ,
2362: DOES> ( n2 -- n1+n2 )
2363: @@ + ;
2364:
2365: 3 curry+ 3+
2366: -2 curry+ 2-
2367: @end example
2368:
2369: @subsubsection The gory details of @code{CREATE..DOES>}
2370: @cindex @code{CREATE} ... @code{DOES>}, details
2371:
2372: doc-does>
2373:
2374: @cindex @code{DOES>} in a separate definition
2375: This means that you need not use @code{CREATE} and @code{DOES>} in the
2376: same definition; E.g., you can put the @code{DOES>}-part in a separate
2377: definition. This allows us to, e.g., select among different DOES>-parts:
2378: @example
2379: : does1
2380: DOES> ( ... -- ... )
2381: ... ;
2382:
2383: : does2
2384: DOES> ( ... -- ... )
2385: ... ;
2386:
2387: : def-word ( ... -- ... )
2388: create ...
2389: IF
2390: does1
2391: ELSE
2392: does2
2393: ENDIF ;
2394: @end example
2395:
2396: @cindex @code{DOES>} in interpretation state
2397: In a standard program you can apply a @code{DOES>}-part only if the last
2398: word was defined with @code{CREATE}. In Gforth, the @code{DOES>}-part
2399: will override the behaviour of the last word defined in any case. In a
2400: standard program, you can use @code{DOES>} only in a colon
2401: definition. In Gforth, you can also use it in interpretation state, in a
2402: kind of one-shot mode:
2403: @example
2404: CREATE name ( ... -- ... )
2405: @var{initialization}
2406: DOES>
2407: @var{code} ;
2408: @end example
2409: This is equivalent to the standard
2410: @example
2411: :noname
2412: DOES>
2413: @var{code} ;
2414: CREATE name EXECUTE ( ... -- ... )
2415: @var{initialization}
2416: @end example
2417:
2418: You can get the address of the body of a word with
2419:
2420: doc->body
2421:
2422: @node Supplying names, Interpretation and Compilation Semantics, User-defined Defining Words, Defining Words
2423: @subsection Supplying names for the defined words
2424: @cindex names for defined words
2425: @cindex defining words, name parameter
2426:
2427: @cindex defining words, name given in a string
2428: By default, defining words take the names for the defined words from the
2429: input stream. Sometimes you want to supply the name from a string. You
2430: can do this with
2431:
2432: doc-nextname
2433:
2434: E.g.,
2435:
2436: @example
2437: s" foo" nextname create
2438: @end example
2439: is equivalent to
2440: @example
2441: create foo
2442: @end example
2443:
2444: @cindex defining words without name
2445: Sometimes you want to define a word without a name. You can do this with
2446:
2447: doc-noname
2448:
2449: @cindex execution token of last defined word
2450: To make any use of the newly defined word, you need its execution
2451: token. You can get it with
2452:
2453: doc-lastxt
2454:
2455: E.g., you can initialize a deferred word with an anonymous colon
2456: definition:
2457: @example
2458: Defer deferred
2459: noname : ( ... -- ... )
2460: ... ;
2461: lastxt IS deferred
2462: @end example
2463:
2464: @code{lastxt} also works when the last word was not defined as
2465: @code{noname}.
2466:
2467: The standard has also recognized the need for anonymous words and
2468: provides
2469:
2470: doc-:noname
2471:
2472: This leaves the execution token for the word on the stack after the
2473: closing @code{;}. You can rewrite the last example with @code{:noname}:
2474: @example
2475: Defer deferred
2476: :noname ( ... -- ... )
2477: ... ;
2478: IS deferred
2479: @end example
2480:
2481: @node Interpretation and Compilation Semantics, , Supplying names, Defining Words
2482: @subsection Interpretation and Compilation Semantics
2483: @cindex semantics, interpretation and compilation
2484:
2485: @cindex interpretation semantics
2486: The @dfn{interpretation semantics} of a word are what the text
2487: interpreter does when it encounters the word in interpret state. It also
2488: appears in some other contexts, e.g., the execution token returned by
2489: @code{' @var{word}} identifies the interpretation semantics of
2490: @var{word} (in other words, @code{' @var{word} execute} is equivalent to
2491: interpret-state text interpretation of @code{@var{word}}).
2492:
2493: @cindex compilation semantics
2494: The @dfn{compilation semantics} of a word are what the text interpreter
2495: does when it encounters the word in compile state. It also appears in
2496: other contexts, e.g, @code{POSTPONE @var{word}} compiles@footnote{In
2497: standard terminology, ``appends to the current definition''.} the
2498: compilation semantics of @var{word}.
2499:
2500: @cindex execution semantics
2501: The standard also talks about @dfn{execution semantics}. They are used
2502: only for defining the interpretation and compilation semantics of many
2503: words. By default, the interpretation semantics of a word are to
2504: @code{execute} its execution semantics, and the compilation semantics of
2505: a word are to @code{compile,} its execution semantics.@footnote{In
2506: standard terminology: The default interpretation semantics are its
2507: execution semantics; the default compilation semantics are to append its
2508: execution semantics to the execution semantics of the current
2509: definition.}
2510:
2511: @cindex immediate words
2512: You can change the compilation semantics into @code{execute}ing the
2513: execution semantics with
2514:
2515: doc-immediate
2516:
2517: @cindex compile-only words
2518: You can remove the interpretation semantics of a word with
2519:
2520: doc-compile-only
2521: doc-restrict
2522:
2523: Note that ticking (@code{'}) compile-only words gives an error
2524: (``Interpreting a compile-only word'').
2525:
2526: Gforth also allows you to define words with arbitrary combinations of
2527: interpretation and compilation semantics.
2528:
2529: doc-interpret/compile:
2530:
2531: This feature was introduced for implementing @code{TO} and @code{S"}. I
2532: recommend that you do not define such words, as cute as they may be:
2533: they make it hard to get at both parts of the word in some contexts.
2534: E.g., assume you want to get an execution token for the compilation
2535: part. Instead, define two words, one that embodies the interpretation
2536: part, and one that embodies the compilation part.
2537:
2538: There is, however, a potentially useful application of this feature:
2539: Providing differing implementations for the default semantics. While
2540: this introduces redundancy and is therefore usually a bad idea, a
2541: performance improvement may be worth the trouble. E.g., consider the
2542: word @code{foobar}:
2543:
2544: @example
2545: : foobar
2546: foo bar ;
2547: @end example
2548:
2549: Let us assume that @code{foobar} is called so frequently that the
2550: calling overhead would take a significant amount of the run-time. We can
2551: optimize it with @code{interpret/compile:}:
2552:
2553: @example
2554: :noname
2555: foo bar ;
2556: :noname
2557: POSTPONE foo POSTPONE bar ;
2558: interpret/compile: foobar
2559: @end example
2560:
2561: This definition has the same interpretation semantics and essentially
2562: the same compilation semantics as the simple definition of
2563: @code{foobar}, but the implementation of the compilation semantics is
2564: more efficient with respect to run-time.
2565:
2566: @cindex state-smart words are a bad idea
2567: Some people try to use state-smart words to emulate the feature provided
2568: by @code{interpret/compile:} (words are state-smart if they check
2569: @code{STATE} during execution). E.g., they would try to code
2570: @code{foobar} like this:
2571:
2572: @example
2573: : foobar
2574: STATE @@
2575: IF ( compilation state )
2576: POSTPONE foo POSTPONE bar
2577: ELSE
2578: foo bar
2579: ENDIF ; immediate
2580: @end example
2581:
2582: While this works if @code{foobar} is processed only by the text
2583: interpreter, it does not work in other contexts (like @code{'} or
2584: @code{POSTPONE}). E.g., @code{' foobar} will produce an execution token
2585: for a state-smart word, not for the interpretation semantics of the
2586: original @code{foobar}; when you execute this execution token (directly
2587: with @code{EXECUTE} or indirectly through @code{COMPILE,}) in compile
2588: state, the result will not be what you expected (i.e., it will not
2589: perform @code{foo bar}). State-smart words are a bad idea. Simply don't
2590: write them!
2591:
2592: @cindex defining words with arbitrary semantics combinations
2593: It is also possible to write defining words that define words with
2594: arbitrary combinations of interpretation and compilation semantics (or,
2595: preferably, arbitrary combinations of implementations of the default
2596: semantics). In general, this looks like:
2597:
2598: @example
2599: : def-word
2600: create-interpret/compile
2601: @var{code1}
2602: interpretation>
2603: @var{code2}
2604: <interpretation
2605: compilation>
2606: @var{code3}
2607: <compilation ;
2608: @end example
2609:
2610: For a @var{word} defined with @code{def-word}, the interpretation
2611: semantics are to push the address of the body of @var{word} and perform
2612: @var{code2}, and the compilation semantics are to push the address of
2613: the body of @var{word} and perform @var{code3}. E.g., @code{constant}
2614: can also be defined like this:
2615:
2616: @example
2617: : constant ( n "name" -- )
2618: create-interpret/compile
2619: ,
2620: interpretation> ( -- n )
2621: @@
2622: <interpretation
2623: compilation> ( compilation. -- ; run-time. -- n )
2624: @@ postpone literal
2625: <compilation ;
2626: @end example
2627:
2628: doc-create-interpret/compile
2629: doc-interpretation>
2630: doc-<interpretation
2631: doc-compilation>
2632: doc-<compilation
2633:
2634: Note that words defined with @code{interpret/compile:} and
2635: @code{create-interpret/compile} have an extended header structure that
2636: differs from other words; however, unless you try to access them with
2637: plain address arithmetic, you should not notice this. Words for
2638: accessing the header structure usually know how to deal with this; e.g.,
2639: @code{' word >body} also gives you the body of a word created with
2640: @code{create-interpret/compile}.
2641:
2642: @c ----------------------------------------------------------
2643: @node Structures, Objects, Defining Words, Words
2644: @section Structures
2645: @cindex structures
2646: @cindex records
2647:
2648: This section presents the structure package that comes with Gforth. A
2649: version of the package implemented in plain ANS Forth is available in
2650: @file{compat/struct.fs}. This package was inspired by a posting on
2651: comp.lang.forth in 1989 (unfortunately I don't remember, by whom;
2652: possibly John Hayes). A version of this section has been published in
2653: ???. Marcel Hendrix provided helpful comments.
2654:
2655: @menu
2656: * Why explicit structure support?::
2657: * Structure Usage::
2658: * Structure Naming Convention::
2659: * Structure Implementation::
2660: * Structure Glossary::
2661: @end menu
2662:
2663: @node Why explicit structure support?, Structure Usage, Structures, Structures
2664: @subsection Why explicit structure support?
2665:
2666: @cindex address arithmetic for structures
2667: @cindex structures using address arithmetic
2668: If we want to use a structure containing several fields, we could simply
2669: reserve memory for it, and access the fields using address arithmetic
2670: (@pxref{Address arithmetic}). As an example, consider a structure with
2671: the following fields
2672:
2673: @table @code
2674: @item a
2675: is a float
2676: @item b
2677: is a cell
2678: @item c
2679: is a float
2680: @end table
2681:
2682: Given the (float-aligned) base address of the structure we get the
2683: address of the field
2684:
2685: @table @code
2686: @item a
2687: without doing anything further.
2688: @item b
2689: with @code{float+}
2690: @item c
2691: with @code{float+ cell+ faligned}
2692: @end table
2693:
2694: It is easy to see that this can become quite tiring.
2695:
2696: Moreover, it is not very readable, because seeing a
2697: @code{cell+} tells us neither which kind of structure is
2698: accessed nor what field is accessed; we have to somehow infer the kind
2699: of structure, and then look up in the documentation, which field of
2700: that structure corresponds to that offset.
2701:
2702: Finally, this kind of address arithmetic also causes maintenance
2703: troubles: If you add or delete a field somewhere in the middle of the
2704: structure, you have to find and change all computations for the fields
2705: afterwards.
2706:
2707: So, instead of using @code{cell+} and friends directly, how
2708: about storing the offsets in constants:
2709:
2710: @example
2711: 0 constant a-offset
2712: 0 float+ constant b-offset
2713: 0 float+ cell+ faligned c-offset
2714: @end example
2715:
2716: Now we can get the address of field @code{x} with @code{x-offset
2717: +}. This is much better in all respects. Of course, you still
2718: have to change all later offset definitions if you add a field. You can
2719: fix this by declaring the offsets in the following way:
2720:
2721: @example
2722: 0 constant a-offset
2723: a-offset float+ constant b-offset
2724: b-offset cell+ faligned constant c-offset
2725: @end example
2726:
2727: Since we always use the offsets with @code{+}, using a defining
2728: word @code{cfield} that includes the @code{+} in the
2729: action of the defined word offers itself:
2730:
2731: @example
2732: : cfield ( n "name" -- )
2733: create ,
2734: does> ( name execution: addr1 -- addr2 )
2735: @@ + ;
2736:
2737: 0 cfield a
2738: 0 a float+ cfield b
2739: 0 b cell+ faligned cfield c
2740: @end example
2741:
2742: Instead of @code{x-offset +}, we now simply write @code{x}.
2743:
2744: The structure field words now can be used quite nicely. However,
2745: their definition is still a bit cumbersome: We have to repeat the
2746: name, the information about size and alignment is distributed before
2747: and after the field definitions etc. The structure package presented
2748: here addresses these problems.
2749:
2750: @node Structure Usage, Structure Naming Convention, Why explicit structure support?, Structures
2751: @subsection Structure Usage
2752: @cindex structure usage
2753:
2754: @cindex @code{field} usage
2755: @cindex @code{struct} usage
2756: @cindex @code{end-struct} usage
2757: You can define a structure for a (data-less) linked list with
2758: @example
2759: struct
2760: cell% field list-next
2761: end-struct list%
2762: @end example
2763:
2764: With the address of the list node on the stack, you can compute the
2765: address of the field that contains the address of the next node with
2766: @code{list-next}. E.g., you can determine the length of a list
2767: with:
2768:
2769: @example
2770: : list-length ( list -- n )
2771: \ "list" is a pointer to the first element of a linked list
2772: \ "n" is the length of the list
2773: 0 begin ( list1 n1 )
2774: over
2775: while ( list1 n1 )
2776: 1+ swap list-next @@ swap
2777: repeat
2778: nip ;
2779: @end example
2780:
2781: You can reserve memory for a list node in the dictionary with
2782: @code{list% %allot}, which leaves the address of the list node on the
2783: stack. For the equivalent allocation on the heap you can use @code{list%
2784: %alloc} (or, for an @code{allocate}-like stack effect (i.e., with ior),
2785: use @code{list% %allocate}). You can also get the the size of a list
2786: node with @code{list% %size} and it's alignment with @code{list%
2787: %alignment}.
2788:
2789: Note that in ANS Forth the body of a @code{create}d word is
2790: @code{aligned} but not necessarily @code{faligned};
2791: therefore, if you do a
2792: @example
2793: create @emph{name} foo% %allot
2794: @end example
2795:
2796: then the memory alloted for @code{foo%} is
2797: guaranteed to start at the body of @code{@emph{name}} only if
2798: @code{foo%} contains only character, cell and double fields.
2799:
2800: @cindex strcutures containing structures
2801: You can also include a structure @code{foo%} as field of
2802: another structure, with:
2803: @example
2804: struct
2805: ...
2806: foo% field ...
2807: ...
2808: end-struct ...
2809: @end example
2810:
2811: @cindex structure extension
2812: @cindex extended records
2813: Instead of starting with an empty structure, you can also extend an
2814: existing structure. E.g., a plain linked list without data, as defined
2815: above, is hardly useful; You can extend it to a linked list of integers,
2816: like this:@footnote{This feature is also known as @emph{extended
2817: records}. It is the main innovation in the Oberon language; in other
2818: words, adding this feature to Modula-2 led Wirth to create a new
2819: language, write a new compiler etc. Adding this feature to Forth just
2820: requires a few lines of code.}
2821:
2822: @example
2823: list%
2824: cell% field intlist-int
2825: end-struct intlist%
2826: @end example
2827:
2828: @code{intlist%} is a structure with two fields:
2829: @code{list-next} and @code{intlist-int}.
2830:
2831: @cindex structures containing arrays
2832: You can specify an array type containing @emph{n} elements of
2833: type @code{foo%} like this:
2834:
2835: @example
2836: foo% @emph{n} *
2837: @end example
2838:
2839: You can use this array type in any place where you can use a normal
2840: type, e.g., when defining a @code{field}, or with
2841: @code{%allot}.
2842:
2843: @cindex first field optimization
2844: The first field is at the base address of a structure and the word
2845: for this field (e.g., @code{list-next}) actually does not change
2846: the address on the stack. You may be tempted to leave it away in the
2847: interest of run-time and space efficiency. This is not necessary,
2848: because the structure package optimizes this case and compiling such
2849: words does not generate any code. So, in the interest of readability
2850: and maintainability you should include the word for the field when
2851: accessing the field.
2852:
2853: @node Structure Naming Convention, Structure Implementation, Structure Usage, Structures
2854: @subsection Structure Naming Convention
2855: @cindex structure naming conventions
2856:
2857: The field names that come to (my) mind are often quite generic, and,
2858: if used, would cause frequent name clashes. E.g., many structures
2859: probably contain a @code{counter} field. The structure names
2860: that come to (my) mind are often also the logical choice for the names
2861: of words that create such a structure.
2862:
2863: Therefore, I have adopted the following naming conventions:
2864:
2865: @itemize @bullet
2866: @cindex field naming convention
2867: @item
2868: The names of fields are of the form
2869: @code{@emph{struct}-@emph{field}}, where
2870: @code{@emph{struct}} is the basic name of the structure, and
2871: @code{@emph{field}} is the basic name of the field. You can
2872: think about field words as converting converts the (address of the)
2873: structure into the (address of the) field.
2874:
2875: @cindex structure naming convention
2876: @item
2877: The names of structures are of the form
2878: @code{@emph{struct}%}, where
2879: @code{@emph{struct}} is the basic name of the structure.
2880: @end itemize
2881:
2882: This naming convention does not work that well for fields of extended
2883: structures; e.g., the integer list structure has a field
2884: @code{intlist-int}, but has @code{list-next}, not
2885: @code{intlist-next}.
2886:
2887: @node Structure Implementation, Structure Glossary, Structure Naming Convention, Structures
2888: @subsection Structure Implementation
2889: @cindex structure implementation
2890: @cindex implementation of structures
2891:
2892: The central idea in the implementation is to pass the data about the
2893: structure being built on the stack, not in some global
2894: variable. Everything else falls into place naturally once this design
2895: decision is made.
2896:
2897: The type description on the stack is of the form @emph{align
2898: size}. Keeping the size on the top-of-stack makes dealing with arrays
2899: very simple.
2900:
2901: @code{field} is a defining word that uses @code{create}
2902: and @code{does>}. The body of the field contains the offset
2903: of the field, and the normal @code{does>} action is
2904:
2905: @example
2906: @ +
2907: @end example
2908:
2909: i.e., add the offset to the address, giving the stack effect
2910: @code{addr1 -- addr2} for a field.
2911:
2912: @cindex first field optimization, implementation
2913: This simple structure is slightly complicated by the optimization
2914: for fields with offset 0, which requires a different
2915: @code{does>}-part (because we cannot rely on there being
2916: something on the stack if such a field is invoked during
2917: compilation). Therefore, we put the different @code{does>}-parts
2918: in separate words, and decide which one to invoke based on the
2919: offset. For a zero offset, the field is basically a noop; it is
2920: immediate, and therefore no code is generated when it is compiled.
2921:
2922: @node Structure Glossary, , Structure Implementation, Structures
2923: @subsection Structure Glossary
2924: @cindex structure glossary
2925:
2926: doc-%align
2927: doc-%alignment
2928: doc-%alloc
2929: doc-%allocate
2930: doc-%allot
2931: doc-cell%
2932: doc-char%
2933: doc-dfloat%
2934: doc-double%
2935: doc-end-struct
2936: doc-field
2937: doc-float%
2938: doc-nalign
2939: doc-sfloat%
2940: doc-%size
2941: doc-struct
2942:
2943: @c -------------------------------------------------------------
2944: @node Objects, Object Oriented Forth, Structures, Words
2945: @section Objects
2946: @cindex objects
2947: @cindex object-oriented programming
2948:
2949: @cindex @file{objects.fs}
2950: @cindex @file{oof.fs}
2951: Gforth comes with three packets for object-oriented programming,
2952: @file{objects.fs}, @file{oof.fs}, and @file{mini-oof.fs}; none of them
2953: is preloaded, so you have to @code{include} them before use. This
2954: section describes the @file{objects.fs} packet. You can find a
2955: description (in German) of @file{oof.fs} in @cite{Object oriented
2956: bigFORTH} by Bernd Paysan, published in @cite{Vierte Dimension} 10(2),
2957: 1994. Both packets are written in ANS Forth and can be used with any
2958: other standard Forth.
2959: @c McKewan's and Zsoter's packages
2960: @c this section is a variant of ...
2961:
2962: This section assumes (in some places) that you have read @ref{Structures}.
2963:
2964: @menu
2965: * Properties of the Objects model::
2966: * Why object-oriented programming?::
2967: * Object-Oriented Terminology::
2968: * Basic Objects Usage::
2969: * The class Object::
2970: * Creating objects::
2971: * Object-Oriented Programming Style::
2972: * Class Binding::
2973: * Method conveniences::
2974: * Classes and Scoping::
2975: * Object Interfaces::
2976: * Objects Implementation::
2977: * Comparison with other object models::
2978: * Objects Glossary::
2979: @end menu
2980:
2981: Marcel Hendrix provided helpful comments on this section. Andras Zsoter
2982: and Bernd Paysan helped me with the related works section.
2983:
2984: @node Properties of the Objects model, Why object-oriented programming?, Objects, Objects
2985: @subsection Properties of the @file{objects.fs} model
2986: @cindex @file{objects.fs} properties
2987:
2988: @itemize @bullet
2989: @item
2990: It is straightforward to pass objects on the stack. Passing
2991: selectors on the stack is a little less convenient, but possible.
2992:
2993: @item
2994: Objects are just data structures in memory, and are referenced by
2995: their address. You can create words for objects with normal defining
2996: words like @code{constant}. Likewise, there is no difference
2997: between instance variables that contain objects and those
2998: that contain other data.
2999:
3000: @item
3001: Late binding is efficient and easy to use.
3002:
3003: @item
3004: It avoids parsing, and thus avoids problems with state-smartness
3005: and reduced extensibility; for convenience there are a few parsing
3006: words, but they have non-parsing counterparts. There are also a few
3007: defining words that parse. This is hard to avoid, because all standard
3008: defining words parse (except @code{:noname}); however, such
3009: words are not as bad as many other parsing words, because they are not
3010: state-smart.
3011:
3012: @item
3013: It does not try to incorporate everything. It does a few things
3014: and does them well (IMO). In particular, I did not intend to support
3015: information hiding with this model (although it has features that may
3016: help); you can use a separate package for achieving this.
3017:
3018: @item
3019: It is layered; you don't have to learn and use all features to use this
3020: model. Only a few features are necessary (@xref{Basic Objects Usage},
3021: @xref{The class Object}, @xref{Creating objects}.), the others
3022: are optional and independent of each other.
3023:
3024: @item
3025: An implementation in ANS Forth is available.
3026:
3027: @end itemize
3028:
3029: I have used the technique, on which this model is based, for
3030: implementing the parser generator Gray; we have also used this technique
3031: in Gforth for implementing the various flavours of wordlists (hashed or
3032: not, case-sensitive or not, special-purpose wordlists for locals etc.).
3033:
3034: @node Why object-oriented programming?, Object-Oriented Terminology, Properties of the Objects model, Objects
3035: @subsection Why object-oriented programming?
3036: @cindex object-oriented programming motivation
3037: @cindex motivation for object-oriented programming
3038:
3039: Often we have to deal with several data structures (@emph{objects}),
3040: that have to be treated similarly in some respects, but differ in
3041: others. Graphical objects are the textbook example: circles,
3042: triangles, dinosaurs, icons, and others, and we may want to add more
3043: during program development. We want to apply some operations to any
3044: graphical object, e.g., @code{draw} for displaying it on the
3045: screen. However, @code{draw} has to do something different for
3046: every kind of object.
3047:
3048: We could implement @code{draw} as a big @code{CASE}
3049: control structure that executes the appropriate code depending on the
3050: kind of object to be drawn. This would be not be very elegant, and,
3051: moreover, we would have to change @code{draw} every time we add
3052: a new kind of graphical object (say, a spaceship).
3053:
3054: What we would rather do is: When defining spaceships, we would tell
3055: the system: "Here's how you @code{draw} a spaceship; you figure
3056: out the rest."
3057:
3058: This is the problem that all systems solve that (rightfully) call
3059: themselves object-oriented, and the object-oriented package I present
3060: here also solves this problem (and not much else).
3061:
3062: @node Object-Oriented Terminology, Basic Objects Usage, Why object-oriented programming?, Objects
3063: @subsection Object-Oriented Terminology
3064: @cindex object-oriented terminology
3065: @cindex terminology for object-oriented programming
3066:
3067: This section is mainly for reference, so you don't have to understand
3068: all of it right away. The terminology is mainly Smalltalk-inspired. In
3069: short:
3070:
3071: @table @emph
3072: @cindex class
3073: @item class
3074: a data structure definition with some extras.
3075:
3076: @cindex object
3077: @item object
3078: an instance of the data structure described by the class definition.
3079:
3080: @cindex instance variables
3081: @item instance variables
3082: fields of the data structure.
3083:
3084: @cindex selector
3085: @cindex method selector
3086: @cindex virtual function
3087: @item selector
3088: (or @emph{method selector}) a word (e.g.,
3089: @code{draw}) for performing an operation on a variety of data
3090: structures (classes). A selector describes @emph{what} operation to
3091: perform. In C++ terminology: a (pure) virtual function.
3092:
3093: @cindex method
3094: @item method
3095: the concrete definition that performs the operation
3096: described by the selector for a specific class. A method specifies
3097: @emph{how} the operation is performed for a specific class.
3098:
3099: @cindex selector invocation
3100: @cindex message send
3101: @cindex invoking a selector
3102: @item selector invocation
3103: a call of a selector. One argument of the call (the TOS (top-of-stack))
3104: is used for determining which method is used. In Smalltalk terminology:
3105: a message (consisting of the selector and the other arguments) is sent
3106: to the object.
3107:
3108: @cindex receiving object
3109: @item receiving object
3110: the object used for determining the method executed by a selector
3111: invocation. In our model it is the object that is on the TOS when the
3112: selector is invoked. (@emph{Receiving} comes from Smalltalks
3113: @emph{message} terminology.)
3114:
3115: @cindex child class
3116: @cindex parent class
3117: @cindex inheritance
3118: @item child class
3119: a class that has (@emph{inherits}) all properties (instance variables,
3120: selectors, methods) from a @emph{parent class}. In Smalltalk
3121: terminology: The subclass inherits from the superclass. In C++
3122: terminology: The derived class inherits from the base class.
3123:
3124: @end table
3125:
3126: @c If you wonder about the message sending terminology, it comes from
3127: @c a time when each object had it's own task and objects communicated via
3128: @c message passing; eventually the Smalltalk developers realized that
3129: @c they can do most things through simple (indirect) calls. They kept the
3130: @c terminology.
3131:
3132: @node Basic Objects Usage, The class Object, Object-Oriented Terminology, Objects
3133: @subsection Basic Objects Usage
3134: @cindex basic objects usage
3135: @cindex objects, basic usage
3136:
3137: You can define a class for graphical objects like this:
3138:
3139: @cindex @code{class} usage
3140: @cindex @code{end-class} usage
3141: @cindex @code{selector} usage
3142: @example
3143: object class \ "object" is the parent class
3144: selector draw ( x y graphical -- )
3145: end-class graphical
3146: @end example
3147:
3148: This code defines a class @code{graphical} with an
3149: operation @code{draw}. We can perform the operation
3150: @code{draw} on any @code{graphical} object, e.g.:
3151:
3152: @example
3153: 100 100 t-rex draw
3154: @end example
3155:
3156: where @code{t-rex} is a word (say, a constant) that produces a
3157: graphical object.
3158:
3159: @cindex abstract class
3160: How do we create a graphical object? With the present definitions,
3161: we cannot create a useful graphical object. The class
3162: @code{graphical} describes graphical objects in general, but not
3163: any concrete graphical object type (C++ users would call it an
3164: @emph{abstract class}); e.g., there is no method for the selector
3165: @code{draw} in the class @code{graphical}.
3166:
3167: For concrete graphical objects, we define child classes of the
3168: class @code{graphical}, e.g.:
3169:
3170: @cindex @code{overrides} usage
3171: @cindex @code{field} usage in class definition
3172: @example
3173: graphical class \ "graphical" is the parent class
3174: cell% field circle-radius
3175:
3176: :noname ( x y circle -- )
3177: circle-radius @@ draw-circle ;
3178: overrides draw
3179:
3180: :noname ( n-radius circle -- )
3181: circle-radius ! ;
3182: overrides construct
3183:
3184: end-class circle
3185: @end example
3186:
3187: Here we define a class @code{circle} as a child of @code{graphical},
3188: with a field @code{circle-radius} (which behaves just like a field in
3189: @pxref{Structures}); it defines new methods for the selectors
3190: @code{draw} and @code{construct} (@code{construct} is defined in
3191: @code{object}, the parent class of @code{graphical}).
3192:
3193: Now we can create a circle on the heap (i.e.,
3194: @code{allocate}d memory) with
3195:
3196: @cindex @code{heap-new} usage
3197: @example
3198: 50 circle heap-new constant my-circle
3199: @end example
3200:
3201: @code{heap-new} invokes @code{construct}, thus
3202: initializing the field @code{circle-radius} with 50. We can draw
3203: this new circle at (100,100) with
3204:
3205: @example
3206: 100 100 my-circle draw
3207: @end example
3208:
3209: @cindex selector invocation, restrictions
3210: @cindex class definition, restrictions
3211: Note: You can invoke a selector only if the object on the TOS
3212: (the receiving object) belongs to the class where the selector was
3213: defined or one of its descendents; e.g., you can invoke
3214: @code{draw} only for objects belonging to @code{graphical}
3215: or its descendents (e.g., @code{circle}). Immediately before
3216: @code{end-class}, the search order has to be the same as
3217: immediately after @code{class}.
3218:
3219: @node The class Object, Creating objects, Basic Objects Usage, Objects
3220: @subsection The class @code{object}
3221: @cindex @code{object} class
3222:
3223: When you define a class, you have to specify a parent class. So how do
3224: you start defining classes? There is one class available from the start:
3225: @code{object}. You can use it as ancestor for all classes. It is the
3226: only class that has no parent. It has two selectors: @code{construct}
3227: and @code{print}.
3228:
3229: @node Creating objects, Object-Oriented Programming Style, The class Object, Objects
3230: @subsection Creating objects
3231: @cindex creating objects
3232: @cindex object creation
3233: @cindex object allocation options
3234:
3235: @cindex @code{heap-new} discussion
3236: @cindex @code{dict-new} discussion
3237: @cindex @code{construct} discussion
3238: You can create and initialize an object of a class on the heap with
3239: @code{heap-new} ( ... class -- object ) and in the dictionary
3240: (allocation with @code{allot}) with @code{dict-new} (
3241: ... class -- object ). Both words invoke @code{construct}, which
3242: consumes the stack items indicated by "..." above.
3243:
3244: @cindex @code{init-object} discussion
3245: @cindex @code{class-inst-size} discussion
3246: If you want to allocate memory for an object yourself, you can get its
3247: alignment and size with @code{class-inst-size 2@@} ( class --
3248: align size ). Once you have memory for an object, you can initialize
3249: it with @code{init-object} ( ... class object -- );
3250: @code{construct} does only a part of the necessary work.
3251:
3252: @node Object-Oriented Programming Style, Class Binding, Creating objects, Objects
3253: @subsection Object-Oriented Programming Style
3254: @cindex object-oriented programming style
3255:
3256: This section is not exhaustive.
3257:
3258: @cindex stack effects of selectors
3259: @cindex selectors and stack effects
3260: In general, it is a good idea to ensure that all methods for the
3261: same selector have the same stack effect: when you invoke a selector,
3262: you often have no idea which method will be invoked, so, unless all
3263: methods have the same stack effect, you will not know the stack effect
3264: of the selector invocation.
3265:
3266: One exception to this rule is methods for the selector
3267: @code{construct}. We know which method is invoked, because we
3268: specify the class to be constructed at the same place. Actually, I
3269: defined @code{construct} as a selector only to give the users a
3270: convenient way to specify initialization. The way it is used, a
3271: mechanism different from selector invocation would be more natural
3272: (but probably would take more code and more space to explain).
3273:
3274: @node Class Binding, Method conveniences, Object-Oriented Programming Style, Objects
3275: @subsection Class Binding
3276: @cindex class binding
3277: @cindex early binding
3278:
3279: @cindex late binding
3280: Normal selector invocations determine the method at run-time depending
3281: on the class of the receiving object (late binding).
3282:
3283: Sometimes we want to invoke a different method. E.g., assume that
3284: you want to use the simple method for @code{print}ing
3285: @code{object}s instead of the possibly long-winded
3286: @code{print} method of the receiver class. You can achieve this
3287: by replacing the invocation of @code{print} with
3288:
3289: @cindex @code{[bind]} usage
3290: @example
3291: [bind] object print
3292: @end example
3293:
3294: in compiled code or
3295:
3296: @cindex @code{bind} usage
3297: @example
3298: bind object print
3299: @end example
3300:
3301: @cindex class binding, alternative to
3302: in interpreted code. Alternatively, you can define the method with a
3303: name (e.g., @code{print-object}), and then invoke it through the
3304: name. Class binding is just a (often more convenient) way to achieve
3305: the same effect; it avoids name clutter and allows you to invoke
3306: methods directly without naming them first.
3307:
3308: @cindex superclass binding
3309: @cindex parent class binding
3310: A frequent use of class binding is this: When we define a method
3311: for a selector, we often want the method to do what the selector does
3312: in the parent class, and a little more. There is a special word for
3313: this purpose: @code{[parent]}; @code{[parent]
3314: @emph{selector}} is equivalent to @code{[bind] @emph{parent
3315: selector}}, where @code{@emph{parent}} is the parent
3316: class of the current class. E.g., a method definition might look like:
3317:
3318: @cindex @code{[parent]} usage
3319: @example
3320: :noname
3321: dup [parent] foo \ do parent's foo on the receiving object
3322: ... \ do some more
3323: ; overrides foo
3324: @end example
3325:
3326: @cindex class binding as optimization
3327: In @cite{Object-oriented programming in ANS Forth} (Forth Dimensions,
3328: March 1997), Andrew McKewan presents class binding as an optimization
3329: technique. I recommend not using it for this purpose unless you are in
3330: an emergency. Late binding is pretty fast with this model anyway, so the
3331: benefit of using class binding is small; the cost of using class binding
3332: where it is not appropriate is reduced maintainability.
3333:
3334: While we are at programming style questions: You should bind
3335: selectors only to ancestor classes of the receiving object. E.g., say,
3336: you know that the receiving object is of class @code{foo} or its
3337: descendents; then you should bind only to @code{foo} and its
3338: ancestors.
3339:
3340: @node Method conveniences, Classes and Scoping, Class Binding, Objects
3341: @subsection Method conveniences
3342: @cindex method conveniences
3343:
3344: In a method you usually access the receiving object pretty often. If
3345: you define the method as a plain colon definition (e.g., with
3346: @code{:noname}), you may have to do a lot of stack
3347: gymnastics. To avoid this, you can define the method with @code{m:
3348: ... ;m}. E.g., you could define the method for
3349: @code{draw}ing a @code{circle} with
3350:
3351: @cindex @code{this} usage
3352: @cindex @code{m:} usage
3353: @cindex @code{;m} usage
3354: @example
3355: m: ( x y circle -- )
3356: ( x y ) this circle-radius @@ draw-circle ;m
3357: @end example
3358:
3359: @cindex @code{exit} in @code{m: ... ;m}
3360: @cindex @code{exitm} discussion
3361: @cindex @code{catch} in @code{m: ... ;m}
3362: When this method is executed, the receiver object is removed from the
3363: stack; you can access it with @code{this} (admittedly, in this
3364: example the use of @code{m: ... ;m} offers no advantage). Note
3365: that I specify the stack effect for the whole method (i.e. including
3366: the receiver object), not just for the code between @code{m:}
3367: and @code{;m}. You cannot use @code{exit} in
3368: @code{m:...;m}; instead, use
3369: @code{exitm}.@footnote{Moreover, for any word that calls
3370: @code{catch} and was defined before loading
3371: @code{objects.fs}, you have to redefine it like I redefined
3372: @code{catch}: @code{: catch this >r catch r> to-this ;}}
3373:
3374: @cindex @code{inst-var} usage
3375: You will frequently use sequences of the form @code{this
3376: @emph{field}} (in the example above: @code{this
3377: circle-radius}). If you use the field only in this way, you can
3378: define it with @code{inst-var} and eliminate the
3379: @code{this} before the field name. E.g., the @code{circle}
3380: class above could also be defined with:
3381:
3382: @example
3383: graphical class
3384: cell% inst-var radius
3385:
3386: m: ( x y circle -- )
3387: radius @@ draw-circle ;m
3388: overrides draw
3389:
3390: m: ( n-radius circle -- )
3391: radius ! ;m
3392: overrides construct
3393:
3394: end-class circle
3395: @end example
3396:
3397: @code{radius} can only be used in @code{circle} and its
3398: descendent classes and inside @code{m:...;m}.
3399:
3400: @cindex @code{inst-value} usage
3401: You can also define fields with @code{inst-value}, which is
3402: to @code{inst-var} what @code{value} is to
3403: @code{variable}. You can change the value of such a field with
3404: @code{[to-inst]}. E.g., we could also define the class
3405: @code{circle} like this:
3406:
3407: @example
3408: graphical class
3409: inst-value radius
3410:
3411: m: ( x y circle -- )
3412: radius draw-circle ;m
3413: overrides draw
3414:
3415: m: ( n-radius circle -- )
3416: [to-inst] radius ;m
3417: overrides construct
3418:
3419: end-class circle
3420: @end example
3421:
3422:
3423: @node Classes and Scoping, Object Interfaces, Method conveniences, Objects
3424: @subsection Classes and Scoping
3425: @cindex classes and scoping
3426: @cindex scoping and classes
3427:
3428: Inheritance is frequent, unlike structure extension. This exacerbates
3429: the problem with the field name convention (@pxref{Structure Naming
3430: Convention}): One always has to remember in which class the field was
3431: originally defined; changing a part of the class structure would require
3432: changes for renaming in otherwise unaffected code.
3433:
3434: @cindex @code{inst-var} visibility
3435: @cindex @code{inst-value} visibility
3436: To solve this problem, I added a scoping mechanism (which was not in my
3437: original charter): A field defined with @code{inst-var} (or
3438: @code{inst-value}) is visible only in the class where it is defined and in
3439: the descendent classes of this class. Using such fields only makes
3440: sense in @code{m:}-defined methods in these classes anyway.
3441:
3442: This scoping mechanism allows us to use the unadorned field name,
3443: because name clashes with unrelated words become much less likely.
3444:
3445: @cindex @code{protected} discussion
3446: @cindex @code{private} discussion
3447: Once we have this mechanism, we can also use it for controlling the
3448: visibility of other words: All words defined after
3449: @code{protected} are visible only in the current class and its
3450: descendents. @code{public} restores the compilation
3451: (i.e. @code{current}) wordlist that was in effect before. If you
3452: have several @code{protected}s without an intervening
3453: @code{public} or @code{set-current}, @code{public}
3454: will restore the compilation wordlist in effect before the first of
3455: these @code{protected}s.
3456:
3457: @node Object Interfaces, Objects Implementation, Classes and Scoping, Objects
3458: @subsection Object Interfaces
3459: @cindex object interfaces
3460: @cindex interfaces for objects
3461:
3462: In this model you can only call selectors defined in the class of the
3463: receiving objects or in one of its ancestors. If you call a selector
3464: with a receiving object that is not in one of these classes, the
3465: result is undefined; if you are lucky, the program crashes
3466: immediately.
3467:
3468: @cindex selectors common to hardly-related classes
3469: Now consider the case when you want to have a selector (or several)
3470: available in two classes: You would have to add the selector to a
3471: common ancestor class, in the worst case to @code{object}. You
3472: may not want to do this, e.g., because someone else is responsible for
3473: this ancestor class.
3474:
3475: The solution for this problem is interfaces. An interface is a
3476: collection of selectors. If a class implements an interface, the
3477: selectors become available to the class and its descendents. A class
3478: can implement an unlimited number of interfaces. For the problem
3479: discussed above, we would define an interface for the selector(s), and
3480: both classes would implement the interface.
3481:
3482: As an example, consider an interface @code{storage} for
3483: writing objects to disk and getting them back, and a class
3484: @code{foo} foo that implements it. The code for this would look
3485: like this:
3486:
3487: @cindex @code{interface} usage
3488: @cindex @code{end-interface} usage
3489: @cindex @code{implementation} usage
3490: @example
3491: interface
3492: selector write ( file object -- )
3493: selector read1 ( file object -- )
3494: end-interface storage
3495:
3496: bar class
3497: storage implementation
3498:
3499: ... overrides write
3500: ... overrides read
3501: ...
3502: end-class foo
3503: @end example
3504:
3505: (I would add a word @code{read} ( file -- object ) that uses
3506: @code{read1} internally, but that's beyond the point illustrated
3507: here.)
3508:
3509: Note that you cannot use @code{protected} in an interface; and
3510: of course you cannot define fields.
3511:
3512: In the Neon model, all selectors are available for all classes;
3513: therefore it does not need interfaces. The price you pay in this model
3514: is slower late binding, and therefore, added complexity to avoid late
3515: binding.
3516:
3517: @node Objects Implementation, Comparison with other object models, Object Interfaces, Objects
3518: @subsection @file{objects.fs} Implementation
3519: @cindex @file{objects.fs} implementation
3520:
3521: @cindex @code{object-map} discussion
3522: An object is a piece of memory, like one of the data structures
3523: described with @code{struct...end-struct}. It has a field
3524: @code{object-map} that points to the method map for the object's
3525: class.
3526:
3527: @cindex method map
3528: @cindex virtual function table
3529: The @emph{method map}@footnote{This is Self terminology; in C++
3530: terminology: virtual function table.} is an array that contains the
3531: execution tokens (XTs) of the methods for the object's class. Each
3532: selector contains an offset into the method maps.
3533:
3534: @cindex @code{selector} implementation, class
3535: @code{selector} is a defining word that uses
3536: @code{create} and @code{does>}. The body of the
3537: selector contains the offset; the @code{does>} action for a
3538: class selector is, basically:
3539:
3540: @example
3541: ( object addr ) @@ over object-map @@ + @@ execute
3542: @end example
3543:
3544: Since @code{object-map} is the first field of the object, it
3545: does not generate any code. As you can see, calling a selector has a
3546: small, constant cost.
3547:
3548: @cindex @code{current-interface} discussion
3549: @cindex class implementation and representation
3550: A class is basically a @code{struct} combined with a method
3551: map. During the class definition the alignment and size of the class
3552: are passed on the stack, just as with @code{struct}s, so
3553: @code{field} can also be used for defining class
3554: fields. However, passing more items on the stack would be
3555: inconvenient, so @code{class} builds a data structure in memory,
3556: which is accessed through the variable
3557: @code{current-interface}. After its definition is complete, the
3558: class is represented on the stack by a pointer (e.g., as parameter for
3559: a child class definition).
3560:
3561: At the start, a new class has the alignment and size of its parent,
3562: and a copy of the parent's method map. Defining new fields extends the
3563: size and alignment; likewise, defining new selectors extends the
3564: method map. @code{overrides} just stores a new XT in the method
3565: map at the offset given by the selector.
3566:
3567: @cindex class binding, implementation
3568: Class binding just gets the XT at the offset given by the selector
3569: from the class's method map and @code{compile,}s (in the case of
3570: @code{[bind]}) it.
3571:
3572: @cindex @code{this} implementation
3573: @cindex @code{catch} and @code{this}
3574: @cindex @code{this} and @code{catch}
3575: I implemented @code{this} as a @code{value}. At the
3576: start of an @code{m:...;m} method the old @code{this} is
3577: stored to the return stack and restored at the end; and the object on
3578: the TOS is stored @code{TO this}. This technique has one
3579: disadvantage: If the user does not leave the method via
3580: @code{;m}, but via @code{throw} or @code{exit},
3581: @code{this} is not restored (and @code{exit} may
3582: crash). To deal with the @code{throw} problem, I have redefined
3583: @code{catch} to save and restore @code{this}; the same
3584: should be done with any word that can catch an exception. As for
3585: @code{exit}, I simply forbid it (as a replacement, there is
3586: @code{exitm}).
3587:
3588: @cindex @code{inst-var} implementation
3589: @code{inst-var} is just the same as @code{field}, with
3590: a different @code{does>} action:
3591: @example
3592: @@ this +
3593: @end example
3594: Similar for @code{inst-value}.
3595:
3596: @cindex class scoping implementation
3597: Each class also has a wordlist that contains the words defined with
3598: @code{inst-var} and @code{inst-value}, and its protected
3599: words. It also has a pointer to its parent. @code{class} pushes
3600: the wordlists of the class an all its ancestors on the search order,
3601: and @code{end-class} drops them.
3602:
3603: @cindex interface implementation
3604: An interface is like a class without fields, parent and protected
3605: words; i.e., it just has a method map. If a class implements an
3606: interface, its method map contains a pointer to the method map of the
3607: interface. The positive offsets in the map are reserved for class
3608: methods, therefore interface map pointers have negative
3609: offsets. Interfaces have offsets that are unique throughout the
3610: system, unlike class selectors, whose offsets are only unique for the
3611: classes where the selector is available (invokable).
3612:
3613: This structure means that interface selectors have to perform one
3614: indirection more than class selectors to find their method. Their body
3615: contains the interface map pointer offset in the class method map, and
3616: the method offset in the interface method map. The
3617: @code{does>} action for an interface selector is, basically:
3618:
3619: @example
3620: ( object selector-body )
3621: 2dup selector-interface @@ ( object selector-body object interface-offset )
3622: swap object-map @@ + @@ ( object selector-body map )
3623: swap selector-offset @@ + @@ execute
3624: @end example
3625:
3626: where @code{object-map} and @code{selector-offset} are
3627: first fields and generate no code.
3628:
3629: As a concrete example, consider the following code:
3630:
3631: @example
3632: interface
3633: selector if1sel1
3634: selector if1sel2
3635: end-interface if1
3636:
3637: object class
3638: if1 implementation
3639: selector cl1sel1
3640: cell% inst-var cl1iv1
3641:
3642: ' m1 overrides construct
3643: ' m2 overrides if1sel1
3644: ' m3 overrides if1sel2
3645: ' m4 overrides cl1sel2
3646: end-class cl1
3647:
3648: create obj1 object dict-new drop
3649: create obj2 cl1 dict-new drop
3650: @end example
3651:
3652: The data structure created by this code (including the data structure
3653: for @code{object}) is shown in the <a
3654: href="objects-implementation.eps">figure</a>, assuming a cell size of 4.
3655:
3656: @node Comparison with other object models, Objects Glossary, Objects Implementation, Objects
3657: @subsection Comparison with other object models
3658: @cindex comparison of object models
3659: @cindex object models, comparison
3660:
3661: Many object-oriented Forth extensions have been proposed (@cite{A survey
3662: of object-oriented Forths} (SIGPLAN Notices, April 1996) by Bradford
3663: J. Rodriguez and W. F. S. Poehlman lists 17). Here I'll discuss the
3664: relation of @file{objects.fs} to two well-known and two closely-related
3665: (by the use of method maps) models.
3666:
3667: @cindex Neon model
3668: The most popular model currently seems to be the Neon model (see
3669: @cite{Object-oriented programming in ANS Forth} (Forth Dimensions, March
3670: 1997) by Andrew McKewan). The Neon model uses a @code{@emph{selector
3671: object}} syntax, which makes it unnatural to pass objects on the
3672: stack. It also requires that the selector parses the input stream (at
3673: compile time); this leads to reduced extensibility and to bugs that are
3674: hard to find. Finally, it allows using every selector to every object;
3675: this eliminates the need for classes, but makes it harder to create
3676: efficient implementations. A longer version of this critique can be
3677: found in @cite{On Standardizing Object-Oriented Forth Extensions} (Forth
3678: Dimensions, May 1997) by Anton Ertl.
3679:
3680: @cindex Pountain's object-oriented model
3681: Another well-known publication is @cite{Object-Oriented Forth} (Academic
3682: Press, London, 1987) by Dick Pountain. However, it is not really about
3683: object-oriented programming, because it hardly deals with late
3684: binding. Instead, it focuses on features like information hiding and
3685: overloading that are characteristic of modular languages like Ada (83).
3686:
3687: @cindex Zsoter's object-oriented model
3688: In @cite{Does late binding have to be slow?} (Forth Dimensions ??? 1996)
3689: Andras Zsoter describes a model that makes heavy use of an active object
3690: (like @code{this} in @file{objects.fs}): The active object is not only
3691: used for accessing all fields, but also specifies the receiving object
3692: of every selector invocation; you have to change the active object
3693: explicitly with @code{@{ ... @}}, whereas in @file{objects.fs} it
3694: changes more or less implicitly at @code{m: ... ;m}. Such a change at
3695: the method entry point is unnecessary with the Zsoter's model, because
3696: the receiving object is the active object already; OTOH, the explicit
3697: change is absolutely necessary in that model, because otherwise no one
3698: could ever change the active object. An ANS Forth implementation of this
3699: model is available at @url{http://www.forth.org/fig/oopf.html}.
3700:
3701: @cindex @file{oof.fs} object model
3702: The @file{oof.fs} model combines information hiding and overloading
3703: resolution (by keeping names in various wordlists) with object-oriented
3704: programming. It sets the active object implicitly on method entry, but
3705: also allows explicit changing (with @code{>o...o>} or with
3706: @code{with...endwith}). It uses parsing and state-smart objects and
3707: classes for resolving overloading and for early binding: the object or
3708: class parses the selector and determines the method from this. If the
3709: selector is not parsed by an object or class, it performs a call to the
3710: selector for the active object (late binding), like Zsoter's model.
3711: Fields are always accessed through the active object. The big
3712: disadvantage of this model is the parsing and the state-smartness, which
3713: reduces extensibility and increases the opportunities for subtle bugs;
3714: essentially, you are only safe if you never tick or @code{postpone} an
3715: object or class.
3716:
3717: @node Objects Glossary, , Comparison with other object models, Objects
3718: @subsection @file{objects.fs} Glossary
3719: @cindex @file{objects.fs} Glossary
3720:
3721: doc-bind
3722: doc-<bind>
3723: doc-bind'
3724: doc-[bind]
3725: doc-class
3726: doc-class->map
3727: doc-class-inst-size
3728: doc-class-override!
3729: doc-construct
3730: doc-current'
3731: doc-[current]
3732: doc-current-interface
3733: doc-dict-new
3734: doc-drop-order
3735: doc-end-class
3736: doc-end-class-noname
3737: doc-end-interface
3738: doc-end-interface-noname
3739: doc-exitm
3740: doc-heap-new
3741: doc-implementation
3742: doc-init-object
3743: doc-inst-value
3744: doc-inst-var
3745: doc-interface
3746: doc-;m
3747: doc-m:
3748: doc-method
3749: doc-object
3750: doc-overrides
3751: doc-[parent]
3752: doc-print
3753: doc-protected
3754: doc-public
3755: doc-push-order
3756: doc-selector
3757: doc-this
3758: doc-<to-inst>
3759: doc-[to-inst]
3760: doc-to-this
3761: doc-xt-new
3762:
3763: @c -------------------------------------------------------------
3764: @node Object Oriented Forth, Mini-OOF, Objects, Words
3765: @section Object oriented Forth
3766: @cindex oof
3767: @cindex object-oriented programming
3768:
3769: @cindex @file{objects.fs}
3770: @cindex @file{oof.fs}
3771: Gforth comes with two packets for object-oriented programming,
3772: @file{objects.fs} and @file{oof.fs}; none of them is preloaded, so you
3773: have to @code{include} them before use. This section describes the
3774: @file{oof.fs} packet. Both packets are written in ANS Forth and can be
3775: used with any other standard Forth (@pxref{Objects}). This section uses
3776: the same rationale why using object-oriented programming, and the same
3777: terminology.
3778:
3779: The packet described in this section is used in bigFORTH since 1991, and
3780: used for two large applications: a chromatographic system used to
3781: create new medicaments, and a graphic user interface library (MINOS).
3782:
3783: @menu
3784: * Properties of the OOF model::
3785: * Basic OOF Usage::
3786: * The base class object::
3787: * Class Declaration::
3788: * Class Implementation::
3789: @end menu
3790:
3791: @node Properties of the OOF model, Basic OOF Usage, Object Oriented Forth, Object Oriented Forth
3792: @subsection Properties of the OOF model
3793: @cindex @file{oof.fs} properties
3794:
3795: @itemize @bullet
3796: @item
3797: This model combines object oriented programming with information
3798: hiding. It helps you writing large application, where scoping is
3799: necessary, because it provides class-oriented scoping.
3800:
3801: @item
3802: Named objects, object pointers, and object arrays can be created,
3803: selector invocation uses the "object selector" syntax. Selector invocation
3804: to objects and/or selectors on the stack is a bit less convenient, but
3805: possible.
3806:
3807: @item
3808: Selector invocation and instance variable usage of the active object is
3809: straight forward, since both make use of the active object.
3810:
3811: @item
3812: Late binding is efficient and easy to use.
3813:
3814: @item
3815: State-smart objects parse selectors. However, extensibility is provided
3816: using a (parsing) selector @code{postpone} and a selector @code{'}.
3817:
3818: @item
3819: An implementation in ANS Forth is available.
3820:
3821: @end itemize
3822:
3823:
3824: @node Basic OOF Usage, The base class object, Properties of the OOF model, Object Oriented Forth
3825: @subsection Basic OOF Usage
3826: @cindex @file{oof.fs} usage
3827:
3828: Here, I use the same example as for @code{objects} (@pxref{Basic Objects Usage}).
3829:
3830: You can define a class for graphical objects like this:
3831:
3832: @cindex @code{class} usage
3833: @cindex @code{class;} usage
3834: @cindex @code{method} usage
3835: @example
3836: object class graphical \ "object" is the parent class
3837: method draw ( x y graphical -- )
3838: class;
3839: @end example
3840:
3841: This code defines a class @code{graphical} with an
3842: operation @code{draw}. We can perform the operation
3843: @code{draw} on any @code{graphical} object, e.g.:
3844:
3845: @example
3846: 100 100 t-rex draw
3847: @end example
3848:
3849: where @code{t-rex} is an object or object pointer, created with e.g.
3850: @code{graphical : trex}.
3851:
3852: @cindex abstract class
3853: How do we create a graphical object? With the present definitions,
3854: we cannot create a useful graphical object. The class
3855: @code{graphical} describes graphical objects in general, but not
3856: any concrete graphical object type (C++ users would call it an
3857: @emph{abstract class}); e.g., there is no method for the selector
3858: @code{draw} in the class @code{graphical}.
3859:
3860: For concrete graphical objects, we define child classes of the
3861: class @code{graphical}, e.g.:
3862:
3863: @example
3864: graphical class circle \ "graphical" is the parent class
3865: cell var circle-radius
3866: how:
3867: : draw ( x y -- )
3868: circle-radius @@ draw-circle ;
3869:
3870: : init ( n-radius -- (
3871: circle-radius ! ;
3872: class;
3873: @end example
3874:
3875: Here we define a class @code{circle} as a child of @code{graphical},
3876: with a field @code{circle-radius}; it defines new methods for the
3877: selectors @code{draw} and @code{init} (@code{init} is defined in
3878: @code{object}, the parent class of @code{graphical}).
3879:
3880: Now we can create a circle in the dictionary with
3881:
3882: @example
3883: 50 circle : my-circle
3884: @end example
3885:
3886: @code{:} invokes @code{init}, thus initializing the field
3887: @code{circle-radius} with 50. We can draw this new circle at (100,100)
3888: with
3889:
3890: @example
3891: 100 100 my-circle draw
3892: @end example
3893:
3894: @cindex selector invocation, restrictions
3895: @cindex class definition, restrictions
3896: Note: You can invoke a selector only if the receiving object belongs to
3897: the class where the selector was defined or one of its descendents;
3898: e.g., you can invoke @code{draw} only for objects belonging to
3899: @code{graphical} or its descendents (e.g., @code{circle}). The scoping
3900: mechanism will check if you try to invoke a selector that is not
3901: defined in this class hierarchy, so you'll get an error at compilation
3902: time.
3903:
3904:
3905: @node The base class object, Class Declaration, Basic OOF Usage, Object Oriented Forth
3906: @subsection The base class @file{object}
3907: @cindex @file{oof.fs} base class
3908:
3909: When you define a class, you have to specify a parent class. So how do
3910: you start defining classes? There is one class available from the start:
3911: @code{object}. You have to use it as ancestor for all classes. It is the
3912: only class that has no parent. Classes are also objects, except that
3913: they don't have instance variables; class manipulation such as
3914: inheritance or changing definitions of a class is handled through
3915: selectors of the class @code{object}.
3916:
3917: @code{object} provides a number of selectors:
3918:
3919: @itemize @bullet
3920: @item
3921: @code{class} for subclassing, @code{definitions} to add definitions
3922: later on, and @code{class?} to get type informations (is the class a
3923: subclass of the class passed on the stack?).
3924: doc---object-class
3925: doc---object-definitions
3926: doc---object-class?
3927:
3928: @item
3929: @code{init} and @code{dispose} as constructor and destroctor of the
3930: object. @code{init} is invocated after the object's memory is allocated,
3931: while @code{dispose} also handles deallocation. Thus if you redefine
3932: @code{dispose}, you have to call the parent's dispose with @code{super
3933: dispose}, too.
3934: doc---object-init
3935: doc---object-dispose
3936:
3937: @item
3938: @code{new}, @code{new[]}, @code{:}, @code{ptr}, @code{asptr}, and
3939: @code{[]} to create named and unnamed objects and object arrays or
3940: object pointers.
3941: doc---object-new
3942: doc---object-new[]
3943: doc---object-:
3944: doc---object-ptr
3945: doc---object-asptr
3946: doc---object-[]
3947:
3948: @item
3949: @code{::} and @code{super} for expicit scoping. You should use expicit
3950: scoping only for super classes or classes with the same set of instance
3951: variables. Explicit scoped selectors use early binding.
3952: doc---object-::
3953: doc---object-super
3954:
3955: @item
3956: @code{self} to get the address of the object
3957: doc---object-self
3958:
3959: @item
3960: @code{bind}, @code{bound}, @code{link}, and @code{is} to assign object
3961: pointers and instance defers.
3962: doc---object-bind
3963: doc---object-bound
3964: doc---object-link
3965: doc---object-is
3966:
3967: @item
3968: @code{'} to obtain selector tokens, @code{send} to invocate selectors
3969: form the stack, and @code{postpone} to generate selector invocation code.
3970: doc---object-'
3971: doc---object-postpone
3972:
3973: @item
3974: @code{with} and @code{endwith} to select the active object from the
3975: stack, and enabling it's scope. Using @code{with} and @code{endwith}
3976: also allows to create code using selector @code{postpone} without being
3977: trapped bye the state-smart objects.
3978: doc---object-with
3979: doc---object-endwith
3980:
3981: @end itemize
3982:
3983: @node Class Declaration, Class Implementation, The base class object, Object Oriented Forth
3984: @subsection Class Declaration
3985: @cindex class declaration
3986:
3987: @itemize @bullet
3988: @item
3989: Instance variables
3990: doc---oof-var
3991:
3992: @item
3993: Object pointers
3994: doc---oof-ptr
3995: doc---oof-asptr
3996:
3997: @item
3998: Instance defers
3999: doc---oof-defer
4000:
4001: @item
4002: Method selectors
4003: doc---oof-early
4004: doc---oof-method
4005:
4006: @item
4007: Class wide variables
4008: doc---oof-static
4009:
4010: @item
4011: End declaration
4012: doc---oof-how:
4013: doc---oof-class;
4014:
4015: @end itemize
4016:
4017: @node Class Implementation, , Class Declaration, Object Oriented Forth
4018: @subsection Class Implementation
4019: @cindex class implementation
4020:
4021: @node Mini-OOF, Tokens for Words, Object Oriented Forth, Words
4022: @section Mini-OOF
4023: @cindex mini-oof
4024:
4025: Gforth's third object oriented Forth package is a 12-liner. It uses a
4026: bit of a mixture of the @file{object.fs} and the @file{oof.fs} syntax,
4027: and reduces to the bare minimum of features.
4028:
4029: @example
4030: : method ( m v -- m' v ) Create over , swap cell+ swap
4031: DOES> ( ... o -- ... ) @ over @ + @ execute ;
4032: : var ( m v size -- m v' ) Create over , +
4033: DOES> ( o -- addr ) @ + ;
4034: : class ( class -- class methods vars ) dup 2@ ;
4035: : end-class ( class methods vars -- )
4036: Create here >r , dup , 2 cells ?DO ['] noop , cell +LOOP
4037: cell+ dup cell+ swap @ 2 - cells r> 2 cells + swap move ;
4038: : defines ( xt class -- ) ' >body @ + ! ;
4039: : new ( class -- o ) here over @ allot swap over ! ;
4040: : :: ( class "name" -- ) ' >body @ + @ compile, ;
4041: Create object 1 cells , 2 cells ,
4042: @end example
4043:
4044: @c -------------------------------------------------------------
4045: @node Tokens for Words, Wordlists, Mini-OOF, Words
4046: @section Tokens for Words
4047: @cindex tokens for words
4048:
4049: This chapter describes the creation and use of tokens that represent
4050: words on the stack (and in data space).
4051:
4052: Named words have interpretation and compilation semantics. Unnamed words
4053: just have execution semantics.
4054:
4055: @cindex execution token
4056: An @dfn{execution token} represents the execution semantics of an
4057: unnamed word. An execution token occupies one cell. As explained in
4058: section @ref{Supplying names}, the execution token of the last words
4059: defined can be produced with
4060:
4061: short-lastxt
4062:
4063: You can perform the semantics represented by an execution token with
4064: doc-execute
4065: You can compile the word with
4066: doc-compile,
4067:
4068: @cindex code field address
4069: @cindex CFA
4070: In Gforth, the abstract data type @emph{execution token} is implemented
4071: as CFA (code field address).
4072:
4073: The interpretation semantics of a named word are also represented by an
4074: execution token. You can get it with
4075:
4076: doc-[']
4077: doc-'
4078:
4079: For literals, you use @code{'} in interpreted code and @code{[']} in
4080: compiled code. Gforth's @code{'} and @code{[']} behave somewhat unusual
4081: by complaining about compile-only words. To get an execution token for a
4082: compiling word @var{X}, use @code{COMP' @var{X} drop} or @code{[COMP']
4083: @var{X} drop}.
4084:
4085: @cindex compilation token
4086: The compilation semantics are represented by a @dfn{compilation token}
4087: consisting of two cells: @var{w xt}. The top cell @var{xt} is an
4088: execution token. The compilation semantics represented by the
4089: compilation token can be performed with @code{execute}, which consumes
4090: the whole compilation token, with an additional stack effect determined
4091: by the represented compilation semantics.
4092:
4093: doc-[comp']
4094: doc-comp'
4095:
4096: You can compile the compilation semantics with @code{postpone,}. I.e.,
4097: @code{COMP' @var{word} POSTPONE,} is equivalent to @code{POSTPONE
4098: @var{word}}.
4099:
4100: doc-postpone,
4101:
4102: At present, the @var{w} part of a compilation token is an execution
4103: token, and the @var{xt} part represents either @code{execute} or
4104: @code{compile,}. However, don't rely on that knowledge, unless necessary;
4105: we may introduce unusual compilation tokens in the future (e.g.,
4106: compilation tokens representing the compilation semantics of literals).
4107:
4108: @cindex name token
4109: @cindex name field address
4110: @cindex NFA
4111: Named words are also represented by the @dfn{name token}. The abstract
4112: data type @emph{name token} is implemented as NFA (name field address).
4113:
4114: doc-find-name
4115: doc-name>int
4116: doc-name?int
4117: doc-name>comp
4118: doc-name>string
4119:
4120: @node Wordlists, Files, Tokens for Words, Words
4121: @section Wordlists
4122:
4123: @node Files, Blocks, Wordlists, Words
4124: @section Files
4125:
4126: @node Blocks, Other I/O, Files, Words
4127: @section Blocks
4128:
4129: @node Other I/O, Programming Tools, Blocks, Words
4130: @section Other I/O
4131:
4132: @node Programming Tools, Assembler and Code Words, Other I/O, Words
4133: @section Programming Tools
4134: @cindex programming tools
4135:
4136: @menu
4137: * Debugging:: Simple and quick.
4138: * Assertions:: Making your programs self-checking.
4139: * Singlestep Debugger:: Executing your program word by word.
4140: @end menu
4141:
4142: @node Debugging, Assertions, Programming Tools, Programming Tools
4143: @subsection Debugging
4144: @cindex debugging
4145:
4146: The simple debugging aids provided in @file{debugs.fs}
4147: are meant to support a different style of debugging than the
4148: tracing/stepping debuggers used in languages with long turn-around
4149: times.
4150:
4151: A much better (faster) way in fast-compiling languages is to add
4152: printing code at well-selected places, let the program run, look at
4153: the output, see where things went wrong, add more printing code, etc.,
4154: until the bug is found.
4155:
4156: The word @code{~~} is easy to insert. It just prints debugging
4157: information (by default the source location and the stack contents). It
4158: is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
4159: query-replace them with nothing). The deferred words
4160: @code{printdebugdata} and @code{printdebugline} control the output of
4161: @code{~~}. The default source location output format works well with
4162: Emacs' compilation mode, so you can step through the program at the
4163: source level using @kbd{C-x `} (the advantage over a stepping debugger
4164: is that you can step in any direction and you know where the crash has
4165: happened or where the strange data has occurred).
4166:
4167: Note that the default actions clobber the contents of the pictured
4168: numeric output string, so you should not use @code{~~}, e.g., between
4169: @code{<#} and @code{#>}.
4170:
4171: doc-~~
4172: doc-printdebugdata
4173: doc-printdebugline
4174:
4175: @node Assertions, Singlestep Debugger, Debugging, Programming Tools
4176: @subsection Assertions
4177: @cindex assertions
4178:
4179: It is a good idea to make your programs self-checking, in particular, if
4180: you use an assumption (e.g., that a certain field of a data structure is
4181: never zero) that may become wrong during maintenance. Gforth supports
4182: assertions for this purpose. They are used like this:
4183:
4184: @example
4185: assert( @var{flag} )
4186: @end example
4187:
4188: The code between @code{assert(} and @code{)} should compute a flag, that
4189: should be true if everything is alright and false otherwise. It should
4190: not change anything else on the stack. The overall stack effect of the
4191: assertion is @code{( -- )}. E.g.
4192:
4193: @example
4194: assert( 1 1 + 2 = ) \ what we learn in school
4195: assert( dup 0<> ) \ assert that the top of stack is not zero
4196: assert( false ) \ this code should not be reached
4197: @end example
4198:
4199: The need for assertions is different at different times. During
4200: debugging, we want more checking, in production we sometimes care more
4201: for speed. Therefore, assertions can be turned off, i.e., the assertion
4202: becomes a comment. Depending on the importance of an assertion and the
4203: time it takes to check it, you may want to turn off some assertions and
4204: keep others turned on. Gforth provides several levels of assertions for
4205: this purpose:
4206:
4207: doc-assert0(
4208: doc-assert1(
4209: doc-assert2(
4210: doc-assert3(
4211: doc-assert(
4212: doc-)
4213:
4214: @code{Assert(} is the same as @code{assert1(}. The variable
4215: @code{assert-level} specifies the highest assertions that are turned
4216: on. I.e., at the default @code{assert-level} of one, @code{assert0(} and
4217: @code{assert1(} assertions perform checking, while @code{assert2(} and
4218: @code{assert3(} assertions are treated as comments.
4219:
4220: Note that the @code{assert-level} is evaluated at compile-time, not at
4221: run-time. I.e., you cannot turn assertions on or off at run-time, you
4222: have to set the @code{assert-level} appropriately before compiling a
4223: piece of code. You can compile several pieces of code at several
4224: @code{assert-level}s (e.g., a trusted library at level 1 and newly
4225: written code at level 3).
4226:
4227: doc-assert-level
4228:
4229: If an assertion fails, a message compatible with Emacs' compilation mode
4230: is produced and the execution is aborted (currently with @code{ABORT"}.
4231: If there is interest, we will introduce a special throw code. But if you
4232: intend to @code{catch} a specific condition, using @code{throw} is
4233: probably more appropriate than an assertion).
4234:
4235: @node Singlestep Debugger, , Assertions, Programming Tools
4236: @subsection Singlestep Debugger
4237: @cindex singlestep Debugger
4238: @cindex debugging Singlestep
4239: @cindex @code{dbg}
4240: @cindex @code{BREAK:}
4241: @cindex @code{BREAK"}
4242:
4243: When a new word is created there's often the need to check whether it behaves
4244: correctly or not. You can do this by typing @code{dbg badword}. This might
4245: look like:
4246: @example
4247: : badword 0 DO i . LOOP ; ok
4248: 2 dbg badword
4249: : badword
4250: Scanning code...
4251:
4252: Nesting debugger ready!
4253:
4254: 400D4738 8049BC4 0 -> [ 2 ] 00002 00000
4255: 400D4740 8049F68 DO -> [ 0 ]
4256: 400D4744 804A0C8 i -> [ 1 ] 00000
4257: 400D4748 400C5E60 . -> 0 [ 0 ]
4258: 400D474C 8049D0C LOOP -> [ 0 ]
4259: 400D4744 804A0C8 i -> [ 1 ] 00001
4260: 400D4748 400C5E60 . -> 1 [ 0 ]
4261: 400D474C 8049D0C LOOP -> [ 0 ]
4262: 400D4758 804B384 ; -> ok
4263: @end example
4264:
4265: Each line displayed is one step. You always have to hit return to
4266: execute the next word that is displayed. If you don't want to execute
4267: the next word in a whole, you have to type @kbd{n} for @code{nest}. Here is
4268: an overview what keys are available:
4269:
4270: @table @i
4271:
4272: @item <return>
4273: Next; Execute the next word.
4274:
4275: @item n
4276: Nest; Single step through next word.
4277:
4278: @item u
4279: Unnest; Stop debugging and execute rest of word. If we got to this word
4280: with nest, continue debugging with the calling word.
4281:
4282: @item d
4283: Done; Stop debugging and execute rest.
4284:
4285: @item s
4286: Stopp; Abort immediately.
4287:
4288: @end table
4289:
4290: Debugging large application with this mechanism is very difficult, because
4291: you have to nest very deep into the program before the interesting part
4292: begins. This takes a lot of time.
4293:
4294: To do it more directly put a @code{BREAK:} command into your source code.
4295: When program execution reaches @code{BREAK:} the single step debugger is
4296: invoked and you have all the features described above.
4297:
4298: If you have more than one part to debug it is useful to know where the
4299: program has stopped at the moment. You can do this by the
4300: @code{BREAK" string"} command. This behaves like @code{BREAK:} except that
4301: string is typed out when the ``breakpoint'' is reached.
4302:
4303: @node Assembler and Code Words, Threading Words, Programming Tools, Words
4304: @section Assembler and Code Words
4305: @cindex assembler
4306: @cindex code words
4307:
4308: Gforth provides some words for defining primitives (words written in
4309: machine code), and for defining the the machine-code equivalent of
4310: @code{DOES>}-based defining words. However, the machine-independent
4311: nature of Gforth poses a few problems: First of all, Gforth runs on
4312: several architectures, so it can provide no standard assembler. What's
4313: worse is that the register allocation not only depends on the processor,
4314: but also on the @code{gcc} version and options used.
4315:
4316: The words that Gforth offers encapsulate some system dependences (e.g., the
4317: header structure), so a system-independent assembler may be used in
4318: Gforth. If you do not have an assembler, you can compile machine code
4319: directly with @code{,} and @code{c,}.
4320:
4321: doc-assembler
4322: doc-code
4323: doc-end-code
4324: doc-;code
4325: doc-flush-icache
4326:
4327: If @code{flush-icache} does not work correctly, @code{code} words
4328: etc. will not work (reliably), either.
4329:
4330: These words are rarely used. Therefore they reside in @code{code.fs},
4331: which is usually not loaded (except @code{flush-icache}, which is always
4332: present). You can load them with @code{require code.fs}.
4333:
4334: @cindex registers of the inner interpreter
4335: In the assembly code you will want to refer to the inner interpreter's
4336: registers (e.g., the data stack pointer) and you may want to use other
4337: registers for temporary storage. Unfortunately, the register allocation
4338: is installation-dependent.
4339:
4340: The easiest solution is to use explicit register declarations
4341: (@pxref{Explicit Reg Vars, , Variables in Specified Registers, gcc.info,
4342: GNU C Manual}) for all of the inner interpreter's registers: You have to
4343: compile Gforth with @code{-DFORCE_REG} (configure option
4344: @code{--enable-force-reg}) and the appropriate declarations must be
4345: present in the @code{machine.h} file (see @code{mips.h} for an example;
4346: you can find a full list of all declarable register symbols with
4347: @code{grep register engine.c}). If you give explicit registers to all
4348: variables that are declared at the beginning of @code{engine()}, you
4349: should be able to use the other caller-saved registers for temporary
4350: storage. Alternatively, you can use the @code{gcc} option
4351: @code{-ffixed-REG} (@pxref{Code Gen Options, , Options for Code
4352: Generation Conventions, gcc.info, GNU C Manual}) to reserve a register
4353: (however, this restriction on register allocation may slow Gforth
4354: significantly).
4355:
4356: If this solution is not viable (e.g., because @code{gcc} does not allow
4357: you to explicitly declare all the registers you need), you have to find
4358: out by looking at the code where the inner interpreter's registers
4359: reside and which registers can be used for temporary storage. You can
4360: get an assembly listing of the engine's code with @code{make engine.s}.
4361:
4362: In any case, it is good practice to abstract your assembly code from the
4363: actual register allocation. E.g., if the data stack pointer resides in
4364: register @code{$17}, create an alias for this register called @code{sp},
4365: and use that in your assembly code.
4366:
4367: @cindex code words, portable
4368: Another option for implementing normal and defining words efficiently
4369: is: adding the wanted functionality to the source of Gforth. For normal
4370: words you just have to edit @file{primitives} (@pxref{Automatic
4371: Generation}), defining words (equivalent to @code{;CODE} words, for fast
4372: defined words) may require changes in @file{engine.c}, @file{kernal.fs},
4373: @file{prims2x.fs}, and possibly @file{cross.fs}.
4374:
4375:
4376: @node Threading Words, Including Files, Assembler and Code Words, Words
4377: @section Threading Words
4378: @cindex threading words
4379:
4380: @cindex code address
4381: These words provide access to code addresses and other threading stuff
4382: in Gforth (and, possibly, other interpretive Forths). It more or less
4383: abstracts away the differences between direct and indirect threading
4384: (and, for direct threading, the machine dependences). However, at
4385: present this wordset is still incomplete. It is also pretty low-level;
4386: some day it will hopefully be made unnecessary by an internals wordset
4387: that abstracts implementation details away completely.
4388:
4389: doc->code-address
4390: doc->does-code
4391: doc-code-address!
4392: doc-does-code!
4393: doc-does-handler!
4394: doc-/does-handler
4395:
4396: The code addresses produced by various defining words are produced by
4397: the following words:
4398:
4399: doc-docol:
4400: doc-docon:
4401: doc-dovar:
4402: doc-douser:
4403: doc-dodefer:
4404: doc-dofield:
4405:
4406: You can recognize words defined by a @code{CREATE}...@code{DOES>} word
4407: with @code{>DOES-CODE}. If the word was defined in that way, the value
4408: returned is different from 0 and identifies the @code{DOES>} used by the
4409: defining word.
4410:
4411: @node Including Files, Include and Require, Threading Words, Words
4412: @section Including Files
4413: @cindex including files
4414:
4415: @menu
4416: * Include and Require::
4417: * Path Handling::
4418: @end menu
4419:
4420: @node Include and Require, Path Handling, Including Files, Including Files
4421: @subsection Include and Requrie
4422:
4423: There a two words to include the source files more intelligently.
4424:
4425: doc-include
4426: doc-require
4427:
4428: @node Path Handling, , Include and Require, Including Files
4429: @subsection Path Handling
4430: @cindex path handling
4431:
4432: In larger program projects it is often neccassary to build up a structured
4433: directory tree. Standard Forth programs are somewhere more central because
4434: they must be accessed from some more other programs. To achieve this it is
4435: possible to manipulate the search path in which Gforth tries to find the
4436: source file.
4437:
4438: doc-fpath+
4439: doc-fpath=
4440: doc-.fpath
4441:
4442: Using fpath and require would look like:
4443:
4444: @example
4445:
4446: fpath= /usr/lib/forth/|./
4447:
4448: require timer.fs
4449:
4450: ...
4451:
4452: @end example
4453:
4454: @cindex ~+
4455: There is another nice feature which is similar to C's @code{include <...>}
4456: and @code{include "..."}. For example: You have a program seperated into
4457: several files in a subdirectory and you want to include some other files
4458: in this subdirectory from within the program. You have to tell Gforth that
4459: you are now looking relative to the directory the current file comes from.
4460: You can tell this Gforth by using the prefix @code{~+/} in front of the
4461: filename. It is also possible to add it to the search path.
4462:
4463: If you have the need to look for a file in the Forth search path, you could
4464: use this Gforth feature in your application.
4465:
4466: doc-open-fpath-file
4467:
4468: It is even possible to use your own search paths. Create a search path like
4469: this:
4470:
4471: @example
4472:
4473: Make a buffer for the path:
4474: create mypath 100 chars , \ maximum length (is checked)
4475: 0 , \ real len
4476: 100 chars allot \ space for path
4477:
4478: @end example
4479:
4480: You have the same functions for the forth search path in an generic version
4481: for different pathes.
4482:
4483: doc-path+
4484: doc-path=
4485: doc-.path
4486: doc-open-path-file
4487:
4488: @c ******************************************************************
4489: @node Tools, ANS conformance, Words, Top
4490: @chapter Tools
4491:
4492: @menu
4493: * ANS Report:: Report the words used, sorted by wordset.
4494: @end menu
4495:
4496: See also @ref{Emacs and Gforth}.
4497:
4498: @node ANS Report, , Tools, Tools
4499: @section @file{ans-report.fs}: Report the words used, sorted by wordset
4500: @cindex @file{ans-report.fs}
4501: @cindex report the words used in your program
4502: @cindex words used in your program
4503:
4504: If you want to label a Forth program as ANS Forth Program, you must
4505: document which wordsets the program uses; for extension wordsets, it is
4506: helpful to list the words the program requires from these wordsets
4507: (because Forth systems are allowed to provide only some words of them).
4508:
4509: The @file{ans-report.fs} tool makes it easy for you to determine which
4510: words from which wordset and which non-ANS words your application
4511: uses. You simply have to include @file{ans-report.fs} before loading the
4512: program you want to check. After loading your program, you can get the
4513: report with @code{print-ans-report}. A typical use is to run this as
4514: batch job like this:
4515: @example
4516: gforth ans-report.fs myprog.fs -e "print-ans-report bye"
4517: @end example
4518:
4519: The output looks like this (for @file{compat/control.fs}):
4520: @example
4521: The program uses the following words
4522: from CORE :
4523: : POSTPONE THEN ; immediate ?dup IF 0=
4524: from BLOCK-EXT :
4525: \
4526: from FILE :
4527: (
4528: @end example
4529:
4530: @subsection Caveats
4531:
4532: Note that @file{ans-report.fs} just checks which words are used, not whether
4533: they are used in an ANS Forth conforming way!
4534:
4535: Some words are defined in several wordsets in the
4536: standard. @file{ans-report.fs} reports them for only one of the
4537: wordsets, and not necessarily the one you expect. It depends on usage
4538: which wordset is the right one to specify. E.g., if you only use the
4539: compilation semantics of @code{S"}, it is a Core word; if you also use
4540: its interpretation semantics, it is a File word.
4541:
4542: @c ******************************************************************
4543: @node ANS conformance, Model, Tools, Top
4544: @chapter ANS conformance
4545: @cindex ANS conformance of Gforth
4546:
4547: To the best of our knowledge, Gforth is an
4548:
4549: ANS Forth System
4550: @itemize @bullet
4551: @item providing the Core Extensions word set
4552: @item providing the Block word set
4553: @item providing the Block Extensions word set
4554: @item providing the Double-Number word set
4555: @item providing the Double-Number Extensions word set
4556: @item providing the Exception word set
4557: @item providing the Exception Extensions word set
4558: @item providing the Facility word set
4559: @item providing @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
4560: @item providing the File Access word set
4561: @item providing the File Access Extensions word set
4562: @item providing the Floating-Point word set
4563: @item providing the Floating-Point Extensions word set
4564: @item providing the Locals word set
4565: @item providing the Locals Extensions word set
4566: @item providing the Memory-Allocation word set
4567: @item providing the Memory-Allocation Extensions word set (that one's easy)
4568: @item providing the Programming-Tools word set
4569: @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
4570: @item providing the Search-Order word set
4571: @item providing the Search-Order Extensions word set
4572: @item providing the String word set
4573: @item providing the String Extensions word set (another easy one)
4574: @end itemize
4575:
4576: @cindex system documentation
4577: In addition, ANS Forth systems are required to document certain
4578: implementation choices. This chapter tries to meet these
4579: requirements. In many cases it gives a way to ask the system for the
4580: information instead of providing the information directly, in
4581: particular, if the information depends on the processor, the operating
4582: system or the installation options chosen, or if they are likely to
4583: change during the maintenance of Gforth.
4584:
4585: @comment The framework for the rest has been taken from pfe.
4586:
4587: @menu
4588: * The Core Words::
4589: * The optional Block word set::
4590: * The optional Double Number word set::
4591: * The optional Exception word set::
4592: * The optional Facility word set::
4593: * The optional File-Access word set::
4594: * The optional Floating-Point word set::
4595: * The optional Locals word set::
4596: * The optional Memory-Allocation word set::
4597: * The optional Programming-Tools word set::
4598: * The optional Search-Order word set::
4599: @end menu
4600:
4601:
4602: @c =====================================================================
4603: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
4604: @comment node-name, next, previous, up
4605: @section The Core Words
4606: @c =====================================================================
4607: @cindex core words, system documentation
4608: @cindex system documentation, core words
4609:
4610: @menu
4611: * core-idef:: Implementation Defined Options
4612: * core-ambcond:: Ambiguous Conditions
4613: * core-other:: Other System Documentation
4614: @end menu
4615:
4616: @c ---------------------------------------------------------------------
4617: @node core-idef, core-ambcond, The Core Words, The Core Words
4618: @subsection Implementation Defined Options
4619: @c ---------------------------------------------------------------------
4620: @cindex core words, implementation-defined options
4621: @cindex implementation-defined options, core words
4622:
4623:
4624: @table @i
4625: @item (Cell) aligned addresses:
4626: @cindex cell-aligned addresses
4627: @cindex aligned addresses
4628: processor-dependent. Gforth's alignment words perform natural alignment
4629: (e.g., an address aligned for a datum of size 8 is divisible by
4630: 8). Unaligned accesses usually result in a @code{-23 THROW}.
4631:
4632: @item @code{EMIT} and non-graphic characters:
4633: @cindex @code{EMIT} and non-graphic characters
4634: @cindex non-graphic characters and @code{EMIT}
4635: The character is output using the C library function (actually, macro)
4636: @code{putc}.
4637:
4638: @item character editing of @code{ACCEPT} and @code{EXPECT}:
4639: @cindex character editing of @code{ACCEPT} and @code{EXPECT}
4640: @cindex editing in @code{ACCEPT} and @code{EXPECT}
4641: @cindex @code{ACCEPT}, editing
4642: @cindex @code{EXPECT}, editing
4643: This is modeled on the GNU readline library (@pxref{Readline
4644: Interaction, , Command Line Editing, readline, The GNU Readline
4645: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
4646: producing a full word completion every time you type it (instead of
4647: producing the common prefix of all completions).
4648:
4649: @item character set:
4650: @cindex character set
4651: The character set of your computer and display device. Gforth is
4652: 8-bit-clean (but some other component in your system may make trouble).
4653:
4654: @item Character-aligned address requirements:
4655: @cindex character-aligned address requirements
4656: installation-dependent. Currently a character is represented by a C
4657: @code{unsigned char}; in the future we might switch to @code{wchar_t}
4658: (Comments on that requested).
4659:
4660: @item character-set extensions and matching of names:
4661: @cindex character-set extensions and matching of names
4662: @cindex case sensitivity for name lookup
4663: @cindex name lookup, case sensitivity
4664: @cindex locale and case sensitivity
4665: Any character except the ASCII NUL charcter can be used in a
4666: name. Matching is case-insensitive (except in @code{TABLE}s). The
4667: matching is performed using the C function @code{strncasecmp}, whose
4668: function is probably influenced by the locale. E.g., the @code{C} locale
4669: does not know about accents and umlauts, so they are matched
4670: case-sensitively in that locale. For portability reasons it is best to
4671: write programs such that they work in the @code{C} locale. Then one can
4672: use libraries written by a Polish programmer (who might use words
4673: containing ISO Latin-2 encoded characters) and by a French programmer
4674: (ISO Latin-1) in the same program (of course, @code{WORDS} will produce
4675: funny results for some of the words (which ones, depends on the font you
4676: are using)). Also, the locale you prefer may not be available in other
4677: operating systems. Hopefully, Unicode will solve these problems one day.
4678:
4679: @item conditions under which control characters match a space delimiter:
4680: @cindex space delimiters
4681: @cindex control characters as delimiters
4682: If @code{WORD} is called with the space character as a delimiter, all
4683: white-space characters (as identified by the C macro @code{isspace()})
4684: are delimiters. @code{PARSE}, on the other hand, treats space like other
4685: delimiters. @code{PARSE-WORD} treats space like @code{WORD}, but behaves
4686: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
4687: interpreter (aka text interpreter) by default, treats all white-space
4688: characters as delimiters.
4689:
4690: @item format of the control flow stack:
4691: @cindex control flow stack, format
4692: The data stack is used as control flow stack. The size of a control flow
4693: stack item in cells is given by the constant @code{cs-item-size}. At the
4694: time of this writing, an item consists of a (pointer to a) locals list
4695: (third), an address in the code (second), and a tag for identifying the
4696: item (TOS). The following tags are used: @code{defstart},
4697: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
4698: @code{scopestart}.
4699:
4700: @item conversion of digits > 35
4701: @cindex digits > 35
4702: The characters @code{[\]^_'} are the digits with the decimal value
4703: 36@minus{}41. There is no way to input many of the larger digits.
4704:
4705: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
4706: @cindex @code{EXPECT}, display after end of input
4707: @cindex @code{ACCEPT}, display after end of input
4708: The cursor is moved to the end of the entered string. If the input is
4709: terminated using the @kbd{Return} key, a space is typed.
4710:
4711: @item exception abort sequence of @code{ABORT"}:
4712: @cindex exception abort sequence of @code{ABORT"}
4713: @cindex @code{ABORT"}, exception abort sequence
4714: The error string is stored into the variable @code{"error} and a
4715: @code{-2 throw} is performed.
4716:
4717: @item input line terminator:
4718: @cindex input line terminator
4719: @cindex line terminator on input
4720: @cindex newline charcter on input
4721: For interactive input, @kbd{C-m} (CR) and @kbd{C-j} (LF) terminate
4722: lines. One of these characters is typically produced when you type the
4723: @kbd{Enter} or @kbd{Return} key.
4724:
4725: @item maximum size of a counted string:
4726: @cindex maximum size of a counted string
4727: @cindex counted string, maximum size
4728: @code{s" /counted-string" environment? drop .}. Currently 255 characters
4729: on all ports, but this may change.
4730:
4731: @item maximum size of a parsed string:
4732: @cindex maximum size of a parsed string
4733: @cindex parsed string, maximum size
4734: Given by the constant @code{/line}. Currently 255 characters.
4735:
4736: @item maximum size of a definition name, in characters:
4737: @cindex maximum size of a definition name, in characters
4738: @cindex name, maximum length
4739: 31
4740:
4741: @item maximum string length for @code{ENVIRONMENT?}, in characters:
4742: @cindex maximum string length for @code{ENVIRONMENT?}, in characters
4743: @cindex @code{ENVIRONMENT?} string length, maximum
4744: 31
4745:
4746: @item method of selecting the user input device:
4747: @cindex user input device, method of selecting
4748: The user input device is the standard input. There is currently no way to
4749: change it from within Gforth. However, the input can typically be
4750: redirected in the command line that starts Gforth.
4751:
4752: @item method of selecting the user output device:
4753: @cindex user output device, method of selecting
4754: @code{EMIT} and @code{TYPE} output to the file-id stored in the value
4755: @code{outfile-id} (@code{stdout} by default). Gforth uses buffered
4756: output, so output on a terminal does not become visible before the next
4757: newline or buffer overflow. Output on non-terminals is invisible until
4758: the buffer overflows.
4759:
4760: @item methods of dictionary compilation:
4761: What are we expected to document here?
4762:
4763: @item number of bits in one address unit:
4764: @cindex number of bits in one address unit
4765: @cindex address unit, size in bits
4766: @code{s" address-units-bits" environment? drop .}. 8 in all current
4767: ports.
4768:
4769: @item number representation and arithmetic:
4770: @cindex number representation and arithmetic
4771: Processor-dependent. Binary two's complement on all current ports.
4772:
4773: @item ranges for integer types:
4774: @cindex ranges for integer types
4775: @cindex integer types, ranges
4776: Installation-dependent. Make environmental queries for @code{MAX-N},
4777: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
4778: unsigned (and positive) types is 0. The lower bound for signed types on
4779: two's complement and one's complement machines machines can be computed
4780: by adding 1 to the upper bound.
4781:
4782: @item read-only data space regions:
4783: @cindex read-only data space regions
4784: @cindex data-space, read-only regions
4785: The whole Forth data space is writable.
4786:
4787: @item size of buffer at @code{WORD}:
4788: @cindex size of buffer at @code{WORD}
4789: @cindex @code{WORD} buffer size
4790: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
4791: shared with the pictured numeric output string. If overwriting
4792: @code{PAD} is acceptable, it is as large as the remaining dictionary
4793: space, although only as much can be sensibly used as fits in a counted
4794: string.
4795:
4796: @item size of one cell in address units:
4797: @cindex cell size
4798: @code{1 cells .}.
4799:
4800: @item size of one character in address units:
4801: @cindex char size
4802: @code{1 chars .}. 1 on all current ports.
4803:
4804: @item size of the keyboard terminal buffer:
4805: @cindex size of the keyboard terminal buffer
4806: @cindex terminal buffer, size
4807: Varies. You can determine the size at a specific time using @code{lp@@
4808: tib - .}. It is shared with the locals stack and TIBs of files that
4809: include the current file. You can change the amount of space for TIBs
4810: and locals stack at Gforth startup with the command line option
4811: @code{-l}.
4812:
4813: @item size of the pictured numeric output buffer:
4814: @cindex size of the pictured numeric output buffer
4815: @cindex pictured numeric output buffer, size
4816: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
4817: shared with @code{WORD}.
4818:
4819: @item size of the scratch area returned by @code{PAD}:
4820: @cindex size of the scratch area returned by @code{PAD}
4821: @cindex @code{PAD} size
4822: The remainder of dictionary space. @code{unused pad here - - .}.
4823:
4824: @item system case-sensitivity characteristics:
4825: @cindex case-sensitivity characteristics
4826: Dictionary searches are case insensitive (except in
4827: @code{TABLE}s). However, as explained above under @i{character-set
4828: extensions}, the matching for non-ASCII characters is determined by the
4829: locale you are using. In the default @code{C} locale all non-ASCII
4830: characters are matched case-sensitively.
4831:
4832: @item system prompt:
4833: @cindex system prompt
4834: @cindex prompt
4835: @code{ ok} in interpret state, @code{ compiled} in compile state.
4836:
4837: @item division rounding:
4838: @cindex division rounding
4839: installation dependent. @code{s" floored" environment? drop .}. We leave
4840: the choice to @code{gcc} (what to use for @code{/}) and to you (whether
4841: to use @code{fm/mod}, @code{sm/rem} or simply @code{/}).
4842:
4843: @item values of @code{STATE} when true:
4844: @cindex @code{STATE} values
4845: -1.
4846:
4847: @item values returned after arithmetic overflow:
4848: On two's complement machines, arithmetic is performed modulo
4849: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
4850: arithmetic (with appropriate mapping for signed types). Division by zero
4851: typically results in a @code{-55 throw} (Floating-point unidentified
4852: fault), although a @code{-10 throw} (divide by zero) would be more
4853: appropriate.
4854:
4855: @item whether the current definition can be found after @t{DOES>}:
4856: @cindex @t{DOES>}, visibility of current definition
4857: No.
4858:
4859: @end table
4860:
4861: @c ---------------------------------------------------------------------
4862: @node core-ambcond, core-other, core-idef, The Core Words
4863: @subsection Ambiguous conditions
4864: @c ---------------------------------------------------------------------
4865: @cindex core words, ambiguous conditions
4866: @cindex ambiguous conditions, core words
4867:
4868: @table @i
4869:
4870: @item a name is neither a word nor a number:
4871: @cindex name not found
4872: @cindex Undefined word
4873: @code{-13 throw} (Undefined word). Actually, @code{-13 bounce}, which
4874: preserves the data and FP stack, so you don't lose more work than
4875: necessary.
4876:
4877: @item a definition name exceeds the maximum length allowed:
4878: @cindex Word name too long
4879: @code{-19 throw} (Word name too long)
4880:
4881: @item addressing a region not inside the various data spaces of the forth system:
4882: @cindex Invalid memory address
4883: The stacks, code space and name space are accessible. Machine code space is
4884: typically readable. Accessing other addresses gives results dependent on
4885: the operating system. On decent systems: @code{-9 throw} (Invalid memory
4886: address).
4887:
4888: @item argument type incompatible with parameter:
4889: @cindex Argument type mismatch
4890: This is usually not caught. Some words perform checks, e.g., the control
4891: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
4892: mismatch).
4893:
4894: @item attempting to obtain the execution token of a word with undefined execution semantics:
4895: @cindex Interpreting a compile-only word, for @code{'} etc.
4896: @cindex execution token of words with undefined execution semantics
4897: @code{-14 throw} (Interpreting a compile-only word). In some cases, you
4898: get an execution token for @code{compile-only-error} (which performs a
4899: @code{-14 throw} when executed).
4900:
4901: @item dividing by zero:
4902: @cindex dividing by zero
4903: @cindex floating point unidentified fault, integer division
4904: @cindex divide by zero
4905: typically results in a @code{-55 throw} (floating point unidentified
4906: fault), although a @code{-10 throw} (divide by zero) would be more
4907: appropriate.
4908:
4909: @item insufficient data stack or return stack space:
4910: @cindex insufficient data stack or return stack space
4911: @cindex stack overflow
4912: @cindex Address alignment exception, stack overflow
4913: @cindex Invalid memory address, stack overflow
4914: Depending on the operating system, the installation, and the invocation
4915: of Gforth, this is either checked by the memory management hardware, or
4916: it is not checked. If it is checked, you typically get a @code{-9 throw}
4917: (Invalid memory address) as soon as the overflow happens. If it is not
4918: check, overflows typically result in mysterious illegal memory accesses,
4919: producing @code{-9 throw} (Invalid memory address) or @code{-23 throw}
4920: (Address alignment exception); they might also destroy the internal data
4921: structure of @code{ALLOCATE} and friends, resulting in various errors in
4922: these words.
4923:
4924: @item insufficient space for loop control parameters:
4925: @cindex insufficient space for loop control parameters
4926: like other return stack overflows.
4927:
4928: @item insufficient space in the dictionary:
4929: @cindex insufficient space in the dictionary
4930: @cindex dictionary overflow
4931: Depending on the operating system, the installation, and the invocation
4932: of Gforth, this is either checked by the memory management hardware, or
4933: it is not checked. Similar results as stack overflows. However,
4934: typically the error appears at a different place when one inserts or
4935: removes code. Also, the @code{THROW} does not relieve the situation (it
4936: does for stack overflows).
4937:
4938: @item interpreting a word with undefined interpretation semantics:
4939: @cindex interpreting a word with undefined interpretation semantics
4940: @cindex Interpreting a compile-only word
4941: For some words, we have defined interpretation semantics. For the
4942: others: @code{-14 throw} (Interpreting a compile-only word).
4943:
4944: @item modifying the contents of the input buffer or a string literal:
4945: @cindex modifying the contents of the input buffer or a string literal
4946: These are located in writable memory and can be modified.
4947:
4948: @item overflow of the pictured numeric output string:
4949: @cindex overflow of the pictured numeric output string
4950: @cindex pictured numeric output string, overflow
4951: Not checked. Runs into the dictionary and destroys it (at least,
4952: partially).
4953:
4954: @item parsed string overflow:
4955: @cindex parsed string overflow
4956: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
4957:
4958: @item producing a result out of range:
4959: @cindex result out of range
4960: On two's complement machines, arithmetic is performed modulo
4961: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
4962: arithmetic (with appropriate mapping for signed types). Division by zero
4963: typically results in a @code{-55 throw} (floatingpoint unidentified
4964: fault), although a @code{-10 throw} (divide by zero) would be more
4965: appropriate. @code{convert} and @code{>number} currently overflow
4966: silently.
4967:
4968: @item reading from an empty data or return stack:
4969: @cindex stack empty
4970: @cindex stack underflow
4971: The data stack is checked by the outer (aka text) interpreter after
4972: every word executed. If it has underflowed, a @code{-4 throw} (Stack
4973: underflow) is performed. Apart from that, stacks may be checked or not,
4974: depending on operating system, installation, and invocation. The
4975: consequences of stack underflows are similar to the consequences of
4976: stack overflows. Note that even if the system uses checking (through the
4977: MMU), your program may have to underflow by a significant number of
4978: stack items to trigger the reaction (the reason for this is that the
4979: MMU, and therefore the checking, works with a page-size granularity).
4980:
4981: @item unexpected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
4982: @cindex unexpected end of the input buffer
4983: @cindex zero-length string as a name
4984: @cindex Attempt to use zero-length string as a name
4985: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
4986: use zero-length string as a name). Words like @code{'} probably will not
4987: find what they search. Note that it is possible to create zero-length
4988: names with @code{nextname} (should it not?).
4989:
4990: @item @code{>IN} greater than input buffer:
4991: @cindex @code{>IN} greater than input buffer
4992: The next invocation of a parsing word returns a string with length 0.
4993:
4994: @item @code{RECURSE} appears after @code{DOES>}:
4995: @cindex @code{RECURSE} appears after @code{DOES>}
4996: Compiles a recursive call to the defining word, not to the defined word.
4997:
4998: @item argument input source different than current input source for @code{RESTORE-INPUT}:
4999: @cindex argument input source different than current input source for @code{RESTORE-INPUT}
5000: @cindex Argument type mismatch, @code{RESTORE-INPUT}
5001: @cindex @code{RESTORE-INPUT}, Argument type mismatch
5002: @code{-12 THROW}. Note that, once an input file is closed (e.g., because
5003: the end of the file was reached), its source-id may be
5004: reused. Therefore, restoring an input source specification referencing a
5005: closed file may lead to unpredictable results instead of a @code{-12
5006: THROW}.
5007:
5008: In the future, Gforth may be able to restore input source specifications
5009: from other than the current input source.
5010:
5011: @item data space containing definitions gets de-allocated:
5012: @cindex data space containing definitions gets de-allocated
5013: Deallocation with @code{allot} is not checked. This typically results in
5014: memory access faults or execution of illegal instructions.
5015:
5016: @item data space read/write with incorrect alignment:
5017: @cindex data space read/write with incorrect alignment
5018: @cindex alignment faults
5019: @cindex Address alignment exception
5020: Processor-dependent. Typically results in a @code{-23 throw} (Address
5021: alignment exception). Under Linux on a 486 or later processor with
5022: alignment turned on, incorrect alignment results in a @code{-9 throw}
5023: (Invalid memory address). There are reportedly some processors with
5024: alignment restrictions that do not report them.
5025:
5026: @item data space pointer not properly aligned, @code{,}, @code{C,}:
5027: @cindex data space pointer not properly aligned, @code{,}, @code{C,}
5028: Like other alignment errors.
5029:
5030: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
5031: Like other stack underflows.
5032:
5033: @item loop control parameters not available:
5034: @cindex loop control parameters not available
5035: Not checked. The counted loop words simply assume that the top of return
5036: stack items are loop control parameters and behave accordingly.
5037:
5038: @item most recent definition does not have a name (@code{IMMEDIATE}):
5039: @cindex most recent definition does not have a name (@code{IMMEDIATE})
5040: @cindex last word was headerless
5041: @code{abort" last word was headerless"}.
5042:
5043: @item name not defined by @code{VALUE} used by @code{TO}:
5044: @cindex name not defined by @code{VALUE} used by @code{TO}
5045: @cindex @code{TO} on non-@code{VALUE}s
5046: @cindex Invalid name argument, @code{TO}
5047: @code{-32 throw} (Invalid name argument) (unless name is a local or was
5048: defined by @code{CONSTANT}; in the latter case it just changes the constant).
5049:
5050: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
5051: @cindex name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]})
5052: @cindex Undefined word, @code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}
5053: @code{-13 throw} (Undefined word)
5054:
5055: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
5056: @cindex parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN})
5057: Gforth behaves as if they were of the same type. I.e., you can predict
5058: the behaviour by interpreting all parameters as, e.g., signed.
5059:
5060: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
5061: @cindex @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}
5062: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} performs the
5063: compilation semantics of @code{TO}.
5064:
5065: @item String longer than a counted string returned by @code{WORD}:
5066: @cindex String longer than a counted string returned by @code{WORD}
5067: @cindex @code{WORD}, string overflow
5068: Not checked. The string will be ok, but the count will, of course,
5069: contain only the least significant bits of the length.
5070:
5071: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
5072: @cindex @code{LSHIFT}, large shift counts
5073: @cindex @code{RSHIFT}, large shift counts
5074: Processor-dependent. Typical behaviours are returning 0 and using only
5075: the low bits of the shift count.
5076:
5077: @item word not defined via @code{CREATE}:
5078: @cindex @code{>BODY} of non-@code{CREATE}d words
5079: @code{>BODY} produces the PFA of the word no matter how it was defined.
5080:
5081: @cindex @code{DOES>} of non-@code{CREATE}d words
5082: @code{DOES>} changes the execution semantics of the last defined word no
5083: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
5084: @code{CREATE , DOES>}.
5085:
5086: @item words improperly used outside @code{<#} and @code{#>}:
5087: Not checked. As usual, you can expect memory faults.
5088:
5089: @end table
5090:
5091:
5092: @c ---------------------------------------------------------------------
5093: @node core-other, , core-ambcond, The Core Words
5094: @subsection Other system documentation
5095: @c ---------------------------------------------------------------------
5096: @cindex other system documentation, core words
5097: @cindex core words, other system documentation
5098:
5099: @table @i
5100: @item nonstandard words using @code{PAD}:
5101: @cindex @code{PAD} use by nonstandard words
5102: None.
5103:
5104: @item operator's terminal facilities available:
5105: @cindex operator's terminal facilities available
5106: After processing the command line, Gforth goes into interactive mode,
5107: and you can give commands to Gforth interactively. The actual facilities
5108: available depend on how you invoke Gforth.
5109:
5110: @item program data space available:
5111: @cindex program data space available
5112: @cindex data space available
5113: @code{UNUSED .} gives the remaining dictionary space. The total
5114: dictionary space can be specified with the @code{-m} switch
5115: (@pxref{Invoking Gforth}) when Gforth starts up.
5116:
5117: @item return stack space available:
5118: @cindex return stack space available
5119: You can compute the total return stack space in cells with
5120: @code{s" RETURN-STACK-CELLS" environment? drop .}. You can specify it at
5121: startup time with the @code{-r} switch (@pxref{Invoking Gforth}).
5122:
5123: @item stack space available:
5124: @cindex stack space available
5125: You can compute the total data stack space in cells with
5126: @code{s" STACK-CELLS" environment? drop .}. You can specify it at
5127: startup time with the @code{-d} switch (@pxref{Invoking Gforth}).
5128:
5129: @item system dictionary space required, in address units:
5130: @cindex system dictionary space required, in address units
5131: Type @code{here forthstart - .} after startup. At the time of this
5132: writing, this gives 80080 (bytes) on a 32-bit system.
5133: @end table
5134:
5135:
5136: @c =====================================================================
5137: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
5138: @section The optional Block word set
5139: @c =====================================================================
5140: @cindex system documentation, block words
5141: @cindex block words, system documentation
5142:
5143: @menu
5144: * block-idef:: Implementation Defined Options
5145: * block-ambcond:: Ambiguous Conditions
5146: * block-other:: Other System Documentation
5147: @end menu
5148:
5149:
5150: @c ---------------------------------------------------------------------
5151: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
5152: @subsection Implementation Defined Options
5153: @c ---------------------------------------------------------------------
5154: @cindex implementation-defined options, block words
5155: @cindex block words, implementation-defined options
5156:
5157: @table @i
5158: @item the format for display by @code{LIST}:
5159: @cindex @code{LIST} display format
5160: First the screen number is displayed, then 16 lines of 64 characters,
5161: each line preceded by the line number.
5162:
5163: @item the length of a line affected by @code{\}:
5164: @cindex length of a line affected by @code{\}
5165: @cindex @code{\}, line length in blocks
5166: 64 characters.
5167: @end table
5168:
5169:
5170: @c ---------------------------------------------------------------------
5171: @node block-ambcond, block-other, block-idef, The optional Block word set
5172: @subsection Ambiguous conditions
5173: @c ---------------------------------------------------------------------
5174: @cindex block words, ambiguous conditions
5175: @cindex ambiguous conditions, block words
5176:
5177: @table @i
5178: @item correct block read was not possible:
5179: @cindex block read not possible
5180: Typically results in a @code{throw} of some OS-derived value (between
5181: -512 and -2048). If the blocks file was just not long enough, blanks are
5182: supplied for the missing portion.
5183:
5184: @item I/O exception in block transfer:
5185: @cindex I/O exception in block transfer
5186: @cindex block transfer, I/O exception
5187: Typically results in a @code{throw} of some OS-derived value (between
5188: -512 and -2048).
5189:
5190: @item invalid block number:
5191: @cindex invalid block number
5192: @cindex block number invalid
5193: @code{-35 throw} (Invalid block number)
5194:
5195: @item a program directly alters the contents of @code{BLK}:
5196: @cindex @code{BLK}, altering @code{BLK}
5197: The input stream is switched to that other block, at the same
5198: position. If the storing to @code{BLK} happens when interpreting
5199: non-block input, the system will get quite confused when the block ends.
5200:
5201: @item no current block buffer for @code{UPDATE}:
5202: @cindex @code{UPDATE}, no current block buffer
5203: @code{UPDATE} has no effect.
5204:
5205: @end table
5206:
5207: @c ---------------------------------------------------------------------
5208: @node block-other, , block-ambcond, The optional Block word set
5209: @subsection Other system documentation
5210: @c ---------------------------------------------------------------------
5211: @cindex other system documentation, block words
5212: @cindex block words, other system documentation
5213:
5214: @table @i
5215: @item any restrictions a multiprogramming system places on the use of buffer addresses:
5216: No restrictions (yet).
5217:
5218: @item the number of blocks available for source and data:
5219: depends on your disk space.
5220:
5221: @end table
5222:
5223:
5224: @c =====================================================================
5225: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
5226: @section The optional Double Number word set
5227: @c =====================================================================
5228: @cindex system documentation, double words
5229: @cindex double words, system documentation
5230:
5231: @menu
5232: * double-ambcond:: Ambiguous Conditions
5233: @end menu
5234:
5235:
5236: @c ---------------------------------------------------------------------
5237: @node double-ambcond, , The optional Double Number word set, The optional Double Number word set
5238: @subsection Ambiguous conditions
5239: @c ---------------------------------------------------------------------
5240: @cindex double words, ambiguous conditions
5241: @cindex ambiguous conditions, double words
5242:
5243: @table @i
5244: @item @var{d} outside of range of @var{n} in @code{D>S}:
5245: @cindex @code{D>S}, @var{d} out of range of @var{n}
5246: The least significant cell of @var{d} is produced.
5247:
5248: @end table
5249:
5250:
5251: @c =====================================================================
5252: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
5253: @section The optional Exception word set
5254: @c =====================================================================
5255: @cindex system documentation, exception words
5256: @cindex exception words, system documentation
5257:
5258: @menu
5259: * exception-idef:: Implementation Defined Options
5260: @end menu
5261:
5262:
5263: @c ---------------------------------------------------------------------
5264: @node exception-idef, , The optional Exception word set, The optional Exception word set
5265: @subsection Implementation Defined Options
5266: @c ---------------------------------------------------------------------
5267: @cindex implementation-defined options, exception words
5268: @cindex exception words, implementation-defined options
5269:
5270: @table @i
5271: @item @code{THROW}-codes used in the system:
5272: @cindex @code{THROW}-codes used in the system
5273: The codes -256@minus{}-511 are used for reporting signals. The mapping
5274: from OS signal numbers to throw codes is -256@minus{}@var{signal}. The
5275: codes -512@minus{}-2047 are used for OS errors (for file and memory
5276: allocation operations). The mapping from OS error numbers to throw codes
5277: is -512@minus{}@code{errno}. One side effect of this mapping is that
5278: undefined OS errors produce a message with a strange number; e.g.,
5279: @code{-1000 THROW} results in @code{Unknown error 488} on my system.
5280: @end table
5281:
5282: @c =====================================================================
5283: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
5284: @section The optional Facility word set
5285: @c =====================================================================
5286: @cindex system documentation, facility words
5287: @cindex facility words, system documentation
5288:
5289: @menu
5290: * facility-idef:: Implementation Defined Options
5291: * facility-ambcond:: Ambiguous Conditions
5292: @end menu
5293:
5294:
5295: @c ---------------------------------------------------------------------
5296: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
5297: @subsection Implementation Defined Options
5298: @c ---------------------------------------------------------------------
5299: @cindex implementation-defined options, facility words
5300: @cindex facility words, implementation-defined options
5301:
5302: @table @i
5303: @item encoding of keyboard events (@code{EKEY}):
5304: @cindex keyboard events, encoding in @code{EKEY}
5305: @cindex @code{EKEY}, encoding of keyboard events
5306: Not yet implemented.
5307:
5308: @item duration of a system clock tick:
5309: @cindex duration of a system clock tick
5310: @cindex clock tick duration
5311: System dependent. With respect to @code{MS}, the time is specified in
5312: microseconds. How well the OS and the hardware implement this, is
5313: another question.
5314:
5315: @item repeatability to be expected from the execution of @code{MS}:
5316: @cindex repeatability to be expected from the execution of @code{MS}
5317: @cindex @code{MS}, repeatability to be expected
5318: System dependent. On Unix, a lot depends on load. If the system is
5319: lightly loaded, and the delay is short enough that Gforth does not get
5320: swapped out, the performance should be acceptable. Under MS-DOS and
5321: other single-tasking systems, it should be good.
5322:
5323: @end table
5324:
5325:
5326: @c ---------------------------------------------------------------------
5327: @node facility-ambcond, , facility-idef, The optional Facility word set
5328: @subsection Ambiguous conditions
5329: @c ---------------------------------------------------------------------
5330: @cindex facility words, ambiguous conditions
5331: @cindex ambiguous conditions, facility words
5332:
5333: @table @i
5334: @item @code{AT-XY} can't be performed on user output device:
5335: @cindex @code{AT-XY} can't be performed on user output device
5336: Largely terminal dependent. No range checks are done on the arguments.
5337: No errors are reported. You may see some garbage appearing, you may see
5338: simply nothing happen.
5339:
5340: @end table
5341:
5342:
5343: @c =====================================================================
5344: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
5345: @section The optional File-Access word set
5346: @c =====================================================================
5347: @cindex system documentation, file words
5348: @cindex file words, system documentation
5349:
5350: @menu
5351: * file-idef:: Implementation Defined Options
5352: * file-ambcond:: Ambiguous Conditions
5353: @end menu
5354:
5355: @c ---------------------------------------------------------------------
5356: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
5357: @subsection Implementation Defined Options
5358: @c ---------------------------------------------------------------------
5359: @cindex implementation-defined options, file words
5360: @cindex file words, implementation-defined options
5361:
5362: @table @i
5363: @item file access methods used:
5364: @cindex file access methods used
5365: @code{R/O}, @code{R/W} and @code{BIN} work as you would
5366: expect. @code{W/O} translates into the C file opening mode @code{w} (or
5367: @code{wb}): The file is cleared, if it exists, and created, if it does
5368: not (with both @code{open-file} and @code{create-file}). Under Unix
5369: @code{create-file} creates a file with 666 permissions modified by your
5370: umask.
5371:
5372: @item file exceptions:
5373: @cindex file exceptions
5374: The file words do not raise exceptions (except, perhaps, memory access
5375: faults when you pass illegal addresses or file-ids).
5376:
5377: @item file line terminator:
5378: @cindex file line terminator
5379: System-dependent. Gforth uses C's newline character as line
5380: terminator. What the actual character code(s) of this are is
5381: system-dependent.
5382:
5383: @item file name format:
5384: @cindex file name format
5385: System dependent. Gforth just uses the file name format of your OS.
5386:
5387: @item information returned by @code{FILE-STATUS}:
5388: @cindex @code{FILE-STATUS}, returned information
5389: @code{FILE-STATUS} returns the most powerful file access mode allowed
5390: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
5391: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
5392: along with the returned mode.
5393:
5394: @item input file state after an exception when including source:
5395: @cindex exception when including source
5396: All files that are left via the exception are closed.
5397:
5398: @item @var{ior} values and meaning:
5399: @cindex @var{ior} values and meaning
5400: The @var{ior}s returned by the file and memory allocation words are
5401: intended as throw codes. They typically are in the range
5402: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
5403: @var{ior}s is -512@minus{}@var{errno}.
5404:
5405: @item maximum depth of file input nesting:
5406: @cindex maximum depth of file input nesting
5407: @cindex file input nesting, maximum depth
5408: limited by the amount of return stack, locals/TIB stack, and the number
5409: of open files available. This should not give you troubles.
5410:
5411: @item maximum size of input line:
5412: @cindex maximum size of input line
5413: @cindex input line size, maximum
5414: @code{/line}. Currently 255.
5415:
5416: @item methods of mapping block ranges to files:
5417: @cindex mapping block ranges to files
5418: @cindex files containing blocks
5419: @cindex blocks in files
5420: By default, blocks are accessed in the file @file{blocks.fb} in the
5421: current working directory. The file can be switched with @code{USE}.
5422:
5423: @item number of string buffers provided by @code{S"}:
5424: @cindex @code{S"}, number of string buffers
5425: 1
5426:
5427: @item size of string buffer used by @code{S"}:
5428: @cindex @code{S"}, size of string buffer
5429: @code{/line}. currently 255.
5430:
5431: @end table
5432:
5433: @c ---------------------------------------------------------------------
5434: @node file-ambcond, , file-idef, The optional File-Access word set
5435: @subsection Ambiguous conditions
5436: @c ---------------------------------------------------------------------
5437: @cindex file words, ambiguous conditions
5438: @cindex ambiguous conditions, file words
5439:
5440: @table @i
5441: @item attempting to position a file outside its boundaries:
5442: @cindex @code{REPOSITION-FILE}, outside the file's boundaries
5443: @code{REPOSITION-FILE} is performed as usual: Afterwards,
5444: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
5445:
5446: @item attempting to read from file positions not yet written:
5447: @cindex reading from file positions not yet written
5448: End-of-file, i.e., zero characters are read and no error is reported.
5449:
5450: @item @var{file-id} is invalid (@code{INCLUDE-FILE}):
5451: @cindex @code{INCLUDE-FILE}, @var{file-id} is invalid
5452: An appropriate exception may be thrown, but a memory fault or other
5453: problem is more probable.
5454:
5455: @item I/O exception reading or closing @var{file-id} (@code{INCLUDE-FILE}, @code{INCLUDED}):
5456: @cindex @code{INCLUDE-FILE}, I/O exception reading or closing @var{file-id}
5457: @cindex @code{INCLUDED}, I/O exception reading or closing @var{file-id}
5458: The @var{ior} produced by the operation, that discovered the problem, is
5459: thrown.
5460:
5461: @item named file cannot be opened (@code{INCLUDED}):
5462: @cindex @code{INCLUDED}, named file cannot be opened
5463: The @var{ior} produced by @code{open-file} is thrown.
5464:
5465: @item requesting an unmapped block number:
5466: @cindex unmapped block numbers
5467: There are no unmapped legal block numbers. On some operating systems,
5468: writing a block with a large number may overflow the file system and
5469: have an error message as consequence.
5470:
5471: @item using @code{source-id} when @code{blk} is non-zero:
5472: @cindex @code{SOURCE-ID}, behaviour when @code{BLK} is non-zero
5473: @code{source-id} performs its function. Typically it will give the id of
5474: the source which loaded the block. (Better ideas?)
5475:
5476: @end table
5477:
5478:
5479: @c =====================================================================
5480: @node The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
5481: @section The optional Floating-Point word set
5482: @c =====================================================================
5483: @cindex system documentation, floating-point words
5484: @cindex floating-point words, system documentation
5485:
5486: @menu
5487: * floating-idef:: Implementation Defined Options
5488: * floating-ambcond:: Ambiguous Conditions
5489: @end menu
5490:
5491:
5492: @c ---------------------------------------------------------------------
5493: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
5494: @subsection Implementation Defined Options
5495: @c ---------------------------------------------------------------------
5496: @cindex implementation-defined options, floating-point words
5497: @cindex floating-point words, implementation-defined options
5498:
5499: @table @i
5500: @item format and range of floating point numbers:
5501: @cindex format and range of floating point numbers
5502: @cindex floating point numbers, format and range
5503: System-dependent; the @code{double} type of C.
5504:
5505: @item results of @code{REPRESENT} when @var{float} is out of range:
5506: @cindex @code{REPRESENT}, results when @var{float} is out of range
5507: System dependent; @code{REPRESENT} is implemented using the C library
5508: function @code{ecvt()} and inherits its behaviour in this respect.
5509:
5510: @item rounding or truncation of floating-point numbers:
5511: @cindex rounding of floating-point numbers
5512: @cindex truncation of floating-point numbers
5513: @cindex floating-point numbers, rounding or truncation
5514: System dependent; the rounding behaviour is inherited from the hosting C
5515: compiler. IEEE-FP-based (i.e., most) systems by default round to
5516: nearest, and break ties by rounding to even (i.e., such that the last
5517: bit of the mantissa is 0).
5518:
5519: @item size of floating-point stack:
5520: @cindex floating-point stack size
5521: @code{s" FLOATING-STACK" environment? drop .} gives the total size of
5522: the floating-point stack (in floats). You can specify this on startup
5523: with the command-line option @code{-f} (@pxref{Invoking Gforth}).
5524:
5525: @item width of floating-point stack:
5526: @cindex floating-point stack width
5527: @code{1 floats}.
5528:
5529: @end table
5530:
5531:
5532: @c ---------------------------------------------------------------------
5533: @node floating-ambcond, , floating-idef, The optional Floating-Point word set
5534: @subsection Ambiguous conditions
5535: @c ---------------------------------------------------------------------
5536: @cindex floating-point words, ambiguous conditions
5537: @cindex ambiguous conditions, floating-point words
5538:
5539: @table @i
5540: @item @code{df@@} or @code{df!} used with an address that is not double-float aligned:
5541: @cindex @code{df@@} or @code{df!} used with an address that is not double-float aligned
5542: System-dependent. Typically results in a @code{-23 THROW} like other
5543: alignment violations.
5544:
5545: @item @code{f@@} or @code{f!} used with an address that is not float aligned:
5546: @cindex @code{f@@} used with an address that is not float aligned
5547: @cindex @code{f!} used with an address that is not float aligned
5548: System-dependent. Typically results in a @code{-23 THROW} like other
5549: alignment violations.
5550:
5551: @item floating-point result out of range:
5552: @cindex floating-point result out of range
5553: System-dependent. Can result in a @code{-55 THROW} (Floating-point
5554: unidentified fault), or can produce a special value representing, e.g.,
5555: Infinity.
5556:
5557: @item @code{sf@@} or @code{sf!} used with an address that is not single-float aligned:
5558: @cindex @code{sf@@} or @code{sf!} used with an address that is not single-float aligned
5559: System-dependent. Typically results in an alignment fault like other
5560: alignment violations.
5561:
5562: @item @code{BASE} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
5563: @cindex @code{BASE} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.})
5564: The floating-point number is converted into decimal nonetheless.
5565:
5566: @item Both arguments are equal to zero (@code{FATAN2}):
5567: @cindex @code{FATAN2}, both arguments are equal to zero
5568: System-dependent. @code{FATAN2} is implemented using the C library
5569: function @code{atan2()}.
5570:
5571: @item Using @code{FTAN} on an argument @var{r1} where cos(@var{r1}) is zero:
5572: @cindex @code{FTAN} on an argument @var{r1} where cos(@var{r1}) is zero
5573: System-dependent. Anyway, typically the cos of @var{r1} will not be zero
5574: because of small errors and the tan will be a very large (or very small)
5575: but finite number.
5576:
5577: @item @var{d} cannot be presented precisely as a float in @code{D>F}:
5578: @cindex @code{D>F}, @var{d} cannot be presented precisely as a float
5579: The result is rounded to the nearest float.
5580:
5581: @item dividing by zero:
5582: @cindex dividing by zero, floating-point
5583: @cindex floating-point dividing by zero
5584: @cindex floating-point unidentified fault, FP divide-by-zero
5585: @code{-55 throw} (Floating-point unidentified fault)
5586:
5587: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
5588: @cindex exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@})
5589: System dependent. On IEEE-FP based systems the number is converted into
5590: an infinity.
5591:
5592: @item @var{float}<1 (@code{FACOSH}):
5593: @cindex @code{FACOSH}, @var{float}<1
5594: @cindex floating-point unidentified fault, @code{FACOSH}
5595: @code{-55 throw} (Floating-point unidentified fault)
5596:
5597: @item @var{float}=<-1 (@code{FLNP1}):
5598: @cindex @code{FLNP1}, @var{float}=<-1
5599: @cindex floating-point unidentified fault, @code{FLNP1}
5600: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
5601: negative infinity is typically produced for @var{float}=-1.
5602:
5603: @item @var{float}=<0 (@code{FLN}, @code{FLOG}):
5604: @cindex @code{FLN}, @var{float}=<0
5605: @cindex @code{FLOG}, @var{float}=<0
5606: @cindex floating-point unidentified fault, @code{FLN} or @code{FLOG}
5607: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
5608: negative infinity is typically produced for @var{float}=0.
5609:
5610: @item @var{float}<0 (@code{FASINH}, @code{FSQRT}):
5611: @cindex @code{FASINH}, @var{float}<0
5612: @cindex @code{FSQRT}, @var{float}<0
5613: @cindex floating-point unidentified fault, @code{FASINH} or @code{FSQRT}
5614: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
5615: produces values for these inputs on my Linux box (Bug in the C library?)
5616:
5617: @item |@var{float}|>1 (@code{FACOS}, @code{FASIN}, @code{FATANH}):
5618: @cindex @code{FACOS}, |@var{float}|>1
5619: @cindex @code{FASIN}, |@var{float}|>1
5620: @cindex @code{FATANH}, |@var{float}|>1
5621: @cindex floating-point unidentified fault, @code{FACOS}, @code{FASIN} or @code{FATANH}
5622: @code{-55 throw} (Floating-point unidentified fault).
5623:
5624: @item integer part of float cannot be represented by @var{d} in @code{F>D}:
5625: @cindex @code{F>D}, integer part of float cannot be represented by @var{d}
5626: @cindex floating-point unidentified fault, @code{F>D}
5627: @code{-55 throw} (Floating-point unidentified fault).
5628:
5629: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
5630: @cindex string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.})
5631: This does not happen.
5632: @end table
5633:
5634: @c =====================================================================
5635: @node The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
5636: @section The optional Locals word set
5637: @c =====================================================================
5638: @cindex system documentation, locals words
5639: @cindex locals words, system documentation
5640:
5641: @menu
5642: * locals-idef:: Implementation Defined Options
5643: * locals-ambcond:: Ambiguous Conditions
5644: @end menu
5645:
5646:
5647: @c ---------------------------------------------------------------------
5648: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
5649: @subsection Implementation Defined Options
5650: @c ---------------------------------------------------------------------
5651: @cindex implementation-defined options, locals words
5652: @cindex locals words, implementation-defined options
5653:
5654: @table @i
5655: @item maximum number of locals in a definition:
5656: @cindex maximum number of locals in a definition
5657: @cindex locals, maximum number in a definition
5658: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
5659: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
5660: characters. The number of locals in a definition is bounded by the size
5661: of locals-buffer, which contains the names of the locals.
5662:
5663: @end table
5664:
5665:
5666: @c ---------------------------------------------------------------------
5667: @node locals-ambcond, , locals-idef, The optional Locals word set
5668: @subsection Ambiguous conditions
5669: @c ---------------------------------------------------------------------
5670: @cindex locals words, ambiguous conditions
5671: @cindex ambiguous conditions, locals words
5672:
5673: @table @i
5674: @item executing a named local in interpretation state:
5675: @cindex local in interpretation state
5676: @cindex Interpreting a compile-only word, for a local
5677: Locals have no interpretation semantics. If you try to perform the
5678: interpretation semantics, you will get a @code{-14 throw} somewhere
5679: (Interpreting a compile-only word). If you perform the compilation
5680: semantics, the locals access will be compiled (irrespective of state).
5681:
5682: @item @var{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
5683: @cindex name not defined by @code{VALUE} or @code{(LOCAL)} used by @code{TO}
5684: @cindex @code{TO} on non-@code{VALUE}s and non-locals
5685: @cindex Invalid name argument, @code{TO}
5686: @code{-32 throw} (Invalid name argument)
5687:
5688: @end table
5689:
5690:
5691: @c =====================================================================
5692: @node The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
5693: @section The optional Memory-Allocation word set
5694: @c =====================================================================
5695: @cindex system documentation, memory-allocation words
5696: @cindex memory-allocation words, system documentation
5697:
5698: @menu
5699: * memory-idef:: Implementation Defined Options
5700: @end menu
5701:
5702:
5703: @c ---------------------------------------------------------------------
5704: @node memory-idef, , The optional Memory-Allocation word set, The optional Memory-Allocation word set
5705: @subsection Implementation Defined Options
5706: @c ---------------------------------------------------------------------
5707: @cindex implementation-defined options, memory-allocation words
5708: @cindex memory-allocation words, implementation-defined options
5709:
5710: @table @i
5711: @item values and meaning of @var{ior}:
5712: @cindex @var{ior} values and meaning
5713: The @var{ior}s returned by the file and memory allocation words are
5714: intended as throw codes. They typically are in the range
5715: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
5716: @var{ior}s is -512@minus{}@var{errno}.
5717:
5718: @end table
5719:
5720: @c =====================================================================
5721: @node The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
5722: @section The optional Programming-Tools word set
5723: @c =====================================================================
5724: @cindex system documentation, programming-tools words
5725: @cindex programming-tools words, system documentation
5726:
5727: @menu
5728: * programming-idef:: Implementation Defined Options
5729: * programming-ambcond:: Ambiguous Conditions
5730: @end menu
5731:
5732:
5733: @c ---------------------------------------------------------------------
5734: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
5735: @subsection Implementation Defined Options
5736: @c ---------------------------------------------------------------------
5737: @cindex implementation-defined options, programming-tools words
5738: @cindex programming-tools words, implementation-defined options
5739:
5740: @table @i
5741: @item ending sequence for input following @code{;CODE} and @code{CODE}:
5742: @cindex @code{;CODE} ending sequence
5743: @cindex @code{CODE} ending sequence
5744: @code{END-CODE}
5745:
5746: @item manner of processing input following @code{;CODE} and @code{CODE}:
5747: @cindex @code{;CODE}, processing input
5748: @cindex @code{CODE}, processing input
5749: The @code{ASSEMBLER} vocabulary is pushed on the search order stack, and
5750: the input is processed by the text interpreter, (starting) in interpret
5751: state.
5752:
5753: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
5754: @cindex @code{ASSEMBLER}, search order capability
5755: The ANS Forth search order word set.
5756:
5757: @item source and format of display by @code{SEE}:
5758: @cindex @code{SEE}, source and format of output
5759: The source for @code{see} is the intermediate code used by the inner
5760: interpreter. The current @code{see} tries to output Forth source code
5761: as well as possible.
5762:
5763: @end table
5764:
5765: @c ---------------------------------------------------------------------
5766: @node programming-ambcond, , programming-idef, The optional Programming-Tools word set
5767: @subsection Ambiguous conditions
5768: @c ---------------------------------------------------------------------
5769: @cindex programming-tools words, ambiguous conditions
5770: @cindex ambiguous conditions, programming-tools words
5771:
5772: @table @i
5773:
5774: @item deleting the compilation wordlist (@code{FORGET}):
5775: @cindex @code{FORGET}, deleting the compilation wordlist
5776: Not implemented (yet).
5777:
5778: @item fewer than @var{u}+1 items on the control flow stack (@code{CS-PICK}, @code{CS-ROLL}):
5779: @cindex @code{CS-PICK}, fewer than @var{u}+1 items on the control flow stack
5780: @cindex @code{CS-ROLL}, fewer than @var{u}+1 items on the control flow stack
5781: @cindex control-flow stack underflow
5782: This typically results in an @code{abort"} with a descriptive error
5783: message (may change into a @code{-22 throw} (Control structure mismatch)
5784: in the future). You may also get a memory access error. If you are
5785: unlucky, this ambiguous condition is not caught.
5786:
5787: @item @var{name} can't be found (@code{FORGET}):
5788: @cindex @code{FORGET}, @var{name} can't be found
5789: Not implemented (yet).
5790:
5791: @item @var{name} not defined via @code{CREATE}:
5792: @cindex @code{;CODE}, @var{name} not defined via @code{CREATE}
5793: @code{;CODE} behaves like @code{DOES>} in this respect, i.e., it changes
5794: the execution semantics of the last defined word no matter how it was
5795: defined.
5796:
5797: @item @code{POSTPONE} applied to @code{[IF]}:
5798: @cindex @code{POSTPONE} applied to @code{[IF]}
5799: @cindex @code{[IF]} and @code{POSTPONE}
5800: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
5801: equivalent to @code{[IF]}.
5802:
5803: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
5804: @cindex @code{[IF]}, end of the input source before matching @code{[ELSE]} or @code{[THEN]}
5805: Continue in the same state of conditional compilation in the next outer
5806: input source. Currently there is no warning to the user about this.
5807:
5808: @item removing a needed definition (@code{FORGET}):
5809: @cindex @code{FORGET}, removing a needed definition
5810: Not implemented (yet).
5811:
5812: @end table
5813:
5814:
5815: @c =====================================================================
5816: @node The optional Search-Order word set, , The optional Programming-Tools word set, ANS conformance
5817: @section The optional Search-Order word set
5818: @c =====================================================================
5819: @cindex system documentation, search-order words
5820: @cindex search-order words, system documentation
5821:
5822: @menu
5823: * search-idef:: Implementation Defined Options
5824: * search-ambcond:: Ambiguous Conditions
5825: @end menu
5826:
5827:
5828: @c ---------------------------------------------------------------------
5829: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
5830: @subsection Implementation Defined Options
5831: @c ---------------------------------------------------------------------
5832: @cindex implementation-defined options, search-order words
5833: @cindex search-order words, implementation-defined options
5834:
5835: @table @i
5836: @item maximum number of word lists in search order:
5837: @cindex maximum number of word lists in search order
5838: @cindex search order, maximum depth
5839: @code{s" wordlists" environment? drop .}. Currently 16.
5840:
5841: @item minimum search order:
5842: @cindex minimum search order
5843: @cindex search order, minimum
5844: @code{root root}.
5845:
5846: @end table
5847:
5848: @c ---------------------------------------------------------------------
5849: @node search-ambcond, , search-idef, The optional Search-Order word set
5850: @subsection Ambiguous conditions
5851: @c ---------------------------------------------------------------------
5852: @cindex search-order words, ambiguous conditions
5853: @cindex ambiguous conditions, search-order words
5854:
5855: @table @i
5856: @item changing the compilation wordlist (during compilation):
5857: @cindex changing the compilation wordlist (during compilation)
5858: @cindex compilation wordlist, change before definition ends
5859: The word is entered into the wordlist that was the compilation wordlist
5860: at the start of the definition. Any changes to the name field (e.g.,
5861: @code{immediate}) or the code field (e.g., when executing @code{DOES>})
5862: are applied to the latest defined word (as reported by @code{last} or
5863: @code{lastxt}), if possible, irrespective of the compilation wordlist.
5864:
5865: @item search order empty (@code{previous}):
5866: @cindex @code{previous}, search order empty
5867: @cindex Vocstack empty, @code{previous}
5868: @code{abort" Vocstack empty"}.
5869:
5870: @item too many word lists in search order (@code{also}):
5871: @cindex @code{also}, too many word lists in search order
5872: @cindex Vocstack full, @code{also}
5873: @code{abort" Vocstack full"}.
5874:
5875: @end table
5876:
5877: @c ***************************************************************
5878: @node Model, Integrating Gforth, ANS conformance, Top
5879: @chapter Model
5880:
5881: This chapter has yet to be written. It will contain information, on
5882: which internal structures you can rely.
5883:
5884: @c ***************************************************************
5885: @node Integrating Gforth, Emacs and Gforth, Model, Top
5886: @chapter Integrating Gforth into C programs
5887:
5888: This is not yet implemented.
5889:
5890: Several people like to use Forth as scripting language for applications
5891: that are otherwise written in C, C++, or some other language.
5892:
5893: The Forth system ATLAST provides facilities for embedding it into
5894: applications; unfortunately it has several disadvantages: most
5895: importantly, it is not based on ANS Forth, and it is apparently dead
5896: (i.e., not developed further and not supported). The facilities
5897: provided by Gforth in this area are inspired by ATLASTs facilities, so
5898: making the switch should not be hard.
5899:
5900: We also tried to design the interface such that it can easily be
5901: implemented by other Forth systems, so that we may one day arrive at a
5902: standardized interface. Such a standard interface would allow you to
5903: replace the Forth system without having to rewrite C code.
5904:
5905: You embed the Gforth interpreter by linking with the library
5906: @code{libgforth.a} (give the compiler the option @code{-lgforth}). All
5907: global symbols in this library that belong to the interface, have the
5908: prefix @code{forth_}. (Global symbols that are used internally have the
5909: prefix @code{gforth_}).
5910:
5911: You can include the declarations of Forth types and the functions and
5912: variables of the interface with @code{#include <forth.h>}.
5913:
5914: Types.
5915:
5916: Variables.
5917:
5918: Data and FP Stack pointer. Area sizes.
5919:
5920: functions.
5921:
5922: forth_init(imagefile)
5923: forth_evaluate(string) exceptions?
5924: forth_goto(address) (or forth_execute(xt)?)
5925: forth_continue() (a corountining mechanism)
5926:
5927: Adding primitives.
5928:
5929: No checking.
5930:
5931: Signals?
5932:
5933: Accessing the Stacks
5934:
5935: @node Emacs and Gforth, Image Files, Integrating Gforth, Top
5936: @chapter Emacs and Gforth
5937: @cindex Emacs and Gforth
5938:
5939: @cindex @file{gforth.el}
5940: @cindex @file{forth.el}
5941: @cindex Rydqvist, Goran
5942: @cindex comment editing commands
5943: @cindex @code{\}, editing with Emacs
5944: @cindex debug tracer editing commands
5945: @cindex @code{~~}, removal with Emacs
5946: @cindex Forth mode in Emacs
5947: Gforth comes with @file{gforth.el}, an improved version of
5948: @file{forth.el} by Goran Rydqvist (included in the TILE package). The
5949: improvements are a better (but still not perfect) handling of
5950: indentation. I have also added comment paragraph filling (@kbd{M-q}),
5951: commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) regions and
5952: removing debugging tracers (@kbd{C-x ~}, @pxref{Debugging}). I left the
5953: stuff I do not use alone, even though some of it only makes sense for
5954: TILE. To get a description of these features, enter Forth mode and type
5955: @kbd{C-h m}.
5956:
5957: @cindex source location of error or debugging output in Emacs
5958: @cindex error output, finding the source location in Emacs
5959: @cindex debugging output, finding the source location in Emacs
5960: In addition, Gforth supports Emacs quite well: The source code locations
5961: given in error messages, debugging output (from @code{~~}) and failed
5962: assertion messages are in the right format for Emacs' compilation mode
5963: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
5964: Manual}) so the source location corresponding to an error or other
5965: message is only a few keystrokes away (@kbd{C-x `} for the next error,
5966: @kbd{C-c C-c} for the error under the cursor).
5967:
5968: @cindex @file{TAGS} file
5969: @cindex @file{etags.fs}
5970: @cindex viewing the source of a word in Emacs
5971: Also, if you @code{include} @file{etags.fs}, a new @file{TAGS} file
5972: (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) will be produced that
5973: contains the definitions of all words defined afterwards. You can then
5974: find the source for a word using @kbd{M-.}. Note that emacs can use
5975: several tags files at the same time (e.g., one for the Gforth sources
5976: and one for your program, @pxref{Select Tags Table,,Selecting a Tags
5977: Table,emacs, Emacs Manual}). The TAGS file for the preloaded words is
5978: @file{$(datadir)/gforth/$(VERSION)/TAGS} (e.g.,
5979: @file{/usr/local/share/gforth/0.2.0/TAGS}).
5980:
5981: @cindex @file{.emacs}
5982: To get all these benefits, add the following lines to your @file{.emacs}
5983: file:
5984:
5985: @example
5986: (autoload 'forth-mode "gforth.el")
5987: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
5988: @end example
5989:
5990: @node Image Files, Engine, Emacs and Gforth, Top
5991: @chapter Image Files
5992: @cindex image files
5993: @cindex @code{.fi} files
5994: @cindex precompiled Forth code
5995: @cindex dictionary in persistent form
5996: @cindex persistent form of dictionary
5997:
5998: An image file is a file containing an image of the Forth dictionary,
5999: i.e., compiled Forth code and data residing in the dictionary. By
6000: convention, we use the extension @code{.fi} for image files.
6001:
6002: @menu
6003: * Image File Background:: Why have image files?
6004: * Non-Relocatable Image Files:: don't always work.
6005: * Data-Relocatable Image Files:: are better.
6006: * Fully Relocatable Image Files:: better yet.
6007: * Stack and Dictionary Sizes:: Setting the default sizes for an image.
6008: * Running Image Files:: @code{gforth -i @var{file}} or @var{file}.
6009: * Modifying the Startup Sequence:: and turnkey applications.
6010: @end menu
6011:
6012: @node Image File Background, Non-Relocatable Image Files, Image Files, Image Files
6013: @section Image File Background
6014: @cindex image file background
6015:
6016: Our Forth system consists not only of primitives, but also of
6017: definitions written in Forth. Since the Forth compiler itself belongs to
6018: those definitions, it is not possible to start the system with the
6019: primitives and the Forth source alone. Therefore we provide the Forth
6020: code as an image file in nearly executable form. At the start of the
6021: system a C routine loads the image file into memory, optionally
6022: relocates the addresses, then sets up the memory (stacks etc.) according
6023: to information in the image file, and starts executing Forth code.
6024:
6025: The image file variants represent different compromises between the
6026: goals of making it easy to generate image files and making them
6027: portable.
6028:
6029: @cindex relocation at run-time
6030: Win32Forth 3.4 and Mitch Bradleys @code{cforth} use relocation at
6031: run-time. This avoids many of the complications discussed below (image
6032: files are data relocatable without further ado), but costs performance
6033: (one addition per memory access).
6034:
6035: @cindex relocation at load-time
6036: By contrast, our loader performs relocation at image load time. The
6037: loader also has to replace tokens standing for primitive calls with the
6038: appropriate code-field addresses (or code addresses in the case of
6039: direct threading).
6040:
6041: There are three kinds of image files, with different degrees of
6042: relocatability: non-relocatable, data-relocatable, and fully relocatable
6043: image files.
6044:
6045: @cindex image file loader
6046: @cindex relocating loader
6047: @cindex loader for image files
6048: These image file variants have several restrictions in common; they are
6049: caused by the design of the image file loader:
6050:
6051: @itemize @bullet
6052: @item
6053: There is only one segment; in particular, this means, that an image file
6054: cannot represent @code{ALLOCATE}d memory chunks (and pointers to
6055: them). And the contents of the stacks are not represented, either.
6056:
6057: @item
6058: The only kinds of relocation supported are: adding the same offset to
6059: all cells that represent data addresses; and replacing special tokens
6060: with code addresses or with pieces of machine code.
6061:
6062: If any complex computations involving addresses are performed, the
6063: results cannot be represented in the image file. Several applications that
6064: use such computations come to mind:
6065: @itemize @minus
6066: @item
6067: Hashing addresses (or data structures which contain addresses) for table
6068: lookup. If you use Gforth's @code{table}s or @code{wordlist}s for this
6069: purpose, you will have no problem, because the hash tables are
6070: recomputed automatically when the system is started. If you use your own
6071: hash tables, you will have to do something similar.
6072:
6073: @item
6074: There's a cute implementation of doubly-linked lists that uses
6075: @code{XOR}ed addresses. You could represent such lists as singly-linked
6076: in the image file, and restore the doubly-linked representation on
6077: startup.@footnote{In my opinion, though, you should think thrice before
6078: using a doubly-linked list (whatever implementation).}
6079:
6080: @item
6081: The code addresses of run-time routines like @code{docol:} cannot be
6082: represented in the image file (because their tokens would be replaced by
6083: machine code in direct threaded implementations). As a workaround,
6084: compute these addresses at run-time with @code{>code-address} from the
6085: executions tokens of appropriate words (see the definitions of
6086: @code{docol:} and friends in @file{kernel.fs}).
6087:
6088: @item
6089: On many architectures addresses are represented in machine code in some
6090: shifted or mangled form. You cannot put @code{CODE} words that contain
6091: absolute addresses in this form in a relocatable image file. Workarounds
6092: are representing the address in some relative form (e.g., relative to
6093: the CFA, which is present in some register), or loading the address from
6094: a place where it is stored in a non-mangled form.
6095: @end itemize
6096: @end itemize
6097:
6098: @node Non-Relocatable Image Files, Data-Relocatable Image Files, Image File Background, Image Files
6099: @section Non-Relocatable Image Files
6100: @cindex non-relocatable image files
6101: @cindex image files, non-relocatable
6102:
6103: These files are simple memory dumps of the dictionary. They are specific
6104: to the executable (i.e., @file{gforth} file) they were created
6105: with. What's worse, they are specific to the place on which the
6106: dictionary resided when the image was created. Now, there is no
6107: guarantee that the dictionary will reside at the same place the next
6108: time you start Gforth, so there's no guarantee that a non-relocatable
6109: image will work the next time (Gforth will complain instead of crashing,
6110: though).
6111:
6112: You can create a non-relocatable image file with
6113:
6114: doc-savesystem
6115:
6116: @node Data-Relocatable Image Files, Fully Relocatable Image Files, Non-Relocatable Image Files, Image Files
6117: @section Data-Relocatable Image Files
6118: @cindex data-relocatable image files
6119: @cindex image files, data-relocatable
6120:
6121: These files contain relocatable data addresses, but fixed code addresses
6122: (instead of tokens). They are specific to the executable (i.e.,
6123: @file{gforth} file) they were created with. For direct threading on some
6124: architectures (e.g., the i386), data-relocatable images do not work. You
6125: get a data-relocatable image, if you use @file{gforthmi} with a
6126: Gforth binary that is not doubly indirect threaded (@pxref{Fully
6127: Relocatable Image Files}).
6128:
6129: @node Fully Relocatable Image Files, Stack and Dictionary Sizes, Data-Relocatable Image Files, Image Files
6130: @section Fully Relocatable Image Files
6131: @cindex fully relocatable image files
6132: @cindex image files, fully relocatable
6133:
6134: @cindex @file{kern*.fi}, relocatability
6135: @cindex @file{gforth.fi}, relocatability
6136: These image files have relocatable data addresses, and tokens for code
6137: addresses. They can be used with different binaries (e.g., with and
6138: without debugging) on the same machine, and even across machines with
6139: the same data formats (byte order, cell size, floating point
6140: format). However, they are usually specific to the version of Gforth
6141: they were created with. The files @file{gforth.fi} and @file{kernl*.fi}
6142: are fully relocatable.
6143:
6144: There are two ways to create a fully relocatable image file:
6145:
6146: @menu
6147: * gforthmi:: The normal way
6148: * cross.fs:: The hard way
6149: @end menu
6150:
6151: @node gforthmi, cross.fs, Fully Relocatable Image Files, Fully Relocatable Image Files
6152: @subsection @file{gforthmi}
6153: @cindex @file{comp-i.fs}
6154: @cindex @file{gforthmi}
6155:
6156: You will usually use @file{gforthmi}. If you want to create an
6157: image @var{file} that contains everything you would load by invoking
6158: Gforth with @code{gforth @var{options}}, you simply say
6159: @example
6160: gforthmi @var{file} @var{options}
6161: @end example
6162:
6163: E.g., if you want to create an image @file{asm.fi} that has the file
6164: @file{asm.fs} loaded in addition to the usual stuff, you could do it
6165: like this:
6166:
6167: @example
6168: gforthmi asm.fi asm.fs
6169: @end example
6170:
6171: @file{gforthmi} works like this: It produces two non-relocatable
6172: images for different addresses and then compares them. Its output
6173: reflects this: first you see the output (if any) of the two Gforth
6174: invocations that produce the nonrelocatable image files, then you see
6175: the output of the comparing program: It displays the offset used for
6176: data addresses and the offset used for code addresses;
6177: moreover, for each cell that cannot be represented correctly in the
6178: image files, it displays a line like the following one:
6179:
6180: @example
6181: 78DC BFFFFA50 BFFFFA40
6182: @end example
6183:
6184: This means that at offset $78dc from @code{forthstart}, one input image
6185: contains $bffffa50, and the other contains $bffffa40. Since these cells
6186: cannot be represented correctly in the output image, you should examine
6187: these places in the dictionary and verify that these cells are dead
6188: (i.e., not read before they are written).
6189:
6190: @cindex @code{savesystem} during @file{gforthmi}
6191: @cindex @code{bye} during @file{gforthmi}
6192: @cindex doubly indirect threaded code
6193: @cindex environment variable @code{GFORTHD}
6194: @cindex @code{GFORTHD} environment variable
6195: @cindex @code{gforth-ditc}
6196: There are a few wrinkles: After processing the passed @var{options}, the
6197: words @code{savesystem} and @code{bye} must be visible. A special doubly
6198: indirect threaded version of the @file{gforth} executable is used for
6199: creating the nonrelocatable images; you can pass the exact filename of
6200: this executable through the environment variable @code{GFORTHD}
6201: (default: @file{gforth-ditc}); if you pass a version that is not doubly
6202: indirect threaded, you will not get a fully relocatable image, but a
6203: data-relocatable image (because there is no code address offset).
6204:
6205: @node cross.fs, , gforthmi, Fully Relocatable Image Files
6206: @subsection @file{cross.fs}
6207: @cindex @file{cross.fs}
6208: @cindex cross-compiler
6209: @cindex metacompiler
6210:
6211: You can also use @code{cross}, a batch compiler that accepts a Forth-like
6212: programming language. This @code{cross} language has to be documented
6213: yet.
6214:
6215: @cindex target compiler
6216: @code{cross} also allows you to create image files for machines with
6217: different data sizes and data formats than the one used for generating
6218: the image file. You can also use it to create an application image that
6219: does not contain a Forth compiler. These features are bought with
6220: restrictions and inconveniences in programming. E.g., addresses have to
6221: be stored in memory with special words (@code{A!}, @code{A,}, etc.) in
6222: order to make the code relocatable.
6223:
6224:
6225: @node Stack and Dictionary Sizes, Running Image Files, Fully Relocatable Image Files, Image Files
6226: @section Stack and Dictionary Sizes
6227: @cindex image file, stack and dictionary sizes
6228: @cindex dictionary size default
6229: @cindex stack size default
6230:
6231: If you invoke Gforth with a command line flag for the size
6232: (@pxref{Invoking Gforth}), the size you specify is stored in the
6233: dictionary. If you save the dictionary with @code{savesystem} or create
6234: an image with @file{gforthmi}, this size will become the default
6235: for the resulting image file. E.g., the following will create a
6236: fully relocatable version of gforth.fi with a 1MB dictionary:
6237:
6238: @example
6239: gforthmi gforth.fi -m 1M
6240: @end example
6241:
6242: In other words, if you want to set the default size for the dictionary
6243: and the stacks of an image, just invoke @file{gforthmi} with the
6244: appropriate options when creating the image.
6245:
6246: @cindex stack size, cache-friendly
6247: Note: For cache-friendly behaviour (i.e., good performance), you should
6248: make the sizes of the stacks modulo, say, 2K, somewhat different. E.g.,
6249: the default stack sizes are: data: 16k (mod 2k=0); fp: 15.5k (mod
6250: 2k=1.5k); return: 15k(mod 2k=1k); locals: 14.5k (mod 2k=0.5k).
6251:
6252: @node Running Image Files, Modifying the Startup Sequence, Stack and Dictionary Sizes, Image Files
6253: @section Running Image Files
6254: @cindex running image files
6255: @cindex invoking image files
6256: @cindex image file invocation
6257:
6258: @cindex -i, invoke image file
6259: @cindex --image file, invoke image file
6260: You can invoke Gforth with an image file @var{image} instead of the
6261: default @file{gforth.fi} with the @code{-i} flag (@pxref{Invoking Gforth}):
6262: @example
6263: gforth -i @var{image}
6264: @end example
6265:
6266: @cindex executable image file
6267: @cindex image files, executable
6268: If your operating system supports starting scripts with a line of the
6269: form @code{#! ...}, you just have to type the image file name to start
6270: Gforth with this image file (note that the file extension @code{.fi} is
6271: just a convention). I.e., to run Gforth with the image file @var{image},
6272: you can just type @var{image} instead of @code{gforth -i @var{image}}.
6273:
6274: doc-#!
6275:
6276: @node Modifying the Startup Sequence, , Running Image Files, Image Files
6277: @section Modifying the Startup Sequence
6278: @cindex startup sequence for image file
6279: @cindex image file initialization sequence
6280: @cindex initialization sequence of image file
6281:
6282: You can add your own initialization to the startup sequence through the
6283: deferred word
6284:
6285: doc-'cold
6286:
6287: @code{'cold} is invoked just before the image-specific command line
6288: processing (by default, loading files and evaluating (@code{-e}) strings)
6289: starts.
6290:
6291: A sequence for adding your initialization usually looks like this:
6292:
6293: @example
6294: :noname
6295: Defers 'cold \ do other initialization stuff (e.g., rehashing wordlists)
6296: ... \ your stuff
6297: ; IS 'cold
6298: @end example
6299:
6300: @cindex turnkey image files
6301: @cindex image files, turnkey applications
6302: You can make a turnkey image by letting @code{'cold} execute a word
6303: (your turnkey application) that never returns; instead, it exits Gforth
6304: via @code{bye} or @code{throw}.
6305:
6306: @cindex command-line arguments, access
6307: @cindex arguments on the command line, access
6308: You can access the (image-specific) command-line arguments through the
6309: variables @code{argc} and @code{argv}. @code{arg} provides conventient
6310: access to @code{argv}.
6311:
6312: doc-argc
6313: doc-argv
6314: doc-arg
6315:
6316: If @code{'cold} exits normally, Gforth processes the command-line
6317: arguments as files to be loaded and strings to be evaluated. Therefore,
6318: @code{'cold} should remove the arguments it has used in this case.
6319:
6320: @c ******************************************************************
6321: @node Engine, Bugs, Image Files, Top
6322: @chapter Engine
6323: @cindex engine
6324: @cindex virtual machine
6325:
6326: Reading this section is not necessary for programming with Gforth. It
6327: may be helpful for finding your way in the Gforth sources.
6328:
6329: The ideas in this section have also been published in the papers
6330: @cite{ANS fig/GNU/??? Forth} (in German) by Bernd Paysan, presented at
6331: the Forth-Tagung '93 and @cite{A Portable Forth Engine} by M. Anton
6332: Ertl, presented at EuroForth '93; the latter is available at
6333: @*@url{http://www.complang.tuwien.ac.at/papers/ertl93.ps.Z}.
6334:
6335: @menu
6336: * Portability::
6337: * Threading::
6338: * Primitives::
6339: * Performance::
6340: @end menu
6341:
6342: @node Portability, Threading, Engine, Engine
6343: @section Portability
6344: @cindex engine portability
6345:
6346: One of the main goals of the effort is availability across a wide range
6347: of personal machines. fig-Forth, and, to a lesser extent, F83, achieved
6348: this goal by manually coding the engine in assembly language for several
6349: then-popular processors. This approach is very labor-intensive and the
6350: results are short-lived due to progress in computer architecture.
6351:
6352: @cindex C, using C for the engine
6353: Others have avoided this problem by coding in C, e.g., Mitch Bradley
6354: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
6355: particularly popular for UNIX-based Forths due to the large variety of
6356: architectures of UNIX machines. Unfortunately an implementation in C
6357: does not mix well with the goals of efficiency and with using
6358: traditional techniques: Indirect or direct threading cannot be expressed
6359: in C, and switch threading, the fastest technique available in C, is
6360: significantly slower. Another problem with C is that it is very
6361: cumbersome to express double integer arithmetic.
6362:
6363: @cindex GNU C for the engine
6364: @cindex long long
6365: Fortunately, there is a portable language that does not have these
6366: limitations: GNU C, the version of C processed by the GNU C compiler
6367: (@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
6368: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
6369: Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
6370: threading possible, its @code{long long} type (@pxref{Long Long, ,
6371: Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forth's
6372: double numbers@footnote{Unfortunately, long longs are not implemented
6373: properly on all machines (e.g., on alpha-osf1, long longs are only 64
6374: bits, the same size as longs (and pointers), but they should be twice as
6375: long according to @pxref{Long Long, , Double-Word Integers, gcc.info, GNU
6376: C Manual}). So, we had to implement doubles in C after all. Still, on
6377: most machines we can use long longs and achieve better performance than
6378: with the emulation package.}. GNU C is available for free on all
6379: important (and many unimportant) UNIX machines, VMS, 80386s running
6380: MS-DOS, the Amiga, and the Atari ST, so a Forth written in GNU C can run
6381: on all these machines.
6382:
6383: Writing in a portable language has the reputation of producing code that
6384: is slower than assembly. For our Forth engine we repeatedly looked at
6385: the code produced by the compiler and eliminated most compiler-induced
6386: inefficiencies by appropriate changes in the source code.
6387:
6388: @cindex explicit register declarations
6389: @cindex --enable-force-reg, configuration flag
6390: @cindex -DFORCE_REG
6391: However, register allocation cannot be portably influenced by the
6392: programmer, leading to some inefficiencies on register-starved
6393: machines. We use explicit register declarations (@pxref{Explicit Reg
6394: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
6395: improve the speed on some machines. They are turned on by using the
6396: configuration flag @code{--enable-force-reg} (@code{gcc} switch
6397: @code{-DFORCE_REG}). Unfortunately, this feature not only depends on the
6398: machine, but also on the compiler version: On some machines some
6399: compiler versions produce incorrect code when certain explicit register
6400: declarations are used. So by default @code{-DFORCE_REG} is not used.
6401:
6402: @node Threading, Primitives, Portability, Engine
6403: @section Threading
6404: @cindex inner interpreter implementation
6405: @cindex threaded code implementation
6406:
6407: @cindex labels as values
6408: GNU C's labels as values extension (available since @code{gcc-2.0},
6409: @pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
6410: makes it possible to take the address of @var{label} by writing
6411: @code{&&@var{label}}. This address can then be used in a statement like
6412: @code{goto *@var{address}}. I.e., @code{goto *&&x} is the same as
6413: @code{goto x}.
6414:
6415: @cindex NEXT, indirect threaded
6416: @cindex indirect threaded inner interpreter
6417: @cindex inner interpreter, indirect threaded
6418: With this feature an indirect threaded NEXT looks like:
6419: @example
6420: cfa = *ip++;
6421: ca = *cfa;
6422: goto *ca;
6423: @end example
6424: @cindex instruction pointer
6425: For those unfamiliar with the names: @code{ip} is the Forth instruction
6426: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
6427: execution token and points to the code field of the next word to be
6428: executed; The @code{ca} (code address) fetched from there points to some
6429: executable code, e.g., a primitive or the colon definition handler
6430: @code{docol}.
6431:
6432: @cindex NEXT, direct threaded
6433: @cindex direct threaded inner interpreter
6434: @cindex inner interpreter, direct threaded
6435: Direct threading is even simpler:
6436: @example
6437: ca = *ip++;
6438: goto *ca;
6439: @end example
6440:
6441: Of course we have packaged the whole thing neatly in macros called
6442: @code{NEXT} and @code{NEXT1} (the part of NEXT after fetching the cfa).
6443:
6444: @menu
6445: * Scheduling::
6446: * Direct or Indirect Threaded?::
6447: * DOES>::
6448: @end menu
6449:
6450: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
6451: @subsection Scheduling
6452: @cindex inner interpreter optimization
6453:
6454: There is a little complication: Pipelined and superscalar processors,
6455: i.e., RISC and some modern CISC machines can process independent
6456: instructions while waiting for the results of an instruction. The
6457: compiler usually reorders (schedules) the instructions in a way that
6458: achieves good usage of these delay slots. However, on our first tries
6459: the compiler did not do well on scheduling primitives. E.g., for
6460: @code{+} implemented as
6461: @example
6462: n=sp[0]+sp[1];
6463: sp++;
6464: sp[0]=n;
6465: NEXT;
6466: @end example
6467: the NEXT comes strictly after the other code, i.e., there is nearly no
6468: scheduling. After a little thought the problem becomes clear: The
6469: compiler cannot know that sp and ip point to different addresses (and
6470: the version of @code{gcc} we used would not know it even if it was
6471: possible), so it could not move the load of the cfa above the store to
6472: the TOS. Indeed the pointers could be the same, if code on or very near
6473: the top of stack were executed. In the interest of speed we chose to
6474: forbid this probably unused ``feature'' and helped the compiler in
6475: scheduling: NEXT is divided into the loading part (@code{NEXT_P1}) and
6476: the goto part (@code{NEXT_P2}). @code{+} now looks like:
6477: @example
6478: n=sp[0]+sp[1];
6479: sp++;
6480: NEXT_P1;
6481: sp[0]=n;
6482: NEXT_P2;
6483: @end example
6484: This can be scheduled optimally by the compiler.
6485:
6486: This division can be turned off with the switch @code{-DCISC_NEXT}. This
6487: switch is on by default on machines that do not profit from scheduling
6488: (e.g., the 80386), in order to preserve registers.
6489:
6490: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
6491: @subsection Direct or Indirect Threaded?
6492: @cindex threading, direct or indirect?
6493:
6494: @cindex -DDIRECT_THREADED
6495: Both! After packaging the nasty details in macro definitions we
6496: realized that we could switch between direct and indirect threading by
6497: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
6498: defining a few machine-specific macros for the direct-threading case.
6499: On the Forth level we also offer access words that hide the
6500: differences between the threading methods (@pxref{Threading Words}).
6501:
6502: Indirect threading is implemented completely machine-independently.
6503: Direct threading needs routines for creating jumps to the executable
6504: code (e.g. to docol or dodoes). These routines are inherently
6505: machine-dependent, but they do not amount to many source lines. I.e.,
6506: even porting direct threading to a new machine is a small effort.
6507:
6508: @cindex --enable-indirect-threaded, configuration flag
6509: @cindex --enable-direct-threaded, configuration flag
6510: The default threading method is machine-dependent. You can enforce a
6511: specific threading method when building Gforth with the configuration
6512: flag @code{--enable-direct-threaded} or
6513: @code{--enable-indirect-threaded}. Note that direct threading is not
6514: supported on all machines.
6515:
6516: @node DOES>, , Direct or Indirect Threaded?, Threading
6517: @subsection DOES>
6518: @cindex @code{DOES>} implementation
6519:
6520: @cindex dodoes routine
6521: @cindex DOES-code
6522: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
6523: the chunk of code executed by every word defined by a
6524: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
6525: the Forth code to be executed, i.e. the code after the
6526: @code{DOES>} (the DOES-code)? There are two solutions:
6527:
6528: In fig-Forth the code field points directly to the dodoes and the
6529: DOES-code address is stored in the cell after the code address (i.e. at
6530: @code{@var{cfa} cell+}). It may seem that this solution is illegal in
6531: the Forth-79 and all later standards, because in fig-Forth this address
6532: lies in the body (which is illegal in these standards). However, by
6533: making the code field larger for all words this solution becomes legal
6534: again. We use this approach for the indirect threaded version and for
6535: direct threading on some machines. Leaving a cell unused in most words
6536: is a bit wasteful, but on the machines we are targeting this is hardly a
6537: problem. The other reason for having a code field size of two cells is
6538: to avoid having different image files for direct and indirect threaded
6539: systems (direct threaded systems require two-cell code fields on many
6540: machines).
6541:
6542: @cindex DOES-handler
6543: The other approach is that the code field points or jumps to the cell
6544: after @code{DOES}. In this variant there is a jump to @code{dodoes} at
6545: this address (the DOES-handler). @code{dodoes} can then get the
6546: DOES-code address by computing the code address, i.e., the address of
6547: the jump to dodoes, and add the length of that jump field. A variant of
6548: this is to have a call to @code{dodoes} after the @code{DOES>}; then the
6549: return address (which can be found in the return register on RISCs) is
6550: the DOES-code address. Since the two cells available in the code field
6551: are used up by the jump to the code address in direct threading on many
6552: architectures, we use this approach for direct threading on these
6553: architectures. We did not want to add another cell to the code field.
6554:
6555: @node Primitives, Performance, Threading, Engine
6556: @section Primitives
6557: @cindex primitives, implementation
6558: @cindex virtual machine instructions, implementation
6559:
6560: @menu
6561: * Automatic Generation::
6562: * TOS Optimization::
6563: * Produced code::
6564: @end menu
6565:
6566: @node Automatic Generation, TOS Optimization, Primitives, Primitives
6567: @subsection Automatic Generation
6568: @cindex primitives, automatic generation
6569:
6570: @cindex @file{prims2x.fs}
6571: Since the primitives are implemented in a portable language, there is no
6572: longer any need to minimize the number of primitives. On the contrary,
6573: having many primitives has an advantage: speed. In order to reduce the
6574: number of errors in primitives and to make programming them easier, we
6575: provide a tool, the primitive generator (@file{prims2x.fs}), that
6576: automatically generates most (and sometimes all) of the C code for a
6577: primitive from the stack effect notation. The source for a primitive
6578: has the following form:
6579:
6580: @cindex primitive source format
6581: @format
6582: @var{Forth-name} @var{stack-effect} @var{category} [@var{pronounc.}]
6583: [@code{""}@var{glossary entry}@code{""}]
6584: @var{C code}
6585: [@code{:}
6586: @var{Forth code}]
6587: @end format
6588:
6589: The items in brackets are optional. The category and glossary fields
6590: are there for generating the documentation, the Forth code is there
6591: for manual implementations on machines without GNU C. E.g., the source
6592: for the primitive @code{+} is:
6593: @example
6594: + n1 n2 -- n core plus
6595: n = n1+n2;
6596: @end example
6597:
6598: This looks like a specification, but in fact @code{n = n1+n2} is C
6599: code. Our primitive generation tool extracts a lot of information from
6600: the stack effect notations@footnote{We use a one-stack notation, even
6601: though we have separate data and floating-point stacks; The separate
6602: notation can be generated easily from the unified notation.}: The number
6603: of items popped from and pushed on the stack, their type, and by what
6604: name they are referred to in the C code. It then generates a C code
6605: prelude and postlude for each primitive. The final C code for @code{+}
6606: looks like this:
6607:
6608: @example
6609: I_plus: /* + ( n1 n2 -- n ) */ /* label, stack effect */
6610: /* */ /* documentation */
6611: @{
6612: DEF_CA /* definition of variable ca (indirect threading) */
6613: Cell n1; /* definitions of variables */
6614: Cell n2;
6615: Cell n;
6616: n1 = (Cell) sp[1]; /* input */
6617: n2 = (Cell) TOS;
6618: sp += 1; /* stack adjustment */
6619: NAME("+") /* debugging output (with -DDEBUG) */
6620: @{
6621: n = n1+n2; /* C code taken from the source */
6622: @}
6623: NEXT_P1; /* NEXT part 1 */
6624: TOS = (Cell)n; /* output */
6625: NEXT_P2; /* NEXT part 2 */
6626: @}
6627: @end example
6628:
6629: This looks long and inefficient, but the GNU C compiler optimizes quite
6630: well and produces optimal code for @code{+} on, e.g., the R3000 and the
6631: HP RISC machines: Defining the @code{n}s does not produce any code, and
6632: using them as intermediate storage also adds no cost.
6633:
6634: There are also other optimizations, that are not illustrated by this
6635: example: Assignments between simple variables are usually for free (copy
6636: propagation). If one of the stack items is not used by the primitive
6637: (e.g. in @code{drop}), the compiler eliminates the load from the stack
6638: (dead code elimination). On the other hand, there are some things that
6639: the compiler does not do, therefore they are performed by
6640: @file{prims2x.fs}: The compiler does not optimize code away that stores
6641: a stack item to the place where it just came from (e.g., @code{over}).
6642:
6643: While programming a primitive is usually easy, there are a few cases
6644: where the programmer has to take the actions of the generator into
6645: account, most notably @code{?dup}, but also words that do not (always)
6646: fall through to NEXT.
6647:
6648: @node TOS Optimization, Produced code, Automatic Generation, Primitives
6649: @subsection TOS Optimization
6650: @cindex TOS optimization for primitives
6651: @cindex primitives, keeping the TOS in a register
6652:
6653: An important optimization for stack machine emulators, e.g., Forth
6654: engines, is keeping one or more of the top stack items in
6655: registers. If a word has the stack effect @var{in1}...@var{inx} @code{--}
6656: @var{out1}...@var{outy}, keeping the top @var{n} items in registers
6657: @itemize @bullet
6658: @item
6659: is better than keeping @var{n-1} items, if @var{x>=n} and @var{y>=n},
6660: due to fewer loads from and stores to the stack.
6661: @item is slower than keeping @var{n-1} items, if @var{x<>y} and @var{x<n} and
6662: @var{y<n}, due to additional moves between registers.
6663: @end itemize
6664:
6665: @cindex -DUSE_TOS
6666: @cindex -DUSE_NO_TOS
6667: In particular, keeping one item in a register is never a disadvantage,
6668: if there are enough registers. Keeping two items in registers is a
6669: disadvantage for frequent words like @code{?branch}, constants,
6670: variables, literals and @code{i}. Therefore our generator only produces
6671: code that keeps zero or one items in registers. The generated C code
6672: covers both cases; the selection between these alternatives is made at
6673: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
6674: code for @code{+} is just a simple variable name in the one-item case,
6675: otherwise it is a macro that expands into @code{sp[0]}. Note that the
6676: GNU C compiler tries to keep simple variables like @code{TOS} in
6677: registers, and it usually succeeds, if there are enough registers.
6678:
6679: @cindex -DUSE_FTOS
6680: @cindex -DUSE_NO_FTOS
6681: The primitive generator performs the TOS optimization for the
6682: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
6683: operations the benefit of this optimization is even larger:
6684: floating-point operations take quite long on most processors, but can be
6685: performed in parallel with other operations as long as their results are
6686: not used. If the FP-TOS is kept in a register, this works. If
6687: it is kept on the stack, i.e., in memory, the store into memory has to
6688: wait for the result of the floating-point operation, lengthening the
6689: execution time of the primitive considerably.
6690:
6691: The TOS optimization makes the automatic generation of primitives a
6692: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
6693: @code{TOS} is not sufficient. There are some special cases to
6694: consider:
6695: @itemize @bullet
6696: @item In the case of @code{dup ( w -- w w )} the generator must not
6697: eliminate the store to the original location of the item on the stack,
6698: if the TOS optimization is turned on.
6699: @item Primitives with stack effects of the form @code{--}
6700: @var{out1}...@var{outy} must store the TOS to the stack at the start.
6701: Likewise, primitives with the stack effect @var{in1}...@var{inx} @code{--}
6702: must load the TOS from the stack at the end. But for the null stack
6703: effect @code{--} no stores or loads should be generated.
6704: @end itemize
6705:
6706: @node Produced code, , TOS Optimization, Primitives
6707: @subsection Produced code
6708: @cindex primitives, assembly code listing
6709:
6710: @cindex @file{engine.s}
6711: To see what assembly code is produced for the primitives on your machine
6712: with your compiler and your flag settings, type @code{make engine.s} and
6713: look at the resulting file @file{engine.s}.
6714:
6715: @node Performance, , Primitives, Engine
6716: @section Performance
6717: @cindex performance of some Forth interpreters
6718: @cindex engine performance
6719: @cindex benchmarking Forth systems
6720: @cindex Gforth performance
6721:
6722: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
6723: impossible to write a significantly faster engine.
6724:
6725: On register-starved machines like the 386 architecture processors
6726: improvements are possible, because @code{gcc} does not utilize the
6727: registers as well as a human, even with explicit register declarations;
6728: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
6729: and hand-tuned it for the 486; this system is 1.19 times faster on the
6730: Sieve benchmark on a 486DX2/66 than Gforth compiled with
6731: @code{gcc-2.6.3} with @code{-DFORCE_REG}.
6732:
6733: @cindex Win32Forth performance
6734: @cindex NT Forth performance
6735: @cindex eforth performance
6736: @cindex ThisForth performance
6737: @cindex PFE performance
6738: @cindex TILE performance
6739: However, this potential advantage of assembly language implementations
6740: is not necessarily realized in complete Forth systems: We compared
6741: Gforth (direct threaded, compiled with @code{gcc-2.6.3} and
6742: @code{-DFORCE_REG}) with Win32Forth 1.2093, LMI's NT Forth (Beta, May
6743: 1994) and Eforth (with and without peephole (aka pinhole) optimization
6744: of the threaded code); all these systems were written in assembly
6745: language. We also compared Gforth with three systems written in C:
6746: PFE-0.9.14 (compiled with @code{gcc-2.6.3} with the default
6747: configuration for Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS
6748: -DUNROLL_NEXT}), ThisForth Beta (compiled with gcc-2.6.3 -O3
6749: -fomit-frame-pointer; ThisForth employs peephole optimization of the
6750: threaded code) and TILE (compiled with @code{make opt}). We benchmarked
6751: Gforth, PFE, ThisForth and TILE on a 486DX2/66 under Linux. Kenneth
6752: O'Heskin kindly provided the results for Win32Forth and NT Forth on a
6753: 486DX2/66 with similar memory performance under Windows NT. Marcel
6754: Hendrix ported Eforth to Linux, then extended it to run the benchmarks,
6755: added the peephole optimizer, ran the benchmarks and reported the
6756: results.
6757:
6758: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
6759: matrix multiplication come from the Stanford integer benchmarks and have
6760: been translated into Forth by Martin Fraeman; we used the versions
6761: included in the TILE Forth package, but with bigger data set sizes; and
6762: a recursive Fibonacci number computation for benchmarking calling
6763: performance. The following table shows the time taken for the benchmarks
6764: scaled by the time taken by Gforth (in other words, it shows the speedup
6765: factor that Gforth achieved over the other systems).
6766:
6767: @example
6768: relative Win32- NT eforth This-
6769: time Gforth Forth Forth eforth +opt PFE Forth TILE
6770: sieve 1.00 1.39 1.14 1.39 0.85 1.58 3.18 8.58
6771: bubble 1.00 1.31 1.41 1.48 0.88 1.50 3.88
6772: matmul 1.00 1.47 1.35 1.46 0.74 1.58 4.09
6773: fib 1.00 1.52 1.34 1.22 0.86 1.74 2.99 4.30
6774: @end example
6775:
6776: You may find the good performance of Gforth compared with the systems
6777: written in assembly language quite surprising. One important reason for
6778: the disappointing performance of these systems is probably that they are
6779: not written optimally for the 486 (e.g., they use the @code{lods}
6780: instruction). In addition, Win32Forth uses a comfortable, but costly
6781: method for relocating the Forth image: like @code{cforth}, it computes
6782: the actual addresses at run time, resulting in two address computations
6783: per NEXT (@pxref{Image File Background}).
6784:
6785: Only Eforth with the peephole optimizer performs comparable to
6786: Gforth. The speedups achieved with peephole optimization of threaded
6787: code are quite remarkable. Adding a peephole optimizer to Gforth should
6788: cause similar speedups.
6789:
6790: The speedup of Gforth over PFE, ThisForth and TILE can be easily
6791: explained with the self-imposed restriction of the latter systems to
6792: standard C, which makes efficient threading impossible (however, the
6793: measured implementation of PFE uses a GNU C extension: @pxref{Global Reg
6794: Vars, , Defining Global Register Variables, gcc.info, GNU C Manual}).
6795: Moreover, current C compilers have a hard time optimizing other aspects
6796: of the ThisForth and the TILE source.
6797:
6798: Note that the performance of Gforth on 386 architecture processors
6799: varies widely with the version of @code{gcc} used. E.g., @code{gcc-2.5.8}
6800: failed to allocate any of the virtual machine registers into real
6801: machine registers by itself and would not work correctly with explicit
6802: register declarations, giving a 1.3 times slower engine (on a 486DX2/66
6803: running the Sieve) than the one measured above.
6804:
6805: Note also that there have been several releases of Win32Forth since the
6806: release presented here, so the results presented here may have little
6807: predictive value for the performance of Win32Forth today.
6808:
6809: @cindex @file{Benchres}
6810: In @cite{Translating Forth to Efficient C} by M. Anton Ertl and Martin
6811: Maierhofer (presented at EuroForth '95), an indirect threaded version of
6812: Gforth is compared with Win32Forth, NT Forth, PFE, and ThisForth; that
6813: version of Gforth is 2%@minus{}8% slower on a 486 than the direct
6814: threaded version used here. The paper available at
6815: @*@url{http://www.complang.tuwien.ac.at/papers/ertl&maierhofer95.ps.gz};
6816: it also contains numbers for some native code systems. You can find a
6817: newer version of these measurements at
6818: @url{http://www.complang.tuwien.ac.at/forth/performance.html}. You can
6819: find numbers for Gforth on various machines in @file{Benchres}.
6820:
6821: @node Bugs, Origin, Engine, Top
6822: @chapter Bugs
6823: @cindex bug reporting
6824:
6825: Known bugs are described in the file BUGS in the Gforth distribution.
6826:
6827: If you find a bug, please send a bug report to
6828: @email{bug-gforth@@gnu.ai.mit.edu}. A bug report should
6829: describe the Gforth version used (it is announced at the start of an
6830: interactive Gforth session), the machine and operating system (on Unix
6831: systems you can use @code{uname -a} to produce this information), the
6832: installation options (send the @file{config.status} file), and a
6833: complete list of changes you (or your installer) have made to the Gforth
6834: sources (if any); it should contain a program (or a sequence of keyboard
6835: commands) that reproduces the bug and a description of what you think
6836: constitutes the buggy behaviour.
6837:
6838: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
6839: to Report Bugs, gcc.info, GNU C Manual}.
6840:
6841:
6842: @node Origin, Word Index, Bugs, Top
6843: @chapter Authors and Ancestors of Gforth
6844:
6845: @section Authors and Contributors
6846: @cindex authors of Gforth
6847: @cindex contributors to Gforth
6848:
6849: The Gforth project was started in mid-1992 by Bernd Paysan and Anton
6850: Ertl. The third major author was Jens Wilke. Lennart Benschop (who was
6851: one of Gforth's first users, in mid-1993) and Stuart Ramsden inspired us
6852: with their continuous feedback. Lennart Benshop contributed
6853: @file{glosgen.fs}, while Stuart Ramsden has been working on automatic
6854: support for calling C libraries. Helpful comments also came from Paul
6855: Kleinrubatscher, Christian Pirker, Dirk Zoller, Marcel Hendrix, John
6856: Wavrik, Barrie Stott and Marc de Groot.
6857:
6858: Gforth also owes a lot to the authors of the tools we used (GCC, CVS,
6859: and autoconf, among others), and to the creators of the Internet: Gforth
6860: was developed across the Internet, and its authors have not met
6861: physically yet.
6862:
6863: @section Pedigree
6864: @cindex Pedigree of Gforth
6865:
6866: Gforth descends from BigForth (1993) and fig-Forth. Gforth and PFE (by
6867: Dirk Zoller) will cross-fertilize each other. Of course, a significant
6868: part of the design of Gforth was prescribed by ANS Forth.
6869:
6870: Bernd Paysan wrote BigForth, a descendent from TurboForth, an unreleased
6871: 32 bit native code version of VolksForth for the Atari ST, written
6872: mostly by Dietrich Weineck.
6873:
6874: VolksForth descends from F83. It was written by Klaus Schleisiek, Bernd
6875: Pennemann, Georg Rehfeld and Dietrich Weineck for the C64 (called
6876: UltraForth there) in the mid-80s and ported to the Atari ST in 1986.
6877:
6878: Henry Laxen and Mike Perry wrote F83 as a model implementation of the
6879: Forth-83 standard. !! Pedigree? When?
6880:
6881: A team led by Bill Ragsdale implemented fig-Forth on many processors in
6882: 1979. Robert Selzer and Bill Ragsdale developed the original
6883: implementation of fig-Forth for the 6502 based on microForth.
6884:
6885: The principal architect of microForth was Dean Sanderson. microForth was
6886: FORTH, Inc.'s first off-the-shelf product. It was developed in 1976 for
6887: the 1802, and subsequently implemented on the 8080, the 6800 and the
6888: Z80.
6889:
6890: All earlier Forth systems were custom-made, usually by Charles Moore,
6891: who discovered (as he puts it) Forth during the late 60s. The first full
6892: Forth existed in 1971.
6893:
6894: A part of the information in this section comes from @cite{The Evolution
6895: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
6896: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
6897: Notices 28(3), 1993. You can find more historical and genealogical
6898: information about Forth there.
6899:
6900: @node Word Index, Concept Index, Origin, Top
6901: @unnumbered Word Index
6902:
6903: This index is as incomplete as the manual. Each word is listed with
6904: stack effect and wordset.
6905:
6906: @printindex fn
6907:
6908: @node Concept Index, , Word Index, Top
6909: @unnumbered Concept and Word Index
6910:
6911: This index is as incomplete as the manual. Not all entries listed are
6912: present verbatim in the text. Only the names are listed for the words
6913: here.
6914:
6915: @printindex cp
6916:
6917: @contents
6918: @bye
6919:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>