]> git.saurik.com Git - bison.git/blob - NEWS
Update.
[bison.git] / NEWS
1 Bison News
2 ----------
3
4 Changes in version 1.49c:
5
6
7 Changes in version 1.49b, 2002-08-12:
8
9 * GLR parsing
10 The declaration
11 %glr-parser
12 causes Bison to produce a Generalized LR (GLR) parser, capable of handling
13 almost any context-free grammar, ambiguous or not. The new declarations
14 %dprec and %merge on grammar rules allow parse-time resolution of
15 ambiguities. Contributed by Paul Hilfinger.
16
17 * Output Directory
18 When not in Yacc compatibility mode, when the output file was not
19 specified, runnning `bison foo/bar.y' created `foo/bar.c'. It
20 now creates `bar.c'.
21
22 * Undefined token
23 The undefined token was systematically mapped to 2 which prevented
24 the use of 2 from the user. This is no longer the case.
25
26 * Unknown token numbers
27 If yylex returned a code out of range, yyparse could die. This is
28 no longer the case.
29
30 * Error token
31 According to POSIX, the error token should be numbered as 256.
32 Bison extends this requirement by making it a preference: *if* the
33 user specified that one of her tokens is numbered 256, then error
34 will be mapped onto another number.
35
36 * Verbose error messages
37 They no longer report `..., expecting error or...' for state where
38 error recovery is possible.
39
40 * End token
41 Defaults to `$end' instead of `$'.
42
43 * Error recovery now conforms to documentation and to POSIX
44 When a Bison-generated parser encounters a syntax error, it now pops
45 the stack until it finds a state that allows shifting the error
46 token. Formerly, it popped the stack until it found a state that
47 allowed some non-error action other than a default reduction on the
48 error token. The new behavior has long been the documented behavior,
49 and has long been required by POSIX. For more details, please see
50 <http://mail.gnu.org/pipermail/bug-bison/2002-May/001452.html>.
51
52 * Traces
53 Popped tokens and nonterminals are now reported.
54
55 * Larger grammars
56 Larger grammars are now supported (larger token numbers, larger grammar
57 size (= sum of the LHS and RHS lengths), larger LALR tables).
58 Formerly, many of these numbers ran afoul of 16-bit limits;
59 now these limits are 32 bits on most hosts.
60
61 * Explicit initial rule
62 Bison used to play hacks with the initial rule, which the user does
63 not write. It is now explicit, and visible in the reports and
64 graphs as rule 0.
65
66 * Useless rules
67 Before, Bison reported the useless rules, but, although not used,
68 included them in the parsers. They are now actually removed.
69
70 * Useless rules, useless nonterminals
71 They are now reported, as a warning, with their locations.
72
73 * Rules never reduced
74 Rules that can never be reduced because of conflicts are now
75 reported.
76
77 * Incorrect `Token not used'
78 On a grammar such as
79
80 %token useless useful
81 %%
82 exp: '0' %prec useful;
83
84 where a token was used to set the precedence of the last rule,
85 bison reported both `useful' and `useless' as useless tokens.
86
87 * Revert the C++ namespace changes introduced in 1.31
88 as they caused too many portability hassles.
89
90 * Default locations
91 By an accident of design, the default computation of @$ was
92 performed after another default computation was performed: @$ = @1.
93 The latter is now removed: YYLLOC_DEFAULT is fully responsible of
94 the computation of @$.
95
96 * Token end-of-file
97 The token end of file may be specified by the user, in which case,
98 the user symbol is used in the reports, the graphs, and the verbose
99 error messages instead of `$end', which remains being the default.
100 For instance
101 %token YYEOF 0
102 or
103 %token YYEOF 0 "end of file"
104
105 * Semantic parser
106 This old option, which has been broken for ages, is removed.
107
108 * New translations
109 Brazilian Portuguese, thanks to Alexandre Folle de Menezes.
110 Croatian, thanks to Denis Lackovic.
111
112 * Incorrect token definitions
113 When fed with `%token 'a' "A"', Bison used to output `#define 'a' 65'.
114
115 * Token definitions as enums
116 Tokens are output both as the traditional #define's, and, provided
117 the compiler supports ANSI C or is a C++ compiler, as enums.
118 This helps debuggers producing symbols instead of values.
119
120 * Reports
121 In addition to --verbose, bison supports --report=THINGS, which
122 produces additional information:
123 - itemset
124 complete the core item sets with their closure
125 - lookahead
126 explicitly associate lookaheads to items
127 - solved
128 describe shift/reduce conflicts solving.
129 Bison used to systematically output this information on top of
130 the report. Solved conflicts are now attached to their states.
131
132 * Type clashes
133 Previous versions don't complain when there is a type clash on
134 the default action if the rule has a mid-rule action, such as in:
135
136 %type <foo> bar
137 %%
138 bar: '0' {} '0';
139
140 This is fixed.
141
142 * GNU M4 is now required when using Bison.
143 \f
144 Changes in version 1.35, 2002-03-25:
145
146 * C Skeleton
147 Some projects use Bison's C parser with C++ compilers, and define
148 YYSTYPE as a class. The recent adjustment of C parsers for data
149 alignment and 64 bit architectures made this impossible.
150
151 Because for the time being no real solution for C++ parser
152 generation exists, kludges were implemented in the parser to
153 maintain this use. In the future, when Bison has C++ parsers, this
154 kludge will be disabled.
155
156 This kludge also addresses some C++ problems when the stack was
157 extended.
158
159 \f
160 Changes in version 1.34, 2002-03-12:
161
162 * File name clashes are detected
163 $ bison foo.y -d -o foo.x
164 fatal error: header and parser would both be named `foo.x'
165
166 * A missing `;' at the end of a rule triggers a warning
167 In accordance with POSIX, and in agreement with other
168 Yacc implementations, Bison will mandate this semicolon in the near
169 future. This eases the implementation of a Bison parser of Bison
170 grammars by making this grammar LALR(1) instead of LR(2). To
171 facilitate the transition, this release introduces a warning.
172
173 * Revert the C++ namespace changes introduced in 1.31, as they caused too
174 many portability hassles.
175
176 * DJGPP support added.
177
178 * Fix test suite portability problems.
179 \f
180 Changes in version 1.33, 2002-02-07:
181
182 * Fix C++ issues
183 Groff could not be compiled for the definition of size_t was lacking
184 under some conditions.
185
186 * Catch invalid @n
187 As is done with $n.
188 \f
189 Changes in version 1.32, 2002-01-23:
190
191 * Fix Yacc output file names
192
193 * Portability fixes
194
195 * Italian, Dutch translations
196 \f
197 Changes in version 1.31, 2002-01-14:
198
199 * Many Bug Fixes
200
201 * GNU Gettext and %expect
202 GNU Gettext asserts 10 s/r conflicts, but there are 7. Now that
203 Bison dies on incorrect %expectations, we fear there will be
204 too many bug reports for Gettext, so _for the time being_, %expect
205 does not trigger an error when the input file is named `plural.y'.
206
207 * Use of alloca in parsers
208 If YYSTACK_USE_ALLOCA is defined to 0, then the parsers will use
209 malloc exclusively. Since 1.29, but was not NEWS'ed.
210
211 alloca is used only when compiled with GCC, to avoid portability
212 problems as on AIX.
213
214 * When the generated parser lacks debugging code, YYDEBUG is now 0
215 (as POSIX requires) instead of being undefined.
216
217 * User Actions
218 Bison has always permitted actions such as { $$ = $1 }: it adds the
219 ending semicolon. Now if in Yacc compatibility mode, the semicolon
220 is no longer output: one has to write { $$ = $1; }.
221
222 * Better C++ compliance
223 The output parsers try to respect C++ namespaces.
224 [This turned out to be a failed experiment, and it was reverted later.]
225
226 * Reduced Grammars
227 Fixed bugs when reporting useless nonterminals.
228
229 * 64 bit hosts
230 The parsers work properly on 64 bit hosts.
231
232 * Error messages
233 Some calls to strerror resulted in scrambled or missing error messages.
234
235 * %expect
236 When the number of shift/reduce conflicts is correct, don't issue
237 any warning.
238
239 * The verbose report includes the rule line numbers.
240
241 * Rule line numbers are fixed in traces.
242
243 * Swedish translation
244
245 * Parse errors
246 Verbose parse error messages from the parsers are better looking.
247 Before: parse error: unexpected `'/'', expecting `"number"' or `'-'' or `'(''
248 Now: parse error: unexpected '/', expecting "number" or '-' or '('
249
250 * Fixed parser memory leaks.
251 When the generated parser was using malloc to extend its stacks, the
252 previous allocations were not freed.
253
254 * Fixed verbose output file.
255 Some newlines were missing.
256 Some conflicts in state descriptions were missing.
257
258 * Fixed conflict report.
259 Option -v was needed to get the result.
260
261 * %expect
262 Was not used.
263 Mismatches are errors, not warnings.
264
265 * Fixed incorrect processing of some invalid input.
266
267 * Fixed CPP guards: 9foo.h uses BISON_9FOO_H instead of 9FOO_H.
268
269 * Fixed some typos in the documentation.
270
271 * %token MY_EOF 0 is supported.
272 Before, MY_EOF was silently renumbered as 257.
273
274 * doc/refcard.tex is updated.
275
276 * %output, %file-prefix, %name-prefix.
277 New.
278
279 * --output
280 New, aliasing `--output-file'.
281 \f
282 Changes in version 1.30, 2001-10-26:
283
284 * `--defines' and `--graph' have now an optionnal argument which is the
285 output file name. `-d' and `-g' do not change, they do not take any
286 argument.
287
288 * `%source_extension' and `%header_extension' are removed, failed
289 experiment.
290
291 * Portability fixes.
292 \f
293 Changes in version 1.29, 2001-09-07:
294
295 * The output file does not define const, as this caused problems when used
296 with common autoconfiguration schemes. If you still use ancient compilers
297 that lack const, compile with the equivalent of the C compiler option
298 `-Dconst='. autoconf's AC_C_CONST macro provides one way to do this.
299
300 * Added `-g' and `--graph'.
301
302 * The Bison manual is now distributed under the terms of the GNU FDL.
303
304 * The input and the output files has automatically a similar extension.
305
306 * Russian translation added.
307
308 * NLS support updated; should hopefully be less troublesome.
309
310 * Added the old Bison reference card.
311
312 * Added `--locations' and `%locations'.
313
314 * Added `-S' and `--skeleton'.
315
316 * `%raw', `-r', `--raw' is disabled.
317
318 * Special characters are escaped when output. This solves the problems
319 of the #line lines with path names including backslashes.
320
321 * New directives.
322 `%yacc', `%fixed_output_files', `%defines', `%no_parser', `%verbose',
323 `%debug', `%source_extension' and `%header_extension'.
324
325 * @$
326 Automatic location tracking.
327 \f
328 Changes in version 1.28, 1999-07-06:
329
330 * Should compile better now with K&R compilers.
331
332 * Added NLS.
333
334 * Fixed a problem with escaping the double quote character.
335
336 * There is now a FAQ.
337 \f
338 Changes in version 1.27:
339
340 * The make rule which prevented bison.simple from being created on
341 some systems has been fixed.
342 \f
343 Changes in version 1.26:
344
345 * Bison now uses automake.
346
347 * New mailing lists: <bug-bison@gnu.org> and <help-bison@gnu.org>.
348
349 * Token numbers now start at 257 as previously documented, not 258.
350
351 * Bison honors the TMPDIR environment variable.
352
353 * A couple of buffer overruns have been fixed.
354
355 * Problems when closing files should now be reported.
356
357 * Generated parsers should now work even on operating systems which do
358 not provide alloca().
359 \f
360 Changes in version 1.25, 1995-10-16:
361
362 * Errors in the input grammar are not fatal; Bison keeps reading
363 the grammar file, and reports all the errors found in it.
364
365 * Tokens can now be specified as multiple-character strings: for
366 example, you could use "<=" for a token which looks like <=, instead
367 of chosing a name like LESSEQ.
368
369 * The %token_table declaration says to write a table of tokens (names
370 and numbers) into the parser file. The yylex function can use this
371 table to recognize multiple-character string tokens, or for other
372 purposes.
373
374 * The %no_lines declaration says not to generate any #line preprocessor
375 directives in the parser file.
376
377 * The %raw declaration says to use internal Bison token numbers, not
378 Yacc-compatible token numbers, when token names are defined as macros.
379
380 * The --no-parser option produces the parser tables without including
381 the parser engine; a project can now use its own parser engine.
382 The actions go into a separate file called NAME.act, in the form of
383 a switch statement body.
384 \f
385 Changes in version 1.23:
386
387 The user can define YYPARSE_PARAM as the name of an argument to be
388 passed into yyparse. The argument should have type void *. It should
389 actually point to an object. Grammar actions can access the variable
390 by casting it to the proper pointer type.
391
392 Line numbers in output file corrected.
393 \f
394 Changes in version 1.22:
395
396 --help option added.
397 \f
398 Changes in version 1.20:
399
400 Output file does not redefine const for C++.
401
402 Local Variables:
403 mode: outline
404 End:
405
406 -----
407
408 Copyright (C) 2001, 2002 Free Software Foundation, Inc.
409
410 This file is part of GNU Autoconf.
411
412 GNU Autoconf is free software; you can redistribute it and/or modify
413 it under the terms of the GNU General Public License as published by
414 the Free Software Foundation; either version 2, or (at your option)
415 any later version.
416
417 GNU Autoconf is distributed in the hope that it will be useful,
418 but WITHOUT ANY WARRANTY; without even the implied warranty of
419 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
420 GNU General Public License for more details.
421
422 You should have received a copy of the GNU General Public License
423 along with autoconf; see the file COPYING. If not, write to
424 the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
425 Boston, MA 02111-1307, USA.