]> git.saurik.com Git - bison.git/blame - NEWS
%printer: support both yyo and yyoutput.
[bison.git] / NEWS
CommitLineData
6780ca7a
DM
1Bison News
2----------
3af4feb2 3
90dcd415
JD
4* Changes in version 2.5.1 (????-??-??):
5
debe2c03 6** Future changes:
041308d0 7
debe2c03 8 The next major release will drop support for K&R C.
ef51bfa7 9
466b4cf2 10** yacc.c: YYBACKUP works as expected.
94556574 11
d834eca0
AD
12** glr.c improvements:
13
14*** Location support is eliminated when not requested:
15
16 GLR parsers used to include location-related code even when
17 locations were not requested, and therefore not even usable.
18
19*** __attribute__ is preserved:
debe2c03
AD
20
21 __attribute__ is no longer disabled when __STRICT_ANSI__ is defined
22 (i.e., when -std is passed to GCC).
23
466b4cf2 24** lalr1.java: several fixes:
041308d0
AD
25
26 The Java parser no longer throws ArrayIndexOutOfBoundsException if
27 the first token leads to a syntax error. Some minor clean ups.
28
ef51bfa7
AD
29** C++11 compatibility:
30
31 C and C++ parsers use nullptr instead of 0 when __cplusplus is
32 201103L or higher.
33
936c88d1
AD
34** C++ locations:
35
36 The position and location constructors (and their initialize
37 methods) accept new arguments for line and column. Several issues
38 in the documentation were fixed.
39
466b4cf2
AD
40** liby is no longer asking for "rpl_fprintf" on some platforms.
41
42** Several improvements have been made to the manual:
43
44 The layout for grammar excerpts was changed to a more compact
45 scheme. Named references are motivated. The description of the
46 automaton description file (*.output) is updated to the current
47 format. Incorrect index entries were fixed. Some other errors were
48 fixed.
49
50** Warnings during the build procedure have been eliminated.
51
52** Several portability problems in the test suite have been fixed:
53
54 This includes warnings with some compilers, unexpected behavior of
55 tools such as diff, warning messages from the test suite itself,
56 etc.
57
b2bc8a1b 58* Changes in version 2.5 (2011-05-14):
006faedf 59
eb8c66bb
JD
60** Grammar symbol names can now contain non-initial dashes:
61
62 Consistently with directives (such as %error-verbose) and with
63 %define variables (e.g. push-pull), grammar symbol names may contain
64 dashes in any position except the beginning. This is a GNU
65 extension over POSIX Yacc. Thus, use of this extension is reported
66 by -Wyacc and rejected in Yacc mode (--yacc).
67
3b1977ea 68** Named references:
348f5608
AR
69
70 Historically, Yacc and Bison have supported positional references
71 ($n, $$) to allow access to symbol values from inside of semantic
72 actions code.
73
74 Starting from this version, Bison can also accept named references.
75 When no ambiguity is possible, original symbol names may be used
76 as named references:
77
ac3297d5 78 if_stmt : "if" cond_expr "then" then_stmt ';'
348f5608
AR
79 { $if_stmt = mk_if_stmt($cond_expr, $then_stmt); }
80
81 In the more common case, explicit names may be declared:
82
ac3297d5 83 stmt[res] : "if" expr[cond] "then" stmt[then] "else" stmt[else] ';'
348f5608
AR
84 { $res = mk_if_stmt($cond, $then, $else); }
85
f840c05a 86 Location information is also accessible using @name syntax. When
348f5608
AR
87 accessing symbol names containing dots or dashes, explicit bracketing
88 ($[sym.1]) must be used.
89
f840c05a 90 These features are experimental in this version. More user feedback
348f5608
AR
91 will help to stabilize them.
92
3b1977ea 93** IELR(1) and canonical LR(1):
34a6c2d1
JD
94
95 IELR(1) is a minimal LR(1) parser table generation algorithm. That
96 is, given any context-free grammar, IELR(1) generates parser tables
723206df 97 with the full language-recognition power of canonical LR(1) but with
3b1977ea
JD
98 nearly the same number of parser states as LALR(1). This reduction
99 in parser states is often an order of magnitude. More importantly,
34a6c2d1
JD
100 because canonical LR(1)'s extra parser states may contain duplicate
101 conflicts in the case of non-LR(1) grammars, the number of conflicts
102 for IELR(1) is often an order of magnitude less as well. This can
103 significantly reduce the complexity of developing of a grammar.
104
105 Bison can now generate IELR(1) and canonical LR(1) parser tables in
106 place of its traditional LALR(1) parser tables, which remain the
107 default. You can specify the type of parser tables in the grammar
108 file with these directives:
109
f37495f6
JD
110 %define lr.type lalr
111 %define lr.type ielr
112 %define lr.type canonical-lr
34a6c2d1 113
6f04ee6c
JD
114 The default-reduction optimization in the parser tables can also be
115 adjusted using `%define lr.default-reductions'. For details on both
116 of these features, see the new section `Tuning LR' in the Bison
117 manual.
34a6c2d1
JD
118
119 These features are experimental. More user feedback will help to
120 stabilize them.
121
6f04ee6c 122** LAC (Lookahead Correction) for syntax error handling:
4c38b19e
JD
123
124 Canonical LR, IELR, and LALR can suffer from a couple of problems
125 upon encountering a syntax error. First, the parser might perform
126 additional parser stack reductions before discovering the syntax
6f04ee6c 127 error. Such reductions can perform user semantic actions that are
4c38b19e
JD
128 unexpected because they are based on an invalid token, and they
129 cause error recovery to begin in a different syntactic context than
130 the one in which the invalid token was encountered. Second, when
6f04ee6c
JD
131 verbose error messages are enabled (with %error-verbose or the
132 obsolete `#define YYERROR_VERBOSE'), the expected token list in the
133 syntax error message can both contain invalid tokens and omit valid
134 tokens.
4c38b19e
JD
135
136 The culprits for the above problems are %nonassoc, default
137 reductions in inconsistent states, and parser state merging. Thus,
138 IELR and LALR suffer the most. Canonical LR can suffer only if
139 %nonassoc is used or if default reductions are enabled for
140 inconsistent states.
141
6f04ee6c
JD
142 LAC is a new mechanism within the parsing algorithm that solves
143 these problems for canonical LR, IELR, and LALR without sacrificing
144 %nonassoc, default reductions, or state merging. When LAC is in
145 use, canonical LR and IELR behave almost exactly the same for both
146 syntactically acceptable and syntactically unacceptable input.
4c38b19e
JD
147 While LALR still does not support the full language-recognition
148 power of canonical LR and IELR, LAC at least enables LALR's syntax
149 error handling to correctly reflect LALR's language-recognition
150 power.
151
152 Currently, LAC is only supported for deterministic parsers in C.
153 You can enable LAC with the following directive:
154
155 %define parse.lac full
156
6f04ee6c
JD
157 See the new section `LAC' in the Bison manual for additional
158 details including a few caveats.
4c38b19e
JD
159
160 LAC is an experimental feature. More user feedback will help to
161 stabilize it.
162
6ee881f6 163** %define improvements:
f37495f6 164
3b1977ea 165*** Can now be invoked via the command line:
4ecd3681 166
34d41938 167 Each of these command-line options
4ecd3681 168
34d41938
JD
169 -D NAME[=VALUE]
170 --define=NAME[=VALUE]
171
172 -F NAME[=VALUE]
173 --force-define=NAME[=VALUE]
4ecd3681
JD
174
175 is equivalent to this grammar file declaration
176
34d41938 177 %define NAME ["VALUE"]
4ecd3681 178
34d41938
JD
179 except that the manner in which Bison processes multiple definitions
180 for the same NAME differs. Most importantly, -F and --force-define
181 quietly override %define, but -D and --define do not. For further
ac3297d5 182 details, see the section `Bison Options' in the Bison manual.
4ecd3681 183
3b1977ea 184*** Variables renamed:
812775a0
JD
185
186 The following %define variables
187
188 api.push_pull
189 lr.keep_unreachable_states
190
191 have been renamed to
192
193 api.push-pull
194 lr.keep-unreachable-states
195
196 The old names are now deprecated but will be maintained indefinitely
197 for backward compatibility.
198
723206df 199*** Values no longer need to be quoted in the grammar file:
f37495f6
JD
200
201 If a %define value is an identifier, it no longer needs to be placed
202 within quotations marks. For example,
203
204 %define api.push-pull "push"
205
206 can be rewritten as
207
208 %define api.push-pull push
209
6ee881f6 210*** Unrecognized variables are now errors not warnings.
c046698e 211
6ee881f6
JD
212*** Multiple invocations for any variable is now an error not a warning.
213
214** Unrecognized %code qualifiers are now errors not warnings.
215
216** Character literals not of length one:
217
218 Previously, Bison quietly converted all character literals to length
219 one. For example, without warning, Bison interpreted the operators in
220 the following grammar to be the same token:
221
222 exp: exp '++'
223 | exp '+' exp
224 ;
225
226 Bison now warns when a character literal is not of length one. In
227 some future release, Bison will start reporting an error instead.
228
229** Destructor calls fixed for lookaheads altered in semantic actions:
230
231 Previously for deterministic parsers in C, if a user semantic action
232 altered yychar, the parser in some cases used the old yychar value to
233 determine which destructor to call for the lookahead upon a syntax
234 error or upon parser return. This bug has been fixed.
235
236** C++ parsers use YYRHSLOC:
237
238 Similarly to the C parsers, the C++ parsers now define the YYRHSLOC
239 macro and use it in the default YYLLOC_DEFAULT. You are encouraged
ac3297d5
JD
240 to use it. If, for instance, your location structure has `first'
241 and `last' members, instead of
6ee881f6
JD
242
243 # define YYLLOC_DEFAULT(Current, Rhs, N) \
244 do \
245 if (N) \
246 { \
247 (Current).first = (Rhs)[1].location.first; \
248 (Current).last = (Rhs)[N].location.last; \
249 } \
250 else \
251 { \
252 (Current).first = (Current).last = (Rhs)[0].location.last; \
253 } \
254 while (false)
255
256 use:
257
258 # define YYLLOC_DEFAULT(Current, Rhs, N) \
259 do \
260 if (N) \
261 { \
262 (Current).first = YYRHSLOC (Rhs, 1).first; \
263 (Current).last = YYRHSLOC (Rhs, N).last; \
264 } \
265 else \
266 { \
267 (Current).first = (Current).last = YYRHSLOC (Rhs, 0).last; \
268 } \
269 while (false)
270
271** YYLLOC_DEFAULT in C++:
272
273 The default implementation of YYLLOC_DEFAULT used to be issued in
274 the header file. It is now output in the implementation file, after
275 the user %code sections so that its #ifndef guard does not try to
276 override the user's YYLLOC_DEFAULT if provided.
c046698e 277
3b1977ea 278** YYFAIL now produces warnings and Java parsers no longer implement it:
62efdd2a
JD
279
280 YYFAIL has existed for many years as an undocumented feature of
281 deterministic parsers in C generated by Bison. More recently, it was
282 a documented feature of Bison's experimental Java parsers. As
283 promised in Bison 2.4.2's NEWS entry, any appearance of YYFAIL in a
284 semantic action now produces a deprecation warning, and Java parsers
285 no longer implement YYFAIL at all. For further details, including a
286 discussion of how to suppress C preprocessor warnings about YYFAIL
287 being unused, see the Bison 2.4.2 NEWS entry.
288
3b1977ea 289** Temporary hack for adding a semicolon to the user action:
a2d05674
JD
290
291 Previously, Bison appended a semicolon to every user action for
292 reductions when the output language defaulted to C (specifically, when
293 neither %yacc, %language, %skeleton, or equivalent command-line
294 options were specified). This allowed actions such as
295
296 exp: exp "+" exp { $$ = $1 + $3 };
297
298 instead of
299
300 exp: exp "+" exp { $$ = $1 + $3; };
301
302 As a first step in removing this misfeature, Bison now issues a
303 warning when it appends a semicolon. Moreover, in cases where Bison
304 cannot easily determine whether a semicolon is needed (for example, an
305 action ending with a cpp directive or a braced compound initializer),
306 it no longer appends one. Thus, the C compiler might now complain
307 about a missing semicolon where it did not before. Future releases of
308 Bison will cease to append semicolons entirely.
309
095a1d11
JD
310** Verbose syntax error message fixes:
311
6f04ee6c
JD
312 When %error-verbose or the obsolete `#define YYERROR_VERBOSE' is
313 specified, syntax error messages produced by the generated parser
314 include the unexpected token as well as a list of expected tokens.
315 The effect of %nonassoc on these verbose messages has been corrected
316 in two ways, but a more complete fix requires LAC, described above:
095a1d11
JD
317
318*** When %nonassoc is used, there can exist parser states that accept no
319 tokens, and so the parser does not always require a lookahead token
320 in order to detect a syntax error. Because no unexpected token or
321 expected tokens can then be reported, the verbose syntax error
322 message described above is suppressed, and the parser instead
ac3297d5 323 reports the simpler message, `syntax error'. Previously, this
095a1d11
JD
324 suppression was sometimes erroneously triggered by %nonassoc when a
325 lookahead was actually required. Now verbose messages are
326 suppressed only when all previous lookaheads have already been
327 shifted or discarded.
328
329*** Previously, the list of expected tokens erroneously included tokens
330 that would actually induce a syntax error because conflicts for them
331 were resolved with %nonassoc in the current parser state. Such
332 tokens are now properly omitted from the list.
333
334*** Expected token lists are still often wrong due to state merging
4c38b19e
JD
335 (from LALR or IELR) and default reductions, which can both add
336 invalid tokens and subtract valid tokens. Canonical LR almost
337 completely fixes this problem by eliminating state merging and
338 default reductions. However, there is one minor problem left even
339 when using canonical LR and even after the fixes above. That is,
340 if the resolution of a conflict with %nonassoc appears in a later
341 parser state than the one at which some syntax error is
342 discovered, the conflicted token is still erroneously included in
343 the expected token list. Bison's new LAC implementation,
344 described above, eliminates this problem and the need for
345 canonical LR. However, LAC is still experimental and is disabled
346 by default.
1fa30307 347
4c2a6e42
JD
348** Java skeleton fixes:
349
350*** A location handling bug has been fixed.
351
352*** The top element of each of the value stack and location stack is now
353 cleared when popped so that it can be garbage collected.
77768165 354
f0e2c228
JD
355*** Parser traces now print the top element of the stack.
356
248dba53
JD
357** -W/--warnings fixes:
358
ac3297d5 359*** Bison now properly recognizes the `no-' versions of categories:
248dba53
JD
360
361 For example, given the following command line, Bison now enables all
362 warnings except warnings for incompatibilities with POSIX Yacc:
363
364 bison -Wall,no-yacc gram.y
365
6f8bdce2
JD
366*** Bison now treats S/R and R/R conflicts like other warnings:
367
368 Previously, conflict reports were independent of Bison's normal
369 warning system. Now, Bison recognizes the warning categories
ac3297d5 370 `conflicts-sr' and `conflicts-rr'. This change has important
6f8bdce2
JD
371 consequences for the -W and --warnings command-line options. For
372 example:
373
374 bison -Wno-conflicts-sr gram.y # S/R conflicts not reported
375 bison -Wno-conflicts-rr gram.y # R/R conflicts not reported
376 bison -Wnone gram.y # no conflicts are reported
377 bison -Werror gram.y # any conflict is an error
378
379 However, as before, if the %expect or %expect-rr directive is
380 specified, an unexpected number of conflicts is an error, and an
381 expected number of conflicts is not reported, so -W and --warning
382 then have no effect on the conflict report.
383
ac3297d5 384*** The `none' category no longer disables a preceding `error':
dab96632
JD
385
386 For example, for the following command line, Bison now reports
387 errors instead of warnings for incompatibilities with POSIX Yacc:
388
389 bison -Werror,none,yacc gram.y
390
b2bc8a1b 391*** The `none' category now disables all Bison warnings:
8ffd7912 392
ac3297d5 393 Previously, the `none' category disabled only Bison warnings for
8ffd7912
JD
394 which there existed a specific -W/--warning category. However,
395 given the following command line, Bison is now guaranteed to
396 suppress all warnings:
397
398 bison -Wnone gram.y
399
9d6af153
JD
400** Precedence directives can now assign token number 0:
401
402 Since Bison 2.3b, which restored the ability of precedence
403 directives to assign token numbers, doing so for token number 0 has
404 produced an assertion failure. For example:
405
406 %left END 0
407
408 This bug has been fixed.
409
7527c744 410* Changes in version 2.4.3 (2010-08-05):
7a9c3cb3 411
4631c34f
JD
412** Bison now obeys -Werror and --warnings=error for warnings about
413 grammar rules that are useless in the parser due to conflicts.
414
7a9c3cb3
JD
415** Problems with spawning M4 on at least FreeBSD 8 and FreeBSD 9 have
416 been fixed.
417
132247cd
JD
418** Failures in the test suite for GCC 4.5 have been fixed.
419
88bb35d6
JD
420** Failures in the test suite for some versions of Sun Studio C++ have
421 been fixed.
422
43fdc9fd
JD
423** Contrary to Bison 2.4.2's NEWS entry, it has been decided that
424 warnings about undefined %prec identifiers will not be converted to
425 errors in Bison 2.5. They will remain warnings, which should be
426 sufficient for POSIX while avoiding backward compatibility issues.
427
cf22447c
JD
428** Minor documentation fixes.
429
ea66d039 430* Changes in version 2.4.2 (2010-03-20):
8defe11b 431
47fa5747
JD
432** Some portability problems that resulted in failures and livelocks
433 in the test suite on some versions of at least Solaris, AIX, HP-UX,
ea66d039
JD
434 RHEL4, and Tru64 have been addressed. As a result, fatal Bison
435 errors should no longer cause M4 to report a broken pipe on the
47fa5747
JD
436 affected platforms.
437
2c203528
JD
438** `%prec IDENTIFIER' requires IDENTIFIER to be defined separately.
439
440 POSIX specifies that an error be reported for any identifier that does
441 not appear on the LHS of a grammar rule and that is not defined by
442 %token, %left, %right, or %nonassoc. Bison 2.3b and later lost this
443 error report for the case when an identifier appears only after a
444 %prec directive. It is now restored. However, for backward
445 compatibility with recent Bison releases, it is only a warning for
446 now. In Bison 2.5 and later, it will return to being an error.
43fdc9fd
JD
447 [Between the 2.4.2 and 2.4.3 releases, it was decided that this
448 warning will not be converted to an error in Bison 2.5.]
2c203528 449
c5196098
EB
450** Detection of GNU M4 1.4.6 or newer during configure is improved.
451
34731471
JD
452** Warnings from gcc's -Wundef option about undefined YYENABLE_NLS,
453 YYLTYPE_IS_TRIVIAL, and __STRICT_ANSI__ in C/C++ parsers are now
454 avoided.
af2ffe5c 455
966aba65
JD
456** %code is now a permanent feature.
457
458 A traditional Yacc prologue directive is written in the form:
459
460 %{CODE%}
461
462 To provide a more flexible alternative, Bison 2.3b introduced the
463 %code directive with the following forms for C/C++:
464
465 %code {CODE}
466 %code requires {CODE}
467 %code provides {CODE}
468 %code top {CODE}
469
470 These forms are now considered permanent features of Bison. See the
471 %code entries in the section "Bison Declaration Summary" in the Bison
472 manual for a summary of their functionality. See the section
473 "Prologue Alternatives" for a detailed discussion including the
474 advantages of %code over the traditional Yacc prologue directive.
475
476 Bison's Java feature as a whole including its current usage of %code
477 is still considered experimental.
478
41d35e54
JD
479** YYFAIL is deprecated and will eventually be removed.
480
481 YYFAIL has existed for many years as an undocumented feature of
482 deterministic parsers in C generated by Bison. Previously, it was
483 documented for Bison's experimental Java parsers. YYFAIL is no longer
484 documented for Java parsers and is formally deprecated in both cases.
485 Users are strongly encouraged to migrate to YYERROR, which is
486 specified by POSIX.
487
488 Like YYERROR, you can invoke YYFAIL from a semantic action in order to
489 induce a syntax error. The most obvious difference from YYERROR is
490 that YYFAIL will automatically invoke yyerror to report the syntax
491 error so that you don't have to. However, there are several other
492 subtle differences between YYERROR and YYFAIL, and YYFAIL suffers from
493 inherent flaws when %error-verbose or `#define YYERROR_VERBOSE' is
494 used. For a more detailed discussion, see:
495
496 http://lists.gnu.org/archive/html/bison-patches/2009-12/msg00024.html
497
498 The upcoming Bison 2.5 will remove YYFAIL from Java parsers, but
499 deterministic parsers in C will continue to implement it. However,
500 because YYFAIL is already flawed, it seems futile to try to make new
501 Bison features compatible with it. Thus, during parser generation,
502 Bison 2.5 will produce a warning whenever it discovers YYFAIL in a
503 rule action. In a later release, YYFAIL will be disabled for
504 %error-verbose and `#define YYERROR_VERBOSE'. Eventually, YYFAIL will
505 be removed altogether.
506
507 There exists at least one case where Bison 2.5's YYFAIL warning will
508 be a false positive. Some projects add phony uses of YYFAIL and other
509 Bison-defined macros for the sole purpose of suppressing C
510 preprocessor warnings (from GCC cpp's -Wunused-macros, for example).
511 To avoid Bison's future warning, such YYFAIL uses can be moved to the
512 epilogue (that is, after the second `%%') in the Bison input file. In
513 this release (2.4.2), Bison already generates its own code to suppress
514 C preprocessor warnings for YYFAIL, so projects can remove their own
515 phony uses of YYFAIL if compatibility with Bison releases prior to
516 2.4.2 is not necessary.
517
dac8cc0d
AD
518** Internationalization.
519
520 Fix a regression introduced in Bison 2.4: Under some circumstances,
521 message translations were not installed although supported by the
522 host system.
523
41930e7a 524* Changes in version 2.4.1 (2008-12-11):
c9ba9e59 525
a957d06c
JD
526** In the GLR defines file, unexpanded M4 macros in the yylval and yylloc
527 declarations have been fixed.
528
738cde3e
AD
529** Temporary hack for adding a semicolon to the user action.
530
531 Bison used to prepend a trailing semicolon at the end of the user
532 action for reductions. This allowed actions such as
533
534 exp: exp "+" exp { $$ = $1 + $3 };
535
536 instead of
537
538 exp: exp "+" exp { $$ = $1 + $3; };
539
d07932ef
JD
540 Some grammars still depend on this `feature'. Bison 2.4.1 restores
541 the previous behavior in the case of C output (specifically, when
542 neither %language or %skeleton or equivalent command-line options
543 are used) to leave more time for grammars depending on the old
544 behavior to be adjusted. Future releases of Bison will disable this
545 feature.
a957d06c
JD
546
547** A few minor improvements to the Bison manual.
c9ba9e59 548
d6fb461d 549* Changes in version 2.4 (2008-11-02):
7bd1665a 550
d6fb461d 551** %language is an experimental feature.
ed4d67dc
JD
552
553 We first introduced this feature in test release 2.3b as a cleaner
554 alternative to %skeleton. Since then, we have discussed the possibility of
555 modifying its effect on Bison's output file names. Thus, in this release,
556 we consider %language to be an experimental feature that will likely evolve
557 in future releases.
7bd1665a 558
d6fb461d 559** Forward compatibility with GNU M4 has been improved.
241fda7a 560
d6fb461d 561** Several bugs in the C++ skeleton and the experimental Java skeleton have been
241fda7a
JD
562 fixed.
563
d6fb461d 564* Changes in version 2.3b (2008-05-27):
35fe0834 565
d6fb461d 566** The quotes around NAME that used to be required in the following directive
d9df47b6
JD
567 are now deprecated:
568
569 %define NAME "VALUE"
570
d6fb461d 571** The directive `%pure-parser' is now deprecated in favor of:
d9df47b6
JD
572
573 %define api.pure
574
575 which has the same effect except that Bison is more careful to warn about
576 unreasonable usage in the latter case.
577
d6fb461d 578** Push Parsing
c373bf8b
JD
579
580 Bison can now generate an LALR(1) parser in C with a push interface. That
ef1b4273 581 is, instead of invoking `yyparse', which pulls tokens from `yylex', you can
c373bf8b
JD
582 push one token at a time to the parser using `yypush_parse', which will
583 return to the caller after processing each token. By default, the push
584 interface is disabled. Either of the following directives will enable it:
585
586 %define api.push_pull "push" // Just push; does not require yylex.
587 %define api.push_pull "both" // Push and pull; requires yylex.
588
589 See the new section `A Push Parser' in the Bison manual for details.
590
59da312b
JD
591 The current push parsing interface is experimental and may evolve. More user
592 feedback will help to stabilize it.
593
d6fb461d 594** The -g and --graph options now output graphs in Graphviz DOT format,
8e55b3aa
JD
595 not VCG format. Like --graph, -g now also takes an optional FILE argument
596 and thus cannot be bundled with other short options.
c373bf8b 597
d6fb461d 598** Java
59da312b
JD
599
600 Bison can now generate an LALR(1) parser in Java. The skeleton is
601 `data/lalr1.java'. Consider using the new %language directive instead of
602 %skeleton to select it.
603
604 See the new section `Java Parsers' in the Bison manual for details.
605
606 The current Java interface is experimental and may evolve. More user
607 feedback will help to stabilize it.
608
d6fb461d 609** %language
59da312b
JD
610
611 This new directive specifies the programming language of the generated
d43f77e7
PB
612 parser, which can be C (the default), C++, or Java. Besides the skeleton
613 that Bison uses, the directive affects the names of the generated files if
614 the grammar file's name ends in ".y".
59da312b 615
d6fb461d 616** XML Automaton Report
59da312b
JD
617
618 Bison can now generate an XML report of the LALR(1) automaton using the new
619 `--xml' option. The current XML schema is experimental and may evolve. More
620 user feedback will help to stabilize it.
c373bf8b 621
d6fb461d 622** The grammar file may now specify the name of the parser header file using
c373bf8b
JD
623 %defines. For example:
624
625 %defines "parser.h"
626
d6fb461d 627** When reporting useless rules, useless nonterminals, and unused terminals,
d80fb37a
JD
628 Bison now employs the terms "useless in grammar" instead of "useless",
629 "useless in parser" instead of "never reduced", and "unused in grammar"
630 instead of "unused".
cff03fb2 631
d6fb461d 632** Unreachable State Removal
c373bf8b
JD
633
634 Previously, Bison sometimes generated parser tables containing unreachable
31984206
JD
635 states. A state can become unreachable during conflict resolution if Bison
636 disables a shift action leading to it from a predecessor state. Bison now:
75ad86ee
JD
637
638 1. Removes unreachable states.
639
640 2. Does not report any conflicts that appeared in unreachable states.
641 WARNING: As a result, you may need to update %expect and %expect-rr
642 directives in existing grammar files.
643
644 3. For any rule used only in such states, Bison now reports the rule as
cff03fb2 645 "useless in parser due to conflicts".
75ad86ee 646
31984206
JD
647 This feature can be disabled with the following directive:
648
649 %define lr.keep_unreachable_states
650
651 See the %define entry in the `Bison Declaration Summary' in the Bison manual
652 for further discussion.
653
d6fb461d 654** Lookahead Set Correction in the `.output' Report
b1cc23c4 655
c373bf8b 656 When instructed to generate a `.output' file including lookahead sets
88c78747
JD
657 (using `--report=lookahead', for example), Bison now prints each reduction's
658 lookahead set only next to the associated state's one item that (1) is
659 associated with the same rule as the reduction and (2) has its dot at the end
660 of its RHS. Previously, Bison also erroneously printed the lookahead set
661 next to all of the state's other items associated with the same rule. This
662 bug affected only the `.output' file and not the generated parser source
663 code.
664
d6fb461d 665** --report-file=FILE is a new option to override the default `.output' file
59da312b 666 name.
1bb2bd75 667
d6fb461d 668** The `=' that used to be required in the following directives is now
02975b9a
JD
669 deprecated:
670
671 %file-prefix "parser"
672 %name-prefix "c_"
673 %output "parser.c"
674
d6fb461d 675** An Alternative to `%{...%}' -- `%code QUALIFIER {CODE}'
c373bf8b
JD
676
677 Bison 2.3a provided a new set of directives as a more flexible alternative to
8e0a5e9e
JD
678 the traditional Yacc prologue blocks. Those have now been consolidated into
679 a single %code directive with an optional qualifier field, which identifies
680 the purpose of the code and thus the location(s) where Bison should generate
681 it:
682
16dc6a9e
JD
683 1. `%code {CODE}' replaces `%after-header {CODE}'
684 2. `%code requires {CODE}' replaces `%start-header {CODE}'
685 3. `%code provides {CODE}' replaces `%end-header {CODE}'
686 4. `%code top {CODE}' replaces `%before-header {CODE}'
8e0a5e9e 687
61fee93e
JD
688 See the %code entries in section `Bison Declaration Summary' in the Bison
689 manual for a summary of the new functionality. See the new section `Prologue
8e0a5e9e
JD
690 Alternatives' for a detailed discussion including the advantages of %code
691 over the traditional Yacc prologues.
692
693 The prologue alternatives are experimental. More user feedback will help to
694 determine whether they should become permanent features.
695
d6fb461d 696** Revised warning: unset or unused mid-rule values
17bd8a73
JD
697
698 Since Bison 2.2, Bison has warned about mid-rule values that are set but not
699 used within any of the actions of the parent rule. For example, Bison warns
700 about unused $2 in:
701
702 exp: '1' { $$ = 1; } '+' exp { $$ = $1 + $4; };
703
704 Now, Bison also warns about mid-rule values that are used but not set. For
705 example, Bison warns about unset $$ in the mid-rule action in:
706
707 exp: '1' { $1 = 1; } '+' exp { $$ = $2 + $4; };
708
709 However, Bison now disables both of these warnings by default since they
710 sometimes prove to be false alarms in existing grammars employing the Yacc
711 constructs $0 or $-N (where N is some positive integer).
712
59da312b 713 To enable these warnings, specify the option `--warnings=midrule-values' or
17bd8a73
JD
714 `-W', which is a synonym for `--warnings=all'.
715
d6fb461d 716** Default %destructor or %printer with `<*>' or `<>'
c373bf8b
JD
717
718 Bison now recognizes two separate kinds of default %destructor's and
12e35840
JD
719 %printer's:
720
721 1. Place `<*>' in a %destructor/%printer symbol list to define a default
722 %destructor/%printer for all grammar symbols for which you have formally
723 declared semantic type tags.
724
3ebecc24 725 2. Place `<>' in a %destructor/%printer symbol list to define a default
12e35840
JD
726 %destructor/%printer for all grammar symbols without declared semantic
727 type tags.
728
729 Bison no longer supports the `%symbol-default' notation from Bison 2.3a.
3ebecc24 730 `<*>' and `<>' combined achieve the same effect with one exception: Bison no
12e35840
JD
731 longer applies any %destructor to a mid-rule value if that mid-rule value is
732 not actually ever referenced using either $$ or $n in a semantic action.
733
85894313
JD
734 The default %destructor's and %printer's are experimental. More user
735 feedback will help to determine whether they should become permanent
736 features.
737
12e35840
JD
738 See the section `Freeing Discarded Symbols' in the Bison manual for further
739 details.
740
d6fb461d 741** %left, %right, and %nonassoc can now declare token numbers. This is required
ab7f29f8
JD
742 by POSIX. However, see the end of section `Operator Precedence' in the Bison
743 manual for a caveat concerning the treatment of literal strings.
744
d6fb461d 745** The nonfunctional --no-parser, -n, and %no-parser options have been
b1cc23c4
JD
746 completely removed from Bison.
747
d6fb461d 748* Changes in version 2.3a, 2006-09-13:
742e4900 749
d6fb461d 750** Instead of %union, you can define and use your own union type
ddc8ede1
PE
751 YYSTYPE if your grammar contains at least one <type> tag.
752 Your YYSTYPE need not be a macro; it can be a typedef.
753 This change is for compatibility with other Yacc implementations,
754 and is required by POSIX.
755
d6fb461d 756** Locations columns and lines start at 1.
cd48d21d
AD
757 In accordance with the GNU Coding Standards and Emacs.
758
d6fb461d 759** You may now declare per-type and default %destructor's and %printer's:
ec5479ce
JD
760
761 For example:
762
b2a0b7ca
JD
763 %union { char *string; }
764 %token <string> STRING1
765 %token <string> STRING2
766 %type <string> string1
767 %type <string> string2
768 %union { char character; }
769 %token <character> CHR
770 %type <character> chr
771 %destructor { free ($$); } %symbol-default
772 %destructor { free ($$); printf ("%d", @$.first_line); } STRING1 string1
773 %destructor { } <character>
774
775 guarantees that, when the parser discards any user-defined symbol that has a
776 semantic type tag other than `<character>', it passes its semantic value to
777 `free'. However, when the parser discards a `STRING1' or a `string1', it
778 also prints its line number to `stdout'. It performs only the second
779 `%destructor' in this case, so it invokes `free' only once.
ec5479ce 780
85894313
JD
781 [Although we failed to mention this here in the 2.3a release, the default
782 %destructor's and %printer's were experimental, and they were rewritten in
783 future versions.]
784
d6fb461d 785** Except for LALR(1) parsers in C with POSIX Yacc emulation enabled (with `-y',
b931235e
JD
786 `--yacc', or `%yacc'), Bison no longer generates #define statements for
787 associating token numbers with token names. Removing the #define statements
788 helps to sanitize the global namespace during preprocessing, but POSIX Yacc
789 requires them. Bison still generates an enum for token names in all cases.
790
d6fb461d 791** Handling of traditional Yacc prologue blocks is now more consistent but
34f98f46 792 potentially incompatible with previous releases of Bison.
9bc0dd67
JD
793
794 As before, you declare prologue blocks in your grammar file with the
795 `%{ ... %}' syntax. To generate the pre-prologue, Bison concatenates all
34f98f46
JD
796 prologue blocks that you've declared before the first %union. To generate
797 the post-prologue, Bison concatenates all prologue blocks that you've
ddc8ede1 798 declared after the first %union.
9bc0dd67 799
34f98f46 800 Previous releases of Bison inserted the pre-prologue into both the header
9bc0dd67
JD
801 file and the code file in all cases except for LALR(1) parsers in C. In the
802 latter case, Bison inserted it only into the code file. For parsers in C++,
803 the point of insertion was before any token definitions (which associate
804 token numbers with names). For parsers in C, the point of insertion was
805 after the token definitions.
806
807 Now, Bison never inserts the pre-prologue into the header file. In the code
808 file, it always inserts it before the token definitions.
809
d6fb461d 810** Bison now provides a more flexible alternative to the traditional Yacc
34f98f46
JD
811 prologue blocks: %before-header, %start-header, %end-header, and
812 %after-header.
813
814 For example, the following declaration order in the grammar file reflects the
815 order in which Bison will output these code blocks. However, you are free to
816 declare these code blocks in your grammar file in whatever order is most
817 convenient for you:
818
819 %before-header {
820 /* Bison treats this block like a pre-prologue block: it inserts it into
821 * the code file before the contents of the header file. It does *not*
822 * insert it into the header file. This is a good place to put
823 * #include's that you want at the top of your code file. A common
824 * example is `#include "system.h"'. */
825 }
826 %start-header {
827 /* Bison inserts this block into both the header file and the code file.
828 * In both files, the point of insertion is before any Bison-generated
829 * token, semantic type, location type, and class definitions. This is a
830 * good place to define %union dependencies, for example. */
9bc0dd67
JD
831 }
832 %union {
34f98f46
JD
833 /* Unlike the traditional Yacc prologue blocks, the output order for the
834 * new %*-header blocks is not affected by their declaration position
835 * relative to any %union in the grammar file. */
9bc0dd67 836 }
34f98f46
JD
837 %end-header {
838 /* Bison inserts this block into both the header file and the code file.
839 * In both files, the point of insertion is after the Bison-generated
840 * definitions. This is a good place to declare or define public
841 * functions or data structures that depend on the Bison-generated
842 * definitions. */
9bc0dd67 843 }
34f98f46
JD
844 %after-header {
845 /* Bison treats this block like a post-prologue block: it inserts it into
846 * the code file after the contents of the header file. It does *not*
847 * insert it into the header file. This is a good place to declare or
848 * define internal functions or data structures that depend on the
849 * Bison-generated definitions. */
850 }
851
852 If you have multiple occurrences of any one of the above declarations, Bison
853 will concatenate the contents in declaration order.
9bc0dd67 854
85894313
JD
855 [Although we failed to mention this here in the 2.3a release, the prologue
856 alternatives were experimental, and they were rewritten in future versions.]
857
d6fb461d 858** The option `--report=look-ahead' has been changed to `--report=lookahead'.
9e6e7ed2
PE
859 The old spelling still works, but is not documented and may be removed
860 in a future release.
742e4900 861
d6fb461d 862* Changes in version 2.3, 2006-06-05:
4ad3ed84 863
d6fb461d 864** GLR grammars should now use `YYRECOVERING ()' instead of `YYRECOVERING',
4ad3ed84
PE
865 for compatibility with LALR(1) grammars.
866
d6fb461d 867** It is now documented that any definition of YYSTYPE or YYLTYPE should
4ad3ed84
PE
868 be to a type name that does not contain parentheses or brackets.
869
d6fb461d 870* Changes in version 2.2, 2006-05-19:
193d7c70 871
d6fb461d 872** The distribution terms for all Bison-generated parsers now permit
193d7c70
PE
873 using the parsers in nonfree programs. Previously, this permission
874 was granted only for Bison-generated LALR(1) parsers in C.
5f4236a0 875
d6fb461d 876** %name-prefix changes the namespace name in C++ outputs.
aa08666d 877
d6fb461d 878** The C++ parsers export their token_type.
5f4236a0 879
d6fb461d 880** Bison now allows multiple %union declarations, and concatenates
d6ca7905
PE
881 their contents together.
882
d6fb461d 883** New warning: unused values
4d7bc38c
PE
884 Right-hand side symbols whose values are not used are reported,
885 if the symbols have destructors. For instance:
affac613 886
8f3596a6 887 exp: exp "?" exp ":" exp { $1 ? $1 : $3; }
721be13c
PE
888 | exp "+" exp
889 ;
affac613 890
8f3596a6
AD
891 will trigger a warning about $$ and $5 in the first rule, and $3 in
892 the second ($1 is copied to $$ by the default rule). This example
4e26c69e 893 most likely contains three errors, and could be rewritten as:
affac613 894
4e26c69e
PE
895 exp: exp "?" exp ":" exp
896 { $$ = $1 ? $3 : $5; free ($1 ? $5 : $3); free ($1); }
897 | exp "+" exp
898 { $$ = $1 ? $1 : $3; if ($1) free ($3); }
721be13c 899 ;
affac613 900
4e26c69e
PE
901 However, if the original actions were really intended, memory leaks
902 and all, the warnings can be suppressed by letting Bison believe the
903 values are used, e.g.:
721be13c 904
8f3596a6 905 exp: exp "?" exp ":" exp { $1 ? $1 : $3; (void) ($$, $5); }
721be13c
PE
906 | exp "+" exp { $$ = $1; (void) $3; }
907 ;
908
84866159
AD
909 If there are mid-rule actions, the warning is issued if no action
910 uses it. The following triggers no warning: $1 and $3 are used.
911
912 exp: exp { push ($1); } '+' exp { push ($3); sum (); };
913
721be13c
PE
914 The warning is intended to help catching lost values and memory leaks.
915 If a value is ignored, its associated memory typically is not reclaimed.
affac613 916
d6fb461d 917** %destructor vs. YYABORT, YYACCEPT, and YYERROR.
9d9b8b70
PE
918 Destructors are now called when user code invokes YYABORT, YYACCEPT,
919 and YYERROR, for all objects on the stack, other than objects
920 corresponding to the right-hand side of the current rule.
a85284cf 921
d6fb461d 922** %expect, %expect-rr
035aa4a0
PE
923 Incorrect numbers of expected conflicts are now actual errors,
924 instead of warnings.
925
d6fb461d 926** GLR, YACC parsers.
4e26c69e
PE
927 The %parse-params are available in the destructors (and the
928 experimental printers) as per the documentation.
4b367315 929
d6fb461d 930** Bison now warns if it finds a stray `$' or `@' in an action.
ad6a9b97 931
d6fb461d 932** %require "VERSION"
4e26c69e
PE
933 This specifies that the grammar file depends on features implemented
934 in Bison version VERSION or higher.
b50d2359 935
d6fb461d 936** lalr1.cc: The token and value types are now class members.
e14d0ab6
AD
937 The tokens were defined as free form enums and cpp macros. YYSTYPE
938 was defined as a free form union. They are now class members:
fb9712a9
AD
939 tokens are enumerations of the `yy::parser::token' struct, and the
940 semantic values have the `yy::parser::semantic_type' type.
941
942 If you do not want or can update to this scheme, the directive
943 `%define "global_tokens_and_yystype" "1"' triggers the global
b50d2359
AD
944 definition of tokens and YYSTYPE. This change is suitable both
945 for previous releases of Bison, and this one.
fb9712a9 946
b50d2359 947 If you wish to update, then make sure older version of Bison will
ab8d9dc5 948 fail using `%require "2.2"'.
fb9712a9 949
d6fb461d 950** DJGPP support added.
193d7c70 951\f
d6fb461d 952* Changes in version 2.1, 2005-09-16:
1ce59070 953
d6fb461d 954** The C++ lalr1.cc skeleton supports %lex-param.
e14d0ab6 955
d6fb461d 956** Bison-generated parsers now support the translation of diagnostics like
baf785db
PE
957 "syntax error" into languages other than English. The default
958 language is still English. For details, please see the new
0410a6e0
PE
959 Internationalization section of the Bison manual. Software
960 distributors should also see the new PACKAGING file. Thanks to
961 Bruno Haible for this new feature.
1ce59070 962
d6fb461d 963** Wording in the Bison-generated parsers has been changed slightly to
1a059451
PE
964 simplify translation. In particular, the message "memory exhausted"
965 has replaced "parser stack overflow", as the old message was not
966 always accurate for modern Bison-generated parsers.
967
d6fb461d 968** Destructors are now called when the parser aborts, for all symbols left
258b75ca
PE
969 behind on the stack. Also, the start symbol is now destroyed after a
970 successful parse. In both cases, the behavior was formerly inconsistent.
971
d6fb461d 972** When generating verbose diagnostics, Bison-generated parsers no longer
72f000b0
PE
973 quote the literal strings associated with tokens. For example, for
974 a syntax error associated with '%token NUM "number"' they might
975 print 'syntax error, unexpected number' instead of 'syntax error,
976 unexpected "number"'.
193d7c70 977\f
d6fb461d 978* Changes in version 2.0, 2004-12-25:
efeed023 979
d6fb461d 980** Possibly-incompatible changes
d7e14fc0 981
82de6b0d
PE
982 - Bison-generated parsers no longer default to using the alloca function
983 (when available) to extend the parser stack, due to widespread
984 problems in unchecked stack-overflow detection. You can "#define
985 YYSTACK_USE_ALLOCA 1" to require the use of alloca, but please read
986 the manual to determine safe values for YYMAXDEPTH in that case.
8dd162d3 987
82de6b0d
PE
988 - Error token location.
989 During error recovery, the location of the syntax error is updated
990 to cover the whole sequence covered by the error token: it includes
991 the shifted symbols thrown away during the first part of the error
992 recovery, and the lookahead rejected during the second part.
18d192f0 993
82de6b0d
PE
994 - Semicolon changes:
995 . Stray semicolons are no longer allowed at the start of a grammar.
996 . Semicolons are now required after in-grammar declarations.
e342c3be 997
82de6b0d
PE
998 - Unescaped newlines are no longer allowed in character constants or
999 string literals. They were never portable, and GCC 3.4.0 has
1000 dropped support for them. Better diagnostics are now generated if
1001 forget a closing quote.
8dd162d3 1002
82de6b0d 1003 - NUL bytes are no longer allowed in Bison string literals, unfortunately.
f74b6f91 1004
d6fb461d 1005** New features
1452af69 1006
82de6b0d 1007 - GLR grammars now support locations.
4febdd96 1008
82de6b0d
PE
1009 - New directive: %initial-action.
1010 This directive allows the user to run arbitrary code (including
1011 initializing @$) from yyparse before parsing starts.
1452af69 1012
82de6b0d
PE
1013 - A new directive "%expect-rr N" specifies the expected number of
1014 reduce/reduce conflicts in GLR parsers.
1452af69 1015
82de6b0d
PE
1016 - %token numbers can now be hexadecimal integers, e.g., `%token FOO 0x12d'.
1017 This is a GNU extension.
4febdd96 1018
82de6b0d 1019 - The option `--report=lookahead' was changed to `--report=look-ahead'.
9e6e7ed2 1020 [However, this was changed back after 2.3.]
1452af69 1021
82de6b0d 1022 - Experimental %destructor support has been added to lalr1.cc.
1452af69 1023
82de6b0d
PE
1024 - New configure option --disable-yacc, to disable installation of the
1025 yacc command and -ly library introduced in 1.875 for POSIX conformance.
6040d338 1026
d6fb461d 1027** Bug fixes
d5a3fe37 1028
82de6b0d
PE
1029 - For now, %expect-count violations are now just warnings, not errors.
1030 This is for compatibility with Bison 1.75 and earlier (when there are
1031 reduce/reduce conflicts) and with Bison 1.30 and earlier (when there
1032 are too many or too few shift/reduce conflicts). However, in future
1033 versions of Bison we plan to improve the %expect machinery so that
1034 these violations will become errors again.
3473d0f8 1035
82de6b0d
PE
1036 - Within Bison itself, numbers (e.g., goto numbers) are no longer
1037 arbitrarily limited to 16-bit counts.
d600ee67 1038
82de6b0d 1039 - Semicolons are now allowed before "|" in grammar rules, as POSIX requires.
d600ee67 1040\f
d6fb461d 1041* Changes in version 1.875, 2003-01-01:
963fcc17 1042
d6fb461d 1043** The documentation license has been upgraded to version 1.2
dc546b0f 1044 of the GNU Free Documentation License.
75eb3bc4 1045
d6fb461d 1046** syntax error processing
75eb3bc4 1047
dc546b0f
PE
1048 - In Yacc-style parsers YYLLOC_DEFAULT is now used to compute error
1049 locations too. This fixes bugs in error-location computation.
75eb3bc4 1050
dc546b0f
PE
1051 - %destructor
1052 It is now possible to reclaim the memory associated to symbols
1053 discarded during error recovery. This feature is still experimental.
20daca06 1054
dc546b0f
PE
1055 - %error-verbose
1056 This new directive is preferred over YYERROR_VERBOSE.
74724a70 1057
dc546b0f
PE
1058 - #defining yyerror to steal internal variables is discouraged.
1059 It is not guaranteed to work forever.
d1de5372 1060
d6fb461d 1061** POSIX conformance
d1de5372 1062
dc546b0f
PE
1063 - Semicolons are once again optional at the end of grammar rules.
1064 This reverts to the behavior of Bison 1.33 and earlier, and improves
1065 compatibility with Yacc.
74724a70 1066
dc546b0f
PE
1067 - `parse error' -> `syntax error'
1068 Bison now uniformly uses the term `syntax error'; formerly, the code
1069 and manual sometimes used the term `parse error' instead. POSIX
1070 requires `syntax error' in diagnostics, and it was thought better to
1071 be consistent.
74724a70 1072
dc546b0f
PE
1073 - The documentation now emphasizes that yylex and yyerror must be
1074 declared before use. C99 requires this.
d1de5372 1075
dc546b0f
PE
1076 - Bison now parses C99 lexical constructs like UCNs and
1077 backslash-newline within C escape sequences, as POSIX 1003.1-2001 requires.
d1de5372 1078
dc546b0f
PE
1079 - File names are properly escaped in C output. E.g., foo\bar.y is
1080 output as "foo\\bar.y".
6780ca7a 1081
dc546b0f
PE
1082 - Yacc command and library now available
1083 The Bison distribution now installs a `yacc' command, as POSIX requires.
1084 Also, Bison now installs a small library liby.a containing
1085 implementations of Yacc-compatible yyerror and main functions.
1086 This library is normally not useful, but POSIX requires it.
6e649e65 1087
dc546b0f 1088 - Type clashes now generate warnings, not errors.
6e649e65 1089
dc546b0f
PE
1090 - If the user does not define YYSTYPE as a macro, Bison now declares it
1091 using typedef instead of defining it as a macro.
1092 For consistency, YYLTYPE is also declared instead of defined.
9501dc6e 1093
d6fb461d 1094** Other compatibility issues
886a425c 1095
dc546b0f
PE
1096 - %union directives can now have a tag before the `{', e.g., the
1097 directive `%union foo {...}' now generates the C code
1098 `typedef union foo { ... } YYSTYPE;'; this is for Yacc compatibility.
1099 The default union tag is `YYSTYPE', for compatibility with Solaris 9 Yacc.
1100 For consistency, YYLTYPE's struct tag is now `YYLTYPE' not `yyltype'.
1101 This is for compatibility with both Yacc and Bison 1.35.
72f889cc 1102
dc546b0f
PE
1103 - `;' is output before the terminating `}' of an action, for
1104 compatibility with Bison 1.35.
886a425c 1105
dc546b0f
PE
1106 - Bison now uses a Yacc-style format for conflict reports, e.g.,
1107 `conflicts: 2 shift/reduce, 1 reduce/reduce'.
437c2d80 1108
dc546b0f
PE
1109 - `yystype' and `yyltype' are now obsolescent macros instead of being
1110 typedefs or tags; they are no longer documented and are planned to be
1111 withdrawn in a future release.
2a8d363a 1112
d6fb461d 1113** GLR parser notes
2a8d363a 1114
dc546b0f
PE
1115 - GLR and inline
1116 Users of Bison have to decide how they handle the portability of the
1117 C keyword `inline'.
959e5f51 1118
dc546b0f
PE
1119 - `parsing stack overflow...' -> `parser stack overflow'
1120 GLR parsers now report `parser stack overflow' as per the Bison manual.
900c5db5 1121
d6fb461d 1122** Bison now warns if it detects conflicting outputs to the same file,
dc546b0f
PE
1123 e.g., it generates a warning for `bison -d -o foo.h foo.y' since
1124 that command outputs both code and header to foo.h.
6e40b4eb 1125
d6fb461d 1126** #line in output files
dc546b0f 1127 - --no-line works properly.
6e40b4eb 1128
d6fb461d 1129** Bison can no longer be built by a K&R C compiler; it requires C89 or
6e40b4eb
AD
1130 later to be built. This change originally took place a few versions
1131 ago, but nobody noticed until we recently asked someone to try
1132 building Bison with a K&R C compiler.
d600ee67 1133\f
d6fb461d 1134* Changes in version 1.75, 2002-10-14:
7933f2b5 1135
d6fb461d 1136** Bison should now work on 64-bit hosts.
7933f2b5 1137
d6fb461d 1138** Indonesian translation thanks to Tedi Heriyanto.
7933f2b5 1139
d6fb461d 1140** GLR parsers
f50adbbd
AD
1141 Fix spurious parse errors.
1142
d6fb461d 1143** Pure parsers
f50adbbd
AD
1144 Some people redefine yyerror to steal yyparse' private variables.
1145 Reenable this trick until an official feature replaces it.
1146
d6fb461d 1147** Type Clashes
d90c934c
AD
1148 In agreement with POSIX and with other Yaccs, leaving a default
1149 action is valid when $$ is untyped, and $1 typed:
1150
1151 untyped: ... typed;
1152
1153 but the converse remains an error:
1154
1155 typed: ... untyped;
1156
d6fb461d 1157** Values of mid-rule actions
d90c934c
AD
1158 The following code:
1159
1160 foo: { ... } { $$ = $1; } ...
1161
1162 was incorrectly rejected: $1 is defined in the second mid-rule
1163 action, and is equal to the $$ of the first mid-rule action.
d600ee67 1164\f
d6fb461d 1165* Changes in version 1.50, 2002-10-04:
adc8c848 1166
d6fb461d 1167** GLR parsing
676385e2
PH
1168 The declaration
1169 %glr-parser
1170 causes Bison to produce a Generalized LR (GLR) parser, capable of handling
1171 almost any context-free grammar, ambiguous or not. The new declarations
e8832397 1172 %dprec and %merge on grammar rules allow parse-time resolution of
676385e2
PH
1173 ambiguities. Contributed by Paul Hilfinger.
1174
7933f2b5 1175 Unfortunately Bison 1.50 does not work properly on 64-bit hosts
420f93c8
PE
1176 like the Alpha, so please stick to 32-bit hosts for now.
1177
d6fb461d 1178** Output Directory
8c165d89 1179 When not in Yacc compatibility mode, when the output file was not
e88dbdbf 1180 specified, running `bison foo/bar.y' created `foo/bar.c'. It
8c165d89
AD
1181 now creates `bar.c'.
1182
d6fb461d 1183** Undefined token
007a50a4 1184 The undefined token was systematically mapped to 2 which prevented
e88dbdbf 1185 the use of 2 by the user. This is no longer the case.
007a50a4 1186
d6fb461d 1187** Unknown token numbers
e88dbdbf 1188 If yylex returned an out of range value, yyparse could die. This is
007a50a4
AD
1189 no longer the case.
1190
d6fb461d 1191** Error token
e88dbdbf 1192 According to POSIX, the error token must be 256.
23c5a174
AD
1193 Bison extends this requirement by making it a preference: *if* the
1194 user specified that one of her tokens is numbered 256, then error
1195 will be mapped onto another number.
1196
d6fb461d 1197** Verbose error messages
e88dbdbf 1198 They no longer report `..., expecting error or...' for states where
217598da
AD
1199 error recovery is possible.
1200
d6fb461d 1201** End token
217598da
AD
1202 Defaults to `$end' instead of `$'.
1203
d6fb461d 1204** Error recovery now conforms to documentation and to POSIX
68cd8af3
PE
1205 When a Bison-generated parser encounters a syntax error, it now pops
1206 the stack until it finds a state that allows shifting the error
1207 token. Formerly, it popped the stack until it found a state that
1208 allowed some non-error action other than a default reduction on the
1209 error token. The new behavior has long been the documented behavior,
1210 and has long been required by POSIX. For more details, please see
337116ba
PE
1211 Paul Eggert, "Reductions during Bison error handling" (2002-05-20)
1212 <http://lists.gnu.org/archive/html/bug-bison/2002-05/msg00038.html>.
68cd8af3 1213
d6fb461d 1214** Traces
5504898e
AD
1215 Popped tokens and nonterminals are now reported.
1216
d6fb461d 1217** Larger grammars
a861a339
PE
1218 Larger grammars are now supported (larger token numbers, larger grammar
1219 size (= sum of the LHS and RHS lengths), larger LALR tables).
1220 Formerly, many of these numbers ran afoul of 16-bit limits;
1221 now these limits are 32 bits on most hosts.
355e7c1c 1222
d6fb461d 1223** Explicit initial rule
643a5994
AD
1224 Bison used to play hacks with the initial rule, which the user does
1225 not write. It is now explicit, and visible in the reports and
1226 graphs as rule 0.
23c5a174 1227
d6fb461d 1228** Useless rules
643a5994 1229 Before, Bison reported the useless rules, but, although not used,
77714df2 1230 included them in the parsers. They are now actually removed.
23c5a174 1231
d6fb461d 1232** Useless rules, useless nonterminals
6b98e4b5
AD
1233 They are now reported, as a warning, with their locations.
1234
d6fb461d 1235** Rules never reduced
e8832397
AD
1236 Rules that can never be reduced because of conflicts are now
1237 reported.
1238
d6fb461d 1239** Incorrect `Token not used'
11652ab3
AD
1240 On a grammar such as
1241
1242 %token useless useful
1243 %%
1244 exp: '0' %prec useful;
1245
1246 where a token was used to set the precedence of the last rule,
1247 bison reported both `useful' and `useless' as useless tokens.
1248
d6fb461d 1249** Revert the C++ namespace changes introduced in 1.31
77714df2 1250 as they caused too many portability hassles.
0179dd65 1251
d6fb461d 1252** Default locations
b2d52318
AD
1253 By an accident of design, the default computation of @$ was
1254 performed after another default computation was performed: @$ = @1.
1255 The latter is now removed: YYLLOC_DEFAULT is fully responsible of
1256 the computation of @$.
adc8c848 1257
d6fb461d 1258** Token end-of-file
b7c49edf
AD
1259 The token end of file may be specified by the user, in which case,
1260 the user symbol is used in the reports, the graphs, and the verbose
a861a339 1261 error messages instead of `$end', which remains being the default.
b7c49edf 1262 For instance
7bd6c77e 1263 %token MYEOF 0
b7c49edf 1264 or
7bd6c77e 1265 %token MYEOF 0 "end of file"
fdbcd8e2 1266
d6fb461d 1267** Semantic parser
fdbcd8e2
AD
1268 This old option, which has been broken for ages, is removed.
1269
d6fb461d 1270** New translations
a861a339 1271 Brazilian Portuguese, thanks to Alexandre Folle de Menezes.
84614e13
AD
1272 Croatian, thanks to Denis Lackovic.
1273
d6fb461d 1274** Incorrect token definitions
e88dbdbf 1275 When given `%token 'a' "A"', Bison used to output `#define 'a' 65'.
b87f8b21 1276
d6fb461d 1277** Token definitions as enums
77714df2
AD
1278 Tokens are output both as the traditional #define's, and, provided
1279 the compiler supports ANSI C or is a C++ compiler, as enums.
e88dbdbf 1280 This lets debuggers display names instead of integers.
77714df2 1281
d6fb461d 1282** Reports
ec3bc396
AD
1283 In addition to --verbose, bison supports --report=THINGS, which
1284 produces additional information:
b408954b
AD
1285 - itemset
1286 complete the core item sets with their closure
9e6e7ed2
PE
1287 - lookahead [changed to `look-ahead' in 1.875e through 2.3, but changed back]
1288 explicitly associate lookahead tokens to items
b408954b
AD
1289 - solved
1290 describe shift/reduce conflicts solving.
1291 Bison used to systematically output this information on top of
1292 the report. Solved conflicts are now attached to their states.
ec3bc396 1293
d6fb461d 1294** Type clashes
9af3fbce
AD
1295 Previous versions don't complain when there is a type clash on
1296 the default action if the rule has a mid-rule action, such as in:
1297
1298 %type <foo> bar
1299 %%
1300 bar: '0' {} '0';
1301
1302 This is fixed.
a861a339 1303
d6fb461d 1304** GNU M4 is now required when using Bison.
f987e9d2 1305\f
d6fb461d 1306* Changes in version 1.35, 2002-03-25:
76551463 1307
d6fb461d 1308** C Skeleton
76551463
AD
1309 Some projects use Bison's C parser with C++ compilers, and define
1310 YYSTYPE as a class. The recent adjustment of C parsers for data
1311 alignment and 64 bit architectures made this impossible.
1312
1313 Because for the time being no real solution for C++ parser
1314 generation exists, kludges were implemented in the parser to
1315 maintain this use. In the future, when Bison has C++ parsers, this
1316 kludge will be disabled.
1317
1318 This kludge also addresses some C++ problems when the stack was
1319 extended.
76551463 1320\f
d6fb461d 1321* Changes in version 1.34, 2002-03-12:
76551463 1322
d6fb461d 1323** File name clashes are detected
76551463
AD
1324 $ bison foo.y -d -o foo.x
1325 fatal error: header and parser would both be named `foo.x'
1326
d6fb461d 1327** A missing `;' at the end of a rule triggers a warning
76551463
AD
1328 In accordance with POSIX, and in agreement with other
1329 Yacc implementations, Bison will mandate this semicolon in the near
1330 future. This eases the implementation of a Bison parser of Bison
1331 grammars by making this grammar LALR(1) instead of LR(2). To
1332 facilitate the transition, this release introduces a warning.
1333
d6fb461d 1334** Revert the C++ namespace changes introduced in 1.31, as they caused too
76551463
AD
1335 many portability hassles.
1336
d6fb461d 1337** DJGPP support added.
76551463 1338
d6fb461d 1339** Fix test suite portability problems.
76551463 1340\f
d6fb461d 1341* Changes in version 1.33, 2002-02-07:
76551463 1342
d6fb461d 1343** Fix C++ issues
76551463
AD
1344 Groff could not be compiled for the definition of size_t was lacking
1345 under some conditions.
1346
d6fb461d 1347** Catch invalid @n
76551463
AD
1348 As is done with $n.
1349\f
d6fb461d 1350* Changes in version 1.32, 2002-01-23:
76551463 1351
d6fb461d 1352** Fix Yacc output file names
76551463 1353
d6fb461d 1354** Portability fixes
76551463 1355
d6fb461d 1356** Italian, Dutch translations
76551463 1357\f
d6fb461d 1358* Changes in version 1.31, 2002-01-14:
52d1aeee 1359
d6fb461d 1360** Many Bug Fixes
52d1aeee 1361
d6fb461d 1362** GNU Gettext and %expect
52d1aeee
MA
1363 GNU Gettext asserts 10 s/r conflicts, but there are 7. Now that
1364 Bison dies on incorrect %expectations, we fear there will be
1365 too many bug reports for Gettext, so _for the time being_, %expect
1366 does not trigger an error when the input file is named `plural.y'.
1367
d6fb461d 1368** Use of alloca in parsers
52d1aeee
MA
1369 If YYSTACK_USE_ALLOCA is defined to 0, then the parsers will use
1370 malloc exclusively. Since 1.29, but was not NEWS'ed.
1371
1372 alloca is used only when compiled with GCC, to avoid portability
1373 problems as on AIX.
1374
d6fb461d 1375** yyparse now returns 2 if memory is exhausted; formerly it dumped core.
b47dbebe 1376
d6fb461d 1377** When the generated parser lacks debugging code, YYDEBUG is now 0
52d1aeee
MA
1378 (as POSIX requires) instead of being undefined.
1379
d6fb461d 1380** User Actions
52d1aeee
MA
1381 Bison has always permitted actions such as { $$ = $1 }: it adds the
1382 ending semicolon. Now if in Yacc compatibility mode, the semicolon
1383 is no longer output: one has to write { $$ = $1; }.
1384
d6fb461d 1385** Better C++ compliance
52d1aeee 1386 The output parsers try to respect C++ namespaces.
76551463 1387 [This turned out to be a failed experiment, and it was reverted later.]
52d1aeee 1388
d6fb461d 1389** Reduced Grammars
52d1aeee
MA
1390 Fixed bugs when reporting useless nonterminals.
1391
d6fb461d 1392** 64 bit hosts
52d1aeee
MA
1393 The parsers work properly on 64 bit hosts.
1394
d6fb461d 1395** Error messages
52d1aeee
MA
1396 Some calls to strerror resulted in scrambled or missing error messages.
1397
d6fb461d 1398** %expect
52d1aeee
MA
1399 When the number of shift/reduce conflicts is correct, don't issue
1400 any warning.
1401
d6fb461d 1402** The verbose report includes the rule line numbers.
52d1aeee 1403
d6fb461d 1404** Rule line numbers are fixed in traces.
52d1aeee 1405
d6fb461d 1406** Swedish translation
52d1aeee 1407
d6fb461d 1408** Parse errors
52d1aeee
MA
1409 Verbose parse error messages from the parsers are better looking.
1410 Before: parse error: unexpected `'/'', expecting `"number"' or `'-'' or `'(''
1411 Now: parse error: unexpected '/', expecting "number" or '-' or '('
1412
d6fb461d 1413** Fixed parser memory leaks.
52d1aeee
MA
1414 When the generated parser was using malloc to extend its stacks, the
1415 previous allocations were not freed.
1416
d6fb461d 1417** Fixed verbose output file.
52d1aeee
MA
1418 Some newlines were missing.
1419 Some conflicts in state descriptions were missing.
1420
d6fb461d 1421** Fixed conflict report.
52d1aeee
MA
1422 Option -v was needed to get the result.
1423
d6fb461d 1424** %expect
52d1aeee
MA
1425 Was not used.
1426 Mismatches are errors, not warnings.
1427
d6fb461d 1428** Fixed incorrect processing of some invalid input.
52d1aeee 1429
d6fb461d 1430** Fixed CPP guards: 9foo.h uses BISON_9FOO_H instead of 9FOO_H.
52d1aeee 1431
d6fb461d 1432** Fixed some typos in the documentation.
52d1aeee 1433
d6fb461d 1434** %token MY_EOF 0 is supported.
52d1aeee
MA
1435 Before, MY_EOF was silently renumbered as 257.
1436
d6fb461d 1437** doc/refcard.tex is updated.
52d1aeee 1438
d6fb461d 1439** %output, %file-prefix, %name-prefix.
52d1aeee
MA
1440 New.
1441
d6fb461d 1442** --output
52d1aeee
MA
1443 New, aliasing `--output-file'.
1444\f
d6fb461d 1445* Changes in version 1.30, 2001-10-26:
342b8b6e 1446
d6fb461d 1447** `--defines' and `--graph' have now an optional argument which is the
fdac0091 1448 output file name. `-d' and `-g' do not change; they do not take any
342b8b6e
AD
1449 argument.
1450
d6fb461d 1451** `%source_extension' and `%header_extension' are removed, failed
342b8b6e
AD
1452 experiment.
1453
d6fb461d 1454** Portability fixes.
f987e9d2 1455\f
d6fb461d 1456* Changes in version 1.29, 2001-09-07:
342b8b6e 1457
d6fb461d 1458** The output file does not define const, as this caused problems when used
342b8b6e
AD
1459 with common autoconfiguration schemes. If you still use ancient compilers
1460 that lack const, compile with the equivalent of the C compiler option
1461 `-Dconst='. autoconf's AC_C_CONST macro provides one way to do this.
1462
d6fb461d 1463** Added `-g' and `--graph'.
f87a2205 1464
d6fb461d 1465** The Bison manual is now distributed under the terms of the GNU FDL.
f2b5126e 1466
d6fb461d 1467** The input and the output files has automatically a similar extension.
234a3be3 1468
d6fb461d 1469** Russian translation added.
f87a2205 1470
d6fb461d 1471** NLS support updated; should hopefully be less troublesome.
f87a2205 1472
d6fb461d 1473** Added the old Bison reference card.
c33638bb 1474
d6fb461d 1475** Added `--locations' and `%locations'.
6deb4447 1476
d6fb461d 1477** Added `-S' and `--skeleton'.
cd5bd6ac 1478
d6fb461d 1479** `%raw', `-r', `--raw' is disabled.
62ab6972 1480
d6fb461d 1481** Special characters are escaped when output. This solves the problems
cd5bd6ac
AD
1482 of the #line lines with path names including backslashes.
1483
d6fb461d 1484** New directives.
4ecbf796
MA
1485 `%yacc', `%fixed_output_files', `%defines', `%no_parser', `%verbose',
1486 `%debug', `%source_extension' and `%header_extension'.
f987e9d2 1487
d6fb461d 1488** @$
f987e9d2 1489 Automatic location tracking.
f87a2205 1490\f
d6fb461d 1491* Changes in version 1.28, 1999-07-06:
d2e00347 1492
d6fb461d 1493** Should compile better now with K&R compilers.
d2e00347 1494
d6fb461d 1495** Added NLS.
d2e00347 1496
d6fb461d 1497** Fixed a problem with escaping the double quote character.
d2e00347 1498
d6fb461d 1499** There is now a FAQ.
d2e00347 1500\f
d6fb461d 1501* Changes in version 1.27:
5c31c3c2 1502
d6fb461d 1503** The make rule which prevented bison.simple from being created on
5c31c3c2
JT
1504 some systems has been fixed.
1505\f
d6fb461d 1506* Changes in version 1.26:
4be07551 1507
d6fb461d 1508** Bison now uses automake.
4be07551 1509
d6fb461d 1510** New mailing lists: <bug-bison@gnu.org> and <help-bison@gnu.org>.
4be07551 1511
d6fb461d 1512** Token numbers now start at 257 as previously documented, not 258.
4be07551 1513
d6fb461d 1514** Bison honors the TMPDIR environment variable.
4be07551 1515
d6fb461d 1516** A couple of buffer overruns have been fixed.
f51dbca1 1517
d6fb461d 1518** Problems when closing files should now be reported.
f51dbca1 1519
d6fb461d 1520** Generated parsers should now work even on operating systems which do
f51dbca1 1521 not provide alloca().
4be07551 1522\f
d6fb461d 1523* Changes in version 1.25, 1995-10-16:
df8878c5 1524
d6fb461d 1525** Errors in the input grammar are not fatal; Bison keeps reading
df8878c5 1526the grammar file, and reports all the errors found in it.
8c44d3ec 1527
d6fb461d 1528** Tokens can now be specified as multiple-character strings: for
df8878c5
RS
1529example, you could use "<=" for a token which looks like <=, instead
1530of chosing a name like LESSEQ.
1531
d6fb461d 1532** The %token_table declaration says to write a table of tokens (names
df8878c5
RS
1533and numbers) into the parser file. The yylex function can use this
1534table to recognize multiple-character string tokens, or for other
1535purposes.
1536
d6fb461d 1537** The %no_lines declaration says not to generate any #line preprocessor
df8878c5
RS
1538directives in the parser file.
1539
d6fb461d 1540** The %raw declaration says to use internal Bison token numbers, not
df8878c5
RS
1541Yacc-compatible token numbers, when token names are defined as macros.
1542
d6fb461d 1543** The --no-parser option produces the parser tables without including
df8878c5
RS
1544the parser engine; a project can now use its own parser engine.
1545The actions go into a separate file called NAME.act, in the form of
1546a switch statement body.
1547\f
d6fb461d 1548* Changes in version 1.23:
6780ca7a 1549
4d019228
DM
1550The user can define YYPARSE_PARAM as the name of an argument to be
1551passed into yyparse. The argument should have type void *. It should
1552actually point to an object. Grammar actions can access the variable
1553by casting it to the proper pointer type.
6780ca7a 1554
6780ca7a 1555Line numbers in output file corrected.
6780ca7a 1556\f
d6fb461d 1557* Changes in version 1.22:
6780ca7a
DM
1558
1559--help option added.
6780ca7a 1560\f
d6fb461d 1561* Changes in version 1.20:
6780ca7a
DM
1562
1563Output file does not redefine const for C++.
9f4503d6
AD
1564
1565Local Variables:
1566mode: outline
1567End:
76551463
AD
1568
1569-----
1570
c932d613 1571Copyright (C) 1995-2012 Free Software Foundation, Inc.
76551463 1572
8defe11b 1573This file is part of Bison, the GNU Parser Generator.
76551463 1574
f16b0819 1575This program is free software: you can redistribute it and/or modify
76551463 1576it under the terms of the GNU General Public License as published by
f16b0819
PE
1577the Free Software Foundation, either version 3 of the License, or
1578(at your option) any later version.
76551463 1579
f16b0819 1580This program is distributed in the hope that it will be useful,
76551463
AD
1581but WITHOUT ANY WARRANTY; without even the implied warranty of
1582MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
1583GNU General Public License for more details.
1584
1585You should have received a copy of the GNU General Public License
f16b0819 1586along with this program. If not, see <http://www.gnu.org/licenses/>.