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