Annotation of gforth/gforth.ds, revision 1.24
1.1 anton 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.)
1.4 anton 4: @setfilename gforth.info
1.17 anton 5: @settitle Gforth Manual
1.4 anton 6: @comment @setchapternewpage odd
1.1 anton 7: @comment %**end of header (This is for running Texinfo on a region.)
8:
9: @ifinfo
1.17 anton 10: This file documents Gforth 0.1
1.1 anton 11:
1.21 anton 12: Copyright @copyright{} 1995 Free Software Foundation, Inc.
1.1 anton 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:
1.4 anton 18: @ignore
1.1 anton 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:
1.4 anton 24: @end ignore
1.1 anton 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:
1.24 ! anton 39: @finalout
1.1 anton 40: @titlepage
41: @sp 10
1.17 anton 42: @center @titlefont{Gforth Manual}
1.1 anton 43: @sp 2
1.17 anton 44: @center for version 0.1
1.1 anton 45: @sp 2
46: @center Anton Ertl
1.17 anton 47: @sp 3
48: @center This manual is under construction
1.1 anton 49:
50: @comment The following two commands start the copyright page.
51: @page
52: @vskip 0pt plus 1filll
1.21 anton 53: Copyright @copyright{} 1995 Free Software Foundation, Inc.
1.1 anton 54:
55: @comment !! Published by ... or You can get a copy of this manual ...
56:
57: Permission is granted to make and distribute verbatim copies of
58: this manual provided the copyright notice and this permission notice
59: are preserved on all copies.
60:
61: Permission is granted to copy and distribute modified versions of this
62: manual under the conditions for verbatim copying, provided also that the
63: sections entitled "Distribution" and "General Public License" are
64: included exactly as in the original, and provided that the entire
65: resulting derived work is distributed under the terms of a permission
66: notice identical to this one.
67:
68: Permission is granted to copy and distribute translations of this manual
69: into another language, under the above conditions for modified versions,
70: except that the sections entitled "Distribution" and "General Public
71: License" may be included in a translation approved by the author instead
72: of in the original English.
73: @end titlepage
74:
75:
76: @node Top, License, (dir), (dir)
77: @ifinfo
1.17 anton 78: Gforth is a free implementation of ANS Forth available on many
1.21 anton 79: personal machines. This manual corresponds to version 0.1.
1.1 anton 80: @end ifinfo
81:
82: @menu
1.4 anton 83: * License::
1.17 anton 84: * Goals:: About the Gforth Project
1.4 anton 85: * Other Books:: Things you might want to read
1.17 anton 86: * Invocation:: Starting Gforth
87: * Words:: Forth words available in Gforth
1.4 anton 88: * ANS conformance:: Implementation-defined options etc.
1.17 anton 89: * Model:: The abstract machine of Gforth
90: * Emacs and Gforth:: The Gforth Mode
1.4 anton 91: * Internals:: Implementation details
92: * Bugs:: How to report them
1.17 anton 93: * Pedigree:: Ancestors of Gforth
1.4 anton 94: * Word Index:: An item for each Forth word
95: * Node Index:: An item for each node
1.1 anton 96: @end menu
97:
98: @node License, Goals, Top, Top
1.20 pazsan 99: @unnumbered GNU GENERAL PUBLIC LICENSE
100: @center Version 2, June 1991
101:
102: @display
103: Copyright @copyright{} 1989, 1991 Free Software Foundation, Inc.
104: 675 Mass Ave, Cambridge, MA 02139, USA
105:
106: Everyone is permitted to copy and distribute verbatim copies
107: of this license document, but changing it is not allowed.
108: @end display
109:
110: @unnumberedsec Preamble
111:
112: The licenses for most software are designed to take away your
113: freedom to share and change it. By contrast, the GNU General Public
114: License is intended to guarantee your freedom to share and change free
115: software---to make sure the software is free for all its users. This
116: General Public License applies to most of the Free Software
117: Foundation's software and to any other program whose authors commit to
118: using it. (Some other Free Software Foundation software is covered by
119: the GNU Library General Public License instead.) You can apply it to
120: your programs, too.
121:
122: When we speak of free software, we are referring to freedom, not
123: price. Our General Public Licenses are designed to make sure that you
124: have the freedom to distribute copies of free software (and charge for
125: this service if you wish), that you receive source code or can get it
126: if you want it, that you can change the software or use pieces of it
127: in new free programs; and that you know you can do these things.
128:
129: To protect your rights, we need to make restrictions that forbid
130: anyone to deny you these rights or to ask you to surrender the rights.
131: These restrictions translate to certain responsibilities for you if you
132: distribute copies of the software, or if you modify it.
133:
134: For example, if you distribute copies of such a program, whether
135: gratis or for a fee, you must give the recipients all the rights that
136: you have. You must make sure that they, too, receive or can get the
137: source code. And you must show them these terms so they know their
138: rights.
139:
140: We protect your rights with two steps: (1) copyright the software, and
141: (2) offer you this license which gives you legal permission to copy,
142: distribute and/or modify the software.
143:
144: Also, for each author's protection and ours, we want to make certain
145: that everyone understands that there is no warranty for this free
146: software. If the software is modified by someone else and passed on, we
147: want its recipients to know that what they have is not the original, so
148: that any problems introduced by others will not reflect on the original
149: authors' reputations.
150:
151: Finally, any free program is threatened constantly by software
152: patents. We wish to avoid the danger that redistributors of a free
153: program will individually obtain patent licenses, in effect making the
154: program proprietary. To prevent this, we have made it clear that any
155: patent must be licensed for everyone's free use or not licensed at all.
156:
157: The precise terms and conditions for copying, distribution and
158: modification follow.
159:
160: @iftex
161: @unnumberedsec TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
162: @end iftex
163: @ifinfo
164: @center TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
165: @end ifinfo
166:
167: @enumerate 0
168: @item
169: This License applies to any program or other work which contains
170: a notice placed by the copyright holder saying it may be distributed
171: under the terms of this General Public License. The ``Program'', below,
172: refers to any such program or work, and a ``work based on the Program''
173: means either the Program or any derivative work under copyright law:
174: that is to say, a work containing the Program or a portion of it,
175: either verbatim or with modifications and/or translated into another
176: language. (Hereinafter, translation is included without limitation in
177: the term ``modification''.) Each licensee is addressed as ``you''.
178:
179: Activities other than copying, distribution and modification are not
180: covered by this License; they are outside its scope. The act of
181: running the Program is not restricted, and the output from the Program
182: is covered only if its contents constitute a work based on the
183: Program (independent of having been made by running the Program).
184: Whether that is true depends on what the Program does.
185:
186: @item
187: You may copy and distribute verbatim copies of the Program's
188: source code as you receive it, in any medium, provided that you
189: conspicuously and appropriately publish on each copy an appropriate
190: copyright notice and disclaimer of warranty; keep intact all the
191: notices that refer to this License and to the absence of any warranty;
192: and give any other recipients of the Program a copy of this License
193: along with the Program.
194:
195: You may charge a fee for the physical act of transferring a copy, and
196: you may at your option offer warranty protection in exchange for a fee.
197:
198: @item
199: You may modify your copy or copies of the Program or any portion
200: of it, thus forming a work based on the Program, and copy and
201: distribute such modifications or work under the terms of Section 1
202: above, provided that you also meet all of these conditions:
203:
204: @enumerate a
205: @item
206: You must cause the modified files to carry prominent notices
207: stating that you changed the files and the date of any change.
208:
209: @item
210: You must cause any work that you distribute or publish, that in
211: whole or in part contains or is derived from the Program or any
212: part thereof, to be licensed as a whole at no charge to all third
213: parties under the terms of this License.
214:
215: @item
216: If the modified program normally reads commands interactively
217: when run, you must cause it, when started running for such
218: interactive use in the most ordinary way, to print or display an
219: announcement including an appropriate copyright notice and a
220: notice that there is no warranty (or else, saying that you provide
221: a warranty) and that users may redistribute the program under
222: these conditions, and telling the user how to view a copy of this
223: License. (Exception: if the Program itself is interactive but
224: does not normally print such an announcement, your work based on
225: the Program is not required to print an announcement.)
226: @end enumerate
227:
228: These requirements apply to the modified work as a whole. If
229: identifiable sections of that work are not derived from the Program,
230: and can be reasonably considered independent and separate works in
231: themselves, then this License, and its terms, do not apply to those
232: sections when you distribute them as separate works. But when you
233: distribute the same sections as part of a whole which is a work based
234: on the Program, the distribution of the whole must be on the terms of
235: this License, whose permissions for other licensees extend to the
236: entire whole, and thus to each and every part regardless of who wrote it.
237:
238: Thus, it is not the intent of this section to claim rights or contest
239: your rights to work written entirely by you; rather, the intent is to
240: exercise the right to control the distribution of derivative or
241: collective works based on the Program.
242:
243: In addition, mere aggregation of another work not based on the Program
244: with the Program (or with a work based on the Program) on a volume of
245: a storage or distribution medium does not bring the other work under
246: the scope of this License.
247:
248: @item
249: You may copy and distribute the Program (or a work based on it,
250: under Section 2) in object code or executable form under the terms of
251: Sections 1 and 2 above provided that you also do one of the following:
252:
253: @enumerate a
254: @item
255: Accompany it with the complete corresponding machine-readable
256: source code, which must be distributed under the terms of Sections
257: 1 and 2 above on a medium customarily used for software interchange; or,
258:
259: @item
260: Accompany it with a written offer, valid for at least three
261: years, to give any third party, for a charge no more than your
262: cost of physically performing source distribution, a complete
263: machine-readable copy of the corresponding source code, to be
264: distributed under the terms of Sections 1 and 2 above on a medium
265: customarily used for software interchange; or,
266:
267: @item
268: Accompany it with the information you received as to the offer
269: to distribute corresponding source code. (This alternative is
270: allowed only for noncommercial distribution and only if you
271: received the program in object code or executable form with such
272: an offer, in accord with Subsection b above.)
273: @end enumerate
274:
275: The source code for a work means the preferred form of the work for
276: making modifications to it. For an executable work, complete source
277: code means all the source code for all modules it contains, plus any
278: associated interface definition files, plus the scripts used to
279: control compilation and installation of the executable. However, as a
280: special exception, the source code distributed need not include
281: anything that is normally distributed (in either source or binary
282: form) with the major components (compiler, kernel, and so on) of the
283: operating system on which the executable runs, unless that component
284: itself accompanies the executable.
285:
286: If distribution of executable or object code is made by offering
287: access to copy from a designated place, then offering equivalent
288: access to copy the source code from the same place counts as
289: distribution of the source code, even though third parties are not
290: compelled to copy the source along with the object code.
291:
292: @item
293: You may not copy, modify, sublicense, or distribute the Program
294: except as expressly provided under this License. Any attempt
295: otherwise to copy, modify, sublicense or distribute the Program is
296: void, and will automatically terminate your rights under this License.
297: However, parties who have received copies, or rights, from you under
298: this License will not have their licenses terminated so long as such
299: parties remain in full compliance.
300:
301: @item
302: You are not required to accept this License, since you have not
303: signed it. However, nothing else grants you permission to modify or
304: distribute the Program or its derivative works. These actions are
305: prohibited by law if you do not accept this License. Therefore, by
306: modifying or distributing the Program (or any work based on the
307: Program), you indicate your acceptance of this License to do so, and
308: all its terms and conditions for copying, distributing or modifying
309: the Program or works based on it.
310:
311: @item
312: Each time you redistribute the Program (or any work based on the
313: Program), the recipient automatically receives a license from the
314: original licensor to copy, distribute or modify the Program subject to
315: these terms and conditions. You may not impose any further
316: restrictions on the recipients' exercise of the rights granted herein.
317: You are not responsible for enforcing compliance by third parties to
318: this License.
319:
320: @item
321: If, as a consequence of a court judgment or allegation of patent
322: infringement or for any other reason (not limited to patent issues),
323: conditions are imposed on you (whether by court order, agreement or
324: otherwise) that contradict the conditions of this License, they do not
325: excuse you from the conditions of this License. If you cannot
326: distribute so as to satisfy simultaneously your obligations under this
327: License and any other pertinent obligations, then as a consequence you
328: may not distribute the Program at all. For example, if a patent
329: license would not permit royalty-free redistribution of the Program by
330: all those who receive copies directly or indirectly through you, then
331: the only way you could satisfy both it and this License would be to
332: refrain entirely from distribution of the Program.
333:
334: If any portion of this section is held invalid or unenforceable under
335: any particular circumstance, the balance of the section is intended to
336: apply and the section as a whole is intended to apply in other
337: circumstances.
338:
339: It is not the purpose of this section to induce you to infringe any
340: patents or other property right claims or to contest validity of any
341: such claims; this section has the sole purpose of protecting the
342: integrity of the free software distribution system, which is
343: implemented by public license practices. Many people have made
344: generous contributions to the wide range of software distributed
345: through that system in reliance on consistent application of that
346: system; it is up to the author/donor to decide if he or she is willing
347: to distribute software through any other system and a licensee cannot
348: impose that choice.
349:
350: This section is intended to make thoroughly clear what is believed to
351: be a consequence of the rest of this License.
352:
353: @item
354: If the distribution and/or use of the Program is restricted in
355: certain countries either by patents or by copyrighted interfaces, the
356: original copyright holder who places the Program under this License
357: may add an explicit geographical distribution limitation excluding
358: those countries, so that distribution is permitted only in or among
359: countries not thus excluded. In such case, this License incorporates
360: the limitation as if written in the body of this License.
361:
362: @item
363: The Free Software Foundation may publish revised and/or new versions
364: of the General Public License from time to time. Such new versions will
365: be similar in spirit to the present version, but may differ in detail to
366: address new problems or concerns.
367:
368: Each version is given a distinguishing version number. If the Program
369: specifies a version number of this License which applies to it and ``any
370: later version'', you have the option of following the terms and conditions
371: either of that version or of any later version published by the Free
372: Software Foundation. If the Program does not specify a version number of
373: this License, you may choose any version ever published by the Free Software
374: Foundation.
375:
376: @item
377: If you wish to incorporate parts of the Program into other free
378: programs whose distribution conditions are different, write to the author
379: to ask for permission. For software which is copyrighted by the Free
380: Software Foundation, write to the Free Software Foundation; we sometimes
381: make exceptions for this. Our decision will be guided by the two goals
382: of preserving the free status of all derivatives of our free software and
383: of promoting the sharing and reuse of software generally.
384:
385: @iftex
386: @heading NO WARRANTY
387: @end iftex
388: @ifinfo
389: @center NO WARRANTY
390: @end ifinfo
391:
392: @item
393: BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
394: FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN
395: OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
396: PROVIDE THE PROGRAM ``AS IS'' WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED
397: OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
398: MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS
399: TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE
400: PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING,
401: REPAIR OR CORRECTION.
402:
403: @item
404: IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
405: WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
406: REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
407: INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
408: OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED
409: TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY
410: YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER
411: PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
412: POSSIBILITY OF SUCH DAMAGES.
413: @end enumerate
414:
415: @iftex
416: @heading END OF TERMS AND CONDITIONS
417: @end iftex
418: @ifinfo
419: @center END OF TERMS AND CONDITIONS
420: @end ifinfo
421:
422: @page
423: @unnumberedsec How to Apply These Terms to Your New Programs
424:
425: If you develop a new program, and you want it to be of the greatest
426: possible use to the public, the best way to achieve this is to make it
427: free software which everyone can redistribute and change under these terms.
428:
429: To do so, attach the following notices to the program. It is safest
430: to attach them to the start of each source file to most effectively
431: convey the exclusion of warranty; and each file should have at least
432: the ``copyright'' line and a pointer to where the full notice is found.
433:
434: @smallexample
435: @var{one line to give the program's name and a brief idea of what it does.}
436: Copyright (C) 19@var{yy} @var{name of author}
437:
438: This program is free software; you can redistribute it and/or modify
439: it under the terms of the GNU General Public License as published by
440: the Free Software Foundation; either version 2 of the License, or
441: (at your option) any later version.
442:
443: This program is distributed in the hope that it will be useful,
444: but WITHOUT ANY WARRANTY; without even the implied warranty of
445: MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
446: GNU General Public License for more details.
447:
448: You should have received a copy of the GNU General Public License
449: along with this program; if not, write to the Free Software
450: Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
451: @end smallexample
452:
453: Also add information on how to contact you by electronic and paper mail.
454:
455: If the program is interactive, make it output a short notice like this
456: when it starts in an interactive mode:
457:
458: @smallexample
459: Gnomovision version 69, Copyright (C) 19@var{yy} @var{name of author}
460: Gnomovision comes with ABSOLUTELY NO WARRANTY; for details
461: type `show w'.
462: This is free software, and you are welcome to redistribute it
463: under certain conditions; type `show c' for details.
464: @end smallexample
465:
466: The hypothetical commands @samp{show w} and @samp{show c} should show
467: the appropriate parts of the General Public License. Of course, the
468: commands you use may be called something other than @samp{show w} and
469: @samp{show c}; they could even be mouse-clicks or menu items---whatever
470: suits your program.
471:
472: You should also get your employer (if you work as a programmer) or your
473: school, if any, to sign a ``copyright disclaimer'' for the program, if
474: necessary. Here is a sample; alter the names:
475:
476: @smallexample
477: Yoyodyne, Inc., hereby disclaims all copyright interest in the program
478: `Gnomovision' (which makes passes at compilers) written by James Hacker.
479:
480: @var{signature of Ty Coon}, 1 April 1989
481: Ty Coon, President of Vice
482: @end smallexample
483:
484: This General Public License does not permit incorporating your program into
485: proprietary programs. If your program is a subroutine library, you may
486: consider it more useful to permit linking proprietary applications with the
487: library. If this is what you want to do, use the GNU Library General
488: Public License instead of this License.
1.1 anton 489:
490: @iftex
1.23 pazsan 491: @node Preface
492: @comment node-name, next, previous, up
1.1 anton 493: @unnumbered Preface
1.23 pazsan 494: @cindex Preface
1.17 anton 495: This manual documents Gforth. The reader is expected to know
1.1 anton 496: Forth. This manual is primarily a reference manual. @xref{Other Books}
497: for introductory material.
498: @end iftex
499:
500: @node Goals, Other Books, License, Top
501: @comment node-name, next, previous, up
1.17 anton 502: @chapter Goals of Gforth
1.1 anton 503: @cindex Goals
1.17 anton 504: The goal of the Gforth Project is to develop a standard model for
1.1 anton 505: ANSI Forth. This can be split into several subgoals:
506:
507: @itemize @bullet
508: @item
1.17 anton 509: Gforth should conform to the ANSI Forth standard.
1.1 anton 510: @item
511: It should be a model, i.e. it should define all the
512: implementation-dependent things.
513: @item
514: It should become standard, i.e. widely accepted and used. This goal
515: is the most difficult one.
516: @end itemize
517:
1.17 anton 518: To achieve these goals Gforth should be
1.1 anton 519: @itemize @bullet
520: @item
521: Similar to previous models (fig-Forth, F83)
522: @item
523: Powerful. It should provide for all the things that are considered
524: necessary today and even some that are not yet considered necessary.
525: @item
526: Efficient. It should not get the reputation of being exceptionally
527: slow.
528: @item
529: Free.
530: @item
531: Available on many machines/easy to port.
532: @end itemize
533:
1.17 anton 534: Have we achieved these goals? Gforth conforms to the ANS Forth
535: standard. It may be considered a model, but we have not yet documented
1.1 anton 536: which parts of the model are stable and which parts we are likely to
1.17 anton 537: change. It certainly has not yet become a de facto standard. It has some
538: similarities and some differences to previous models. It has some
539: powerful features, but not yet everything that we envisioned. We
540: certainly have achieved our execution speed goals (@pxref{Performance}).
541: It is free and available on many machines.
1.1 anton 542:
543: @node Other Books, Invocation, Goals, Top
544: @chapter Other books on ANS Forth
545:
546: As the standard is relatively new, there are not many books out yet. It
1.17 anton 547: is not recommended to learn Forth by using Gforth and a book that is
1.1 anton 548: not written for ANS Forth, as you will not know your mistakes from the
549: deviations of the book.
550:
551: There is, of course, the standard, the definite reference if you want to
1.19 anton 552: write ANS Forth programs. It is available in printed form from the
553: National Standards Institute Sales Department (Tel.: USA (212) 642-4900;
554: Fax.: USA (212) 302-1286) as document @cite{X3.215-1994} for about $200. You
555: can also get it from Global Engineering Documents (Tel.: USA (800)
556: 854-7179; Fax.: (303) 843-9880) for about $300.
557:
558: @cite{dpANS6}, the last draft of the standard, which was then submitted to ANSI
559: for publication is available electronically and for free in some MS Word
560: format, and it has been converted to HTML. Some pointers to these
561: versions can be found through
1.24 ! anton 562: @*@file{http://www.complang.tuwien.ac.at/projects/forth.html}.
1.1 anton 563:
1.21 anton 564: @cite{Forth: The new model} by Jack Woehr (Prentice-Hall, 1993) is an
1.1 anton 565: introductory book based on a draft version of the standard. It does not
566: cover the whole standard. It also contains interesting background
567: information (Jack Woehr was in the ANS Forth Technical Committe). It is
568: not appropriate for complete newbies, but programmers experienced in
569: other languages should find it ok.
570:
571: @node Invocation, Words, Other Books, Top
572: @chapter Invocation
573:
574: You will usually just say @code{gforth}. In many other cases the default
1.17 anton 575: Gforth image will be invoked like this:
1.1 anton 576:
577: @example
578: gforth [files] [-e forth-code]
579: @end example
580:
581: executing the contents of the files and the Forth code in the order they
582: are given.
583:
584: In general, the command line looks like this:
585:
586: @example
587: gforth [initialization options] [image-specific options]
588: @end example
589:
590: The initialization options must come before the rest of the command
591: line. They are:
592:
593: @table @code
594: @item --image-file @var{file}
1.20 pazsan 595: @item -i @var{file}
1.1 anton 596: Loads the Forth image @var{file} instead of the default
597: @file{gforth.fi}.
598:
599: @item --path @var{path}
1.20 pazsan 600: @item -p @var{path}
1.1 anton 601: Uses @var{path} for searching the image file and Forth source code
602: files instead of the default in the environment variable
603: @code{GFORTHPATH} or the path specified at installation time (typically
604: @file{/usr/local/lib/gforth:.}). A path is given as a @code{:}-separated
605: list.
606:
607: @item --dictionary-size @var{size}
608: @item -m @var{size}
609: Allocate @var{size} space for the Forth dictionary space instead of
610: using the default specified in the image (typically 256K). The
611: @var{size} specification consists of an integer and a unit (e.g.,
612: @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
613: size, in this case Cells), @code{k} (kilobytes), and @code{M}
614: (Megabytes). If no unit is specified, @code{e} is used.
615:
616: @item --data-stack-size @var{size}
617: @item -d @var{size}
618: Allocate @var{size} space for the data stack instead of using the
619: default specified in the image (typically 16K).
620:
621: @item --return-stack-size @var{size}
622: @item -r @var{size}
623: Allocate @var{size} space for the return stack instead of using the
624: default specified in the image (typically 16K).
625:
626: @item --fp-stack-size @var{size}
627: @item -f @var{size}
628: Allocate @var{size} space for the floating point stack instead of
629: using the default specified in the image (typically 16K). In this case
630: the unit specifier @code{e} refers to floating point numbers.
631:
632: @item --locals-stack-size @var{size}
633: @item -l @var{size}
634: Allocate @var{size} space for the locals stack instead of using the
635: default specified in the image (typically 16K).
636:
637: @end table
638:
639: As explained above, the image-specific command-line arguments for the
640: default image @file{gforth.fi} consist of a sequence of filenames and
641: @code{-e @var{forth-code}} options that are interpreted in the seqence
642: in which they are given. The @code{-e @var{forth-code}} or
643: @code{--evaluate @var{forth-code}} option evaluates the forth
644: code. This option takes only one argument; if you want to evaluate more
645: Forth words, you have to quote them or use several @code{-e}s. To exit
646: after processing the command line (instead of entering interactive mode)
647: append @code{-e bye} to the command line.
648:
1.22 anton 649: If you have several versions of Gforth installed, @code{gforth} will
650: invoke the version that was installed last. @code{gforth-@var{version}}
651: invokes a specific version. You may want to use the option
652: @code{--path}, if your environment contains the variable
653: @code{GFORTHPATH}.
654:
1.1 anton 655: Not yet implemented:
656: On startup the system first executes the system initialization file
657: (unless the option @code{--no-init-file} is given; note that the system
658: resulting from using this option may not be ANS Forth conformant). Then
659: the user initialization file @file{.gforth.fs} is executed, unless the
660: option @code{--no-rc} is given; this file is first searched in @file{.},
661: then in @file{~}, then in the normal path (see above).
662:
1.4 anton 663: @node Words, ANS conformance, Invocation, Top
1.1 anton 664: @chapter Forth Words
665:
666: @menu
1.4 anton 667: * Notation::
668: * Arithmetic::
669: * Stack Manipulation::
670: * Memory access::
671: * Control Structures::
672: * Locals::
673: * Defining Words::
674: * Wordlists::
675: * Files::
676: * Blocks::
677: * Other I/O::
678: * Programming Tools::
1.18 anton 679: * Assembler and Code words::
1.4 anton 680: * Threading Words::
1.1 anton 681: @end menu
682:
683: @node Notation, Arithmetic, Words, Words
684: @section Notation
685:
686: The Forth words are described in this section in the glossary notation
687: that has become a de-facto standard for Forth texts, i.e.
688:
1.4 anton 689: @format
1.1 anton 690: @var{word} @var{Stack effect} @var{wordset} @var{pronunciation}
1.4 anton 691: @end format
1.1 anton 692: @var{Description}
693:
694: @table @var
695: @item word
1.17 anton 696: The name of the word. BTW, Gforth is case insensitive, so you can
1.14 anton 697: type the words in in lower case (However, @pxref{core-idef}).
1.1 anton 698:
699: @item Stack effect
700: The stack effect is written in the notation @code{@var{before} --
701: @var{after}}, where @var{before} and @var{after} describe the top of
702: stack entries before and after the execution of the word. The rest of
703: the stack is not touched by the word. The top of stack is rightmost,
1.17 anton 704: i.e., a stack sequence is written as it is typed in. Note that Gforth
1.1 anton 705: uses a separate floating point stack, but a unified stack
706: notation. Also, return stack effects are not shown in @var{stack
707: effect}, but in @var{Description}. The name of a stack item describes
708: the type and/or the function of the item. See below for a discussion of
709: the types.
710:
1.19 anton 711: All words have two stack effects: A compile-time stack effect and a
712: run-time stack effect. The compile-time stack-effect of most words is
713: @var{ -- }. If the compile-time stack-effect of a word deviates from
714: this standard behaviour, or the word does other unusual things at
715: compile time, both stack effects are shown; otherwise only the run-time
716: stack effect is shown.
717:
1.1 anton 718: @item pronunciation
719: How the word is pronounced
720:
721: @item wordset
722: The ANS Forth standard is divided into several wordsets. A standard
723: system need not support all of them. So, the fewer wordsets your program
724: uses the more portable it will be in theory. However, we suspect that
725: most ANS Forth systems on personal machines will feature all
726: wordsets. Words that are not defined in the ANS standard have
1.19 anton 727: @code{gforth} or @code{gforth-internal} as wordset. @code{gforth}
728: describes words that will work in future releases of Gforth;
729: @code{gforth-internal} words are more volatile. Environmental query
730: strings are also displayed like words; you can recognize them by the
731: @code{environment} in the wordset field.
1.1 anton 732:
733: @item Description
734: A description of the behaviour of the word.
735: @end table
736:
1.4 anton 737: The type of a stack item is specified by the character(s) the name
738: starts with:
1.1 anton 739:
740: @table @code
741: @item f
742: Bool, i.e. @code{false} or @code{true}.
743: @item c
744: Char
745: @item w
746: Cell, can contain an integer or an address
747: @item n
748: signed integer
749: @item u
750: unsigned integer
751: @item d
752: double sized signed integer
753: @item ud
754: double sized unsigned integer
755: @item r
756: Float
757: @item a_
758: Cell-aligned address
759: @item c_
760: Char-aligned address (note that a Char is two bytes in Windows NT)
761: @item f_
762: Float-aligned address
763: @item df_
764: Address aligned for IEEE double precision float
765: @item sf_
766: Address aligned for IEEE single precision float
767: @item xt
768: Execution token, same size as Cell
769: @item wid
770: Wordlist ID, same size as Cell
771: @item f83name
772: Pointer to a name structure
773: @end table
774:
1.4 anton 775: @node Arithmetic, Stack Manipulation, Notation, Words
1.1 anton 776: @section Arithmetic
777: Forth arithmetic is not checked, i.e., you will not hear about integer
778: overflow on addition or multiplication, you may hear about division by
779: zero if you are lucky. The operator is written after the operands, but
780: the operands are still in the original order. I.e., the infix @code{2-1}
781: corresponds to @code{2 1 -}. Forth offers a variety of division
782: operators. If you perform division with potentially negative operands,
783: you do not want to use @code{/} or @code{/mod} with its undefined
784: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
1.4 anton 785: former, @pxref{Mixed precision}).
786:
787: @menu
788: * Single precision::
789: * Bitwise operations::
790: * Mixed precision:: operations with single and double-cell integers
791: * Double precision:: Double-cell integer arithmetic
792: * Floating Point::
793: @end menu
1.1 anton 794:
1.4 anton 795: @node Single precision, Bitwise operations, Arithmetic, Arithmetic
1.1 anton 796: @subsection Single precision
797: doc-+
798: doc--
799: doc-*
800: doc-/
801: doc-mod
802: doc-/mod
803: doc-negate
804: doc-abs
805: doc-min
806: doc-max
807:
1.4 anton 808: @node Bitwise operations, Mixed precision, Single precision, Arithmetic
1.1 anton 809: @subsection Bitwise operations
810: doc-and
811: doc-or
812: doc-xor
813: doc-invert
814: doc-2*
815: doc-2/
816:
1.4 anton 817: @node Mixed precision, Double precision, Bitwise operations, Arithmetic
1.1 anton 818: @subsection Mixed precision
819: doc-m+
820: doc-*/
821: doc-*/mod
822: doc-m*
823: doc-um*
824: doc-m*/
825: doc-um/mod
826: doc-fm/mod
827: doc-sm/rem
828:
1.4 anton 829: @node Double precision, Floating Point, Mixed precision, Arithmetic
1.1 anton 830: @subsection Double precision
1.16 anton 831:
832: The outer (aka text) interpreter converts numbers containing a dot into
833: a double precision number. Note that only numbers with the dot as last
834: character are standard-conforming.
835:
1.1 anton 836: doc-d+
837: doc-d-
838: doc-dnegate
839: doc-dabs
840: doc-dmin
841: doc-dmax
842:
1.4 anton 843: @node Floating Point, , Double precision, Arithmetic
844: @subsection Floating Point
1.16 anton 845:
846: The format of floating point numbers recognized by the outer (aka text)
847: interpreter is: a signed decimal number, possibly containing a decimal
848: point (@code{.}), followed by @code{E} or @code{e}, optionally followed
849: by a signed integer (the exponent). E.g., @code{1e} ist the same as
850: @code{+1.0e+1}. Note that a number without @code{e}
851: is not interpreted as floating-point number, but as double (if the
852: number contains a @code{.}) or single precision integer. Also,
853: conversions between string and floating point numbers always use base
854: 10, irrespective of the value of @code{BASE}. If @code{BASE} contains a
855: value greater then 14, the @code{E} may be interpreted as digit and the
856: number will be interpreted as integer, unless it has a signed exponent
857: (both @code{+} and @code{-} are allowed as signs).
1.4 anton 858:
859: Angles in floating point operations are given in radians (a full circle
1.17 anton 860: has 2 pi radians). Note, that Gforth has a separate floating point
1.4 anton 861: stack, but we use the unified notation.
862:
863: Floating point numbers have a number of unpleasant surprises for the
864: unwary (e.g., floating point addition is not associative) and even a few
865: for the wary. You should not use them unless you know what you are doing
866: or you don't care that the results you get are totally bogus. If you
867: want to learn about the problems of floating point numbers (and how to
1.11 anton 868: avoid them), you might start with @cite{David Goldberg, What Every
1.6 anton 869: Computer Scientist Should Know About Floating-Point Arithmetic, ACM
870: Computing Surveys 23(1):5@minus{}48, March 1991}.
1.4 anton 871:
872: doc-f+
873: doc-f-
874: doc-f*
875: doc-f/
876: doc-fnegate
877: doc-fabs
878: doc-fmax
879: doc-fmin
880: doc-floor
881: doc-fround
882: doc-f**
883: doc-fsqrt
884: doc-fexp
885: doc-fexpm1
886: doc-fln
887: doc-flnp1
888: doc-flog
1.6 anton 889: doc-falog
1.4 anton 890: doc-fsin
891: doc-fcos
892: doc-fsincos
893: doc-ftan
894: doc-fasin
895: doc-facos
896: doc-fatan
897: doc-fatan2
898: doc-fsinh
899: doc-fcosh
900: doc-ftanh
901: doc-fasinh
902: doc-facosh
903: doc-fatanh
904:
905: @node Stack Manipulation, Memory access, Arithmetic, Words
1.1 anton 906: @section Stack Manipulation
907:
1.17 anton 908: Gforth has a data stack (aka parameter stack) for characters, cells,
1.1 anton 909: addresses, and double cells, a floating point stack for floating point
910: numbers, a return stack for storing the return addresses of colon
911: definitions and other data, and a locals stack for storing local
912: variables. Note that while every sane Forth has a separate floating
913: point stack, this is not strictly required; an ANS Forth system could
914: theoretically keep floating point numbers on the data stack. As an
915: additional difficulty, you don't know how many cells a floating point
916: number takes. It is reportedly possible to write words in a way that
917: they work also for a unified stack model, but we do not recommend trying
1.4 anton 918: it. Instead, just say that your program has an environmental dependency
919: on a separate FP stack.
920:
921: Also, a Forth system is allowed to keep the local variables on the
1.1 anton 922: return stack. This is reasonable, as local variables usually eliminate
923: the need to use the return stack explicitly. So, if you want to produce
924: a standard complying program and if you are using local variables in a
925: word, forget about return stack manipulations in that word (see the
926: standard document for the exact rules).
927:
1.4 anton 928: @menu
929: * Data stack::
930: * Floating point stack::
931: * Return stack::
932: * Locals stack::
933: * Stack pointer manipulation::
934: @end menu
935:
936: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
1.1 anton 937: @subsection Data stack
938: doc-drop
939: doc-nip
940: doc-dup
941: doc-over
942: doc-tuck
943: doc-swap
944: doc-rot
945: doc--rot
946: doc-?dup
947: doc-pick
948: doc-roll
949: doc-2drop
950: doc-2nip
951: doc-2dup
952: doc-2over
953: doc-2tuck
954: doc-2swap
955: doc-2rot
956:
1.4 anton 957: @node Floating point stack, Return stack, Data stack, Stack Manipulation
1.1 anton 958: @subsection Floating point stack
959: doc-fdrop
960: doc-fnip
961: doc-fdup
962: doc-fover
963: doc-ftuck
964: doc-fswap
965: doc-frot
966:
1.4 anton 967: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
1.1 anton 968: @subsection Return stack
969: doc->r
970: doc-r>
971: doc-r@
972: doc-rdrop
973: doc-2>r
974: doc-2r>
975: doc-2r@
976: doc-2rdrop
977:
1.4 anton 978: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
1.1 anton 979: @subsection Locals stack
980:
1.4 anton 981: @node Stack pointer manipulation, , Locals stack, Stack Manipulation
1.1 anton 982: @subsection Stack pointer manipulation
983: doc-sp@
984: doc-sp!
985: doc-fp@
986: doc-fp!
987: doc-rp@
988: doc-rp!
989: doc-lp@
990: doc-lp!
991:
1.4 anton 992: @node Memory access, Control Structures, Stack Manipulation, Words
1.1 anton 993: @section Memory access
994:
1.4 anton 995: @menu
996: * Stack-Memory transfers::
997: * Address arithmetic::
998: * Memory block access::
999: @end menu
1000:
1001: @node Stack-Memory transfers, Address arithmetic, Memory access, Memory access
1.1 anton 1002: @subsection Stack-Memory transfers
1003:
1004: doc-@
1005: doc-!
1006: doc-+!
1007: doc-c@
1008: doc-c!
1009: doc-2@
1010: doc-2!
1011: doc-f@
1012: doc-f!
1013: doc-sf@
1014: doc-sf!
1015: doc-df@
1016: doc-df!
1017:
1.4 anton 1018: @node Address arithmetic, Memory block access, Stack-Memory transfers, Memory access
1.1 anton 1019: @subsection Address arithmetic
1020:
1021: ANS Forth does not specify the sizes of the data types. Instead, it
1022: offers a number of words for computing sizes and doing address
1023: arithmetic. Basically, address arithmetic is performed in terms of
1024: address units (aus); on most systems the address unit is one byte. Note
1025: that a character may have more than one au, so @code{chars} is no noop
1026: (on systems where it is a noop, it compiles to nothing).
1027:
1028: ANS Forth also defines words for aligning addresses for specific
1029: addresses. Many computers require that accesses to specific data types
1030: must only occur at specific addresses; e.g., that cells may only be
1031: accessed at addresses divisible by 4. Even if a machine allows unaligned
1032: accesses, it can usually perform aligned accesses faster.
1033:
1.17 anton 1034: For the performance-conscious: alignment operations are usually only
1.1 anton 1035: necessary during the definition of a data structure, not during the
1036: (more frequent) accesses to it.
1037:
1038: ANS Forth defines no words for character-aligning addresses. This is not
1039: an oversight, but reflects the fact that addresses that are not
1040: char-aligned have no use in the standard and therefore will not be
1041: created.
1042:
1043: The standard guarantees that addresses returned by @code{CREATE}d words
1.17 anton 1044: are cell-aligned; in addition, Gforth guarantees that these addresses
1.1 anton 1045: are aligned for all purposes.
1046:
1.9 anton 1047: Note that the standard defines a word @code{char}, which has nothing to
1048: do with address arithmetic.
1049:
1.1 anton 1050: doc-chars
1051: doc-char+
1052: doc-cells
1053: doc-cell+
1054: doc-align
1055: doc-aligned
1056: doc-floats
1057: doc-float+
1058: doc-falign
1059: doc-faligned
1060: doc-sfloats
1061: doc-sfloat+
1062: doc-sfalign
1063: doc-sfaligned
1064: doc-dfloats
1065: doc-dfloat+
1066: doc-dfalign
1067: doc-dfaligned
1.10 anton 1068: doc-maxalign
1069: doc-maxaligned
1070: doc-cfalign
1071: doc-cfaligned
1.1 anton 1072: doc-address-unit-bits
1073:
1.4 anton 1074: @node Memory block access, , Address arithmetic, Memory access
1.1 anton 1075: @subsection Memory block access
1076:
1077: doc-move
1078: doc-erase
1079:
1080: While the previous words work on address units, the rest works on
1081: characters.
1082:
1083: doc-cmove
1084: doc-cmove>
1085: doc-fill
1086: doc-blank
1087:
1.4 anton 1088: @node Control Structures, Locals, Memory access, Words
1.1 anton 1089: @section Control Structures
1090:
1091: Control structures in Forth cannot be used in interpret state, only in
1092: compile state, i.e., in a colon definition. We do not like this
1093: limitation, but have not seen a satisfying way around it yet, although
1094: many schemes have been proposed.
1095:
1.4 anton 1096: @menu
1097: * Selection::
1098: * Simple Loops::
1099: * Counted Loops::
1100: * Arbitrary control structures::
1101: * Calls and returns::
1102: * Exception Handling::
1103: @end menu
1104:
1105: @node Selection, Simple Loops, Control Structures, Control Structures
1.1 anton 1106: @subsection Selection
1107:
1108: @example
1109: @var{flag}
1110: IF
1111: @var{code}
1112: ENDIF
1113: @end example
1114: or
1115: @example
1116: @var{flag}
1117: IF
1118: @var{code1}
1119: ELSE
1120: @var{code2}
1121: ENDIF
1122: @end example
1123:
1.4 anton 1124: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
1.1 anton 1125: standard, and @code{ENDIF} is not, although it is quite popular. We
1126: recommend using @code{ENDIF}, because it is less confusing for people
1127: who also know other languages (and is not prone to reinforcing negative
1128: prejudices against Forth in these people). Adding @code{ENDIF} to a
1129: system that only supplies @code{THEN} is simple:
1130: @example
1131: : endif POSTPONE then ; immediate
1132: @end example
1133:
1134: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
1135: (adv.)} has the following meanings:
1136: @quotation
1137: ... 2b: following next after in order ... 3d: as a necessary consequence
1138: (if you were there, then you saw them).
1139: @end quotation
1140: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
1141: and many other programming languages has the meaning 3d.]
1142:
1143: We also provide the words @code{?dup-if} and @code{?dup-0=-if}, so you
1144: can avoid using @code{?dup}.
1145:
1146: @example
1147: @var{n}
1148: CASE
1149: @var{n1} OF @var{code1} ENDOF
1150: @var{n2} OF @var{code2} ENDOF
1.4 anton 1151: @dots{}
1.1 anton 1152: ENDCASE
1153: @end example
1154:
1155: Executes the first @var{codei}, where the @var{ni} is equal to
1156: @var{n}. A default case can be added by simply writing the code after
1157: the last @code{ENDOF}. It may use @var{n}, which is on top of the stack,
1158: but must not consume it.
1159:
1.4 anton 1160: @node Simple Loops, Counted Loops, Selection, Control Structures
1.1 anton 1161: @subsection Simple Loops
1162:
1163: @example
1164: BEGIN
1165: @var{code1}
1166: @var{flag}
1167: WHILE
1168: @var{code2}
1169: REPEAT
1170: @end example
1171:
1172: @var{code1} is executed and @var{flag} is computed. If it is true,
1173: @var{code2} is executed and the loop is restarted; If @var{flag} is false, execution continues after the @code{REPEAT}.
1174:
1175: @example
1176: BEGIN
1177: @var{code}
1178: @var{flag}
1179: UNTIL
1180: @end example
1181:
1182: @var{code} is executed. The loop is restarted if @code{flag} is false.
1183:
1184: @example
1185: BEGIN
1186: @var{code}
1187: AGAIN
1188: @end example
1189:
1190: This is an endless loop.
1191:
1.4 anton 1192: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
1.1 anton 1193: @subsection Counted Loops
1194:
1195: The basic counted loop is:
1196: @example
1197: @var{limit} @var{start}
1198: ?DO
1199: @var{body}
1200: LOOP
1201: @end example
1202:
1203: This performs one iteration for every integer, starting from @var{start}
1204: and up to, but excluding @var{limit}. The counter, aka index, can be
1205: accessed with @code{i}. E.g., the loop
1206: @example
1207: 10 0 ?DO
1208: i .
1209: LOOP
1210: @end example
1211: prints
1212: @example
1213: 0 1 2 3 4 5 6 7 8 9
1214: @end example
1215: The index of the innermost loop can be accessed with @code{i}, the index
1216: of the next loop with @code{j}, and the index of the third loop with
1217: @code{k}.
1218:
1219: The loop control data are kept on the return stack, so there are some
1220: restrictions on mixing return stack accesses and counted loop
1221: words. E.g., if you put values on the return stack outside the loop, you
1222: cannot read them inside the loop. If you put values on the return stack
1223: within a loop, you have to remove them before the end of the loop and
1224: before accessing the index of the loop.
1225:
1226: There are several variations on the counted loop:
1227:
1228: @code{LEAVE} leaves the innermost counted loop immediately.
1229:
1.18 anton 1230: If @var{start} is greater than @var{limit}, a @code{?DO} loop is entered
1231: (and @code{LOOP} iterates until they become equal by wrap-around
1232: arithmetic). This behaviour is usually not what you want. Therefore,
1233: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
1234: @code{?DO}), which do not enter the loop if @var{start} is greater than
1235: @var{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
1236: unsigned loop parameters. These words can be implemented easily on
1237: standard systems, so using them does not make your programs hard to
1238: port; e.g.:
1239: @example
1240: : +DO ( compile-time: -- do-sys; run-time: n1 n2 -- )
1241: POSTPONE over POSTPONE min POSTPONE ?DO ; immediate
1242: @end example
1243:
1.1 anton 1244: @code{LOOP} can be replaced with @code{@var{n} +LOOP}; this updates the
1245: index by @var{n} instead of by 1. The loop is terminated when the border
1246: between @var{limit-1} and @var{limit} is crossed. E.g.:
1247:
1.18 anton 1248: @code{4 0 +DO i . 2 +LOOP} prints @code{0 2}
1.1 anton 1249:
1.18 anton 1250: @code{4 1 +DO i . 2 +LOOP} prints @code{1 3}
1.1 anton 1251:
1252: The behaviour of @code{@var{n} +LOOP} is peculiar when @var{n} is negative:
1253:
1.2 anton 1254: @code{-1 0 ?DO i . -1 +LOOP} prints @code{0 -1}
1.1 anton 1255:
1.2 anton 1256: @code{ 0 0 ?DO i . -1 +LOOP} prints nothing
1.1 anton 1257:
1.18 anton 1258: Therefore we recommend avoiding @code{@var{n} +LOOP} with negative
1259: @var{n}. One alternative is @code{@var{u} -LOOP}, which reduces the
1260: index by @var{u} each iteration. The loop is terminated when the border
1261: between @var{limit+1} and @var{limit} is crossed. Gforth also provides
1262: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
1.1 anton 1263:
1.18 anton 1264: @code{-2 0 -DO i . 1 -LOOP} prints @code{0 -1}
1.1 anton 1265:
1.18 anton 1266: @code{-1 0 -DO i . 1 -LOOP} prints @code{0}
1.1 anton 1267:
1.18 anton 1268: @code{ 0 0 -DO i . 1 -LOOP} prints nothing
1.1 anton 1269:
1.18 anton 1270: Another alternative is @code{@var{n} S+LOOP}, where the negative
1271: case behaves symmetrical to the positive case:
1.1 anton 1272:
1.18 anton 1273: @code{-2 0 -DO i . -1 S+LOOP} prints @code{0 -1}
1274:
1275: The loop is terminated when the border between @var{limit@minus{}sgn(n)}
1276: and @var{limit} is crossed. Unfortunately, neither @code{-LOOP} nor
1277: @code{S+LOOP} are part of the ANS Forth standard, and they are not easy
1278: to implement using standard words. If you want to write standard
1279: programs, just avoid counting down.
1280:
1281: @code{?DO} can also be replaced by @code{DO}. @code{DO} always enters
1282: the loop, independent of the loop parameters. Do not use @code{DO}, even
1283: if you know that the loop is entered in any case. Such knowledge tends
1284: to become invalid during maintenance of a program, and then the
1285: @code{DO} will make trouble.
1.1 anton 1286:
1287: @code{UNLOOP} is used to prepare for an abnormal loop exit, e.g., via
1288: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
1289: return stack so @code{EXIT} can get to its return address.
1290:
1291: Another counted loop is
1292: @example
1293: @var{n}
1294: FOR
1295: @var{body}
1296: NEXT
1297: @end example
1298: This is the preferred loop of native code compiler writers who are too
1.17 anton 1299: lazy to optimize @code{?DO} loops properly. In Gforth, this loop
1.1 anton 1300: iterates @var{n+1} times; @code{i} produces values starting with @var{n}
1301: and ending with 0. Other Forth systems may behave differently, even if
1302: they support @code{FOR} loops.
1303:
1.4 anton 1304: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
1.2 anton 1305: @subsection Arbitrary control structures
1306:
1307: ANS Forth permits and supports using control structures in a non-nested
1308: way. Information about incomplete control structures is stored on the
1309: control-flow stack. This stack may be implemented on the Forth data
1.17 anton 1310: stack, and this is what we have done in Gforth.
1.2 anton 1311:
1312: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
1313: entry represents a backward branch target. A few words are the basis for
1314: building any control structure possible (except control structures that
1315: need storage, like calls, coroutines, and backtracking).
1316:
1.3 anton 1317: doc-if
1318: doc-ahead
1319: doc-then
1320: doc-begin
1321: doc-until
1322: doc-again
1323: doc-cs-pick
1324: doc-cs-roll
1.2 anton 1325:
1.17 anton 1326: On many systems control-flow stack items take one word, in Gforth they
1.2 anton 1327: currently take three (this may change in the future). Therefore it is a
1328: really good idea to manipulate the control flow stack with
1329: @code{cs-pick} and @code{cs-roll}, not with data stack manipulation
1330: words.
1331:
1332: Some standard control structure words are built from these words:
1333:
1.3 anton 1334: doc-else
1335: doc-while
1336: doc-repeat
1.2 anton 1337:
1338: Counted loop words constitute a separate group of words:
1339:
1.3 anton 1340: doc-?do
1.18 anton 1341: doc-+do
1342: doc-u+do
1343: doc--do
1344: doc-u-do
1.3 anton 1345: doc-do
1346: doc-for
1347: doc-loop
1348: doc-s+loop
1349: doc-+loop
1.18 anton 1350: doc--loop
1.3 anton 1351: doc-next
1352: doc-leave
1353: doc-?leave
1354: doc-unloop
1.10 anton 1355: doc-done
1.2 anton 1356:
1357: The standard does not allow using @code{cs-pick} and @code{cs-roll} on
1358: @i{do-sys}. Our system allows it, but it's your job to ensure that for
1359: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
1.3 anton 1360: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
1361: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
1.7 pazsan 1362: resolved (by using one of the loop-ending words or @code{DONE}).
1.2 anton 1363:
1364: Another group of control structure words are
1365:
1.3 anton 1366: doc-case
1367: doc-endcase
1368: doc-of
1369: doc-endof
1.2 anton 1370:
1371: @i{case-sys} and @i{of-sys} cannot be processed using @code{cs-pick} and
1372: @code{cs-roll}.
1373:
1.3 anton 1374: @subsubsection Programming Style
1375:
1376: In order to ensure readability we recommend that you do not create
1377: arbitrary control structures directly, but define new control structure
1378: words for the control structure you want and use these words in your
1379: program.
1380:
1381: E.g., instead of writing
1382:
1383: @example
1384: begin
1385: ...
1386: if [ 1 cs-roll ]
1387: ...
1388: again then
1389: @end example
1390:
1391: we recommend defining control structure words, e.g.,
1392:
1393: @example
1394: : while ( dest -- orig dest )
1395: POSTPONE if
1396: 1 cs-roll ; immediate
1397:
1398: : repeat ( orig dest -- )
1399: POSTPONE again
1400: POSTPONE then ; immediate
1401: @end example
1402:
1403: and then using these to create the control structure:
1404:
1405: @example
1406: begin
1407: ...
1408: while
1409: ...
1410: repeat
1411: @end example
1412:
1413: That's much easier to read, isn't it? Of course, @code{BEGIN} and
1414: @code{WHILE} are predefined, so in this example it would not be
1415: necessary to define them.
1416:
1.4 anton 1417: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
1.3 anton 1418: @subsection Calls and returns
1419:
1420: A definition can be called simply be writing the name of the
1.17 anton 1421: definition. When the end of the definition is reached, it returns. An
1422: earlier return can be forced using
1.3 anton 1423:
1424: doc-exit
1425:
1426: Don't forget to clean up the return stack and @code{UNLOOP} any
1427: outstanding @code{?DO}...@code{LOOP}s before @code{EXIT}ing. The
1428: primitive compiled by @code{EXIT} is
1429:
1430: doc-;s
1431:
1.4 anton 1432: @node Exception Handling, , Calls and returns, Control Structures
1.3 anton 1433: @subsection Exception Handling
1434:
1435: doc-catch
1436: doc-throw
1437:
1.4 anton 1438: @node Locals, Defining Words, Control Structures, Words
1.1 anton 1439: @section Locals
1440:
1.2 anton 1441: Local variables can make Forth programming more enjoyable and Forth
1442: programs easier to read. Unfortunately, the locals of ANS Forth are
1443: laden with restrictions. Therefore, we provide not only the ANS Forth
1444: locals wordset, but also our own, more powerful locals wordset (we
1445: implemented the ANS Forth locals wordset through our locals wordset).
1446:
1.24 ! anton 1447: The ideas in this section have also been published in the paper
! 1448: @cite{Automatic Scoping of Local Variables} by M. Anton Ertl, presented
! 1449: at EuroForth '94; it is available at
! 1450: @*@file{http://www.complang.tuwien.ac.at/papers/ertl94l.ps.gz}.
! 1451:
1.2 anton 1452: @menu
1.17 anton 1453: * Gforth locals::
1.4 anton 1454: * ANS Forth locals::
1.2 anton 1455: @end menu
1456:
1.17 anton 1457: @node Gforth locals, ANS Forth locals, Locals, Locals
1458: @subsection Gforth locals
1.2 anton 1459:
1460: Locals can be defined with
1461:
1462: @example
1463: @{ local1 local2 ... -- comment @}
1464: @end example
1465: or
1466: @example
1467: @{ local1 local2 ... @}
1468: @end example
1469:
1470: E.g.,
1471: @example
1472: : max @{ n1 n2 -- n3 @}
1473: n1 n2 > if
1474: n1
1475: else
1476: n2
1477: endif ;
1478: @end example
1479:
1480: The similarity of locals definitions with stack comments is intended. A
1481: locals definition often replaces the stack comment of a word. The order
1482: of the locals corresponds to the order in a stack comment and everything
1483: after the @code{--} is really a comment.
1484:
1485: This similarity has one disadvantage: It is too easy to confuse locals
1486: declarations with stack comments, causing bugs and making them hard to
1487: find. However, this problem can be avoided by appropriate coding
1488: conventions: Do not use both notations in the same program. If you do,
1489: they should be distinguished using additional means, e.g. by position.
1490:
1491: The name of the local may be preceded by a type specifier, e.g.,
1492: @code{F:} for a floating point value:
1493:
1494: @example
1495: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
1496: \ complex multiplication
1497: Ar Br f* Ai Bi f* f-
1498: Ar Bi f* Ai Br f* f+ ;
1499: @end example
1500:
1.17 anton 1501: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
1.2 anton 1502: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
1503: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
1504: with @code{W:}, @code{D:} etc.) produces its value and can be changed
1505: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
1506: produces its address (which becomes invalid when the variable's scope is
1507: left). E.g., the standard word @code{emit} can be defined in therms of
1508: @code{type} like this:
1509:
1510: @example
1511: : emit @{ C^ char* -- @}
1512: char* 1 type ;
1513: @end example
1514:
1515: A local without type specifier is a @code{W:} local. Both flavours of
1516: locals are initialized with values from the data or FP stack.
1517:
1518: Currently there is no way to define locals with user-defined data
1519: structures, but we are working on it.
1520:
1.17 anton 1521: Gforth allows defining locals everywhere in a colon definition. This
1.7 pazsan 1522: poses the following questions:
1.2 anton 1523:
1.4 anton 1524: @menu
1525: * Where are locals visible by name?::
1.14 anton 1526: * How long do locals live?::
1.4 anton 1527: * Programming Style::
1528: * Implementation::
1529: @end menu
1530:
1.17 anton 1531: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
1.2 anton 1532: @subsubsection Where are locals visible by name?
1533:
1534: Basically, the answer is that locals are visible where you would expect
1535: it in block-structured languages, and sometimes a little longer. If you
1536: want to restrict the scope of a local, enclose its definition in
1537: @code{SCOPE}...@code{ENDSCOPE}.
1538:
1539: doc-scope
1540: doc-endscope
1541:
1542: These words behave like control structure words, so you can use them
1543: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
1544: arbitrary ways.
1545:
1546: If you want a more exact answer to the visibility question, here's the
1547: basic principle: A local is visible in all places that can only be
1548: reached through the definition of the local@footnote{In compiler
1549: construction terminology, all places dominated by the definition of the
1550: local.}. In other words, it is not visible in places that can be reached
1551: without going through the definition of the local. E.g., locals defined
1552: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
1553: defined in @code{BEGIN}...@code{UNTIL} are visible after the
1554: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
1555:
1556: The reasoning behind this solution is: We want to have the locals
1557: visible as long as it is meaningful. The user can always make the
1558: visibility shorter by using explicit scoping. In a place that can
1559: only be reached through the definition of a local, the meaning of a
1560: local name is clear. In other places it is not: How is the local
1561: initialized at the control flow path that does not contain the
1562: definition? Which local is meant, if the same name is defined twice in
1563: two independent control flow paths?
1564:
1565: This should be enough detail for nearly all users, so you can skip the
1566: rest of this section. If you relly must know all the gory details and
1567: options, read on.
1568:
1569: In order to implement this rule, the compiler has to know which places
1570: are unreachable. It knows this automatically after @code{AHEAD},
1571: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
1572: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
1573: compiler that the control flow never reaches that place. If
1574: @code{UNREACHABLE} is not used where it could, the only consequence is
1575: that the visibility of some locals is more limited than the rule above
1576: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
1577: lie to the compiler), buggy code will be produced.
1578:
1579: Another problem with this rule is that at @code{BEGIN}, the compiler
1.3 anton 1580: does not know which locals will be visible on the incoming
1581: back-edge. All problems discussed in the following are due to this
1582: ignorance of the compiler (we discuss the problems using @code{BEGIN}
1583: loops as examples; the discussion also applies to @code{?DO} and other
1.2 anton 1584: loops). Perhaps the most insidious example is:
1585: @example
1586: AHEAD
1587: BEGIN
1588: x
1589: [ 1 CS-ROLL ] THEN
1.4 anton 1590: @{ x @}
1.2 anton 1591: ...
1592: UNTIL
1593: @end example
1594:
1595: This should be legal according to the visibility rule. The use of
1596: @code{x} can only be reached through the definition; but that appears
1597: textually below the use.
1598:
1599: From this example it is clear that the visibility rules cannot be fully
1600: implemented without major headaches. Our implementation treats common
1601: cases as advertised and the exceptions are treated in a safe way: The
1602: compiler makes a reasonable guess about the locals visible after a
1603: @code{BEGIN}; if it is too pessimistic, the
1604: user will get a spurious error about the local not being defined; if the
1605: compiler is too optimistic, it will notice this later and issue a
1606: warning. In the case above the compiler would complain about @code{x}
1607: being undefined at its use. You can see from the obscure examples in
1608: this section that it takes quite unusual control structures to get the
1609: compiler into trouble, and even then it will often do fine.
1610:
1611: If the @code{BEGIN} is reachable from above, the most optimistic guess
1612: is that all locals visible before the @code{BEGIN} will also be
1613: visible after the @code{BEGIN}. This guess is valid for all loops that
1614: are entered only through the @code{BEGIN}, in particular, for normal
1615: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
1616: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
1617: compiler. When the branch to the @code{BEGIN} is finally generated by
1618: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
1619: warns the user if it was too optimisitic:
1620: @example
1621: IF
1.4 anton 1622: @{ x @}
1.2 anton 1623: BEGIN
1624: \ x ?
1625: [ 1 cs-roll ] THEN
1626: ...
1627: UNTIL
1628: @end example
1629:
1630: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
1631: optimistically assumes that it lives until the @code{THEN}. It notices
1632: this difference when it compiles the @code{UNTIL} and issues a
1633: warning. The user can avoid the warning, and make sure that @code{x}
1634: is not used in the wrong area by using explicit scoping:
1635: @example
1636: IF
1637: SCOPE
1.4 anton 1638: @{ x @}
1.2 anton 1639: ENDSCOPE
1640: BEGIN
1641: [ 1 cs-roll ] THEN
1642: ...
1643: UNTIL
1644: @end example
1645:
1646: Since the guess is optimistic, there will be no spurious error messages
1647: about undefined locals.
1648:
1649: If the @code{BEGIN} is not reachable from above (e.g., after
1650: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
1651: optimistic guess, as the locals visible after the @code{BEGIN} may be
1652: defined later. Therefore, the compiler assumes that no locals are
1.17 anton 1653: visible after the @code{BEGIN}. However, the user can use
1.2 anton 1654: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
1.17 anton 1655: visible at the BEGIN as at the point where the top control-flow stack
1656: item was created.
1.2 anton 1657:
1658: doc-assume-live
1659:
1660: E.g.,
1661: @example
1.4 anton 1662: @{ x @}
1.2 anton 1663: AHEAD
1664: ASSUME-LIVE
1665: BEGIN
1666: x
1667: [ 1 CS-ROLL ] THEN
1668: ...
1669: UNTIL
1670: @end example
1671:
1672: Other cases where the locals are defined before the @code{BEGIN} can be
1673: handled by inserting an appropriate @code{CS-ROLL} before the
1674: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
1675: behind the @code{ASSUME-LIVE}).
1676:
1677: Cases where locals are defined after the @code{BEGIN} (but should be
1678: visible immediately after the @code{BEGIN}) can only be handled by
1679: rearranging the loop. E.g., the ``most insidious'' example above can be
1680: arranged into:
1681: @example
1682: BEGIN
1.4 anton 1683: @{ x @}
1.2 anton 1684: ... 0=
1685: WHILE
1686: x
1687: REPEAT
1688: @end example
1689:
1.17 anton 1690: @node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
1.2 anton 1691: @subsubsection How long do locals live?
1692:
1693: The right answer for the lifetime question would be: A local lives at
1694: least as long as it can be accessed. For a value-flavoured local this
1695: means: until the end of its visibility. However, a variable-flavoured
1696: local could be accessed through its address far beyond its visibility
1697: scope. Ultimately, this would mean that such locals would have to be
1698: garbage collected. Since this entails un-Forth-like implementation
1699: complexities, I adopted the same cowardly solution as some other
1700: languages (e.g., C): The local lives only as long as it is visible;
1701: afterwards its address is invalid (and programs that access it
1702: afterwards are erroneous).
1703:
1.17 anton 1704: @node Programming Style, Implementation, How long do locals live?, Gforth locals
1.2 anton 1705: @subsubsection Programming Style
1706:
1707: The freedom to define locals anywhere has the potential to change
1708: programming styles dramatically. In particular, the need to use the
1709: return stack for intermediate storage vanishes. Moreover, all stack
1710: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
1711: determined arguments) can be eliminated: If the stack items are in the
1712: wrong order, just write a locals definition for all of them; then
1713: write the items in the order you want.
1714:
1715: This seems a little far-fetched and eliminating stack manipulations is
1.4 anton 1716: unlikely to become a conscious programming objective. Still, the number
1717: of stack manipulations will be reduced dramatically if local variables
1.17 anton 1718: are used liberally (e.g., compare @code{max} in @ref{Gforth locals} with
1.4 anton 1719: a traditional implementation of @code{max}).
1.2 anton 1720:
1721: This shows one potential benefit of locals: making Forth programs more
1722: readable. Of course, this benefit will only be realized if the
1723: programmers continue to honour the principle of factoring instead of
1724: using the added latitude to make the words longer.
1725:
1726: Using @code{TO} can and should be avoided. Without @code{TO},
1727: every value-flavoured local has only a single assignment and many
1728: advantages of functional languages apply to Forth. I.e., programs are
1729: easier to analyse, to optimize and to read: It is clear from the
1730: definition what the local stands for, it does not turn into something
1731: different later.
1732:
1733: E.g., a definition using @code{TO} might look like this:
1734: @example
1735: : strcmp @{ addr1 u1 addr2 u2 -- n @}
1736: u1 u2 min 0
1737: ?do
1738: addr1 c@ addr2 c@ - ?dup
1739: if
1740: unloop exit
1741: then
1742: addr1 char+ TO addr1
1743: addr2 char+ TO addr2
1744: loop
1745: u1 u2 - ;
1746: @end example
1747: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
1748: every loop iteration. @code{strcmp} is a typical example of the
1749: readability problems of using @code{TO}. When you start reading
1750: @code{strcmp}, you think that @code{addr1} refers to the start of the
1751: string. Only near the end of the loop you realize that it is something
1752: else.
1753:
1754: This can be avoided by defining two locals at the start of the loop that
1755: are initialized with the right value for the current iteration.
1756: @example
1757: : strcmp @{ addr1 u1 addr2 u2 -- n @}
1758: addr1 addr2
1759: u1 u2 min 0
1760: ?do @{ s1 s2 @}
1761: s1 c@ s2 c@ - ?dup
1762: if
1763: unloop exit
1764: then
1765: s1 char+ s2 char+
1766: loop
1767: 2drop
1768: u1 u2 - ;
1769: @end example
1770: Here it is clear from the start that @code{s1} has a different value
1771: in every loop iteration.
1772:
1.17 anton 1773: @node Implementation, , Programming Style, Gforth locals
1.2 anton 1774: @subsubsection Implementation
1775:
1.17 anton 1776: Gforth uses an extra locals stack. The most compelling reason for
1.2 anton 1777: this is that the return stack is not float-aligned; using an extra stack
1778: also eliminates the problems and restrictions of using the return stack
1779: as locals stack. Like the other stacks, the locals stack grows toward
1780: lower addresses. A few primitives allow an efficient implementation:
1781:
1782: doc-@local#
1783: doc-f@local#
1784: doc-laddr#
1785: doc-lp+!#
1786: doc-lp!
1787: doc->l
1788: doc-f>l
1789:
1790: In addition to these primitives, some specializations of these
1791: primitives for commonly occurring inline arguments are provided for
1792: efficiency reasons, e.g., @code{@@local0} as specialization of
1793: @code{@@local#} for the inline argument 0. The following compiling words
1794: compile the right specialized version, or the general version, as
1795: appropriate:
1796:
1.12 anton 1797: doc-compile-@local
1798: doc-compile-f@local
1.2 anton 1799: doc-compile-lp+!
1800:
1801: Combinations of conditional branches and @code{lp+!#} like
1802: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
1803: is taken) are provided for efficiency and correctness in loops.
1804:
1805: A special area in the dictionary space is reserved for keeping the
1806: local variable names. @code{@{} switches the dictionary pointer to this
1807: area and @code{@}} switches it back and generates the locals
1808: initializing code. @code{W:} etc.@ are normal defining words. This
1809: special area is cleared at the start of every colon definition.
1810:
1.17 anton 1811: A special feature of Gforth's dictionary is used to implement the
1.2 anton 1812: definition of locals without type specifiers: every wordlist (aka
1813: vocabulary) has its own methods for searching
1.4 anton 1814: etc. (@pxref{Wordlists}). For the present purpose we defined a wordlist
1.2 anton 1815: with a special search method: When it is searched for a word, it
1816: actually creates that word using @code{W:}. @code{@{} changes the search
1817: order to first search the wordlist containing @code{@}}, @code{W:} etc.,
1818: and then the wordlist for defining locals without type specifiers.
1819:
1820: The lifetime rules support a stack discipline within a colon
1821: definition: The lifetime of a local is either nested with other locals
1822: lifetimes or it does not overlap them.
1823:
1824: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
1825: pointer manipulation is generated. Between control structure words
1826: locals definitions can push locals onto the locals stack. @code{AGAIN}
1827: is the simplest of the other three control flow words. It has to
1828: restore the locals stack depth of the corresponding @code{BEGIN}
1829: before branching. The code looks like this:
1830: @format
1831: @code{lp+!#} current-locals-size @minus{} dest-locals-size
1832: @code{branch} <begin>
1833: @end format
1834:
1835: @code{UNTIL} is a little more complicated: If it branches back, it
1836: must adjust the stack just like @code{AGAIN}. But if it falls through,
1837: the locals stack must not be changed. The compiler generates the
1838: following code:
1839: @format
1840: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
1841: @end format
1842: The locals stack pointer is only adjusted if the branch is taken.
1843:
1844: @code{THEN} can produce somewhat inefficient code:
1845: @format
1846: @code{lp+!#} current-locals-size @minus{} orig-locals-size
1847: <orig target>:
1848: @code{lp+!#} orig-locals-size @minus{} new-locals-size
1849: @end format
1850: The second @code{lp+!#} adjusts the locals stack pointer from the
1.4 anton 1851: level at the @var{orig} point to the level after the @code{THEN}. The
1.2 anton 1852: first @code{lp+!#} adjusts the locals stack pointer from the current
1853: level to the level at the orig point, so the complete effect is an
1854: adjustment from the current level to the right level after the
1855: @code{THEN}.
1856:
1857: In a conventional Forth implementation a dest control-flow stack entry
1858: is just the target address and an orig entry is just the address to be
1859: patched. Our locals implementation adds a wordlist to every orig or dest
1860: item. It is the list of locals visible (or assumed visible) at the point
1861: described by the entry. Our implementation also adds a tag to identify
1862: the kind of entry, in particular to differentiate between live and dead
1863: (reachable and unreachable) orig entries.
1864:
1865: A few unusual operations have to be performed on locals wordlists:
1866:
1867: doc-common-list
1868: doc-sub-list?
1869: doc-list-size
1870:
1871: Several features of our locals wordlist implementation make these
1872: operations easy to implement: The locals wordlists are organised as
1873: linked lists; the tails of these lists are shared, if the lists
1874: contain some of the same locals; and the address of a name is greater
1875: than the address of the names behind it in the list.
1876:
1877: Another important implementation detail is the variable
1878: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
1879: determine if they can be reached directly or only through the branch
1880: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
1881: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
1882: definition, by @code{BEGIN} and usually by @code{THEN}.
1883:
1884: Counted loops are similar to other loops in most respects, but
1885: @code{LEAVE} requires special attention: It performs basically the same
1886: service as @code{AHEAD}, but it does not create a control-flow stack
1887: entry. Therefore the information has to be stored elsewhere;
1888: traditionally, the information was stored in the target fields of the
1889: branches created by the @code{LEAVE}s, by organizing these fields into a
1890: linked list. Unfortunately, this clever trick does not provide enough
1891: space for storing our extended control flow information. Therefore, we
1892: introduce another stack, the leave stack. It contains the control-flow
1893: stack entries for all unresolved @code{LEAVE}s.
1894:
1895: Local names are kept until the end of the colon definition, even if
1896: they are no longer visible in any control-flow path. In a few cases
1897: this may lead to increased space needs for the locals name area, but
1898: usually less than reclaiming this space would cost in code size.
1899:
1900:
1.17 anton 1901: @node ANS Forth locals, , Gforth locals, Locals
1.2 anton 1902: @subsection ANS Forth locals
1903:
1904: The ANS Forth locals wordset does not define a syntax for locals, but
1905: words that make it possible to define various syntaxes. One of the
1.17 anton 1906: possible syntaxes is a subset of the syntax we used in the Gforth locals
1.2 anton 1907: wordset, i.e.:
1908:
1909: @example
1910: @{ local1 local2 ... -- comment @}
1911: @end example
1912: or
1913: @example
1914: @{ local1 local2 ... @}
1915: @end example
1916:
1917: The order of the locals corresponds to the order in a stack comment. The
1918: restrictions are:
1.1 anton 1919:
1.2 anton 1920: @itemize @bullet
1921: @item
1.17 anton 1922: Locals can only be cell-sized values (no type specifiers are allowed).
1.2 anton 1923: @item
1924: Locals can be defined only outside control structures.
1925: @item
1926: Locals can interfere with explicit usage of the return stack. For the
1927: exact (and long) rules, see the standard. If you don't use return stack
1.17 anton 1928: accessing words in a definition using locals, you will be all right. The
1.2 anton 1929: purpose of this rule is to make locals implementation on the return
1930: stack easier.
1931: @item
1932: The whole definition must be in one line.
1933: @end itemize
1934:
1935: Locals defined in this way behave like @code{VALUE}s
1.4 anton 1936: (@xref{Values}). I.e., they are initialized from the stack. Using their
1.2 anton 1937: name produces their value. Their value can be changed using @code{TO}.
1938:
1.17 anton 1939: Since this syntax is supported by Gforth directly, you need not do
1.2 anton 1940: anything to use it. If you want to port a program using this syntax to
1941: another ANS Forth system, use @file{anslocal.fs} to implement the syntax
1942: on the other system.
1943:
1944: Note that a syntax shown in the standard, section A.13 looks
1945: similar, but is quite different in having the order of locals
1946: reversed. Beware!
1947:
1948: The ANS Forth locals wordset itself consists of the following word
1949:
1950: doc-(local)
1951:
1952: The ANS Forth locals extension wordset defines a syntax, but it is so
1953: awful that we strongly recommend not to use it. We have implemented this
1.17 anton 1954: syntax to make porting to Gforth easy, but do not document it here. The
1.2 anton 1955: problem with this syntax is that the locals are defined in an order
1956: reversed with respect to the standard stack comment notation, making
1957: programs harder to read, and easier to misread and miswrite. The only
1958: merit of this syntax is that it is easy to implement using the ANS Forth
1959: locals wordset.
1.3 anton 1960:
1.4 anton 1961: @node Defining Words, Wordlists, Locals, Words
1962: @section Defining Words
1963:
1.14 anton 1964: @menu
1965: * Values::
1966: @end menu
1967:
1.4 anton 1968: @node Values, , Defining Words, Defining Words
1969: @subsection Values
1970:
1971: @node Wordlists, Files, Defining Words, Words
1972: @section Wordlists
1973:
1974: @node Files, Blocks, Wordlists, Words
1975: @section Files
1976:
1977: @node Blocks, Other I/O, Files, Words
1978: @section Blocks
1979:
1980: @node Other I/O, Programming Tools, Blocks, Words
1981: @section Other I/O
1982:
1.18 anton 1983: @node Programming Tools, Assembler and Code words, Other I/O, Words
1.4 anton 1984: @section Programming Tools
1985:
1.5 anton 1986: @menu
1987: * Debugging:: Simple and quick.
1988: * Assertions:: Making your programs self-checking.
1989: @end menu
1990:
1991: @node Debugging, Assertions, Programming Tools, Programming Tools
1.4 anton 1992: @subsection Debugging
1993:
1994: The simple debugging aids provided in @file{debugging.fs}
1995: are meant to support a different style of debugging than the
1996: tracing/stepping debuggers used in languages with long turn-around
1997: times.
1998:
1999: A much better (faster) way in fast-compilig languages is to add
2000: printing code at well-selected places, let the program run, look at
2001: the output, see where things went wrong, add more printing code, etc.,
2002: until the bug is found.
2003:
2004: The word @code{~~} is easy to insert. It just prints debugging
2005: information (by default the source location and the stack contents). It
2006: is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
2007: query-replace them with nothing). The deferred words
2008: @code{printdebugdata} and @code{printdebugline} control the output of
2009: @code{~~}. The default source location output format works well with
2010: Emacs' compilation mode, so you can step through the program at the
1.5 anton 2011: source level using @kbd{C-x `} (the advantage over a stepping debugger
2012: is that you can step in any direction and you know where the crash has
2013: happened or where the strange data has occurred).
1.4 anton 2014:
2015: Note that the default actions clobber the contents of the pictured
2016: numeric output string, so you should not use @code{~~}, e.g., between
2017: @code{<#} and @code{#>}.
2018:
2019: doc-~~
2020: doc-printdebugdata
2021: doc-printdebugline
2022:
1.5 anton 2023: @node Assertions, , Debugging, Programming Tools
1.4 anton 2024: @subsection Assertions
2025:
1.5 anton 2026: It is a good idea to make your programs self-checking, in particular, if
2027: you use an assumption (e.g., that a certain field of a data structure is
1.17 anton 2028: never zero) that may become wrong during maintenance. Gforth supports
1.5 anton 2029: assertions for this purpose. They are used like this:
2030:
2031: @example
2032: assert( @var{flag} )
2033: @end example
2034:
2035: The code between @code{assert(} and @code{)} should compute a flag, that
2036: should be true if everything is alright and false otherwise. It should
2037: not change anything else on the stack. The overall stack effect of the
2038: assertion is @code{( -- )}. E.g.
2039:
2040: @example
2041: assert( 1 1 + 2 = ) \ what we learn in school
2042: assert( dup 0<> ) \ assert that the top of stack is not zero
2043: assert( false ) \ this code should not be reached
2044: @end example
2045:
2046: The need for assertions is different at different times. During
2047: debugging, we want more checking, in production we sometimes care more
2048: for speed. Therefore, assertions can be turned off, i.e., the assertion
2049: becomes a comment. Depending on the importance of an assertion and the
2050: time it takes to check it, you may want to turn off some assertions and
1.17 anton 2051: keep others turned on. Gforth provides several levels of assertions for
1.5 anton 2052: this purpose:
2053:
2054: doc-assert0(
2055: doc-assert1(
2056: doc-assert2(
2057: doc-assert3(
2058: doc-assert(
2059: doc-)
2060:
2061: @code{Assert(} is the same as @code{assert1(}. The variable
2062: @code{assert-level} specifies the highest assertions that are turned
2063: on. I.e., at the default @code{assert-level} of one, @code{assert0(} and
2064: @code{assert1(} assertions perform checking, while @code{assert2(} and
2065: @code{assert3(} assertions are treated as comments.
2066:
2067: Note that the @code{assert-level} is evaluated at compile-time, not at
2068: run-time. I.e., you cannot turn assertions on or off at run-time, you
2069: have to set the @code{assert-level} appropriately before compiling a
2070: piece of code. You can compile several pieces of code at several
2071: @code{assert-level}s (e.g., a trusted library at level 1 and newly
2072: written code at level 3).
2073:
2074: doc-assert-level
2075:
2076: If an assertion fails, a message compatible with Emacs' compilation mode
2077: is produced and the execution is aborted (currently with @code{ABORT"}.
2078: If there is interest, we will introduce a special throw code. But if you
2079: intend to @code{catch} a specific condition, using @code{throw} is
2080: probably more appropriate than an assertion).
2081:
1.18 anton 2082: @node Assembler and Code words, Threading Words, Programming Tools, Words
2083: @section Assembler and Code words
2084:
2085: Gforth provides some words for defining primitives (words written in
2086: machine code), and for defining the the machine-code equivalent of
2087: @code{DOES>}-based defining words. However, the machine-independent
2088: nature of Gforth poses a few problems: First of all. Gforth runs on
2089: several architectures, so it can provide no standard assembler. What's
2090: worse is that the register allocation not only depends on the processor,
2091: but also on the gcc version and options used.
2092:
2093: The words Gforth offers encapsulate some system dependences (e.g., the
2094: header structure), so a system-independent assembler may be used in
2095: Gforth. If you do not have an assembler, you can compile machine code
2096: directly with @code{,} and @code{c,}.
2097:
2098: doc-assembler
2099: doc-code
2100: doc-end-code
2101: doc-;code
2102: doc-flush-icache
2103:
2104: If @code{flush-icache} does not work correctly, @code{code} words
2105: etc. will not work (reliably), either.
2106:
2107: These words are rarely used. Therefore they reside in @code{code.fs},
2108: which is usually not loaded (except @code{flush-icache}, which is always
1.19 anton 2109: present). You can load them with @code{require code.fs}.
1.18 anton 2110:
2111: Another option for implementing normal and defining words efficiently
2112: is: adding the wanted functionality to the source of Gforth. For normal
2113: words you just have to edit @file{primitives}, defining words (for fast
2114: defined words) probably require changes in @file{engine.c},
2115: @file{kernal.fs}, @file{prims2x.fs}, and possibly @file{cross.fs}.
2116:
2117:
2118: @node Threading Words, , Assembler and Code words, Words
1.4 anton 2119: @section Threading Words
2120:
2121: These words provide access to code addresses and other threading stuff
1.17 anton 2122: in Gforth (and, possibly, other interpretive Forths). It more or less
1.4 anton 2123: abstracts away the differences between direct and indirect threading
2124: (and, for direct threading, the machine dependences). However, at
2125: present this wordset is still inclomplete. It is also pretty low-level;
2126: some day it will hopefully be made unnecessary by an internals words set
2127: that abstracts implementation details away completely.
2128:
2129: doc->code-address
2130: doc->does-code
2131: doc-code-address!
2132: doc-does-code!
2133: doc-does-handler!
2134: doc-/does-handler
2135:
1.18 anton 2136: The code addresses produced by various defining words are produced by
2137: the following words:
1.14 anton 2138:
1.18 anton 2139: doc-docol:
2140: doc-docon:
2141: doc-dovar:
2142: doc-douser:
2143: doc-dodefer:
2144: doc-dofield:
2145:
2146: Currently there is no installation-independent way for recogizing words
2147: defined by a @code{CREATE}...@code{DOES>} word; however, once you know
2148: that a word is defined by a @code{CREATE}...@code{DOES>} word, you can
2149: use @code{>DOES-CODE}.
1.14 anton 2150:
1.4 anton 2151: @node ANS conformance, Model, Words, Top
2152: @chapter ANS conformance
2153:
1.17 anton 2154: To the best of our knowledge, Gforth is an
1.14 anton 2155:
1.15 anton 2156: ANS Forth System
2157: @itemize
2158: @item providing the Core Extensions word set
2159: @item providing the Block word set
2160: @item providing the Block Extensions word set
2161: @item providing the Double-Number word set
2162: @item providing the Double-Number Extensions word set
2163: @item providing the Exception word set
2164: @item providing the Exception Extensions word set
2165: @item providing the Facility word set
2166: @item providing @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
2167: @item providing the File Access word set
2168: @item providing the File Access Extensions word set
2169: @item providing the Floating-Point word set
2170: @item providing the Floating-Point Extensions word set
2171: @item providing the Locals word set
2172: @item providing the Locals Extensions word set
2173: @item providing the Memory-Allocation word set
2174: @item providing the Memory-Allocation Extensions word set (that one's easy)
2175: @item providing the Programming-Tools word set
1.18 anton 2176: @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
1.15 anton 2177: @item providing the Search-Order word set
2178: @item providing the Search-Order Extensions word set
2179: @item providing the String word set
2180: @item providing the String Extensions word set (another easy one)
2181: @end itemize
2182:
2183: In addition, ANS Forth systems are required to document certain
2184: implementation choices. This chapter tries to meet these
2185: requirements. In many cases it gives a way to ask the system for the
2186: information instead of providing the information directly, in
2187: particular, if the information depends on the processor, the operating
2188: system or the installation options chosen, or if they are likely to
1.17 anton 2189: change during the maintenance of Gforth.
1.15 anton 2190:
1.14 anton 2191: @comment The framework for the rest has been taken from pfe.
2192:
2193: @menu
2194: * The Core Words::
2195: * The optional Block word set::
2196: * The optional Double Number word set::
2197: * The optional Exception word set::
2198: * The optional Facility word set::
2199: * The optional File-Access word set::
2200: * The optional Floating-Point word set::
2201: * The optional Locals word set::
2202: * The optional Memory-Allocation word set::
2203: * The optional Programming-Tools word set::
2204: * The optional Search-Order word set::
2205: @end menu
2206:
2207:
2208: @c =====================================================================
2209: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
2210: @comment node-name, next, previous, up
2211: @section The Core Words
2212: @c =====================================================================
2213:
2214: @menu
1.15 anton 2215: * core-idef:: Implementation Defined Options
2216: * core-ambcond:: Ambiguous Conditions
2217: * core-other:: Other System Documentation
1.14 anton 2218: @end menu
2219:
2220: @c ---------------------------------------------------------------------
2221: @node core-idef, core-ambcond, The Core Words, The Core Words
2222: @subsection Implementation Defined Options
2223: @c ---------------------------------------------------------------------
2224:
2225: @table @i
2226:
2227: @item (Cell) aligned addresses:
1.17 anton 2228: processor-dependent. Gforth's alignment words perform natural alignment
1.14 anton 2229: (e.g., an address aligned for a datum of size 8 is divisible by
2230: 8). Unaligned accesses usually result in a @code{-23 THROW}.
2231:
2232: @item @code{EMIT} and non-graphic characters:
2233: The character is output using the C library function (actually, macro)
2234: @code{putchar}.
2235:
2236: @item character editing of @code{ACCEPT} and @code{EXPECT}:
2237: This is modeled on the GNU readline library (@pxref{Readline
2238: Interaction, , Command Line Editing, readline, The GNU Readline
2239: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
2240: producing a full word completion every time you type it (instead of
2241: producing the common prefix of all completions).
2242:
2243: @item character set:
2244: The character set of your computer and display device. Gforth is
2245: 8-bit-clean (but some other component in your system may make trouble).
2246:
2247: @item Character-aligned address requirements:
2248: installation-dependent. Currently a character is represented by a C
2249: @code{unsigned char}; in the future we might switch to @code{wchar_t}
2250: (Comments on that requested).
2251:
2252: @item character-set extensions and matching of names:
1.17 anton 2253: Any character except the ASCII NUL charcter can be used in a
2254: name. Matching is case-insensitive. The matching is performed using the
2255: C function @code{strncasecmp}, whose function is probably influenced by
2256: the locale. E.g., the @code{C} locale does not know about accents and
1.14 anton 2257: umlauts, so they are matched case-sensitively in that locale. For
2258: portability reasons it is best to write programs such that they work in
2259: the @code{C} locale. Then one can use libraries written by a Polish
2260: programmer (who might use words containing ISO Latin-2 encoded
2261: characters) and by a French programmer (ISO Latin-1) in the same program
2262: (of course, @code{WORDS} will produce funny results for some of the
2263: words (which ones, depends on the font you are using)). Also, the locale
2264: you prefer may not be available in other operating systems. Hopefully,
2265: Unicode will solve these problems one day.
2266:
2267: @item conditions under which control characters match a space delimiter:
2268: If @code{WORD} is called with the space character as a delimiter, all
2269: white-space characters (as identified by the C macro @code{isspace()})
2270: are delimiters. @code{PARSE}, on the other hand, treats space like other
2271: delimiters. @code{PARSE-WORD} treats space like @code{WORD}, but behaves
2272: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
2273: interpreter (aka text interpreter) by default, treats all white-space
2274: characters as delimiters.
2275:
2276: @item format of the control flow stack:
2277: The data stack is used as control flow stack. The size of a control flow
2278: stack item in cells is given by the constant @code{cs-item-size}. At the
2279: time of this writing, an item consists of a (pointer to a) locals list
2280: (third), an address in the code (second), and a tag for identifying the
2281: item (TOS). The following tags are used: @code{defstart},
2282: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
2283: @code{scopestart}.
2284:
2285: @item conversion of digits > 35
2286: The characters @code{[\]^_'} are the digits with the decimal value
2287: 36@minus{}41. There is no way to input many of the larger digits.
2288:
2289: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
2290: The cursor is moved to the end of the entered string. If the input is
2291: terminated using the @kbd{Return} key, a space is typed.
2292:
2293: @item exception abort sequence of @code{ABORT"}:
2294: The error string is stored into the variable @code{"error} and a
2295: @code{-2 throw} is performed.
2296:
2297: @item input line terminator:
2298: For interactive input, @kbd{C-m} and @kbd{C-j} terminate lines. One of
2299: these characters is typically produced when you type the @kbd{Enter} or
2300: @kbd{Return} key.
2301:
2302: @item maximum size of a counted string:
2303: @code{s" /counted-string" environment? drop .}. Currently 255 characters
2304: on all ports, but this may change.
2305:
2306: @item maximum size of a parsed string:
2307: Given by the constant @code{/line}. Currently 255 characters.
2308:
2309: @item maximum size of a definition name, in characters:
2310: 31
2311:
2312: @item maximum string length for @code{ENVIRONMENT?}, in characters:
2313: 31
2314:
2315: @item method of selecting the user input device:
1.17 anton 2316: The user input device is the standard input. There is currently no way to
2317: change it from within Gforth. However, the input can typically be
2318: redirected in the command line that starts Gforth.
1.14 anton 2319:
2320: @item method of selecting the user output device:
2321: The user output device is the standard output. It cannot be redirected
1.17 anton 2322: from within Gforth, but typically from the command line that starts
2323: Gforth. Gforth uses buffered output, so output on a terminal does not
1.14 anton 2324: become visible before the next newline or buffer overflow. Output on
2325: non-terminals is invisible until the buffer overflows.
2326:
2327: @item methods of dictionary compilation:
1.17 anton 2328: What are we expected to document here?
1.14 anton 2329:
2330: @item number of bits in one address unit:
2331: @code{s" address-units-bits" environment? drop .}. 8 in all current
2332: ports.
2333:
2334: @item number representation and arithmetic:
2335: Processor-dependent. Binary two's complement on all current ports.
2336:
2337: @item ranges for integer types:
2338: Installation-dependent. Make environmental queries for @code{MAX-N},
2339: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
2340: unsigned (and positive) types is 0. The lower bound for signed types on
2341: two's complement and one's complement machines machines can be computed
2342: by adding 1 to the upper bound.
2343:
2344: @item read-only data space regions:
2345: The whole Forth data space is writable.
2346:
2347: @item size of buffer at @code{WORD}:
2348: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
2349: shared with the pictured numeric output string. If overwriting
2350: @code{PAD} is acceptable, it is as large as the remaining dictionary
2351: space, although only as much can be sensibly used as fits in a counted
2352: string.
2353:
2354: @item size of one cell in address units:
2355: @code{1 cells .}.
2356:
2357: @item size of one character in address units:
2358: @code{1 chars .}. 1 on all current ports.
2359:
2360: @item size of the keyboard terminal buffer:
2361: Varies. You can determine the size at a specific time using @code{lp@
2362: tib - .}. It is shared with the locals stack and TIBs of files that
2363: include the current file. You can change the amount of space for TIBs
1.17 anton 2364: and locals stack at Gforth startup with the command line option
1.14 anton 2365: @code{-l}.
2366:
2367: @item size of the pictured numeric output buffer:
2368: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
2369: shared with @code{WORD}.
2370:
2371: @item size of the scratch area returned by @code{PAD}:
2372: The remainder of dictionary space. You can even use the unused part of
2373: the data stack space. The current size can be computed with @code{sp@
2374: pad - .}.
2375:
2376: @item system case-sensitivity characteristics:
2377: Dictionary searches are case insensitive. However, as explained above
2378: under @i{character-set extensions}, the matching for non-ASCII
2379: characters is determined by the locale you are using. In the default
2380: @code{C} locale all non-ASCII characters are matched case-sensitively.
2381:
2382: @item system prompt:
2383: @code{ ok} in interpret state, @code{ compiled} in compile state.
2384:
2385: @item division rounding:
2386: installation dependent. @code{s" floored" environment? drop .}. We leave
2387: the choice to gcc (what to use for @code{/}) and to you (whether to use
2388: @code{fm/mod}, @code{sm/rem} or simply @code{/}).
2389:
2390: @item values of @code{STATE} when true:
2391: -1.
2392:
2393: @item values returned after arithmetic overflow:
2394: On two's complement machines, arithmetic is performed modulo
2395: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
2396: arithmetic (with appropriate mapping for signed types). Division by zero
2397: typically results in a @code{-55 throw} (floatingpoint unidentified
2398: fault), although a @code{-10 throw} (divide by zero) would be more
2399: appropriate.
2400:
2401: @item whether the current definition can be found after @t{DOES>}:
2402: No.
2403:
2404: @end table
2405:
2406: @c ---------------------------------------------------------------------
2407: @node core-ambcond, core-other, core-idef, The Core Words
2408: @subsection Ambiguous conditions
2409: @c ---------------------------------------------------------------------
2410:
2411: @table @i
2412:
2413: @item a name is neither a word nor a number:
2414: @code{-13 throw} (Undefined word)
2415:
2416: @item a definition name exceeds the maximum length allowed:
2417: @code{-19 throw} (Word name too long)
2418:
2419: @item addressing a region not inside the various data spaces of the forth system:
2420: The stacks, code space and name space are accessible. Machine code space is
2421: typically readable. Accessing other addresses gives results dependent on
2422: the operating system. On decent systems: @code{-9 throw} (Invalid memory
2423: address).
2424:
2425: @item argument type incompatible with parameter:
2426: This is usually not caught. Some words perform checks, e.g., the control
2427: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
2428: mismatch).
2429:
2430: @item attempting to obtain the execution token of a word with undefined execution semantics:
2431: You get an execution token representing the compilation semantics
2432: instead.
2433:
2434: @item dividing by zero:
2435: typically results in a @code{-55 throw} (floating point unidentified
2436: fault), although a @code{-10 throw} (divide by zero) would be more
2437: appropriate.
2438:
2439: @item insufficient data stack or return stack space:
2440: Not checked. This typically results in mysterious illegal memory
2441: accesses, producing @code{-9 throw} (Invalid memory address) or
2442: @code{-23 throw} (Address alignment exception).
2443:
2444: @item insufficient space for loop control parameters:
2445: like other return stack overflows.
2446:
2447: @item insufficient space in the dictionary:
2448: Not checked. Similar results as stack overflows. However, typically the
2449: error appears at a different place when one inserts or removes code.
2450:
2451: @item interpreting a word with undefined interpretation semantics:
2452: For some words, we defined interpretation semantics. For the others:
2453: @code{-14 throw} (Interpreting a compile-only word). Note that this is
2454: checked only by the outer (aka text) interpreter; if the word is
2455: @code{execute}d in some other way, it will typically perform it's
2456: compilation semantics even in interpret state. (We could change @code{'}
2457: and relatives not to give the xt of such words, but we think that would
2458: be too restrictive).
2459:
2460: @item modifying the contents of the input buffer or a string literal:
2461: These are located in writable memory and can be modified.
2462:
2463: @item overflow of the pictured numeric output string:
2464: Not checked.
2465:
2466: @item parsed string overflow:
2467: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
2468:
2469: @item producing a result out of range:
2470: On two's complement machines, arithmetic is performed modulo
2471: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
2472: arithmetic (with appropriate mapping for signed types). Division by zero
2473: typically results in a @code{-55 throw} (floatingpoint unidentified
2474: fault), although a @code{-10 throw} (divide by zero) would be more
2475: appropriate. @code{convert} and @code{>number} currently overflow
2476: silently.
2477:
2478: @item reading from an empty data or return stack:
2479: The data stack is checked by the outer (aka text) interpreter after
2480: every word executed. If it has underflowed, a @code{-4 throw} (Stack
2481: underflow) is performed. Apart from that, the stacks are not checked and
2482: underflows can result in similar behaviour as overflows (of adjacent
2483: stacks).
2484:
2485: @item unexepected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
2486: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
2487: use zero-length string as a name). Words like @code{'} probably will not
2488: find what they search. Note that it is possible to create zero-length
2489: names with @code{nextname} (should it not?).
2490:
2491: @item @code{>IN} greater than input buffer:
2492: The next invocation of a parsing word returns a string wih length 0.
2493:
2494: @item @code{RECURSE} appears after @code{DOES>}:
2495: Compiles a recursive call to the defining word not to the defined word.
2496:
2497: @item argument input source different than current input source for @code{RESTORE-INPUT}:
2498: !!???If the argument input source is a valid input source then it gets
1.19 anton 2499: restored. Otherwise causes @code{-12 THROW}, which, unless caught, issues
1.14 anton 2500: the message "argument type mismatch" and aborts.
2501:
2502: @item data space containing definitions gets de-allocated:
2503: Deallocation with @code{allot} is not checked. This typically resuls in
2504: memory access faults or execution of illegal instructions.
2505:
2506: @item data space read/write with incorrect alignment:
2507: Processor-dependent. Typically results in a @code{-23 throw} (Address
2508: alignment exception). Under Linux on a 486 or later processor with
2509: alignment turned on, incorrect alignment results in a @code{-9 throw}
2510: (Invalid memory address). There are reportedly some processors with
2511: alignment restrictions that do not report them.
2512:
2513: @item data space pointer not properly aligned, @code{,}, @code{C,}:
2514: Like other alignment errors.
2515:
2516: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
2517: Not checked. May cause an illegal memory access.
2518:
2519: @item loop control parameters not available:
2520: Not checked. The counted loop words simply assume that the top of return
2521: stack items are loop control parameters and behave accordingly.
2522:
2523: @item most recent definition does not have a name (@code{IMMEDIATE}):
2524: @code{abort" last word was headerless"}.
2525:
2526: @item name not defined by @code{VALUE} used by @code{TO}:
2527: @code{-32 throw} (Invalid name argument)
2528:
1.15 anton 2529: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
1.14 anton 2530: @code{-13 throw} (Undefined word)
2531:
2532: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
2533: Gforth behaves as if they were of the same type. I.e., you can predict
2534: the behaviour by interpreting all parameters as, e.g., signed.
2535:
2536: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
2537: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} is equivalent to
2538: @code{TO}.
2539:
2540: @item String longer than a counted string returned by @code{WORD}:
2541: Not checked. The string will be ok, but the count will, of course,
2542: contain only the least significant bits of the length.
2543:
1.15 anton 2544: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
1.14 anton 2545: Processor-dependent. Typical behaviours are returning 0 and using only
2546: the low bits of the shift count.
2547:
2548: @item word not defined via @code{CREATE}:
2549: @code{>BODY} produces the PFA of the word no matter how it was defined.
2550:
2551: @code{DOES>} changes the execution semantics of the last defined word no
2552: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
2553: @code{CREATE , DOES>}.
2554:
2555: @item words improperly used outside @code{<#} and @code{#>}:
2556: Not checked. As usual, you can expect memory faults.
2557:
2558: @end table
2559:
2560:
2561: @c ---------------------------------------------------------------------
2562: @node core-other, , core-ambcond, The Core Words
2563: @subsection Other system documentation
2564: @c ---------------------------------------------------------------------
2565:
2566: @table @i
2567:
2568: @item nonstandard words using @code{PAD}:
2569: None.
2570:
2571: @item operator's terminal facilities available:
2572: !!??
2573:
2574: @item program data space available:
2575: @code{sp@ here - .} gives the space remaining for dictionary and data
2576: stack together.
2577:
2578: @item return stack space available:
2579: !!??
2580:
2581: @item stack space available:
2582: @code{sp@ here - .} gives the space remaining for dictionary and data
2583: stack together.
2584:
2585: @item system dictionary space required, in address units:
2586: Type @code{here forthstart - .} after startup. At the time of this
2587: writing, this gives 70108 (bytes) on a 32-bit system.
2588: @end table
2589:
2590:
2591: @c =====================================================================
2592: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
2593: @section The optional Block word set
2594: @c =====================================================================
2595:
2596: @menu
1.15 anton 2597: * block-idef:: Implementation Defined Options
2598: * block-ambcond:: Ambiguous Conditions
2599: * block-other:: Other System Documentation
1.14 anton 2600: @end menu
2601:
2602:
2603: @c ---------------------------------------------------------------------
2604: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
2605: @subsection Implementation Defined Options
2606: @c ---------------------------------------------------------------------
2607:
2608: @table @i
2609:
2610: @item the format for display by @code{LIST}:
2611: First the screen number is displayed, then 16 lines of 64 characters,
2612: each line preceded by the line number.
2613:
2614: @item the length of a line affected by @code{\}:
2615: 64 characters.
2616: @end table
2617:
2618:
2619: @c ---------------------------------------------------------------------
2620: @node block-ambcond, block-other, block-idef, The optional Block word set
2621: @subsection Ambiguous conditions
2622: @c ---------------------------------------------------------------------
2623:
2624: @table @i
2625:
2626: @item correct block read was not possible:
2627: Typically results in a @code{throw} of some OS-derived value (between
2628: -512 and -2048). If the blocks file was just not long enough, blanks are
2629: supplied for the missing portion.
2630:
2631: @item I/O exception in block transfer:
2632: Typically results in a @code{throw} of some OS-derived value (between
2633: -512 and -2048).
2634:
2635: @item invalid block number:
2636: @code{-35 throw} (Invalid block number)
2637:
2638: @item a program directly alters the contents of @code{BLK}:
2639: The input stream is switched to that other block, at the same
2640: position. If the storing to @code{BLK} happens when interpreting
2641: non-block input, the system will get quite confused when the block ends.
2642:
2643: @item no current block buffer for @code{UPDATE}:
2644: @code{UPDATE} has no effect.
2645:
2646: @end table
2647:
2648:
2649: @c ---------------------------------------------------------------------
2650: @node block-other, , block-ambcond, The optional Block word set
2651: @subsection Other system documentation
2652: @c ---------------------------------------------------------------------
2653:
2654: @table @i
2655:
2656: @item any restrictions a multiprogramming system places on the use of buffer addresses:
2657: No restrictions (yet).
2658:
2659: @item the number of blocks available for source and data:
2660: depends on your disk space.
2661:
2662: @end table
2663:
2664:
2665: @c =====================================================================
2666: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
2667: @section The optional Double Number word set
2668: @c =====================================================================
2669:
2670: @menu
1.15 anton 2671: * double-ambcond:: Ambiguous Conditions
1.14 anton 2672: @end menu
2673:
2674:
2675: @c ---------------------------------------------------------------------
1.15 anton 2676: @node double-ambcond, , The optional Double Number word set, The optional Double Number word set
1.14 anton 2677: @subsection Ambiguous conditions
2678: @c ---------------------------------------------------------------------
2679:
2680: @table @i
2681:
1.15 anton 2682: @item @var{d} outside of range of @var{n} in @code{D>S}:
1.14 anton 2683: The least significant cell of @var{d} is produced.
2684:
2685: @end table
2686:
2687:
2688: @c =====================================================================
2689: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
2690: @section The optional Exception word set
2691: @c =====================================================================
2692:
2693: @menu
1.15 anton 2694: * exception-idef:: Implementation Defined Options
1.14 anton 2695: @end menu
2696:
2697:
2698: @c ---------------------------------------------------------------------
1.15 anton 2699: @node exception-idef, , The optional Exception word set, The optional Exception word set
1.14 anton 2700: @subsection Implementation Defined Options
2701: @c ---------------------------------------------------------------------
2702:
2703: @table @i
2704: @item @code{THROW}-codes used in the system:
2705: The codes -256@minus{}-511 are used for reporting signals (see
2706: @file{errore.fs}). The codes -512@minus{}-2047 are used for OS errors
2707: (for file and memory allocation operations). The mapping from OS error
2708: numbers to throw code is -512@minus{}@var{errno}. One side effect of
2709: this mapping is that undefined OS errors produce a message with a
2710: strange number; e.g., @code{-1000 THROW} results in @code{Unknown error
2711: 488} on my system.
2712: @end table
2713:
2714: @c =====================================================================
2715: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
2716: @section The optional Facility word set
2717: @c =====================================================================
2718:
2719: @menu
1.15 anton 2720: * facility-idef:: Implementation Defined Options
2721: * facility-ambcond:: Ambiguous Conditions
1.14 anton 2722: @end menu
2723:
2724:
2725: @c ---------------------------------------------------------------------
2726: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
2727: @subsection Implementation Defined Options
2728: @c ---------------------------------------------------------------------
2729:
2730: @table @i
2731:
2732: @item encoding of keyboard events (@code{EKEY}):
2733: Not yet implemeted.
2734:
2735: @item duration of a system clock tick
2736: System dependent. With respect to @code{MS}, the time is specified in
2737: microseconds. How well the OS and the hardware implement this, is
2738: another question.
2739:
2740: @item repeatability to be expected from the execution of @code{MS}:
2741: System dependent. On Unix, a lot depends on load. If the system is
1.17 anton 2742: lightly loaded, and the delay is short enough that Gforth does not get
1.14 anton 2743: swapped out, the performance should be acceptable. Under MS-DOS and
2744: other single-tasking systems, it should be good.
2745:
2746: @end table
2747:
2748:
2749: @c ---------------------------------------------------------------------
1.15 anton 2750: @node facility-ambcond, , facility-idef, The optional Facility word set
1.14 anton 2751: @subsection Ambiguous conditions
2752: @c ---------------------------------------------------------------------
2753:
2754: @table @i
2755:
2756: @item @code{AT-XY} can't be performed on user output device:
2757: Largely terminal dependant. No range checks are done on the arguments.
2758: No errors are reported. You may see some garbage appearing, you may see
2759: simply nothing happen.
2760:
2761: @end table
2762:
2763:
2764: @c =====================================================================
2765: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
2766: @section The optional File-Access word set
2767: @c =====================================================================
2768:
2769: @menu
1.15 anton 2770: * file-idef:: Implementation Defined Options
2771: * file-ambcond:: Ambiguous Conditions
1.14 anton 2772: @end menu
2773:
2774:
2775: @c ---------------------------------------------------------------------
2776: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
2777: @subsection Implementation Defined Options
2778: @c ---------------------------------------------------------------------
2779:
2780: @table @i
2781:
2782: @item File access methods used:
2783: @code{R/O}, @code{R/W} and @code{BIN} work as you would
2784: expect. @code{W/O} translates into the C file opening mode @code{w} (or
2785: @code{wb}): The file is cleared, if it exists, and created, if it does
1.15 anton 2786: not (both with @code{open-file} and @code{create-file}). Under Unix
1.14 anton 2787: @code{create-file} creates a file with 666 permissions modified by your
2788: umask.
2789:
2790: @item file exceptions:
2791: The file words do not raise exceptions (except, perhaps, memory access
2792: faults when you pass illegal addresses or file-ids).
2793:
2794: @item file line terminator:
2795: System-dependent. Gforth uses C's newline character as line
2796: terminator. What the actual character code(s) of this are is
2797: system-dependent.
2798:
2799: @item file name format
2800: System dependent. Gforth just uses the file name format of your OS.
2801:
2802: @item information returned by @code{FILE-STATUS}:
2803: @code{FILE-STATUS} returns the most powerful file access mode allowed
2804: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
2805: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
2806: along with the retured mode.
2807:
2808: @item input file state after an exception when including source:
2809: All files that are left via the exception are closed.
2810:
2811: @item @var{ior} values and meaning:
1.15 anton 2812: The @var{ior}s returned by the file and memory allocation words are
2813: intended as throw codes. They typically are in the range
2814: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
2815: @var{ior}s is -512@minus{}@var{errno}.
1.14 anton 2816:
2817: @item maximum depth of file input nesting:
2818: limited by the amount of return stack, locals/TIB stack, and the number
2819: of open files available. This should not give you troubles.
2820:
2821: @item maximum size of input line:
2822: @code{/line}. Currently 255.
2823:
2824: @item methods of mapping block ranges to files:
2825: Currently, the block words automatically access the file
2826: @file{blocks.fb} in the currend working directory. More sophisticated
2827: methods could be implemented if there is demand (and a volunteer).
2828:
2829: @item number of string buffers provided by @code{S"}:
2830: 1
2831:
2832: @item size of string buffer used by @code{S"}:
2833: @code{/line}. currently 255.
2834:
2835: @end table
2836:
2837: @c ---------------------------------------------------------------------
1.15 anton 2838: @node file-ambcond, , file-idef, The optional File-Access word set
1.14 anton 2839: @subsection Ambiguous conditions
2840: @c ---------------------------------------------------------------------
2841:
2842: @table @i
2843:
2844: @item attempting to position a file outside it's boundaries:
2845: @code{REPOSITION-FILE} is performed as usual: Afterwards,
2846: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
2847:
2848: @item attempting to read from file positions not yet written:
2849: End-of-file, i.e., zero characters are read and no error is reported.
2850:
2851: @item @var{file-id} is invalid (@code{INCLUDE-FILE}):
2852: An appropriate exception may be thrown, but a memory fault or other
2853: problem is more probable.
2854:
2855: @item I/O exception reading or closing @var{file-id} (@code{include-file}, @code{included}):
2856: The @var{ior} produced by the operation, that discovered the problem, is
2857: thrown.
2858:
2859: @item named file cannot be opened (@code{included}):
2860: The @var{ior} produced by @code{open-file} is thrown.
2861:
2862: @item requesting an unmapped block number:
2863: There are no unmapped legal block numbers. On some operating systems,
2864: writing a block with a large number may overflow the file system and
2865: have an error message as consequence.
2866:
2867: @item using @code{source-id} when @code{blk} is non-zero:
2868: @code{source-id} performs its function. Typically it will give the id of
2869: the source which loaded the block. (Better ideas?)
2870:
2871: @end table
2872:
2873:
2874: @c =====================================================================
2875: @node The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
1.15 anton 2876: @section The optional Floating-Point word set
1.14 anton 2877: @c =====================================================================
2878:
2879: @menu
1.15 anton 2880: * floating-idef:: Implementation Defined Options
2881: * floating-ambcond:: Ambiguous Conditions
1.14 anton 2882: @end menu
2883:
2884:
2885: @c ---------------------------------------------------------------------
2886: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
2887: @subsection Implementation Defined Options
2888: @c ---------------------------------------------------------------------
2889:
2890: @table @i
2891:
1.15 anton 2892: @item format and range of floating point numbers:
2893: System-dependent; the @code{double} type of C.
1.14 anton 2894:
1.15 anton 2895: @item results of @code{REPRESENT} when @var{float} is out of range:
2896: System dependent; @code{REPRESENT} is implemented using the C library
2897: function @code{ecvt()} and inherits its behaviour in this respect.
1.14 anton 2898:
1.15 anton 2899: @item rounding or truncation of floating-point numbers:
2900: What's the question?!!
1.14 anton 2901:
1.15 anton 2902: @item size of floating-point stack:
2903: @code{s" FLOATING-STACK" environment? drop .}. Can be changed at startup
2904: with the command-line option @code{-f}.
1.14 anton 2905:
1.15 anton 2906: @item width of floating-point stack:
2907: @code{1 floats}.
1.14 anton 2908:
2909: @end table
2910:
2911:
2912: @c ---------------------------------------------------------------------
1.15 anton 2913: @node floating-ambcond, , floating-idef, The optional Floating-Point word set
2914: @subsection Ambiguous conditions
1.14 anton 2915: @c ---------------------------------------------------------------------
2916:
2917: @table @i
2918:
1.15 anton 2919: @item @code{df@@} or @code{df!} used with an address that is not double-float aligned:
2920: System-dependent. Typically results in an alignment fault like other
2921: alignment violations.
1.14 anton 2922:
1.15 anton 2923: @item @code{f@@} or @code{f!} used with an address that is not float aligned:
2924: System-dependent. Typically results in an alignment fault like other
2925: alignment violations.
1.14 anton 2926:
1.15 anton 2927: @item Floating-point result out of range:
2928: System-dependent. Can result in a @code{-55 THROW} (Floating-point
2929: unidentified fault), or can produce a special value representing, e.g.,
2930: Infinity.
1.14 anton 2931:
1.15 anton 2932: @item @code{sf@@} or @code{sf!} used with an address that is not single-float aligned:
2933: System-dependent. Typically results in an alignment fault like other
2934: alignment violations.
1.14 anton 2935:
1.15 anton 2936: @item BASE is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
2937: The floating-point number is converted into decimal nonetheless.
1.14 anton 2938:
1.15 anton 2939: @item Both arguments are equal to zero (@code{FATAN2}):
2940: System-dependent. @code{FATAN2} is implemented using the C library
2941: function @code{atan2()}.
1.14 anton 2942:
1.15 anton 2943: @item Using ftan on an argument @var{r1} where cos(@var{r1}) is zero:
2944: System-dependent. Anyway, typically the cos of @var{r1} will not be zero
2945: because of small errors and the tan will be a very large (or very small)
2946: but finite number.
1.14 anton 2947:
1.15 anton 2948: @item @var{d} cannot be presented precisely as a float in @code{D>F}:
2949: The result is rounded to the nearest float.
1.14 anton 2950:
1.15 anton 2951: @item dividing by zero:
2952: @code{-55 throw} (Floating-point unidentified fault)
1.14 anton 2953:
1.15 anton 2954: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
2955: System dependent. On IEEE-FP based systems the number is converted into
2956: an infinity.
1.14 anton 2957:
1.15 anton 2958: @item @var{float}<1 (@code{facosh}):
2959: @code{-55 throw} (Floating-point unidentified fault)
1.14 anton 2960:
1.15 anton 2961: @item @var{float}=<-1 (@code{flnp1}):
2962: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
2963: negative infinity is typically produced for @var{float}=-1.
1.14 anton 2964:
1.15 anton 2965: @item @var{float}=<0 (@code{fln}, @code{flog}):
2966: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
2967: negative infinity is typically produced for @var{float}=0.
1.14 anton 2968:
1.15 anton 2969: @item @var{float}<0 (@code{fasinh}, @code{fsqrt}):
2970: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
2971: produces values for these inputs on my Linux box (Bug in the C library?)
1.14 anton 2972:
1.15 anton 2973: @item |@var{float}|>1 (@code{facos}, @code{fasin}, @code{fatanh}):
2974: @code{-55 throw} (Floating-point unidentified fault).
1.14 anton 2975:
1.15 anton 2976: @item integer part of float cannot be represented by @var{d} in @code{f>d}:
2977: @code{-55 throw} (Floating-point unidentified fault).
1.14 anton 2978:
1.15 anton 2979: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
2980: This does not happen.
2981: @end table
1.14 anton 2982:
2983:
2984:
2985: @c =====================================================================
1.15 anton 2986: @node The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
2987: @section The optional Locals word set
1.14 anton 2988: @c =====================================================================
2989:
2990: @menu
1.15 anton 2991: * locals-idef:: Implementation Defined Options
2992: * locals-ambcond:: Ambiguous Conditions
1.14 anton 2993: @end menu
2994:
2995:
2996: @c ---------------------------------------------------------------------
1.15 anton 2997: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
1.14 anton 2998: @subsection Implementation Defined Options
2999: @c ---------------------------------------------------------------------
3000:
3001: @table @i
3002:
1.15 anton 3003: @item maximum number of locals in a definition:
3004: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
3005: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
3006: characters. The number of locals in a definition is bounded by the size
3007: of locals-buffer, which contains the names of the locals.
1.14 anton 3008:
3009: @end table
3010:
3011:
3012: @c ---------------------------------------------------------------------
1.15 anton 3013: @node locals-ambcond, , locals-idef, The optional Locals word set
1.14 anton 3014: @subsection Ambiguous conditions
3015: @c ---------------------------------------------------------------------
3016:
3017: @table @i
3018:
1.15 anton 3019: @item executing a named local in interpretation state:
3020: @code{-14 throw} (Interpreting a compile-only word).
1.14 anton 3021:
1.15 anton 3022: @item @var{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
3023: @code{-32 throw} (Invalid name argument)
1.14 anton 3024:
3025: @end table
3026:
3027:
3028: @c =====================================================================
1.15 anton 3029: @node The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
3030: @section The optional Memory-Allocation word set
1.14 anton 3031: @c =====================================================================
3032:
3033: @menu
1.15 anton 3034: * memory-idef:: Implementation Defined Options
1.14 anton 3035: @end menu
3036:
3037:
3038: @c ---------------------------------------------------------------------
1.15 anton 3039: @node memory-idef, , The optional Memory-Allocation word set, The optional Memory-Allocation word set
1.14 anton 3040: @subsection Implementation Defined Options
3041: @c ---------------------------------------------------------------------
3042:
3043: @table @i
3044:
1.15 anton 3045: @item values and meaning of @var{ior}:
3046: The @var{ior}s returned by the file and memory allocation words are
3047: intended as throw codes. They typically are in the range
3048: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
3049: @var{ior}s is -512@minus{}@var{errno}.
1.14 anton 3050:
3051: @end table
3052:
3053: @c =====================================================================
1.15 anton 3054: @node The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
3055: @section The optional Programming-Tools word set
1.14 anton 3056: @c =====================================================================
3057:
3058: @menu
1.15 anton 3059: * programming-idef:: Implementation Defined Options
3060: * programming-ambcond:: Ambiguous Conditions
1.14 anton 3061: @end menu
3062:
3063:
3064: @c ---------------------------------------------------------------------
1.15 anton 3065: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
1.14 anton 3066: @subsection Implementation Defined Options
3067: @c ---------------------------------------------------------------------
3068:
3069: @table @i
3070:
1.15 anton 3071: @item ending sequence for input following @code{;code} and @code{code}:
3072: Not implemented (yet).
1.14 anton 3073:
1.15 anton 3074: @item manner of processing input following @code{;code} and @code{code}:
3075: Not implemented (yet).
3076:
3077: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
3078: Not implemented (yet). If they were implemented, they would use the
3079: search order wordset.
3080:
3081: @item source and format of display by @code{SEE}:
3082: The source for @code{see} is the intermediate code used by the inner
3083: interpreter. The current @code{see} tries to output Forth source code
3084: as well as possible.
3085:
1.14 anton 3086: @end table
3087:
3088: @c ---------------------------------------------------------------------
1.15 anton 3089: @node programming-ambcond, , programming-idef, The optional Programming-Tools word set
1.14 anton 3090: @subsection Ambiguous conditions
3091: @c ---------------------------------------------------------------------
3092:
3093: @table @i
3094:
1.15 anton 3095: @item deleting the compilation wordlist (@code{FORGET}):
3096: Not implemented (yet).
1.14 anton 3097:
1.15 anton 3098: @item fewer than @var{u}+1 items on the control flow stack (@code{CS-PICK}, @code{CS-ROLL}):
3099: This typically results in an @code{abort"} with a descriptive error
3100: message (may change into a @code{-22 throw} (Control structure mismatch)
3101: in the future). You may also get a memory access error. If you are
3102: unlucky, this ambiguous condition is not caught.
3103:
3104: @item @var{name} can't be found (@code{forget}):
3105: Not implemented (yet).
1.14 anton 3106:
1.15 anton 3107: @item @var{name} not defined via @code{CREATE}:
3108: @code{;code} is not implemented (yet). If it were, it would behave like
3109: @code{DOES>} in this respect, i.e., change the execution semantics of
3110: the last defined word no matter how it was defined.
1.14 anton 3111:
1.15 anton 3112: @item @code{POSTPONE} applied to @code{[IF]}:
3113: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
3114: equivalent to @code{[IF]}.
1.14 anton 3115:
1.15 anton 3116: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
3117: Continue in the same state of conditional compilation in the next outer
3118: input source. Currently there is no warning to the user about this.
1.14 anton 3119:
1.15 anton 3120: @item removing a needed definition (@code{FORGET}):
3121: Not implemented (yet).
1.14 anton 3122:
3123: @end table
3124:
3125:
3126: @c =====================================================================
1.15 anton 3127: @node The optional Search-Order word set, , The optional Programming-Tools word set, ANS conformance
3128: @section The optional Search-Order word set
1.14 anton 3129: @c =====================================================================
3130:
3131: @menu
1.15 anton 3132: * search-idef:: Implementation Defined Options
3133: * search-ambcond:: Ambiguous Conditions
1.14 anton 3134: @end menu
3135:
3136:
3137: @c ---------------------------------------------------------------------
1.15 anton 3138: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
1.14 anton 3139: @subsection Implementation Defined Options
3140: @c ---------------------------------------------------------------------
3141:
3142: @table @i
3143:
1.15 anton 3144: @item maximum number of word lists in search order:
3145: @code{s" wordlists" environment? drop .}. Currently 16.
3146:
3147: @item minimum search order:
3148: @code{root root}.
1.14 anton 3149:
3150: @end table
3151:
3152: @c ---------------------------------------------------------------------
1.15 anton 3153: @node search-ambcond, , search-idef, The optional Search-Order word set
1.14 anton 3154: @subsection Ambiguous conditions
3155: @c ---------------------------------------------------------------------
3156:
3157: @table @i
3158:
1.15 anton 3159: @item changing the compilation wordlist (during compilation):
3160: The definition is put into the wordlist that is the compilation wordlist
3161: when @code{REVEAL} is executed (by @code{;}, @code{DOES>},
3162: @code{RECURSIVE}, etc.).
1.14 anton 3163:
1.15 anton 3164: @item search order empty (@code{previous}):
3165: @code{abort" Vocstack empty"}.
1.14 anton 3166:
1.15 anton 3167: @item too many word lists in search order (@code{also}):
3168: @code{abort" Vocstack full"}.
1.14 anton 3169:
3170: @end table
1.13 anton 3171:
3172:
1.17 anton 3173: @node Model, Emacs and Gforth, ANS conformance, Top
1.4 anton 3174: @chapter Model
3175:
1.17 anton 3176: @node Emacs and Gforth, Internals, Model, Top
3177: @chapter Emacs and Gforth
1.4 anton 3178:
1.17 anton 3179: Gforth comes with @file{gforth.el}, an improved version of
1.4 anton 3180: @file{forth.el} by Goran Rydqvist (icluded in the TILE package). The
3181: improvements are a better (but still not perfect) handling of
3182: indentation. I have also added comment paragraph filling (@kbd{M-q}),
1.8 anton 3183: commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) regions and
3184: removing debugging tracers (@kbd{C-x ~}, @pxref{Debugging}). I left the
3185: stuff I do not use alone, even though some of it only makes sense for
3186: TILE. To get a description of these features, enter Forth mode and type
3187: @kbd{C-h m}.
1.4 anton 3188:
1.17 anton 3189: In addition, Gforth supports Emacs quite well: The source code locations
1.4 anton 3190: given in error messages, debugging output (from @code{~~}) and failed
3191: assertion messages are in the right format for Emacs' compilation mode
3192: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
3193: Manual}) so the source location corresponding to an error or other
3194: message is only a few keystrokes away (@kbd{C-x `} for the next error,
3195: @kbd{C-c C-c} for the error under the cursor).
3196:
3197: Also, if you @code{include} @file{etags.fs}, a new @file{TAGS} file
3198: (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) will be produced that
3199: contains the definitions of all words defined afterwards. You can then
3200: find the source for a word using @kbd{M-.}. Note that emacs can use
1.17 anton 3201: several tags files at the same time (e.g., one for the Gforth sources
1.4 anton 3202: and one for your program).
3203:
3204: To get all these benefits, add the following lines to your @file{.emacs}
3205: file:
3206:
3207: @example
3208: (autoload 'forth-mode "gforth.el")
3209: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
3210: @end example
3211:
1.17 anton 3212: @node Internals, Bugs, Emacs and Gforth, Top
1.3 anton 3213: @chapter Internals
3214:
1.17 anton 3215: Reading this section is not necessary for programming with Gforth. It
3216: should be helpful for finding your way in the Gforth sources.
1.3 anton 3217:
1.24 ! anton 3218: The ideas in this section have also been published in the papers
! 3219: @cite{ANS fig/GNU/??? Forth} (in German) by Bernd Paysan, presented at
! 3220: the Forth-Tagung '93 and @cite{A Portable Forth Engine} by M. Anton
! 3221: Ertl, presented at EuroForth '93; the latter is available at
! 3222: @*@file{http://www.complang.tuwien.ac.at/papers/ertl93.ps.Z}.
! 3223:
1.4 anton 3224: @menu
3225: * Portability::
3226: * Threading::
3227: * Primitives::
3228: * System Architecture::
1.17 anton 3229: * Performance::
1.4 anton 3230: @end menu
3231:
3232: @node Portability, Threading, Internals, Internals
1.3 anton 3233: @section Portability
3234:
3235: One of the main goals of the effort is availability across a wide range
3236: of personal machines. fig-Forth, and, to a lesser extent, F83, achieved
3237: this goal by manually coding the engine in assembly language for several
3238: then-popular processors. This approach is very labor-intensive and the
3239: results are short-lived due to progress in computer architecture.
3240:
3241: Others have avoided this problem by coding in C, e.g., Mitch Bradley
3242: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
3243: particularly popular for UNIX-based Forths due to the large variety of
3244: architectures of UNIX machines. Unfortunately an implementation in C
3245: does not mix well with the goals of efficiency and with using
3246: traditional techniques: Indirect or direct threading cannot be expressed
3247: in C, and switch threading, the fastest technique available in C, is
3248: significantly slower. Another problem with C is that it's very
3249: cumbersome to express double integer arithmetic.
3250:
3251: Fortunately, there is a portable language that does not have these
3252: limitations: GNU C, the version of C processed by the GNU C compiler
3253: (@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
3254: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
3255: Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
3256: threading possible, its @code{long long} type (@pxref{Long Long, ,
3257: Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forths
3258: double numbers. GNU C is available for free on all important (and many
3259: unimportant) UNIX machines, VMS, 80386s running MS-DOS, the Amiga, and
3260: the Atari ST, so a Forth written in GNU C can run on all these
1.17 anton 3261: machines.
1.3 anton 3262:
3263: Writing in a portable language has the reputation of producing code that
3264: is slower than assembly. For our Forth engine we repeatedly looked at
3265: the code produced by the compiler and eliminated most compiler-induced
3266: inefficiencies by appropriate changes in the source-code.
3267:
3268: However, register allocation cannot be portably influenced by the
3269: programmer, leading to some inefficiencies on register-starved
3270: machines. We use explicit register declarations (@pxref{Explicit Reg
3271: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
3272: improve the speed on some machines. They are turned on by using the
3273: @code{gcc} switch @code{-DFORCE_REG}. Unfortunately, this feature not
3274: only depends on the machine, but also on the compiler version: On some
3275: machines some compiler versions produce incorrect code when certain
3276: explicit register declarations are used. So by default
3277: @code{-DFORCE_REG} is not used.
3278:
1.4 anton 3279: @node Threading, Primitives, Portability, Internals
1.3 anton 3280: @section Threading
3281:
3282: GNU C's labels as values extension (available since @code{gcc-2.0},
3283: @pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
3284: makes it possible to take the address of @var{label} by writing
3285: @code{&&@var{label}}. This address can then be used in a statement like
3286: @code{goto *@var{address}}. I.e., @code{goto *&&x} is the same as
3287: @code{goto x}.
3288:
3289: With this feature an indirect threaded NEXT looks like:
3290: @example
3291: cfa = *ip++;
3292: ca = *cfa;
3293: goto *ca;
3294: @end example
3295: For those unfamiliar with the names: @code{ip} is the Forth instruction
3296: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
3297: execution token and points to the code field of the next word to be
3298: executed; The @code{ca} (code address) fetched from there points to some
3299: executable code, e.g., a primitive or the colon definition handler
3300: @code{docol}.
3301:
3302: Direct threading is even simpler:
3303: @example
3304: ca = *ip++;
3305: goto *ca;
3306: @end example
3307:
3308: Of course we have packaged the whole thing neatly in macros called
3309: @code{NEXT} and @code{NEXT1} (the part of NEXT after fetching the cfa).
3310:
1.4 anton 3311: @menu
3312: * Scheduling::
3313: * Direct or Indirect Threaded?::
3314: * DOES>::
3315: @end menu
3316:
3317: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
1.3 anton 3318: @subsection Scheduling
3319:
3320: There is a little complication: Pipelined and superscalar processors,
3321: i.e., RISC and some modern CISC machines can process independent
3322: instructions while waiting for the results of an instruction. The
3323: compiler usually reorders (schedules) the instructions in a way that
3324: achieves good usage of these delay slots. However, on our first tries
3325: the compiler did not do well on scheduling primitives. E.g., for
3326: @code{+} implemented as
3327: @example
3328: n=sp[0]+sp[1];
3329: sp++;
3330: sp[0]=n;
3331: NEXT;
3332: @end example
3333: the NEXT comes strictly after the other code, i.e., there is nearly no
3334: scheduling. After a little thought the problem becomes clear: The
3335: compiler cannot know that sp and ip point to different addresses (and
1.4 anton 3336: the version of @code{gcc} we used would not know it even if it was
3337: possible), so it could not move the load of the cfa above the store to
3338: the TOS. Indeed the pointers could be the same, if code on or very near
3339: the top of stack were executed. In the interest of speed we chose to
3340: forbid this probably unused ``feature'' and helped the compiler in
3341: scheduling: NEXT is divided into the loading part (@code{NEXT_P1}) and
3342: the goto part (@code{NEXT_P2}). @code{+} now looks like:
1.3 anton 3343: @example
3344: n=sp[0]+sp[1];
3345: sp++;
3346: NEXT_P1;
3347: sp[0]=n;
3348: NEXT_P2;
3349: @end example
1.4 anton 3350: This can be scheduled optimally by the compiler.
1.3 anton 3351:
3352: This division can be turned off with the switch @code{-DCISC_NEXT}. This
3353: switch is on by default on machines that do not profit from scheduling
3354: (e.g., the 80386), in order to preserve registers.
3355:
1.4 anton 3356: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
1.3 anton 3357: @subsection Direct or Indirect Threaded?
3358:
3359: Both! After packaging the nasty details in macro definitions we
3360: realized that we could switch between direct and indirect threading by
3361: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
3362: defining a few machine-specific macros for the direct-threading case.
3363: On the Forth level we also offer access words that hide the
3364: differences between the threading methods (@pxref{Threading Words}).
3365:
3366: Indirect threading is implemented completely
3367: machine-independently. Direct threading needs routines for creating
3368: jumps to the executable code (e.g. to docol or dodoes). These routines
3369: are inherently machine-dependent, but they do not amount to many source
3370: lines. I.e., even porting direct threading to a new machine is a small
3371: effort.
3372:
1.4 anton 3373: @node DOES>, , Direct or Indirect Threaded?, Threading
1.3 anton 3374: @subsection DOES>
3375: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
3376: the chunk of code executed by every word defined by a
3377: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
3378: the Forth code to be executed, i.e. the code after the @code{DOES>} (the
3379: DOES-code)? There are two solutions:
3380:
3381: In fig-Forth the code field points directly to the dodoes and the
3382: DOES-code address is stored in the cell after the code address
3383: (i.e. at cfa cell+). It may seem that this solution is illegal in the
3384: Forth-79 and all later standards, because in fig-Forth this address
3385: lies in the body (which is illegal in these standards). However, by
3386: making the code field larger for all words this solution becomes legal
3387: again. We use this approach for the indirect threaded version. Leaving
3388: a cell unused in most words is a bit wasteful, but on the machines we
3389: are targetting this is hardly a problem. The other reason for having a
3390: code field size of two cells is to avoid having different image files
1.4 anton 3391: for direct and indirect threaded systems (@pxref{System Architecture}).
1.3 anton 3392:
3393: The other approach is that the code field points or jumps to the cell
3394: after @code{DOES}. In this variant there is a jump to @code{dodoes} at
3395: this address. @code{dodoes} can then get the DOES-code address by
3396: computing the code address, i.e., the address of the jump to dodoes,
3397: and add the length of that jump field. A variant of this is to have a
3398: call to @code{dodoes} after the @code{DOES>}; then the return address
3399: (which can be found in the return register on RISCs) is the DOES-code
3400: address. Since the two cells available in the code field are usually
3401: used up by the jump to the code address in direct threading, we use
3402: this approach for direct threading. We did not want to add another
3403: cell to the code field.
3404:
1.4 anton 3405: @node Primitives, System Architecture, Threading, Internals
1.3 anton 3406: @section Primitives
3407:
1.4 anton 3408: @menu
3409: * Automatic Generation::
3410: * TOS Optimization::
3411: * Produced code::
3412: @end menu
3413:
3414: @node Automatic Generation, TOS Optimization, Primitives, Primitives
1.3 anton 3415: @subsection Automatic Generation
3416:
3417: Since the primitives are implemented in a portable language, there is no
3418: longer any need to minimize the number of primitives. On the contrary,
3419: having many primitives is an advantage: speed. In order to reduce the
3420: number of errors in primitives and to make programming them easier, we
3421: provide a tool, the primitive generator (@file{prims2x.fs}), that
3422: automatically generates most (and sometimes all) of the C code for a
3423: primitive from the stack effect notation. The source for a primitive
3424: has the following form:
3425:
3426: @format
3427: @var{Forth-name} @var{stack-effect} @var{category} [@var{pronounc.}]
3428: [@code{""}@var{glossary entry}@code{""}]
3429: @var{C code}
3430: [@code{:}
3431: @var{Forth code}]
3432: @end format
3433:
3434: The items in brackets are optional. The category and glossary fields
3435: are there for generating the documentation, the Forth code is there
3436: for manual implementations on machines without GNU C. E.g., the source
3437: for the primitive @code{+} is:
3438: @example
3439: + n1 n2 -- n core plus
3440: n = n1+n2;
3441: @end example
3442:
3443: This looks like a specification, but in fact @code{n = n1+n2} is C
3444: code. Our primitive generation tool extracts a lot of information from
3445: the stack effect notations@footnote{We use a one-stack notation, even
3446: though we have separate data and floating-point stacks; The separate
3447: notation can be generated easily from the unified notation.}: The number
3448: of items popped from and pushed on the stack, their type, and by what
3449: name they are referred to in the C code. It then generates a C code
3450: prelude and postlude for each primitive. The final C code for @code{+}
3451: looks like this:
3452:
3453: @example
3454: I_plus: /* + ( n1 n2 -- n ) */ /* label, stack effect */
3455: /* */ /* documentation */
1.4 anton 3456: @{
1.3 anton 3457: DEF_CA /* definition of variable ca (indirect threading) */
3458: Cell n1; /* definitions of variables */
3459: Cell n2;
3460: Cell n;
3461: n1 = (Cell) sp[1]; /* input */
3462: n2 = (Cell) TOS;
3463: sp += 1; /* stack adjustment */
3464: NAME("+") /* debugging output (with -DDEBUG) */
1.4 anton 3465: @{
1.3 anton 3466: n = n1+n2; /* C code taken from the source */
1.4 anton 3467: @}
1.3 anton 3468: NEXT_P1; /* NEXT part 1 */
3469: TOS = (Cell)n; /* output */
3470: NEXT_P2; /* NEXT part 2 */
1.4 anton 3471: @}
1.3 anton 3472: @end example
3473:
3474: This looks long and inefficient, but the GNU C compiler optimizes quite
3475: well and produces optimal code for @code{+} on, e.g., the R3000 and the
3476: HP RISC machines: Defining the @code{n}s does not produce any code, and
3477: using them as intermediate storage also adds no cost.
3478:
3479: There are also other optimizations, that are not illustrated by this
3480: example: Assignments between simple variables are usually for free (copy
3481: propagation). If one of the stack items is not used by the primitive
3482: (e.g. in @code{drop}), the compiler eliminates the load from the stack
3483: (dead code elimination). On the other hand, there are some things that
3484: the compiler does not do, therefore they are performed by
3485: @file{prims2x.fs}: The compiler does not optimize code away that stores
3486: a stack item to the place where it just came from (e.g., @code{over}).
3487:
3488: While programming a primitive is usually easy, there are a few cases
3489: where the programmer has to take the actions of the generator into
3490: account, most notably @code{?dup}, but also words that do not (always)
3491: fall through to NEXT.
3492:
1.4 anton 3493: @node TOS Optimization, Produced code, Automatic Generation, Primitives
1.3 anton 3494: @subsection TOS Optimization
3495:
3496: An important optimization for stack machine emulators, e.g., Forth
3497: engines, is keeping one or more of the top stack items in
1.4 anton 3498: registers. If a word has the stack effect @var{in1}...@var{inx} @code{--}
3499: @var{out1}...@var{outy}, keeping the top @var{n} items in registers
1.3 anton 3500: @itemize
3501: @item
3502: is better than keeping @var{n-1} items, if @var{x>=n} and @var{y>=n},
3503: due to fewer loads from and stores to the stack.
3504: @item is slower than keeping @var{n-1} items, if @var{x<>y} and @var{x<n} and
3505: @var{y<n}, due to additional moves between registers.
3506: @end itemize
3507:
3508: In particular, keeping one item in a register is never a disadvantage,
3509: if there are enough registers. Keeping two items in registers is a
3510: disadvantage for frequent words like @code{?branch}, constants,
3511: variables, literals and @code{i}. Therefore our generator only produces
3512: code that keeps zero or one items in registers. The generated C code
3513: covers both cases; the selection between these alternatives is made at
3514: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
3515: code for @code{+} is just a simple variable name in the one-item case,
3516: otherwise it is a macro that expands into @code{sp[0]}. Note that the
3517: GNU C compiler tries to keep simple variables like @code{TOS} in
3518: registers, and it usually succeeds, if there are enough registers.
3519:
3520: The primitive generator performs the TOS optimization for the
3521: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
3522: operations the benefit of this optimization is even larger:
3523: floating-point operations take quite long on most processors, but can be
3524: performed in parallel with other operations as long as their results are
3525: not used. If the FP-TOS is kept in a register, this works. If
3526: it is kept on the stack, i.e., in memory, the store into memory has to
3527: wait for the result of the floating-point operation, lengthening the
3528: execution time of the primitive considerably.
3529:
3530: The TOS optimization makes the automatic generation of primitives a
3531: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
3532: @code{TOS} is not sufficient. There are some special cases to
3533: consider:
3534: @itemize
3535: @item In the case of @code{dup ( w -- w w )} the generator must not
3536: eliminate the store to the original location of the item on the stack,
3537: if the TOS optimization is turned on.
1.4 anton 3538: @item Primitives with stack effects of the form @code{--}
3539: @var{out1}...@var{outy} must store the TOS to the stack at the start.
3540: Likewise, primitives with the stack effect @var{in1}...@var{inx} @code{--}
1.3 anton 3541: must load the TOS from the stack at the end. But for the null stack
3542: effect @code{--} no stores or loads should be generated.
3543: @end itemize
3544:
1.4 anton 3545: @node Produced code, , TOS Optimization, Primitives
1.3 anton 3546: @subsection Produced code
3547:
3548: To see what assembly code is produced for the primitives on your machine
3549: with your compiler and your flag settings, type @code{make engine.s} and
1.4 anton 3550: look at the resulting file @file{engine.s}.
1.3 anton 3551:
1.17 anton 3552: @node System Architecture, Performance, Primitives, Internals
1.3 anton 3553: @section System Architecture
3554:
3555: Our Forth system consists not only of primitives, but also of
3556: definitions written in Forth. Since the Forth compiler itself belongs
3557: to those definitions, it is not possible to start the system with the
3558: primitives and the Forth source alone. Therefore we provide the Forth
3559: code as an image file in nearly executable form. At the start of the
3560: system a C routine loads the image file into memory, sets up the
3561: memory (stacks etc.) according to information in the image file, and
3562: starts executing Forth code.
3563:
3564: The image file format is a compromise between the goals of making it
3565: easy to generate image files and making them portable. The easiest way
3566: to generate an image file is to just generate a memory dump. However,
3567: this kind of image file cannot be used on a different machine, or on
3568: the next version of the engine on the same machine, it even might not
3569: work with the same engine compiled by a different version of the C
3570: compiler. We would like to have as few versions of the image file as
3571: possible, because we do not want to distribute many versions of the
3572: same image file, and to make it easy for the users to use their image
3573: files on many machines. We currently need to create a different image
3574: file for machines with different cell sizes and different byte order
1.17 anton 3575: (little- or big-endian)@footnote{We are considering adding information to the
1.3 anton 3576: image file that enables the loader to change the byte order.}.
3577:
3578: Forth code that is going to end up in a portable image file has to
1.4 anton 3579: comply to some restrictions: addresses have to be stored in memory with
3580: special words (@code{A!}, @code{A,}, etc.) in order to make the code
3581: relocatable. Cells, floats, etc., have to be stored at the natural
3582: alignment boundaries@footnote{E.g., store floats (8 bytes) at an address
3583: dividable by~8. This happens automatically in our system when you use
3584: the ANS Forth alignment words.}, in order to avoid alignment faults on
3585: machines with stricter alignment. The image file is produced by a
3586: metacompiler (@file{cross.fs}).
1.3 anton 3587:
3588: So, unlike the image file of Mitch Bradleys @code{cforth}, our image
3589: file is not directly executable, but has to undergo some manipulations
3590: during loading. Address relocation is performed at image load-time, not
3591: at run-time. The loader also has to replace tokens standing for
3592: primitive calls with the appropriate code-field addresses (or code
3593: addresses in the case of direct threading).
1.4 anton 3594:
1.17 anton 3595: @node Performance, , System Architecture, Internals
3596: @section Performance
3597:
3598: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
3599: impossible to write a significantly faster engine.
3600:
3601: On register-starved machines like the 386 architecture processors
3602: improvements are possible, because @code{gcc} does not utilize the
3603: registers as well as a human, even with explicit register declarations;
3604: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
3605: and hand-tuned it for the 486; this system is 1.19 times faster on the
3606: Sieve benchmark on a 486DX2/66 than Gforth compiled with
3607: @code{gcc-2.6.3} with @code{-DFORCE_REG}.
3608:
3609: However, this potential advantage of assembly language implementations
3610: is not necessarily realized in complete Forth systems: We compared
3611: Gforth (compiled with @code{gcc-2.6.3} and @code{-DFORCE_REG}) with
1.18 anton 3612: Win32Forth 1.2093 and LMI's NT Forth (Beta, May 1994), two systems
3613: written in assembly, and with two systems written in C: PFE-0.9.11
3614: (compiled with @code{gcc-2.6.3} with the default configuration for
3615: Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS}) and ThisForth Beta
3616: (compiled with gcc-2.6.3 -O3 -fomit-frame-pointer). We benchmarked
3617: Gforth, PFE and ThisForth on a 486DX2/66 under Linux. Kenneth O'Heskin
3618: kindly provided the results for Win32Forth and NT Forth on a 486DX2/66
3619: with similar memory performance under Windows NT.
1.17 anton 3620:
3621: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
3622: matrix multiplication come from the Stanford integer benchmarks and have
3623: been translated into Forth by Martin Fraeman; we used the versions
3624: included in the TILE Forth package; and a recursive Fibonacci number
1.24 ! anton 3625: computation for benchmarking calling performance. The following table shows
1.17 anton 3626: the time taken for the benchmarks scaled by the time taken by Gforth (in
3627: other words, it shows the speedup factor that Gforth achieved over the
3628: other systems).
3629:
3630: @example
3631: relative Win32- NT This-
3632: time Gforth Forth Forth PFE Forth
3633: sieve 1.00 1.30 1.07 1.67 2.98
3634: bubble 1.00 1.30 1.40 1.66
3635: matmul 1.00 1.40 1.29 2.24
3636: fib 1.00 1.44 1.26 1.82 2.82
3637: @end example
3638:
3639: You may find the good performance of Gforth compared with the systems
3640: written in assembly language quite surprising. One important reason for
3641: the disappointing performance of these systems is probably that they are
3642: not written optimally for the 486 (e.g., they use the @code{lods}
3643: instruction). In addition, Win32Forth uses a comfortable, but costly
3644: method for relocating the Forth image: like @code{cforth}, it computes
3645: the actual addresses at run time, resulting in two address computations
3646: per NEXT (@pxref{System Architecture}).
3647:
3648: The speedup of Gforth over PFE and ThisForth can be easily explained
3649: with the self-imposed restriction to standard C (although the measured
3650: implementation of PFE uses a GNU C extension: global register
3651: variables), which makes efficient threading impossible. Moreover,
3652: current C compilers have a hard time optimizing other aspects of the
3653: ThisForth source.
3654:
3655: Note that the performance of Gforth on 386 architecture processors
3656: varies widely with the version of @code{gcc} used. E.g., @code{gcc-2.5.8}
3657: failed to allocate any of the virtual machine registers into real
3658: machine registers by itself and would not work correctly with explicit
3659: register declarations, giving a 1.3 times slower engine (on a 486DX2/66
3660: running the Sieve) than the one measured above.
3661:
1.24 ! anton 3662: The numbers in this section have also been published in the paper
! 3663: @cite{Translating Forth to Efficient C} by M. Anton Ertl and Martin
! 3664: Maierhofer, presented at EuroForth '95. It is available at
! 3665: @*@file{http://www.complang.tuwien.ac.at/papers/ertl&maierhofer95.ps.gz};
! 3666: it also contains numbers for some native code systems. You can find
! 3667: numbers for Gforth on various machines in @file{Benchres}.
! 3668:
1.4 anton 3669: @node Bugs, Pedigree, Internals, Top
3670: @chapter Bugs
3671:
1.17 anton 3672: Known bugs are described in the file BUGS in the Gforth distribution.
3673:
1.24 ! anton 3674: If you find a bug, please send a bug report to
! 3675: @code{gforth-bugs@@mips.complang.tuwien.ac.at}. A bug report should
1.17 anton 3676: describe the Gforth version used (it is announced at the start of an
3677: interactive Gforth session), the machine and operating system (on Unix
3678: systems you can use @code{uname -a} to produce this information), the
1.24 ! anton 3679: installation options (send the @code{config.status} file), and a
! 3680: complete list of changes you (or your installer) have made to the Gforth
! 3681: sources (if any); it should contain a program (or a sequence of keyboard
! 3682: commands) that reproduces the bug and a description of what you think
! 3683: constitutes the buggy behaviour.
1.17 anton 3684:
3685: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
3686: to Report Bugs, gcc.info, GNU C Manual}.
3687:
3688:
1.4 anton 3689: @node Pedigree, Word Index, Bugs, Top
3690: @chapter Pedigree
3691:
1.17 anton 3692: Gforth descends from BigForth (1993) and fig-Forth. Gforth and PFE (by
1.24 ! anton 3693: Dirk Zoller) will cross-fertilize each other. Of course, a significant
! 3694: part of the design of Gforth was prescribed by ANS Forth.
1.17 anton 3695:
1.23 pazsan 3696: Bernd Paysan wrote BigForth, a descendent from TurboForth, an unreleased
3697: 32 bit native code version of VolksForth for the Atari ST, written
3698: mostly by Dietrich Weineck.
3699:
3700: VolksForth descends from F83. It was written by Klaus Schleisiek, Bernd
3701: Pennemann, Georg Rehfeld and Dietrich Weineck for the C64 (called
1.24 ! anton 3702: UltraForth there) in the mid-80s and ported to the Atari ST in 1986.
1.17 anton 3703:
3704: Laxen and Perry wrote F83 as a model implementation of the
3705: Forth-83 standard. !! Pedigree? When?
3706:
3707: A team led by Bill Ragsdale implemented fig-Forth on many processors in
1.24 ! anton 3708: 1979. Robert Selzer and Bill Ragsdale developed the original
! 3709: implementation of fig-Forth for the 6502 based on microForth.
! 3710:
! 3711: The principal architect of microForth was Dean Sanderson. microForth was
! 3712: FORTH, Inc.'s first off-the-shelf product. It was developped in 1976 for
! 3713: the 1802, and subsequently implemented on the 8080, the 6800 and the
! 3714: Z80.
1.17 anton 3715:
1.24 ! anton 3716: All earlier Forth systems were custom-made, usually by Charles Moore,
! 3717: who discovered (as he puts it) Forth in the late 60s.
1.17 anton 3718:
3719: A part of the information in this section comes from @cite{The Evolution
3720: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
3721: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
3722: Notices 28(3), 1993. You can find more historical and genealogical
3723: information about Forth there.
3724:
1.4 anton 3725: @node Word Index, Node Index, Pedigree, Top
3726: @chapter Word Index
3727:
1.18 anton 3728: This index is as incomplete as the manual. Each word is listed with
3729: stack effect and wordset.
1.17 anton 3730:
3731: @printindex fn
3732:
1.4 anton 3733: @node Node Index, , Word Index, Top
3734: @chapter Node Index
1.17 anton 3735:
3736: This index is even less complete than the manual.
1.1 anton 3737:
3738: @contents
3739: @bye
3740:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>