]> git.saurik.com Git - bison.git/blame - NEWS
doc: fix index.
[bison.git] / NEWS
CommitLineData
6780ca7a
DM
1Bison News
2----------
3af4feb2 3
90dcd415
JD
4* Changes in version 2.5.1 (????-??-??):
5
041308d0 6** Several portability problems in the test suite have been fixed:
74516381 7
041308d0
AD
8 This includes warnings with some compilers, unexpected behavior of
9 tools such as diff, warning messages from the test suite itself,
10 etc.
11
ef51bfa7
AD
12 __attribute__ is not longer disabled when __STRICT_ANSI__ is defined
13 (i.e., when -std is passed to GCC).
14
041308d0
AD
15** Warnings during the build procedure have been eliminated.
16
17** Many minor improvements have been made to the manual:
18
19 The layout for grammar has changed to a more compact scheme. Named
20 references are motivated. The description of the automaton
21 description file is updated.
908c8647 22
94556574
AD
23** YYBACKUP works as expected.
24
041308d0
AD
25** liby is no longer asking for "rpl_fprintf" on some platforms.
26
27** Several Java fixes:
28
29 The Java parser no longer throws ArrayIndexOutOfBoundsException if
30 the first token leads to a syntax error. Some minor clean ups.
31
ef51bfa7
AD
32** C++11 compatibility:
33
34 C and C++ parsers use nullptr instead of 0 when __cplusplus is
35 201103L or higher.
36
936c88d1
AD
37** C++ locations:
38
39 The position and location constructors (and their initialize
40 methods) accept new arguments for line and column. Several issues
41 in the documentation were fixed.
42
b2bc8a1b 43* Changes in version 2.5 (2011-05-14):
006faedf 44
eb8c66bb
JD
45** Grammar symbol names can now contain non-initial dashes:
46
47 Consistently with directives (such as %error-verbose) and with
48 %define variables (e.g. push-pull), grammar symbol names may contain
49 dashes in any position except the beginning. This is a GNU
50 extension over POSIX Yacc. Thus, use of this extension is reported
51 by -Wyacc and rejected in Yacc mode (--yacc).
52
3b1977ea 53** Named references:
348f5608
AR
54
55 Historically, Yacc and Bison have supported positional references
56 ($n, $$) to allow access to symbol values from inside of semantic
57 actions code.
58
59 Starting from this version, Bison can also accept named references.
60 When no ambiguity is possible, original symbol names may be used
61 as named references:
62
ac3297d5 63 if_stmt : "if" cond_expr "then" then_stmt ';'
348f5608
AR
64 { $if_stmt = mk_if_stmt($cond_expr, $then_stmt); }
65
66 In the more common case, explicit names may be declared:
67
ac3297d5 68 stmt[res] : "if" expr[cond] "then" stmt[then] "else" stmt[else] ';'
348f5608
AR
69 { $res = mk_if_stmt($cond, $then, $else); }
70
f840c05a 71 Location information is also accessible using @name syntax. When
348f5608
AR
72 accessing symbol names containing dots or dashes, explicit bracketing
73 ($[sym.1]) must be used.
74
f840c05a 75 These features are experimental in this version. More user feedback
348f5608
AR
76 will help to stabilize them.
77
3b1977ea 78** IELR(1) and canonical LR(1):
34a6c2d1
JD
79
80 IELR(1) is a minimal LR(1) parser table generation algorithm. That
81 is, given any context-free grammar, IELR(1) generates parser tables
723206df 82 with the full language-recognition power of canonical LR(1) but with
3b1977ea
JD
83 nearly the same number of parser states as LALR(1). This reduction
84 in parser states is often an order of magnitude. More importantly,
34a6c2d1
JD
85 because canonical LR(1)'s extra parser states may contain duplicate
86 conflicts in the case of non-LR(1) grammars, the number of conflicts
87 for IELR(1) is often an order of magnitude less as well. This can
88 significantly reduce the complexity of developing of a grammar.
89
90 Bison can now generate IELR(1) and canonical LR(1) parser tables in
91 place of its traditional LALR(1) parser tables, which remain the
92 default. You can specify the type of parser tables in the grammar
93 file with these directives:
94
f37495f6
JD
95 %define lr.type lalr
96 %define lr.type ielr
97 %define lr.type canonical-lr
34a6c2d1 98
6f04ee6c
JD
99 The default-reduction optimization in the parser tables can also be
100 adjusted using `%define lr.default-reductions'. For details on both
101 of these features, see the new section `Tuning LR' in the Bison
102 manual.
34a6c2d1
JD
103
104 These features are experimental. More user feedback will help to
105 stabilize them.
106
6f04ee6c 107** LAC (Lookahead Correction) for syntax error handling:
4c38b19e
JD
108
109 Canonical LR, IELR, and LALR can suffer from a couple of problems
110 upon encountering a syntax error. First, the parser might perform
111 additional parser stack reductions before discovering the syntax
6f04ee6c 112 error. Such reductions can perform user semantic actions that are
4c38b19e
JD
113 unexpected because they are based on an invalid token, and they
114 cause error recovery to begin in a different syntactic context than
115 the one in which the invalid token was encountered. Second, when
6f04ee6c
JD
116 verbose error messages are enabled (with %error-verbose or the
117 obsolete `#define YYERROR_VERBOSE'), the expected token list in the
118 syntax error message can both contain invalid tokens and omit valid
119 tokens.
4c38b19e
JD
120
121 The culprits for the above problems are %nonassoc, default
122 reductions in inconsistent states, and parser state merging. Thus,
123 IELR and LALR suffer the most. Canonical LR can suffer only if
124 %nonassoc is used or if default reductions are enabled for
125 inconsistent states.
126
6f04ee6c
JD
127 LAC is a new mechanism within the parsing algorithm that solves
128 these problems for canonical LR, IELR, and LALR without sacrificing
129 %nonassoc, default reductions, or state merging. When LAC is in
130 use, canonical LR and IELR behave almost exactly the same for both
131 syntactically acceptable and syntactically unacceptable input.
4c38b19e
JD
132 While LALR still does not support the full language-recognition
133 power of canonical LR and IELR, LAC at least enables LALR's syntax
134 error handling to correctly reflect LALR's language-recognition
135 power.
136
137 Currently, LAC is only supported for deterministic parsers in C.
138 You can enable LAC with the following directive:
139
140 %define parse.lac full
141
6f04ee6c
JD
142 See the new section `LAC' in the Bison manual for additional
143 details including a few caveats.
4c38b19e
JD
144
145 LAC is an experimental feature. More user feedback will help to
146 stabilize it.
147
6ee881f6 148** %define improvements:
f37495f6 149
3b1977ea 150*** Can now be invoked via the command line:
4ecd3681 151
34d41938 152 Each of these command-line options
4ecd3681 153
34d41938
JD
154 -D NAME[=VALUE]
155 --define=NAME[=VALUE]
156
157 -F NAME[=VALUE]
158 --force-define=NAME[=VALUE]
4ecd3681
JD
159
160 is equivalent to this grammar file declaration
161
34d41938 162 %define NAME ["VALUE"]
4ecd3681 163
34d41938
JD
164 except that the manner in which Bison processes multiple definitions
165 for the same NAME differs. Most importantly, -F and --force-define
166 quietly override %define, but -D and --define do not. For further
ac3297d5 167 details, see the section `Bison Options' in the Bison manual.
4ecd3681 168
3b1977ea 169*** Variables renamed:
812775a0
JD
170
171 The following %define variables
172
173 api.push_pull
174 lr.keep_unreachable_states
175
176 have been renamed to
177
178 api.push-pull
179 lr.keep-unreachable-states
180
181 The old names are now deprecated but will be maintained indefinitely
182 for backward compatibility.
183
723206df 184*** Values no longer need to be quoted in the grammar file:
f37495f6
JD
185
186 If a %define value is an identifier, it no longer needs to be placed
187 within quotations marks. For example,
188
189 %define api.push-pull "push"
190
191 can be rewritten as
192
193 %define api.push-pull push
194
6ee881f6 195*** Unrecognized variables are now errors not warnings.
c046698e 196
6ee881f6
JD
197*** Multiple invocations for any variable is now an error not a warning.
198
199** Unrecognized %code qualifiers are now errors not warnings.
200
201** Character literals not of length one:
202
203 Previously, Bison quietly converted all character literals to length
204 one. For example, without warning, Bison interpreted the operators in
205 the following grammar to be the same token:
206
207 exp: exp '++'
208 | exp '+' exp
209 ;
210
211 Bison now warns when a character literal is not of length one. In
212 some future release, Bison will start reporting an error instead.
213
214** Destructor calls fixed for lookaheads altered in semantic actions:
215
216 Previously for deterministic parsers in C, if a user semantic action
217 altered yychar, the parser in some cases used the old yychar value to
218 determine which destructor to call for the lookahead upon a syntax
219 error or upon parser return. This bug has been fixed.
220
221** C++ parsers use YYRHSLOC:
222
223 Similarly to the C parsers, the C++ parsers now define the YYRHSLOC
224 macro and use it in the default YYLLOC_DEFAULT. You are encouraged
ac3297d5
JD
225 to use it. If, for instance, your location structure has `first'
226 and `last' members, instead of
6ee881f6
JD
227
228 # define YYLLOC_DEFAULT(Current, Rhs, N) \
229 do \
230 if (N) \
231 { \
232 (Current).first = (Rhs)[1].location.first; \
233 (Current).last = (Rhs)[N].location.last; \
234 } \
235 else \
236 { \
237 (Current).first = (Current).last = (Rhs)[0].location.last; \
238 } \
239 while (false)
240
241 use:
242
243 # define YYLLOC_DEFAULT(Current, Rhs, N) \
244 do \
245 if (N) \
246 { \
247 (Current).first = YYRHSLOC (Rhs, 1).first; \
248 (Current).last = YYRHSLOC (Rhs, N).last; \
249 } \
250 else \
251 { \
252 (Current).first = (Current).last = YYRHSLOC (Rhs, 0).last; \
253 } \
254 while (false)
255
256** YYLLOC_DEFAULT in C++:
257
258 The default implementation of YYLLOC_DEFAULT used to be issued in
259 the header file. It is now output in the implementation file, after
260 the user %code sections so that its #ifndef guard does not try to
261 override the user's YYLLOC_DEFAULT if provided.
c046698e 262
3b1977ea 263** YYFAIL now produces warnings and Java parsers no longer implement it:
62efdd2a
JD
264
265 YYFAIL has existed for many years as an undocumented feature of
266 deterministic parsers in C generated by Bison. More recently, it was
267 a documented feature of Bison's experimental Java parsers. As
268 promised in Bison 2.4.2's NEWS entry, any appearance of YYFAIL in a
269 semantic action now produces a deprecation warning, and Java parsers
270 no longer implement YYFAIL at all. For further details, including a
271 discussion of how to suppress C preprocessor warnings about YYFAIL
272 being unused, see the Bison 2.4.2 NEWS entry.
273
3b1977ea 274** Temporary hack for adding a semicolon to the user action:
a2d05674
JD
275
276 Previously, Bison appended a semicolon to every user action for
277 reductions when the output language defaulted to C (specifically, when
278 neither %yacc, %language, %skeleton, or equivalent command-line
279 options were specified). This allowed actions such as
280
281 exp: exp "+" exp { $$ = $1 + $3 };
282
283 instead of
284
285 exp: exp "+" exp { $$ = $1 + $3; };
286
287 As a first step in removing this misfeature, Bison now issues a
288 warning when it appends a semicolon. Moreover, in cases where Bison
289 cannot easily determine whether a semicolon is needed (for example, an
290 action ending with a cpp directive or a braced compound initializer),
291 it no longer appends one. Thus, the C compiler might now complain
292 about a missing semicolon where it did not before. Future releases of
293 Bison will cease to append semicolons entirely.
294
095a1d11
JD
295** Verbose syntax error message fixes:
296
6f04ee6c
JD
297 When %error-verbose or the obsolete `#define YYERROR_VERBOSE' is
298 specified, syntax error messages produced by the generated parser
299 include the unexpected token as well as a list of expected tokens.
300 The effect of %nonassoc on these verbose messages has been corrected
301 in two ways, but a more complete fix requires LAC, described above:
095a1d11
JD
302
303*** When %nonassoc is used, there can exist parser states that accept no
304 tokens, and so the parser does not always require a lookahead token
305 in order to detect a syntax error. Because no unexpected token or
306 expected tokens can then be reported, the verbose syntax error
307 message described above is suppressed, and the parser instead
ac3297d5 308 reports the simpler message, `syntax error'. Previously, this
095a1d11
JD
309 suppression was sometimes erroneously triggered by %nonassoc when a
310 lookahead was actually required. Now verbose messages are
311 suppressed only when all previous lookaheads have already been
312 shifted or discarded.
313
314*** Previously, the list of expected tokens erroneously included tokens
315 that would actually induce a syntax error because conflicts for them
316 were resolved with %nonassoc in the current parser state. Such
317 tokens are now properly omitted from the list.
318
319*** Expected token lists are still often wrong due to state merging
4c38b19e
JD
320 (from LALR or IELR) and default reductions, which can both add
321 invalid tokens and subtract valid tokens. Canonical LR almost
322 completely fixes this problem by eliminating state merging and
323 default reductions. However, there is one minor problem left even
324 when using canonical LR and even after the fixes above. That is,
325 if the resolution of a conflict with %nonassoc appears in a later
326 parser state than the one at which some syntax error is
327 discovered, the conflicted token is still erroneously included in
328 the expected token list. Bison's new LAC implementation,
329 described above, eliminates this problem and the need for
330 canonical LR. However, LAC is still experimental and is disabled
331 by default.
1fa30307 332
4c2a6e42
JD
333** Java skeleton fixes:
334
335*** A location handling bug has been fixed.
336
337*** The top element of each of the value stack and location stack is now
338 cleared when popped so that it can be garbage collected.
77768165 339
f0e2c228
JD
340*** Parser traces now print the top element of the stack.
341
248dba53
JD
342** -W/--warnings fixes:
343
ac3297d5 344*** Bison now properly recognizes the `no-' versions of categories:
248dba53
JD
345
346 For example, given the following command line, Bison now enables all
347 warnings except warnings for incompatibilities with POSIX Yacc:
348
349 bison -Wall,no-yacc gram.y
350
6f8bdce2
JD
351*** Bison now treats S/R and R/R conflicts like other warnings:
352
353 Previously, conflict reports were independent of Bison's normal
354 warning system. Now, Bison recognizes the warning categories
ac3297d5 355 `conflicts-sr' and `conflicts-rr'. This change has important
6f8bdce2
JD
356 consequences for the -W and --warnings command-line options. For
357 example:
358
359 bison -Wno-conflicts-sr gram.y # S/R conflicts not reported
360 bison -Wno-conflicts-rr gram.y # R/R conflicts not reported
361 bison -Wnone gram.y # no conflicts are reported
362 bison -Werror gram.y # any conflict is an error
363
364 However, as before, if the %expect or %expect-rr directive is
365 specified, an unexpected number of conflicts is an error, and an
366 expected number of conflicts is not reported, so -W and --warning
367 then have no effect on the conflict report.
368
ac3297d5 369*** The `none' category no longer disables a preceding `error':
dab96632
JD
370
371 For example, for the following command line, Bison now reports
372 errors instead of warnings for incompatibilities with POSIX Yacc:
373
374 bison -Werror,none,yacc gram.y
375
b2bc8a1b 376*** The `none' category now disables all Bison warnings:
8ffd7912 377
ac3297d5 378 Previously, the `none' category disabled only Bison warnings for
8ffd7912
JD
379 which there existed a specific -W/--warning category. However,
380 given the following command line, Bison is now guaranteed to
381 suppress all warnings:
382
383 bison -Wnone gram.y
384
9d6af153
JD
385** Precedence directives can now assign token number 0:
386
387 Since Bison 2.3b, which restored the ability of precedence
388 directives to assign token numbers, doing so for token number 0 has
389 produced an assertion failure. For example:
390
391 %left END 0
392
393 This bug has been fixed.
394
7527c744 395* Changes in version 2.4.3 (2010-08-05):
7a9c3cb3 396
4631c34f
JD
397** Bison now obeys -Werror and --warnings=error for warnings about
398 grammar rules that are useless in the parser due to conflicts.
399
7a9c3cb3
JD
400** Problems with spawning M4 on at least FreeBSD 8 and FreeBSD 9 have
401 been fixed.
402
132247cd
JD
403** Failures in the test suite for GCC 4.5 have been fixed.
404
88bb35d6
JD
405** Failures in the test suite for some versions of Sun Studio C++ have
406 been fixed.
407
43fdc9fd
JD
408** Contrary to Bison 2.4.2's NEWS entry, it has been decided that
409 warnings about undefined %prec identifiers will not be converted to
410 errors in Bison 2.5. They will remain warnings, which should be
411 sufficient for POSIX while avoiding backward compatibility issues.
412
cf22447c
JD
413** Minor documentation fixes.
414
ea66d039 415* Changes in version 2.4.2 (2010-03-20):
8defe11b 416
47fa5747
JD
417** Some portability problems that resulted in failures and livelocks
418 in the test suite on some versions of at least Solaris, AIX, HP-UX,
ea66d039
JD
419 RHEL4, and Tru64 have been addressed. As a result, fatal Bison
420 errors should no longer cause M4 to report a broken pipe on the
47fa5747
JD
421 affected platforms.
422
2c203528
JD
423** `%prec IDENTIFIER' requires IDENTIFIER to be defined separately.
424
425 POSIX specifies that an error be reported for any identifier that does
426 not appear on the LHS of a grammar rule and that is not defined by
427 %token, %left, %right, or %nonassoc. Bison 2.3b and later lost this
428 error report for the case when an identifier appears only after a
429 %prec directive. It is now restored. However, for backward
430 compatibility with recent Bison releases, it is only a warning for
431 now. In Bison 2.5 and later, it will return to being an error.
43fdc9fd
JD
432 [Between the 2.4.2 and 2.4.3 releases, it was decided that this
433 warning will not be converted to an error in Bison 2.5.]
2c203528 434
c5196098
EB
435** Detection of GNU M4 1.4.6 or newer during configure is improved.
436
34731471
JD
437** Warnings from gcc's -Wundef option about undefined YYENABLE_NLS,
438 YYLTYPE_IS_TRIVIAL, and __STRICT_ANSI__ in C/C++ parsers are now
439 avoided.
af2ffe5c 440
966aba65
JD
441** %code is now a permanent feature.
442
443 A traditional Yacc prologue directive is written in the form:
444
445 %{CODE%}
446
447 To provide a more flexible alternative, Bison 2.3b introduced the
448 %code directive with the following forms for C/C++:
449
450 %code {CODE}
451 %code requires {CODE}
452 %code provides {CODE}
453 %code top {CODE}
454
455 These forms are now considered permanent features of Bison. See the
456 %code entries in the section "Bison Declaration Summary" in the Bison
457 manual for a summary of their functionality. See the section
458 "Prologue Alternatives" for a detailed discussion including the
459 advantages of %code over the traditional Yacc prologue directive.
460
461 Bison's Java feature as a whole including its current usage of %code
462 is still considered experimental.
463
41d35e54
JD
464** YYFAIL is deprecated and will eventually be removed.
465
466 YYFAIL has existed for many years as an undocumented feature of
467 deterministic parsers in C generated by Bison. Previously, it was
468 documented for Bison's experimental Java parsers. YYFAIL is no longer
469 documented for Java parsers and is formally deprecated in both cases.
470 Users are strongly encouraged to migrate to YYERROR, which is
471 specified by POSIX.
472
473 Like YYERROR, you can invoke YYFAIL from a semantic action in order to
474 induce a syntax error. The most obvious difference from YYERROR is
475 that YYFAIL will automatically invoke yyerror to report the syntax
476 error so that you don't have to. However, there are several other
477 subtle differences between YYERROR and YYFAIL, and YYFAIL suffers from
478 inherent flaws when %error-verbose or `#define YYERROR_VERBOSE' is
479 used. For a more detailed discussion, see:
480
481 http://lists.gnu.org/archive/html/bison-patches/2009-12/msg00024.html
482
483 The upcoming Bison 2.5 will remove YYFAIL from Java parsers, but
484 deterministic parsers in C will continue to implement it. However,
485 because YYFAIL is already flawed, it seems futile to try to make new
486 Bison features compatible with it. Thus, during parser generation,
487 Bison 2.5 will produce a warning whenever it discovers YYFAIL in a
488 rule action. In a later release, YYFAIL will be disabled for
489 %error-verbose and `#define YYERROR_VERBOSE'. Eventually, YYFAIL will
490 be removed altogether.
491
492 There exists at least one case where Bison 2.5's YYFAIL warning will
493 be a false positive. Some projects add phony uses of YYFAIL and other
494 Bison-defined macros for the sole purpose of suppressing C
495 preprocessor warnings (from GCC cpp's -Wunused-macros, for example).
496 To avoid Bison's future warning, such YYFAIL uses can be moved to the
497 epilogue (that is, after the second `%%') in the Bison input file. In
498 this release (2.4.2), Bison already generates its own code to suppress
499 C preprocessor warnings for YYFAIL, so projects can remove their own
500 phony uses of YYFAIL if compatibility with Bison releases prior to
501 2.4.2 is not necessary.
502
dac8cc0d
AD
503** Internationalization.
504
505 Fix a regression introduced in Bison 2.4: Under some circumstances,
506 message translations were not installed although supported by the
507 host system.
508
41930e7a 509* Changes in version 2.4.1 (2008-12-11):
c9ba9e59 510
a957d06c
JD
511** In the GLR defines file, unexpanded M4 macros in the yylval and yylloc
512 declarations have been fixed.
513
738cde3e
AD
514** Temporary hack for adding a semicolon to the user action.
515
516 Bison used to prepend a trailing semicolon at the end of the user
517 action for reductions. This allowed actions such as
518
519 exp: exp "+" exp { $$ = $1 + $3 };
520
521 instead of
522
523 exp: exp "+" exp { $$ = $1 + $3; };
524
d07932ef
JD
525 Some grammars still depend on this `feature'. Bison 2.4.1 restores
526 the previous behavior in the case of C output (specifically, when
527 neither %language or %skeleton or equivalent command-line options
528 are used) to leave more time for grammars depending on the old
529 behavior to be adjusted. Future releases of Bison will disable this
530 feature.
a957d06c
JD
531
532** A few minor improvements to the Bison manual.
c9ba9e59 533
d6fb461d 534* Changes in version 2.4 (2008-11-02):
7bd1665a 535
d6fb461d 536** %language is an experimental feature.
ed4d67dc
JD
537
538 We first introduced this feature in test release 2.3b as a cleaner
539 alternative to %skeleton. Since then, we have discussed the possibility of
540 modifying its effect on Bison's output file names. Thus, in this release,
541 we consider %language to be an experimental feature that will likely evolve
542 in future releases.
7bd1665a 543
d6fb461d 544** Forward compatibility with GNU M4 has been improved.
241fda7a 545
d6fb461d 546** Several bugs in the C++ skeleton and the experimental Java skeleton have been
241fda7a
JD
547 fixed.
548
d6fb461d 549* Changes in version 2.3b (2008-05-27):
35fe0834 550
d6fb461d 551** The quotes around NAME that used to be required in the following directive
d9df47b6
JD
552 are now deprecated:
553
554 %define NAME "VALUE"
555
d6fb461d 556** The directive `%pure-parser' is now deprecated in favor of:
d9df47b6
JD
557
558 %define api.pure
559
560 which has the same effect except that Bison is more careful to warn about
561 unreasonable usage in the latter case.
562
d6fb461d 563** Push Parsing
c373bf8b
JD
564
565 Bison can now generate an LALR(1) parser in C with a push interface. That
ef1b4273 566 is, instead of invoking `yyparse', which pulls tokens from `yylex', you can
c373bf8b
JD
567 push one token at a time to the parser using `yypush_parse', which will
568 return to the caller after processing each token. By default, the push
569 interface is disabled. Either of the following directives will enable it:
570
571 %define api.push_pull "push" // Just push; does not require yylex.
572 %define api.push_pull "both" // Push and pull; requires yylex.
573
574 See the new section `A Push Parser' in the Bison manual for details.
575
59da312b
JD
576 The current push parsing interface is experimental and may evolve. More user
577 feedback will help to stabilize it.
578
d6fb461d 579** The -g and --graph options now output graphs in Graphviz DOT format,
8e55b3aa
JD
580 not VCG format. Like --graph, -g now also takes an optional FILE argument
581 and thus cannot be bundled with other short options.
c373bf8b 582
d6fb461d 583** Java
59da312b
JD
584
585 Bison can now generate an LALR(1) parser in Java. The skeleton is
586 `data/lalr1.java'. Consider using the new %language directive instead of
587 %skeleton to select it.
588
589 See the new section `Java Parsers' in the Bison manual for details.
590
591 The current Java interface is experimental and may evolve. More user
592 feedback will help to stabilize it.
593
d6fb461d 594** %language
59da312b
JD
595
596 This new directive specifies the programming language of the generated
d43f77e7
PB
597 parser, which can be C (the default), C++, or Java. Besides the skeleton
598 that Bison uses, the directive affects the names of the generated files if
599 the grammar file's name ends in ".y".
59da312b 600
d6fb461d 601** XML Automaton Report
59da312b
JD
602
603 Bison can now generate an XML report of the LALR(1) automaton using the new
604 `--xml' option. The current XML schema is experimental and may evolve. More
605 user feedback will help to stabilize it.
c373bf8b 606
d6fb461d 607** The grammar file may now specify the name of the parser header file using
c373bf8b
JD
608 %defines. For example:
609
610 %defines "parser.h"
611
d6fb461d 612** When reporting useless rules, useless nonterminals, and unused terminals,
d80fb37a
JD
613 Bison now employs the terms "useless in grammar" instead of "useless",
614 "useless in parser" instead of "never reduced", and "unused in grammar"
615 instead of "unused".
cff03fb2 616
d6fb461d 617** Unreachable State Removal
c373bf8b
JD
618
619 Previously, Bison sometimes generated parser tables containing unreachable
31984206
JD
620 states. A state can become unreachable during conflict resolution if Bison
621 disables a shift action leading to it from a predecessor state. Bison now:
75ad86ee
JD
622
623 1. Removes unreachable states.
624
625 2. Does not report any conflicts that appeared in unreachable states.
626 WARNING: As a result, you may need to update %expect and %expect-rr
627 directives in existing grammar files.
628
629 3. For any rule used only in such states, Bison now reports the rule as
cff03fb2 630 "useless in parser due to conflicts".
75ad86ee 631
31984206
JD
632 This feature can be disabled with the following directive:
633
634 %define lr.keep_unreachable_states
635
636 See the %define entry in the `Bison Declaration Summary' in the Bison manual
637 for further discussion.
638
d6fb461d 639** Lookahead Set Correction in the `.output' Report
b1cc23c4 640
c373bf8b 641 When instructed to generate a `.output' file including lookahead sets
88c78747
JD
642 (using `--report=lookahead', for example), Bison now prints each reduction's
643 lookahead set only next to the associated state's one item that (1) is
644 associated with the same rule as the reduction and (2) has its dot at the end
645 of its RHS. Previously, Bison also erroneously printed the lookahead set
646 next to all of the state's other items associated with the same rule. This
647 bug affected only the `.output' file and not the generated parser source
648 code.
649
d6fb461d 650** --report-file=FILE is a new option to override the default `.output' file
59da312b 651 name.
1bb2bd75 652
d6fb461d 653** The `=' that used to be required in the following directives is now
02975b9a
JD
654 deprecated:
655
656 %file-prefix "parser"
657 %name-prefix "c_"
658 %output "parser.c"
659
d6fb461d 660** An Alternative to `%{...%}' -- `%code QUALIFIER {CODE}'
c373bf8b
JD
661
662 Bison 2.3a provided a new set of directives as a more flexible alternative to
8e0a5e9e
JD
663 the traditional Yacc prologue blocks. Those have now been consolidated into
664 a single %code directive with an optional qualifier field, which identifies
665 the purpose of the code and thus the location(s) where Bison should generate
666 it:
667
16dc6a9e
JD
668 1. `%code {CODE}' replaces `%after-header {CODE}'
669 2. `%code requires {CODE}' replaces `%start-header {CODE}'
670 3. `%code provides {CODE}' replaces `%end-header {CODE}'
671 4. `%code top {CODE}' replaces `%before-header {CODE}'
8e0a5e9e 672
61fee93e
JD
673 See the %code entries in section `Bison Declaration Summary' in the Bison
674 manual for a summary of the new functionality. See the new section `Prologue
8e0a5e9e
JD
675 Alternatives' for a detailed discussion including the advantages of %code
676 over the traditional Yacc prologues.
677
678 The prologue alternatives are experimental. More user feedback will help to
679 determine whether they should become permanent features.
680
d6fb461d 681** Revised warning: unset or unused mid-rule values
17bd8a73
JD
682
683 Since Bison 2.2, Bison has warned about mid-rule values that are set but not
684 used within any of the actions of the parent rule. For example, Bison warns
685 about unused $2 in:
686
687 exp: '1' { $$ = 1; } '+' exp { $$ = $1 + $4; };
688
689 Now, Bison also warns about mid-rule values that are used but not set. For
690 example, Bison warns about unset $$ in the mid-rule action in:
691
692 exp: '1' { $1 = 1; } '+' exp { $$ = $2 + $4; };
693
694 However, Bison now disables both of these warnings by default since they
695 sometimes prove to be false alarms in existing grammars employing the Yacc
696 constructs $0 or $-N (where N is some positive integer).
697
59da312b 698 To enable these warnings, specify the option `--warnings=midrule-values' or
17bd8a73
JD
699 `-W', which is a synonym for `--warnings=all'.
700
d6fb461d 701** Default %destructor or %printer with `<*>' or `<>'
c373bf8b
JD
702
703 Bison now recognizes two separate kinds of default %destructor's and
12e35840
JD
704 %printer's:
705
706 1. Place `<*>' in a %destructor/%printer symbol list to define a default
707 %destructor/%printer for all grammar symbols for which you have formally
708 declared semantic type tags.
709
3ebecc24 710 2. Place `<>' in a %destructor/%printer symbol list to define a default
12e35840
JD
711 %destructor/%printer for all grammar symbols without declared semantic
712 type tags.
713
714 Bison no longer supports the `%symbol-default' notation from Bison 2.3a.
3ebecc24 715 `<*>' and `<>' combined achieve the same effect with one exception: Bison no
12e35840
JD
716 longer applies any %destructor to a mid-rule value if that mid-rule value is
717 not actually ever referenced using either $$ or $n in a semantic action.
718
85894313
JD
719 The default %destructor's and %printer's are experimental. More user
720 feedback will help to determine whether they should become permanent
721 features.
722
12e35840
JD
723 See the section `Freeing Discarded Symbols' in the Bison manual for further
724 details.
725
d6fb461d 726** %left, %right, and %nonassoc can now declare token numbers. This is required
ab7f29f8
JD
727 by POSIX. However, see the end of section `Operator Precedence' in the Bison
728 manual for a caveat concerning the treatment of literal strings.
729
d6fb461d 730** The nonfunctional --no-parser, -n, and %no-parser options have been
b1cc23c4
JD
731 completely removed from Bison.
732
d6fb461d 733* Changes in version 2.3a, 2006-09-13:
742e4900 734
d6fb461d 735** Instead of %union, you can define and use your own union type
ddc8ede1
PE
736 YYSTYPE if your grammar contains at least one <type> tag.
737 Your YYSTYPE need not be a macro; it can be a typedef.
738 This change is for compatibility with other Yacc implementations,
739 and is required by POSIX.
740
d6fb461d 741** Locations columns and lines start at 1.
cd48d21d
AD
742 In accordance with the GNU Coding Standards and Emacs.
743
d6fb461d 744** You may now declare per-type and default %destructor's and %printer's:
ec5479ce
JD
745
746 For example:
747
b2a0b7ca
JD
748 %union { char *string; }
749 %token <string> STRING1
750 %token <string> STRING2
751 %type <string> string1
752 %type <string> string2
753 %union { char character; }
754 %token <character> CHR
755 %type <character> chr
756 %destructor { free ($$); } %symbol-default
757 %destructor { free ($$); printf ("%d", @$.first_line); } STRING1 string1
758 %destructor { } <character>
759
760 guarantees that, when the parser discards any user-defined symbol that has a
761 semantic type tag other than `<character>', it passes its semantic value to
762 `free'. However, when the parser discards a `STRING1' or a `string1', it
763 also prints its line number to `stdout'. It performs only the second
764 `%destructor' in this case, so it invokes `free' only once.
ec5479ce 765
85894313
JD
766 [Although we failed to mention this here in the 2.3a release, the default
767 %destructor's and %printer's were experimental, and they were rewritten in
768 future versions.]
769
d6fb461d 770** Except for LALR(1) parsers in C with POSIX Yacc emulation enabled (with `-y',
b931235e
JD
771 `--yacc', or `%yacc'), Bison no longer generates #define statements for
772 associating token numbers with token names. Removing the #define statements
773 helps to sanitize the global namespace during preprocessing, but POSIX Yacc
774 requires them. Bison still generates an enum for token names in all cases.
775
d6fb461d 776** Handling of traditional Yacc prologue blocks is now more consistent but
34f98f46 777 potentially incompatible with previous releases of Bison.
9bc0dd67
JD
778
779 As before, you declare prologue blocks in your grammar file with the
780 `%{ ... %}' syntax. To generate the pre-prologue, Bison concatenates all
34f98f46
JD
781 prologue blocks that you've declared before the first %union. To generate
782 the post-prologue, Bison concatenates all prologue blocks that you've
ddc8ede1 783 declared after the first %union.
9bc0dd67 784
34f98f46 785 Previous releases of Bison inserted the pre-prologue into both the header
9bc0dd67
JD
786 file and the code file in all cases except for LALR(1) parsers in C. In the
787 latter case, Bison inserted it only into the code file. For parsers in C++,
788 the point of insertion was before any token definitions (which associate
789 token numbers with names). For parsers in C, the point of insertion was
790 after the token definitions.
791
792 Now, Bison never inserts the pre-prologue into the header file. In the code
793 file, it always inserts it before the token definitions.
794
d6fb461d 795** Bison now provides a more flexible alternative to the traditional Yacc
34f98f46
JD
796 prologue blocks: %before-header, %start-header, %end-header, and
797 %after-header.
798
799 For example, the following declaration order in the grammar file reflects the
800 order in which Bison will output these code blocks. However, you are free to
801 declare these code blocks in your grammar file in whatever order is most
802 convenient for you:
803
804 %before-header {
805 /* Bison treats this block like a pre-prologue block: it inserts it into
806 * the code file before the contents of the header file. It does *not*
807 * insert it into the header file. This is a good place to put
808 * #include's that you want at the top of your code file. A common
809 * example is `#include "system.h"'. */
810 }
811 %start-header {
812 /* Bison inserts this block into both the header file and the code file.
813 * In both files, the point of insertion is before any Bison-generated
814 * token, semantic type, location type, and class definitions. This is a
815 * good place to define %union dependencies, for example. */
9bc0dd67
JD
816 }
817 %union {
34f98f46
JD
818 /* Unlike the traditional Yacc prologue blocks, the output order for the
819 * new %*-header blocks is not affected by their declaration position
820 * relative to any %union in the grammar file. */
9bc0dd67 821 }
34f98f46
JD
822 %end-header {
823 /* Bison inserts this block into both the header file and the code file.
824 * In both files, the point of insertion is after the Bison-generated
825 * definitions. This is a good place to declare or define public
826 * functions or data structures that depend on the Bison-generated
827 * definitions. */
9bc0dd67 828 }
34f98f46
JD
829 %after-header {
830 /* Bison treats this block like a post-prologue block: it inserts it into
831 * the code file after the contents of the header file. It does *not*
832 * insert it into the header file. This is a good place to declare or
833 * define internal functions or data structures that depend on the
834 * Bison-generated definitions. */
835 }
836
837 If you have multiple occurrences of any one of the above declarations, Bison
838 will concatenate the contents in declaration order.
9bc0dd67 839
85894313
JD
840 [Although we failed to mention this here in the 2.3a release, the prologue
841 alternatives were experimental, and they were rewritten in future versions.]
842
d6fb461d 843** The option `--report=look-ahead' has been changed to `--report=lookahead'.
9e6e7ed2
PE
844 The old spelling still works, but is not documented and may be removed
845 in a future release.
742e4900 846
d6fb461d 847* Changes in version 2.3, 2006-06-05:
4ad3ed84 848
d6fb461d 849** GLR grammars should now use `YYRECOVERING ()' instead of `YYRECOVERING',
4ad3ed84
PE
850 for compatibility with LALR(1) grammars.
851
d6fb461d 852** It is now documented that any definition of YYSTYPE or YYLTYPE should
4ad3ed84
PE
853 be to a type name that does not contain parentheses or brackets.
854
d6fb461d 855* Changes in version 2.2, 2006-05-19:
193d7c70 856
d6fb461d 857** The distribution terms for all Bison-generated parsers now permit
193d7c70
PE
858 using the parsers in nonfree programs. Previously, this permission
859 was granted only for Bison-generated LALR(1) parsers in C.
5f4236a0 860
d6fb461d 861** %name-prefix changes the namespace name in C++ outputs.
aa08666d 862
d6fb461d 863** The C++ parsers export their token_type.
5f4236a0 864
d6fb461d 865** Bison now allows multiple %union declarations, and concatenates
d6ca7905
PE
866 their contents together.
867
d6fb461d 868** New warning: unused values
4d7bc38c
PE
869 Right-hand side symbols whose values are not used are reported,
870 if the symbols have destructors. For instance:
affac613 871
8f3596a6 872 exp: exp "?" exp ":" exp { $1 ? $1 : $3; }
721be13c
PE
873 | exp "+" exp
874 ;
affac613 875
8f3596a6
AD
876 will trigger a warning about $$ and $5 in the first rule, and $3 in
877 the second ($1 is copied to $$ by the default rule). This example
4e26c69e 878 most likely contains three errors, and could be rewritten as:
affac613 879
4e26c69e
PE
880 exp: exp "?" exp ":" exp
881 { $$ = $1 ? $3 : $5; free ($1 ? $5 : $3); free ($1); }
882 | exp "+" exp
883 { $$ = $1 ? $1 : $3; if ($1) free ($3); }
721be13c 884 ;
affac613 885
4e26c69e
PE
886 However, if the original actions were really intended, memory leaks
887 and all, the warnings can be suppressed by letting Bison believe the
888 values are used, e.g.:
721be13c 889
8f3596a6 890 exp: exp "?" exp ":" exp { $1 ? $1 : $3; (void) ($$, $5); }
721be13c
PE
891 | exp "+" exp { $$ = $1; (void) $3; }
892 ;
893
84866159
AD
894 If there are mid-rule actions, the warning is issued if no action
895 uses it. The following triggers no warning: $1 and $3 are used.
896
897 exp: exp { push ($1); } '+' exp { push ($3); sum (); };
898
721be13c
PE
899 The warning is intended to help catching lost values and memory leaks.
900 If a value is ignored, its associated memory typically is not reclaimed.
affac613 901
d6fb461d 902** %destructor vs. YYABORT, YYACCEPT, and YYERROR.
9d9b8b70
PE
903 Destructors are now called when user code invokes YYABORT, YYACCEPT,
904 and YYERROR, for all objects on the stack, other than objects
905 corresponding to the right-hand side of the current rule.
a85284cf 906
d6fb461d 907** %expect, %expect-rr
035aa4a0
PE
908 Incorrect numbers of expected conflicts are now actual errors,
909 instead of warnings.
910
d6fb461d 911** GLR, YACC parsers.
4e26c69e
PE
912 The %parse-params are available in the destructors (and the
913 experimental printers) as per the documentation.
4b367315 914
d6fb461d 915** Bison now warns if it finds a stray `$' or `@' in an action.
ad6a9b97 916
d6fb461d 917** %require "VERSION"
4e26c69e
PE
918 This specifies that the grammar file depends on features implemented
919 in Bison version VERSION or higher.
b50d2359 920
d6fb461d 921** lalr1.cc: The token and value types are now class members.
e14d0ab6
AD
922 The tokens were defined as free form enums and cpp macros. YYSTYPE
923 was defined as a free form union. They are now class members:
fb9712a9
AD
924 tokens are enumerations of the `yy::parser::token' struct, and the
925 semantic values have the `yy::parser::semantic_type' type.
926
927 If you do not want or can update to this scheme, the directive
928 `%define "global_tokens_and_yystype" "1"' triggers the global
b50d2359
AD
929 definition of tokens and YYSTYPE. This change is suitable both
930 for previous releases of Bison, and this one.
fb9712a9 931
b50d2359 932 If you wish to update, then make sure older version of Bison will
ab8d9dc5 933 fail using `%require "2.2"'.
fb9712a9 934
d6fb461d 935** DJGPP support added.
193d7c70 936\f
d6fb461d 937* Changes in version 2.1, 2005-09-16:
1ce59070 938
d6fb461d 939** The C++ lalr1.cc skeleton supports %lex-param.
e14d0ab6 940
d6fb461d 941** Bison-generated parsers now support the translation of diagnostics like
baf785db
PE
942 "syntax error" into languages other than English. The default
943 language is still English. For details, please see the new
0410a6e0
PE
944 Internationalization section of the Bison manual. Software
945 distributors should also see the new PACKAGING file. Thanks to
946 Bruno Haible for this new feature.
1ce59070 947
d6fb461d 948** Wording in the Bison-generated parsers has been changed slightly to
1a059451
PE
949 simplify translation. In particular, the message "memory exhausted"
950 has replaced "parser stack overflow", as the old message was not
951 always accurate for modern Bison-generated parsers.
952
d6fb461d 953** Destructors are now called when the parser aborts, for all symbols left
258b75ca
PE
954 behind on the stack. Also, the start symbol is now destroyed after a
955 successful parse. In both cases, the behavior was formerly inconsistent.
956
d6fb461d 957** When generating verbose diagnostics, Bison-generated parsers no longer
72f000b0
PE
958 quote the literal strings associated with tokens. For example, for
959 a syntax error associated with '%token NUM "number"' they might
960 print 'syntax error, unexpected number' instead of 'syntax error,
961 unexpected "number"'.
193d7c70 962\f
d6fb461d 963* Changes in version 2.0, 2004-12-25:
efeed023 964
d6fb461d 965** Possibly-incompatible changes
d7e14fc0 966
82de6b0d
PE
967 - Bison-generated parsers no longer default to using the alloca function
968 (when available) to extend the parser stack, due to widespread
969 problems in unchecked stack-overflow detection. You can "#define
970 YYSTACK_USE_ALLOCA 1" to require the use of alloca, but please read
971 the manual to determine safe values for YYMAXDEPTH in that case.
8dd162d3 972
82de6b0d
PE
973 - Error token location.
974 During error recovery, the location of the syntax error is updated
975 to cover the whole sequence covered by the error token: it includes
976 the shifted symbols thrown away during the first part of the error
977 recovery, and the lookahead rejected during the second part.
18d192f0 978
82de6b0d
PE
979 - Semicolon changes:
980 . Stray semicolons are no longer allowed at the start of a grammar.
981 . Semicolons are now required after in-grammar declarations.
e342c3be 982
82de6b0d
PE
983 - Unescaped newlines are no longer allowed in character constants or
984 string literals. They were never portable, and GCC 3.4.0 has
985 dropped support for them. Better diagnostics are now generated if
986 forget a closing quote.
8dd162d3 987
82de6b0d 988 - NUL bytes are no longer allowed in Bison string literals, unfortunately.
f74b6f91 989
d6fb461d 990** New features
1452af69 991
82de6b0d 992 - GLR grammars now support locations.
4febdd96 993
82de6b0d
PE
994 - New directive: %initial-action.
995 This directive allows the user to run arbitrary code (including
996 initializing @$) from yyparse before parsing starts.
1452af69 997
82de6b0d
PE
998 - A new directive "%expect-rr N" specifies the expected number of
999 reduce/reduce conflicts in GLR parsers.
1452af69 1000
82de6b0d
PE
1001 - %token numbers can now be hexadecimal integers, e.g., `%token FOO 0x12d'.
1002 This is a GNU extension.
4febdd96 1003
82de6b0d 1004 - The option `--report=lookahead' was changed to `--report=look-ahead'.
9e6e7ed2 1005 [However, this was changed back after 2.3.]
1452af69 1006
82de6b0d 1007 - Experimental %destructor support has been added to lalr1.cc.
1452af69 1008
82de6b0d
PE
1009 - New configure option --disable-yacc, to disable installation of the
1010 yacc command and -ly library introduced in 1.875 for POSIX conformance.
6040d338 1011
d6fb461d 1012** Bug fixes
d5a3fe37 1013
82de6b0d
PE
1014 - For now, %expect-count violations are now just warnings, not errors.
1015 This is for compatibility with Bison 1.75 and earlier (when there are
1016 reduce/reduce conflicts) and with Bison 1.30 and earlier (when there
1017 are too many or too few shift/reduce conflicts). However, in future
1018 versions of Bison we plan to improve the %expect machinery so that
1019 these violations will become errors again.
3473d0f8 1020
82de6b0d
PE
1021 - Within Bison itself, numbers (e.g., goto numbers) are no longer
1022 arbitrarily limited to 16-bit counts.
d600ee67 1023
82de6b0d 1024 - Semicolons are now allowed before "|" in grammar rules, as POSIX requires.
d600ee67 1025\f
d6fb461d 1026* Changes in version 1.875, 2003-01-01:
963fcc17 1027
d6fb461d 1028** The documentation license has been upgraded to version 1.2
dc546b0f 1029 of the GNU Free Documentation License.
75eb3bc4 1030
d6fb461d 1031** syntax error processing
75eb3bc4 1032
dc546b0f
PE
1033 - In Yacc-style parsers YYLLOC_DEFAULT is now used to compute error
1034 locations too. This fixes bugs in error-location computation.
75eb3bc4 1035
dc546b0f
PE
1036 - %destructor
1037 It is now possible to reclaim the memory associated to symbols
1038 discarded during error recovery. This feature is still experimental.
20daca06 1039
dc546b0f
PE
1040 - %error-verbose
1041 This new directive is preferred over YYERROR_VERBOSE.
74724a70 1042
dc546b0f
PE
1043 - #defining yyerror to steal internal variables is discouraged.
1044 It is not guaranteed to work forever.
d1de5372 1045
d6fb461d 1046** POSIX conformance
d1de5372 1047
dc546b0f
PE
1048 - Semicolons are once again optional at the end of grammar rules.
1049 This reverts to the behavior of Bison 1.33 and earlier, and improves
1050 compatibility with Yacc.
74724a70 1051
dc546b0f
PE
1052 - `parse error' -> `syntax error'
1053 Bison now uniformly uses the term `syntax error'; formerly, the code
1054 and manual sometimes used the term `parse error' instead. POSIX
1055 requires `syntax error' in diagnostics, and it was thought better to
1056 be consistent.
74724a70 1057
dc546b0f
PE
1058 - The documentation now emphasizes that yylex and yyerror must be
1059 declared before use. C99 requires this.
d1de5372 1060
dc546b0f
PE
1061 - Bison now parses C99 lexical constructs like UCNs and
1062 backslash-newline within C escape sequences, as POSIX 1003.1-2001 requires.
d1de5372 1063
dc546b0f
PE
1064 - File names are properly escaped in C output. E.g., foo\bar.y is
1065 output as "foo\\bar.y".
6780ca7a 1066
dc546b0f
PE
1067 - Yacc command and library now available
1068 The Bison distribution now installs a `yacc' command, as POSIX requires.
1069 Also, Bison now installs a small library liby.a containing
1070 implementations of Yacc-compatible yyerror and main functions.
1071 This library is normally not useful, but POSIX requires it.
6e649e65 1072
dc546b0f 1073 - Type clashes now generate warnings, not errors.
6e649e65 1074
dc546b0f
PE
1075 - If the user does not define YYSTYPE as a macro, Bison now declares it
1076 using typedef instead of defining it as a macro.
1077 For consistency, YYLTYPE is also declared instead of defined.
9501dc6e 1078
d6fb461d 1079** Other compatibility issues
886a425c 1080
dc546b0f
PE
1081 - %union directives can now have a tag before the `{', e.g., the
1082 directive `%union foo {...}' now generates the C code
1083 `typedef union foo { ... } YYSTYPE;'; this is for Yacc compatibility.
1084 The default union tag is `YYSTYPE', for compatibility with Solaris 9 Yacc.
1085 For consistency, YYLTYPE's struct tag is now `YYLTYPE' not `yyltype'.
1086 This is for compatibility with both Yacc and Bison 1.35.
72f889cc 1087
dc546b0f
PE
1088 - `;' is output before the terminating `}' of an action, for
1089 compatibility with Bison 1.35.
886a425c 1090
dc546b0f
PE
1091 - Bison now uses a Yacc-style format for conflict reports, e.g.,
1092 `conflicts: 2 shift/reduce, 1 reduce/reduce'.
437c2d80 1093
dc546b0f
PE
1094 - `yystype' and `yyltype' are now obsolescent macros instead of being
1095 typedefs or tags; they are no longer documented and are planned to be
1096 withdrawn in a future release.
2a8d363a 1097
d6fb461d 1098** GLR parser notes
2a8d363a 1099
dc546b0f
PE
1100 - GLR and inline
1101 Users of Bison have to decide how they handle the portability of the
1102 C keyword `inline'.
959e5f51 1103
dc546b0f
PE
1104 - `parsing stack overflow...' -> `parser stack overflow'
1105 GLR parsers now report `parser stack overflow' as per the Bison manual.
900c5db5 1106
d6fb461d 1107** Bison now warns if it detects conflicting outputs to the same file,
dc546b0f
PE
1108 e.g., it generates a warning for `bison -d -o foo.h foo.y' since
1109 that command outputs both code and header to foo.h.
6e40b4eb 1110
d6fb461d 1111** #line in output files
dc546b0f 1112 - --no-line works properly.
6e40b4eb 1113
d6fb461d 1114** Bison can no longer be built by a K&R C compiler; it requires C89 or
6e40b4eb
AD
1115 later to be built. This change originally took place a few versions
1116 ago, but nobody noticed until we recently asked someone to try
1117 building Bison with a K&R C compiler.
d600ee67 1118\f
d6fb461d 1119* Changes in version 1.75, 2002-10-14:
7933f2b5 1120
d6fb461d 1121** Bison should now work on 64-bit hosts.
7933f2b5 1122
d6fb461d 1123** Indonesian translation thanks to Tedi Heriyanto.
7933f2b5 1124
d6fb461d 1125** GLR parsers
f50adbbd
AD
1126 Fix spurious parse errors.
1127
d6fb461d 1128** Pure parsers
f50adbbd
AD
1129 Some people redefine yyerror to steal yyparse' private variables.
1130 Reenable this trick until an official feature replaces it.
1131
d6fb461d 1132** Type Clashes
d90c934c
AD
1133 In agreement with POSIX and with other Yaccs, leaving a default
1134 action is valid when $$ is untyped, and $1 typed:
1135
1136 untyped: ... typed;
1137
1138 but the converse remains an error:
1139
1140 typed: ... untyped;
1141
d6fb461d 1142** Values of mid-rule actions
d90c934c
AD
1143 The following code:
1144
1145 foo: { ... } { $$ = $1; } ...
1146
1147 was incorrectly rejected: $1 is defined in the second mid-rule
1148 action, and is equal to the $$ of the first mid-rule action.
d600ee67 1149\f
d6fb461d 1150* Changes in version 1.50, 2002-10-04:
adc8c848 1151
d6fb461d 1152** GLR parsing
676385e2
PH
1153 The declaration
1154 %glr-parser
1155 causes Bison to produce a Generalized LR (GLR) parser, capable of handling
1156 almost any context-free grammar, ambiguous or not. The new declarations
e8832397 1157 %dprec and %merge on grammar rules allow parse-time resolution of
676385e2
PH
1158 ambiguities. Contributed by Paul Hilfinger.
1159
7933f2b5 1160 Unfortunately Bison 1.50 does not work properly on 64-bit hosts
420f93c8
PE
1161 like the Alpha, so please stick to 32-bit hosts for now.
1162
d6fb461d 1163** Output Directory
8c165d89 1164 When not in Yacc compatibility mode, when the output file was not
e88dbdbf 1165 specified, running `bison foo/bar.y' created `foo/bar.c'. It
8c165d89
AD
1166 now creates `bar.c'.
1167
d6fb461d 1168** Undefined token
007a50a4 1169 The undefined token was systematically mapped to 2 which prevented
e88dbdbf 1170 the use of 2 by the user. This is no longer the case.
007a50a4 1171
d6fb461d 1172** Unknown token numbers
e88dbdbf 1173 If yylex returned an out of range value, yyparse could die. This is
007a50a4
AD
1174 no longer the case.
1175
d6fb461d 1176** Error token
e88dbdbf 1177 According to POSIX, the error token must be 256.
23c5a174
AD
1178 Bison extends this requirement by making it a preference: *if* the
1179 user specified that one of her tokens is numbered 256, then error
1180 will be mapped onto another number.
1181
d6fb461d 1182** Verbose error messages
e88dbdbf 1183 They no longer report `..., expecting error or...' for states where
217598da
AD
1184 error recovery is possible.
1185
d6fb461d 1186** End token
217598da
AD
1187 Defaults to `$end' instead of `$'.
1188
d6fb461d 1189** Error recovery now conforms to documentation and to POSIX
68cd8af3
PE
1190 When a Bison-generated parser encounters a syntax error, it now pops
1191 the stack until it finds a state that allows shifting the error
1192 token. Formerly, it popped the stack until it found a state that
1193 allowed some non-error action other than a default reduction on the
1194 error token. The new behavior has long been the documented behavior,
1195 and has long been required by POSIX. For more details, please see
337116ba
PE
1196 Paul Eggert, "Reductions during Bison error handling" (2002-05-20)
1197 <http://lists.gnu.org/archive/html/bug-bison/2002-05/msg00038.html>.
68cd8af3 1198
d6fb461d 1199** Traces
5504898e
AD
1200 Popped tokens and nonterminals are now reported.
1201
d6fb461d 1202** Larger grammars
a861a339
PE
1203 Larger grammars are now supported (larger token numbers, larger grammar
1204 size (= sum of the LHS and RHS lengths), larger LALR tables).
1205 Formerly, many of these numbers ran afoul of 16-bit limits;
1206 now these limits are 32 bits on most hosts.
355e7c1c 1207
d6fb461d 1208** Explicit initial rule
643a5994
AD
1209 Bison used to play hacks with the initial rule, which the user does
1210 not write. It is now explicit, and visible in the reports and
1211 graphs as rule 0.
23c5a174 1212
d6fb461d 1213** Useless rules
643a5994 1214 Before, Bison reported the useless rules, but, although not used,
77714df2 1215 included them in the parsers. They are now actually removed.
23c5a174 1216
d6fb461d 1217** Useless rules, useless nonterminals
6b98e4b5
AD
1218 They are now reported, as a warning, with their locations.
1219
d6fb461d 1220** Rules never reduced
e8832397
AD
1221 Rules that can never be reduced because of conflicts are now
1222 reported.
1223
d6fb461d 1224** Incorrect `Token not used'
11652ab3
AD
1225 On a grammar such as
1226
1227 %token useless useful
1228 %%
1229 exp: '0' %prec useful;
1230
1231 where a token was used to set the precedence of the last rule,
1232 bison reported both `useful' and `useless' as useless tokens.
1233
d6fb461d 1234** Revert the C++ namespace changes introduced in 1.31
77714df2 1235 as they caused too many portability hassles.
0179dd65 1236
d6fb461d 1237** Default locations
b2d52318
AD
1238 By an accident of design, the default computation of @$ was
1239 performed after another default computation was performed: @$ = @1.
1240 The latter is now removed: YYLLOC_DEFAULT is fully responsible of
1241 the computation of @$.
adc8c848 1242
d6fb461d 1243** Token end-of-file
b7c49edf
AD
1244 The token end of file may be specified by the user, in which case,
1245 the user symbol is used in the reports, the graphs, and the verbose
a861a339 1246 error messages instead of `$end', which remains being the default.
b7c49edf 1247 For instance
7bd6c77e 1248 %token MYEOF 0
b7c49edf 1249 or
7bd6c77e 1250 %token MYEOF 0 "end of file"
fdbcd8e2 1251
d6fb461d 1252** Semantic parser
fdbcd8e2
AD
1253 This old option, which has been broken for ages, is removed.
1254
d6fb461d 1255** New translations
a861a339 1256 Brazilian Portuguese, thanks to Alexandre Folle de Menezes.
84614e13
AD
1257 Croatian, thanks to Denis Lackovic.
1258
d6fb461d 1259** Incorrect token definitions
e88dbdbf 1260 When given `%token 'a' "A"', Bison used to output `#define 'a' 65'.
b87f8b21 1261
d6fb461d 1262** Token definitions as enums
77714df2
AD
1263 Tokens are output both as the traditional #define's, and, provided
1264 the compiler supports ANSI C or is a C++ compiler, as enums.
e88dbdbf 1265 This lets debuggers display names instead of integers.
77714df2 1266
d6fb461d 1267** Reports
ec3bc396
AD
1268 In addition to --verbose, bison supports --report=THINGS, which
1269 produces additional information:
b408954b
AD
1270 - itemset
1271 complete the core item sets with their closure
9e6e7ed2
PE
1272 - lookahead [changed to `look-ahead' in 1.875e through 2.3, but changed back]
1273 explicitly associate lookahead tokens to items
b408954b
AD
1274 - solved
1275 describe shift/reduce conflicts solving.
1276 Bison used to systematically output this information on top of
1277 the report. Solved conflicts are now attached to their states.
ec3bc396 1278
d6fb461d 1279** Type clashes
9af3fbce
AD
1280 Previous versions don't complain when there is a type clash on
1281 the default action if the rule has a mid-rule action, such as in:
1282
1283 %type <foo> bar
1284 %%
1285 bar: '0' {} '0';
1286
1287 This is fixed.
a861a339 1288
d6fb461d 1289** GNU M4 is now required when using Bison.
f987e9d2 1290\f
d6fb461d 1291* Changes in version 1.35, 2002-03-25:
76551463 1292
d6fb461d 1293** C Skeleton
76551463
AD
1294 Some projects use Bison's C parser with C++ compilers, and define
1295 YYSTYPE as a class. The recent adjustment of C parsers for data
1296 alignment and 64 bit architectures made this impossible.
1297
1298 Because for the time being no real solution for C++ parser
1299 generation exists, kludges were implemented in the parser to
1300 maintain this use. In the future, when Bison has C++ parsers, this
1301 kludge will be disabled.
1302
1303 This kludge also addresses some C++ problems when the stack was
1304 extended.
76551463 1305\f
d6fb461d 1306* Changes in version 1.34, 2002-03-12:
76551463 1307
d6fb461d 1308** File name clashes are detected
76551463
AD
1309 $ bison foo.y -d -o foo.x
1310 fatal error: header and parser would both be named `foo.x'
1311
d6fb461d 1312** A missing `;' at the end of a rule triggers a warning
76551463
AD
1313 In accordance with POSIX, and in agreement with other
1314 Yacc implementations, Bison will mandate this semicolon in the near
1315 future. This eases the implementation of a Bison parser of Bison
1316 grammars by making this grammar LALR(1) instead of LR(2). To
1317 facilitate the transition, this release introduces a warning.
1318
d6fb461d 1319** Revert the C++ namespace changes introduced in 1.31, as they caused too
76551463
AD
1320 many portability hassles.
1321
d6fb461d 1322** DJGPP support added.
76551463 1323
d6fb461d 1324** Fix test suite portability problems.
76551463 1325\f
d6fb461d 1326* Changes in version 1.33, 2002-02-07:
76551463 1327
d6fb461d 1328** Fix C++ issues
76551463
AD
1329 Groff could not be compiled for the definition of size_t was lacking
1330 under some conditions.
1331
d6fb461d 1332** Catch invalid @n
76551463
AD
1333 As is done with $n.
1334\f
d6fb461d 1335* Changes in version 1.32, 2002-01-23:
76551463 1336
d6fb461d 1337** Fix Yacc output file names
76551463 1338
d6fb461d 1339** Portability fixes
76551463 1340
d6fb461d 1341** Italian, Dutch translations
76551463 1342\f
d6fb461d 1343* Changes in version 1.31, 2002-01-14:
52d1aeee 1344
d6fb461d 1345** Many Bug Fixes
52d1aeee 1346
d6fb461d 1347** GNU Gettext and %expect
52d1aeee
MA
1348 GNU Gettext asserts 10 s/r conflicts, but there are 7. Now that
1349 Bison dies on incorrect %expectations, we fear there will be
1350 too many bug reports for Gettext, so _for the time being_, %expect
1351 does not trigger an error when the input file is named `plural.y'.
1352
d6fb461d 1353** Use of alloca in parsers
52d1aeee
MA
1354 If YYSTACK_USE_ALLOCA is defined to 0, then the parsers will use
1355 malloc exclusively. Since 1.29, but was not NEWS'ed.
1356
1357 alloca is used only when compiled with GCC, to avoid portability
1358 problems as on AIX.
1359
d6fb461d 1360** yyparse now returns 2 if memory is exhausted; formerly it dumped core.
b47dbebe 1361
d6fb461d 1362** When the generated parser lacks debugging code, YYDEBUG is now 0
52d1aeee
MA
1363 (as POSIX requires) instead of being undefined.
1364
d6fb461d 1365** User Actions
52d1aeee
MA
1366 Bison has always permitted actions such as { $$ = $1 }: it adds the
1367 ending semicolon. Now if in Yacc compatibility mode, the semicolon
1368 is no longer output: one has to write { $$ = $1; }.
1369
d6fb461d 1370** Better C++ compliance
52d1aeee 1371 The output parsers try to respect C++ namespaces.
76551463 1372 [This turned out to be a failed experiment, and it was reverted later.]
52d1aeee 1373
d6fb461d 1374** Reduced Grammars
52d1aeee
MA
1375 Fixed bugs when reporting useless nonterminals.
1376
d6fb461d 1377** 64 bit hosts
52d1aeee
MA
1378 The parsers work properly on 64 bit hosts.
1379
d6fb461d 1380** Error messages
52d1aeee
MA
1381 Some calls to strerror resulted in scrambled or missing error messages.
1382
d6fb461d 1383** %expect
52d1aeee
MA
1384 When the number of shift/reduce conflicts is correct, don't issue
1385 any warning.
1386
d6fb461d 1387** The verbose report includes the rule line numbers.
52d1aeee 1388
d6fb461d 1389** Rule line numbers are fixed in traces.
52d1aeee 1390
d6fb461d 1391** Swedish translation
52d1aeee 1392
d6fb461d 1393** Parse errors
52d1aeee
MA
1394 Verbose parse error messages from the parsers are better looking.
1395 Before: parse error: unexpected `'/'', expecting `"number"' or `'-'' or `'(''
1396 Now: parse error: unexpected '/', expecting "number" or '-' or '('
1397
d6fb461d 1398** Fixed parser memory leaks.
52d1aeee
MA
1399 When the generated parser was using malloc to extend its stacks, the
1400 previous allocations were not freed.
1401
d6fb461d 1402** Fixed verbose output file.
52d1aeee
MA
1403 Some newlines were missing.
1404 Some conflicts in state descriptions were missing.
1405
d6fb461d 1406** Fixed conflict report.
52d1aeee
MA
1407 Option -v was needed to get the result.
1408
d6fb461d 1409** %expect
52d1aeee
MA
1410 Was not used.
1411 Mismatches are errors, not warnings.
1412
d6fb461d 1413** Fixed incorrect processing of some invalid input.
52d1aeee 1414
d6fb461d 1415** Fixed CPP guards: 9foo.h uses BISON_9FOO_H instead of 9FOO_H.
52d1aeee 1416
d6fb461d 1417** Fixed some typos in the documentation.
52d1aeee 1418
d6fb461d 1419** %token MY_EOF 0 is supported.
52d1aeee
MA
1420 Before, MY_EOF was silently renumbered as 257.
1421
d6fb461d 1422** doc/refcard.tex is updated.
52d1aeee 1423
d6fb461d 1424** %output, %file-prefix, %name-prefix.
52d1aeee
MA
1425 New.
1426
d6fb461d 1427** --output
52d1aeee
MA
1428 New, aliasing `--output-file'.
1429\f
d6fb461d 1430* Changes in version 1.30, 2001-10-26:
342b8b6e 1431
d6fb461d 1432** `--defines' and `--graph' have now an optional argument which is the
fdac0091 1433 output file name. `-d' and `-g' do not change; they do not take any
342b8b6e
AD
1434 argument.
1435
d6fb461d 1436** `%source_extension' and `%header_extension' are removed, failed
342b8b6e
AD
1437 experiment.
1438
d6fb461d 1439** Portability fixes.
f987e9d2 1440\f
d6fb461d 1441* Changes in version 1.29, 2001-09-07:
342b8b6e 1442
d6fb461d 1443** The output file does not define const, as this caused problems when used
342b8b6e
AD
1444 with common autoconfiguration schemes. If you still use ancient compilers
1445 that lack const, compile with the equivalent of the C compiler option
1446 `-Dconst='. autoconf's AC_C_CONST macro provides one way to do this.
1447
d6fb461d 1448** Added `-g' and `--graph'.
f87a2205 1449
d6fb461d 1450** The Bison manual is now distributed under the terms of the GNU FDL.
f2b5126e 1451
d6fb461d 1452** The input and the output files has automatically a similar extension.
234a3be3 1453
d6fb461d 1454** Russian translation added.
f87a2205 1455
d6fb461d 1456** NLS support updated; should hopefully be less troublesome.
f87a2205 1457
d6fb461d 1458** Added the old Bison reference card.
c33638bb 1459
d6fb461d 1460** Added `--locations' and `%locations'.
6deb4447 1461
d6fb461d 1462** Added `-S' and `--skeleton'.
cd5bd6ac 1463
d6fb461d 1464** `%raw', `-r', `--raw' is disabled.
62ab6972 1465
d6fb461d 1466** Special characters are escaped when output. This solves the problems
cd5bd6ac
AD
1467 of the #line lines with path names including backslashes.
1468
d6fb461d 1469** New directives.
4ecbf796
MA
1470 `%yacc', `%fixed_output_files', `%defines', `%no_parser', `%verbose',
1471 `%debug', `%source_extension' and `%header_extension'.
f987e9d2 1472
d6fb461d 1473** @$
f987e9d2 1474 Automatic location tracking.
f87a2205 1475\f
d6fb461d 1476* Changes in version 1.28, 1999-07-06:
d2e00347 1477
d6fb461d 1478** Should compile better now with K&R compilers.
d2e00347 1479
d6fb461d 1480** Added NLS.
d2e00347 1481
d6fb461d 1482** Fixed a problem with escaping the double quote character.
d2e00347 1483
d6fb461d 1484** There is now a FAQ.
d2e00347 1485\f
d6fb461d 1486* Changes in version 1.27:
5c31c3c2 1487
d6fb461d 1488** The make rule which prevented bison.simple from being created on
5c31c3c2
JT
1489 some systems has been fixed.
1490\f
d6fb461d 1491* Changes in version 1.26:
4be07551 1492
d6fb461d 1493** Bison now uses automake.
4be07551 1494
d6fb461d 1495** New mailing lists: <bug-bison@gnu.org> and <help-bison@gnu.org>.
4be07551 1496
d6fb461d 1497** Token numbers now start at 257 as previously documented, not 258.
4be07551 1498
d6fb461d 1499** Bison honors the TMPDIR environment variable.
4be07551 1500
d6fb461d 1501** A couple of buffer overruns have been fixed.
f51dbca1 1502
d6fb461d 1503** Problems when closing files should now be reported.
f51dbca1 1504
d6fb461d 1505** Generated parsers should now work even on operating systems which do
f51dbca1 1506 not provide alloca().
4be07551 1507\f
d6fb461d 1508* Changes in version 1.25, 1995-10-16:
df8878c5 1509
d6fb461d 1510** Errors in the input grammar are not fatal; Bison keeps reading
df8878c5 1511the grammar file, and reports all the errors found in it.
8c44d3ec 1512
d6fb461d 1513** Tokens can now be specified as multiple-character strings: for
df8878c5
RS
1514example, you could use "<=" for a token which looks like <=, instead
1515of chosing a name like LESSEQ.
1516
d6fb461d 1517** The %token_table declaration says to write a table of tokens (names
df8878c5
RS
1518and numbers) into the parser file. The yylex function can use this
1519table to recognize multiple-character string tokens, or for other
1520purposes.
1521
d6fb461d 1522** The %no_lines declaration says not to generate any #line preprocessor
df8878c5
RS
1523directives in the parser file.
1524
d6fb461d 1525** The %raw declaration says to use internal Bison token numbers, not
df8878c5
RS
1526Yacc-compatible token numbers, when token names are defined as macros.
1527
d6fb461d 1528** The --no-parser option produces the parser tables without including
df8878c5
RS
1529the parser engine; a project can now use its own parser engine.
1530The actions go into a separate file called NAME.act, in the form of
1531a switch statement body.
1532\f
d6fb461d 1533* Changes in version 1.23:
6780ca7a 1534
4d019228
DM
1535The user can define YYPARSE_PARAM as the name of an argument to be
1536passed into yyparse. The argument should have type void *. It should
1537actually point to an object. Grammar actions can access the variable
1538by casting it to the proper pointer type.
6780ca7a 1539
6780ca7a 1540Line numbers in output file corrected.
6780ca7a 1541\f
d6fb461d 1542* Changes in version 1.22:
6780ca7a
DM
1543
1544--help option added.
6780ca7a 1545\f
d6fb461d 1546* Changes in version 1.20:
6780ca7a
DM
1547
1548Output file does not redefine const for C++.
9f4503d6
AD
1549
1550Local Variables:
1551mode: outline
1552End:
76551463
AD
1553
1554-----
1555
c932d613 1556Copyright (C) 1995-2012 Free Software Foundation, Inc.
76551463 1557
8defe11b 1558This file is part of Bison, the GNU Parser Generator.
76551463 1559
f16b0819 1560This program is free software: you can redistribute it and/or modify
76551463 1561it under the terms of the GNU General Public License as published by
f16b0819
PE
1562the Free Software Foundation, either version 3 of the License, or
1563(at your option) any later version.
76551463 1564
f16b0819 1565This program is distributed in the hope that it will be useful,
76551463
AD
1566but WITHOUT ANY WARRANTY; without even the implied warranty of
1567MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
1568GNU General Public License for more details.
1569
1570You should have received a copy of the GNU General Public License
f16b0819 1571along with this program. If not, see <http://www.gnu.org/licenses/>.