[gforth] / gforth / Attic / gforth.ds

gforth: gforth/Attic/gforth.ds

1 :	anton	1.1	\input texinfo @c --texinfo--
2 :			@comment The source is gforth.ds, from which gforth.texi is generated
3 :			@comment %**start of header (This is for running Texinfo on a region.)
4 :	anton	1.4	@setfilename gforth.info
5 :	anton	1.17	@settitle Gforth Manual
6 :	anton	1.4	@comment @setchapternewpage odd
7 :	anton	1.1	@comment %**end of header (This is for running Texinfo on a region.)
8 :
9 :			@ifinfo
10 :	anton	1.30	This file documents Gforth 0.2
11 :	anton	1.1
12 :	anton	1.32	Copyright @copyright{} 1995,1996 Free Software Foundation, Inc.
13 :	anton	1.1
14 :			Permission is granted to make and distribute verbatim copies of
15 :			this manual provided the copyright notice and this permission notice
16 :			are preserved on all copies.
17 :
18 :	anton	1.4	@ignore
19 :	anton	1.1	Permission is granted to process this file through TeX and print the
20 :			results, provided the printed document carries a copying permission
21 :			notice identical to this one except for the removal of this paragraph
22 :			(this paragraph not being relevant to the printed manual).
23 :
24 :	anton	1.4	@end ignore
25 :	anton	1.1	Permission is granted to copy and distribute modified versions of this
26 :			manual under the conditions for verbatim copying, provided also that the
27 :			sections entitled "Distribution" and "General Public License" are
28 :			included exactly as in the original, and provided that the entire
29 :			resulting derived work is distributed under the terms of a permission
30 :			notice identical to this one.
31 :
32 :			Permission is granted to copy and distribute translations of this manual
33 :			into another language, under the above conditions for modified versions,
34 :			except that the sections entitled "Distribution" and "General Public
35 :			License" may be included in a translation approved by the author instead
36 :			of in the original English.
37 :			@end ifinfo
38 :
39 :	anton	1.24	@finalout
40 :	anton	1.1	@titlepage
41 :			@sp 10
42 :	anton	1.17	@center @titlefont{Gforth Manual}
43 :	anton	1.1	@sp 2
44 :	anton	1.30	@center for version 0.2
45 :	anton	1.1	@sp 2
46 :			@center Anton Ertl
47 :	anton	1.25	@center Bernd Paysan
48 :	anton	1.17	@sp 3
49 :			@center This manual is under construction
50 :	anton	1.1
51 :			@comment The following two commands start the copyright page.
52 :			@page
53 :			@vskip 0pt plus 1filll
54 :	anton	1.32	Copyright @copyright{} 1995,1996 Free Software Foundation, Inc.
55 :	anton	1.1
56 :			@comment !! Published by ... or You can get a copy of this manual ...
57 :
58 :			Permission is granted to make and distribute verbatim copies of
59 :			this manual provided the copyright notice and this permission notice
60 :			are preserved on all copies.
61 :
62 :			Permission is granted to copy and distribute modified versions of this
63 :			manual under the conditions for verbatim copying, provided also that the
64 :			sections entitled "Distribution" and "General Public License" are
65 :			included exactly as in the original, and provided that the entire
66 :			resulting derived work is distributed under the terms of a permission
67 :			notice identical to this one.
68 :
69 :			Permission is granted to copy and distribute translations of this manual
70 :			into another language, under the above conditions for modified versions,
71 :			except that the sections entitled "Distribution" and "General Public
72 :			License" may be included in a translation approved by the author instead
73 :			of in the original English.
74 :			@end titlepage
75 :
76 :
77 :			@node Top, License, (dir), (dir)
78 :			@ifinfo
79 :	anton	1.17	Gforth is a free implementation of ANS Forth available on many
80 :	anton	1.30	personal machines. This manual corresponds to version 0.2.
81 :	anton	1.1	@end ifinfo
82 :
83 :			@menu
84 :	anton	1.4	* License::
85 :	anton	1.17	* Goals:: About the Gforth Project
86 :	anton	1.4	* Other Books:: Things you might want to read
87 :	anton	1.17	* Invocation:: Starting Gforth
88 :			* Words:: Forth words available in Gforth
89 :	anton	1.4	* ANS conformance:: Implementation-defined options etc.
90 :	anton	1.17	* Model:: The abstract machine of Gforth
91 :			* Emacs and Gforth:: The Gforth Mode
92 :	anton	1.4	* Internals:: Implementation details
93 :			* Bugs:: How to report them
94 :	anton	1.29	* Origin:: Authors and ancestors of Gforth
95 :	anton	1.4	* Word Index:: An item for each Forth word
96 :			* Node Index:: An item for each node
97 :	anton	1.1	@end menu
98 :
99 :			@node License, Goals, Top, Top
100 :	pazsan	1.20	@unnumbered GNU GENERAL PUBLIC LICENSE
101 :			@center Version 2, June 1991
102 :
103 :			@display
104 :			Copyright @copyright{} 1989, 1991 Free Software Foundation, Inc.
105 :			675 Mass Ave, Cambridge, MA 02139, USA
106 :
107 :			Everyone is permitted to copy and distribute verbatim copies
108 :			of this license document, but changing it is not allowed.
109 :			@end display
110 :
111 :			@unnumberedsec Preamble
112 :
113 :			The licenses for most software are designed to take away your
114 :			freedom to share and change it. By contrast, the GNU General Public
115 :			License is intended to guarantee your freedom to share and change free
116 :			software---to make sure the software is free for all its users. This
117 :			General Public License applies to most of the Free Software
118 :			Foundation's software and to any other program whose authors commit to
119 :			using it. (Some other Free Software Foundation software is covered by
120 :			the GNU Library General Public License instead.) You can apply it to
121 :			your programs, too.
122 :
123 :			When we speak of free software, we are referring to freedom, not
124 :			price. Our General Public Licenses are designed to make sure that you
125 :			have the freedom to distribute copies of free software (and charge for
126 :			this service if you wish), that you receive source code or can get it
127 :			if you want it, that you can change the software or use pieces of it
128 :			in new free programs; and that you know you can do these things.
129 :
130 :			To protect your rights, we need to make restrictions that forbid
131 :			anyone to deny you these rights or to ask you to surrender the rights.
132 :			These restrictions translate to certain responsibilities for you if you
133 :			distribute copies of the software, or if you modify it.
134 :
135 :			For example, if you distribute copies of such a program, whether
136 :			gratis or for a fee, you must give the recipients all the rights that
137 :			you have. You must make sure that they, too, receive or can get the
138 :			source code. And you must show them these terms so they know their
139 :			rights.
140 :
141 :			We protect your rights with two steps: (1) copyright the software, and
142 :			(2) offer you this license which gives you legal permission to copy,
143 :			distribute and/or modify the software.
144 :
145 :			Also, for each author's protection and ours, we want to make certain
146 :			that everyone understands that there is no warranty for this free
147 :			software. If the software is modified by someone else and passed on, we
148 :			want its recipients to know that what they have is not the original, so
149 :			that any problems introduced by others will not reflect on the original
150 :			authors' reputations.
151 :
152 :			Finally, any free program is threatened constantly by software
153 :			patents. We wish to avoid the danger that redistributors of a free
154 :			program will individually obtain patent licenses, in effect making the
155 :			program proprietary. To prevent this, we have made it clear that any
156 :			patent must be licensed for everyone's free use or not licensed at all.
157 :
158 :			The precise terms and conditions for copying, distribution and
159 :			modification follow.
160 :
161 :			@iftex
162 :			@unnumberedsec TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
163 :			@end iftex
164 :			@ifinfo
165 :			@center TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
166 :			@end ifinfo
167 :
168 :			@enumerate 0
169 :			@item
170 :			This License applies to any program or other work which contains
171 :			a notice placed by the copyright holder saying it may be distributed
172 :			under the terms of this General Public License. The ``Program'', below,
173 :			refers to any such program or work, and a ``work based on the Program''
174 :			means either the Program or any derivative work under copyright law:
175 :			that is to say, a work containing the Program or a portion of it,
176 :			either verbatim or with modifications and/or translated into another
177 :			language. (Hereinafter, translation is included without limitation in
178 :			the term ``modification''.) Each licensee is addressed as ``you''.
179 :
180 :			Activities other than copying, distribution and modification are not
181 :			covered by this License; they are outside its scope. The act of
182 :			running the Program is not restricted, and the output from the Program
183 :			is covered only if its contents constitute a work based on the
184 :			Program (independent of having been made by running the Program).
185 :			Whether that is true depends on what the Program does.
186 :
187 :			@item
188 :			You may copy and distribute verbatim copies of the Program's
189 :			source code as you receive it, in any medium, provided that you
190 :			conspicuously and appropriately publish on each copy an appropriate
191 :			copyright notice and disclaimer of warranty; keep intact all the
192 :			notices that refer to this License and to the absence of any warranty;
193 :			and give any other recipients of the Program a copy of this License
194 :			along with the Program.
195 :
196 :			You may charge a fee for the physical act of transferring a copy, and
197 :			you may at your option offer warranty protection in exchange for a fee.
198 :
199 :			@item
200 :			You may modify your copy or copies of the Program or any portion
201 :			of it, thus forming a work based on the Program, and copy and
202 :			distribute such modifications or work under the terms of Section 1
203 :			above, provided that you also meet all of these conditions:
204 :
205 :			@enumerate a
206 :			@item
207 :			You must cause the modified files to carry prominent notices
208 :			stating that you changed the files and the date of any change.
209 :
210 :			@item
211 :			You must cause any work that you distribute or publish, that in
212 :			whole or in part contains or is derived from the Program or any
213 :			part thereof, to be licensed as a whole at no charge to all third
214 :			parties under the terms of this License.
215 :
216 :			@item
217 :			If the modified program normally reads commands interactively
218 :			when run, you must cause it, when started running for such
219 :			interactive use in the most ordinary way, to print or display an
220 :			announcement including an appropriate copyright notice and a
221 :			notice that there is no warranty (or else, saying that you provide
222 :			a warranty) and that users may redistribute the program under
223 :			these conditions, and telling the user how to view a copy of this
224 :			License. (Exception: if the Program itself is interactive but
225 :			does not normally print such an announcement, your work based on
226 :			the Program is not required to print an announcement.)
227 :			@end enumerate
228 :
229 :			These requirements apply to the modified work as a whole. If
230 :			identifiable sections of that work are not derived from the Program,
231 :			and can be reasonably considered independent and separate works in
232 :			themselves, then this License, and its terms, do not apply to those
233 :			sections when you distribute them as separate works. But when you
234 :			distribute the same sections as part of a whole which is a work based
235 :			on the Program, the distribution of the whole must be on the terms of
236 :			this License, whose permissions for other licensees extend to the
237 :			entire whole, and thus to each and every part regardless of who wrote it.
238 :
239 :			Thus, it is not the intent of this section to claim rights or contest
240 :			your rights to work written entirely by you; rather, the intent is to
241 :			exercise the right to control the distribution of derivative or
242 :			collective works based on the Program.
243 :
244 :			In addition, mere aggregation of another work not based on the Program
245 :			with the Program (or with a work based on the Program) on a volume of
246 :			a storage or distribution medium does not bring the other work under
247 :			the scope of this License.
248 :
249 :			@item
250 :			You may copy and distribute the Program (or a work based on it,
251 :			under Section 2) in object code or executable form under the terms of
252 :			Sections 1 and 2 above provided that you also do one of the following:
253 :
254 :			@enumerate a
255 :			@item
256 :			Accompany it with the complete corresponding machine-readable
257 :			source code, which must be distributed under the terms of Sections
258 :			1 and 2 above on a medium customarily used for software interchange; or,
259 :
260 :			@item
261 :			Accompany it with a written offer, valid for at least three
262 :			years, to give any third party, for a charge no more than your
263 :			cost of physically performing source distribution, a complete
264 :			machine-readable copy of the corresponding source code, to be
265 :			distributed under the terms of Sections 1 and 2 above on a medium
266 :			customarily used for software interchange; or,
267 :
268 :			@item
269 :			Accompany it with the information you received as to the offer
270 :			to distribute corresponding source code. (This alternative is
271 :			allowed only for noncommercial distribution and only if you
272 :			received the program in object code or executable form with such
273 :			an offer, in accord with Subsection b above.)
274 :			@end enumerate
275 :
276 :			The source code for a work means the preferred form of the work for
277 :			making modifications to it. For an executable work, complete source
278 :			code means all the source code for all modules it contains, plus any
279 :			associated interface definition files, plus the scripts used to
280 :			control compilation and installation of the executable. However, as a
281 :			special exception, the source code distributed need not include
282 :			anything that is normally distributed (in either source or binary
283 :			form) with the major components (compiler, kernel, and so on) of the
284 :			operating system on which the executable runs, unless that component
285 :			itself accompanies the executable.
286 :
287 :			If distribution of executable or object code is made by offering
288 :			access to copy from a designated place, then offering equivalent
289 :			access to copy the source code from the same place counts as
290 :			distribution of the source code, even though third parties are not
291 :			compelled to copy the source along with the object code.
292 :
293 :			@item
294 :			You may not copy, modify, sublicense, or distribute the Program
295 :			except as expressly provided under this License. Any attempt
296 :			otherwise to copy, modify, sublicense or distribute the Program is
297 :			void, and will automatically terminate your rights under this License.
298 :			However, parties who have received copies, or rights, from you under
299 :			this License will not have their licenses terminated so long as such
300 :			parties remain in full compliance.
301 :
302 :			@item
303 :			You are not required to accept this License, since you have not
304 :			signed it. However, nothing else grants you permission to modify or
305 :			distribute the Program or its derivative works. These actions are
306 :			prohibited by law if you do not accept this License. Therefore, by
307 :			modifying or distributing the Program (or any work based on the
308 :			Program), you indicate your acceptance of this License to do so, and
309 :			all its terms and conditions for copying, distributing or modifying
310 :			the Program or works based on it.
311 :
312 :			@item
313 :			Each time you redistribute the Program (or any work based on the
314 :			Program), the recipient automatically receives a license from the
315 :			original licensor to copy, distribute or modify the Program subject to
316 :			these terms and conditions. You may not impose any further
317 :			restrictions on the recipients' exercise of the rights granted herein.
318 :			You are not responsible for enforcing compliance by third parties to
319 :			this License.
320 :
321 :			@item
322 :			If, as a consequence of a court judgment or allegation of patent
323 :			infringement or for any other reason (not limited to patent issues),
324 :			conditions are imposed on you (whether by court order, agreement or
325 :			otherwise) that contradict the conditions of this License, they do not
326 :			excuse you from the conditions of this License. If you cannot
327 :			distribute so as to satisfy simultaneously your obligations under this
328 :			License and any other pertinent obligations, then as a consequence you
329 :			may not distribute the Program at all. For example, if a patent
330 :			license would not permit royalty-free redistribution of the Program by
331 :			all those who receive copies directly or indirectly through you, then
332 :			the only way you could satisfy both it and this License would be to
333 :			refrain entirely from distribution of the Program.
334 :
335 :			If any portion of this section is held invalid or unenforceable under
336 :			any particular circumstance, the balance of the section is intended to
337 :			apply and the section as a whole is intended to apply in other
338 :			circumstances.
339 :
340 :			It is not the purpose of this section to induce you to infringe any
341 :			patents or other property right claims or to contest validity of any
342 :			such claims; this section has the sole purpose of protecting the
343 :			integrity of the free software distribution system, which is
344 :			implemented by public license practices. Many people have made
345 :			generous contributions to the wide range of software distributed
346 :			through that system in reliance on consistent application of that
347 :			system; it is up to the author/donor to decide if he or she is willing
348 :			to distribute software through any other system and a licensee cannot
349 :			impose that choice.
350 :
351 :			This section is intended to make thoroughly clear what is believed to
352 :			be a consequence of the rest of this License.
353 :
354 :			@item
355 :			If the distribution and/or use of the Program is restricted in
356 :			certain countries either by patents or by copyrighted interfaces, the
357 :			original copyright holder who places the Program under this License
358 :			may add an explicit geographical distribution limitation excluding
359 :			those countries, so that distribution is permitted only in or among
360 :			countries not thus excluded. In such case, this License incorporates
361 :			the limitation as if written in the body of this License.
362 :
363 :			@item
364 :			The Free Software Foundation may publish revised and/or new versions
365 :			of the General Public License from time to time. Such new versions will
366 :			be similar in spirit to the present version, but may differ in detail to
367 :			address new problems or concerns.
368 :
369 :			Each version is given a distinguishing version number. If the Program
370 :			specifies a version number of this License which applies to it and ``any
371 :			later version'', you have the option of following the terms and conditions
372 :			either of that version or of any later version published by the Free
373 :			Software Foundation. If the Program does not specify a version number of
374 :			this License, you may choose any version ever published by the Free Software
375 :			Foundation.
376 :
377 :			@item
378 :			If you wish to incorporate parts of the Program into other free
379 :			programs whose distribution conditions are different, write to the author
380 :			to ask for permission. For software which is copyrighted by the Free
381 :			Software Foundation, write to the Free Software Foundation; we sometimes
382 :			make exceptions for this. Our decision will be guided by the two goals
383 :			of preserving the free status of all derivatives of our free software and
384 :			of promoting the sharing and reuse of software generally.
385 :
386 :			@iftex
387 :			@heading NO WARRANTY
388 :			@end iftex
389 :			@ifinfo
390 :			@center NO WARRANTY
391 :			@end ifinfo
392 :
393 :			@item
394 :			BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
395 :			FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN
396 :			OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
397 :			PROVIDE THE PROGRAM ``AS IS'' WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED
398 :			OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
399 :			MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS
400 :			TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE
401 :			PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING,
402 :			REPAIR OR CORRECTION.
403 :
404 :			@item
405 :			IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
406 :			WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
407 :			REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
408 :			INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
409 :			OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED
410 :			TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY
411 :			YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER
412 :			PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
413 :			POSSIBILITY OF SUCH DAMAGES.
414 :			@end enumerate
415 :
416 :			@iftex
417 :			@heading END OF TERMS AND CONDITIONS
418 :			@end iftex
419 :			@ifinfo
420 :			@center END OF TERMS AND CONDITIONS
421 :			@end ifinfo
422 :
423 :			@page
424 :			@unnumberedsec How to Apply These Terms to Your New Programs
425 :
426 :			If you develop a new program, and you want it to be of the greatest
427 :			possible use to the public, the best way to achieve this is to make it
428 :			free software which everyone can redistribute and change under these terms.
429 :
430 :			To do so, attach the following notices to the program. It is safest
431 :			to attach them to the start of each source file to most effectively
432 :			convey the exclusion of warranty; and each file should have at least
433 :			the ``copyright'' line and a pointer to where the full notice is found.
434 :
435 :			@smallexample
436 :			@var{one line to give the program's name and a brief idea of what it does.}
437 :			Copyright (C) 19@var{yy} @var{name of author}
438 :
439 :			This program is free software; you can redistribute it and/or modify
440 :			it under the terms of the GNU General Public License as published by
441 :			the Free Software Foundation; either version 2 of the License, or
442 :			(at your option) any later version.
443 :
444 :			This program is distributed in the hope that it will be useful,
445 :			but WITHOUT ANY WARRANTY; without even the implied warranty of
446 :			MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
447 :			GNU General Public License for more details.
448 :
449 :			You should have received a copy of the GNU General Public License
450 :			along with this program; if not, write to the Free Software
451 :			Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
452 :			@end smallexample
453 :
454 :			Also add information on how to contact you by electronic and paper mail.
455 :
456 :			If the program is interactive, make it output a short notice like this
457 :			when it starts in an interactive mode:
458 :
459 :			@smallexample
460 :			Gnomovision version 69, Copyright (C) 19@var{yy} @var{name of author}
461 :			Gnomovision comes with ABSOLUTELY NO WARRANTY; for details
462 :			type `show w'.
463 :			This is free software, and you are welcome to redistribute it
464 :			under certain conditions; type `show c' for details.
465 :			@end smallexample
466 :
467 :			The hypothetical commands @samp{show w} and @samp{show c} should show
468 :			the appropriate parts of the General Public License. Of course, the
469 :			commands you use may be called something other than @samp{show w} and
470 :			@samp{show c}; they could even be mouse-clicks or menu items---whatever
471 :			suits your program.
472 :
473 :			You should also get your employer (if you work as a programmer) or your
474 :			school, if any, to sign a ``copyright disclaimer'' for the program, if
475 :			necessary. Here is a sample; alter the names:
476 :
477 :			@smallexample
478 :			Yoyodyne, Inc., hereby disclaims all copyright interest in the program
479 :			`Gnomovision' (which makes passes at compilers) written by James Hacker.
480 :
481 :			@var{signature of Ty Coon}, 1 April 1989
482 :			Ty Coon, President of Vice
483 :			@end smallexample
484 :
485 :			This General Public License does not permit incorporating your program into
486 :			proprietary programs. If your program is a subroutine library, you may
487 :			consider it more useful to permit linking proprietary applications with the
488 :			library. If this is what you want to do, use the GNU Library General
489 :			Public License instead of this License.
490 :	anton	1.1
491 :			@iftex
492 :	pazsan	1.23	@node Preface
493 :			@comment node-name, next, previous, up
494 :	anton	1.1	@unnumbered Preface
495 :	pazsan	1.23	@cindex Preface
496 :	anton	1.17	This manual documents Gforth. The reader is expected to know
497 :	anton	1.1	Forth. This manual is primarily a reference manual. @xref{Other Books}
498 :			for introductory material.
499 :			@end iftex
500 :
501 :			@node Goals, Other Books, License, Top
502 :			@comment node-name, next, previous, up
503 :	anton	1.17	@chapter Goals of Gforth
504 :	anton	1.1	@cindex Goals
505 :	anton	1.17	The goal of the Gforth Project is to develop a standard model for
506 :	anton	1.1	ANSI Forth. This can be split into several subgoals:
507 :
508 :			@itemize @bullet
509 :			@item
510 :	anton	1.17	Gforth should conform to the ANSI Forth standard.
511 :	anton	1.1	@item
512 :			It should be a model, i.e. it should define all the
513 :			implementation-dependent things.
514 :			@item
515 :			It should become standard, i.e. widely accepted and used. This goal
516 :			is the most difficult one.
517 :			@end itemize
518 :
519 :	anton	1.17	To achieve these goals Gforth should be
520 :	anton	1.1	@itemize @bullet
521 :			@item
522 :			Similar to previous models (fig-Forth, F83)
523 :			@item
524 :			Powerful. It should provide for all the things that are considered
525 :			necessary today and even some that are not yet considered necessary.
526 :			@item
527 :			Efficient. It should not get the reputation of being exceptionally
528 :			slow.
529 :			@item
530 :			Free.
531 :			@item
532 :			Available on many machines/easy to port.
533 :			@end itemize
534 :
535 :	anton	1.17	Have we achieved these goals? Gforth conforms to the ANS Forth
536 :			standard. It may be considered a model, but we have not yet documented
537 :	anton	1.1	which parts of the model are stable and which parts we are likely to
538 :	anton	1.17	change. It certainly has not yet become a de facto standard. It has some
539 :			similarities and some differences to previous models. It has some
540 :			powerful features, but not yet everything that we envisioned. We
541 :			certainly have achieved our execution speed goals (@pxref{Performance}).
542 :			It is free and available on many machines.
543 :	anton	1.1
544 :			@node Other Books, Invocation, Goals, Top
545 :			@chapter Other books on ANS Forth
546 :
547 :			As the standard is relatively new, there are not many books out yet. It
548 :	anton	1.17	is not recommended to learn Forth by using Gforth and a book that is
549 :	anton	1.1	not written for ANS Forth, as you will not know your mistakes from the
550 :			deviations of the book.
551 :
552 :			There is, of course, the standard, the definite reference if you want to
553 :	anton	1.19	write ANS Forth programs. It is available in printed form from the
554 :			National Standards Institute Sales Department (Tel.: USA (212) 642-4900;
555 :			Fax.: USA (212) 302-1286) as document @cite{X3.215-1994} for about $200. You
556 :			can also get it from Global Engineering Documents (Tel.: USA (800)
557 :			854-7179; Fax.: (303) 843-9880) for about $300.
558 :
559 :			@cite{dpANS6}, the last draft of the standard, which was then submitted to ANSI
560 :			for publication is available electronically and for free in some MS Word
561 :			format, and it has been converted to HTML. Some pointers to these
562 :			versions can be found through
563 :	anton	1.24	@*@file{http://www.complang.tuwien.ac.at/projects/forth.html}.
564 :	anton	1.1
565 :	anton	1.21	@cite{Forth: The new model} by Jack Woehr (Prentice-Hall, 1993) is an
566 :	anton	1.1	introductory book based on a draft version of the standard. It does not
567 :			cover the whole standard. It also contains interesting background
568 :			information (Jack Woehr was in the ANS Forth Technical Committe). It is
569 :			not appropriate for complete newbies, but programmers experienced in
570 :			other languages should find it ok.
571 :
572 :			@node Invocation, Words, Other Books, Top
573 :			@chapter Invocation
574 :
575 :			You will usually just say @code{gforth}. In many other cases the default
576 :	anton	1.17	Gforth image will be invoked like this:
577 :	anton	1.1
578 :			@example
579 :			gforth [files] [-e forth-code]
580 :			@end example
581 :
582 :			executing the contents of the files and the Forth code in the order they
583 :			are given.
584 :
585 :			In general, the command line looks like this:
586 :
587 :			@example
588 :			gforth [initialization options] [image-specific options]
589 :			@end example
590 :
591 :			The initialization options must come before the rest of the command
592 :			line. They are:
593 :
594 :			@table @code
595 :			@item --image-file @var{file}
596 :	pazsan	1.20	@item -i @var{file}
597 :	anton	1.1	Loads the Forth image @var{file} instead of the default
598 :			@file{gforth.fi}.
599 :
600 :			@item --path @var{path}
601 :	pazsan	1.20	@item -p @var{path}
602 :	anton	1.1	Uses @var{path} for searching the image file and Forth source code
603 :			files instead of the default in the environment variable
604 :			@code{GFORTHPATH} or the path specified at installation time (typically
605 :			@file{/usr/local/lib/gforth:.}). A path is given as a @code{:}-separated
606 :			list.
607 :
608 :			@item --dictionary-size @var{size}
609 :			@item -m @var{size}
610 :			Allocate @var{size} space for the Forth dictionary space instead of
611 :			using the default specified in the image (typically 256K). The
612 :			@var{size} specification consists of an integer and a unit (e.g.,
613 :			@code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
614 :			size, in this case Cells), @code{k} (kilobytes), and @code{M}
615 :			(Megabytes). If no unit is specified, @code{e} is used.
616 :
617 :			@item --data-stack-size @var{size}
618 :			@item -d @var{size}
619 :			Allocate @var{size} space for the data stack instead of using the
620 :			default specified in the image (typically 16K).
621 :
622 :			@item --return-stack-size @var{size}
623 :			@item -r @var{size}
624 :			Allocate @var{size} space for the return stack instead of using the
625 :			default specified in the image (typically 16K).
626 :
627 :			@item --fp-stack-size @var{size}
628 :			@item -f @var{size}
629 :			Allocate @var{size} space for the floating point stack instead of
630 :			using the default specified in the image (typically 16K). In this case
631 :			the unit specifier @code{e} refers to floating point numbers.
632 :
633 :			@item --locals-stack-size @var{size}
634 :			@item -l @var{size}
635 :			Allocate @var{size} space for the locals stack instead of using the
636 :			default specified in the image (typically 16K).
637 :
638 :			@end table
639 :
640 :			As explained above, the image-specific command-line arguments for the
641 :			default image @file{gforth.fi} consist of a sequence of filenames and
642 :			@code{-e @var{forth-code}} options that are interpreted in the seqence
643 :			in which they are given. The @code{-e @var{forth-code}} or
644 :			@code{--evaluate @var{forth-code}} option evaluates the forth
645 :			code. This option takes only one argument; if you want to evaluate more
646 :			Forth words, you have to quote them or use several @code{-e}s. To exit
647 :			after processing the command line (instead of entering interactive mode)
648 :			append @code{-e bye} to the command line.
649 :
650 :	anton	1.22	If you have several versions of Gforth installed, @code{gforth} will
651 :			invoke the version that was installed last. @code{gforth-@var{version}}
652 :			invokes a specific version. You may want to use the option
653 :			@code{--path}, if your environment contains the variable
654 :			@code{GFORTHPATH}.
655 :
656 :	anton	1.1	Not yet implemented:
657 :			On startup the system first executes the system initialization file
658 :			(unless the option @code{--no-init-file} is given; note that the system
659 :			resulting from using this option may not be ANS Forth conformant). Then
660 :			the user initialization file @file{.gforth.fs} is executed, unless the
661 :			option @code{--no-rc} is given; this file is first searched in @file{.},
662 :			then in @file{~}, then in the normal path (see above).
663 :
664 :	anton	1.4	@node Words, ANS conformance, Invocation, Top
665 :	anton	1.1	@chapter Forth Words
666 :
667 :			@menu
668 :	anton	1.4	* Notation::
669 :			* Arithmetic::
670 :			* Stack Manipulation::
671 :			* Memory access::
672 :			* Control Structures::
673 :			* Locals::
674 :			* Defining Words::
675 :			* Wordlists::
676 :			* Files::
677 :			* Blocks::
678 :			* Other I/O::
679 :			* Programming Tools::
680 :	anton	1.18	* Assembler and Code words::
681 :	anton	1.4	* Threading Words::
682 :	anton	1.1	@end menu
683 :
684 :			@node Notation, Arithmetic, Words, Words
685 :			@section Notation
686 :
687 :			The Forth words are described in this section in the glossary notation
688 :			that has become a de-facto standard for Forth texts, i.e.
689 :
690 :	anton	1.4	@format
691 :	anton	1.1	@var{word} @var{Stack effect} @var{wordset} @var{pronunciation}
692 :	anton	1.4	@end format
693 :	anton	1.1	@var{Description}
694 :
695 :			@table @var
696 :			@item word
697 :	anton	1.17	The name of the word. BTW, Gforth is case insensitive, so you can
698 :	anton	1.14	type the words in in lower case (However, @pxref{core-idef}).
699 :	anton	1.1
700 :			@item Stack effect
701 :			The stack effect is written in the notation @code{@var{before} --
702 :			@var{after}}, where @var{before} and @var{after} describe the top of
703 :			stack entries before and after the execution of the word. The rest of
704 :			the stack is not touched by the word. The top of stack is rightmost,
705 :	anton	1.17	i.e., a stack sequence is written as it is typed in. Note that Gforth
706 :	anton	1.1	uses a separate floating point stack, but a unified stack
707 :			notation. Also, return stack effects are not shown in @var{stack
708 :			effect}, but in @var{Description}. The name of a stack item describes
709 :			the type and/or the function of the item. See below for a discussion of
710 :			the types.
711 :
712 :	anton	1.19	All words have two stack effects: A compile-time stack effect and a
713 :			run-time stack effect. The compile-time stack-effect of most words is
714 :			@var{ -- }. If the compile-time stack-effect of a word deviates from
715 :			this standard behaviour, or the word does other unusual things at
716 :			compile time, both stack effects are shown; otherwise only the run-time
717 :			stack effect is shown.
718 :
719 :	anton	1.1	@item pronunciation
720 :			How the word is pronounced
721 :
722 :			@item wordset
723 :			The ANS Forth standard is divided into several wordsets. A standard
724 :			system need not support all of them. So, the fewer wordsets your program
725 :			uses the more portable it will be in theory. However, we suspect that
726 :			most ANS Forth systems on personal machines will feature all
727 :			wordsets. Words that are not defined in the ANS standard have
728 :	anton	1.19	@code{gforth} or @code{gforth-internal} as wordset. @code{gforth}
729 :			describes words that will work in future releases of Gforth;
730 :			@code{gforth-internal} words are more volatile. Environmental query
731 :			strings are also displayed like words; you can recognize them by the
732 :			@code{environment} in the wordset field.
733 :	anton	1.1
734 :			@item Description
735 :			A description of the behaviour of the word.
736 :			@end table
737 :
738 :	anton	1.4	The type of a stack item is specified by the character(s) the name
739 :			starts with:
740 :	anton	1.1
741 :			@table @code
742 :			@item f
743 :			Bool, i.e. @code{false} or @code{true}.
744 :			@item c
745 :			Char
746 :			@item w
747 :			Cell, can contain an integer or an address
748 :			@item n
749 :			signed integer
750 :			@item u
751 :			unsigned integer
752 :			@item d
753 :			double sized signed integer
754 :			@item ud
755 :			double sized unsigned integer
756 :			@item r
757 :			Float
758 :			@item a_
759 :			Cell-aligned address
760 :			@item c_
761 :			Char-aligned address (note that a Char is two bytes in Windows NT)
762 :			@item f_
763 :			Float-aligned address
764 :			@item df_
765 :			Address aligned for IEEE double precision float
766 :			@item sf_
767 :			Address aligned for IEEE single precision float
768 :			@item xt
769 :			Execution token, same size as Cell
770 :			@item wid
771 :			Wordlist ID, same size as Cell
772 :			@item f83name
773 :			Pointer to a name structure
774 :			@end table
775 :
776 :	anton	1.4	@node Arithmetic, Stack Manipulation, Notation, Words
777 :	anton	1.1	@section Arithmetic
778 :			Forth arithmetic is not checked, i.e., you will not hear about integer
779 :			overflow on addition or multiplication, you may hear about division by
780 :			zero if you are lucky. The operator is written after the operands, but
781 :			the operands are still in the original order. I.e., the infix @code{2-1}
782 :			corresponds to @code{2 1 -}. Forth offers a variety of division
783 :			operators. If you perform division with potentially negative operands,
784 :			you do not want to use @code{/} or @code{/mod} with its undefined
785 :			behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
786 :	anton	1.4	former, @pxref{Mixed precision}).
787 :
788 :			@menu
789 :			* Single precision::
790 :			* Bitwise operations::
791 :			* Mixed precision:: operations with single and double-cell integers
792 :			* Double precision:: Double-cell integer arithmetic
793 :			* Floating Point::
794 :			@end menu
795 :	anton	1.1
796 :	anton	1.4	@node Single precision, Bitwise operations, Arithmetic, Arithmetic
797 :	anton	1.1	@subsection Single precision
798 :			doc-+
799 :			doc--
800 :			doc-*
801 :			doc-/
802 :			doc-mod
803 :			doc-/mod
804 :			doc-negate
805 :			doc-abs
806 :			doc-min
807 :			doc-max
808 :
809 :	anton	1.4	@node Bitwise operations, Mixed precision, Single precision, Arithmetic
810 :	anton	1.1	@subsection Bitwise operations
811 :			doc-and
812 :			doc-or
813 :			doc-xor
814 :			doc-invert
815 :			doc-2*
816 :			doc-2/
817 :
818 :	anton	1.4	@node Mixed precision, Double precision, Bitwise operations, Arithmetic
819 :	anton	1.1	@subsection Mixed precision
820 :			doc-m+
821 :			doc-*/
822 :			doc-*/mod
823 :			doc-m*
824 :			doc-um*
825 :			doc-m*/
826 :			doc-um/mod
827 :			doc-fm/mod
828 :			doc-sm/rem
829 :
830 :	anton	1.4	@node Double precision, Floating Point, Mixed precision, Arithmetic
831 :	anton	1.1	@subsection Double precision
832 :	anton	1.16
833 :			The outer (aka text) interpreter converts numbers containing a dot into
834 :			a double precision number. Note that only numbers with the dot as last
835 :			character are standard-conforming.
836 :
837 :	anton	1.1	doc-d+
838 :			doc-d-
839 :			doc-dnegate
840 :			doc-dabs
841 :			doc-dmin
842 :			doc-dmax
843 :
844 :	anton	1.4	@node Floating Point, , Double precision, Arithmetic
845 :			@subsection Floating Point
846 :	anton	1.16
847 :			The format of floating point numbers recognized by the outer (aka text)
848 :			interpreter is: a signed decimal number, possibly containing a decimal
849 :			point (@code{.}), followed by @code{E} or @code{e}, optionally followed
850 :			by a signed integer (the exponent). E.g., @code{1e} ist the same as
851 :			@code{+1.0e+1}. Note that a number without @code{e}
852 :			is not interpreted as floating-point number, but as double (if the
853 :			number contains a @code{.}) or single precision integer. Also,
854 :			conversions between string and floating point numbers always use base
855 :			10, irrespective of the value of @code{BASE}. If @code{BASE} contains a
856 :			value greater then 14, the @code{E} may be interpreted as digit and the
857 :			number will be interpreted as integer, unless it has a signed exponent
858 :			(both @code{+} and @code{-} are allowed as signs).
859 :	anton	1.4
860 :			Angles in floating point operations are given in radians (a full circle
861 :	anton	1.17	has 2 pi radians). Note, that Gforth has a separate floating point
862 :	anton	1.4	stack, but we use the unified notation.
863 :
864 :			Floating point numbers have a number of unpleasant surprises for the
865 :			unwary (e.g., floating point addition is not associative) and even a few
866 :			for the wary. You should not use them unless you know what you are doing
867 :			or you don't care that the results you get are totally bogus. If you
868 :			want to learn about the problems of floating point numbers (and how to
869 :	anton	1.11	avoid them), you might start with @cite{David Goldberg, What Every
870 :	anton	1.6	Computer Scientist Should Know About Floating-Point Arithmetic, ACM
871 :			Computing Surveys 23(1):5@minus{}48, March 1991}.
872 :	anton	1.4
873 :			doc-f+
874 :			doc-f-
875 :			doc-f*
876 :			doc-f/
877 :			doc-fnegate
878 :			doc-fabs
879 :			doc-fmax
880 :			doc-fmin
881 :			doc-floor
882 :			doc-fround
883 :			doc-f**
884 :			doc-fsqrt
885 :			doc-fexp
886 :			doc-fexpm1
887 :			doc-fln
888 :			doc-flnp1
889 :			doc-flog
890 :	anton	1.6	doc-falog
891 :	anton	1.4	doc-fsin
892 :			doc-fcos
893 :			doc-fsincos
894 :			doc-ftan
895 :			doc-fasin
896 :			doc-facos
897 :			doc-fatan
898 :			doc-fatan2
899 :			doc-fsinh
900 :			doc-fcosh
901 :			doc-ftanh
902 :			doc-fasinh
903 :			doc-facosh
904 :			doc-fatanh
905 :
906 :			@node Stack Manipulation, Memory access, Arithmetic, Words
907 :	anton	1.1	@section Stack Manipulation
908 :
909 :	anton	1.17	Gforth has a data stack (aka parameter stack) for characters, cells,
910 :	anton	1.1	addresses, and double cells, a floating point stack for floating point
911 :			numbers, a return stack for storing the return addresses of colon
912 :			definitions and other data, and a locals stack for storing local
913 :			variables. Note that while every sane Forth has a separate floating
914 :			point stack, this is not strictly required; an ANS Forth system could
915 :			theoretically keep floating point numbers on the data stack. As an
916 :			additional difficulty, you don't know how many cells a floating point
917 :			number takes. It is reportedly possible to write words in a way that
918 :			they work also for a unified stack model, but we do not recommend trying
919 :	anton	1.4	it. Instead, just say that your program has an environmental dependency
920 :			on a separate FP stack.
921 :
922 :			Also, a Forth system is allowed to keep the local variables on the
923 :	anton	1.1	return stack. This is reasonable, as local variables usually eliminate
924 :			the need to use the return stack explicitly. So, if you want to produce
925 :			a standard complying program and if you are using local variables in a
926 :			word, forget about return stack manipulations in that word (see the
927 :			standard document for the exact rules).
928 :
929 :	anton	1.4	@menu
930 :			* Data stack::
931 :			* Floating point stack::
932 :			* Return stack::
933 :			* Locals stack::
934 :			* Stack pointer manipulation::
935 :			@end menu
936 :
937 :			@node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
938 :	anton	1.1	@subsection Data stack
939 :			doc-drop
940 :			doc-nip
941 :			doc-dup
942 :			doc-over
943 :			doc-tuck
944 :			doc-swap
945 :			doc-rot
946 :			doc--rot
947 :			doc-?dup
948 :			doc-pick
949 :			doc-roll
950 :			doc-2drop
951 :			doc-2nip
952 :			doc-2dup
953 :			doc-2over
954 :			doc-2tuck
955 :			doc-2swap
956 :			doc-2rot
957 :
958 :	anton	1.4	@node Floating point stack, Return stack, Data stack, Stack Manipulation
959 :	anton	1.1	@subsection Floating point stack
960 :			doc-fdrop
961 :			doc-fnip
962 :			doc-fdup
963 :			doc-fover
964 :			doc-ftuck
965 :			doc-fswap
966 :			doc-frot
967 :
968 :	anton	1.4	@node Return stack, Locals stack, Floating point stack, Stack Manipulation
969 :	anton	1.1	@subsection Return stack
970 :			doc->r
971 :			doc-r>
972 :			doc-r@
973 :			doc-rdrop
974 :			doc-2>r
975 :			doc-2r>
976 :			doc-2r@
977 :			doc-2rdrop
978 :
979 :	anton	1.4	@node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
980 :	anton	1.1	@subsection Locals stack
981 :
982 :	anton	1.4	@node Stack pointer manipulation, , Locals stack, Stack Manipulation
983 :	anton	1.1	@subsection Stack pointer manipulation
984 :			doc-sp@
985 :			doc-sp!
986 :			doc-fp@
987 :			doc-fp!
988 :			doc-rp@
989 :			doc-rp!
990 :			doc-lp@
991 :			doc-lp!
992 :
993 :	anton	1.4	@node Memory access, Control Structures, Stack Manipulation, Words
994 :	anton	1.1	@section Memory access
995 :
996 :	anton	1.4	@menu
997 :			* Stack-Memory transfers::
998 :			* Address arithmetic::
999 :			* Memory block access::
1000 :			@end menu
1001 :
1002 :			@node Stack-Memory transfers, Address arithmetic, Memory access, Memory access
1003 :	anton	1.1	@subsection Stack-Memory transfers
1004 :
1005 :			doc-@
1006 :			doc-!
1007 :			doc-+!
1008 :			doc-c@
1009 :			doc-c!
1010 :			doc-2@
1011 :			doc-2!
1012 :			doc-f@
1013 :			doc-f!
1014 :			doc-sf@
1015 :			doc-sf!
1016 :			doc-df@
1017 :			doc-df!
1018 :
1019 :	anton	1.4	@node Address arithmetic, Memory block access, Stack-Memory transfers, Memory access
1020 :	anton	1.1	@subsection Address arithmetic
1021 :
1022 :			ANS Forth does not specify the sizes of the data types. Instead, it
1023 :			offers a number of words for computing sizes and doing address
1024 :			arithmetic. Basically, address arithmetic is performed in terms of
1025 :			address units (aus); on most systems the address unit is one byte. Note
1026 :			that a character may have more than one au, so @code{chars} is no noop
1027 :			(on systems where it is a noop, it compiles to nothing).
1028 :
1029 :			ANS Forth also defines words for aligning addresses for specific
1030 :			addresses. Many computers require that accesses to specific data types
1031 :			must only occur at specific addresses; e.g., that cells may only be
1032 :			accessed at addresses divisible by 4. Even if a machine allows unaligned
1033 :			accesses, it can usually perform aligned accesses faster.
1034 :
1035 :	anton	1.17	For the performance-conscious: alignment operations are usually only
1036 :	anton	1.1	necessary during the definition of a data structure, not during the
1037 :			(more frequent) accesses to it.
1038 :
1039 :			ANS Forth defines no words for character-aligning addresses. This is not
1040 :			an oversight, but reflects the fact that addresses that are not
1041 :			char-aligned have no use in the standard and therefore will not be
1042 :			created.
1043 :
1044 :			The standard guarantees that addresses returned by @code{CREATE}d words
1045 :	anton	1.17	are cell-aligned; in addition, Gforth guarantees that these addresses
1046 :	anton	1.1	are aligned for all purposes.
1047 :
1048 :	anton	1.9	Note that the standard defines a word @code{char}, which has nothing to
1049 :			do with address arithmetic.
1050 :
1051 :	anton	1.1	doc-chars
1052 :			doc-char+
1053 :			doc-cells
1054 :			doc-cell+
1055 :			doc-align
1056 :			doc-aligned
1057 :			doc-floats
1058 :			doc-float+
1059 :			doc-falign
1060 :			doc-faligned
1061 :			doc-sfloats
1062 :			doc-sfloat+
1063 :			doc-sfalign
1064 :			doc-sfaligned
1065 :			doc-dfloats
1066 :			doc-dfloat+
1067 :			doc-dfalign
1068 :			doc-dfaligned
1069 :	anton	1.10	doc-maxalign
1070 :			doc-maxaligned
1071 :			doc-cfalign
1072 :			doc-cfaligned
1073 :	anton	1.1	doc-address-unit-bits
1074 :
1075 :	anton	1.4	@node Memory block access, , Address arithmetic, Memory access
1076 :	anton	1.1	@subsection Memory block access
1077 :
1078 :			doc-move
1079 :			doc-erase
1080 :
1081 :			While the previous words work on address units, the rest works on
1082 :			characters.
1083 :
1084 :			doc-cmove
1085 :			doc-cmove>
1086 :			doc-fill
1087 :			doc-blank
1088 :
1089 :	anton	1.4	@node Control Structures, Locals, Memory access, Words
1090 :	anton	1.1	@section Control Structures
1091 :
1092 :			Control structures in Forth cannot be used in interpret state, only in
1093 :			compile state, i.e., in a colon definition. We do not like this
1094 :			limitation, but have not seen a satisfying way around it yet, although
1095 :			many schemes have been proposed.
1096 :
1097 :	anton	1.4	@menu
1098 :			* Selection::
1099 :			* Simple Loops::
1100 :			* Counted Loops::
1101 :			* Arbitrary control structures::
1102 :			* Calls and returns::
1103 :			* Exception Handling::
1104 :			@end menu
1105 :
1106 :			@node Selection, Simple Loops, Control Structures, Control Structures
1107 :	anton	1.1	@subsection Selection
1108 :
1109 :			@example
1110 :			@var{flag}
1111 :			IF
1112 :			@var{code}
1113 :			ENDIF
1114 :			@end example
1115 :			or
1116 :			@example
1117 :			@var{flag}
1118 :			IF
1119 :			@var{code1}
1120 :			ELSE
1121 :			@var{code2}
1122 :			ENDIF
1123 :			@end example
1124 :
1125 :	anton	1.4	You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
1126 :	anton	1.1	standard, and @code{ENDIF} is not, although it is quite popular. We
1127 :			recommend using @code{ENDIF}, because it is less confusing for people
1128 :			who also know other languages (and is not prone to reinforcing negative
1129 :			prejudices against Forth in these people). Adding @code{ENDIF} to a
1130 :			system that only supplies @code{THEN} is simple:
1131 :			@example
1132 :			: endif POSTPONE then ; immediate
1133 :			@end example
1134 :
1135 :			[According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
1136 :			(adv.)} has the following meanings:
1137 :			@quotation
1138 :			... 2b: following next after in order ... 3d: as a necessary consequence
1139 :			(if you were there, then you saw them).
1140 :			@end quotation
1141 :			Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
1142 :			and many other programming languages has the meaning 3d.]
1143 :
1144 :	anton	1.31	Gforth also provides the words @code{?dup-if} and @code{?dup-0=-if}, so
1145 :			you can avoid using @code{?dup}. Using these alternatives is also more
1146 :			efficient than using @code{?dup}. Definitions in plain standard Forth
1147 :			for @code{ENDIF}, @code{?DUP-IF} and @code{?DUP-0=-IF} are provided in
1148 :			@file{compat/control.fs}.
1149 :	anton	1.1
1150 :			@example
1151 :			@var{n}
1152 :			CASE
1153 :			@var{n1} OF @var{code1} ENDOF
1154 :			@var{n2} OF @var{code2} ENDOF
1155 :	anton	1.4	@dots{}
1156 :	anton	1.1	ENDCASE
1157 :			@end example
1158 :
1159 :			Executes the first @var{codei}, where the @var{ni} is equal to
1160 :			@var{n}. A default case can be added by simply writing the code after
1161 :			the last @code{ENDOF}. It may use @var{n}, which is on top of the stack,
1162 :			but must not consume it.
1163 :
1164 :	anton	1.4	@node Simple Loops, Counted Loops, Selection, Control Structures
1165 :	anton	1.1	@subsection Simple Loops
1166 :
1167 :			@example
1168 :			BEGIN
1169 :			@var{code1}
1170 :			@var{flag}
1171 :			WHILE
1172 :			@var{code2}
1173 :			REPEAT
1174 :			@end example
1175 :
1176 :			@var{code1} is executed and @var{flag} is computed. If it is true,
1177 :			@var{code2} is executed and the loop is restarted; If @var{flag} is false, execution continues after the @code{REPEAT}.
1178 :
1179 :			@example
1180 :			BEGIN
1181 :			@var{code}
1182 :			@var{flag}
1183 :			UNTIL
1184 :			@end example
1185 :
1186 :			@var{code} is executed. The loop is restarted if @code{flag} is false.
1187 :
1188 :			@example
1189 :			BEGIN
1190 :			@var{code}
1191 :			AGAIN
1192 :			@end example
1193 :
1194 :			This is an endless loop.
1195 :
1196 :	anton	1.4	@node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
1197 :	anton	1.1	@subsection Counted Loops
1198 :
1199 :			The basic counted loop is:
1200 :			@example
1201 :			@var{limit} @var{start}
1202 :			?DO
1203 :			@var{body}
1204 :			LOOP
1205 :			@end example
1206 :
1207 :			This performs one iteration for every integer, starting from @var{start}
1208 :			and up to, but excluding @var{limit}. The counter, aka index, can be
1209 :			accessed with @code{i}. E.g., the loop
1210 :			@example
1211 :			10 0 ?DO
1212 :			i .
1213 :			LOOP
1214 :			@end example
1215 :			prints
1216 :			@example
1217 :			0 1 2 3 4 5 6 7 8 9
1218 :			@end example
1219 :			The index of the innermost loop can be accessed with @code{i}, the index
1220 :			of the next loop with @code{j}, and the index of the third loop with
1221 :			@code{k}.
1222 :
1223 :			The loop control data are kept on the return stack, so there are some
1224 :			restrictions on mixing return stack accesses and counted loop
1225 :			words. E.g., if you put values on the return stack outside the loop, you
1226 :			cannot read them inside the loop. If you put values on the return stack
1227 :			within a loop, you have to remove them before the end of the loop and
1228 :			before accessing the index of the loop.
1229 :
1230 :			There are several variations on the counted loop:
1231 :
1232 :			@code{LEAVE} leaves the innermost counted loop immediately.
1233 :
1234 :	anton	1.18	If @var{start} is greater than @var{limit}, a @code{?DO} loop is entered
1235 :			(and @code{LOOP} iterates until they become equal by wrap-around
1236 :			arithmetic). This behaviour is usually not what you want. Therefore,
1237 :			Gforth offers @code{+DO} and @code{U+DO} (as replacements for
1238 :			@code{?DO}), which do not enter the loop if @var{start} is greater than
1239 :			@var{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
1240 :	anton	1.30	unsigned loop parameters.
1241 :	anton	1.18
1242 :	anton	1.1	@code{LOOP} can be replaced with @code{@var{n} +LOOP}; this updates the
1243 :			index by @var{n} instead of by 1. The loop is terminated when the border
1244 :			between @var{limit-1} and @var{limit} is crossed. E.g.:
1245 :
1246 :	anton	1.18	@code{4 0 +DO i . 2 +LOOP} prints @code{0 2}
1247 :	anton	1.1
1248 :	anton	1.18	@code{4 1 +DO i . 2 +LOOP} prints @code{1 3}
1249 :	anton	1.1
1250 :			The behaviour of @code{@var{n} +LOOP} is peculiar when @var{n} is negative:
1251 :
1252 :	anton	1.2	@code{-1 0 ?DO i . -1 +LOOP} prints @code{0 -1}
1253 :	anton	1.1
1254 :	anton	1.2	@code{ 0 0 ?DO i . -1 +LOOP} prints nothing
1255 :	anton	1.1
1256 :	anton	1.18	Therefore we recommend avoiding @code{@var{n} +LOOP} with negative
1257 :			@var{n}. One alternative is @code{@var{u} -LOOP}, which reduces the
1258 :			index by @var{u} each iteration. The loop is terminated when the border
1259 :			between @var{limit+1} and @var{limit} is crossed. Gforth also provides
1260 :			@code{-DO} and @code{U-DO} for down-counting loops. E.g.:
1261 :	anton	1.1
1262 :	anton	1.18	@code{-2 0 -DO i . 1 -LOOP} prints @code{0 -1}
1263 :	anton	1.1
1264 :	anton	1.18	@code{-1 0 -DO i . 1 -LOOP} prints @code{0}
1265 :	anton	1.1
1266 :	anton	1.18	@code{ 0 0 -DO i . 1 -LOOP} prints nothing
1267 :	anton	1.1
1268 :	anton	1.30	Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
1269 :			@code{-LOOP} are not in the ANS Forth standard. However, an
1270 :			implementation for these words that uses only standard words is provided
1271 :			in @file{compat/loops.fs}.
1272 :	anton	1.18
1273 :			@code{?DO} can also be replaced by @code{DO}. @code{DO} always enters
1274 :			the loop, independent of the loop parameters. Do not use @code{DO}, even
1275 :			if you know that the loop is entered in any case. Such knowledge tends
1276 :			to become invalid during maintenance of a program, and then the
1277 :			@code{DO} will make trouble.
1278 :	anton	1.1
1279 :			@code{UNLOOP} is used to prepare for an abnormal loop exit, e.g., via
1280 :			@code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
1281 :			return stack so @code{EXIT} can get to its return address.
1282 :
1283 :			Another counted loop is
1284 :			@example
1285 :			@var{n}
1286 :			FOR
1287 :			@var{body}
1288 :			NEXT
1289 :			@end example
1290 :			This is the preferred loop of native code compiler writers who are too
1291 :	anton	1.17	lazy to optimize @code{?DO} loops properly. In Gforth, this loop
1292 :	anton	1.1	iterates @var{n+1} times; @code{i} produces values starting with @var{n}
1293 :			and ending with 0. Other Forth systems may behave differently, even if
1294 :	anton	1.30	they support @code{FOR} loops. To avoid problems, don't use @code{FOR}
1295 :			loops.
1296 :	anton	1.1
1297 :	anton	1.4	@node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
1298 :	anton	1.2	@subsection Arbitrary control structures
1299 :
1300 :			ANS Forth permits and supports using control structures in a non-nested
1301 :			way. Information about incomplete control structures is stored on the
1302 :			control-flow stack. This stack may be implemented on the Forth data
1303 :	anton	1.17	stack, and this is what we have done in Gforth.
1304 :	anton	1.2
1305 :			An @i{orig} entry represents an unresolved forward branch, a @i{dest}
1306 :			entry represents a backward branch target. A few words are the basis for
1307 :			building any control structure possible (except control structures that
1308 :			need storage, like calls, coroutines, and backtracking).
1309 :
1310 :	anton	1.3	doc-if
1311 :			doc-ahead
1312 :			doc-then
1313 :			doc-begin
1314 :			doc-until
1315 :			doc-again
1316 :			doc-cs-pick
1317 :			doc-cs-roll
1318 :	anton	1.2
1319 :	anton	1.17	On many systems control-flow stack items take one word, in Gforth they
1320 :	anton	1.2	currently take three (this may change in the future). Therefore it is a
1321 :			really good idea to manipulate the control flow stack with
1322 :			@code{cs-pick} and @code{cs-roll}, not with data stack manipulation
1323 :			words.
1324 :
1325 :			Some standard control structure words are built from these words:
1326 :
1327 :	anton	1.3	doc-else
1328 :			doc-while
1329 :			doc-repeat
1330 :	anton	1.2
1331 :	anton	1.31	Gforth adds some more control-structure words:
1332 :
1333 :			doc-endif
1334 :			doc-?dup-if
1335 :			doc-?dup-0=-if
1336 :
1337 :	anton	1.2	Counted loop words constitute a separate group of words:
1338 :
1339 :	anton	1.3	doc-?do
1340 :	anton	1.18	doc-+do
1341 :			doc-u+do
1342 :			doc--do
1343 :			doc-u-do
1344 :	anton	1.3	doc-do
1345 :			doc-for
1346 :			doc-loop
1347 :			doc-+loop
1348 :	anton	1.18	doc--loop
1349 :	anton	1.3	doc-next
1350 :			doc-leave
1351 :			doc-?leave
1352 :			doc-unloop
1353 :	anton	1.10	doc-done
1354 :	anton	1.2
1355 :			The standard does not allow using @code{cs-pick} and @code{cs-roll} on
1356 :			@i{do-sys}. Our system allows it, but it's your job to ensure that for
1357 :			every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
1358 :	anton	1.3	through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
1359 :			fall-through path). Also, you have to ensure that all @code{LEAVE}s are
1360 :	pazsan	1.7	resolved (by using one of the loop-ending words or @code{DONE}).
1361 :	anton	1.2
1362 :			Another group of control structure words are
1363 :
1364 :	anton	1.3	doc-case
1365 :			doc-endcase
1366 :			doc-of
1367 :			doc-endof
1368 :	anton	1.2
1369 :			@i{case-sys} and @i{of-sys} cannot be processed using @code{cs-pick} and
1370 :			@code{cs-roll}.
1371 :
1372 :	anton	1.3	@subsubsection Programming Style
1373 :
1374 :			In order to ensure readability we recommend that you do not create
1375 :			arbitrary control structures directly, but define new control structure
1376 :			words for the control structure you want and use these words in your
1377 :			program.
1378 :
1379 :			E.g., instead of writing
1380 :
1381 :			@example
1382 :			begin
1383 :			...
1384 :			if [ 1 cs-roll ]
1385 :			...
1386 :			again then
1387 :			@end example
1388 :
1389 :			we recommend defining control structure words, e.g.,
1390 :
1391 :			@example
1392 :			: while ( dest -- orig dest )
1393 :			POSTPONE if
1394 :			1 cs-roll ; immediate
1395 :
1396 :			: repeat ( orig dest -- )
1397 :			POSTPONE again
1398 :			POSTPONE then ; immediate
1399 :			@end example
1400 :
1401 :			and then using these to create the control structure:
1402 :
1403 :			@example
1404 :			begin
1405 :			...
1406 :			while
1407 :			...
1408 :			repeat
1409 :			@end example
1410 :
1411 :	anton	1.30	That's much easier to read, isn't it? Of course, @code{REPEAT} and
1412 :	anton	1.3	@code{WHILE} are predefined, so in this example it would not be
1413 :			necessary to define them.
1414 :
1415 :	anton	1.4	@node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
1416 :	anton	1.3	@subsection Calls and returns
1417 :
1418 :			A definition can be called simply be writing the name of the
1419 :	anton	1.17	definition. When the end of the definition is reached, it returns. An
1420 :			earlier return can be forced using
1421 :	anton	1.3
1422 :			doc-exit
1423 :
1424 :			Don't forget to clean up the return stack and @code{UNLOOP} any
1425 :			outstanding @code{?DO}...@code{LOOP}s before @code{EXIT}ing. The
1426 :			primitive compiled by @code{EXIT} is
1427 :
1428 :			doc-;s
1429 :
1430 :	anton	1.4	@node Exception Handling, , Calls and returns, Control Structures
1431 :	anton	1.3	@subsection Exception Handling
1432 :
1433 :			doc-catch
1434 :			doc-throw
1435 :
1436 :	anton	1.4	@node Locals, Defining Words, Control Structures, Words
1437 :	anton	1.1	@section Locals
1438 :
1439 :	anton	1.2	Local variables can make Forth programming more enjoyable and Forth
1440 :			programs easier to read. Unfortunately, the locals of ANS Forth are
1441 :			laden with restrictions. Therefore, we provide not only the ANS Forth
1442 :			locals wordset, but also our own, more powerful locals wordset (we
1443 :			implemented the ANS Forth locals wordset through our locals wordset).
1444 :
1445 :	anton	1.24	The ideas in this section have also been published in the paper
1446 :			@cite{Automatic Scoping of Local Variables} by M. Anton Ertl, presented
1447 :			at EuroForth '94; it is available at
1448 :			@*@file{http://www.complang.tuwien.ac.at/papers/ertl94l.ps.gz}.
1449 :
1450 :	anton	1.2	@menu
1451 :	anton	1.17	* Gforth locals::
1452 :	anton	1.4	* ANS Forth locals::
1453 :	anton	1.2	@end menu
1454 :
1455 :	anton	1.17	@node Gforth locals, ANS Forth locals, Locals, Locals
1456 :			@subsection Gforth locals
1457 :	anton	1.2
1458 :			Locals can be defined with
1459 :
1460 :			@example
1461 :			@{ local1 local2 ... -- comment @}
1462 :			@end example
1463 :			or
1464 :			@example
1465 :			@{ local1 local2 ... @}
1466 :			@end example
1467 :
1468 :			E.g.,
1469 :			@example
1470 :			: max @{ n1 n2 -- n3 @}
1471 :			n1 n2 > if
1472 :			n1
1473 :			else
1474 :			n2
1475 :			endif ;
1476 :			@end example
1477 :
1478 :			The similarity of locals definitions with stack comments is intended. A
1479 :			locals definition often replaces the stack comment of a word. The order
1480 :			of the locals corresponds to the order in a stack comment and everything
1481 :			after the @code{--} is really a comment.
1482 :
1483 :			This similarity has one disadvantage: It is too easy to confuse locals
1484 :			declarations with stack comments, causing bugs and making them hard to
1485 :			find. However, this problem can be avoided by appropriate coding
1486 :			conventions: Do not use both notations in the same program. If you do,
1487 :			they should be distinguished using additional means, e.g. by position.
1488 :
1489 :			The name of the local may be preceded by a type specifier, e.g.,
1490 :			@code{F:} for a floating point value:
1491 :
1492 :			@example
1493 :			: CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
1494 :			\ complex multiplication
1495 :			Ar Br f* Ai Bi f* f-
1496 :			Ar Bi f* Ai Br f* f+ ;
1497 :			@end example
1498 :
1499 :	anton	1.17	Gforth currently supports cells (@code{W:}, @code{W^}), doubles
1500 :	anton	1.2	(@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
1501 :			(@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
1502 :			with @code{W:}, @code{D:} etc.) produces its value and can be changed
1503 :			with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
1504 :			produces its address (which becomes invalid when the variable's scope is
1505 :			left). E.g., the standard word @code{emit} can be defined in therms of
1506 :			@code{type} like this:
1507 :
1508 :			@example
1509 :			: emit @{ C^ char* -- @}
1510 :			char* 1 type ;
1511 :			@end example
1512 :
1513 :			A local without type specifier is a @code{W:} local. Both flavours of
1514 :			locals are initialized with values from the data or FP stack.
1515 :
1516 :			Currently there is no way to define locals with user-defined data
1517 :			structures, but we are working on it.
1518 :
1519 :	anton	1.17	Gforth allows defining locals everywhere in a colon definition. This
1520 :	pazsan	1.7	poses the following questions:
1521 :	anton	1.2
1522 :	anton	1.4	@menu
1523 :			* Where are locals visible by name?::
1524 :	anton	1.14	* How long do locals live?::
1525 :	anton	1.4	* Programming Style::
1526 :			* Implementation::
1527 :			@end menu
1528 :
1529 :	anton	1.17	@node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
1530 :	anton	1.2	@subsubsection Where are locals visible by name?
1531 :
1532 :			Basically, the answer is that locals are visible where you would expect
1533 :			it in block-structured languages, and sometimes a little longer. If you
1534 :			want to restrict the scope of a local, enclose its definition in
1535 :			@code{SCOPE}...@code{ENDSCOPE}.
1536 :
1537 :			doc-scope
1538 :			doc-endscope
1539 :
1540 :			These words behave like control structure words, so you can use them
1541 :			with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
1542 :			arbitrary ways.
1543 :
1544 :			If you want a more exact answer to the visibility question, here's the
1545 :			basic principle: A local is visible in all places that can only be
1546 :			reached through the definition of the local@footnote{In compiler
1547 :			construction terminology, all places dominated by the definition of the
1548 :			local.}. In other words, it is not visible in places that can be reached
1549 :			without going through the definition of the local. E.g., locals defined
1550 :			in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
1551 :			defined in @code{BEGIN}...@code{UNTIL} are visible after the
1552 :			@code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
1553 :
1554 :			The reasoning behind this solution is: We want to have the locals
1555 :			visible as long as it is meaningful. The user can always make the
1556 :			visibility shorter by using explicit scoping. In a place that can
1557 :			only be reached through the definition of a local, the meaning of a
1558 :			local name is clear. In other places it is not: How is the local
1559 :			initialized at the control flow path that does not contain the
1560 :			definition? Which local is meant, if the same name is defined twice in
1561 :			two independent control flow paths?
1562 :
1563 :			This should be enough detail for nearly all users, so you can skip the
1564 :			rest of this section. If you relly must know all the gory details and
1565 :			options, read on.
1566 :
1567 :			In order to implement this rule, the compiler has to know which places
1568 :			are unreachable. It knows this automatically after @code{AHEAD},
1569 :			@code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
1570 :			most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
1571 :			compiler that the control flow never reaches that place. If
1572 :			@code{UNREACHABLE} is not used where it could, the only consequence is
1573 :			that the visibility of some locals is more limited than the rule above
1574 :			says. If @code{UNREACHABLE} is used where it should not (i.e., if you
1575 :			lie to the compiler), buggy code will be produced.
1576 :
1577 :			Another problem with this rule is that at @code{BEGIN}, the compiler
1578 :	anton	1.3	does not know which locals will be visible on the incoming
1579 :			back-edge. All problems discussed in the following are due to this
1580 :			ignorance of the compiler (we discuss the problems using @code{BEGIN}
1581 :			loops as examples; the discussion also applies to @code{?DO} and other
1582 :	anton	1.2	loops). Perhaps the most insidious example is:
1583 :			@example
1584 :			AHEAD
1585 :			BEGIN
1586 :			x
1587 :			[ 1 CS-ROLL ] THEN
1588 :	anton	1.4	@{ x @}
1589 :	anton	1.2	...
1590 :			UNTIL
1591 :			@end example
1592 :
1593 :			This should be legal according to the visibility rule. The use of
1594 :			@code{x} can only be reached through the definition; but that appears
1595 :			textually below the use.
1596 :
1597 :			From this example it is clear that the visibility rules cannot be fully
1598 :			implemented without major headaches. Our implementation treats common
1599 :			cases as advertised and the exceptions are treated in a safe way: The
1600 :			compiler makes a reasonable guess about the locals visible after a
1601 :			@code{BEGIN}; if it is too pessimistic, the
1602 :			user will get a spurious error about the local not being defined; if the
1603 :			compiler is too optimistic, it will notice this later and issue a
1604 :			warning. In the case above the compiler would complain about @code{x}
1605 :			being undefined at its use. You can see from the obscure examples in
1606 :			this section that it takes quite unusual control structures to get the
1607 :			compiler into trouble, and even then it will often do fine.
1608 :
1609 :			If the @code{BEGIN} is reachable from above, the most optimistic guess
1610 :			is that all locals visible before the @code{BEGIN} will also be
1611 :			visible after the @code{BEGIN}. This guess is valid for all loops that
1612 :			are entered only through the @code{BEGIN}, in particular, for normal
1613 :			@code{BEGIN}...@code{WHILE}...@code{REPEAT} and
1614 :			@code{BEGIN}...@code{UNTIL} loops and it is implemented in our
1615 :			compiler. When the branch to the @code{BEGIN} is finally generated by
1616 :			@code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
1617 :			warns the user if it was too optimisitic:
1618 :			@example
1619 :			IF
1620 :	anton	1.4	@{ x @}
1621 :	anton	1.2	BEGIN
1622 :			\ x ?
1623 :			[ 1 cs-roll ] THEN
1624 :			...
1625 :			UNTIL
1626 :			@end example
1627 :
1628 :			Here, @code{x} lives only until the @code{BEGIN}, but the compiler
1629 :			optimistically assumes that it lives until the @code{THEN}. It notices
1630 :			this difference when it compiles the @code{UNTIL} and issues a
1631 :			warning. The user can avoid the warning, and make sure that @code{x}
1632 :			is not used in the wrong area by using explicit scoping:
1633 :			@example
1634 :			IF
1635 :			SCOPE
1636 :	anton	1.4	@{ x @}
1637 :	anton	1.2	ENDSCOPE
1638 :			BEGIN
1639 :			[ 1 cs-roll ] THEN
1640 :			...
1641 :			UNTIL
1642 :			@end example
1643 :
1644 :			Since the guess is optimistic, there will be no spurious error messages
1645 :			about undefined locals.
1646 :
1647 :			If the @code{BEGIN} is not reachable from above (e.g., after
1648 :			@code{AHEAD} or @code{EXIT}), the compiler cannot even make an
1649 :			optimistic guess, as the locals visible after the @code{BEGIN} may be
1650 :			defined later. Therefore, the compiler assumes that no locals are
1651 :	anton	1.17	visible after the @code{BEGIN}. However, the user can use
1652 :	anton	1.2	@code{ASSUME-LIVE} to make the compiler assume that the same locals are
1653 :	anton	1.17	visible at the BEGIN as at the point where the top control-flow stack
1654 :			item was created.
1655 :	anton	1.2
1656 :			doc-assume-live
1657 :
1658 :			E.g.,
1659 :			@example
1660 :	anton	1.4	@{ x @}
1661 :	anton	1.2	AHEAD
1662 :			ASSUME-LIVE
1663 :			BEGIN
1664 :			x
1665 :			[ 1 CS-ROLL ] THEN
1666 :			...
1667 :			UNTIL
1668 :			@end example
1669 :
1670 :			Other cases where the locals are defined before the @code{BEGIN} can be
1671 :			handled by inserting an appropriate @code{CS-ROLL} before the
1672 :			@code{ASSUME-LIVE} (and changing the control-flow stack manipulation
1673 :			behind the @code{ASSUME-LIVE}).
1674 :
1675 :			Cases where locals are defined after the @code{BEGIN} (but should be
1676 :			visible immediately after the @code{BEGIN}) can only be handled by
1677 :			rearranging the loop. E.g., the ``most insidious'' example above can be
1678 :			arranged into:
1679 :			@example
1680 :			BEGIN
1681 :	anton	1.4	@{ x @}
1682 :	anton	1.2	... 0=
1683 :			WHILE
1684 :			x
1685 :			REPEAT
1686 :			@end example
1687 :
1688 :	anton	1.17	@node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
1689 :	anton	1.2	@subsubsection How long do locals live?
1690 :
1691 :			The right answer for the lifetime question would be: A local lives at
1692 :			least as long as it can be accessed. For a value-flavoured local this
1693 :			means: until the end of its visibility. However, a variable-flavoured
1694 :			local could be accessed through its address far beyond its visibility
1695 :			scope. Ultimately, this would mean that such locals would have to be
1696 :			garbage collected. Since this entails un-Forth-like implementation
1697 :			complexities, I adopted the same cowardly solution as some other
1698 :			languages (e.g., C): The local lives only as long as it is visible;
1699 :			afterwards its address is invalid (and programs that access it
1700 :			afterwards are erroneous).
1701 :
1702 :	anton	1.17	@node Programming Style, Implementation, How long do locals live?, Gforth locals
1703 :	anton	1.2	@subsubsection Programming Style
1704 :
1705 :			The freedom to define locals anywhere has the potential to change
1706 :			programming styles dramatically. In particular, the need to use the
1707 :			return stack for intermediate storage vanishes. Moreover, all stack
1708 :			manipulations (except @code{PICK}s and @code{ROLL}s with run-time
1709 :			determined arguments) can be eliminated: If the stack items are in the
1710 :			wrong order, just write a locals definition for all of them; then
1711 :			write the items in the order you want.
1712 :
1713 :			This seems a little far-fetched and eliminating stack manipulations is
1714 :	anton	1.4	unlikely to become a conscious programming objective. Still, the number
1715 :			of stack manipulations will be reduced dramatically if local variables
1716 :	anton	1.17	are used liberally (e.g., compare @code{max} in @ref{Gforth locals} with
1717 :	anton	1.4	a traditional implementation of @code{max}).
1718 :	anton	1.2
1719 :			This shows one potential benefit of locals: making Forth programs more
1720 :			readable. Of course, this benefit will only be realized if the
1721 :			programmers continue to honour the principle of factoring instead of
1722 :			using the added latitude to make the words longer.
1723 :
1724 :			Using @code{TO} can and should be avoided. Without @code{TO},
1725 :			every value-flavoured local has only a single assignment and many
1726 :			advantages of functional languages apply to Forth. I.e., programs are
1727 :			easier to analyse, to optimize and to read: It is clear from the
1728 :			definition what the local stands for, it does not turn into something
1729 :			different later.
1730 :
1731 :			E.g., a definition using @code{TO} might look like this:
1732 :			@example
1733 :			: strcmp @{ addr1 u1 addr2 u2 -- n @}
1734 :			u1 u2 min 0
1735 :			?do
1736 :	anton	1.31	addr1 c@ addr2 c@ -
1737 :			?dup-if
1738 :	anton	1.2	unloop exit
1739 :			then
1740 :			addr1 char+ TO addr1
1741 :			addr2 char+ TO addr2
1742 :			loop
1743 :			u1 u2 - ;
1744 :			@end example
1745 :			Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
1746 :			every loop iteration. @code{strcmp} is a typical example of the
1747 :			readability problems of using @code{TO}. When you start reading
1748 :			@code{strcmp}, you think that @code{addr1} refers to the start of the
1749 :			string. Only near the end of the loop you realize that it is something
1750 :			else.
1751 :
1752 :			This can be avoided by defining two locals at the start of the loop that
1753 :			are initialized with the right value for the current iteration.
1754 :			@example
1755 :			: strcmp @{ addr1 u1 addr2 u2 -- n @}
1756 :			addr1 addr2
1757 :			u1 u2 min 0
1758 :			?do @{ s1 s2 @}
1759 :	anton	1.31	s1 c@ s2 c@ -
1760 :			?dup-if
1761 :	anton	1.2	unloop exit
1762 :			then
1763 :			s1 char+ s2 char+
1764 :			loop
1765 :			2drop
1766 :			u1 u2 - ;
1767 :			@end example
1768 :			Here it is clear from the start that @code{s1} has a different value
1769 :			in every loop iteration.
1770 :
1771 :	anton	1.17	@node Implementation, , Programming Style, Gforth locals
1772 :	anton	1.2	@subsubsection Implementation
1773 :
1774 :	anton	1.17	Gforth uses an extra locals stack. The most compelling reason for
1775 :	anton	1.2	this is that the return stack is not float-aligned; using an extra stack
1776 :			also eliminates the problems and restrictions of using the return stack
1777 :			as locals stack. Like the other stacks, the locals stack grows toward
1778 :			lower addresses. A few primitives allow an efficient implementation:
1779 :
1780 :			doc-@local#
1781 :			doc-f@local#
1782 :			doc-laddr#
1783 :			doc-lp+!#
1784 :			doc-lp!
1785 :			doc->l
1786 :			doc-f>l
1787 :
1788 :			In addition to these primitives, some specializations of these
1789 :			primitives for commonly occurring inline arguments are provided for
1790 :			efficiency reasons, e.g., @code{@@local0} as specialization of
1791 :			@code{@@local#} for the inline argument 0. The following compiling words
1792 :			compile the right specialized version, or the general version, as
1793 :			appropriate:
1794 :
1795 :	anton	1.12	doc-compile-@local
1796 :			doc-compile-f@local
1797 :	anton	1.2	doc-compile-lp+!
1798 :
1799 :			Combinations of conditional branches and @code{lp+!#} like
1800 :			@code{?branch-lp+!#} (the locals pointer is only changed if the branch
1801 :			is taken) are provided for efficiency and correctness in loops.
1802 :
1803 :			A special area in the dictionary space is reserved for keeping the
1804 :			local variable names. @code{@{} switches the dictionary pointer to this
1805 :			area and @code{@}} switches it back and generates the locals
1806 :			initializing code. @code{W:} etc.@ are normal defining words. This
1807 :			special area is cleared at the start of every colon definition.
1808 :
1809 :	anton	1.17	A special feature of Gforth's dictionary is used to implement the
1810 :	anton	1.2	definition of locals without type specifiers: every wordlist (aka
1811 :			vocabulary) has its own methods for searching
1812 :	anton	1.4	etc. (@pxref{Wordlists}). For the present purpose we defined a wordlist
1813 :	anton	1.2	with a special search method: When it is searched for a word, it
1814 :			actually creates that word using @code{W:}. @code{@{} changes the search
1815 :			order to first search the wordlist containing @code{@}}, @code{W:} etc.,
1816 :			and then the wordlist for defining locals without type specifiers.
1817 :
1818 :			The lifetime rules support a stack discipline within a colon
1819 :			definition: The lifetime of a local is either nested with other locals
1820 :			lifetimes or it does not overlap them.
1821 :
1822 :			At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
1823 :			pointer manipulation is generated. Between control structure words
1824 :			locals definitions can push locals onto the locals stack. @code{AGAIN}
1825 :			is the simplest of the other three control flow words. It has to
1826 :			restore the locals stack depth of the corresponding @code{BEGIN}
1827 :			before branching. The code looks like this:
1828 :			@format
1829 :			@code{lp+!#} current-locals-size @minus{} dest-locals-size
1830 :			@code{branch} <begin>
1831 :			@end format
1832 :
1833 :			@code{UNTIL} is a little more complicated: If it branches back, it
1834 :			must adjust the stack just like @code{AGAIN}. But if it falls through,
1835 :			the locals stack must not be changed. The compiler generates the
1836 :			following code:
1837 :			@format
1838 :			@code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
1839 :			@end format
1840 :			The locals stack pointer is only adjusted if the branch is taken.
1841 :
1842 :			@code{THEN} can produce somewhat inefficient code:
1843 :			@format
1844 :			@code{lp+!#} current-locals-size @minus{} orig-locals-size
1845 :			<orig target>:
1846 :			@code{lp+!#} orig-locals-size @minus{} new-locals-size
1847 :			@end format
1848 :			The second @code{lp+!#} adjusts the locals stack pointer from the
1849 :	anton	1.4	level at the @var{orig} point to the level after the @code{THEN}. The
1850 :	anton	1.2	first @code{lp+!#} adjusts the locals stack pointer from the current
1851 :			level to the level at the orig point, so the complete effect is an
1852 :			adjustment from the current level to the right level after the
1853 :			@code{THEN}.
1854 :
1855 :			In a conventional Forth implementation a dest control-flow stack entry
1856 :			is just the target address and an orig entry is just the address to be
1857 :			patched. Our locals implementation adds a wordlist to every orig or dest
1858 :			item. It is the list of locals visible (or assumed visible) at the point
1859 :			described by the entry. Our implementation also adds a tag to identify
1860 :			the kind of entry, in particular to differentiate between live and dead
1861 :			(reachable and unreachable) orig entries.
1862 :
1863 :			A few unusual operations have to be performed on locals wordlists:
1864 :
1865 :			doc-common-list
1866 :			doc-sub-list?
1867 :			doc-list-size
1868 :
1869 :			Several features of our locals wordlist implementation make these
1870 :			operations easy to implement: The locals wordlists are organised as
1871 :			linked lists; the tails of these lists are shared, if the lists
1872 :			contain some of the same locals; and the address of a name is greater
1873 :			than the address of the names behind it in the list.
1874 :
1875 :			Another important implementation detail is the variable
1876 :			@code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
1877 :			determine if they can be reached directly or only through the branch
1878 :			that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
1879 :			@code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
1880 :			definition, by @code{BEGIN} and usually by @code{THEN}.
1881 :
1882 :			Counted loops are similar to other loops in most respects, but
1883 :			@code{LEAVE} requires special attention: It performs basically the same
1884 :			service as @code{AHEAD}, but it does not create a control-flow stack
1885 :			entry. Therefore the information has to be stored elsewhere;
1886 :			traditionally, the information was stored in the target fields of the
1887 :			branches created by the @code{LEAVE}s, by organizing these fields into a
1888 :			linked list. Unfortunately, this clever trick does not provide enough
1889 :			space for storing our extended control flow information. Therefore, we
1890 :			introduce another stack, the leave stack. It contains the control-flow
1891 :			stack entries for all unresolved @code{LEAVE}s.
1892 :
1893 :			Local names are kept until the end of the colon definition, even if
1894 :			they are no longer visible in any control-flow path. In a few cases
1895 :			this may lead to increased space needs for the locals name area, but
1896 :			usually less than reclaiming this space would cost in code size.
1897 :
1898 :
1899 :	anton	1.17	@node ANS Forth locals, , Gforth locals, Locals
1900 :	anton	1.2	@subsection ANS Forth locals
1901 :
1902 :			The ANS Forth locals wordset does not define a syntax for locals, but
1903 :			words that make it possible to define various syntaxes. One of the
1904 :	anton	1.17	possible syntaxes is a subset of the syntax we used in the Gforth locals
1905 :	anton	1.2	wordset, i.e.:
1906 :
1907 :			@example
1908 :			@{ local1 local2 ... -- comment @}
1909 :			@end example
1910 :			or
1911 :			@example
1912 :			@{ local1 local2 ... @}
1913 :			@end example
1914 :
1915 :			The order of the locals corresponds to the order in a stack comment. The
1916 :			restrictions are:
1917 :	anton	1.1
1918 :	anton	1.2	@itemize @bullet
1919 :			@item
1920 :	anton	1.17	Locals can only be cell-sized values (no type specifiers are allowed).
1921 :	anton	1.2	@item
1922 :			Locals can be defined only outside control structures.
1923 :			@item
1924 :			Locals can interfere with explicit usage of the return stack. For the
1925 :			exact (and long) rules, see the standard. If you don't use return stack
1926 :	anton	1.17	accessing words in a definition using locals, you will be all right. The
1927 :	anton	1.2	purpose of this rule is to make locals implementation on the return
1928 :			stack easier.
1929 :			@item
1930 :			The whole definition must be in one line.
1931 :			@end itemize
1932 :
1933 :			Locals defined in this way behave like @code{VALUE}s
1934 :	anton	1.4	(@xref{Values}). I.e., they are initialized from the stack. Using their
1935 :	anton	1.2	name produces their value. Their value can be changed using @code{TO}.
1936 :
1937 :	anton	1.17	Since this syntax is supported by Gforth directly, you need not do
1938 :	anton	1.2	anything to use it. If you want to port a program using this syntax to
1939 :	anton	1.30	another ANS Forth system, use @file{compat/anslocal.fs} to implement the
1940 :			syntax on the other system.
1941 :	anton	1.2
1942 :			Note that a syntax shown in the standard, section A.13 looks
1943 :			similar, but is quite different in having the order of locals
1944 :			reversed. Beware!
1945 :
1946 :			The ANS Forth locals wordset itself consists of the following word
1947 :
1948 :			doc-(local)
1949 :
1950 :			The ANS Forth locals extension wordset defines a syntax, but it is so
1951 :			awful that we strongly recommend not to use it. We have implemented this
1952 :	anton	1.17	syntax to make porting to Gforth easy, but do not document it here. The
1953 :	anton	1.2	problem with this syntax is that the locals are defined in an order
1954 :			reversed with respect to the standard stack comment notation, making
1955 :			programs harder to read, and easier to misread and miswrite. The only
1956 :			merit of this syntax is that it is easy to implement using the ANS Forth
1957 :			locals wordset.
1958 :	anton	1.3
1959 :	anton	1.4	@node Defining Words, Wordlists, Locals, Words
1960 :			@section Defining Words
1961 :
1962 :	anton	1.14	@menu
1963 :			* Values::
1964 :			@end menu
1965 :
1966 :	anton	1.4	@node Values, , Defining Words, Defining Words
1967 :			@subsection Values
1968 :
1969 :			@node Wordlists, Files, Defining Words, Words
1970 :			@section Wordlists
1971 :
1972 :			@node Files, Blocks, Wordlists, Words
1973 :			@section Files
1974 :
1975 :			@node Blocks, Other I/O, Files, Words
1976 :			@section Blocks
1977 :
1978 :			@node Other I/O, Programming Tools, Blocks, Words
1979 :			@section Other I/O
1980 :
1981 :	anton	1.18	@node Programming Tools, Assembler and Code words, Other I/O, Words
1982 :	anton	1.4	@section Programming Tools
1983 :
1984 :	anton	1.5	@menu
1985 :			* Debugging:: Simple and quick.
1986 :			* Assertions:: Making your programs self-checking.
1987 :			@end menu
1988 :
1989 :			@node Debugging, Assertions, Programming Tools, Programming Tools
1990 :	anton	1.4	@subsection Debugging
1991 :
1992 :			The simple debugging aids provided in @file{debugging.fs}
1993 :			are meant to support a different style of debugging than the
1994 :			tracing/stepping debuggers used in languages with long turn-around
1995 :			times.
1996 :
1997 :			A much better (faster) way in fast-compilig languages is to add
1998 :			printing code at well-selected places, let the program run, look at
1999 :			the output, see where things went wrong, add more printing code, etc.,
2000 :			until the bug is found.
2001 :
2002 :			The word @code{~~} is easy to insert. It just prints debugging
2003 :			information (by default the source location and the stack contents). It
2004 :			is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
2005 :			query-replace them with nothing). The deferred words
2006 :			@code{printdebugdata} and @code{printdebugline} control the output of
2007 :			@code{~~}. The default source location output format works well with
2008 :			Emacs' compilation mode, so you can step through the program at the
2009 :	anton	1.5	source level using @kbd{C-x `} (the advantage over a stepping debugger
2010 :			is that you can step in any direction and you know where the crash has
2011 :			happened or where the strange data has occurred).
2012 :	anton	1.4
2013 :			Note that the default actions clobber the contents of the pictured
2014 :			numeric output string, so you should not use @code{~~}, e.g., between
2015 :			@code{<#} and @code{#>}.
2016 :
2017 :			doc-~~
2018 :			doc-printdebugdata
2019 :			doc-printdebugline
2020 :
2021 :	anton	1.5	@node Assertions, , Debugging, Programming Tools
2022 :	anton	1.4	@subsection Assertions
2023 :
2024 :	anton	1.5	It is a good idea to make your programs self-checking, in particular, if
2025 :			you use an assumption (e.g., that a certain field of a data structure is
2026 :	anton	1.17	never zero) that may become wrong during maintenance. Gforth supports
2027 :	anton	1.5	assertions for this purpose. They are used like this:
2028 :
2029 :			@example
2030 :			assert( @var{flag} )
2031 :			@end example
2032 :
2033 :			The code between @code{assert(} and @code{)} should compute a flag, that
2034 :			should be true if everything is alright and false otherwise. It should
2035 :			not change anything else on the stack. The overall stack effect of the
2036 :			assertion is @code{( -- )}. E.g.
2037 :
2038 :			@example
2039 :			assert( 1 1 + 2 = ) \ what we learn in school
2040 :			assert( dup 0<> ) \ assert that the top of stack is not zero
2041 :			assert( false ) \ this code should not be reached
2042 :			@end example
2043 :
2044 :			The need for assertions is different at different times. During
2045 :			debugging, we want more checking, in production we sometimes care more
2046 :			for speed. Therefore, assertions can be turned off, i.e., the assertion
2047 :			becomes a comment. Depending on the importance of an assertion and the
2048 :			time it takes to check it, you may want to turn off some assertions and
2049 :	anton	1.17	keep others turned on. Gforth provides several levels of assertions for
2050 :	anton	1.5	this purpose:
2051 :
2052 :			doc-assert0(
2053 :			doc-assert1(
2054 :			doc-assert2(
2055 :			doc-assert3(
2056 :			doc-assert(
2057 :			doc-)
2058 :
2059 :			@code{Assert(} is the same as @code{assert1(}. The variable
2060 :			@code{assert-level} specifies the highest assertions that are turned
2061 :			on. I.e., at the default @code{assert-level} of one, @code{assert0(} and
2062 :			@code{assert1(} assertions perform checking, while @code{assert2(} and
2063 :			@code{assert3(} assertions are treated as comments.
2064 :
2065 :			Note that the @code{assert-level} is evaluated at compile-time, not at
2066 :			run-time. I.e., you cannot turn assertions on or off at run-time, you
2067 :			have to set the @code{assert-level} appropriately before compiling a
2068 :			piece of code. You can compile several pieces of code at several
2069 :			@code{assert-level}s (e.g., a trusted library at level 1 and newly
2070 :			written code at level 3).
2071 :
2072 :			doc-assert-level
2073 :
2074 :			If an assertion fails, a message compatible with Emacs' compilation mode
2075 :			is produced and the execution is aborted (currently with @code{ABORT"}.
2076 :			If there is interest, we will introduce a special throw code. But if you
2077 :			intend to @code{catch} a specific condition, using @code{throw} is
2078 :			probably more appropriate than an assertion).
2079 :
2080 :	anton	1.18	@node Assembler and Code words, Threading Words, Programming Tools, Words
2081 :			@section Assembler and Code words
2082 :
2083 :			Gforth provides some words for defining primitives (words written in
2084 :			machine code), and for defining the the machine-code equivalent of
2085 :			@code{DOES>}-based defining words. However, the machine-independent
2086 :			nature of Gforth poses a few problems: First of all. Gforth runs on
2087 :			several architectures, so it can provide no standard assembler. What's
2088 :			worse is that the register allocation not only depends on the processor,
2089 :	anton	1.25	but also on the @code{gcc} version and options used.
2090 :	anton	1.18
2091 :	anton	1.25	The words that Gforth offers encapsulate some system dependences (e.g., the
2092 :	anton	1.18	header structure), so a system-independent assembler may be used in
2093 :			Gforth. If you do not have an assembler, you can compile machine code
2094 :			directly with @code{,} and @code{c,}.
2095 :
2096 :			doc-assembler
2097 :			doc-code
2098 :			doc-end-code
2099 :			doc-;code
2100 :			doc-flush-icache
2101 :
2102 :			If @code{flush-icache} does not work correctly, @code{code} words
2103 :			etc. will not work (reliably), either.
2104 :
2105 :			These words are rarely used. Therefore they reside in @code{code.fs},
2106 :			which is usually not loaded (except @code{flush-icache}, which is always
2107 :	anton	1.19	present). You can load them with @code{require code.fs}.
2108 :	anton	1.18
2109 :	anton	1.25	In the assembly code you will want to refer to the inner interpreter's
2110 :			registers (e.g., the data stack pointer) and you may want to use other
2111 :			registers for temporary storage. Unfortunately, the register allocation
2112 :			is installation-dependent.
2113 :
2114 :			The easiest solution is to use explicit register declarations
2115 :			(@pxref{Explicit Reg Vars, , Variables in Specified Registers, gcc.info,
2116 :			GNU C Manual}) for all of the inner interpreter's registers: You have to
2117 :			compile Gforth with @code{-DFORCE_REG} (configure option
2118 :			@code{--enable-force-reg}) and the appropriate declarations must be
2119 :			present in the @code{machine.h} file (see @code{mips.h} for an example;
2120 :			you can find a full list of all declarable register symbols with
2121 :			@code{grep register engine.c}). If you give explicit registers to all
2122 :			variables that are declared at the beginning of @code{engine()}, you
2123 :			should be able to use the other caller-saved registers for temporary
2124 :			storage. Alternatively, you can use the @code{gcc} option
2125 :			@code{-ffixed-REG} (@pxref{Code Gen Options, , Options for Code
2126 :			Generation Conventions, gcc.info, GNU C Manual}) to reserve a register
2127 :			(however, this restriction on register allocation may slow Gforth
2128 :			significantly).
2129 :
2130 :			If this solution is not viable (e.g., because @code{gcc} does not allow
2131 :			you to explicitly declare all the registers you need), you have to find
2132 :			out by looking at the code where the inner interpreter's registers
2133 :			reside and which registers can be used for temporary storage. You can
2134 :			get an assembly listing of the engine's code with @code{make engine.s}.
2135 :
2136 :			In any case, it is good practice to abstract your assembly code from the
2137 :			actual register allocation. E.g., if the data stack pointer resides in
2138 :			register @code{$17}, create an alias for this register called @code{sp},
2139 :			and use that in your assembly code.
2140 :
2141 :	anton	1.18	Another option for implementing normal and defining words efficiently
2142 :			is: adding the wanted functionality to the source of Gforth. For normal
2143 :			words you just have to edit @file{primitives}, defining words (for fast
2144 :	anton	1.25	defined words) may require changes in @file{engine.c},
2145 :	anton	1.18	@file{kernal.fs}, @file{prims2x.fs}, and possibly @file{cross.fs}.
2146 :
2147 :
2148 :			@node Threading Words, , Assembler and Code words, Words
2149 :	anton	1.4	@section Threading Words
2150 :
2151 :			These words provide access to code addresses and other threading stuff
2152 :	anton	1.17	in Gforth (and, possibly, other interpretive Forths). It more or less
2153 :	anton	1.4	abstracts away the differences between direct and indirect threading
2154 :			(and, for direct threading, the machine dependences). However, at
2155 :			present this wordset is still inclomplete. It is also pretty low-level;
2156 :			some day it will hopefully be made unnecessary by an internals words set
2157 :			that abstracts implementation details away completely.
2158 :
2159 :			doc->code-address
2160 :			doc->does-code
2161 :			doc-code-address!
2162 :			doc-does-code!
2163 :			doc-does-handler!
2164 :			doc-/does-handler
2165 :
2166 :	anton	1.18	The code addresses produced by various defining words are produced by
2167 :			the following words:
2168 :	anton	1.14
2169 :	anton	1.18	doc-docol:
2170 :			doc-docon:
2171 :			doc-dovar:
2172 :			doc-douser:
2173 :			doc-dodefer:
2174 :			doc-dofield:
2175 :
2176 :			Currently there is no installation-independent way for recogizing words
2177 :			defined by a @code{CREATE}...@code{DOES>} word; however, once you know
2178 :			that a word is defined by a @code{CREATE}...@code{DOES>} word, you can
2179 :			use @code{>DOES-CODE}.
2180 :	anton	1.14
2181 :	anton	1.4	@node ANS conformance, Model, Words, Top
2182 :			@chapter ANS conformance
2183 :
2184 :	anton	1.17	To the best of our knowledge, Gforth is an
2185 :	anton	1.14
2186 :	anton	1.15	ANS Forth System
2187 :			@itemize
2188 :			@item providing the Core Extensions word set
2189 :			@item providing the Block word set
2190 :			@item providing the Block Extensions word set
2191 :			@item providing the Double-Number word set
2192 :			@item providing the Double-Number Extensions word set
2193 :			@item providing the Exception word set
2194 :			@item providing the Exception Extensions word set
2195 :			@item providing the Facility word set
2196 :			@item providing @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
2197 :			@item providing the File Access word set
2198 :			@item providing the File Access Extensions word set
2199 :			@item providing the Floating-Point word set
2200 :			@item providing the Floating-Point Extensions word set
2201 :			@item providing the Locals word set
2202 :			@item providing the Locals Extensions word set
2203 :			@item providing the Memory-Allocation word set
2204 :			@item providing the Memory-Allocation Extensions word set (that one's easy)
2205 :			@item providing the Programming-Tools word set
2206 :	anton	1.18	@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
2207 :	anton	1.15	@item providing the Search-Order word set
2208 :			@item providing the Search-Order Extensions word set
2209 :			@item providing the String word set
2210 :			@item providing the String Extensions word set (another easy one)
2211 :			@end itemize
2212 :
2213 :			In addition, ANS Forth systems are required to document certain
2214 :			implementation choices. This chapter tries to meet these
2215 :			requirements. In many cases it gives a way to ask the system for the
2216 :			information instead of providing the information directly, in
2217 :			particular, if the information depends on the processor, the operating
2218 :			system or the installation options chosen, or if they are likely to
2219 :	anton	1.17	change during the maintenance of Gforth.
2220 :	anton	1.15
2221 :	anton	1.14	@comment The framework for the rest has been taken from pfe.
2222 :
2223 :			@menu
2224 :			* The Core Words::
2225 :			* The optional Block word set::
2226 :			* The optional Double Number word set::
2227 :			* The optional Exception word set::
2228 :			* The optional Facility word set::
2229 :			* The optional File-Access word set::
2230 :			* The optional Floating-Point word set::
2231 :			* The optional Locals word set::
2232 :			* The optional Memory-Allocation word set::
2233 :			* The optional Programming-Tools word set::
2234 :			* The optional Search-Order word set::
2235 :			@end menu
2236 :
2237 :
2238 :			@c =====================================================================
2239 :			@node The Core Words, The optional Block word set, ANS conformance, ANS conformance
2240 :			@comment node-name, next, previous, up
2241 :			@section The Core Words
2242 :			@c =====================================================================
2243 :
2244 :			@menu
2245 :	anton	1.15	* core-idef:: Implementation Defined Options
2246 :			* core-ambcond:: Ambiguous Conditions
2247 :			* core-other:: Other System Documentation
2248 :	anton	1.14	@end menu
2249 :
2250 :			@c ---------------------------------------------------------------------
2251 :			@node core-idef, core-ambcond, The Core Words, The Core Words
2252 :			@subsection Implementation Defined Options
2253 :			@c ---------------------------------------------------------------------
2254 :
2255 :			@table @i
2256 :
2257 :			@item (Cell) aligned addresses:
2258 :	anton	1.17	processor-dependent. Gforth's alignment words perform natural alignment
2259 :	anton	1.14	(e.g., an address aligned for a datum of size 8 is divisible by
2260 :			8). Unaligned accesses usually result in a @code{-23 THROW}.
2261 :
2262 :			@item @code{EMIT} and non-graphic characters:
2263 :			The character is output using the C library function (actually, macro)
2264 :			@code{putchar}.
2265 :
2266 :			@item character editing of @code{ACCEPT} and @code{EXPECT}:
2267 :			This is modeled on the GNU readline library (@pxref{Readline
2268 :			Interaction, , Command Line Editing, readline, The GNU Readline
2269 :			Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
2270 :			producing a full word completion every time you type it (instead of
2271 :			producing the common prefix of all completions).
2272 :
2273 :			@item character set:
2274 :			The character set of your computer and display device. Gforth is
2275 :			8-bit-clean (but some other component in your system may make trouble).
2276 :
2277 :			@item Character-aligned address requirements:
2278 :			installation-dependent. Currently a character is represented by a C
2279 :			@code{unsigned char}; in the future we might switch to @code{wchar_t}
2280 :			(Comments on that requested).
2281 :
2282 :			@item character-set extensions and matching of names:
2283 :	anton	1.17	Any character except the ASCII NUL charcter can be used in a
2284 :			name. Matching is case-insensitive. The matching is performed using the
2285 :			C function @code{strncasecmp}, whose function is probably influenced by
2286 :			the locale. E.g., the @code{C} locale does not know about accents and
2287 :	anton	1.14	umlauts, so they are matched case-sensitively in that locale. For
2288 :			portability reasons it is best to write programs such that they work in
2289 :			the @code{C} locale. Then one can use libraries written by a Polish
2290 :			programmer (who might use words containing ISO Latin-2 encoded
2291 :			characters) and by a French programmer (ISO Latin-1) in the same program
2292 :			(of course, @code{WORDS} will produce funny results for some of the
2293 :			words (which ones, depends on the font you are using)). Also, the locale
2294 :			you prefer may not be available in other operating systems. Hopefully,
2295 :			Unicode will solve these problems one day.
2296 :
2297 :			@item conditions under which control characters match a space delimiter:
2298 :			If @code{WORD} is called with the space character as a delimiter, all
2299 :			white-space characters (as identified by the C macro @code{isspace()})
2300 :			are delimiters. @code{PARSE}, on the other hand, treats space like other
2301 :			delimiters. @code{PARSE-WORD} treats space like @code{WORD}, but behaves
2302 :			like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
2303 :			interpreter (aka text interpreter) by default, treats all white-space
2304 :			characters as delimiters.
2305 :
2306 :			@item format of the control flow stack:
2307 :			The data stack is used as control flow stack. The size of a control flow
2308 :			stack item in cells is given by the constant @code{cs-item-size}. At the
2309 :			time of this writing, an item consists of a (pointer to a) locals list
2310 :			(third), an address in the code (second), and a tag for identifying the
2311 :			item (TOS). The following tags are used: @code{defstart},
2312 :			@code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
2313 :			@code{scopestart}.
2314 :
2315 :			@item conversion of digits > 35
2316 :			The characters @code{[\]^_'} are the digits with the decimal value
2317 :			36@minus{}41. There is no way to input many of the larger digits.
2318 :
2319 :			@item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
2320 :			The cursor is moved to the end of the entered string. If the input is
2321 :			terminated using the @kbd{Return} key, a space is typed.
2322 :
2323 :			@item exception abort sequence of @code{ABORT"}:
2324 :			The error string is stored into the variable @code{"error} and a
2325 :			@code{-2 throw} is performed.
2326 :
2327 :			@item input line terminator:
2328 :			For interactive input, @kbd{C-m} and @kbd{C-j} terminate lines. One of
2329 :			these characters is typically produced when you type the @kbd{Enter} or
2330 :			@kbd{Return} key.
2331 :
2332 :			@item maximum size of a counted string:
2333 :			@code{s" /counted-string" environment? drop .}. Currently 255 characters
2334 :			on all ports, but this may change.
2335 :
2336 :			@item maximum size of a parsed string:
2337 :			Given by the constant @code{/line}. Currently 255 characters.
2338 :
2339 :			@item maximum size of a definition name, in characters:
2340 :			31
2341 :
2342 :			@item maximum string length for @code{ENVIRONMENT?}, in characters:
2343 :			31
2344 :
2345 :			@item method of selecting the user input device:
2346 :	anton	1.17	The user input device is the standard input. There is currently no way to
2347 :			change it from within Gforth. However, the input can typically be
2348 :			redirected in the command line that starts Gforth.
2349 :	anton	1.14
2350 :			@item method of selecting the user output device:
2351 :			The user output device is the standard output. It cannot be redirected
2352 :	anton	1.17	from within Gforth, but typically from the command line that starts
2353 :			Gforth. Gforth uses buffered output, so output on a terminal does not
2354 :	anton	1.14	become visible before the next newline or buffer overflow. Output on
2355 :			non-terminals is invisible until the buffer overflows.
2356 :
2357 :			@item methods of dictionary compilation:
2358 :	anton	1.17	What are we expected to document here?
2359 :	anton	1.14
2360 :			@item number of bits in one address unit:
2361 :			@code{s" address-units-bits" environment? drop .}. 8 in all current
2362 :			ports.
2363 :
2364 :			@item number representation and arithmetic:
2365 :			Processor-dependent. Binary two's complement on all current ports.
2366 :
2367 :			@item ranges for integer types:
2368 :			Installation-dependent. Make environmental queries for @code{MAX-N},
2369 :			@code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
2370 :			unsigned (and positive) types is 0. The lower bound for signed types on
2371 :			two's complement and one's complement machines machines can be computed
2372 :			by adding 1 to the upper bound.
2373 :
2374 :			@item read-only data space regions:
2375 :			The whole Forth data space is writable.
2376 :
2377 :			@item size of buffer at @code{WORD}:
2378 :			@code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
2379 :			shared with the pictured numeric output string. If overwriting
2380 :			@code{PAD} is acceptable, it is as large as the remaining dictionary
2381 :			space, although only as much can be sensibly used as fits in a counted
2382 :			string.
2383 :
2384 :			@item size of one cell in address units:
2385 :			@code{1 cells .}.
2386 :
2387 :			@item size of one character in address units:
2388 :			@code{1 chars .}. 1 on all current ports.
2389 :
2390 :			@item size of the keyboard terminal buffer:
2391 :			Varies. You can determine the size at a specific time using @code{lp@
2392 :			tib - .}. It is shared with the locals stack and TIBs of files that
2393 :			include the current file. You can change the amount of space for TIBs
2394 :	anton	1.17	and locals stack at Gforth startup with the command line option
2395 :	anton	1.14	@code{-l}.
2396 :
2397 :			@item size of the pictured numeric output buffer:
2398 :			@code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
2399 :			shared with @code{WORD}.
2400 :
2401 :			@item size of the scratch area returned by @code{PAD}:
2402 :			The remainder of dictionary space. You can even use the unused part of
2403 :			the data stack space. The current size can be computed with @code{sp@
2404 :			pad - .}.
2405 :
2406 :			@item system case-sensitivity characteristics:
2407 :			Dictionary searches are case insensitive. However, as explained above
2408 :			under @i{character-set extensions}, the matching for non-ASCII
2409 :			characters is determined by the locale you are using. In the default
2410 :			@code{C} locale all non-ASCII characters are matched case-sensitively.
2411 :
2412 :			@item system prompt:
2413 :			@code{ ok} in interpret state, @code{ compiled} in compile state.
2414 :
2415 :			@item division rounding:
2416 :			installation dependent. @code{s" floored" environment? drop .}. We leave
2417 :	anton	1.25	the choice to @code{gcc} (what to use for @code{/}) and to you (whether to use
2418 :	anton	1.14	@code{fm/mod}, @code{sm/rem} or simply @code{/}).
2419 :
2420 :			@item values of @code{STATE} when true:
2421 :			-1.
2422 :
2423 :			@item values returned after arithmetic overflow:
2424 :			On two's complement machines, arithmetic is performed modulo
2425 :			2bits-per-cell for single arithmetic and 4bits-per-cell for double
2426 :			arithmetic (with appropriate mapping for signed types). Division by zero
2427 :			typically results in a @code{-55 throw} (floatingpoint unidentified
2428 :			fault), although a @code{-10 throw} (divide by zero) would be more
2429 :			appropriate.
2430 :
2431 :			@item whether the current definition can be found after @t{DOES>}:
2432 :			No.
2433 :
2434 :			@end table
2435 :
2436 :			@c ---------------------------------------------------------------------
2437 :			@node core-ambcond, core-other, core-idef, The Core Words
2438 :			@subsection Ambiguous conditions
2439 :			@c ---------------------------------------------------------------------
2440 :
2441 :			@table @i
2442 :
2443 :			@item a name is neither a word nor a number:
2444 :			@code{-13 throw} (Undefined word)
2445 :
2446 :			@item a definition name exceeds the maximum length allowed:
2447 :			@code{-19 throw} (Word name too long)
2448 :
2449 :			@item addressing a region not inside the various data spaces of the forth system:
2450 :			The stacks, code space and name space are accessible. Machine code space is
2451 :			typically readable. Accessing other addresses gives results dependent on
2452 :			the operating system. On decent systems: @code{-9 throw} (Invalid memory
2453 :			address).
2454 :
2455 :			@item argument type incompatible with parameter:
2456 :			This is usually not caught. Some words perform checks, e.g., the control
2457 :			flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
2458 :			mismatch).
2459 :
2460 :			@item attempting to obtain the execution token of a word with undefined execution semantics:
2461 :			You get an execution token representing the compilation semantics
2462 :			instead.
2463 :
2464 :			@item dividing by zero:
2465 :			typically results in a @code{-55 throw} (floating point unidentified
2466 :			fault), although a @code{-10 throw} (divide by zero) would be more
2467 :			appropriate.
2468 :
2469 :			@item insufficient data stack or return stack space:
2470 :			Not checked. This typically results in mysterious illegal memory
2471 :			accesses, producing @code{-9 throw} (Invalid memory address) or
2472 :			@code{-23 throw} (Address alignment exception).
2473 :
2474 :			@item insufficient space for loop control parameters:
2475 :			like other return stack overflows.
2476 :
2477 :			@item insufficient space in the dictionary:
2478 :			Not checked. Similar results as stack overflows. However, typically the
2479 :			error appears at a different place when one inserts or removes code.
2480 :
2481 :			@item interpreting a word with undefined interpretation semantics:
2482 :			For some words, we defined interpretation semantics. For the others:
2483 :			@code{-14 throw} (Interpreting a compile-only word). Note that this is
2484 :			checked only by the outer (aka text) interpreter; if the word is
2485 :			@code{execute}d in some other way, it will typically perform it's
2486 :			compilation semantics even in interpret state. (We could change @code{'}
2487 :			and relatives not to give the xt of such words, but we think that would
2488 :			be too restrictive).
2489 :
2490 :			@item modifying the contents of the input buffer or a string literal:
2491 :			These are located in writable memory and can be modified.
2492 :
2493 :			@item overflow of the pictured numeric output string:
2494 :			Not checked.
2495 :
2496 :			@item parsed string overflow:
2497 :			@code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
2498 :
2499 :			@item producing a result out of range:
2500 :			On two's complement machines, arithmetic is performed modulo
2501 :			2bits-per-cell for single arithmetic and 4bits-per-cell for double
2502 :			arithmetic (with appropriate mapping for signed types). Division by zero
2503 :			typically results in a @code{-55 throw} (floatingpoint unidentified
2504 :			fault), although a @code{-10 throw} (divide by zero) would be more
2505 :			appropriate. @code{convert} and @code{>number} currently overflow
2506 :			silently.
2507 :
2508 :			@item reading from an empty data or return stack:
2509 :			The data stack is checked by the outer (aka text) interpreter after
2510 :			every word executed. If it has underflowed, a @code{-4 throw} (Stack
2511 :			underflow) is performed. Apart from that, the stacks are not checked and
2512 :			underflows can result in similar behaviour as overflows (of adjacent
2513 :			stacks).
2514 :
2515 :			@item unexepected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
2516 :			@code{Create} and its descendants perform a @code{-16 throw} (Attempt to
2517 :			use zero-length string as a name). Words like @code{'} probably will not
2518 :			find what they search. Note that it is possible to create zero-length
2519 :			names with @code{nextname} (should it not?).
2520 :
2521 :			@item @code{>IN} greater than input buffer:
2522 :			The next invocation of a parsing word returns a string wih length 0.
2523 :
2524 :			@item @code{RECURSE} appears after @code{DOES>}:
2525 :			Compiles a recursive call to the defining word not to the defined word.
2526 :
2527 :			@item argument input source different than current input source for @code{RESTORE-INPUT}:
2528 :	anton	1.27	@code{-12 THROW}. Note that, once an input file is closed (e.g., because
2529 :			the end of the file was reached), its source-id may be
2530 :			reused. Therefore, restoring an input source specification referencing a
2531 :			closed file may lead to unpredictable results instead of a @code{-12
2532 :			THROW}.
2533 :
2534 :			In the future, Gforth may be able to retore input source specifications
2535 :			from other than the current input soruce.
2536 :	anton	1.14
2537 :			@item data space containing definitions gets de-allocated:
2538 :			Deallocation with @code{allot} is not checked. This typically resuls in
2539 :			memory access faults or execution of illegal instructions.
2540 :
2541 :			@item data space read/write with incorrect alignment:
2542 :			Processor-dependent. Typically results in a @code{-23 throw} (Address
2543 :			alignment exception). Under Linux on a 486 or later processor with
2544 :			alignment turned on, incorrect alignment results in a @code{-9 throw}
2545 :			(Invalid memory address). There are reportedly some processors with
2546 :			alignment restrictions that do not report them.
2547 :
2548 :			@item data space pointer not properly aligned, @code{,}, @code{C,}:
2549 :			Like other alignment errors.
2550 :
2551 :			@item less than u+2 stack items (@code{PICK} and @code{ROLL}):
2552 :			Not checked. May cause an illegal memory access.
2553 :
2554 :			@item loop control parameters not available:
2555 :			Not checked. The counted loop words simply assume that the top of return
2556 :			stack items are loop control parameters and behave accordingly.
2557 :
2558 :			@item most recent definition does not have a name (@code{IMMEDIATE}):
2559 :			@code{abort" last word was headerless"}.
2560 :
2561 :			@item name not defined by @code{VALUE} used by @code{TO}:
2562 :			@code{-32 throw} (Invalid name argument)
2563 :
2564 :	anton	1.15	@item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
2565 :	anton	1.14	@code{-13 throw} (Undefined word)
2566 :
2567 :			@item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
2568 :			Gforth behaves as if they were of the same type. I.e., you can predict
2569 :			the behaviour by interpreting all parameters as, e.g., signed.
2570 :
2571 :			@item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
2572 :			Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} is equivalent to
2573 :			@code{TO}.
2574 :
2575 :			@item String longer than a counted string returned by @code{WORD}:
2576 :			Not checked. The string will be ok, but the count will, of course,
2577 :			contain only the least significant bits of the length.
2578 :
2579 :	anton	1.15	@item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
2580 :	anton	1.14	Processor-dependent. Typical behaviours are returning 0 and using only
2581 :			the low bits of the shift count.
2582 :
2583 :			@item word not defined via @code{CREATE}:
2584 :			@code{>BODY} produces the PFA of the word no matter how it was defined.
2585 :
2586 :			@code{DOES>} changes the execution semantics of the last defined word no
2587 :			matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
2588 :			@code{CREATE , DOES>}.
2589 :
2590 :			@item words improperly used outside @code{<#} and @code{#>}:
2591 :			Not checked. As usual, you can expect memory faults.
2592 :
2593 :			@end table
2594 :
2595 :
2596 :			@c ---------------------------------------------------------------------
2597 :			@node core-other, , core-ambcond, The Core Words
2598 :			@subsection Other system documentation
2599 :			@c ---------------------------------------------------------------------
2600 :
2601 :			@table @i
2602 :
2603 :			@item nonstandard words using @code{PAD}:
2604 :			None.
2605 :
2606 :			@item operator's terminal facilities available:
2607 :	anton	1.26	After processing the command line, Gforth goes into interactive mode,
2608 :			and you can give commands to Gforth interactively. The actual facilities
2609 :			available depend on how you invoke Gforth.
2610 :	anton	1.14
2611 :			@item program data space available:
2612 :			@code{sp@ here - .} gives the space remaining for dictionary and data
2613 :			stack together.
2614 :
2615 :			@item return stack space available:
2616 :	anton	1.26	By default 16 KBytes. The default can be overridden with the @code{-r}
2617 :			switch (@pxref{Invocation}) when Gforth starts up.
2618 :	anton	1.14
2619 :			@item stack space available:
2620 :			@code{sp@ here - .} gives the space remaining for dictionary and data
2621 :			stack together.
2622 :
2623 :			@item system dictionary space required, in address units:
2624 :			Type @code{here forthstart - .} after startup. At the time of this
2625 :			writing, this gives 70108 (bytes) on a 32-bit system.
2626 :			@end table
2627 :
2628 :
2629 :			@c =====================================================================
2630 :			@node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
2631 :			@section The optional Block word set
2632 :			@c =====================================================================
2633 :
2634 :			@menu
2635 :	anton	1.15	* block-idef:: Implementation Defined Options
2636 :			* block-ambcond:: Ambiguous Conditions
2637 :			* block-other:: Other System Documentation
2638 :	anton	1.14	@end menu
2639 :
2640 :
2641 :			@c ---------------------------------------------------------------------
2642 :			@node block-idef, block-ambcond, The optional Block word set, The optional Block word set
2643 :			@subsection Implementation Defined Options
2644 :			@c ---------------------------------------------------------------------
2645 :
2646 :			@table @i
2647 :
2648 :			@item the format for display by @code{LIST}:
2649 :			First the screen number is displayed, then 16 lines of 64 characters,
2650 :			each line preceded by the line number.
2651 :
2652 :			@item the length of a line affected by @code{\}:
2653 :			64 characters.
2654 :			@end table
2655 :
2656 :
2657 :			@c ---------------------------------------------------------------------
2658 :			@node block-ambcond, block-other, block-idef, The optional Block word set
2659 :			@subsection Ambiguous conditions
2660 :			@c ---------------------------------------------------------------------
2661 :
2662 :			@table @i
2663 :
2664 :			@item correct block read was not possible:
2665 :			Typically results in a @code{throw} of some OS-derived value (between
2666 :			-512 and -2048). If the blocks file was just not long enough, blanks are
2667 :			supplied for the missing portion.
2668 :
2669 :			@item I/O exception in block transfer:
2670 :			Typically results in a @code{throw} of some OS-derived value (between
2671 :			-512 and -2048).
2672 :
2673 :			@item invalid block number:
2674 :			@code{-35 throw} (Invalid block number)
2675 :
2676 :			@item a program directly alters the contents of @code{BLK}:
2677 :			The input stream is switched to that other block, at the same
2678 :			position. If the storing to @code{BLK} happens when interpreting
2679 :			non-block input, the system will get quite confused when the block ends.
2680 :
2681 :			@item no current block buffer for @code{UPDATE}:
2682 :			@code{UPDATE} has no effect.
2683 :
2684 :			@end table
2685 :
2686 :
2687 :			@c ---------------------------------------------------------------------
2688 :			@node block-other, , block-ambcond, The optional Block word set
2689 :			@subsection Other system documentation
2690 :			@c ---------------------------------------------------------------------
2691 :
2692 :			@table @i
2693 :
2694 :			@item any restrictions a multiprogramming system places on the use of buffer addresses:
2695 :			No restrictions (yet).
2696 :
2697 :			@item the number of blocks available for source and data:
2698 :			depends on your disk space.
2699 :
2700 :			@end table
2701 :
2702 :
2703 :			@c =====================================================================
2704 :			@node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
2705 :			@section The optional Double Number word set
2706 :			@c =====================================================================
2707 :
2708 :			@menu
2709 :	anton	1.15	* double-ambcond:: Ambiguous Conditions
2710 :	anton	1.14	@end menu
2711 :
2712 :
2713 :			@c ---------------------------------------------------------------------
2714 :	anton	1.15	@node double-ambcond, , The optional Double Number word set, The optional Double Number word set
2715 :	anton	1.14	@subsection Ambiguous conditions
2716 :			@c ---------------------------------------------------------------------
2717 :
2718 :			@table @i
2719 :
2720 :	anton	1.15	@item @var{d} outside of range of @var{n} in @code{D>S}:
2721 :	anton	1.14	The least significant cell of @var{d} is produced.
2722 :
2723 :			@end table
2724 :
2725 :
2726 :			@c =====================================================================
2727 :			@node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
2728 :			@section The optional Exception word set
2729 :			@c =====================================================================
2730 :
2731 :			@menu
2732 :	anton	1.15	* exception-idef:: Implementation Defined Options
2733 :	anton	1.14	@end menu
2734 :
2735 :
2736 :			@c ---------------------------------------------------------------------
2737 :	anton	1.15	@node exception-idef, , The optional Exception word set, The optional Exception word set
2738 :	anton	1.14	@subsection Implementation Defined Options
2739 :			@c ---------------------------------------------------------------------
2740 :
2741 :			@table @i
2742 :			@item @code{THROW}-codes used in the system:
2743 :			The codes -256@minus{}-511 are used for reporting signals (see
2744 :			@file{errore.fs}). The codes -512@minus{}-2047 are used for OS errors
2745 :			(for file and memory allocation operations). The mapping from OS error
2746 :			numbers to throw code is -512@minus{}@var{errno}. One side effect of
2747 :			this mapping is that undefined OS errors produce a message with a
2748 :			strange number; e.g., @code{-1000 THROW} results in @code{Unknown error
2749 :			488} on my system.
2750 :			@end table
2751 :
2752 :			@c =====================================================================
2753 :			@node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
2754 :			@section The optional Facility word set
2755 :			@c =====================================================================
2756 :
2757 :			@menu
2758 :	anton	1.15	* facility-idef:: Implementation Defined Options
2759 :			* facility-ambcond:: Ambiguous Conditions
2760 :	anton	1.14	@end menu
2761 :
2762 :
2763 :			@c ---------------------------------------------------------------------
2764 :			@node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
2765 :			@subsection Implementation Defined Options
2766 :			@c ---------------------------------------------------------------------
2767 :
2768 :			@table @i
2769 :
2770 :			@item encoding of keyboard events (@code{EKEY}):
2771 :			Not yet implemeted.
2772 :
2773 :			@item duration of a system clock tick
2774 :			System dependent. With respect to @code{MS}, the time is specified in
2775 :			microseconds. How well the OS and the hardware implement this, is
2776 :			another question.
2777 :
2778 :			@item repeatability to be expected from the execution of @code{MS}:
2779 :			System dependent. On Unix, a lot depends on load. If the system is
2780 :	anton	1.17	lightly loaded, and the delay is short enough that Gforth does not get
2781 :	anton	1.14	swapped out, the performance should be acceptable. Under MS-DOS and
2782 :			other single-tasking systems, it should be good.
2783 :
2784 :			@end table
2785 :
2786 :
2787 :			@c ---------------------------------------------------------------------
2788 :	anton	1.15	@node facility-ambcond, , facility-idef, The optional Facility word set
2789 :	anton	1.14	@subsection Ambiguous conditions
2790 :			@c ---------------------------------------------------------------------
2791 :
2792 :			@table @i
2793 :
2794 :			@item @code{AT-XY} can't be performed on user output device:
2795 :			Largely terminal dependant. No range checks are done on the arguments.
2796 :			No errors are reported. You may see some garbage appearing, you may see
2797 :			simply nothing happen.
2798 :
2799 :			@end table
2800 :
2801 :
2802 :			@c =====================================================================
2803 :			@node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
2804 :			@section The optional File-Access word set
2805 :			@c =====================================================================
2806 :
2807 :			@menu
2808 :	anton	1.15	* file-idef:: Implementation Defined Options
2809 :			* file-ambcond:: Ambiguous Conditions
2810 :	anton	1.14	@end menu
2811 :
2812 :
2813 :			@c ---------------------------------------------------------------------
2814 :			@node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
2815 :			@subsection Implementation Defined Options
2816 :			@c ---------------------------------------------------------------------
2817 :
2818 :			@table @i
2819 :
2820 :			@item File access methods used:
2821 :			@code{R/O}, @code{R/W} and @code{BIN} work as you would
2822 :			expect. @code{W/O} translates into the C file opening mode @code{w} (or
2823 :			@code{wb}): The file is cleared, if it exists, and created, if it does
2824 :	anton	1.15	not (both with @code{open-file} and @code{create-file}). Under Unix
2825 :	anton	1.14	@code{create-file} creates a file with 666 permissions modified by your
2826 :			umask.
2827 :
2828 :			@item file exceptions:
2829 :			The file words do not raise exceptions (except, perhaps, memory access
2830 :			faults when you pass illegal addresses or file-ids).
2831 :
2832 :			@item file line terminator:
2833 :			System-dependent. Gforth uses C's newline character as line
2834 :			terminator. What the actual character code(s) of this are is
2835 :			system-dependent.
2836 :
2837 :			@item file name format
2838 :			System dependent. Gforth just uses the file name format of your OS.
2839 :
2840 :			@item information returned by @code{FILE-STATUS}:
2841 :			@code{FILE-STATUS} returns the most powerful file access mode allowed
2842 :			for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
2843 :			cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
2844 :			along with the retured mode.
2845 :
2846 :			@item input file state after an exception when including source:
2847 :			All files that are left via the exception are closed.
2848 :
2849 :			@item @var{ior} values and meaning:
2850 :	anton	1.15	The @var{ior}s returned by the file and memory allocation words are
2851 :			intended as throw codes. They typically are in the range
2852 :			-512@minus{}-2047 of OS errors. The mapping from OS error numbers to
2853 :			@var{ior}s is -512@minus{}@var{errno}.
2854 :	anton	1.14
2855 :			@item maximum depth of file input nesting:
2856 :			limited by the amount of return stack, locals/TIB stack, and the number
2857 :			of open files available. This should not give you troubles.
2858 :
2859 :			@item maximum size of input line:
2860 :			@code{/line}. Currently 255.
2861 :
2862 :			@item methods of mapping block ranges to files:
2863 :			Currently, the block words automatically access the file
2864 :			@file{blocks.fb} in the currend working directory. More sophisticated
2865 :			methods could be implemented if there is demand (and a volunteer).
2866 :
2867 :			@item number of string buffers provided by @code{S"}:
2868 :			1
2869 :
2870 :			@item size of string buffer used by @code{S"}:
2871 :			@code{/line}. currently 255.
2872 :
2873 :			@end table
2874 :
2875 :			@c ---------------------------------------------------------------------
2876 :	anton	1.15	@node file-ambcond, , file-idef, The optional File-Access word set
2877 :	anton	1.14	@subsection Ambiguous conditions
2878 :			@c ---------------------------------------------------------------------
2879 :
2880 :			@table @i
2881 :
2882 :			@item attempting to position a file outside it's boundaries:
2883 :			@code{REPOSITION-FILE} is performed as usual: Afterwards,
2884 :			@code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
2885 :
2886 :			@item attempting to read from file positions not yet written:
2887 :			End-of-file, i.e., zero characters are read and no error is reported.
2888 :
2889 :			@item @var{file-id} is invalid (@code{INCLUDE-FILE}):
2890 :			An appropriate exception may be thrown, but a memory fault or other
2891 :			problem is more probable.
2892 :
2893 :			@item I/O exception reading or closing @var{file-id} (@code{include-file}, @code{included}):
2894 :			The @var{ior} produced by the operation, that discovered the problem, is
2895 :			thrown.
2896 :
2897 :			@item named file cannot be opened (@code{included}):
2898 :			The @var{ior} produced by @code{open-file} is thrown.
2899 :
2900 :			@item requesting an unmapped block number:
2901 :			There are no unmapped legal block numbers. On some operating systems,
2902 :			writing a block with a large number may overflow the file system and
2903 :			have an error message as consequence.
2904 :
2905 :			@item using @code{source-id} when @code{blk} is non-zero:
2906 :			@code{source-id} performs its function. Typically it will give the id of
2907 :			the source which loaded the block. (Better ideas?)
2908 :
2909 :			@end table
2910 :
2911 :
2912 :			@c =====================================================================
2913 :			@node The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
2914 :	anton	1.15	@section The optional Floating-Point word set
2915 :	anton	1.14	@c =====================================================================
2916 :
2917 :			@menu
2918 :	anton	1.15	* floating-idef:: Implementation Defined Options
2919 :			* floating-ambcond:: Ambiguous Conditions
2920 :	anton	1.14	@end menu
2921 :
2922 :
2923 :			@c ---------------------------------------------------------------------
2924 :			@node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
2925 :			@subsection Implementation Defined Options
2926 :			@c ---------------------------------------------------------------------
2927 :
2928 :			@table @i
2929 :
2930 :	anton	1.15	@item format and range of floating point numbers:
2931 :			System-dependent; the @code{double} type of C.
2932 :	anton	1.14
2933 :	anton	1.15	@item results of @code{REPRESENT} when @var{float} is out of range:
2934 :			System dependent; @code{REPRESENT} is implemented using the C library
2935 :			function @code{ecvt()} and inherits its behaviour in this respect.
2936 :	anton	1.14
2937 :	anton	1.15	@item rounding or truncation of floating-point numbers:
2938 :	anton	1.26	System dependent; the rounding behaviour is inherited from the hosting C
2939 :			compiler. IEEE-FP-based (i.e., most) systems by default round to
2940 :			nearest, and break ties by rounding to even (i.e., such that the last
2941 :			bit of the mantissa is 0).
2942 :	anton	1.14
2943 :	anton	1.15	@item size of floating-point stack:
2944 :			@code{s" FLOATING-STACK" environment? drop .}. Can be changed at startup
2945 :			with the command-line option @code{-f}.
2946 :	anton	1.14
2947 :	anton	1.15	@item width of floating-point stack:
2948 :			@code{1 floats}.
2949 :	anton	1.14
2950 :			@end table
2951 :
2952 :
2953 :			@c ---------------------------------------------------------------------
2954 :	anton	1.15	@node floating-ambcond, , floating-idef, The optional Floating-Point word set
2955 :			@subsection Ambiguous conditions
2956 :	anton	1.14	@c ---------------------------------------------------------------------
2957 :
2958 :			@table @i
2959 :
2960 :	anton	1.15	@item @code{df@@} or @code{df!} used with an address that is not double-float aligned:
2961 :			System-dependent. Typically results in an alignment fault like other
2962 :			alignment violations.
2963 :	anton	1.14
2964 :	anton	1.15	@item @code{f@@} or @code{f!} used with an address that is not float aligned:
2965 :			System-dependent. Typically results in an alignment fault like other
2966 :			alignment violations.
2967 :	anton	1.14
2968 :	anton	1.15	@item Floating-point result out of range:
2969 :			System-dependent. Can result in a @code{-55 THROW} (Floating-point
2970 :			unidentified fault), or can produce a special value representing, e.g.,
2971 :			Infinity.
2972 :	anton	1.14
2973 :	anton	1.15	@item @code{sf@@} or @code{sf!} used with an address that is not single-float aligned:
2974 :			System-dependent. Typically results in an alignment fault like other
2975 :			alignment violations.
2976 :	anton	1.14
2977 :	anton	1.15	@item BASE is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
2978 :			The floating-point number is converted into decimal nonetheless.
2979 :	anton	1.14
2980 :	anton	1.15	@item Both arguments are equal to zero (@code{FATAN2}):
2981 :			System-dependent. @code{FATAN2} is implemented using the C library
2982 :			function @code{atan2()}.
2983 :	anton	1.14
2984 :	anton	1.15	@item Using ftan on an argument @var{r1} where cos(@var{r1}) is zero:
2985 :			System-dependent. Anyway, typically the cos of @var{r1} will not be zero
2986 :			because of small errors and the tan will be a very large (or very small)
2987 :			but finite number.
2988 :	anton	1.14
2989 :	anton	1.15	@item @var{d} cannot be presented precisely as a float in @code{D>F}:
2990 :			The result is rounded to the nearest float.
2991 :	anton	1.14
2992 :	anton	1.15	@item dividing by zero:
2993 :			@code{-55 throw} (Floating-point unidentified fault)
2994 :	anton	1.14
2995 :	anton	1.15	@item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
2996 :			System dependent. On IEEE-FP based systems the number is converted into
2997 :			an infinity.
2998 :	anton	1.14
2999 :	anton	1.15	@item @var{float}<1 (@code{facosh}):
3000 :			@code{-55 throw} (Floating-point unidentified fault)
3001 :	anton	1.14
3002 :	anton	1.15	@item @var{float}=<-1 (@code{flnp1}):
3003 :			@code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
3004 :			negative infinity is typically produced for @var{float}=-1.
3005 :	anton	1.14
3006 :	anton	1.15	@item @var{float}=<0 (@code{fln}, @code{flog}):
3007 :			@code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
3008 :			negative infinity is typically produced for @var{float}=0.
3009 :	anton	1.14
3010 :	anton	1.15	@item @var{float}<0 (@code{fasinh}, @code{fsqrt}):
3011 :			@code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
3012 :			produces values for these inputs on my Linux box (Bug in the C library?)
3013 :	anton	1.14
3014 :	anton	1.15	@item \|@var{float}\|>1 (@code{facos}, @code{fasin}, @code{fatanh}):
3015 :			@code{-55 throw} (Floating-point unidentified fault).
3016 :	anton	1.14
3017 :	anton	1.15	@item integer part of float cannot be represented by @var{d} in @code{f>d}:
3018 :			@code{-55 throw} (Floating-point unidentified fault).
3019 :	anton	1.14
3020 :	anton	1.15	@item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
3021 :			This does not happen.
3022 :			@end table
3023 :	anton	1.14
3024 :
3025 :
3026 :			@c =====================================================================
3027 :	anton	1.15	@node The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
3028 :			@section The optional Locals word set
3029 :	anton	1.14	@c =====================================================================
3030 :
3031 :			@menu
3032 :	anton	1.15	* locals-idef:: Implementation Defined Options
3033 :			* locals-ambcond:: Ambiguous Conditions
3034 :	anton	1.14	@end menu
3035 :
3036 :
3037 :			@c ---------------------------------------------------------------------
3038 :	anton	1.15	@node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
3039 :	anton	1.14	@subsection Implementation Defined Options
3040 :			@c ---------------------------------------------------------------------
3041 :
3042 :			@table @i
3043 :
3044 :	anton	1.15	@item maximum number of locals in a definition:
3045 :			@code{s" #locals" environment? drop .}. Currently 15. This is a lower
3046 :			bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
3047 :			characters. The number of locals in a definition is bounded by the size
3048 :			of locals-buffer, which contains the names of the locals.
3049 :	anton	1.14
3050 :			@end table
3051 :
3052 :
3053 :			@c ---------------------------------------------------------------------
3054 :	anton	1.15	@node locals-ambcond, , locals-idef, The optional Locals word set
3055 :	anton	1.14	@subsection Ambiguous conditions
3056 :			@c ---------------------------------------------------------------------
3057 :
3058 :			@table @i
3059 :
3060 :	anton	1.15	@item executing a named local in interpretation state:
3061 :			@code{-14 throw} (Interpreting a compile-only word).
3062 :	anton	1.14
3063 :	anton	1.15	@item @var{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
3064 :			@code{-32 throw} (Invalid name argument)
3065 :	anton	1.14
3066 :			@end table
3067 :
3068 :
3069 :			@c =====================================================================
3070 :	anton	1.15	@node The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
3071 :			@section The optional Memory-Allocation word set
3072 :	anton	1.14	@c =====================================================================
3073 :
3074 :			@menu
3075 :	anton	1.15	* memory-idef:: Implementation Defined Options
3076 :	anton	1.14	@end menu
3077 :
3078 :
3079 :			@c ---------------------------------------------------------------------
3080 :	anton	1.15	@node memory-idef, , The optional Memory-Allocation word set, The optional Memory-Allocation word set
3081 :	anton	1.14	@subsection Implementation Defined Options
3082 :			@c ---------------------------------------------------------------------
3083 :
3084 :			@table @i
3085 :
3086 :	anton	1.15	@item values and meaning of @var{ior}:
3087 :			The @var{ior}s returned by the file and memory allocation words are
3088 :			intended as throw codes. They typically are in the range
3089 :			-512@minus{}-2047 of OS errors. The mapping from OS error numbers to
3090 :			@var{ior}s is -512@minus{}@var{errno}.
3091 :	anton	1.14
3092 :			@end table
3093 :
3094 :			@c =====================================================================
3095 :	anton	1.15	@node The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
3096 :			@section The optional Programming-Tools word set
3097 :	anton	1.14	@c =====================================================================
3098 :
3099 :			@menu
3100 :	anton	1.15	* programming-idef:: Implementation Defined Options
3101 :			* programming-ambcond:: Ambiguous Conditions
3102 :	anton	1.14	@end menu
3103 :
3104 :
3105 :			@c ---------------------------------------------------------------------
3106 :	anton	1.15	@node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
3107 :	anton	1.14	@subsection Implementation Defined Options
3108 :			@c ---------------------------------------------------------------------
3109 :
3110 :			@table @i
3111 :
3112 :	anton	1.15	@item ending sequence for input following @code{;code} and @code{code}:
3113 :			Not implemented (yet).
3114 :	anton	1.14
3115 :	anton	1.15	@item manner of processing input following @code{;code} and @code{code}:
3116 :			Not implemented (yet).
3117 :
3118 :			@item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
3119 :			Not implemented (yet). If they were implemented, they would use the
3120 :			search order wordset.
3121 :
3122 :			@item source and format of display by @code{SEE}:
3123 :			The source for @code{see} is the intermediate code used by the inner
3124 :			interpreter. The current @code{see} tries to output Forth source code
3125 :			as well as possible.
3126 :
3127 :	anton	1.14	@end table
3128 :
3129 :			@c ---------------------------------------------------------------------
3130 :	anton	1.15	@node programming-ambcond, , programming-idef, The optional Programming-Tools word set
3131 :	anton	1.14	@subsection Ambiguous conditions
3132 :			@c ---------------------------------------------------------------------
3133 :
3134 :			@table @i
3135 :
3136 :	anton	1.15	@item deleting the compilation wordlist (@code{FORGET}):
3137 :			Not implemented (yet).
3138 :	anton	1.14
3139 :	anton	1.15	@item fewer than @var{u}+1 items on the control flow stack (@code{CS-PICK}, @code{CS-ROLL}):
3140 :			This typically results in an @code{abort"} with a descriptive error
3141 :			message (may change into a @code{-22 throw} (Control structure mismatch)
3142 :			in the future). You may also get a memory access error. If you are
3143 :			unlucky, this ambiguous condition is not caught.
3144 :
3145 :			@item @var{name} can't be found (@code{forget}):
3146 :			Not implemented (yet).
3147 :	anton	1.14
3148 :	anton	1.15	@item @var{name} not defined via @code{CREATE}:
3149 :			@code{;code} is not implemented (yet). If it were, it would behave like
3150 :			@code{DOES>} in this respect, i.e., change the execution semantics of
3151 :			the last defined word no matter how it was defined.
3152 :	anton	1.14
3153 :	anton	1.15	@item @code{POSTPONE} applied to @code{[IF]}:
3154 :			After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
3155 :			equivalent to @code{[IF]}.
3156 :	anton	1.14
3157 :	anton	1.15	@item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
3158 :			Continue in the same state of conditional compilation in the next outer
3159 :			input source. Currently there is no warning to the user about this.
3160 :	anton	1.14
3161 :	anton	1.15	@item removing a needed definition (@code{FORGET}):
3162 :			Not implemented (yet).
3163 :	anton	1.14
3164 :			@end table
3165 :
3166 :
3167 :			@c =====================================================================
3168 :	anton	1.15	@node The optional Search-Order word set, , The optional Programming-Tools word set, ANS conformance
3169 :			@section The optional Search-Order word set
3170 :	anton	1.14	@c =====================================================================
3171 :
3172 :			@menu
3173 :	anton	1.15	* search-idef:: Implementation Defined Options
3174 :			* search-ambcond:: Ambiguous Conditions
3175 :	anton	1.14	@end menu
3176 :
3177 :
3178 :			@c ---------------------------------------------------------------------
3179 :	anton	1.15	@node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
3180 :	anton	1.14	@subsection Implementation Defined Options
3181 :			@c ---------------------------------------------------------------------
3182 :
3183 :			@table @i
3184 :
3185 :	anton	1.15	@item maximum number of word lists in search order:
3186 :			@code{s" wordlists" environment? drop .}. Currently 16.
3187 :
3188 :			@item minimum search order:
3189 :			@code{root root}.
3190 :	anton	1.14
3191 :			@end table
3192 :
3193 :			@c ---------------------------------------------------------------------
3194 :	anton	1.15	@node search-ambcond, , search-idef, The optional Search-Order word set
3195 :	anton	1.14	@subsection Ambiguous conditions
3196 :			@c ---------------------------------------------------------------------
3197 :
3198 :			@table @i
3199 :
3200 :	anton	1.15	@item changing the compilation wordlist (during compilation):
3201 :			The definition is put into the wordlist that is the compilation wordlist
3202 :			when @code{REVEAL} is executed (by @code{;}, @code{DOES>},
3203 :			@code{RECURSIVE}, etc.).
3204 :	anton	1.14
3205 :	anton	1.15	@item search order empty (@code{previous}):
3206 :			@code{abort" Vocstack empty"}.
3207 :	anton	1.14
3208 :	anton	1.15	@item too many word lists in search order (@code{also}):
3209 :			@code{abort" Vocstack full"}.
3210 :	anton	1.14
3211 :			@end table
3212 :	anton	1.13
3213 :
3214 :	anton	1.17	@node Model, Emacs and Gforth, ANS conformance, Top
3215 :	anton	1.4	@chapter Model
3216 :
3217 :	anton	1.17	@node Emacs and Gforth, Internals, Model, Top
3218 :			@chapter Emacs and Gforth
3219 :	anton	1.4
3220 :	anton	1.17	Gforth comes with @file{gforth.el}, an improved version of
3221 :	anton	1.4	@file{forth.el} by Goran Rydqvist (icluded in the TILE package). The
3222 :			improvements are a better (but still not perfect) handling of
3223 :			indentation. I have also added comment paragraph filling (@kbd{M-q}),
3224 :	anton	1.8	commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) regions and
3225 :			removing debugging tracers (@kbd{C-x ~}, @pxref{Debugging}). I left the
3226 :			stuff I do not use alone, even though some of it only makes sense for
3227 :			TILE. To get a description of these features, enter Forth mode and type
3228 :			@kbd{C-h m}.
3229 :	anton	1.4
3230 :	anton	1.17	In addition, Gforth supports Emacs quite well: The source code locations
3231 :	anton	1.4	given in error messages, debugging output (from @code{~~}) and failed
3232 :			assertion messages are in the right format for Emacs' compilation mode
3233 :			(@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
3234 :			Manual}) so the source location corresponding to an error or other
3235 :			message is only a few keystrokes away (@kbd{C-x `} for the next error,
3236 :			@kbd{C-c C-c} for the error under the cursor).
3237 :
3238 :			Also, if you @code{include} @file{etags.fs}, a new @file{TAGS} file
3239 :			(@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) will be produced that
3240 :			contains the definitions of all words defined afterwards. You can then
3241 :			find the source for a word using @kbd{M-.}. Note that emacs can use
3242 :	anton	1.17	several tags files at the same time (e.g., one for the Gforth sources
3243 :	anton	1.28	and one for your program, @pxref{Select Tags Table,,Selecting a Tags
3244 :			Table,emacs, Emacs Manual}). The TAGS file for the preloaded words is
3245 :			@file{$(datadir)/gforth/$(VERSION)/TAGS} (e.g.,
3246 :			@file{/usr/local/share/gforth/0.2/TAGS}).
3247 :	anton	1.4
3248 :			To get all these benefits, add the following lines to your @file{.emacs}
3249 :			file:
3250 :
3251 :			@example
3252 :			(autoload 'forth-mode "gforth.el")
3253 :			(setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
3254 :			@end example
3255 :
3256 :	anton	1.17	@node Internals, Bugs, Emacs and Gforth, Top
3257 :	anton	1.3	@chapter Internals
3258 :
3259 :	anton	1.17	Reading this section is not necessary for programming with Gforth. It
3260 :			should be helpful for finding your way in the Gforth sources.
3261 :	anton	1.3
3262 :	anton	1.24	The ideas in this section have also been published in the papers
3263 :			@cite{ANS fig/GNU/??? Forth} (in German) by Bernd Paysan, presented at
3264 :			the Forth-Tagung '93 and @cite{A Portable Forth Engine} by M. Anton
3265 :			Ertl, presented at EuroForth '93; the latter is available at
3266 :			@*@file{http://www.complang.tuwien.ac.at/papers/ertl93.ps.Z}.
3267 :
3268 :	anton	1.4	@menu
3269 :			* Portability::
3270 :			* Threading::
3271 :			* Primitives::
3272 :			* System Architecture::
3273 :	anton	1.17	* Performance::
3274 :	anton	1.4	@end menu
3275 :
3276 :			@node Portability, Threading, Internals, Internals
3277 :	anton	1.3	@section Portability
3278 :
3279 :			One of the main goals of the effort is availability across a wide range
3280 :			of personal machines. fig-Forth, and, to a lesser extent, F83, achieved
3281 :			this goal by manually coding the engine in assembly language for several
3282 :			then-popular processors. This approach is very labor-intensive and the
3283 :			results are short-lived due to progress in computer architecture.
3284 :
3285 :			Others have avoided this problem by coding in C, e.g., Mitch Bradley
3286 :			(cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
3287 :			particularly popular for UNIX-based Forths due to the large variety of
3288 :			architectures of UNIX machines. Unfortunately an implementation in C
3289 :			does not mix well with the goals of efficiency and with using
3290 :			traditional techniques: Indirect or direct threading cannot be expressed
3291 :			in C, and switch threading, the fastest technique available in C, is
3292 :			significantly slower. Another problem with C is that it's very
3293 :			cumbersome to express double integer arithmetic.
3294 :
3295 :			Fortunately, there is a portable language that does not have these
3296 :			limitations: GNU C, the version of C processed by the GNU C compiler
3297 :			(@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
3298 :			GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
3299 :			Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
3300 :			threading possible, its @code{long long} type (@pxref{Long Long, ,
3301 :			Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forths
3302 :	anton	1.32	double numbers@footnote{Unfortunately, long longs are not implemented
3303 :			properly on all machines (e.g., on alpha-osf1, long longs are only 64
3304 :			bits, the same size as longs (and pointers), but they should be twice as
3305 :			long according to @ref{Long Long, , Double-Word Integers, gcc.info, GNU
3306 :			C Manual}). So, we had to implement doubles in C after all. Still, on
3307 :			most machines we can use long longs and achieve better performance than
3308 :			with the emulation package.}. GNU C is available for free on all
3309 :			important (and many unimportant) UNIX machines, VMS, 80386s running
3310 :			MS-DOS, the Amiga, and the Atari ST, so a Forth written in GNU C can run
3311 :			on all these machines.
3312 :	anton	1.3
3313 :			Writing in a portable language has the reputation of producing code that
3314 :			is slower than assembly. For our Forth engine we repeatedly looked at
3315 :			the code produced by the compiler and eliminated most compiler-induced
3316 :			inefficiencies by appropriate changes in the source-code.
3317 :
3318 :			However, register allocation cannot be portably influenced by the
3319 :			programmer, leading to some inefficiencies on register-starved
3320 :			machines. We use explicit register declarations (@pxref{Explicit Reg
3321 :			Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
3322 :			improve the speed on some machines. They are turned on by using the
3323 :			@code{gcc} switch @code{-DFORCE_REG}. Unfortunately, this feature not
3324 :			only depends on the machine, but also on the compiler version: On some
3325 :			machines some compiler versions produce incorrect code when certain
3326 :			explicit register declarations are used. So by default
3327 :			@code{-DFORCE_REG} is not used.
3328 :
3329 :	anton	1.4	@node Threading, Primitives, Portability, Internals
3330 :	anton	1.3	@section Threading
3331 :
3332 :			GNU C's labels as values extension (available since @code{gcc-2.0},
3333 :			@pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
3334 :			makes it possible to take the address of @var{label} by writing
3335 :			@code{&&@var{label}}. This address can then be used in a statement like
3336 :			@code{goto @var{address}}. I.e., @code{goto &&x} is the same as
3337 :			@code{goto x}.
3338 :
3339 :			With this feature an indirect threaded NEXT looks like:
3340 :			@example
3341 :			cfa = *ip++;
3342 :			ca = *cfa;
3343 :			goto *ca;
3344 :			@end example
3345 :			For those unfamiliar with the names: @code{ip} is the Forth instruction
3346 :			pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
3347 :			execution token and points to the code field of the next word to be
3348 :			executed; The @code{ca} (code address) fetched from there points to some
3349 :			executable code, e.g., a primitive or the colon definition handler
3350 :			@code{docol}.
3351 :
3352 :			Direct threading is even simpler:
3353 :			@example
3354 :			ca = *ip++;
3355 :			goto *ca;
3356 :			@end example
3357 :
3358 :			Of course we have packaged the whole thing neatly in macros called
3359 :			@code{NEXT} and @code{NEXT1} (the part of NEXT after fetching the cfa).
3360 :
3361 :	anton	1.4	@menu
3362 :			* Scheduling::
3363 :			* Direct or Indirect Threaded?::
3364 :			* DOES>::
3365 :			@end menu
3366 :
3367 :			@node Scheduling, Direct or Indirect Threaded?, Threading, Threading
3368 :	anton	1.3	@subsection Scheduling
3369 :
3370 :			There is a little complication: Pipelined and superscalar processors,
3371 :			i.e., RISC and some modern CISC machines can process independent
3372 :			instructions while waiting for the results of an instruction. The
3373 :			compiler usually reorders (schedules) the instructions in a way that
3374 :			achieves good usage of these delay slots. However, on our first tries
3375 :			the compiler did not do well on scheduling primitives. E.g., for
3376 :			@code{+} implemented as
3377 :			@example
3378 :			n=sp[0]+sp[1];
3379 :			sp++;
3380 :			sp[0]=n;
3381 :			NEXT;
3382 :			@end example
3383 :			the NEXT comes strictly after the other code, i.e., there is nearly no
3384 :			scheduling. After a little thought the problem becomes clear: The
3385 :			compiler cannot know that sp and ip point to different addresses (and
3386 :	anton	1.4	the version of @code{gcc} we used would not know it even if it was
3387 :			possible), so it could not move the load of the cfa above the store to
3388 :			the TOS. Indeed the pointers could be the same, if code on or very near
3389 :			the top of stack were executed. In the interest of speed we chose to
3390 :			forbid this probably unused ``feature'' and helped the compiler in
3391 :			scheduling: NEXT is divided into the loading part (@code{NEXT_P1}) and
3392 :			the goto part (@code{NEXT_P2}). @code{+} now looks like:
3393 :	anton	1.3	@example
3394 :			n=sp[0]+sp[1];
3395 :			sp++;
3396 :			NEXT_P1;
3397 :			sp[0]=n;
3398 :			NEXT_P2;
3399 :			@end example
3400 :	anton	1.4	This can be scheduled optimally by the compiler.
3401 :	anton	1.3
3402 :			This division can be turned off with the switch @code{-DCISC_NEXT}. This
3403 :			switch is on by default on machines that do not profit from scheduling
3404 :			(e.g., the 80386), in order to preserve registers.
3405 :
3406 :	anton	1.4	@node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
3407 :	anton	1.3	@subsection Direct or Indirect Threaded?
3408 :
3409 :			Both! After packaging the nasty details in macro definitions we
3410 :			realized that we could switch between direct and indirect threading by
3411 :			simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
3412 :			defining a few machine-specific macros for the direct-threading case.
3413 :			On the Forth level we also offer access words that hide the
3414 :			differences between the threading methods (@pxref{Threading Words}).
3415 :
3416 :			Indirect threading is implemented completely
3417 :			machine-independently. Direct threading needs routines for creating
3418 :			jumps to the executable code (e.g. to docol or dodoes). These routines
3419 :			are inherently machine-dependent, but they do not amount to many source
3420 :			lines. I.e., even porting direct threading to a new machine is a small
3421 :			effort.
3422 :
3423 :	anton	1.4	@node DOES>, , Direct or Indirect Threaded?, Threading
3424 :	anton	1.3	@subsection DOES>
3425 :			One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
3426 :			the chunk of code executed by every word defined by a
3427 :			@code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
3428 :			the Forth code to be executed, i.e. the code after the @code{DOES>} (the
3429 :			DOES-code)? There are two solutions:
3430 :
3431 :			In fig-Forth the code field points directly to the dodoes and the
3432 :			DOES-code address is stored in the cell after the code address
3433 :			(i.e. at cfa cell+). It may seem that this solution is illegal in the
3434 :			Forth-79 and all later standards, because in fig-Forth this address
3435 :			lies in the body (which is illegal in these standards). However, by
3436 :			making the code field larger for all words this solution becomes legal
3437 :			again. We use this approach for the indirect threaded version. Leaving
3438 :			a cell unused in most words is a bit wasteful, but on the machines we
3439 :			are targetting this is hardly a problem. The other reason for having a
3440 :			code field size of two cells is to avoid having different image files
3441 :	anton	1.4	for direct and indirect threaded systems (@pxref{System Architecture}).
3442 :	anton	1.3
3443 :			The other approach is that the code field points or jumps to the cell
3444 :			after @code{DOES}. In this variant there is a jump to @code{dodoes} at
3445 :			this address. @code{dodoes} can then get the DOES-code address by
3446 :			computing the code address, i.e., the address of the jump to dodoes,
3447 :			and add the length of that jump field. A variant of this is to have a
3448 :			call to @code{dodoes} after the @code{DOES>}; then the return address
3449 :			(which can be found in the return register on RISCs) is the DOES-code
3450 :			address. Since the two cells available in the code field are usually
3451 :			used up by the jump to the code address in direct threading, we use
3452 :			this approach for direct threading. We did not want to add another
3453 :			cell to the code field.
3454 :
3455 :	anton	1.4	@node Primitives, System Architecture, Threading, Internals
3456 :	anton	1.3	@section Primitives
3457 :
3458 :	anton	1.4	@menu
3459 :			* Automatic Generation::
3460 :			* TOS Optimization::
3461 :			* Produced code::
3462 :			@end menu
3463 :
3464 :			@node Automatic Generation, TOS Optimization, Primitives, Primitives
3465 :	anton	1.3	@subsection Automatic Generation
3466 :
3467 :			Since the primitives are implemented in a portable language, there is no
3468 :			longer any need to minimize the number of primitives. On the contrary,
3469 :			having many primitives is an advantage: speed. In order to reduce the
3470 :			number of errors in primitives and to make programming them easier, we
3471 :			provide a tool, the primitive generator (@file{prims2x.fs}), that
3472 :			automatically generates most (and sometimes all) of the C code for a
3473 :			primitive from the stack effect notation. The source for a primitive
3474 :			has the following form:
3475 :
3476 :			@format
3477 :			@var{Forth-name} @var{stack-effect} @var{category} [@var{pronounc.}]
3478 :			[@code{""}@var{glossary entry}@code{""}]
3479 :			@var{C code}
3480 :			[@code{:}
3481 :			@var{Forth code}]
3482 :			@end format
3483 :
3484 :			The items in brackets are optional. The category and glossary fields
3485 :			are there for generating the documentation, the Forth code is there
3486 :			for manual implementations on machines without GNU C. E.g., the source
3487 :			for the primitive @code{+} is:
3488 :			@example
3489 :			+ n1 n2 -- n core plus
3490 :			n = n1+n2;
3491 :			@end example
3492 :
3493 :			This looks like a specification, but in fact @code{n = n1+n2} is C
3494 :			code. Our primitive generation tool extracts a lot of information from
3495 :			the stack effect notations@footnote{We use a one-stack notation, even
3496 :			though we have separate data and floating-point stacks; The separate
3497 :			notation can be generated easily from the unified notation.}: The number
3498 :			of items popped from and pushed on the stack, their type, and by what
3499 :			name they are referred to in the C code. It then generates a C code
3500 :			prelude and postlude for each primitive. The final C code for @code{+}
3501 :			looks like this:
3502 :
3503 :			@example
3504 :			I_plus: /* + ( n1 n2 -- n ) / / label, stack effect */
3505 :			/* / / documentation */
3506 :	anton	1.4	@{
3507 :	anton	1.3	DEF_CA /* definition of variable ca (indirect threading) */
3508 :			Cell n1; /* definitions of variables */
3509 :			Cell n2;
3510 :			Cell n;
3511 :			n1 = (Cell) sp[1]; /* input */
3512 :			n2 = (Cell) TOS;
3513 :			sp += 1; /* stack adjustment */
3514 :			NAME("+") /* debugging output (with -DDEBUG) */
3515 :	anton	1.4	@{
3516 :	anton	1.3	n = n1+n2; /* C code taken from the source */
3517 :	anton	1.4	@}
3518 :	anton	1.3	NEXT_P1; /* NEXT part 1 */
3519 :			TOS = (Cell)n; /* output */
3520 :			NEXT_P2; /* NEXT part 2 */
3521 :	anton	1.4	@}
3522 :	anton	1.3	@end example
3523 :
3524 :			This looks long and inefficient, but the GNU C compiler optimizes quite
3525 :			well and produces optimal code for @code{+} on, e.g., the R3000 and the
3526 :			HP RISC machines: Defining the @code{n}s does not produce any code, and
3527 :			using them as intermediate storage also adds no cost.
3528 :
3529 :			There are also other optimizations, that are not illustrated by this
3530 :			example: Assignments between simple variables are usually for free (copy
3531 :			propagation). If one of the stack items is not used by the primitive
3532 :			(e.g. in @code{drop}), the compiler eliminates the load from the stack
3533 :			(dead code elimination). On the other hand, there are some things that
3534 :			the compiler does not do, therefore they are performed by
3535 :			@file{prims2x.fs}: The compiler does not optimize code away that stores
3536 :			a stack item to the place where it just came from (e.g., @code{over}).
3537 :
3538 :			While programming a primitive is usually easy, there are a few cases
3539 :			where the programmer has to take the actions of the generator into
3540 :			account, most notably @code{?dup}, but also words that do not (always)
3541 :			fall through to NEXT.
3542 :
3543 :	anton	1.4	@node TOS Optimization, Produced code, Automatic Generation, Primitives
3544 :	anton	1.3	@subsection TOS Optimization
3545 :
3546 :			An important optimization for stack machine emulators, e.g., Forth
3547 :			engines, is keeping one or more of the top stack items in
3548 :	anton	1.4	registers. If a word has the stack effect @var{in1}...@var{inx} @code{--}
3549 :			@var{out1}...@var{outy}, keeping the top @var{n} items in registers
3550 :	anton	1.3	@itemize
3551 :			@item
3552 :			is better than keeping @var{n-1} items, if @var{x>=n} and @var{y>=n},
3553 :			due to fewer loads from and stores to the stack.
3554 :			@item is slower than keeping @var{n-1} items, if @var{x<>y} and @var{x<n} and
3555 :			@var{y<n}, due to additional moves between registers.
3556 :			@end itemize
3557 :
3558 :			In particular, keeping one item in a register is never a disadvantage,
3559 :			if there are enough registers. Keeping two items in registers is a
3560 :			disadvantage for frequent words like @code{?branch}, constants,
3561 :			variables, literals and @code{i}. Therefore our generator only produces
3562 :			code that keeps zero or one items in registers. The generated C code
3563 :			covers both cases; the selection between these alternatives is made at
3564 :			C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
3565 :			code for @code{+} is just a simple variable name in the one-item case,
3566 :			otherwise it is a macro that expands into @code{sp[0]}. Note that the
3567 :			GNU C compiler tries to keep simple variables like @code{TOS} in
3568 :			registers, and it usually succeeds, if there are enough registers.
3569 :
3570 :			The primitive generator performs the TOS optimization for the
3571 :			floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
3572 :			operations the benefit of this optimization is even larger:
3573 :			floating-point operations take quite long on most processors, but can be
3574 :			performed in parallel with other operations as long as their results are
3575 :			not used. If the FP-TOS is kept in a register, this works. If
3576 :			it is kept on the stack, i.e., in memory, the store into memory has to
3577 :			wait for the result of the floating-point operation, lengthening the
3578 :			execution time of the primitive considerably.
3579 :
3580 :			The TOS optimization makes the automatic generation of primitives a
3581 :			bit more complicated. Just replacing all occurrences of @code{sp[0]} by
3582 :			@code{TOS} is not sufficient. There are some special cases to
3583 :			consider:
3584 :			@itemize
3585 :			@item In the case of @code{dup ( w -- w w )} the generator must not
3586 :			eliminate the store to the original location of the item on the stack,
3587 :			if the TOS optimization is turned on.
3588 :	anton	1.4	@item Primitives with stack effects of the form @code{--}
3589 :			@var{out1}...@var{outy} must store the TOS to the stack at the start.
3590 :			Likewise, primitives with the stack effect @var{in1}...@var{inx} @code{--}
3591 :	anton	1.3	must load the TOS from the stack at the end. But for the null stack
3592 :			effect @code{--} no stores or loads should be generated.
3593 :			@end itemize
3594 :
3595 :	anton	1.4	@node Produced code, , TOS Optimization, Primitives
3596 :	anton	1.3	@subsection Produced code
3597 :
3598 :			To see what assembly code is produced for the primitives on your machine
3599 :			with your compiler and your flag settings, type @code{make engine.s} and
3600 :	anton	1.4	look at the resulting file @file{engine.s}.
3601 :	anton	1.3
3602 :	anton	1.17	@node System Architecture, Performance, Primitives, Internals
3603 :	anton	1.3	@section System Architecture
3604 :
3605 :			Our Forth system consists not only of primitives, but also of
3606 :			definitions written in Forth. Since the Forth compiler itself belongs
3607 :			to those definitions, it is not possible to start the system with the
3608 :			primitives and the Forth source alone. Therefore we provide the Forth
3609 :			code as an image file in nearly executable form. At the start of the
3610 :			system a C routine loads the image file into memory, sets up the
3611 :			memory (stacks etc.) according to information in the image file, and
3612 :			starts executing Forth code.
3613 :
3614 :			The image file format is a compromise between the goals of making it
3615 :			easy to generate image files and making them portable. The easiest way
3616 :			to generate an image file is to just generate a memory dump. However,
3617 :			this kind of image file cannot be used on a different machine, or on
3618 :			the next version of the engine on the same machine, it even might not
3619 :			work with the same engine compiled by a different version of the C
3620 :			compiler. We would like to have as few versions of the image file as
3621 :			possible, because we do not want to distribute many versions of the
3622 :			same image file, and to make it easy for the users to use their image
3623 :			files on many machines. We currently need to create a different image
3624 :			file for machines with different cell sizes and different byte order
3625 :	anton	1.17	(little- or big-endian)@footnote{We are considering adding information to the
3626 :	anton	1.3	image file that enables the loader to change the byte order.}.
3627 :
3628 :			Forth code that is going to end up in a portable image file has to
3629 :	anton	1.4	comply to some restrictions: addresses have to be stored in memory with
3630 :			special words (@code{A!}, @code{A,}, etc.) in order to make the code
3631 :			relocatable. Cells, floats, etc., have to be stored at the natural
3632 :			alignment boundaries@footnote{E.g., store floats (8 bytes) at an address
3633 :			dividable by~8. This happens automatically in our system when you use
3634 :			the ANS Forth alignment words.}, in order to avoid alignment faults on
3635 :			machines with stricter alignment. The image file is produced by a
3636 :			metacompiler (@file{cross.fs}).
3637 :	anton	1.3
3638 :			So, unlike the image file of Mitch Bradleys @code{cforth}, our image
3639 :			file is not directly executable, but has to undergo some manipulations
3640 :			during loading. Address relocation is performed at image load-time, not
3641 :			at run-time. The loader also has to replace tokens standing for
3642 :			primitive calls with the appropriate code-field addresses (or code
3643 :			addresses in the case of direct threading).
3644 :	anton	1.4
3645 :	anton	1.17	@node Performance, , System Architecture, Internals
3646 :			@section Performance
3647 :
3648 :			On RISCs the Gforth engine is very close to optimal; i.e., it is usually
3649 :			impossible to write a significantly faster engine.
3650 :
3651 :			On register-starved machines like the 386 architecture processors
3652 :			improvements are possible, because @code{gcc} does not utilize the
3653 :			registers as well as a human, even with explicit register declarations;
3654 :			e.g., Bernd Beuster wrote a Forth system fragment in assembly language
3655 :			and hand-tuned it for the 486; this system is 1.19 times faster on the
3656 :			Sieve benchmark on a 486DX2/66 than Gforth compiled with
3657 :			@code{gcc-2.6.3} with @code{-DFORCE_REG}.
3658 :
3659 :			However, this potential advantage of assembly language implementations
3660 :			is not necessarily realized in complete Forth systems: We compared
3661 :	anton	1.26	Gforth (direct threaded, compiled with @code{gcc-2.6.3} and
3662 :			@code{-DFORCE_REG}) with Win32Forth 1.2093, LMI's NT Forth (Beta, May
3663 :			1994) and Eforth (with and without peephole (aka pinhole) optimization
3664 :			of the threaded code); all these systems were written in assembly
3665 :	anton	1.30	language. We also compared Gforth with three systems written in C:
3666 :	anton	1.32	PFE-0.9.14 (compiled with @code{gcc-2.6.3} with the default
3667 :			configuration for Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS
3668 :			-DUNROLL_NEXT}), ThisForth Beta (compiled with gcc-2.6.3 -O3
3669 :			-fomit-frame-pointer; ThisForth employs peephole optimization of the
3670 :			threaded code) and TILE (compiled with @code{make opt}). We benchmarked
3671 :			Gforth, PFE, ThisForth and TILE on a 486DX2/66 under Linux. Kenneth
3672 :			O'Heskin kindly provided the results for Win32Forth and NT Forth on a
3673 :			486DX2/66 with similar memory performance under Windows NT. Marcel
3674 :			Hendrix ported Eforth to Linux, then extended it to run the benchmarks,
3675 :			added the peephole optimizer, ran the benchmarks and reported the
3676 :			results.
3677 :	anton	1.17
3678 :			We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
3679 :			matrix multiplication come from the Stanford integer benchmarks and have
3680 :			been translated into Forth by Martin Fraeman; we used the versions
3681 :	anton	1.30	included in the TILE Forth package, but with bigger data set sizes; and
3682 :			a recursive Fibonacci number computation for benchmarking calling
3683 :			performance. The following table shows the time taken for the benchmarks
3684 :			scaled by the time taken by Gforth (in other words, it shows the speedup
3685 :			factor that Gforth achieved over the other systems).
3686 :	anton	1.17
3687 :			@example
3688 :	anton	1.30	relative Win32- NT eforth This-
3689 :			time Gforth Forth Forth eforth +opt PFE Forth TILE
3690 :	anton	1.32	sieve 1.00 1.39 1.14 1.39 0.85 1.58 3.18 8.58
3691 :			bubble 1.00 1.31 1.41 1.48 0.88 1.50 3.88
3692 :			matmul 1.00 1.47 1.35 1.46 1.16 1.58 4.09
3693 :			fib 1.00 1.52 1.34 1.22 1.13 1.74 2.99 4.30
3694 :	anton	1.17	@end example
3695 :
3696 :			You may find the good performance of Gforth compared with the systems
3697 :			written in assembly language quite surprising. One important reason for
3698 :			the disappointing performance of these systems is probably that they are
3699 :			not written optimally for the 486 (e.g., they use the @code{lods}
3700 :			instruction). In addition, Win32Forth uses a comfortable, but costly
3701 :			method for relocating the Forth image: like @code{cforth}, it computes
3702 :			the actual addresses at run time, resulting in two address computations
3703 :			per NEXT (@pxref{System Architecture}).
3704 :
3705 :	anton	1.26	Only Eforth with the peephole optimizer performs comparable to
3706 :			Gforth. The speedups achieved with peephole optimization of threaded
3707 :			code are quite remarkable. Adding a peephole optimizer to Gforth should
3708 :			cause similar speedups.
3709 :
3710 :	anton	1.30	The speedup of Gforth over PFE, ThisForth and TILE can be easily
3711 :			explained with the self-imposed restriction to standard C, which makes
3712 :			efficient threading impossible (however, the measured implementation of
3713 :			PFE uses a GNU C extension: @ref{Global Reg Vars, , Defining Global
3714 :			Register Variables, gcc.info, GNU C Manual}). Moreover, current C
3715 :			compilers have a hard time optimizing other aspects of the ThisForth
3716 :			and the TILE source.
3717 :	anton	1.17
3718 :			Note that the performance of Gforth on 386 architecture processors
3719 :			varies widely with the version of @code{gcc} used. E.g., @code{gcc-2.5.8}
3720 :			failed to allocate any of the virtual machine registers into real
3721 :			machine registers by itself and would not work correctly with explicit
3722 :			register declarations, giving a 1.3 times slower engine (on a 486DX2/66
3723 :			running the Sieve) than the one measured above.
3724 :
3725 :	anton	1.26	In @cite{Translating Forth to Efficient C} by M. Anton Ertl and Martin
3726 :			Maierhofer (presented at EuroForth '95), an indirect threaded version of
3727 :			Gforth is compared with Win32Forth, NT Forth, PFE, and ThisForth; that
3728 :			version of Gforth is 2\%@minus{}8\% slower on a 486 than the version
3729 :			used here. The paper available at
3730 :	anton	1.24	@*@file{http://www.complang.tuwien.ac.at/papers/ertl&maierhofer95.ps.gz};
3731 :			it also contains numbers for some native code systems. You can find
3732 :			numbers for Gforth on various machines in @file{Benchres}.
3733 :
3734 :	anton	1.29	@node Bugs, Origin, Internals, Top
3735 :	anton	1.4	@chapter Bugs
3736 :
3737 :	anton	1.17	Known bugs are described in the file BUGS in the Gforth distribution.
3738 :
3739 :	anton	1.24	If you find a bug, please send a bug report to
3740 :	anton	1.32	@code{bug-gforth@@gnu.ai.mit.edu}. A bug report should
3741 :	anton	1.17	describe the Gforth version used (it is announced at the start of an
3742 :			interactive Gforth session), the machine and operating system (on Unix
3743 :			systems you can use @code{uname -a} to produce this information), the
3744 :	anton	1.24	installation options (send the @code{config.status} file), and a
3745 :			complete list of changes you (or your installer) have made to the Gforth
3746 :			sources (if any); it should contain a program (or a sequence of keyboard
3747 :			commands) that reproduces the bug and a description of what you think
3748 :			constitutes the buggy behaviour.
3749 :	anton	1.17
3750 :			For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
3751 :			to Report Bugs, gcc.info, GNU C Manual}.
3752 :
3753 :
3754 :	anton	1.29	@node Origin, Word Index, Bugs, Top
3755 :			@chapter Authors and Ancestors of Gforth
3756 :
3757 :			@section Authors and Contributors
3758 :
3759 :			The Gforth project was started in mid-1992 by Bernd Paysan and Anton
3760 :	anton	1.30	Ertl. The third major author was Jens Wilke. Lennart Benschop (who was
3761 :			one of Gforth's first users, in mid-1993) and Stuart Ramsden inspired us
3762 :			with their continuous feedback. Lennart Benshop contributed
3763 :	anton	1.29	@file{glosgen.fs}, while Stuart Ramsden has been working on automatic
3764 :			support for calling C libraries. Helpful comments also came from Paul
3765 :			Kleinrubatscher, Christian Pirker, Dirk Zoller and Marcel Hendrix.
3766 :
3767 :	anton	1.30	Gforth also owes a lot to the authors of the tools we used (GCC, CVS,
3768 :			and autoconf, among others), and to the creators of the Internet: Gforth
3769 :			was developed across the Internet, and its authors have not met
3770 :			physically yet.
3771 :
3772 :	anton	1.29	@section Pedigree
3773 :	anton	1.4
3774 :	anton	1.17	Gforth descends from BigForth (1993) and fig-Forth. Gforth and PFE (by
3775 :	anton	1.24	Dirk Zoller) will cross-fertilize each other. Of course, a significant
3776 :			part of the design of Gforth was prescribed by ANS Forth.
3777 :	anton	1.17
3778 :	pazsan	1.23	Bernd Paysan wrote BigForth, a descendent from TurboForth, an unreleased
3779 :			32 bit native code version of VolksForth for the Atari ST, written
3780 :			mostly by Dietrich Weineck.
3781 :
3782 :			VolksForth descends from F83. It was written by Klaus Schleisiek, Bernd
3783 :			Pennemann, Georg Rehfeld and Dietrich Weineck for the C64 (called
3784 :	anton	1.24	UltraForth there) in the mid-80s and ported to the Atari ST in 1986.
3785 :	anton	1.17
3786 :	anton	1.25	Hennry Laxen and Mike Perry wrote F83 as a model implementation of the
3787 :	anton	1.17	Forth-83 standard. !! Pedigree? When?
3788 :
3789 :			A team led by Bill Ragsdale implemented fig-Forth on many processors in
3790 :	anton	1.24	1979. Robert Selzer and Bill Ragsdale developed the original
3791 :			implementation of fig-Forth for the 6502 based on microForth.
3792 :
3793 :			The principal architect of microForth was Dean Sanderson. microForth was
3794 :			FORTH, Inc.'s first off-the-shelf product. It was developped in 1976 for
3795 :			the 1802, and subsequently implemented on the 8080, the 6800 and the
3796 :			Z80.
3797 :	anton	1.17
3798 :	anton	1.24	All earlier Forth systems were custom-made, usually by Charles Moore,
3799 :	anton	1.30	who discovered (as he puts it) Forth during the late 60s. The first full
3800 :			Forth existed in 1971.
3801 :	anton	1.17
3802 :			A part of the information in this section comes from @cite{The Evolution
3803 :			of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
3804 :			H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
3805 :			Notices 28(3), 1993. You can find more historical and genealogical
3806 :			information about Forth there.
3807 :
3808 :	anton	1.29	@node Word Index, Node Index, Origin, Top
3809 :	anton	1.4	@chapter Word Index
3810 :
3811 :	anton	1.18	This index is as incomplete as the manual. Each word is listed with
3812 :			stack effect and wordset.
3813 :	anton	1.17
3814 :			@printindex fn
3815 :
3816 :	anton	1.4	@node Node Index, , Word Index, Top
3817 :			@chapter Node Index
3818 :	anton	1.17
3819 :			This index is even less complete than the manual.
3820 :	anton	1.1
3821 :			@contents
3822 :			@bye
3823 :

CVS Admin

gforth: gforth/Attic/gforth.ds

ViewCVS and CVS Help