1: \input texinfo @c -*-texinfo-*-
2: @comment The source is gforth.ds, from which gforth.texi is generated
3: @comment %**start of header (This is for running Texinfo on a region.)
4: @setfilename gforth.info
5: @settitle Gforth Manual
6: @comment @setchapternewpage odd
7: @comment %**end of header (This is for running Texinfo on a region.)
8:
9: @ifinfo
10: This file documents Gforth 0.2
11:
12: Copyright @copyright{} 1995,1996 Free Software Foundation, Inc.
13:
14: Permission is granted to make and distribute verbatim copies of
15: this manual provided the copyright notice and this permission notice
16: are preserved on all copies.
17:
18: @ignore
19: Permission is granted to process this file through TeX and print the
20: results, provided the printed document carries a copying permission
21: notice identical to this one except for the removal of this paragraph
22: (this paragraph not being relevant to the printed manual).
23:
24: @end ignore
25: Permission is granted to copy and distribute modified versions of this
26: manual under the conditions for verbatim copying, provided also that the
27: sections entitled "Distribution" and "General Public License" are
28: included exactly as in the original, and provided that the entire
29: resulting derived work is distributed under the terms of a permission
30: notice identical to this one.
31:
32: Permission is granted to copy and distribute translations of this manual
33: into another language, under the above conditions for modified versions,
34: except that the sections entitled "Distribution" and "General Public
35: License" may be included in a translation approved by the author instead
36: of in the original English.
37: @end ifinfo
38:
39: @finalout
40: @titlepage
41: @sp 10
42: @center @titlefont{Gforth Manual}
43: @sp 2
44: @center for version 0.2
45: @sp 2
46: @center Anton Ertl
47: @center Bernd Paysan
48: @sp 3
49: @center This manual is under construction
50:
51: @comment The following two commands start the copyright page.
52: @page
53: @vskip 0pt plus 1filll
54: Copyright @copyright{} 1995,1996 Free Software Foundation, Inc.
55:
56: @comment !! Published by ... or You can get a copy of this manual ...
57:
58: Permission is granted to make and distribute verbatim copies of
59: this manual provided the copyright notice and this permission notice
60: are preserved on all copies.
61:
62: Permission is granted to copy and distribute modified versions of this
63: manual under the conditions for verbatim copying, provided also that the
64: sections entitled "Distribution" and "General Public License" are
65: included exactly as in the original, and provided that the entire
66: resulting derived work is distributed under the terms of a permission
67: notice identical to this one.
68:
69: Permission is granted to copy and distribute translations of this manual
70: into another language, under the above conditions for modified versions,
71: except that the sections entitled "Distribution" and "General Public
72: License" may be included in a translation approved by the author instead
73: of in the original English.
74: @end titlepage
75:
76:
77: @node Top, License, (dir), (dir)
78: @ifinfo
79: Gforth is a free implementation of ANS Forth available on many
80: personal machines. This manual corresponds to version 0.2.
81: @end ifinfo
82:
83: @menu
84: * License::
85: * Goals:: About the Gforth Project
86: * Other Books:: Things you might want to read
87: * Invocation:: Starting Gforth
88: * Words:: Forth words available in Gforth
89: * ANS conformance:: Implementation-defined options etc.
90: * Model:: The abstract machine of Gforth
91: * Integrating Gforth:: Forth as scripting language for applications.
92: * Emacs and Gforth:: The Gforth Mode
93: * Internals:: Implementation details
94: * Bugs:: How to report them
95: * Origin:: Authors and ancestors of Gforth
96: * Word Index:: An item for each Forth word
97: * Node Index:: An item for each node
98: @end menu
99:
100: @node License, Goals, Top, Top
101: @unnumbered GNU GENERAL PUBLIC LICENSE
102: @center Version 2, June 1991
103:
104: @display
105: Copyright @copyright{} 1989, 1991 Free Software Foundation, Inc.
106: 675 Mass Ave, Cambridge, MA 02139, USA
107:
108: Everyone is permitted to copy and distribute verbatim copies
109: of this license document, but changing it is not allowed.
110: @end display
111:
112: @unnumberedsec Preamble
113:
114: The licenses for most software are designed to take away your
115: freedom to share and change it. By contrast, the GNU General Public
116: License is intended to guarantee your freedom to share and change free
117: software---to make sure the software is free for all its users. This
118: General Public License applies to most of the Free Software
119: Foundation's software and to any other program whose authors commit to
120: using it. (Some other Free Software Foundation software is covered by
121: the GNU Library General Public License instead.) You can apply it to
122: your programs, too.
123:
124: When we speak of free software, we are referring to freedom, not
125: price. Our General Public Licenses are designed to make sure that you
126: have the freedom to distribute copies of free software (and charge for
127: this service if you wish), that you receive source code or can get it
128: if you want it, that you can change the software or use pieces of it
129: in new free programs; and that you know you can do these things.
130:
131: To protect your rights, we need to make restrictions that forbid
132: anyone to deny you these rights or to ask you to surrender the rights.
133: These restrictions translate to certain responsibilities for you if you
134: distribute copies of the software, or if you modify it.
135:
136: For example, if you distribute copies of such a program, whether
137: gratis or for a fee, you must give the recipients all the rights that
138: you have. You must make sure that they, too, receive or can get the
139: source code. And you must show them these terms so they know their
140: rights.
141:
142: We protect your rights with two steps: (1) copyright the software, and
143: (2) offer you this license which gives you legal permission to copy,
144: distribute and/or modify the software.
145:
146: Also, for each author's protection and ours, we want to make certain
147: that everyone understands that there is no warranty for this free
148: software. If the software is modified by someone else and passed on, we
149: want its recipients to know that what they have is not the original, so
150: that any problems introduced by others will not reflect on the original
151: authors' reputations.
152:
153: Finally, any free program is threatened constantly by software
154: patents. We wish to avoid the danger that redistributors of a free
155: program will individually obtain patent licenses, in effect making the
156: program proprietary. To prevent this, we have made it clear that any
157: patent must be licensed for everyone's free use or not licensed at all.
158:
159: The precise terms and conditions for copying, distribution and
160: modification follow.
161:
162: @iftex
163: @unnumberedsec TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
164: @end iftex
165: @ifinfo
166: @center TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
167: @end ifinfo
168:
169: @enumerate 0
170: @item
171: This License applies to any program or other work which contains
172: a notice placed by the copyright holder saying it may be distributed
173: under the terms of this General Public License. The ``Program'', below,
174: refers to any such program or work, and a ``work based on the Program''
175: means either the Program or any derivative work under copyright law:
176: that is to say, a work containing the Program or a portion of it,
177: either verbatim or with modifications and/or translated into another
178: language. (Hereinafter, translation is included without limitation in
179: the term ``modification''.) Each licensee is addressed as ``you''.
180:
181: Activities other than copying, distribution and modification are not
182: covered by this License; they are outside its scope. The act of
183: running the Program is not restricted, and the output from the Program
184: is covered only if its contents constitute a work based on the
185: Program (independent of having been made by running the Program).
186: Whether that is true depends on what the Program does.
187:
188: @item
189: You may copy and distribute verbatim copies of the Program's
190: source code as you receive it, in any medium, provided that you
191: conspicuously and appropriately publish on each copy an appropriate
192: copyright notice and disclaimer of warranty; keep intact all the
193: notices that refer to this License and to the absence of any warranty;
194: and give any other recipients of the Program a copy of this License
195: along with the Program.
196:
197: You may charge a fee for the physical act of transferring a copy, and
198: you may at your option offer warranty protection in exchange for a fee.
199:
200: @item
201: You may modify your copy or copies of the Program or any portion
202: of it, thus forming a work based on the Program, and copy and
203: distribute such modifications or work under the terms of Section 1
204: above, provided that you also meet all of these conditions:
205:
206: @enumerate a
207: @item
208: You must cause the modified files to carry prominent notices
209: stating that you changed the files and the date of any change.
210:
211: @item
212: You must cause any work that you distribute or publish, that in
213: whole or in part contains or is derived from the Program or any
214: part thereof, to be licensed as a whole at no charge to all third
215: parties under the terms of this License.
216:
217: @item
218: If the modified program normally reads commands interactively
219: when run, you must cause it, when started running for such
220: interactive use in the most ordinary way, to print or display an
221: announcement including an appropriate copyright notice and a
222: notice that there is no warranty (or else, saying that you provide
223: a warranty) and that users may redistribute the program under
224: these conditions, and telling the user how to view a copy of this
225: License. (Exception: if the Program itself is interactive but
226: does not normally print such an announcement, your work based on
227: the Program is not required to print an announcement.)
228: @end enumerate
229:
230: These requirements apply to the modified work as a whole. If
231: identifiable sections of that work are not derived from the Program,
232: and can be reasonably considered independent and separate works in
233: themselves, then this License, and its terms, do not apply to those
234: sections when you distribute them as separate works. But when you
235: distribute the same sections as part of a whole which is a work based
236: on the Program, the distribution of the whole must be on the terms of
237: this License, whose permissions for other licensees extend to the
238: entire whole, and thus to each and every part regardless of who wrote it.
239:
240: Thus, it is not the intent of this section to claim rights or contest
241: your rights to work written entirely by you; rather, the intent is to
242: exercise the right to control the distribution of derivative or
243: collective works based on the Program.
244:
245: In addition, mere aggregation of another work not based on the Program
246: with the Program (or with a work based on the Program) on a volume of
247: a storage or distribution medium does not bring the other work under
248: the scope of this License.
249:
250: @item
251: You may copy and distribute the Program (or a work based on it,
252: under Section 2) in object code or executable form under the terms of
253: Sections 1 and 2 above provided that you also do one of the following:
254:
255: @enumerate a
256: @item
257: Accompany it with the complete corresponding machine-readable
258: source code, which must be distributed under the terms of Sections
259: 1 and 2 above on a medium customarily used for software interchange; or,
260:
261: @item
262: Accompany it with a written offer, valid for at least three
263: years, to give any third party, for a charge no more than your
264: cost of physically performing source distribution, a complete
265: machine-readable copy of the corresponding source code, to be
266: distributed under the terms of Sections 1 and 2 above on a medium
267: customarily used for software interchange; or,
268:
269: @item
270: Accompany it with the information you received as to the offer
271: to distribute corresponding source code. (This alternative is
272: allowed only for noncommercial distribution and only if you
273: received the program in object code or executable form with such
274: an offer, in accord with Subsection b above.)
275: @end enumerate
276:
277: The source code for a work means the preferred form of the work for
278: making modifications to it. For an executable work, complete source
279: code means all the source code for all modules it contains, plus any
280: associated interface definition files, plus the scripts used to
281: control compilation and installation of the executable. However, as a
282: special exception, the source code distributed need not include
283: anything that is normally distributed (in either source or binary
284: form) with the major components (compiler, kernel, and so on) of the
285: operating system on which the executable runs, unless that component
286: itself accompanies the executable.
287:
288: If distribution of executable or object code is made by offering
289: access to copy from a designated place, then offering equivalent
290: access to copy the source code from the same place counts as
291: distribution of the source code, even though third parties are not
292: compelled to copy the source along with the object code.
293:
294: @item
295: You may not copy, modify, sublicense, or distribute the Program
296: except as expressly provided under this License. Any attempt
297: otherwise to copy, modify, sublicense or distribute the Program is
298: void, and will automatically terminate your rights under this License.
299: However, parties who have received copies, or rights, from you under
300: this License will not have their licenses terminated so long as such
301: parties remain in full compliance.
302:
303: @item
304: You are not required to accept this License, since you have not
305: signed it. However, nothing else grants you permission to modify or
306: distribute the Program or its derivative works. These actions are
307: prohibited by law if you do not accept this License. Therefore, by
308: modifying or distributing the Program (or any work based on the
309: Program), you indicate your acceptance of this License to do so, and
310: all its terms and conditions for copying, distributing or modifying
311: the Program or works based on it.
312:
313: @item
314: Each time you redistribute the Program (or any work based on the
315: Program), the recipient automatically receives a license from the
316: original licensor to copy, distribute or modify the Program subject to
317: these terms and conditions. You may not impose any further
318: restrictions on the recipients' exercise of the rights granted herein.
319: You are not responsible for enforcing compliance by third parties to
320: this License.
321:
322: @item
323: If, as a consequence of a court judgment or allegation of patent
324: infringement or for any other reason (not limited to patent issues),
325: conditions are imposed on you (whether by court order, agreement or
326: otherwise) that contradict the conditions of this License, they do not
327: excuse you from the conditions of this License. If you cannot
328: distribute so as to satisfy simultaneously your obligations under this
329: License and any other pertinent obligations, then as a consequence you
330: may not distribute the Program at all. For example, if a patent
331: license would not permit royalty-free redistribution of the Program by
332: all those who receive copies directly or indirectly through you, then
333: the only way you could satisfy both it and this License would be to
334: refrain entirely from distribution of the Program.
335:
336: If any portion of this section is held invalid or unenforceable under
337: any particular circumstance, the balance of the section is intended to
338: apply and the section as a whole is intended to apply in other
339: circumstances.
340:
341: It is not the purpose of this section to induce you to infringe any
342: patents or other property right claims or to contest validity of any
343: such claims; this section has the sole purpose of protecting the
344: integrity of the free software distribution system, which is
345: implemented by public license practices. Many people have made
346: generous contributions to the wide range of software distributed
347: through that system in reliance on consistent application of that
348: system; it is up to the author/donor to decide if he or she is willing
349: to distribute software through any other system and a licensee cannot
350: impose that choice.
351:
352: This section is intended to make thoroughly clear what is believed to
353: be a consequence of the rest of this License.
354:
355: @item
356: If the distribution and/or use of the Program is restricted in
357: certain countries either by patents or by copyrighted interfaces, the
358: original copyright holder who places the Program under this License
359: may add an explicit geographical distribution limitation excluding
360: those countries, so that distribution is permitted only in or among
361: countries not thus excluded. In such case, this License incorporates
362: the limitation as if written in the body of this License.
363:
364: @item
365: The Free Software Foundation may publish revised and/or new versions
366: of the General Public License from time to time. Such new versions will
367: be similar in spirit to the present version, but may differ in detail to
368: address new problems or concerns.
369:
370: Each version is given a distinguishing version number. If the Program
371: specifies a version number of this License which applies to it and ``any
372: later version'', you have the option of following the terms and conditions
373: either of that version or of any later version published by the Free
374: Software Foundation. If the Program does not specify a version number of
375: this License, you may choose any version ever published by the Free Software
376: Foundation.
377:
378: @item
379: If you wish to incorporate parts of the Program into other free
380: programs whose distribution conditions are different, write to the author
381: to ask for permission. For software which is copyrighted by the Free
382: Software Foundation, write to the Free Software Foundation; we sometimes
383: make exceptions for this. Our decision will be guided by the two goals
384: of preserving the free status of all derivatives of our free software and
385: of promoting the sharing and reuse of software generally.
386:
387: @iftex
388: @heading NO WARRANTY
389: @end iftex
390: @ifinfo
391: @center NO WARRANTY
392: @end ifinfo
393:
394: @item
395: BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
396: FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN
397: OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
398: PROVIDE THE PROGRAM ``AS IS'' WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED
399: OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
400: MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS
401: TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE
402: PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING,
403: REPAIR OR CORRECTION.
404:
405: @item
406: IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
407: WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
408: REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
409: INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
410: OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED
411: TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY
412: YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER
413: PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
414: POSSIBILITY OF SUCH DAMAGES.
415: @end enumerate
416:
417: @iftex
418: @heading END OF TERMS AND CONDITIONS
419: @end iftex
420: @ifinfo
421: @center END OF TERMS AND CONDITIONS
422: @end ifinfo
423:
424: @page
425: @unnumberedsec How to Apply These Terms to Your New Programs
426:
427: If you develop a new program, and you want it to be of the greatest
428: possible use to the public, the best way to achieve this is to make it
429: free software which everyone can redistribute and change under these terms.
430:
431: To do so, attach the following notices to the program. It is safest
432: to attach them to the start of each source file to most effectively
433: convey the exclusion of warranty; and each file should have at least
434: the ``copyright'' line and a pointer to where the full notice is found.
435:
436: @smallexample
437: @var{one line to give the program's name and a brief idea of what it does.}
438: Copyright (C) 19@var{yy} @var{name of author}
439:
440: This program is free software; you can redistribute it and/or modify
441: it under the terms of the GNU General Public License as published by
442: the Free Software Foundation; either version 2 of the License, or
443: (at your option) any later version.
444:
445: This program is distributed in the hope that it will be useful,
446: but WITHOUT ANY WARRANTY; without even the implied warranty of
447: MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
448: GNU General Public License for more details.
449:
450: You should have received a copy of the GNU General Public License
451: along with this program; if not, write to the Free Software
452: Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
453: @end smallexample
454:
455: Also add information on how to contact you by electronic and paper mail.
456:
457: If the program is interactive, make it output a short notice like this
458: when it starts in an interactive mode:
459:
460: @smallexample
461: Gnomovision version 69, Copyright (C) 19@var{yy} @var{name of author}
462: Gnomovision comes with ABSOLUTELY NO WARRANTY; for details
463: type `show w'.
464: This is free software, and you are welcome to redistribute it
465: under certain conditions; type `show c' for details.
466: @end smallexample
467:
468: The hypothetical commands @samp{show w} and @samp{show c} should show
469: the appropriate parts of the General Public License. Of course, the
470: commands you use may be called something other than @samp{show w} and
471: @samp{show c}; they could even be mouse-clicks or menu items---whatever
472: suits your program.
473:
474: You should also get your employer (if you work as a programmer) or your
475: school, if any, to sign a ``copyright disclaimer'' for the program, if
476: necessary. Here is a sample; alter the names:
477:
478: @smallexample
479: Yoyodyne, Inc., hereby disclaims all copyright interest in the program
480: `Gnomovision' (which makes passes at compilers) written by James Hacker.
481:
482: @var{signature of Ty Coon}, 1 April 1989
483: Ty Coon, President of Vice
484: @end smallexample
485:
486: This General Public License does not permit incorporating your program into
487: proprietary programs. If your program is a subroutine library, you may
488: consider it more useful to permit linking proprietary applications with the
489: library. If this is what you want to do, use the GNU Library General
490: Public License instead of this License.
491:
492: @iftex
493: @node Preface
494: @comment node-name, next, previous, up
495: @unnumbered Preface
496: @cindex Preface
497: This manual documents Gforth. The reader is expected to know
498: Forth. This manual is primarily a reference manual. @xref{Other Books}
499: for introductory material.
500: @end iftex
501:
502: @node Goals, Other Books, License, Top
503: @comment node-name, next, previous, up
504: @chapter Goals of Gforth
505: @cindex Goals
506: The goal of the Gforth Project is to develop a standard model for
507: ANSI Forth. This can be split into several subgoals:
508:
509: @itemize @bullet
510: @item
511: Gforth should conform to the ANSI Forth standard.
512: @item
513: It should be a model, i.e. it should define all the
514: implementation-dependent things.
515: @item
516: It should become standard, i.e. widely accepted and used. This goal
517: is the most difficult one.
518: @end itemize
519:
520: To achieve these goals Gforth should be
521: @itemize @bullet
522: @item
523: Similar to previous models (fig-Forth, F83)
524: @item
525: Powerful. It should provide for all the things that are considered
526: necessary today and even some that are not yet considered necessary.
527: @item
528: Efficient. It should not get the reputation of being exceptionally
529: slow.
530: @item
531: Free.
532: @item
533: Available on many machines/easy to port.
534: @end itemize
535:
536: Have we achieved these goals? Gforth conforms to the ANS Forth
537: standard. It may be considered a model, but we have not yet documented
538: which parts of the model are stable and which parts we are likely to
539: change. It certainly has not yet become a de facto standard. It has some
540: similarities and some differences to previous models. It has some
541: powerful features, but not yet everything that we envisioned. We
542: certainly have achieved our execution speed goals (@pxref{Performance}).
543: It is free and available on many machines.
544:
545: @node Other Books, Invocation, Goals, Top
546: @chapter Other books on ANS Forth
547:
548: As the standard is relatively new, there are not many books out yet. It
549: is not recommended to learn Forth by using Gforth and a book that is
550: not written for ANS Forth, as you will not know your mistakes from the
551: deviations of the book.
552:
553: There is, of course, the standard, the definite reference if you want to
554: write ANS Forth programs. It is available in printed form from the
555: National Standards Institute Sales Department (Tel.: USA (212) 642-4900;
556: Fax.: USA (212) 302-1286) as document @cite{X3.215-1994} for about $200. You
557: can also get it from Global Engineering Documents (Tel.: USA (800)
558: 854-7179; Fax.: (303) 843-9880) for about $300.
559:
560: @cite{dpANS6}, the last draft of the standard, which was then submitted to ANSI
561: for publication is available electronically and for free in some MS Word
562: format, and it has been converted to HTML. Some pointers to these
563: versions can be found through
564: @*@file{http://www.complang.tuwien.ac.at/projects/forth.html}.
565:
566: @cite{Forth: The new model} by Jack Woehr (Prentice-Hall, 1993) is an
567: introductory book based on a draft version of the standard. It does not
568: cover the whole standard. It also contains interesting background
569: information (Jack Woehr was in the ANS Forth Technical Committe). It is
570: not appropriate for complete newbies, but programmers experienced in
571: other languages should find it ok.
572:
573: @node Invocation, Words, Other Books, Top
574: @chapter Invocation
575:
576: You will usually just say @code{gforth}. In many other cases the default
577: Gforth image will be invoked like this:
578:
579: @example
580: gforth [files] [-e forth-code]
581: @end example
582:
583: executing the contents of the files and the Forth code in the order they
584: are given.
585:
586: In general, the command line looks like this:
587:
588: @example
589: gforth [initialization options] [image-specific options]
590: @end example
591:
592: The initialization options must come before the rest of the command
593: line. They are:
594:
595: @table @code
596: @item --image-file @var{file}
597: @item -i @var{file}
598: Loads the Forth image @var{file} instead of the default
599: @file{gforth.fi}.
600:
601: @item --path @var{path}
602: @item -p @var{path}
603: Uses @var{path} for searching the image file and Forth source code
604: files instead of the default in the environment variable
605: @code{GFORTHPATH} or the path specified at installation time (typically
606: @file{/usr/local/lib/gforth:.}). A path is given as a @code{:}-separated
607: list.
608:
609: @item --dictionary-size @var{size}
610: @item -m @var{size}
611: Allocate @var{size} space for the Forth dictionary space instead of
612: using the default specified in the image (typically 256K). The
613: @var{size} specification consists of an integer and a unit (e.g.,
614: @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
615: size, in this case Cells), @code{k} (kilobytes), and @code{M}
616: (Megabytes). If no unit is specified, @code{e} is used.
617:
618: @item --data-stack-size @var{size}
619: @item -d @var{size}
620: Allocate @var{size} space for the data stack instead of using the
621: default specified in the image (typically 16K).
622:
623: @item --return-stack-size @var{size}
624: @item -r @var{size}
625: Allocate @var{size} space for the return stack instead of using the
626: default specified in the image (typically 16K).
627:
628: @item --fp-stack-size @var{size}
629: @item -f @var{size}
630: Allocate @var{size} space for the floating point stack instead of
631: using the default specified in the image (typically 16K). In this case
632: the unit specifier @code{e} refers to floating point numbers.
633:
634: @item --locals-stack-size @var{size}
635: @item -l @var{size}
636: Allocate @var{size} space for the locals stack instead of using the
637: default specified in the image (typically 16K).
638:
639: @end table
640:
641: As explained above, the image-specific command-line arguments for the
642: default image @file{gforth.fi} consist of a sequence of filenames and
643: @code{-e @var{forth-code}} options that are interpreted in the seqence
644: in which they are given. The @code{-e @var{forth-code}} or
645: @code{--evaluate @var{forth-code}} option evaluates the forth
646: code. This option takes only one argument; if you want to evaluate more
647: Forth words, you have to quote them or use several @code{-e}s. To exit
648: after processing the command line (instead of entering interactive mode)
649: append @code{-e bye} to the command line.
650:
651: If you have several versions of Gforth installed, @code{gforth} will
652: invoke the version that was installed last. @code{gforth-@var{version}}
653: invokes a specific version. You may want to use the option
654: @code{--path}, if your environment contains the variable
655: @code{GFORTHPATH}.
656:
657: Not yet implemented:
658: On startup the system first executes the system initialization file
659: (unless the option @code{--no-init-file} is given; note that the system
660: resulting from using this option may not be ANS Forth conformant). Then
661: the user initialization file @file{.gforth.fs} is executed, unless the
662: option @code{--no-rc} is given; this file is first searched in @file{.},
663: then in @file{~}, then in the normal path (see above).
664:
665: @node Words, ANS conformance, Invocation, Top
666: @chapter Forth Words
667:
668: @menu
669: * Notation::
670: * Arithmetic::
671: * Stack Manipulation::
672: * Memory access::
673: * Control Structures::
674: * Locals::
675: * Defining Words::
676: * Wordlists::
677: * Files::
678: * Blocks::
679: * Other I/O::
680: * Programming Tools::
681: * Assembler and Code words::
682: * Threading Words::
683: @end menu
684:
685: @node Notation, Arithmetic, Words, Words
686: @section Notation
687:
688: The Forth words are described in this section in the glossary notation
689: that has become a de-facto standard for Forth texts, i.e.
690:
691: @format
692: @var{word} @var{Stack effect} @var{wordset} @var{pronunciation}
693: @end format
694: @var{Description}
695:
696: @table @var
697: @item word
698: The name of the word. BTW, Gforth is case insensitive, so you can
699: type the words in in lower case (However, @pxref{core-idef}).
700:
701: @item Stack effect
702: The stack effect is written in the notation @code{@var{before} --
703: @var{after}}, where @var{before} and @var{after} describe the top of
704: stack entries before and after the execution of the word. The rest of
705: the stack is not touched by the word. The top of stack is rightmost,
706: i.e., a stack sequence is written as it is typed in. Note that Gforth
707: uses a separate floating point stack, but a unified stack
708: notation. Also, return stack effects are not shown in @var{stack
709: effect}, but in @var{Description}. The name of a stack item describes
710: the type and/or the function of the item. See below for a discussion of
711: the types.
712:
713: All words have two stack effects: A compile-time stack effect and a
714: run-time stack effect. The compile-time stack-effect of most words is
715: @var{ -- }. If the compile-time stack-effect of a word deviates from
716: this standard behaviour, or the word does other unusual things at
717: compile time, both stack effects are shown; otherwise only the run-time
718: stack effect is shown.
719:
720: @item pronunciation
721: How the word is pronounced
722:
723: @item wordset
724: The ANS Forth standard is divided into several wordsets. A standard
725: system need not support all of them. So, the fewer wordsets your program
726: uses the more portable it will be in theory. However, we suspect that
727: most ANS Forth systems on personal machines will feature all
728: wordsets. Words that are not defined in the ANS standard have
729: @code{gforth} or @code{gforth-internal} as wordset. @code{gforth}
730: describes words that will work in future releases of Gforth;
731: @code{gforth-internal} words are more volatile. Environmental query
732: strings are also displayed like words; you can recognize them by the
733: @code{environment} in the wordset field.
734:
735: @item Description
736: A description of the behaviour of the word.
737: @end table
738:
739: The type of a stack item is specified by the character(s) the name
740: starts with:
741:
742: @table @code
743: @item f
744: Bool, i.e. @code{false} or @code{true}.
745: @item c
746: Char
747: @item w
748: Cell, can contain an integer or an address
749: @item n
750: signed integer
751: @item u
752: unsigned integer
753: @item d
754: double sized signed integer
755: @item ud
756: double sized unsigned integer
757: @item r
758: Float (on the FP stack)
759: @item a_
760: Cell-aligned address
761: @item c_
762: Char-aligned address (note that a Char may have two bytes in Windows NT)
763: @item f_
764: Float-aligned address
765: @item df_
766: Address aligned for IEEE double precision float
767: @item sf_
768: Address aligned for IEEE single precision float
769: @item xt
770: Execution token, same size as Cell
771: @item wid
772: Wordlist ID, same size as Cell
773: @item f83name
774: Pointer to a name structure
775: @item "
776: string in the input stream (not the stack). The terminating character is
777: a blank by default. If it is not a blank, it is shown in @code{<>}
778: quotes.
779:
780: @end table
781:
782: @node Arithmetic, Stack Manipulation, Notation, Words
783: @section Arithmetic
784: Forth arithmetic is not checked, i.e., you will not hear about integer
785: overflow on addition or multiplication, you may hear about division by
786: zero if you are lucky. The operator is written after the operands, but
787: the operands are still in the original order. I.e., the infix @code{2-1}
788: corresponds to @code{2 1 -}. Forth offers a variety of division
789: operators. If you perform division with potentially negative operands,
790: you do not want to use @code{/} or @code{/mod} with its undefined
791: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
792: former, @pxref{Mixed precision}).
793:
794: @menu
795: * Single precision::
796: * Bitwise operations::
797: * Mixed precision:: operations with single and double-cell integers
798: * Double precision:: Double-cell integer arithmetic
799: * Floating Point::
800: @end menu
801:
802: @node Single precision, Bitwise operations, Arithmetic, Arithmetic
803: @subsection Single precision
804: doc-+
805: doc--
806: doc-*
807: doc-/
808: doc-mod
809: doc-/mod
810: doc-negate
811: doc-abs
812: doc-min
813: doc-max
814:
815: @node Bitwise operations, Mixed precision, Single precision, Arithmetic
816: @subsection Bitwise operations
817: doc-and
818: doc-or
819: doc-xor
820: doc-invert
821: doc-2*
822: doc-2/
823:
824: @node Mixed precision, Double precision, Bitwise operations, Arithmetic
825: @subsection Mixed precision
826: doc-m+
827: doc-*/
828: doc-*/mod
829: doc-m*
830: doc-um*
831: doc-m*/
832: doc-um/mod
833: doc-fm/mod
834: doc-sm/rem
835:
836: @node Double precision, Floating Point, Mixed precision, Arithmetic
837: @subsection Double precision
838:
839: The outer (aka text) interpreter converts numbers containing a dot into
840: a double precision number. Note that only numbers with the dot as last
841: character are standard-conforming.
842:
843: doc-d+
844: doc-d-
845: doc-dnegate
846: doc-dabs
847: doc-dmin
848: doc-dmax
849:
850: @node Floating Point, , Double precision, Arithmetic
851: @subsection Floating Point
852:
853: The format of floating point numbers recognized by the outer (aka text)
854: interpreter is: a signed decimal number, possibly containing a decimal
855: point (@code{.}), followed by @code{E} or @code{e}, optionally followed
856: by a signed integer (the exponent). E.g., @code{1e} ist the same as
857: @code{+1.0e+0}. Note that a number without @code{e}
858: is not interpreted as floating-point number, but as double (if the
859: number contains a @code{.}) or single precision integer. Also,
860: conversions between string and floating point numbers always use base
861: 10, irrespective of the value of @code{BASE}. If @code{BASE} contains a
862: value greater then 14, the @code{E} may be interpreted as digit and the
863: number will be interpreted as integer, unless it has a signed exponent
864: (both @code{+} and @code{-} are allowed as signs).
865:
866: Angles in floating point operations are given in radians (a full circle
867: has 2 pi radians). Note, that Gforth has a separate floating point
868: stack, but we use the unified notation.
869:
870: Floating point numbers have a number of unpleasant surprises for the
871: unwary (e.g., floating point addition is not associative) and even a few
872: for the wary. You should not use them unless you know what you are doing
873: or you don't care that the results you get are totally bogus. If you
874: want to learn about the problems of floating point numbers (and how to
875: avoid them), you might start with @cite{David Goldberg, What Every
876: Computer Scientist Should Know About Floating-Point Arithmetic, ACM
877: Computing Surveys 23(1):5@minus{}48, March 1991}.
878:
879: doc-f+
880: doc-f-
881: doc-f*
882: doc-f/
883: doc-fnegate
884: doc-fabs
885: doc-fmax
886: doc-fmin
887: doc-floor
888: doc-fround
889: doc-f**
890: doc-fsqrt
891: doc-fexp
892: doc-fexpm1
893: doc-fln
894: doc-flnp1
895: doc-flog
896: doc-falog
897: doc-fsin
898: doc-fcos
899: doc-fsincos
900: doc-ftan
901: doc-fasin
902: doc-facos
903: doc-fatan
904: doc-fatan2
905: doc-fsinh
906: doc-fcosh
907: doc-ftanh
908: doc-fasinh
909: doc-facosh
910: doc-fatanh
911:
912: @node Stack Manipulation, Memory access, Arithmetic, Words
913: @section Stack Manipulation
914:
915: Gforth has a data stack (aka parameter stack) for characters, cells,
916: addresses, and double cells, a floating point stack for floating point
917: numbers, a return stack for storing the return addresses of colon
918: definitions and other data, and a locals stack for storing local
919: variables. Note that while every sane Forth has a separate floating
920: point stack, this is not strictly required; an ANS Forth system could
921: theoretically keep floating point numbers on the data stack. As an
922: additional difficulty, you don't know how many cells a floating point
923: number takes. It is reportedly possible to write words in a way that
924: they work also for a unified stack model, but we do not recommend trying
925: it. Instead, just say that your program has an environmental dependency
926: on a separate FP stack.
927:
928: Also, a Forth system is allowed to keep the local variables on the
929: return stack. This is reasonable, as local variables usually eliminate
930: the need to use the return stack explicitly. So, if you want to produce
931: a standard complying program and if you are using local variables in a
932: word, forget about return stack manipulations in that word (see the
933: standard document for the exact rules).
934:
935: @menu
936: * Data stack::
937: * Floating point stack::
938: * Return stack::
939: * Locals stack::
940: * Stack pointer manipulation::
941: @end menu
942:
943: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
944: @subsection Data stack
945: doc-drop
946: doc-nip
947: doc-dup
948: doc-over
949: doc-tuck
950: doc-swap
951: doc-rot
952: doc--rot
953: doc-?dup
954: doc-pick
955: doc-roll
956: doc-2drop
957: doc-2nip
958: doc-2dup
959: doc-2over
960: doc-2tuck
961: doc-2swap
962: doc-2rot
963:
964: @node Floating point stack, Return stack, Data stack, Stack Manipulation
965: @subsection Floating point stack
966: doc-fdrop
967: doc-fnip
968: doc-fdup
969: doc-fover
970: doc-ftuck
971: doc-fswap
972: doc-frot
973:
974: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
975: @subsection Return stack
976: doc->r
977: doc-r>
978: doc-r@
979: doc-rdrop
980: doc-2>r
981: doc-2r>
982: doc-2r@
983: doc-2rdrop
984:
985: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
986: @subsection Locals stack
987:
988: @node Stack pointer manipulation, , Locals stack, Stack Manipulation
989: @subsection Stack pointer manipulation
990: doc-sp@
991: doc-sp!
992: doc-fp@
993: doc-fp!
994: doc-rp@
995: doc-rp!
996: doc-lp@
997: doc-lp!
998:
999: @node Memory access, Control Structures, Stack Manipulation, Words
1000: @section Memory access
1001:
1002: @menu
1003: * Stack-Memory transfers::
1004: * Address arithmetic::
1005: * Memory block access::
1006: @end menu
1007:
1008: @node Stack-Memory transfers, Address arithmetic, Memory access, Memory access
1009: @subsection Stack-Memory transfers
1010:
1011: doc-@
1012: doc-!
1013: doc-+!
1014: doc-c@
1015: doc-c!
1016: doc-2@
1017: doc-2!
1018: doc-f@
1019: doc-f!
1020: doc-sf@
1021: doc-sf!
1022: doc-df@
1023: doc-df!
1024:
1025: @node Address arithmetic, Memory block access, Stack-Memory transfers, Memory access
1026: @subsection Address arithmetic
1027:
1028: ANS Forth does not specify the sizes of the data types. Instead, it
1029: offers a number of words for computing sizes and doing address
1030: arithmetic. Basically, address arithmetic is performed in terms of
1031: address units (aus); on most systems the address unit is one byte. Note
1032: that a character may have more than one au, so @code{chars} is no noop
1033: (on systems where it is a noop, it compiles to nothing).
1034:
1035: ANS Forth also defines words for aligning addresses for specific
1036: addresses. Many computers require that accesses to specific data types
1037: must only occur at specific addresses; e.g., that cells may only be
1038: accessed at addresses divisible by 4. Even if a machine allows unaligned
1039: accesses, it can usually perform aligned accesses faster.
1040:
1041: For the performance-conscious: alignment operations are usually only
1042: necessary during the definition of a data structure, not during the
1043: (more frequent) accesses to it.
1044:
1045: ANS Forth defines no words for character-aligning addresses. This is not
1046: an oversight, but reflects the fact that addresses that are not
1047: char-aligned have no use in the standard and therefore will not be
1048: created.
1049:
1050: The standard guarantees that addresses returned by @code{CREATE}d words
1051: are cell-aligned; in addition, Gforth guarantees that these addresses
1052: are aligned for all purposes.
1053:
1054: Note that the standard defines a word @code{char}, which has nothing to
1055: do with address arithmetic.
1056:
1057: doc-chars
1058: doc-char+
1059: doc-cells
1060: doc-cell+
1061: doc-align
1062: doc-aligned
1063: doc-floats
1064: doc-float+
1065: doc-falign
1066: doc-faligned
1067: doc-sfloats
1068: doc-sfloat+
1069: doc-sfalign
1070: doc-sfaligned
1071: doc-dfloats
1072: doc-dfloat+
1073: doc-dfalign
1074: doc-dfaligned
1075: doc-maxalign
1076: doc-maxaligned
1077: doc-cfalign
1078: doc-cfaligned
1079: doc-address-unit-bits
1080:
1081: @node Memory block access, , Address arithmetic, Memory access
1082: @subsection Memory block access
1083:
1084: doc-move
1085: doc-erase
1086:
1087: While the previous words work on address units, the rest works on
1088: characters.
1089:
1090: doc-cmove
1091: doc-cmove>
1092: doc-fill
1093: doc-blank
1094:
1095: @node Control Structures, Locals, Memory access, Words
1096: @section Control Structures
1097:
1098: Control structures in Forth cannot be used in interpret state, only in
1099: compile state, i.e., in a colon definition. We do not like this
1100: limitation, but have not seen a satisfying way around it yet, although
1101: many schemes have been proposed.
1102:
1103: @menu
1104: * Selection::
1105: * Simple Loops::
1106: * Counted Loops::
1107: * Arbitrary control structures::
1108: * Calls and returns::
1109: * Exception Handling::
1110: @end menu
1111:
1112: @node Selection, Simple Loops, Control Structures, Control Structures
1113: @subsection Selection
1114:
1115: @example
1116: @var{flag}
1117: IF
1118: @var{code}
1119: ENDIF
1120: @end example
1121: or
1122: @example
1123: @var{flag}
1124: IF
1125: @var{code1}
1126: ELSE
1127: @var{code2}
1128: ENDIF
1129: @end example
1130:
1131: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
1132: standard, and @code{ENDIF} is not, although it is quite popular. We
1133: recommend using @code{ENDIF}, because it is less confusing for people
1134: who also know other languages (and is not prone to reinforcing negative
1135: prejudices against Forth in these people). Adding @code{ENDIF} to a
1136: system that only supplies @code{THEN} is simple:
1137: @example
1138: : endif POSTPONE then ; immediate
1139: @end example
1140:
1141: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
1142: (adv.)} has the following meanings:
1143: @quotation
1144: ... 2b: following next after in order ... 3d: as a necessary consequence
1145: (if you were there, then you saw them).
1146: @end quotation
1147: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
1148: and many other programming languages has the meaning 3d.]
1149:
1150: Gforth also provides the words @code{?dup-if} and @code{?dup-0=-if}, so
1151: you can avoid using @code{?dup}. Using these alternatives is also more
1152: efficient than using @code{?dup}. Definitions in plain standard Forth
1153: for @code{ENDIF}, @code{?DUP-IF} and @code{?DUP-0=-IF} are provided in
1154: @file{compat/control.fs}.
1155:
1156: @example
1157: @var{n}
1158: CASE
1159: @var{n1} OF @var{code1} ENDOF
1160: @var{n2} OF @var{code2} ENDOF
1161: @dots{}
1162: ENDCASE
1163: @end example
1164:
1165: Executes the first @var{codei}, where the @var{ni} is equal to
1166: @var{n}. A default case can be added by simply writing the code after
1167: the last @code{ENDOF}. It may use @var{n}, which is on top of the stack,
1168: but must not consume it.
1169:
1170: @node Simple Loops, Counted Loops, Selection, Control Structures
1171: @subsection Simple Loops
1172:
1173: @example
1174: BEGIN
1175: @var{code1}
1176: @var{flag}
1177: WHILE
1178: @var{code2}
1179: REPEAT
1180: @end example
1181:
1182: @var{code1} is executed and @var{flag} is computed. If it is true,
1183: @var{code2} is executed and the loop is restarted; If @var{flag} is false, execution continues after the @code{REPEAT}.
1184:
1185: @example
1186: BEGIN
1187: @var{code}
1188: @var{flag}
1189: UNTIL
1190: @end example
1191:
1192: @var{code} is executed. The loop is restarted if @code{flag} is false.
1193:
1194: @example
1195: BEGIN
1196: @var{code}
1197: AGAIN
1198: @end example
1199:
1200: This is an endless loop.
1201:
1202: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
1203: @subsection Counted Loops
1204:
1205: The basic counted loop is:
1206: @example
1207: @var{limit} @var{start}
1208: ?DO
1209: @var{body}
1210: LOOP
1211: @end example
1212:
1213: This performs one iteration for every integer, starting from @var{start}
1214: and up to, but excluding @var{limit}. The counter, aka index, can be
1215: accessed with @code{i}. E.g., the loop
1216: @example
1217: 10 0 ?DO
1218: i .
1219: LOOP
1220: @end example
1221: prints
1222: @example
1223: 0 1 2 3 4 5 6 7 8 9
1224: @end example
1225: The index of the innermost loop can be accessed with @code{i}, the index
1226: of the next loop with @code{j}, and the index of the third loop with
1227: @code{k}.
1228:
1229: The loop control data are kept on the return stack, so there are some
1230: restrictions on mixing return stack accesses and counted loop
1231: words. E.g., if you put values on the return stack outside the loop, you
1232: cannot read them inside the loop. If you put values on the return stack
1233: within a loop, you have to remove them before the end of the loop and
1234: before accessing the index of the loop.
1235:
1236: There are several variations on the counted loop:
1237:
1238: @code{LEAVE} leaves the innermost counted loop immediately.
1239:
1240: If @var{start} is greater than @var{limit}, a @code{?DO} loop is entered
1241: (and @code{LOOP} iterates until they become equal by wrap-around
1242: arithmetic). This behaviour is usually not what you want. Therefore,
1243: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
1244: @code{?DO}), which do not enter the loop if @var{start} is greater than
1245: @var{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
1246: unsigned loop parameters.
1247:
1248: @code{LOOP} can be replaced with @code{@var{n} +LOOP}; this updates the
1249: index by @var{n} instead of by 1. The loop is terminated when the border
1250: between @var{limit-1} and @var{limit} is crossed. E.g.:
1251:
1252: @code{4 0 +DO i . 2 +LOOP} prints @code{0 2}
1253:
1254: @code{4 1 +DO i . 2 +LOOP} prints @code{1 3}
1255:
1256: The behaviour of @code{@var{n} +LOOP} is peculiar when @var{n} is negative:
1257:
1258: @code{-1 0 ?DO i . -1 +LOOP} prints @code{0 -1}
1259:
1260: @code{ 0 0 ?DO i . -1 +LOOP} prints nothing
1261:
1262: Therefore we recommend avoiding @code{@var{n} +LOOP} with negative
1263: @var{n}. One alternative is @code{@var{u} -LOOP}, which reduces the
1264: index by @var{u} each iteration. The loop is terminated when the border
1265: between @var{limit+1} and @var{limit} is crossed. Gforth also provides
1266: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
1267:
1268: @code{-2 0 -DO i . 1 -LOOP} prints @code{0 -1}
1269:
1270: @code{-1 0 -DO i . 1 -LOOP} prints @code{0}
1271:
1272: @code{ 0 0 -DO i . 1 -LOOP} prints nothing
1273:
1274: Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
1275: @code{-LOOP} are not in the ANS Forth standard. However, an
1276: implementation for these words that uses only standard words is provided
1277: in @file{compat/loops.fs}.
1278:
1279: @code{?DO} can also be replaced by @code{DO}. @code{DO} always enters
1280: the loop, independent of the loop parameters. Do not use @code{DO}, even
1281: if you know that the loop is entered in any case. Such knowledge tends
1282: to become invalid during maintenance of a program, and then the
1283: @code{DO} will make trouble.
1284:
1285: @code{UNLOOP} is used to prepare for an abnormal loop exit, e.g., via
1286: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
1287: return stack so @code{EXIT} can get to its return address.
1288:
1289: Another counted loop is
1290: @example
1291: @var{n}
1292: FOR
1293: @var{body}
1294: NEXT
1295: @end example
1296: This is the preferred loop of native code compiler writers who are too
1297: lazy to optimize @code{?DO} loops properly. In Gforth, this loop
1298: iterates @var{n+1} times; @code{i} produces values starting with @var{n}
1299: and ending with 0. Other Forth systems may behave differently, even if
1300: they support @code{FOR} loops. To avoid problems, don't use @code{FOR}
1301: loops.
1302:
1303: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
1304: @subsection Arbitrary control structures
1305:
1306: ANS Forth permits and supports using control structures in a non-nested
1307: way. Information about incomplete control structures is stored on the
1308: control-flow stack. This stack may be implemented on the Forth data
1309: stack, and this is what we have done in Gforth.
1310:
1311: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
1312: entry represents a backward branch target. A few words are the basis for
1313: building any control structure possible (except control structures that
1314: need storage, like calls, coroutines, and backtracking).
1315:
1316: doc-if
1317: doc-ahead
1318: doc-then
1319: doc-begin
1320: doc-until
1321: doc-again
1322: doc-cs-pick
1323: doc-cs-roll
1324:
1325: On many systems control-flow stack items take one word, in Gforth they
1326: currently take three (this may change in the future). Therefore it is a
1327: really good idea to manipulate the control flow stack with
1328: @code{cs-pick} and @code{cs-roll}, not with data stack manipulation
1329: words.
1330:
1331: Some standard control structure words are built from these words:
1332:
1333: doc-else
1334: doc-while
1335: doc-repeat
1336:
1337: Gforth adds some more control-structure words:
1338:
1339: doc-endif
1340: doc-?dup-if
1341: doc-?dup-0=-if
1342:
1343: Counted loop words constitute a separate group of words:
1344:
1345: doc-?do
1346: doc-+do
1347: doc-u+do
1348: doc--do
1349: doc-u-do
1350: doc-do
1351: doc-for
1352: doc-loop
1353: doc-+loop
1354: doc--loop
1355: doc-next
1356: doc-leave
1357: doc-?leave
1358: doc-unloop
1359: doc-done
1360:
1361: The standard does not allow using @code{cs-pick} and @code{cs-roll} on
1362: @i{do-sys}. Our system allows it, but it's your job to ensure that for
1363: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
1364: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
1365: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
1366: resolved (by using one of the loop-ending words or @code{DONE}).
1367:
1368: Another group of control structure words are
1369:
1370: doc-case
1371: doc-endcase
1372: doc-of
1373: doc-endof
1374:
1375: @i{case-sys} and @i{of-sys} cannot be processed using @code{cs-pick} and
1376: @code{cs-roll}.
1377:
1378: @subsubsection Programming Style
1379:
1380: In order to ensure readability we recommend that you do not create
1381: arbitrary control structures directly, but define new control structure
1382: words for the control structure you want and use these words in your
1383: program.
1384:
1385: E.g., instead of writing
1386:
1387: @example
1388: begin
1389: ...
1390: if [ 1 cs-roll ]
1391: ...
1392: again then
1393: @end example
1394:
1395: we recommend defining control structure words, e.g.,
1396:
1397: @example
1398: : while ( dest -- orig dest )
1399: POSTPONE if
1400: 1 cs-roll ; immediate
1401:
1402: : repeat ( orig dest -- )
1403: POSTPONE again
1404: POSTPONE then ; immediate
1405: @end example
1406:
1407: and then using these to create the control structure:
1408:
1409: @example
1410: begin
1411: ...
1412: while
1413: ...
1414: repeat
1415: @end example
1416:
1417: That's much easier to read, isn't it? Of course, @code{REPEAT} and
1418: @code{WHILE} are predefined, so in this example it would not be
1419: necessary to define them.
1420:
1421: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
1422: @subsection Calls and returns
1423:
1424: A definition can be called simply be writing the name of the
1425: definition. When the end of the definition is reached, it returns. An
1426: earlier return can be forced using
1427:
1428: doc-exit
1429:
1430: Don't forget to clean up the return stack and @code{UNLOOP} any
1431: outstanding @code{?DO}...@code{LOOP}s before @code{EXIT}ing. The
1432: primitive compiled by @code{EXIT} is
1433:
1434: doc-;s
1435:
1436: @node Exception Handling, , Calls and returns, Control Structures
1437: @subsection Exception Handling
1438:
1439: doc-catch
1440: doc-throw
1441:
1442: @node Locals, Defining Words, Control Structures, Words
1443: @section Locals
1444:
1445: Local variables can make Forth programming more enjoyable and Forth
1446: programs easier to read. Unfortunately, the locals of ANS Forth are
1447: laden with restrictions. Therefore, we provide not only the ANS Forth
1448: locals wordset, but also our own, more powerful locals wordset (we
1449: implemented the ANS Forth locals wordset through our locals wordset).
1450:
1451: The ideas in this section have also been published in the paper
1452: @cite{Automatic Scoping of Local Variables} by M. Anton Ertl, presented
1453: at EuroForth '94; it is available at
1454: @*@file{http://www.complang.tuwien.ac.at/papers/ertl94l.ps.gz}.
1455:
1456: @menu
1457: * Gforth locals::
1458: * ANS Forth locals::
1459: @end menu
1460:
1461: @node Gforth locals, ANS Forth locals, Locals, Locals
1462: @subsection Gforth locals
1463:
1464: Locals can be defined with
1465:
1466: @example
1467: @{ local1 local2 ... -- comment @}
1468: @end example
1469: or
1470: @example
1471: @{ local1 local2 ... @}
1472: @end example
1473:
1474: E.g.,
1475: @example
1476: : max @{ n1 n2 -- n3 @}
1477: n1 n2 > if
1478: n1
1479: else
1480: n2
1481: endif ;
1482: @end example
1483:
1484: The similarity of locals definitions with stack comments is intended. A
1485: locals definition often replaces the stack comment of a word. The order
1486: of the locals corresponds to the order in a stack comment and everything
1487: after the @code{--} is really a comment.
1488:
1489: This similarity has one disadvantage: It is too easy to confuse locals
1490: declarations with stack comments, causing bugs and making them hard to
1491: find. However, this problem can be avoided by appropriate coding
1492: conventions: Do not use both notations in the same program. If you do,
1493: they should be distinguished using additional means, e.g. by position.
1494:
1495: The name of the local may be preceded by a type specifier, e.g.,
1496: @code{F:} for a floating point value:
1497:
1498: @example
1499: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
1500: \ complex multiplication
1501: Ar Br f* Ai Bi f* f-
1502: Ar Bi f* Ai Br f* f+ ;
1503: @end example
1504:
1505: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
1506: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
1507: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
1508: with @code{W:}, @code{D:} etc.) produces its value and can be changed
1509: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
1510: produces its address (which becomes invalid when the variable's scope is
1511: left). E.g., the standard word @code{emit} can be defined in therms of
1512: @code{type} like this:
1513:
1514: @example
1515: : emit @{ C^ char* -- @}
1516: char* 1 type ;
1517: @end example
1518:
1519: A local without type specifier is a @code{W:} local. Both flavours of
1520: locals are initialized with values from the data or FP stack.
1521:
1522: Currently there is no way to define locals with user-defined data
1523: structures, but we are working on it.
1524:
1525: Gforth allows defining locals everywhere in a colon definition. This
1526: poses the following questions:
1527:
1528: @menu
1529: * Where are locals visible by name?::
1530: * How long do locals live?::
1531: * Programming Style::
1532: * Implementation::
1533: @end menu
1534:
1535: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
1536: @subsubsection Where are locals visible by name?
1537:
1538: Basically, the answer is that locals are visible where you would expect
1539: it in block-structured languages, and sometimes a little longer. If you
1540: want to restrict the scope of a local, enclose its definition in
1541: @code{SCOPE}...@code{ENDSCOPE}.
1542:
1543: doc-scope
1544: doc-endscope
1545:
1546: These words behave like control structure words, so you can use them
1547: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
1548: arbitrary ways.
1549:
1550: If you want a more exact answer to the visibility question, here's the
1551: basic principle: A local is visible in all places that can only be
1552: reached through the definition of the local@footnote{In compiler
1553: construction terminology, all places dominated by the definition of the
1554: local.}. In other words, it is not visible in places that can be reached
1555: without going through the definition of the local. E.g., locals defined
1556: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
1557: defined in @code{BEGIN}...@code{UNTIL} are visible after the
1558: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
1559:
1560: The reasoning behind this solution is: We want to have the locals
1561: visible as long as it is meaningful. The user can always make the
1562: visibility shorter by using explicit scoping. In a place that can
1563: only be reached through the definition of a local, the meaning of a
1564: local name is clear. In other places it is not: How is the local
1565: initialized at the control flow path that does not contain the
1566: definition? Which local is meant, if the same name is defined twice in
1567: two independent control flow paths?
1568:
1569: This should be enough detail for nearly all users, so you can skip the
1570: rest of this section. If you relly must know all the gory details and
1571: options, read on.
1572:
1573: In order to implement this rule, the compiler has to know which places
1574: are unreachable. It knows this automatically after @code{AHEAD},
1575: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
1576: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
1577: compiler that the control flow never reaches that place. If
1578: @code{UNREACHABLE} is not used where it could, the only consequence is
1579: that the visibility of some locals is more limited than the rule above
1580: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
1581: lie to the compiler), buggy code will be produced.
1582:
1583: Another problem with this rule is that at @code{BEGIN}, the compiler
1584: does not know which locals will be visible on the incoming
1585: back-edge. All problems discussed in the following are due to this
1586: ignorance of the compiler (we discuss the problems using @code{BEGIN}
1587: loops as examples; the discussion also applies to @code{?DO} and other
1588: loops). Perhaps the most insidious example is:
1589: @example
1590: AHEAD
1591: BEGIN
1592: x
1593: [ 1 CS-ROLL ] THEN
1594: @{ x @}
1595: ...
1596: UNTIL
1597: @end example
1598:
1599: This should be legal according to the visibility rule. The use of
1600: @code{x} can only be reached through the definition; but that appears
1601: textually below the use.
1602:
1603: From this example it is clear that the visibility rules cannot be fully
1604: implemented without major headaches. Our implementation treats common
1605: cases as advertised and the exceptions are treated in a safe way: The
1606: compiler makes a reasonable guess about the locals visible after a
1607: @code{BEGIN}; if it is too pessimistic, the
1608: user will get a spurious error about the local not being defined; if the
1609: compiler is too optimistic, it will notice this later and issue a
1610: warning. In the case above the compiler would complain about @code{x}
1611: being undefined at its use. You can see from the obscure examples in
1612: this section that it takes quite unusual control structures to get the
1613: compiler into trouble, and even then it will often do fine.
1614:
1615: If the @code{BEGIN} is reachable from above, the most optimistic guess
1616: is that all locals visible before the @code{BEGIN} will also be
1617: visible after the @code{BEGIN}. This guess is valid for all loops that
1618: are entered only through the @code{BEGIN}, in particular, for normal
1619: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
1620: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
1621: compiler. When the branch to the @code{BEGIN} is finally generated by
1622: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
1623: warns the user if it was too optimisitic:
1624: @example
1625: IF
1626: @{ x @}
1627: BEGIN
1628: \ x ?
1629: [ 1 cs-roll ] THEN
1630: ...
1631: UNTIL
1632: @end example
1633:
1634: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
1635: optimistically assumes that it lives until the @code{THEN}. It notices
1636: this difference when it compiles the @code{UNTIL} and issues a
1637: warning. The user can avoid the warning, and make sure that @code{x}
1638: is not used in the wrong area by using explicit scoping:
1639: @example
1640: IF
1641: SCOPE
1642: @{ x @}
1643: ENDSCOPE
1644: BEGIN
1645: [ 1 cs-roll ] THEN
1646: ...
1647: UNTIL
1648: @end example
1649:
1650: Since the guess is optimistic, there will be no spurious error messages
1651: about undefined locals.
1652:
1653: If the @code{BEGIN} is not reachable from above (e.g., after
1654: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
1655: optimistic guess, as the locals visible after the @code{BEGIN} may be
1656: defined later. Therefore, the compiler assumes that no locals are
1657: visible after the @code{BEGIN}. However, the user can use
1658: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
1659: visible at the BEGIN as at the point where the top control-flow stack
1660: item was created.
1661:
1662: doc-assume-live
1663:
1664: E.g.,
1665: @example
1666: @{ x @}
1667: AHEAD
1668: ASSUME-LIVE
1669: BEGIN
1670: x
1671: [ 1 CS-ROLL ] THEN
1672: ...
1673: UNTIL
1674: @end example
1675:
1676: Other cases where the locals are defined before the @code{BEGIN} can be
1677: handled by inserting an appropriate @code{CS-ROLL} before the
1678: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
1679: behind the @code{ASSUME-LIVE}).
1680:
1681: Cases where locals are defined after the @code{BEGIN} (but should be
1682: visible immediately after the @code{BEGIN}) can only be handled by
1683: rearranging the loop. E.g., the ``most insidious'' example above can be
1684: arranged into:
1685: @example
1686: BEGIN
1687: @{ x @}
1688: ... 0=
1689: WHILE
1690: x
1691: REPEAT
1692: @end example
1693:
1694: @node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
1695: @subsubsection How long do locals live?
1696:
1697: The right answer for the lifetime question would be: A local lives at
1698: least as long as it can be accessed. For a value-flavoured local this
1699: means: until the end of its visibility. However, a variable-flavoured
1700: local could be accessed through its address far beyond its visibility
1701: scope. Ultimately, this would mean that such locals would have to be
1702: garbage collected. Since this entails un-Forth-like implementation
1703: complexities, I adopted the same cowardly solution as some other
1704: languages (e.g., C): The local lives only as long as it is visible;
1705: afterwards its address is invalid (and programs that access it
1706: afterwards are erroneous).
1707:
1708: @node Programming Style, Implementation, How long do locals live?, Gforth locals
1709: @subsubsection Programming Style
1710:
1711: The freedom to define locals anywhere has the potential to change
1712: programming styles dramatically. In particular, the need to use the
1713: return stack for intermediate storage vanishes. Moreover, all stack
1714: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
1715: determined arguments) can be eliminated: If the stack items are in the
1716: wrong order, just write a locals definition for all of them; then
1717: write the items in the order you want.
1718:
1719: This seems a little far-fetched and eliminating stack manipulations is
1720: unlikely to become a conscious programming objective. Still, the number
1721: of stack manipulations will be reduced dramatically if local variables
1722: are used liberally (e.g., compare @code{max} in @ref{Gforth locals} with
1723: a traditional implementation of @code{max}).
1724:
1725: This shows one potential benefit of locals: making Forth programs more
1726: readable. Of course, this benefit will only be realized if the
1727: programmers continue to honour the principle of factoring instead of
1728: using the added latitude to make the words longer.
1729:
1730: Using @code{TO} can and should be avoided. Without @code{TO},
1731: every value-flavoured local has only a single assignment and many
1732: advantages of functional languages apply to Forth. I.e., programs are
1733: easier to analyse, to optimize and to read: It is clear from the
1734: definition what the local stands for, it does not turn into something
1735: different later.
1736:
1737: E.g., a definition using @code{TO} might look like this:
1738: @example
1739: : strcmp @{ addr1 u1 addr2 u2 -- n @}
1740: u1 u2 min 0
1741: ?do
1742: addr1 c@@ addr2 c@@ -
1743: ?dup-if
1744: unloop exit
1745: then
1746: addr1 char+ TO addr1
1747: addr2 char+ TO addr2
1748: loop
1749: u1 u2 - ;
1750: @end example
1751: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
1752: every loop iteration. @code{strcmp} is a typical example of the
1753: readability problems of using @code{TO}. When you start reading
1754: @code{strcmp}, you think that @code{addr1} refers to the start of the
1755: string. Only near the end of the loop you realize that it is something
1756: else.
1757:
1758: This can be avoided by defining two locals at the start of the loop that
1759: are initialized with the right value for the current iteration.
1760: @example
1761: : strcmp @{ addr1 u1 addr2 u2 -- n @}
1762: addr1 addr2
1763: u1 u2 min 0
1764: ?do @{ s1 s2 @}
1765: s1 c@@ s2 c@@ -
1766: ?dup-if
1767: unloop exit
1768: then
1769: s1 char+ s2 char+
1770: loop
1771: 2drop
1772: u1 u2 - ;
1773: @end example
1774: Here it is clear from the start that @code{s1} has a different value
1775: in every loop iteration.
1776:
1777: @node Implementation, , Programming Style, Gforth locals
1778: @subsubsection Implementation
1779:
1780: Gforth uses an extra locals stack. The most compelling reason for
1781: this is that the return stack is not float-aligned; using an extra stack
1782: also eliminates the problems and restrictions of using the return stack
1783: as locals stack. Like the other stacks, the locals stack grows toward
1784: lower addresses. A few primitives allow an efficient implementation:
1785:
1786: doc-@local#
1787: doc-f@local#
1788: doc-laddr#
1789: doc-lp+!#
1790: doc-lp!
1791: doc->l
1792: doc-f>l
1793:
1794: In addition to these primitives, some specializations of these
1795: primitives for commonly occurring inline arguments are provided for
1796: efficiency reasons, e.g., @code{@@local0} as specialization of
1797: @code{@@local#} for the inline argument 0. The following compiling words
1798: compile the right specialized version, or the general version, as
1799: appropriate:
1800:
1801: doc-compile-@local
1802: doc-compile-f@local
1803: doc-compile-lp+!
1804:
1805: Combinations of conditional branches and @code{lp+!#} like
1806: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
1807: is taken) are provided for efficiency and correctness in loops.
1808:
1809: A special area in the dictionary space is reserved for keeping the
1810: local variable names. @code{@{} switches the dictionary pointer to this
1811: area and @code{@}} switches it back and generates the locals
1812: initializing code. @code{W:} etc.@ are normal defining words. This
1813: special area is cleared at the start of every colon definition.
1814:
1815: A special feature of Gforth's dictionary is used to implement the
1816: definition of locals without type specifiers: every wordlist (aka
1817: vocabulary) has its own methods for searching
1818: etc. (@pxref{Wordlists}). For the present purpose we defined a wordlist
1819: with a special search method: When it is searched for a word, it
1820: actually creates that word using @code{W:}. @code{@{} changes the search
1821: order to first search the wordlist containing @code{@}}, @code{W:} etc.,
1822: and then the wordlist for defining locals without type specifiers.
1823:
1824: The lifetime rules support a stack discipline within a colon
1825: definition: The lifetime of a local is either nested with other locals
1826: lifetimes or it does not overlap them.
1827:
1828: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
1829: pointer manipulation is generated. Between control structure words
1830: locals definitions can push locals onto the locals stack. @code{AGAIN}
1831: is the simplest of the other three control flow words. It has to
1832: restore the locals stack depth of the corresponding @code{BEGIN}
1833: before branching. The code looks like this:
1834: @format
1835: @code{lp+!#} current-locals-size @minus{} dest-locals-size
1836: @code{branch} <begin>
1837: @end format
1838:
1839: @code{UNTIL} is a little more complicated: If it branches back, it
1840: must adjust the stack just like @code{AGAIN}. But if it falls through,
1841: the locals stack must not be changed. The compiler generates the
1842: following code:
1843: @format
1844: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
1845: @end format
1846: The locals stack pointer is only adjusted if the branch is taken.
1847:
1848: @code{THEN} can produce somewhat inefficient code:
1849: @format
1850: @code{lp+!#} current-locals-size @minus{} orig-locals-size
1851: <orig target>:
1852: @code{lp+!#} orig-locals-size @minus{} new-locals-size
1853: @end format
1854: The second @code{lp+!#} adjusts the locals stack pointer from the
1855: level at the @var{orig} point to the level after the @code{THEN}. The
1856: first @code{lp+!#} adjusts the locals stack pointer from the current
1857: level to the level at the orig point, so the complete effect is an
1858: adjustment from the current level to the right level after the
1859: @code{THEN}.
1860:
1861: In a conventional Forth implementation a dest control-flow stack entry
1862: is just the target address and an orig entry is just the address to be
1863: patched. Our locals implementation adds a wordlist to every orig or dest
1864: item. It is the list of locals visible (or assumed visible) at the point
1865: described by the entry. Our implementation also adds a tag to identify
1866: the kind of entry, in particular to differentiate between live and dead
1867: (reachable and unreachable) orig entries.
1868:
1869: A few unusual operations have to be performed on locals wordlists:
1870:
1871: doc-common-list
1872: doc-sub-list?
1873: doc-list-size
1874:
1875: Several features of our locals wordlist implementation make these
1876: operations easy to implement: The locals wordlists are organised as
1877: linked lists; the tails of these lists are shared, if the lists
1878: contain some of the same locals; and the address of a name is greater
1879: than the address of the names behind it in the list.
1880:
1881: Another important implementation detail is the variable
1882: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
1883: determine if they can be reached directly or only through the branch
1884: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
1885: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
1886: definition, by @code{BEGIN} and usually by @code{THEN}.
1887:
1888: Counted loops are similar to other loops in most respects, but
1889: @code{LEAVE} requires special attention: It performs basically the same
1890: service as @code{AHEAD}, but it does not create a control-flow stack
1891: entry. Therefore the information has to be stored elsewhere;
1892: traditionally, the information was stored in the target fields of the
1893: branches created by the @code{LEAVE}s, by organizing these fields into a
1894: linked list. Unfortunately, this clever trick does not provide enough
1895: space for storing our extended control flow information. Therefore, we
1896: introduce another stack, the leave stack. It contains the control-flow
1897: stack entries for all unresolved @code{LEAVE}s.
1898:
1899: Local names are kept until the end of the colon definition, even if
1900: they are no longer visible in any control-flow path. In a few cases
1901: this may lead to increased space needs for the locals name area, but
1902: usually less than reclaiming this space would cost in code size.
1903:
1904:
1905: @node ANS Forth locals, , Gforth locals, Locals
1906: @subsection ANS Forth locals
1907:
1908: The ANS Forth locals wordset does not define a syntax for locals, but
1909: words that make it possible to define various syntaxes. One of the
1910: possible syntaxes is a subset of the syntax we used in the Gforth locals
1911: wordset, i.e.:
1912:
1913: @example
1914: @{ local1 local2 ... -- comment @}
1915: @end example
1916: or
1917: @example
1918: @{ local1 local2 ... @}
1919: @end example
1920:
1921: The order of the locals corresponds to the order in a stack comment. The
1922: restrictions are:
1923:
1924: @itemize @bullet
1925: @item
1926: Locals can only be cell-sized values (no type specifiers are allowed).
1927: @item
1928: Locals can be defined only outside control structures.
1929: @item
1930: Locals can interfere with explicit usage of the return stack. For the
1931: exact (and long) rules, see the standard. If you don't use return stack
1932: accessing words in a definition using locals, you will be all right. The
1933: purpose of this rule is to make locals implementation on the return
1934: stack easier.
1935: @item
1936: The whole definition must be in one line.
1937: @end itemize
1938:
1939: Locals defined in this way behave like @code{VALUE}s (@xref{Simple
1940: Defining Words}). I.e., they are initialized from the stack. Using their
1941: name produces their value. Their value can be changed using @code{TO}.
1942:
1943: Since this syntax is supported by Gforth directly, you need not do
1944: anything to use it. If you want to port a program using this syntax to
1945: another ANS Forth system, use @file{compat/anslocal.fs} to implement the
1946: syntax on the other system.
1947:
1948: Note that a syntax shown in the standard, section A.13 looks
1949: similar, but is quite different in having the order of locals
1950: reversed. Beware!
1951:
1952: The ANS Forth locals wordset itself consists of the following word
1953:
1954: doc-(local)
1955:
1956: The ANS Forth locals extension wordset defines a syntax, but it is so
1957: awful that we strongly recommend not to use it. We have implemented this
1958: syntax to make porting to Gforth easy, but do not document it here. The
1959: problem with this syntax is that the locals are defined in an order
1960: reversed with respect to the standard stack comment notation, making
1961: programs harder to read, and easier to misread and miswrite. The only
1962: merit of this syntax is that it is easy to implement using the ANS Forth
1963: locals wordset.
1964:
1965: @node Defining Words, Wordlists, Locals, Words
1966: @section Defining Words
1967:
1968: @menu
1969: * Simple Defining Words::
1970: * Colon Definitions::
1971: * User-defined Defining Words::
1972: * Supplying names::
1973: * Interpretation and Compilation Semantics::
1974: @end menu
1975:
1976: @node Simple Defining Words, Colon Definitions, Defining Words, Defining Words
1977: @subsection Simple Defining Words
1978:
1979: doc-constant
1980: doc-2constant
1981: doc-fconstant
1982: doc-variable
1983: doc-2variable
1984: doc-fvariable
1985: doc-create
1986: doc-user
1987: doc-value
1988: doc-to
1989: doc-defer
1990: doc-is
1991:
1992: @node Colon Definitions, User-defined Defining Words, Simple Defining Words, Defining Words
1993: @subsection Colon Definitions
1994:
1995: @example
1996: : name ( ... -- ... )
1997: word1 word2 word3 ;
1998: @end example
1999:
2000: creates a word called @code{name}, that, upon execution, executes
2001: @code{word1 word2 word3}. @code{name} is a @dfn{(colon) definition}.
2002:
2003: The explanation above is somewhat superficial. @xref{Interpretation and
2004: Compilation Semantics} for an in-depth discussion of some of the issues
2005: involved.
2006:
2007: doc-:
2008: doc-;
2009:
2010: @node User-defined Defining Words, Supplying names, Colon Definitions, Defining Words
2011: @subsection User-defined Defining Words
2012:
2013: You can create new defining words simply by wrapping defining-time code
2014: around existing defining words and putting the sequence in a colon
2015: definition.
2016:
2017: If you want the words defined with your defining words to behave
2018: differently from words defined with standard defining words, you can
2019: write your defining word like this:
2020:
2021: @example
2022: : def-word ( "name" -- )
2023: Create @var{code1}
2024: DOES> ( ... -- ... )
2025: @var{code2} ;
2026:
2027: def-word name
2028: @end example
2029:
2030: Technically, this fragment defines a defining word @code{def-word}, and
2031: a word @code{name}; when you execute @code{name}, the address of the
2032: body of @code{name} is put on the data stack and @var{code2} is executed
2033: (the address of the body of @code{name} is the address @code{HERE}
2034: returns immediately after the @code{CREATE}).
2035:
2036: In other words, if you make the following definitions:
2037:
2038: @example
2039: : def-word1 ( "name" -- )
2040: Create @var{code1} ;
2041:
2042: : action1 ( ... -- ... )
2043: @var{code2} ;
2044:
2045: def-word name1
2046: @end example
2047:
2048: Using @code{name1 action1} is equivalent to using @code{name}.
2049:
2050: E.g., you can implement @code{Constant} in this way:
2051:
2052: @example
2053: : constant ( w "name" -- )
2054: create ,
2055: DOES> ( -- w )
2056: @@ ;
2057: @end example
2058:
2059: When you create a constant with @code{5 constant five}, first a new word
2060: @code{five} is created, then the value 5 is laid down in the body of
2061: @code{five} with @code{,}. When @code{five} is invoked, the address of
2062: the body is put on the stack, and @code{@@} retrieves the value 5.
2063:
2064: In the example above the stack comment after the @code{DOES>} specifies
2065: the stack effect of the defined words, not the stack effect of the
2066: following code (the following code expects the address of the body on
2067: the top of stack, which is not reflected in the stack comment). This is
2068: the convention that I use and recommend (it clashes a bit with using
2069: locals declarations for stack effect specification, though).
2070:
2071: @subsubsection Applications of @code{CREATE..DOES>}
2072:
2073: You may wonder how to use this feature. Here are some usage patterns:
2074:
2075: When you see a sequence of code occurring several times, and you can
2076: identify a meaning, you will factor it out as a colon definition. When
2077: you see similar colon definitions, you can factor them using
2078: @code{CREATE..DOES>}. E.g., an assembler usually defines several words
2079: that look very similar:
2080: @example
2081: : ori, ( reg-taget reg-source n -- )
2082: 0 asm-reg-reg-imm ;
2083: : andi, ( reg-taget reg-source n -- )
2084: 1 asm-reg-reg-imm ;
2085: @end example
2086:
2087: This could be factored with:
2088: @example
2089: : reg-reg-imm ( op-code -- )
2090: create ,
2091: DOES> ( reg-taget reg-source n -- )
2092: @@ asm-reg-reg-imm ;
2093:
2094: 0 reg-reg-imm ori,
2095: 1 reg-reg-imm andi,
2096: @end example
2097:
2098: Another view of @code{CREATE..DOES>} is to consider it as a crude way to
2099: supply a part of the parameters for a word (known as @dfn{currying} in
2100: the functional language community). E.g., @code{+} needs two
2101: parameters. Creating versions of @code{+} with one parameter fixed can
2102: be done like this:
2103: @example
2104: : curry+ ( n1 -- )
2105: create ,
2106: DOES> ( n2 -- n1+n2 )
2107: @@ + ;
2108:
2109: 3 curry+ 3+
2110: -2 curry+ 2-
2111: @end example
2112:
2113: @subsubsection The gory details of @code{CREATE..DOES>}
2114:
2115: doc-does>
2116:
2117: This means that you need not use @code{CREATE} and @code{DOES>} in the
2118: same definition; E.g., you can put the @code{DOES>}-part in a separate
2119: definition. This allows us to, e.g., select among different DOES>-parts:
2120: @example
2121: : does1
2122: DOES> ( ... -- ... )
2123: ... ;
2124:
2125: : does2
2126: DOES> ( ... -- ... )
2127: ... ;
2128:
2129: : def-word ( ... -- ... )
2130: create ...
2131: IF
2132: does1
2133: ELSE
2134: does2
2135: ENDIF ;
2136: @end example
2137:
2138: In a standard program you can apply a @code{DOES>}-part only if the last
2139: word was defined with @code{CREATE}. In Gforth, the @code{DOES>}-part
2140: will override the behaviour of the last word defined in any case. In a
2141: standard program, you can use @code{DOES>} only in a colon
2142: definition. In Gforth, you can also use it in interpretation state, in a
2143: kind of one-shot mode:
2144: @example
2145: CREATE name ( ... -- ... )
2146: @var{initialization}
2147: DOES>
2148: @var{code} ;
2149: @end example
2150: This is equivalwent to the standard
2151: @example
2152: :noname
2153: DOES>
2154: @var{code} ;
2155: CREATE name EXECUTE ( ... -- ... )
2156: @var{initialization}
2157: @end example
2158:
2159: You can get the address of the body of a word with
2160:
2161: doc->body
2162:
2163: @node Supplying names, Interpretation and Compilation Semantics, User-defined Defining Words, Defining Words
2164: @subsection Supplying names for the defined words
2165:
2166: By default, defining words take the names for the defined words from the
2167: input stream. Sometimes you want to supply the name from a string. You
2168: can do this with
2169:
2170: doc-nextname
2171:
2172: E.g.,
2173:
2174: @example
2175: s" foo" nextname create
2176: @end example
2177: is equivalent to
2178: @example
2179: create foo
2180: @end example
2181:
2182: Sometimes you want to define a word without a name. You can do this with
2183:
2184: doc-noname
2185:
2186: To make any use of the newly defined word, you need its execution
2187: token. You can get it with
2188:
2189: doc-lastxt
2190:
2191: E.g., you can initialize a deferred word with an anonymous colon
2192: definition:
2193: @example
2194: Defer deferred
2195: noname : ( ... -- ... )
2196: ... ;
2197: lastxt IS deferred
2198: @end example
2199:
2200: @code{lastxt} also works when the last word was not defined as
2201: @code{noname}.
2202:
2203: The standard has also recognized the need for anonymous words and
2204: provides
2205:
2206: doc-:noname
2207:
2208: This leaves the execution token for the word on the stack after the
2209: closing @code{;}. You can rewrite the last example with @code{:noname}:
2210: @example
2211: Defer deferred
2212: :noname ( ... -- ... )
2213: ... ;
2214: IS deferred
2215: @end example
2216:
2217: @node Interpretation and Compilation Semantics, , Supplying names, Defining Words
2218: @subsection Interpretation and Compilation Semantics
2219:
2220: The @dfn{interpretation semantics} of a word are what the text
2221: interpreter does when it encounters the word in interpret state. It also
2222: appears in some other contexts, e.g., the execution token returned by
2223: @code{' @var{word}} identifies the interpretation semantics of
2224: @var{word} (in other words, @code{' @var{word} execute} is equivalent to
2225: interpret-state text interpretation of @code{@var{word}}).
2226:
2227: The @dfn{compilation semantics} of a word are what the text interpreter
2228: does when it encounters the word in compile state. It also appears in
2229: other contexts, e.g, @code{POSTPONE @var{word}} compiles@footnote{In
2230: standard terminology, ``appends to the current definition''.} the
2231: compilation semantics of @var{word}.
2232:
2233: The standard also talks about @dfn{execution semantics}. They are used
2234: only for defining the interpretation and compilation semantics of many
2235: words. By default, the interpretation semantics of a word are to
2236: @code{execute} its execution semantics, and the compilation semantics of
2237: a word are to @code{compile,} its execution semantics.@footnote{In
2238: standard terminology: The default interpretation semantics are its
2239: execution semantics; the default compilation semantics are to append its
2240: execution semantics to the execution semantics of the current
2241: definition.}
2242:
2243: You can change the compilation semantics into @code{execute}ing the
2244: execution semantics with
2245:
2246: doc-immediate
2247:
2248: You can remove the interpretation semantics of a word with
2249:
2250: doc-compile-only
2251: doc-restrict
2252:
2253: Note that ticking (@code{'}) compile-only words gives an error
2254: (``Interpreting a compile-only word'').
2255:
2256: Gforth also allows you to define words with arbitrary combinations of
2257: interpretation and compilation semantics.
2258:
2259: doc-interpret/compile:
2260:
2261: This feature was introduced for implementing @code{TO} and @code{S"}. I
2262: recommend that you do not define such words, as cute as they may be:
2263: they make it hard to get at both parts of the word in some contexts.
2264: E.g., assume you want to get an execution token for the compilation
2265: part. Instead, define two words, one that embodies the interpretation
2266: part, and one that embodies the compilation part.
2267:
2268: There is, however, a potentially useful application of this feature:
2269: Providing differing implementations for the default semantics. While
2270: this introduces redundancy and is therefore usually a bad idea, a
2271: performance improvement may be worth the trouble. E.g., consider the
2272: word @code{foobar}:
2273:
2274: @example
2275: : foobar
2276: foo bar ;
2277: @end example
2278:
2279: Let us assume that @code{foobar} is called so frequently that the
2280: calling overhead would take a significant amount of the run-time. We can
2281: optimize it with @code{interpret/compile:}:
2282:
2283: @example
2284: :noname
2285: foo bar ;
2286: :noname
2287: POSTPONE foo POSTPONE bar ;
2288: interpret/compile: foobar
2289: @end example
2290:
2291: This definition has the same interpretation semantics and essentially
2292: the same compilation semantics as the simple definition of
2293: @code{foobar}, but the implementation of the compilation semantics is
2294: more efficient with respect to run-time.
2295:
2296: Some people try to use state-smart words to emulate the feature provided
2297: by @code{interpret/compile:} (words are state-smart if they check
2298: @code{STATE} during execution). E.g., they would try to code
2299: @code{foobar} like this:
2300:
2301: @example
2302: : foobar
2303: STATE @@
2304: IF ( compilation state )
2305: POSTPONE foo POSTPONE bar
2306: ELSE
2307: foo bar
2308: ENDIF ; immediate
2309: @end example
2310:
2311: While this works if @code{foobar} is processed only by the text
2312: interpreter, it does not work in other contexts (like @code{'} or
2313: @code{POSTPONE}). E.g., @code{' foobar} will produce an execution token
2314: for a state-smart word, not for the interpretation semantics of the
2315: original @code{foobar}; when you execute this execution token (directly
2316: with @code{EXECUTE} or indirectly through @code{COMPILE,}) in compile
2317: state, the result will not be what you expected (i.e., it will not
2318: perform @code{foo bar}). State-smart words are a bad idea. Simply don't
2319: write them!
2320:
2321: It is also possible to write defining words that define words with
2322: arbitrary combinations of interpretation and compilation semantics (or,
2323: preferably, arbitrary combinations of implementations of the default
2324: semantics). In general, this looks like:
2325:
2326: @example
2327: : def-word
2328: create-interpret/compile
2329: @var{code1}
2330: interpretation>
2331: @var{code2}
2332: <interpretation
2333: compilation>
2334: @var{code3}
2335: <compilation ;
2336: @end example
2337:
2338: For a @var{word} defined with @code{def-word}, the interpretation
2339: semantics are to push the address of the body of @var{word} and perform
2340: @var{code2}, and the compilation semantics are to push the address of
2341: the body of @var{word} and perform @var{code3}. E.g., @code{constant}
2342: can also be defined like this:
2343:
2344: @example
2345: : constant ( n "name" -- )
2346: create-interpret/compile
2347: ,
2348: interpretation> ( -- n )
2349: @@
2350: <interpretation
2351: compilation> ( compilation. -- ; run-time. -- n )
2352: @@ postpone literal
2353: <compilation ;
2354: @end example
2355:
2356: doc-create-interpret/compile
2357: doc-interpretation>
2358: doc-<interpretation
2359: doc-compilation>
2360: doc-<compilation
2361:
2362: Note that words defined with @code{interpret/compile:} and
2363: @code{create-interpret/compile} have an extended header structure that
2364: differs from other words; however, unless you try to access them with
2365: plain address arithmetic, you should not notice this. Words for
2366: accessing the header structure usually know how to deal with this; e.g.,
2367: @code{' word >body} also gives you the body of a word created with
2368: @code{create-interpret/compile}.
2369:
2370: @node Wordlists, Files, Defining Words, Words
2371: @section Wordlists
2372:
2373: @node Files, Blocks, Wordlists, Words
2374: @section Files
2375:
2376: @node Blocks, Other I/O, Files, Words
2377: @section Blocks
2378:
2379: @node Other I/O, Programming Tools, Blocks, Words
2380: @section Other I/O
2381:
2382: @node Programming Tools, Assembler and Code words, Other I/O, Words
2383: @section Programming Tools
2384:
2385: @menu
2386: * Debugging:: Simple and quick.
2387: * Assertions:: Making your programs self-checking.
2388: @end menu
2389:
2390: @node Debugging, Assertions, Programming Tools, Programming Tools
2391: @subsection Debugging
2392:
2393: The simple debugging aids provided in @file{debugging.fs}
2394: are meant to support a different style of debugging than the
2395: tracing/stepping debuggers used in languages with long turn-around
2396: times.
2397:
2398: A much better (faster) way in fast-compilig languages is to add
2399: printing code at well-selected places, let the program run, look at
2400: the output, see where things went wrong, add more printing code, etc.,
2401: until the bug is found.
2402:
2403: The word @code{~~} is easy to insert. It just prints debugging
2404: information (by default the source location and the stack contents). It
2405: is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
2406: query-replace them with nothing). The deferred words
2407: @code{printdebugdata} and @code{printdebugline} control the output of
2408: @code{~~}. The default source location output format works well with
2409: Emacs' compilation mode, so you can step through the program at the
2410: source level using @kbd{C-x `} (the advantage over a stepping debugger
2411: is that you can step in any direction and you know where the crash has
2412: happened or where the strange data has occurred).
2413:
2414: Note that the default actions clobber the contents of the pictured
2415: numeric output string, so you should not use @code{~~}, e.g., between
2416: @code{<#} and @code{#>}.
2417:
2418: doc-~~
2419: doc-printdebugdata
2420: doc-printdebugline
2421:
2422: @node Assertions, , Debugging, Programming Tools
2423: @subsection Assertions
2424:
2425: It is a good idea to make your programs self-checking, in particular, if
2426: you use an assumption (e.g., that a certain field of a data structure is
2427: never zero) that may become wrong during maintenance. Gforth supports
2428: assertions for this purpose. They are used like this:
2429:
2430: @example
2431: assert( @var{flag} )
2432: @end example
2433:
2434: The code between @code{assert(} and @code{)} should compute a flag, that
2435: should be true if everything is alright and false otherwise. It should
2436: not change anything else on the stack. The overall stack effect of the
2437: assertion is @code{( -- )}. E.g.
2438:
2439: @example
2440: assert( 1 1 + 2 = ) \ what we learn in school
2441: assert( dup 0<> ) \ assert that the top of stack is not zero
2442: assert( false ) \ this code should not be reached
2443: @end example
2444:
2445: The need for assertions is different at different times. During
2446: debugging, we want more checking, in production we sometimes care more
2447: for speed. Therefore, assertions can be turned off, i.e., the assertion
2448: becomes a comment. Depending on the importance of an assertion and the
2449: time it takes to check it, you may want to turn off some assertions and
2450: keep others turned on. Gforth provides several levels of assertions for
2451: this purpose:
2452:
2453: doc-assert0(
2454: doc-assert1(
2455: doc-assert2(
2456: doc-assert3(
2457: doc-assert(
2458: doc-)
2459:
2460: @code{Assert(} is the same as @code{assert1(}. The variable
2461: @code{assert-level} specifies the highest assertions that are turned
2462: on. I.e., at the default @code{assert-level} of one, @code{assert0(} and
2463: @code{assert1(} assertions perform checking, while @code{assert2(} and
2464: @code{assert3(} assertions are treated as comments.
2465:
2466: Note that the @code{assert-level} is evaluated at compile-time, not at
2467: run-time. I.e., you cannot turn assertions on or off at run-time, you
2468: have to set the @code{assert-level} appropriately before compiling a
2469: piece of code. You can compile several pieces of code at several
2470: @code{assert-level}s (e.g., a trusted library at level 1 and newly
2471: written code at level 3).
2472:
2473: doc-assert-level
2474:
2475: If an assertion fails, a message compatible with Emacs' compilation mode
2476: is produced and the execution is aborted (currently with @code{ABORT"}.
2477: If there is interest, we will introduce a special throw code. But if you
2478: intend to @code{catch} a specific condition, using @code{throw} is
2479: probably more appropriate than an assertion).
2480:
2481: @node Assembler and Code words, Threading Words, Programming Tools, Words
2482: @section Assembler and Code words
2483:
2484: Gforth provides some words for defining primitives (words written in
2485: machine code), and for defining the the machine-code equivalent of
2486: @code{DOES>}-based defining words. However, the machine-independent
2487: nature of Gforth poses a few problems: First of all. Gforth runs on
2488: several architectures, so it can provide no standard assembler. What's
2489: worse is that the register allocation not only depends on the processor,
2490: but also on the @code{gcc} version and options used.
2491:
2492: The words that Gforth offers encapsulate some system dependences (e.g., the
2493: header structure), so a system-independent assembler may be used in
2494: Gforth. If you do not have an assembler, you can compile machine code
2495: directly with @code{,} and @code{c,}.
2496:
2497: doc-assembler
2498: doc-code
2499: doc-end-code
2500: doc-;code
2501: doc-flush-icache
2502:
2503: If @code{flush-icache} does not work correctly, @code{code} words
2504: etc. will not work (reliably), either.
2505:
2506: These words are rarely used. Therefore they reside in @code{code.fs},
2507: which is usually not loaded (except @code{flush-icache}, which is always
2508: present). You can load them with @code{require code.fs}.
2509:
2510: In the assembly code you will want to refer to the inner interpreter's
2511: registers (e.g., the data stack pointer) and you may want to use other
2512: registers for temporary storage. Unfortunately, the register allocation
2513: is installation-dependent.
2514:
2515: The easiest solution is to use explicit register declarations
2516: (@pxref{Explicit Reg Vars, , Variables in Specified Registers, gcc.info,
2517: GNU C Manual}) for all of the inner interpreter's registers: You have to
2518: compile Gforth with @code{-DFORCE_REG} (configure option
2519: @code{--enable-force-reg}) and the appropriate declarations must be
2520: present in the @code{machine.h} file (see @code{mips.h} for an example;
2521: you can find a full list of all declarable register symbols with
2522: @code{grep register engine.c}). If you give explicit registers to all
2523: variables that are declared at the beginning of @code{engine()}, you
2524: should be able to use the other caller-saved registers for temporary
2525: storage. Alternatively, you can use the @code{gcc} option
2526: @code{-ffixed-REG} (@pxref{Code Gen Options, , Options for Code
2527: Generation Conventions, gcc.info, GNU C Manual}) to reserve a register
2528: (however, this restriction on register allocation may slow Gforth
2529: significantly).
2530:
2531: If this solution is not viable (e.g., because @code{gcc} does not allow
2532: you to explicitly declare all the registers you need), you have to find
2533: out by looking at the code where the inner interpreter's registers
2534: reside and which registers can be used for temporary storage. You can
2535: get an assembly listing of the engine's code with @code{make engine.s}.
2536:
2537: In any case, it is good practice to abstract your assembly code from the
2538: actual register allocation. E.g., if the data stack pointer resides in
2539: register @code{$17}, create an alias for this register called @code{sp},
2540: and use that in your assembly code.
2541:
2542: Another option for implementing normal and defining words efficiently
2543: is: adding the wanted functionality to the source of Gforth. For normal
2544: words you just have to edit @file{primitives} (@pxref{Automatic
2545: Generation}), defining words (equivalent to @code{;CODE} words, for fast
2546: defined words) may require changes in @file{engine.c}, @file{kernal.fs},
2547: @file{prims2x.fs}, and possibly @file{cross.fs}.
2548:
2549:
2550: @node Threading Words, , Assembler and Code words, Words
2551: @section Threading Words
2552:
2553: These words provide access to code addresses and other threading stuff
2554: in Gforth (and, possibly, other interpretive Forths). It more or less
2555: abstracts away the differences between direct and indirect threading
2556: (and, for direct threading, the machine dependences). However, at
2557: present this wordset is still inclomplete. It is also pretty low-level;
2558: some day it will hopefully be made unnecessary by an internals words set
2559: that abstracts implementation details away completely.
2560:
2561: doc->code-address
2562: doc->does-code
2563: doc-code-address!
2564: doc-does-code!
2565: doc-does-handler!
2566: doc-/does-handler
2567:
2568: The code addresses produced by various defining words are produced by
2569: the following words:
2570:
2571: doc-docol:
2572: doc-docon:
2573: doc-dovar:
2574: doc-douser:
2575: doc-dodefer:
2576: doc-dofield:
2577:
2578: You can recognize words defined by a @code{CREATE}...@code{DOES>} word
2579: with @code{>DOES-CODE}. If the word was defined in that way, the value
2580: returned is different from 0 and identifies the @code{DOES>} used by the
2581: defining word.
2582:
2583: @node ANS conformance, Model, Words, Top
2584: @chapter ANS conformance
2585:
2586: To the best of our knowledge, Gforth is an
2587:
2588: ANS Forth System
2589: @itemize @bullet
2590: @item providing the Core Extensions word set
2591: @item providing the Block word set
2592: @item providing the Block Extensions word set
2593: @item providing the Double-Number word set
2594: @item providing the Double-Number Extensions word set
2595: @item providing the Exception word set
2596: @item providing the Exception Extensions word set
2597: @item providing the Facility word set
2598: @item providing @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
2599: @item providing the File Access word set
2600: @item providing the File Access Extensions word set
2601: @item providing the Floating-Point word set
2602: @item providing the Floating-Point Extensions word set
2603: @item providing the Locals word set
2604: @item providing the Locals Extensions word set
2605: @item providing the Memory-Allocation word set
2606: @item providing the Memory-Allocation Extensions word set (that one's easy)
2607: @item providing the Programming-Tools word set
2608: @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
2609: @item providing the Search-Order word set
2610: @item providing the Search-Order Extensions word set
2611: @item providing the String word set
2612: @item providing the String Extensions word set (another easy one)
2613: @end itemize
2614:
2615: In addition, ANS Forth systems are required to document certain
2616: implementation choices. This chapter tries to meet these
2617: requirements. In many cases it gives a way to ask the system for the
2618: information instead of providing the information directly, in
2619: particular, if the information depends on the processor, the operating
2620: system or the installation options chosen, or if they are likely to
2621: change during the maintenance of Gforth.
2622:
2623: @comment The framework for the rest has been taken from pfe.
2624:
2625: @menu
2626: * The Core Words::
2627: * The optional Block word set::
2628: * The optional Double Number word set::
2629: * The optional Exception word set::
2630: * The optional Facility word set::
2631: * The optional File-Access word set::
2632: * The optional Floating-Point word set::
2633: * The optional Locals word set::
2634: * The optional Memory-Allocation word set::
2635: * The optional Programming-Tools word set::
2636: * The optional Search-Order word set::
2637: @end menu
2638:
2639:
2640: @c =====================================================================
2641: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
2642: @comment node-name, next, previous, up
2643: @section The Core Words
2644: @c =====================================================================
2645:
2646: @menu
2647: * core-idef:: Implementation Defined Options
2648: * core-ambcond:: Ambiguous Conditions
2649: * core-other:: Other System Documentation
2650: @end menu
2651:
2652: @c ---------------------------------------------------------------------
2653: @node core-idef, core-ambcond, The Core Words, The Core Words
2654: @subsection Implementation Defined Options
2655: @c ---------------------------------------------------------------------
2656:
2657: @table @i
2658:
2659: @item (Cell) aligned addresses:
2660: processor-dependent. Gforth's alignment words perform natural alignment
2661: (e.g., an address aligned for a datum of size 8 is divisible by
2662: 8). Unaligned accesses usually result in a @code{-23 THROW}.
2663:
2664: @item @code{EMIT} and non-graphic characters:
2665: The character is output using the C library function (actually, macro)
2666: @code{putc}.
2667:
2668: @item character editing of @code{ACCEPT} and @code{EXPECT}:
2669: This is modeled on the GNU readline library (@pxref{Readline
2670: Interaction, , Command Line Editing, readline, The GNU Readline
2671: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
2672: producing a full word completion every time you type it (instead of
2673: producing the common prefix of all completions).
2674:
2675: @item character set:
2676: The character set of your computer and display device. Gforth is
2677: 8-bit-clean (but some other component in your system may make trouble).
2678:
2679: @item Character-aligned address requirements:
2680: installation-dependent. Currently a character is represented by a C
2681: @code{unsigned char}; in the future we might switch to @code{wchar_t}
2682: (Comments on that requested).
2683:
2684: @item character-set extensions and matching of names:
2685: Any character except the ASCII NUL charcter can be used in a
2686: name. Matching is case-insensitive (except in @code{TABLE}s. The
2687: matching is performed using the C function @code{strncasecmp}, whose
2688: function is probably influenced by the locale. E.g., the @code{C} locale
2689: does not know about accents and umlauts, so they are matched
2690: case-sensitively in that locale. For portability reasons it is best to
2691: write programs such that they work in the @code{C} locale. Then one can
2692: use libraries written by a Polish programmer (who might use words
2693: containing ISO Latin-2 encoded characters) and by a French programmer
2694: (ISO Latin-1) in the same program (of course, @code{WORDS} will produce
2695: funny results for some of the words (which ones, depends on the font you
2696: are using)). Also, the locale you prefer may not be available in other
2697: operating systems. Hopefully, Unicode will solve these problems one day.
2698:
2699: @item conditions under which control characters match a space delimiter:
2700: If @code{WORD} is called with the space character as a delimiter, all
2701: white-space characters (as identified by the C macro @code{isspace()})
2702: are delimiters. @code{PARSE}, on the other hand, treats space like other
2703: delimiters. @code{PARSE-WORD} treats space like @code{WORD}, but behaves
2704: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
2705: interpreter (aka text interpreter) by default, treats all white-space
2706: characters as delimiters.
2707:
2708: @item format of the control flow stack:
2709: The data stack is used as control flow stack. The size of a control flow
2710: stack item in cells is given by the constant @code{cs-item-size}. At the
2711: time of this writing, an item consists of a (pointer to a) locals list
2712: (third), an address in the code (second), and a tag for identifying the
2713: item (TOS). The following tags are used: @code{defstart},
2714: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
2715: @code{scopestart}.
2716:
2717: @item conversion of digits > 35
2718: The characters @code{[\]^_'} are the digits with the decimal value
2719: 36@minus{}41. There is no way to input many of the larger digits.
2720:
2721: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
2722: The cursor is moved to the end of the entered string. If the input is
2723: terminated using the @kbd{Return} key, a space is typed.
2724:
2725: @item exception abort sequence of @code{ABORT"}:
2726: The error string is stored into the variable @code{"error} and a
2727: @code{-2 throw} is performed.
2728:
2729: @item input line terminator:
2730: For interactive input, @kbd{C-m} (CR) and @kbd{C-j} (LF) terminate
2731: lines. One of these characters is typically produced when you type the
2732: @kbd{Enter} or @kbd{Return} key.
2733:
2734: @item maximum size of a counted string:
2735: @code{s" /counted-string" environment? drop .}. Currently 255 characters
2736: on all ports, but this may change.
2737:
2738: @item maximum size of a parsed string:
2739: Given by the constant @code{/line}. Currently 255 characters.
2740:
2741: @item maximum size of a definition name, in characters:
2742: 31
2743:
2744: @item maximum string length for @code{ENVIRONMENT?}, in characters:
2745: 31
2746:
2747: @item method of selecting the user input device:
2748: The user input device is the standard input. There is currently no way to
2749: change it from within Gforth. However, the input can typically be
2750: redirected in the command line that starts Gforth.
2751:
2752: @item method of selecting the user output device:
2753: @code{EMIT} and @code{TYPE} output to the file-id stored in the value
2754: @code{outfile-id} (@code{stdout} by default). Gforth uses buffered
2755: output, so output on a terminal does not become visible before the next
2756: newline or buffer overflow. Output on non-terminals is invisible until
2757: the buffer overflows.
2758:
2759: @item methods of dictionary compilation:
2760: What are we expected to document here?
2761:
2762: @item number of bits in one address unit:
2763: @code{s" address-units-bits" environment? drop .}. 8 in all current
2764: ports.
2765:
2766: @item number representation and arithmetic:
2767: Processor-dependent. Binary two's complement on all current ports.
2768:
2769: @item ranges for integer types:
2770: Installation-dependent. Make environmental queries for @code{MAX-N},
2771: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
2772: unsigned (and positive) types is 0. The lower bound for signed types on
2773: two's complement and one's complement machines machines can be computed
2774: by adding 1 to the upper bound.
2775:
2776: @item read-only data space regions:
2777: The whole Forth data space is writable.
2778:
2779: @item size of buffer at @code{WORD}:
2780: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
2781: shared with the pictured numeric output string. If overwriting
2782: @code{PAD} is acceptable, it is as large as the remaining dictionary
2783: space, although only as much can be sensibly used as fits in a counted
2784: string.
2785:
2786: @item size of one cell in address units:
2787: @code{1 cells .}.
2788:
2789: @item size of one character in address units:
2790: @code{1 chars .}. 1 on all current ports.
2791:
2792: @item size of the keyboard terminal buffer:
2793: Varies. You can determine the size at a specific time using @code{lp@@
2794: tib - .}. It is shared with the locals stack and TIBs of files that
2795: include the current file. You can change the amount of space for TIBs
2796: and locals stack at Gforth startup with the command line option
2797: @code{-l}.
2798:
2799: @item size of the pictured numeric output buffer:
2800: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
2801: shared with @code{WORD}.
2802:
2803: @item size of the scratch area returned by @code{PAD}:
2804: The remainder of dictionary space. You can even use the unused part of
2805: the data stack space. The current size can be computed with @code{sp@@
2806: pad - .}.
2807:
2808: @item system case-sensitivity characteristics:
2809: Dictionary searches are case insensitive (except in
2810: @code{TABLE}s). However, as explained above under @i{character-set
2811: extensions}, the matching for non-ASCII characters is determined by the
2812: locale you are using. In the default @code{C} locale all non-ASCII
2813: characters are matched case-sensitively.
2814:
2815: @item system prompt:
2816: @code{ ok} in interpret state, @code{ compiled} in compile state.
2817:
2818: @item division rounding:
2819: installation dependent. @code{s" floored" environment? drop .}. We leave
2820: the choice to @code{gcc} (what to use for @code{/}) and to you (whether to use
2821: @code{fm/mod}, @code{sm/rem} or simply @code{/}).
2822:
2823: @item values of @code{STATE} when true:
2824: -1.
2825:
2826: @item values returned after arithmetic overflow:
2827: On two's complement machines, arithmetic is performed modulo
2828: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
2829: arithmetic (with appropriate mapping for signed types). Division by zero
2830: typically results in a @code{-55 throw} (Floating-point unidentified
2831: fault), although a @code{-10 throw} (divide by zero) would be more
2832: appropriate.
2833:
2834: @item whether the current definition can be found after @t{DOES>}:
2835: No.
2836:
2837: @end table
2838:
2839: @c ---------------------------------------------------------------------
2840: @node core-ambcond, core-other, core-idef, The Core Words
2841: @subsection Ambiguous conditions
2842: @c ---------------------------------------------------------------------
2843:
2844: @table @i
2845:
2846: @item a name is neither a word nor a number:
2847: @code{-13 throw} (Undefined word). Actually, @code{-13 bounce}, which
2848: preserves the data and FP stack, so you don't lose more work than
2849: necessary.
2850:
2851: @item a definition name exceeds the maximum length allowed:
2852: @code{-19 throw} (Word name too long)
2853:
2854: @item addressing a region not inside the various data spaces of the forth system:
2855: The stacks, code space and name space are accessible. Machine code space is
2856: typically readable. Accessing other addresses gives results dependent on
2857: the operating system. On decent systems: @code{-9 throw} (Invalid memory
2858: address).
2859:
2860: @item argument type incompatible with parameter:
2861: This is usually not caught. Some words perform checks, e.g., the control
2862: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
2863: mismatch).
2864:
2865: @item attempting to obtain the execution token of a word with undefined execution semantics:
2866: @code{-14 throw} (Interpreting a compile-only word). In some cases, you
2867: get an execution token for @code{compile-only-error} (which performs a
2868: @code{-14 throw} when executed).
2869:
2870: @item dividing by zero:
2871: typically results in a @code{-55 throw} (floating point unidentified
2872: fault), although a @code{-10 throw} (divide by zero) would be more
2873: appropriate.
2874:
2875: @item insufficient data stack or return stack space:
2876: Not checked. This typically results in mysterious illegal memory
2877: accesses, producing @code{-9 throw} (Invalid memory address) or
2878: @code{-23 throw} (Address alignment exception).
2879:
2880: @item insufficient space for loop control parameters:
2881: like other return stack overflows.
2882:
2883: @item insufficient space in the dictionary:
2884: Not checked. Similar results as stack overflows. However, typically the
2885: error appears at a different place when one inserts or removes code.
2886:
2887: @item interpreting a word with undefined interpretation semantics:
2888: For some words, we defined interpretation semantics. For the others:
2889: @code{-14 throw} (Interpreting a compile-only word).
2890:
2891: @item modifying the contents of the input buffer or a string literal:
2892: These are located in writable memory and can be modified.
2893:
2894: @item overflow of the pictured numeric output string:
2895: Not checked.
2896:
2897: @item parsed string overflow:
2898: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
2899:
2900: @item producing a result out of range:
2901: On two's complement machines, arithmetic is performed modulo
2902: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
2903: arithmetic (with appropriate mapping for signed types). Division by zero
2904: typically results in a @code{-55 throw} (floatingpoint unidentified
2905: fault), although a @code{-10 throw} (divide by zero) would be more
2906: appropriate. @code{convert} and @code{>number} currently overflow
2907: silently.
2908:
2909: @item reading from an empty data or return stack:
2910: The data stack is checked by the outer (aka text) interpreter after
2911: every word executed. If it has underflowed, a @code{-4 throw} (Stack
2912: underflow) is performed. Apart from that, the stacks are not checked and
2913: underflows can result in similar behaviour as overflows (of adjacent
2914: stacks).
2915:
2916: @item unexpected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
2917: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
2918: use zero-length string as a name). Words like @code{'} probably will not
2919: find what they search. Note that it is possible to create zero-length
2920: names with @code{nextname} (should it not?).
2921:
2922: @item @code{>IN} greater than input buffer:
2923: The next invocation of a parsing word returns a string wih length 0.
2924:
2925: @item @code{RECURSE} appears after @code{DOES>}:
2926: Compiles a recursive call to the defining word, not to the defined word.
2927:
2928: @item argument input source different than current input source for @code{RESTORE-INPUT}:
2929: @code{-12 THROW}. Note that, once an input file is closed (e.g., because
2930: the end of the file was reached), its source-id may be
2931: reused. Therefore, restoring an input source specification referencing a
2932: closed file may lead to unpredictable results instead of a @code{-12
2933: THROW}.
2934:
2935: In the future, Gforth may be able to restore input source specifications
2936: from other than the current input soruce.
2937:
2938: @item data space containing definitions gets de-allocated:
2939: Deallocation with @code{allot} is not checked. This typically resuls in
2940: memory access faults or execution of illegal instructions.
2941:
2942: @item data space read/write with incorrect alignment:
2943: Processor-dependent. Typically results in a @code{-23 throw} (Address
2944: alignment exception). Under Linux on a 486 or later processor with
2945: alignment turned on, incorrect alignment results in a @code{-9 throw}
2946: (Invalid memory address). There are reportedly some processors with
2947: alignment restrictions that do not report them.
2948:
2949: @item data space pointer not properly aligned, @code{,}, @code{C,}:
2950: Like other alignment errors.
2951:
2952: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
2953: Not checked. May cause an illegal memory access.
2954:
2955: @item loop control parameters not available:
2956: Not checked. The counted loop words simply assume that the top of return
2957: stack items are loop control parameters and behave accordingly.
2958:
2959: @item most recent definition does not have a name (@code{IMMEDIATE}):
2960: @code{abort" last word was headerless"}.
2961:
2962: @item name not defined by @code{VALUE} used by @code{TO}:
2963: @code{-32 throw} (Invalid name argument) (unless name was defined by
2964: @code{CONSTANT}; then it just changes the constant).
2965:
2966: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
2967: @code{-13 throw} (Undefined word)
2968:
2969: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
2970: Gforth behaves as if they were of the same type. I.e., you can predict
2971: the behaviour by interpreting all parameters as, e.g., signed.
2972:
2973: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
2974: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} performs the
2975: compilation semantics of @code{TO}.
2976:
2977: @item String longer than a counted string returned by @code{WORD}:
2978: Not checked. The string will be ok, but the count will, of course,
2979: contain only the least significant bits of the length.
2980:
2981: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
2982: Processor-dependent. Typical behaviours are returning 0 and using only
2983: the low bits of the shift count.
2984:
2985: @item word not defined via @code{CREATE}:
2986: @code{>BODY} produces the PFA of the word no matter how it was defined.
2987:
2988: @code{DOES>} changes the execution semantics of the last defined word no
2989: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
2990: @code{CREATE , DOES>}.
2991:
2992: @item words improperly used outside @code{<#} and @code{#>}:
2993: Not checked. As usual, you can expect memory faults.
2994:
2995: @end table
2996:
2997:
2998: @c ---------------------------------------------------------------------
2999: @node core-other, , core-ambcond, The Core Words
3000: @subsection Other system documentation
3001: @c ---------------------------------------------------------------------
3002:
3003: @table @i
3004:
3005: @item nonstandard words using @code{PAD}:
3006: None.
3007:
3008: @item operator's terminal facilities available:
3009: After processing the command line, Gforth goes into interactive mode,
3010: and you can give commands to Gforth interactively. The actual facilities
3011: available depend on how you invoke Gforth.
3012:
3013: @item program data space available:
3014: @code{sp@@ here - .} gives the space remaining for dictionary and data
3015: stack together.
3016:
3017: @item return stack space available:
3018: By default 16 KBytes. The default can be overridden with the @code{-r}
3019: switch (@pxref{Invocation}) when Gforth starts up.
3020:
3021: @item stack space available:
3022: @code{sp@@ here - .} gives the space remaining for dictionary and data
3023: stack together.
3024:
3025: @item system dictionary space required, in address units:
3026: Type @code{here forthstart - .} after startup. At the time of this
3027: writing, this gives 70108 (bytes) on a 32-bit system.
3028: @end table
3029:
3030:
3031: @c =====================================================================
3032: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
3033: @section The optional Block word set
3034: @c =====================================================================
3035:
3036: @menu
3037: * block-idef:: Implementation Defined Options
3038: * block-ambcond:: Ambiguous Conditions
3039: * block-other:: Other System Documentation
3040: @end menu
3041:
3042:
3043: @c ---------------------------------------------------------------------
3044: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
3045: @subsection Implementation Defined Options
3046: @c ---------------------------------------------------------------------
3047:
3048: @table @i
3049:
3050: @item the format for display by @code{LIST}:
3051: First the screen number is displayed, then 16 lines of 64 characters,
3052: each line preceded by the line number.
3053:
3054: @item the length of a line affected by @code{\}:
3055: 64 characters.
3056: @end table
3057:
3058:
3059: @c ---------------------------------------------------------------------
3060: @node block-ambcond, block-other, block-idef, The optional Block word set
3061: @subsection Ambiguous conditions
3062: @c ---------------------------------------------------------------------
3063:
3064: @table @i
3065:
3066: @item correct block read was not possible:
3067: Typically results in a @code{throw} of some OS-derived value (between
3068: -512 and -2048). If the blocks file was just not long enough, blanks are
3069: supplied for the missing portion.
3070:
3071: @item I/O exception in block transfer:
3072: Typically results in a @code{throw} of some OS-derived value (between
3073: -512 and -2048).
3074:
3075: @item invalid block number:
3076: @code{-35 throw} (Invalid block number)
3077:
3078: @item a program directly alters the contents of @code{BLK}:
3079: The input stream is switched to that other block, at the same
3080: position. If the storing to @code{BLK} happens when interpreting
3081: non-block input, the system will get quite confused when the block ends.
3082:
3083: @item no current block buffer for @code{UPDATE}:
3084: @code{UPDATE} has no effect.
3085:
3086: @end table
3087:
3088:
3089: @c ---------------------------------------------------------------------
3090: @node block-other, , block-ambcond, The optional Block word set
3091: @subsection Other system documentation
3092: @c ---------------------------------------------------------------------
3093:
3094: @table @i
3095:
3096: @item any restrictions a multiprogramming system places on the use of buffer addresses:
3097: No restrictions (yet).
3098:
3099: @item the number of blocks available for source and data:
3100: depends on your disk space.
3101:
3102: @end table
3103:
3104:
3105: @c =====================================================================
3106: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
3107: @section The optional Double Number word set
3108: @c =====================================================================
3109:
3110: @menu
3111: * double-ambcond:: Ambiguous Conditions
3112: @end menu
3113:
3114:
3115: @c ---------------------------------------------------------------------
3116: @node double-ambcond, , The optional Double Number word set, The optional Double Number word set
3117: @subsection Ambiguous conditions
3118: @c ---------------------------------------------------------------------
3119:
3120: @table @i
3121:
3122: @item @var{d} outside of range of @var{n} in @code{D>S}:
3123: The least significant cell of @var{d} is produced.
3124:
3125: @end table
3126:
3127:
3128: @c =====================================================================
3129: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
3130: @section The optional Exception word set
3131: @c =====================================================================
3132:
3133: @menu
3134: * exception-idef:: Implementation Defined Options
3135: @end menu
3136:
3137:
3138: @c ---------------------------------------------------------------------
3139: @node exception-idef, , The optional Exception word set, The optional Exception word set
3140: @subsection Implementation Defined Options
3141: @c ---------------------------------------------------------------------
3142:
3143: @table @i
3144: @item @code{THROW}-codes used in the system:
3145: The codes -256@minus{}-511 are used for reporting signals (see
3146: @file{errore.fs}). The codes -512@minus{}-2047 are used for OS errors
3147: (for file and memory allocation operations). The mapping from OS error
3148: numbers to throw code is -512@minus{}@var{errno}. One side effect of
3149: this mapping is that undefined OS errors produce a message with a
3150: strange number; e.g., @code{-1000 THROW} results in @code{Unknown error
3151: 488} on my system.
3152: @end table
3153:
3154: @c =====================================================================
3155: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
3156: @section The optional Facility word set
3157: @c =====================================================================
3158:
3159: @menu
3160: * facility-idef:: Implementation Defined Options
3161: * facility-ambcond:: Ambiguous Conditions
3162: @end menu
3163:
3164:
3165: @c ---------------------------------------------------------------------
3166: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
3167: @subsection Implementation Defined Options
3168: @c ---------------------------------------------------------------------
3169:
3170: @table @i
3171:
3172: @item encoding of keyboard events (@code{EKEY}):
3173: Not yet implemeted.
3174:
3175: @item duration of a system clock tick
3176: System dependent. With respect to @code{MS}, the time is specified in
3177: microseconds. How well the OS and the hardware implement this, is
3178: another question.
3179:
3180: @item repeatability to be expected from the execution of @code{MS}:
3181: System dependent. On Unix, a lot depends on load. If the system is
3182: lightly loaded, and the delay is short enough that Gforth does not get
3183: swapped out, the performance should be acceptable. Under MS-DOS and
3184: other single-tasking systems, it should be good.
3185:
3186: @end table
3187:
3188:
3189: @c ---------------------------------------------------------------------
3190: @node facility-ambcond, , facility-idef, The optional Facility word set
3191: @subsection Ambiguous conditions
3192: @c ---------------------------------------------------------------------
3193:
3194: @table @i
3195:
3196: @item @code{AT-XY} can't be performed on user output device:
3197: Largely terminal dependant. No range checks are done on the arguments.
3198: No errors are reported. You may see some garbage appearing, you may see
3199: simply nothing happen.
3200:
3201: @end table
3202:
3203:
3204: @c =====================================================================
3205: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
3206: @section The optional File-Access word set
3207: @c =====================================================================
3208:
3209: @menu
3210: * file-idef:: Implementation Defined Options
3211: * file-ambcond:: Ambiguous Conditions
3212: @end menu
3213:
3214:
3215: @c ---------------------------------------------------------------------
3216: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
3217: @subsection Implementation Defined Options
3218: @c ---------------------------------------------------------------------
3219:
3220: @table @i
3221:
3222: @item File access methods used:
3223: @code{R/O}, @code{R/W} and @code{BIN} work as you would
3224: expect. @code{W/O} translates into the C file opening mode @code{w} (or
3225: @code{wb}): The file is cleared, if it exists, and created, if it does
3226: not (both with @code{open-file} and @code{create-file}). Under Unix
3227: @code{create-file} creates a file with 666 permissions modified by your
3228: umask.
3229:
3230: @item file exceptions:
3231: The file words do not raise exceptions (except, perhaps, memory access
3232: faults when you pass illegal addresses or file-ids).
3233:
3234: @item file line terminator:
3235: System-dependent. Gforth uses C's newline character as line
3236: terminator. What the actual character code(s) of this are is
3237: system-dependent.
3238:
3239: @item file name format
3240: System dependent. Gforth just uses the file name format of your OS.
3241:
3242: @item information returned by @code{FILE-STATUS}:
3243: @code{FILE-STATUS} returns the most powerful file access mode allowed
3244: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
3245: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
3246: along with the retured mode.
3247:
3248: @item input file state after an exception when including source:
3249: All files that are left via the exception are closed.
3250:
3251: @item @var{ior} values and meaning:
3252: The @var{ior}s returned by the file and memory allocation words are
3253: intended as throw codes. They typically are in the range
3254: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
3255: @var{ior}s is -512@minus{}@var{errno}.
3256:
3257: @item maximum depth of file input nesting:
3258: limited by the amount of return stack, locals/TIB stack, and the number
3259: of open files available. This should not give you troubles.
3260:
3261: @item maximum size of input line:
3262: @code{/line}. Currently 255.
3263:
3264: @item methods of mapping block ranges to files:
3265: Currently, the block words automatically access the file
3266: @file{blocks.fb} in the currend working directory. More sophisticated
3267: methods could be implemented if there is demand (and a volunteer).
3268:
3269: @item number of string buffers provided by @code{S"}:
3270: 1
3271:
3272: @item size of string buffer used by @code{S"}:
3273: @code{/line}. currently 255.
3274:
3275: @end table
3276:
3277: @c ---------------------------------------------------------------------
3278: @node file-ambcond, , file-idef, The optional File-Access word set
3279: @subsection Ambiguous conditions
3280: @c ---------------------------------------------------------------------
3281:
3282: @table @i
3283:
3284: @item attempting to position a file outside it's boundaries:
3285: @code{REPOSITION-FILE} is performed as usual: Afterwards,
3286: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
3287:
3288: @item attempting to read from file positions not yet written:
3289: End-of-file, i.e., zero characters are read and no error is reported.
3290:
3291: @item @var{file-id} is invalid (@code{INCLUDE-FILE}):
3292: An appropriate exception may be thrown, but a memory fault or other
3293: problem is more probable.
3294:
3295: @item I/O exception reading or closing @var{file-id} (@code{include-file}, @code{included}):
3296: The @var{ior} produced by the operation, that discovered the problem, is
3297: thrown.
3298:
3299: @item named file cannot be opened (@code{included}):
3300: The @var{ior} produced by @code{open-file} is thrown.
3301:
3302: @item requesting an unmapped block number:
3303: There are no unmapped legal block numbers. On some operating systems,
3304: writing a block with a large number may overflow the file system and
3305: have an error message as consequence.
3306:
3307: @item using @code{source-id} when @code{blk} is non-zero:
3308: @code{source-id} performs its function. Typically it will give the id of
3309: the source which loaded the block. (Better ideas?)
3310:
3311: @end table
3312:
3313:
3314: @c =====================================================================
3315: @node The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
3316: @section The optional Floating-Point word set
3317: @c =====================================================================
3318:
3319: @menu
3320: * floating-idef:: Implementation Defined Options
3321: * floating-ambcond:: Ambiguous Conditions
3322: @end menu
3323:
3324:
3325: @c ---------------------------------------------------------------------
3326: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
3327: @subsection Implementation Defined Options
3328: @c ---------------------------------------------------------------------
3329:
3330: @table @i
3331:
3332: @item format and range of floating point numbers:
3333: System-dependent; the @code{double} type of C.
3334:
3335: @item results of @code{REPRESENT} when @var{float} is out of range:
3336: System dependent; @code{REPRESENT} is implemented using the C library
3337: function @code{ecvt()} and inherits its behaviour in this respect.
3338:
3339: @item rounding or truncation of floating-point numbers:
3340: System dependent; the rounding behaviour is inherited from the hosting C
3341: compiler. IEEE-FP-based (i.e., most) systems by default round to
3342: nearest, and break ties by rounding to even (i.e., such that the last
3343: bit of the mantissa is 0).
3344:
3345: @item size of floating-point stack:
3346: @code{s" FLOATING-STACK" environment? drop .}. Can be changed at startup
3347: with the command-line option @code{-f}.
3348:
3349: @item width of floating-point stack:
3350: @code{1 floats}.
3351:
3352: @end table
3353:
3354:
3355: @c ---------------------------------------------------------------------
3356: @node floating-ambcond, , floating-idef, The optional Floating-Point word set
3357: @subsection Ambiguous conditions
3358: @c ---------------------------------------------------------------------
3359:
3360: @table @i
3361:
3362: @item @code{df@@} or @code{df!} used with an address that is not double-float aligned:
3363: System-dependent. Typically results in an alignment fault like other
3364: alignment violations.
3365:
3366: @item @code{f@@} or @code{f!} used with an address that is not float aligned:
3367: System-dependent. Typically results in an alignment fault like other
3368: alignment violations.
3369:
3370: @item Floating-point result out of range:
3371: System-dependent. Can result in a @code{-55 THROW} (Floating-point
3372: unidentified fault), or can produce a special value representing, e.g.,
3373: Infinity.
3374:
3375: @item @code{sf@@} or @code{sf!} used with an address that is not single-float aligned:
3376: System-dependent. Typically results in an alignment fault like other
3377: alignment violations.
3378:
3379: @item BASE is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
3380: The floating-point number is converted into decimal nonetheless.
3381:
3382: @item Both arguments are equal to zero (@code{FATAN2}):
3383: System-dependent. @code{FATAN2} is implemented using the C library
3384: function @code{atan2()}.
3385:
3386: @item Using ftan on an argument @var{r1} where cos(@var{r1}) is zero:
3387: System-dependent. Anyway, typically the cos of @var{r1} will not be zero
3388: because of small errors and the tan will be a very large (or very small)
3389: but finite number.
3390:
3391: @item @var{d} cannot be presented precisely as a float in @code{D>F}:
3392: The result is rounded to the nearest float.
3393:
3394: @item dividing by zero:
3395: @code{-55 throw} (Floating-point unidentified fault)
3396:
3397: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
3398: System dependent. On IEEE-FP based systems the number is converted into
3399: an infinity.
3400:
3401: @item @var{float}<1 (@code{facosh}):
3402: @code{-55 throw} (Floating-point unidentified fault)
3403:
3404: @item @var{float}=<-1 (@code{flnp1}):
3405: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
3406: negative infinity is typically produced for @var{float}=-1.
3407:
3408: @item @var{float}=<0 (@code{fln}, @code{flog}):
3409: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
3410: negative infinity is typically produced for @var{float}=0.
3411:
3412: @item @var{float}<0 (@code{fasinh}, @code{fsqrt}):
3413: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
3414: produces values for these inputs on my Linux box (Bug in the C library?)
3415:
3416: @item |@var{float}|>1 (@code{facos}, @code{fasin}, @code{fatanh}):
3417: @code{-55 throw} (Floating-point unidentified fault).
3418:
3419: @item integer part of float cannot be represented by @var{d} in @code{f>d}:
3420: @code{-55 throw} (Floating-point unidentified fault).
3421:
3422: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
3423: This does not happen.
3424: @end table
3425:
3426:
3427:
3428: @c =====================================================================
3429: @node The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
3430: @section The optional Locals word set
3431: @c =====================================================================
3432:
3433: @menu
3434: * locals-idef:: Implementation Defined Options
3435: * locals-ambcond:: Ambiguous Conditions
3436: @end menu
3437:
3438:
3439: @c ---------------------------------------------------------------------
3440: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
3441: @subsection Implementation Defined Options
3442: @c ---------------------------------------------------------------------
3443:
3444: @table @i
3445:
3446: @item maximum number of locals in a definition:
3447: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
3448: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
3449: characters. The number of locals in a definition is bounded by the size
3450: of locals-buffer, which contains the names of the locals.
3451:
3452: @end table
3453:
3454:
3455: @c ---------------------------------------------------------------------
3456: @node locals-ambcond, , locals-idef, The optional Locals word set
3457: @subsection Ambiguous conditions
3458: @c ---------------------------------------------------------------------
3459:
3460: @table @i
3461:
3462: @item executing a named local in interpretation state:
3463: @code{-14 throw} (Interpreting a compile-only word).
3464:
3465: @item @var{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
3466: @code{-32 throw} (Invalid name argument)
3467:
3468: @end table
3469:
3470:
3471: @c =====================================================================
3472: @node The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
3473: @section The optional Memory-Allocation word set
3474: @c =====================================================================
3475:
3476: @menu
3477: * memory-idef:: Implementation Defined Options
3478: @end menu
3479:
3480:
3481: @c ---------------------------------------------------------------------
3482: @node memory-idef, , The optional Memory-Allocation word set, The optional Memory-Allocation word set
3483: @subsection Implementation Defined Options
3484: @c ---------------------------------------------------------------------
3485:
3486: @table @i
3487:
3488: @item values and meaning of @var{ior}:
3489: The @var{ior}s returned by the file and memory allocation words are
3490: intended as throw codes. They typically are in the range
3491: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
3492: @var{ior}s is -512@minus{}@var{errno}.
3493:
3494: @end table
3495:
3496: @c =====================================================================
3497: @node The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
3498: @section The optional Programming-Tools word set
3499: @c =====================================================================
3500:
3501: @menu
3502: * programming-idef:: Implementation Defined Options
3503: * programming-ambcond:: Ambiguous Conditions
3504: @end menu
3505:
3506:
3507: @c ---------------------------------------------------------------------
3508: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
3509: @subsection Implementation Defined Options
3510: @c ---------------------------------------------------------------------
3511:
3512: @table @i
3513:
3514: @item ending sequence for input following @code{;code} and @code{code}:
3515: Not implemented (yet).
3516:
3517: @item manner of processing input following @code{;code} and @code{code}:
3518: Not implemented (yet).
3519:
3520: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
3521: Not implemented (yet). If they were implemented, they would use the
3522: search order wordset.
3523:
3524: @item source and format of display by @code{SEE}:
3525: The source for @code{see} is the intermediate code used by the inner
3526: interpreter. The current @code{see} tries to output Forth source code
3527: as well as possible.
3528:
3529: @end table
3530:
3531: @c ---------------------------------------------------------------------
3532: @node programming-ambcond, , programming-idef, The optional Programming-Tools word set
3533: @subsection Ambiguous conditions
3534: @c ---------------------------------------------------------------------
3535:
3536: @table @i
3537:
3538: @item deleting the compilation wordlist (@code{FORGET}):
3539: Not implemented (yet).
3540:
3541: @item fewer than @var{u}+1 items on the control flow stack (@code{CS-PICK}, @code{CS-ROLL}):
3542: This typically results in an @code{abort"} with a descriptive error
3543: message (may change into a @code{-22 throw} (Control structure mismatch)
3544: in the future). You may also get a memory access error. If you are
3545: unlucky, this ambiguous condition is not caught.
3546:
3547: @item @var{name} can't be found (@code{forget}):
3548: Not implemented (yet).
3549:
3550: @item @var{name} not defined via @code{CREATE}:
3551: @code{;code} is not implemented (yet). If it were, it would behave like
3552: @code{DOES>} in this respect, i.e., change the execution semantics of
3553: the last defined word no matter how it was defined.
3554:
3555: @item @code{POSTPONE} applied to @code{[IF]}:
3556: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
3557: equivalent to @code{[IF]}.
3558:
3559: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
3560: Continue in the same state of conditional compilation in the next outer
3561: input source. Currently there is no warning to the user about this.
3562:
3563: @item removing a needed definition (@code{FORGET}):
3564: Not implemented (yet).
3565:
3566: @end table
3567:
3568:
3569: @c =====================================================================
3570: @node The optional Search-Order word set, , The optional Programming-Tools word set, ANS conformance
3571: @section The optional Search-Order word set
3572: @c =====================================================================
3573:
3574: @menu
3575: * search-idef:: Implementation Defined Options
3576: * search-ambcond:: Ambiguous Conditions
3577: @end menu
3578:
3579:
3580: @c ---------------------------------------------------------------------
3581: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
3582: @subsection Implementation Defined Options
3583: @c ---------------------------------------------------------------------
3584:
3585: @table @i
3586:
3587: @item maximum number of word lists in search order:
3588: @code{s" wordlists" environment? drop .}. Currently 16.
3589:
3590: @item minimum search order:
3591: @code{root root}.
3592:
3593: @end table
3594:
3595: @c ---------------------------------------------------------------------
3596: @node search-ambcond, , search-idef, The optional Search-Order word set
3597: @subsection Ambiguous conditions
3598: @c ---------------------------------------------------------------------
3599:
3600: @table @i
3601:
3602: @item changing the compilation wordlist (during compilation):
3603: The word is entered into the wordlist that was the compilation wordlist
3604: at the start of the definition. Any changes to the name field (e.g.,
3605: @code{immediate}) or the code field (e.g., when executing @code{DOES>})
3606: are applied to the latest defined word (as reported by @code{last} or
3607: @code{lastxt}), if possible, irrespective of the compilation wordlist.
3608:
3609: @item search order empty (@code{previous}):
3610: @code{abort" Vocstack empty"}.
3611:
3612: @item too many word lists in search order (@code{also}):
3613: @code{abort" Vocstack full"}.
3614:
3615: @end table
3616:
3617: @node Model, Integrating Gforth, ANS conformance, Top
3618: @chapter Model
3619:
3620: This chapter has yet to be written. It will contain information, on
3621: which internal structures you can rely.
3622:
3623: @node Integrating Gforth, Emacs and Gforth, Model, Top
3624: @chapter Integrating Gforth into C programs
3625:
3626: This is not yet implemented.
3627:
3628: Several people like to use Forth as scripting language for applications
3629: that are otherwise written in C, C++, or some other language.
3630:
3631: The Forth system ATLAST provides facilities for embedding it into
3632: applications; unfortunately it has several disadvantages: most
3633: importantly, it is not based on ANS Forth, and it is apparently dead
3634: (i.e., not developed further and not supported). The facilities
3635: provided by Gforth in this area are inspired by ATLASTs facilities, so
3636: making the switch should not be hard.
3637:
3638: We also tried to design the interface such that it can easily be
3639: implemented by other Forth systems, so that we may one day arrive at a
3640: standardized interface. Such a standard interface would allow you to
3641: replace the Forth system without having to rewrite C code.
3642:
3643: You embed the Gforth interpreter by linking with the library
3644: @code{libgforth.a} (give the compiler the option @code{-lgforth}). All
3645: global symbols in this library that belong to the interface, have the
3646: prefix @code{forth_}. (Global symbols that are used internally have the
3647: prefix @code{gforth_}).
3648:
3649: You can include the declarations of Forth types and the functions and
3650: variables of the interface with @code{#include <forth.h>}.
3651:
3652: Types.
3653:
3654: Variables.
3655:
3656: Data and FP Stack pointer. Area sizes.
3657:
3658: functions.
3659:
3660: forth_init(imagefile)
3661: forth_evaluate(string) exceptions?
3662: forth_goto(address) (or forth_execute(xt)?)
3663: forth_continue() (a corountining mechanism)
3664:
3665: Adding primitives.
3666:
3667: No checking.
3668:
3669: Signals?
3670:
3671: Accessing the Stacks
3672:
3673: @node Emacs and Gforth, Internals, Integrating Gforth, Top
3674: @chapter Emacs and Gforth
3675:
3676: Gforth comes with @file{gforth.el}, an improved version of
3677: @file{forth.el} by Goran Rydqvist (included in the TILE package). The
3678: improvements are a better (but still not perfect) handling of
3679: indentation. I have also added comment paragraph filling (@kbd{M-q}),
3680: commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) regions and
3681: removing debugging tracers (@kbd{C-x ~}, @pxref{Debugging}). I left the
3682: stuff I do not use alone, even though some of it only makes sense for
3683: TILE. To get a description of these features, enter Forth mode and type
3684: @kbd{C-h m}.
3685:
3686: In addition, Gforth supports Emacs quite well: The source code locations
3687: given in error messages, debugging output (from @code{~~}) and failed
3688: assertion messages are in the right format for Emacs' compilation mode
3689: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
3690: Manual}) so the source location corresponding to an error or other
3691: message is only a few keystrokes away (@kbd{C-x `} for the next error,
3692: @kbd{C-c C-c} for the error under the cursor).
3693:
3694: Also, if you @code{include} @file{etags.fs}, a new @file{TAGS} file
3695: (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) will be produced that
3696: contains the definitions of all words defined afterwards. You can then
3697: find the source for a word using @kbd{M-.}. Note that emacs can use
3698: several tags files at the same time (e.g., one for the Gforth sources
3699: and one for your program, @pxref{Select Tags Table,,Selecting a Tags
3700: Table,emacs, Emacs Manual}). The TAGS file for the preloaded words is
3701: @file{$(datadir)/gforth/$(VERSION)/TAGS} (e.g.,
3702: @file{/usr/local/share/gforth/0.2.0/TAGS}).
3703:
3704: To get all these benefits, add the following lines to your @file{.emacs}
3705: file:
3706:
3707: @example
3708: (autoload 'forth-mode "gforth.el")
3709: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
3710: @end example
3711:
3712: @node Internals, Bugs, Emacs and Gforth, Top
3713: @chapter Internals
3714:
3715: Reading this section is not necessary for programming with Gforth. It
3716: should be helpful for finding your way in the Gforth sources.
3717:
3718: The ideas in this section have also been published in the papers
3719: @cite{ANS fig/GNU/??? Forth} (in German) by Bernd Paysan, presented at
3720: the Forth-Tagung '93 and @cite{A Portable Forth Engine} by M. Anton
3721: Ertl, presented at EuroForth '93; the latter is available at
3722: @*@file{http://www.complang.tuwien.ac.at/papers/ertl93.ps.Z}.
3723:
3724: @menu
3725: * Portability::
3726: * Threading::
3727: * Primitives::
3728: * System Architecture::
3729: * Performance::
3730: @end menu
3731:
3732: @node Portability, Threading, Internals, Internals
3733: @section Portability
3734:
3735: One of the main goals of the effort is availability across a wide range
3736: of personal machines. fig-Forth, and, to a lesser extent, F83, achieved
3737: this goal by manually coding the engine in assembly language for several
3738: then-popular processors. This approach is very labor-intensive and the
3739: results are short-lived due to progress in computer architecture.
3740:
3741: Others have avoided this problem by coding in C, e.g., Mitch Bradley
3742: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
3743: particularly popular for UNIX-based Forths due to the large variety of
3744: architectures of UNIX machines. Unfortunately an implementation in C
3745: does not mix well with the goals of efficiency and with using
3746: traditional techniques: Indirect or direct threading cannot be expressed
3747: in C, and switch threading, the fastest technique available in C, is
3748: significantly slower. Another problem with C is that it's very
3749: cumbersome to express double integer arithmetic.
3750:
3751: Fortunately, there is a portable language that does not have these
3752: limitations: GNU C, the version of C processed by the GNU C compiler
3753: (@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
3754: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
3755: Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
3756: threading possible, its @code{long long} type (@pxref{Long Long, ,
3757: Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forth's
3758: double numbers@footnote{Unfortunately, long longs are not implemented
3759: properly on all machines (e.g., on alpha-osf1, long longs are only 64
3760: bits, the same size as longs (and pointers), but they should be twice as
3761: long according to @ref{Long Long, , Double-Word Integers, gcc.info, GNU
3762: C Manual}). So, we had to implement doubles in C after all. Still, on
3763: most machines we can use long longs and achieve better performance than
3764: with the emulation package.}. GNU C is available for free on all
3765: important (and many unimportant) UNIX machines, VMS, 80386s running
3766: MS-DOS, the Amiga, and the Atari ST, so a Forth written in GNU C can run
3767: on all these machines.
3768:
3769: Writing in a portable language has the reputation of producing code that
3770: is slower than assembly. For our Forth engine we repeatedly looked at
3771: the code produced by the compiler and eliminated most compiler-induced
3772: inefficiencies by appropriate changes in the source-code.
3773:
3774: However, register allocation cannot be portably influenced by the
3775: programmer, leading to some inefficiencies on register-starved
3776: machines. We use explicit register declarations (@pxref{Explicit Reg
3777: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
3778: improve the speed on some machines. They are turned on by using the
3779: @code{gcc} switch @code{-DFORCE_REG}. Unfortunately, this feature not
3780: only depends on the machine, but also on the compiler version: On some
3781: machines some compiler versions produce incorrect code when certain
3782: explicit register declarations are used. So by default
3783: @code{-DFORCE_REG} is not used.
3784:
3785: @node Threading, Primitives, Portability, Internals
3786: @section Threading
3787:
3788: GNU C's labels as values extension (available since @code{gcc-2.0},
3789: @pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
3790: makes it possible to take the address of @var{label} by writing
3791: @code{&&@var{label}}. This address can then be used in a statement like
3792: @code{goto *@var{address}}. I.e., @code{goto *&&x} is the same as
3793: @code{goto x}.
3794:
3795: With this feature an indirect threaded NEXT looks like:
3796: @example
3797: cfa = *ip++;
3798: ca = *cfa;
3799: goto *ca;
3800: @end example
3801: For those unfamiliar with the names: @code{ip} is the Forth instruction
3802: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
3803: execution token and points to the code field of the next word to be
3804: executed; The @code{ca} (code address) fetched from there points to some
3805: executable code, e.g., a primitive or the colon definition handler
3806: @code{docol}.
3807:
3808: Direct threading is even simpler:
3809: @example
3810: ca = *ip++;
3811: goto *ca;
3812: @end example
3813:
3814: Of course we have packaged the whole thing neatly in macros called
3815: @code{NEXT} and @code{NEXT1} (the part of NEXT after fetching the cfa).
3816:
3817: @menu
3818: * Scheduling::
3819: * Direct or Indirect Threaded?::
3820: * DOES>::
3821: @end menu
3822:
3823: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
3824: @subsection Scheduling
3825:
3826: There is a little complication: Pipelined and superscalar processors,
3827: i.e., RISC and some modern CISC machines can process independent
3828: instructions while waiting for the results of an instruction. The
3829: compiler usually reorders (schedules) the instructions in a way that
3830: achieves good usage of these delay slots. However, on our first tries
3831: the compiler did not do well on scheduling primitives. E.g., for
3832: @code{+} implemented as
3833: @example
3834: n=sp[0]+sp[1];
3835: sp++;
3836: sp[0]=n;
3837: NEXT;
3838: @end example
3839: the NEXT comes strictly after the other code, i.e., there is nearly no
3840: scheduling. After a little thought the problem becomes clear: The
3841: compiler cannot know that sp and ip point to different addresses (and
3842: the version of @code{gcc} we used would not know it even if it was
3843: possible), so it could not move the load of the cfa above the store to
3844: the TOS. Indeed the pointers could be the same, if code on or very near
3845: the top of stack were executed. In the interest of speed we chose to
3846: forbid this probably unused ``feature'' and helped the compiler in
3847: scheduling: NEXT is divided into the loading part (@code{NEXT_P1}) and
3848: the goto part (@code{NEXT_P2}). @code{+} now looks like:
3849: @example
3850: n=sp[0]+sp[1];
3851: sp++;
3852: NEXT_P1;
3853: sp[0]=n;
3854: NEXT_P2;
3855: @end example
3856: This can be scheduled optimally by the compiler.
3857:
3858: This division can be turned off with the switch @code{-DCISC_NEXT}. This
3859: switch is on by default on machines that do not profit from scheduling
3860: (e.g., the 80386), in order to preserve registers.
3861:
3862: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
3863: @subsection Direct or Indirect Threaded?
3864:
3865: Both! After packaging the nasty details in macro definitions we
3866: realized that we could switch between direct and indirect threading by
3867: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
3868: defining a few machine-specific macros for the direct-threading case.
3869: On the Forth level we also offer access words that hide the
3870: differences between the threading methods (@pxref{Threading Words}).
3871:
3872: Indirect threading is implemented completely
3873: machine-independently. Direct threading needs routines for creating
3874: jumps to the executable code (e.g. to docol or dodoes). These routines
3875: are inherently machine-dependent, but they do not amount to many source
3876: lines. I.e., even porting direct threading to a new machine is a small
3877: effort.
3878:
3879: @node DOES>, , Direct or Indirect Threaded?, Threading
3880: @subsection DOES>
3881: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
3882: the chunk of code executed by every word defined by a
3883: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
3884: the Forth code to be executed, i.e. the code after the @code{DOES>} (the
3885: DOES-code)? There are two solutions:
3886:
3887: In fig-Forth the code field points directly to the dodoes and the
3888: DOES-code address is stored in the cell after the code address
3889: (i.e. at cfa cell+). It may seem that this solution is illegal in the
3890: Forth-79 and all later standards, because in fig-Forth this address
3891: lies in the body (which is illegal in these standards). However, by
3892: making the code field larger for all words this solution becomes legal
3893: again. We use this approach for the indirect threaded version. Leaving
3894: a cell unused in most words is a bit wasteful, but on the machines we
3895: are targetting this is hardly a problem. The other reason for having a
3896: code field size of two cells is to avoid having different image files
3897: for direct and indirect threaded systems (@pxref{System Architecture}).
3898:
3899: The other approach is that the code field points or jumps to the cell
3900: after @code{DOES}. In this variant there is a jump to @code{dodoes} at
3901: this address. @code{dodoes} can then get the DOES-code address by
3902: computing the code address, i.e., the address of the jump to dodoes,
3903: and add the length of that jump field. A variant of this is to have a
3904: call to @code{dodoes} after the @code{DOES>}; then the return address
3905: (which can be found in the return register on RISCs) is the DOES-code
3906: address. Since the two cells available in the code field are usually
3907: used up by the jump to the code address in direct threading, we use
3908: this approach for direct threading. We did not want to add another
3909: cell to the code field.
3910:
3911: @node Primitives, System Architecture, Threading, Internals
3912: @section Primitives
3913:
3914: @menu
3915: * Automatic Generation::
3916: * TOS Optimization::
3917: * Produced code::
3918: @end menu
3919:
3920: @node Automatic Generation, TOS Optimization, Primitives, Primitives
3921: @subsection Automatic Generation
3922:
3923: Since the primitives are implemented in a portable language, there is no
3924: longer any need to minimize the number of primitives. On the contrary,
3925: having many primitives is an advantage: speed. In order to reduce the
3926: number of errors in primitives and to make programming them easier, we
3927: provide a tool, the primitive generator (@file{prims2x.fs}), that
3928: automatically generates most (and sometimes all) of the C code for a
3929: primitive from the stack effect notation. The source for a primitive
3930: has the following form:
3931:
3932: @format
3933: @var{Forth-name} @var{stack-effect} @var{category} [@var{pronounc.}]
3934: [@code{""}@var{glossary entry}@code{""}]
3935: @var{C code}
3936: [@code{:}
3937: @var{Forth code}]
3938: @end format
3939:
3940: The items in brackets are optional. The category and glossary fields
3941: are there for generating the documentation, the Forth code is there
3942: for manual implementations on machines without GNU C. E.g., the source
3943: for the primitive @code{+} is:
3944: @example
3945: + n1 n2 -- n core plus
3946: n = n1+n2;
3947: @end example
3948:
3949: This looks like a specification, but in fact @code{n = n1+n2} is C
3950: code. Our primitive generation tool extracts a lot of information from
3951: the stack effect notations@footnote{We use a one-stack notation, even
3952: though we have separate data and floating-point stacks; The separate
3953: notation can be generated easily from the unified notation.}: The number
3954: of items popped from and pushed on the stack, their type, and by what
3955: name they are referred to in the C code. It then generates a C code
3956: prelude and postlude for each primitive. The final C code for @code{+}
3957: looks like this:
3958:
3959: @example
3960: I_plus: /* + ( n1 n2 -- n ) */ /* label, stack effect */
3961: /* */ /* documentation */
3962: @{
3963: DEF_CA /* definition of variable ca (indirect threading) */
3964: Cell n1; /* definitions of variables */
3965: Cell n2;
3966: Cell n;
3967: n1 = (Cell) sp[1]; /* input */
3968: n2 = (Cell) TOS;
3969: sp += 1; /* stack adjustment */
3970: NAME("+") /* debugging output (with -DDEBUG) */
3971: @{
3972: n = n1+n2; /* C code taken from the source */
3973: @}
3974: NEXT_P1; /* NEXT part 1 */
3975: TOS = (Cell)n; /* output */
3976: NEXT_P2; /* NEXT part 2 */
3977: @}
3978: @end example
3979:
3980: This looks long and inefficient, but the GNU C compiler optimizes quite
3981: well and produces optimal code for @code{+} on, e.g., the R3000 and the
3982: HP RISC machines: Defining the @code{n}s does not produce any code, and
3983: using them as intermediate storage also adds no cost.
3984:
3985: There are also other optimizations, that are not illustrated by this
3986: example: Assignments between simple variables are usually for free (copy
3987: propagation). If one of the stack items is not used by the primitive
3988: (e.g. in @code{drop}), the compiler eliminates the load from the stack
3989: (dead code elimination). On the other hand, there are some things that
3990: the compiler does not do, therefore they are performed by
3991: @file{prims2x.fs}: The compiler does not optimize code away that stores
3992: a stack item to the place where it just came from (e.g., @code{over}).
3993:
3994: While programming a primitive is usually easy, there are a few cases
3995: where the programmer has to take the actions of the generator into
3996: account, most notably @code{?dup}, but also words that do not (always)
3997: fall through to NEXT.
3998:
3999: @node TOS Optimization, Produced code, Automatic Generation, Primitives
4000: @subsection TOS Optimization
4001:
4002: An important optimization for stack machine emulators, e.g., Forth
4003: engines, is keeping one or more of the top stack items in
4004: registers. If a word has the stack effect @var{in1}...@var{inx} @code{--}
4005: @var{out1}...@var{outy}, keeping the top @var{n} items in registers
4006: @itemize @bullet
4007: @item
4008: is better than keeping @var{n-1} items, if @var{x>=n} and @var{y>=n},
4009: due to fewer loads from and stores to the stack.
4010: @item is slower than keeping @var{n-1} items, if @var{x<>y} and @var{x<n} and
4011: @var{y<n}, due to additional moves between registers.
4012: @end itemize
4013:
4014: In particular, keeping one item in a register is never a disadvantage,
4015: if there are enough registers. Keeping two items in registers is a
4016: disadvantage for frequent words like @code{?branch}, constants,
4017: variables, literals and @code{i}. Therefore our generator only produces
4018: code that keeps zero or one items in registers. The generated C code
4019: covers both cases; the selection between these alternatives is made at
4020: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
4021: code for @code{+} is just a simple variable name in the one-item case,
4022: otherwise it is a macro that expands into @code{sp[0]}. Note that the
4023: GNU C compiler tries to keep simple variables like @code{TOS} in
4024: registers, and it usually succeeds, if there are enough registers.
4025:
4026: The primitive generator performs the TOS optimization for the
4027: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
4028: operations the benefit of this optimization is even larger:
4029: floating-point operations take quite long on most processors, but can be
4030: performed in parallel with other operations as long as their results are
4031: not used. If the FP-TOS is kept in a register, this works. If
4032: it is kept on the stack, i.e., in memory, the store into memory has to
4033: wait for the result of the floating-point operation, lengthening the
4034: execution time of the primitive considerably.
4035:
4036: The TOS optimization makes the automatic generation of primitives a
4037: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
4038: @code{TOS} is not sufficient. There are some special cases to
4039: consider:
4040: @itemize @bullet
4041: @item In the case of @code{dup ( w -- w w )} the generator must not
4042: eliminate the store to the original location of the item on the stack,
4043: if the TOS optimization is turned on.
4044: @item Primitives with stack effects of the form @code{--}
4045: @var{out1}...@var{outy} must store the TOS to the stack at the start.
4046: Likewise, primitives with the stack effect @var{in1}...@var{inx} @code{--}
4047: must load the TOS from the stack at the end. But for the null stack
4048: effect @code{--} no stores or loads should be generated.
4049: @end itemize
4050:
4051: @node Produced code, , TOS Optimization, Primitives
4052: @subsection Produced code
4053:
4054: To see what assembly code is produced for the primitives on your machine
4055: with your compiler and your flag settings, type @code{make engine.s} and
4056: look at the resulting file @file{engine.s}.
4057:
4058: @node System Architecture, Performance, Primitives, Internals
4059: @section System Architecture
4060:
4061: Our Forth system consists not only of primitives, but also of
4062: definitions written in Forth. Since the Forth compiler itself belongs
4063: to those definitions, it is not possible to start the system with the
4064: primitives and the Forth source alone. Therefore we provide the Forth
4065: code as an image file in nearly executable form. At the start of the
4066: system a C routine loads the image file into memory, sets up the
4067: memory (stacks etc.) according to information in the image file, and
4068: starts executing Forth code.
4069:
4070: The image file format is a compromise between the goals of making it
4071: easy to generate image files and making them portable. The easiest way
4072: to generate an image file is to just generate a memory dump. However,
4073: this kind of image file cannot be used on a different machine, or on
4074: the next version of the engine on the same machine, it even might not
4075: work with the same engine compiled by a different version of the C
4076: compiler. We would like to have as few versions of the image file as
4077: possible, because we do not want to distribute many versions of the
4078: same image file, and to make it easy for the users to use their image
4079: files on many machines. We currently need to create a different image
4080: file for machines with different cell sizes and different byte order
4081: (little- or big-endian)@footnote{We are considering adding information to the
4082: image file that enables the loader to change the byte order.}.
4083:
4084: Forth code that is going to end up in a portable image file has to
4085: comply to some restrictions: addresses have to be stored in memory with
4086: special words (@code{A!}, @code{A,}, etc.) in order to make the code
4087: relocatable. Cells, floats, etc., have to be stored at the natural
4088: alignment boundaries@footnote{E.g., store floats (8 bytes) at an address
4089: dividable by~8. This happens automatically in our system when you use
4090: the ANS Forth alignment words.}, in order to avoid alignment faults on
4091: machines with stricter alignment. The image file is produced by a
4092: metacompiler (@file{cross.fs}).
4093:
4094: So, unlike the image file of Mitch Bradleys @code{cforth}, our image
4095: file is not directly executable, but has to undergo some manipulations
4096: during loading. Address relocation is performed at image load-time, not
4097: at run-time. The loader also has to replace tokens standing for
4098: primitive calls with the appropriate code-field addresses (or code
4099: addresses in the case of direct threading).
4100:
4101: @node Performance, , System Architecture, Internals
4102: @section Performance
4103:
4104: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
4105: impossible to write a significantly faster engine.
4106:
4107: On register-starved machines like the 386 architecture processors
4108: improvements are possible, because @code{gcc} does not utilize the
4109: registers as well as a human, even with explicit register declarations;
4110: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
4111: and hand-tuned it for the 486; this system is 1.19 times faster on the
4112: Sieve benchmark on a 486DX2/66 than Gforth compiled with
4113: @code{gcc-2.6.3} with @code{-DFORCE_REG}.
4114:
4115: However, this potential advantage of assembly language implementations
4116: is not necessarily realized in complete Forth systems: We compared
4117: Gforth (direct threaded, compiled with @code{gcc-2.6.3} and
4118: @code{-DFORCE_REG}) with Win32Forth 1.2093, LMI's NT Forth (Beta, May
4119: 1994) and Eforth (with and without peephole (aka pinhole) optimization
4120: of the threaded code); all these systems were written in assembly
4121: language. We also compared Gforth with three systems written in C:
4122: PFE-0.9.14 (compiled with @code{gcc-2.6.3} with the default
4123: configuration for Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS
4124: -DUNROLL_NEXT}), ThisForth Beta (compiled with gcc-2.6.3 -O3
4125: -fomit-frame-pointer; ThisForth employs peephole optimization of the
4126: threaded code) and TILE (compiled with @code{make opt}). We benchmarked
4127: Gforth, PFE, ThisForth and TILE on a 486DX2/66 under Linux. Kenneth
4128: O'Heskin kindly provided the results for Win32Forth and NT Forth on a
4129: 486DX2/66 with similar memory performance under Windows NT. Marcel
4130: Hendrix ported Eforth to Linux, then extended it to run the benchmarks,
4131: added the peephole optimizer, ran the benchmarks and reported the
4132: results.
4133:
4134: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
4135: matrix multiplication come from the Stanford integer benchmarks and have
4136: been translated into Forth by Martin Fraeman; we used the versions
4137: included in the TILE Forth package, but with bigger data set sizes; and
4138: a recursive Fibonacci number computation for benchmarking calling
4139: performance. The following table shows the time taken for the benchmarks
4140: scaled by the time taken by Gforth (in other words, it shows the speedup
4141: factor that Gforth achieved over the other systems).
4142:
4143: @example
4144: relative Win32- NT eforth This-
4145: time Gforth Forth Forth eforth +opt PFE Forth TILE
4146: sieve 1.00 1.39 1.14 1.39 0.85 1.58 3.18 8.58
4147: bubble 1.00 1.31 1.41 1.48 0.88 1.50 3.88
4148: matmul 1.00 1.47 1.35 1.46 1.16 1.58 4.09
4149: fib 1.00 1.52 1.34 1.22 1.13 1.74 2.99 4.30
4150: @end example
4151:
4152: You may find the good performance of Gforth compared with the systems
4153: written in assembly language quite surprising. One important reason for
4154: the disappointing performance of these systems is probably that they are
4155: not written optimally for the 486 (e.g., they use the @code{lods}
4156: instruction). In addition, Win32Forth uses a comfortable, but costly
4157: method for relocating the Forth image: like @code{cforth}, it computes
4158: the actual addresses at run time, resulting in two address computations
4159: per NEXT (@pxref{System Architecture}).
4160:
4161: Only Eforth with the peephole optimizer performs comparable to
4162: Gforth. The speedups achieved with peephole optimization of threaded
4163: code are quite remarkable. Adding a peephole optimizer to Gforth should
4164: cause similar speedups.
4165:
4166: The speedup of Gforth over PFE, ThisForth and TILE can be easily
4167: explained with the self-imposed restriction to standard C, which makes
4168: efficient threading impossible (however, the measured implementation of
4169: PFE uses a GNU C extension: @ref{Global Reg Vars, , Defining Global
4170: Register Variables, gcc.info, GNU C Manual}). Moreover, current C
4171: compilers have a hard time optimizing other aspects of the ThisForth
4172: and the TILE source.
4173:
4174: Note that the performance of Gforth on 386 architecture processors
4175: varies widely with the version of @code{gcc} used. E.g., @code{gcc-2.5.8}
4176: failed to allocate any of the virtual machine registers into real
4177: machine registers by itself and would not work correctly with explicit
4178: register declarations, giving a 1.3 times slower engine (on a 486DX2/66
4179: running the Sieve) than the one measured above.
4180:
4181: In @cite{Translating Forth to Efficient C} by M. Anton Ertl and Martin
4182: Maierhofer (presented at EuroForth '95), an indirect threaded version of
4183: Gforth is compared with Win32Forth, NT Forth, PFE, and ThisForth; that
4184: version of Gforth is 2\%@minus{}8\% slower on a 486 than the version
4185: used here. The paper available at
4186: @*@file{http://www.complang.tuwien.ac.at/papers/ertl&maierhofer95.ps.gz};
4187: it also contains numbers for some native code systems. You can find
4188: numbers for Gforth on various machines in @file{Benchres}.
4189:
4190: @node Bugs, Origin, Internals, Top
4191: @chapter Bugs
4192:
4193: Known bugs are described in the file BUGS in the Gforth distribution.
4194:
4195: If you find a bug, please send a bug report to
4196: @code{bug-gforth@@gnu.ai.mit.edu}. A bug report should
4197: describe the Gforth version used (it is announced at the start of an
4198: interactive Gforth session), the machine and operating system (on Unix
4199: systems you can use @code{uname -a} to produce this information), the
4200: installation options (send the @code{config.status} file), and a
4201: complete list of changes you (or your installer) have made to the Gforth
4202: sources (if any); it should contain a program (or a sequence of keyboard
4203: commands) that reproduces the bug and a description of what you think
4204: constitutes the buggy behaviour.
4205:
4206: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
4207: to Report Bugs, gcc.info, GNU C Manual}.
4208:
4209:
4210: @node Origin, Word Index, Bugs, Top
4211: @chapter Authors and Ancestors of Gforth
4212:
4213: @section Authors and Contributors
4214:
4215: The Gforth project was started in mid-1992 by Bernd Paysan and Anton
4216: Ertl. The third major author was Jens Wilke. Lennart Benschop (who was
4217: one of Gforth's first users, in mid-1993) and Stuart Ramsden inspired us
4218: with their continuous feedback. Lennart Benshop contributed
4219: @file{glosgen.fs}, while Stuart Ramsden has been working on automatic
4220: support for calling C libraries. Helpful comments also came from Paul
4221: Kleinrubatscher, Christian Pirker, Dirk Zoller and Marcel Hendrix.
4222:
4223: Gforth also owes a lot to the authors of the tools we used (GCC, CVS,
4224: and autoconf, among others), and to the creators of the Internet: Gforth
4225: was developed across the Internet, and its authors have not met
4226: physically yet.
4227:
4228: @section Pedigree
4229:
4230: Gforth descends from BigForth (1993) and fig-Forth. Gforth and PFE (by
4231: Dirk Zoller) will cross-fertilize each other. Of course, a significant
4232: part of the design of Gforth was prescribed by ANS Forth.
4233:
4234: Bernd Paysan wrote BigForth, a descendent from TurboForth, an unreleased
4235: 32 bit native code version of VolksForth for the Atari ST, written
4236: mostly by Dietrich Weineck.
4237:
4238: VolksForth descends from F83. It was written by Klaus Schleisiek, Bernd
4239: Pennemann, Georg Rehfeld and Dietrich Weineck for the C64 (called
4240: UltraForth there) in the mid-80s and ported to the Atari ST in 1986.
4241:
4242: Henry Laxen and Mike Perry wrote F83 as a model implementation of the
4243: Forth-83 standard. !! Pedigree? When?
4244:
4245: A team led by Bill Ragsdale implemented fig-Forth on many processors in
4246: 1979. Robert Selzer and Bill Ragsdale developed the original
4247: implementation of fig-Forth for the 6502 based on microForth.
4248:
4249: The principal architect of microForth was Dean Sanderson. microForth was
4250: FORTH, Inc.'s first off-the-shelf product. It was developped in 1976 for
4251: the 1802, and subsequently implemented on the 8080, the 6800 and the
4252: Z80.
4253:
4254: All earlier Forth systems were custom-made, usually by Charles Moore,
4255: who discovered (as he puts it) Forth during the late 60s. The first full
4256: Forth existed in 1971.
4257:
4258: A part of the information in this section comes from @cite{The Evolution
4259: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
4260: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
4261: Notices 28(3), 1993. You can find more historical and genealogical
4262: information about Forth there.
4263:
4264: @node Word Index, Node Index, Origin, Top
4265: @chapter Word Index
4266:
4267: This index is as incomplete as the manual. Each word is listed with
4268: stack effect and wordset.
4269:
4270: @printindex fn
4271:
4272: @node Node Index, , Word Index, Top
4273: @chapter Node Index
4274:
4275: This index is even less complete than the manual.
4276:
4277: @contents
4278: @bye
4279:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>
![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to gforth.ds CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [gforth] / gforth