]> git.saurik.com Git - bison.git/blame - NEWS
doc: discuss named references after locations.
[bison.git] / NEWS
CommitLineData
6780ca7a
DM
1Bison News
2----------
3af4feb2 3
90dcd415
JD
4* Changes in version 2.5.1 (????-??-??):
5
908c8647
JD
6** Minor improvements have been made to the manual.
7
b2bc8a1b 8* Changes in version 2.5 (2011-05-14):
006faedf 9
eb8c66bb
JD
10** Grammar symbol names can now contain non-initial dashes:
11
12 Consistently with directives (such as %error-verbose) and with
13 %define variables (e.g. push-pull), grammar symbol names may contain
14 dashes in any position except the beginning. This is a GNU
15 extension over POSIX Yacc. Thus, use of this extension is reported
16 by -Wyacc and rejected in Yacc mode (--yacc).
17
3b1977ea 18** Named references:
348f5608
AR
19
20 Historically, Yacc and Bison have supported positional references
21 ($n, $$) to allow access to symbol values from inside of semantic
22 actions code.
23
24 Starting from this version, Bison can also accept named references.
25 When no ambiguity is possible, original symbol names may be used
26 as named references:
27
ac3297d5 28 if_stmt : "if" cond_expr "then" then_stmt ';'
348f5608
AR
29 { $if_stmt = mk_if_stmt($cond_expr, $then_stmt); }
30
31 In the more common case, explicit names may be declared:
32
ac3297d5 33 stmt[res] : "if" expr[cond] "then" stmt[then] "else" stmt[else] ';'
348f5608
AR
34 { $res = mk_if_stmt($cond, $then, $else); }
35
f840c05a 36 Location information is also accessible using @name syntax. When
348f5608
AR
37 accessing symbol names containing dots or dashes, explicit bracketing
38 ($[sym.1]) must be used.
39
f840c05a 40 These features are experimental in this version. More user feedback
348f5608
AR
41 will help to stabilize them.
42
3b1977ea 43** IELR(1) and canonical LR(1):
34a6c2d1
JD
44
45 IELR(1) is a minimal LR(1) parser table generation algorithm. That
46 is, given any context-free grammar, IELR(1) generates parser tables
723206df 47 with the full language-recognition power of canonical LR(1) but with
3b1977ea
JD
48 nearly the same number of parser states as LALR(1). This reduction
49 in parser states is often an order of magnitude. More importantly,
34a6c2d1
JD
50 because canonical LR(1)'s extra parser states may contain duplicate
51 conflicts in the case of non-LR(1) grammars, the number of conflicts
52 for IELR(1) is often an order of magnitude less as well. This can
53 significantly reduce the complexity of developing of a grammar.
54
55 Bison can now generate IELR(1) and canonical LR(1) parser tables in
56 place of its traditional LALR(1) parser tables, which remain the
57 default. You can specify the type of parser tables in the grammar
58 file with these directives:
59
f37495f6
JD
60 %define lr.type lalr
61 %define lr.type ielr
62 %define lr.type canonical-lr
34a6c2d1 63
6f04ee6c
JD
64 The default-reduction optimization in the parser tables can also be
65 adjusted using `%define lr.default-reductions'. For details on both
66 of these features, see the new section `Tuning LR' in the Bison
67 manual.
34a6c2d1
JD
68
69 These features are experimental. More user feedback will help to
70 stabilize them.
71
6f04ee6c 72** LAC (Lookahead Correction) for syntax error handling:
4c38b19e
JD
73
74 Canonical LR, IELR, and LALR can suffer from a couple of problems
75 upon encountering a syntax error. First, the parser might perform
76 additional parser stack reductions before discovering the syntax
6f04ee6c 77 error. Such reductions can perform user semantic actions that are
4c38b19e
JD
78 unexpected because they are based on an invalid token, and they
79 cause error recovery to begin in a different syntactic context than
80 the one in which the invalid token was encountered. Second, when
6f04ee6c
JD
81 verbose error messages are enabled (with %error-verbose or the
82 obsolete `#define YYERROR_VERBOSE'), the expected token list in the
83 syntax error message can both contain invalid tokens and omit valid
84 tokens.
4c38b19e
JD
85
86 The culprits for the above problems are %nonassoc, default
87 reductions in inconsistent states, and parser state merging. Thus,
88 IELR and LALR suffer the most. Canonical LR can suffer only if
89 %nonassoc is used or if default reductions are enabled for
90 inconsistent states.
91
6f04ee6c
JD
92 LAC is a new mechanism within the parsing algorithm that solves
93 these problems for canonical LR, IELR, and LALR without sacrificing
94 %nonassoc, default reductions, or state merging. When LAC is in
95 use, canonical LR and IELR behave almost exactly the same for both
96 syntactically acceptable and syntactically unacceptable input.
4c38b19e
JD
97 While LALR still does not support the full language-recognition
98 power of canonical LR and IELR, LAC at least enables LALR's syntax
99 error handling to correctly reflect LALR's language-recognition
100 power.
101
102 Currently, LAC is only supported for deterministic parsers in C.
103 You can enable LAC with the following directive:
104
105 %define parse.lac full
106
6f04ee6c
JD
107 See the new section `LAC' in the Bison manual for additional
108 details including a few caveats.
4c38b19e
JD
109
110 LAC is an experimental feature. More user feedback will help to
111 stabilize it.
112
6ee881f6 113** %define improvements:
f37495f6 114
3b1977ea 115*** Can now be invoked via the command line:
4ecd3681 116
34d41938 117 Each of these command-line options
4ecd3681 118
34d41938
JD
119 -D NAME[=VALUE]
120 --define=NAME[=VALUE]
121
122 -F NAME[=VALUE]
123 --force-define=NAME[=VALUE]
4ecd3681
JD
124
125 is equivalent to this grammar file declaration
126
34d41938 127 %define NAME ["VALUE"]
4ecd3681 128
34d41938
JD
129 except that the manner in which Bison processes multiple definitions
130 for the same NAME differs. Most importantly, -F and --force-define
131 quietly override %define, but -D and --define do not. For further
ac3297d5 132 details, see the section `Bison Options' in the Bison manual.
4ecd3681 133
3b1977ea 134*** Variables renamed:
812775a0
JD
135
136 The following %define variables
137
138 api.push_pull
139 lr.keep_unreachable_states
140
141 have been renamed to
142
143 api.push-pull
144 lr.keep-unreachable-states
145
146 The old names are now deprecated but will be maintained indefinitely
147 for backward compatibility.
148
723206df 149*** Values no longer need to be quoted in the grammar file:
f37495f6
JD
150
151 If a %define value is an identifier, it no longer needs to be placed
152 within quotations marks. For example,
153
154 %define api.push-pull "push"
155
156 can be rewritten as
157
158 %define api.push-pull push
159
6ee881f6 160*** Unrecognized variables are now errors not warnings.
c046698e 161
6ee881f6
JD
162*** Multiple invocations for any variable is now an error not a warning.
163
164** Unrecognized %code qualifiers are now errors not warnings.
165
166** Character literals not of length one:
167
168 Previously, Bison quietly converted all character literals to length
169 one. For example, without warning, Bison interpreted the operators in
170 the following grammar to be the same token:
171
172 exp: exp '++'
173 | exp '+' exp
174 ;
175
176 Bison now warns when a character literal is not of length one. In
177 some future release, Bison will start reporting an error instead.
178
179** Destructor calls fixed for lookaheads altered in semantic actions:
180
181 Previously for deterministic parsers in C, if a user semantic action
182 altered yychar, the parser in some cases used the old yychar value to
183 determine which destructor to call for the lookahead upon a syntax
184 error or upon parser return. This bug has been fixed.
185
186** C++ parsers use YYRHSLOC:
187
188 Similarly to the C parsers, the C++ parsers now define the YYRHSLOC
189 macro and use it in the default YYLLOC_DEFAULT. You are encouraged
ac3297d5
JD
190 to use it. If, for instance, your location structure has `first'
191 and `last' members, instead of
6ee881f6
JD
192
193 # define YYLLOC_DEFAULT(Current, Rhs, N) \
194 do \
195 if (N) \
196 { \
197 (Current).first = (Rhs)[1].location.first; \
198 (Current).last = (Rhs)[N].location.last; \
199 } \
200 else \
201 { \
202 (Current).first = (Current).last = (Rhs)[0].location.last; \
203 } \
204 while (false)
205
206 use:
207
208 # define YYLLOC_DEFAULT(Current, Rhs, N) \
209 do \
210 if (N) \
211 { \
212 (Current).first = YYRHSLOC (Rhs, 1).first; \
213 (Current).last = YYRHSLOC (Rhs, N).last; \
214 } \
215 else \
216 { \
217 (Current).first = (Current).last = YYRHSLOC (Rhs, 0).last; \
218 } \
219 while (false)
220
221** YYLLOC_DEFAULT in C++:
222
223 The default implementation of YYLLOC_DEFAULT used to be issued in
224 the header file. It is now output in the implementation file, after
225 the user %code sections so that its #ifndef guard does not try to
226 override the user's YYLLOC_DEFAULT if provided.
c046698e 227
3b1977ea 228** YYFAIL now produces warnings and Java parsers no longer implement it:
62efdd2a
JD
229
230 YYFAIL has existed for many years as an undocumented feature of
231 deterministic parsers in C generated by Bison. More recently, it was
232 a documented feature of Bison's experimental Java parsers. As
233 promised in Bison 2.4.2's NEWS entry, any appearance of YYFAIL in a
234 semantic action now produces a deprecation warning, and Java parsers
235 no longer implement YYFAIL at all. For further details, including a
236 discussion of how to suppress C preprocessor warnings about YYFAIL
237 being unused, see the Bison 2.4.2 NEWS entry.
238
3b1977ea 239** Temporary hack for adding a semicolon to the user action:
a2d05674
JD
240
241 Previously, Bison appended a semicolon to every user action for
242 reductions when the output language defaulted to C (specifically, when
243 neither %yacc, %language, %skeleton, or equivalent command-line
244 options were specified). This allowed actions such as
245
246 exp: exp "+" exp { $$ = $1 + $3 };
247
248 instead of
249
250 exp: exp "+" exp { $$ = $1 + $3; };
251
252 As a first step in removing this misfeature, Bison now issues a
253 warning when it appends a semicolon. Moreover, in cases where Bison
254 cannot easily determine whether a semicolon is needed (for example, an
255 action ending with a cpp directive or a braced compound initializer),
256 it no longer appends one. Thus, the C compiler might now complain
257 about a missing semicolon where it did not before. Future releases of
258 Bison will cease to append semicolons entirely.
259
095a1d11
JD
260** Verbose syntax error message fixes:
261
6f04ee6c
JD
262 When %error-verbose or the obsolete `#define YYERROR_VERBOSE' is
263 specified, syntax error messages produced by the generated parser
264 include the unexpected token as well as a list of expected tokens.
265 The effect of %nonassoc on these verbose messages has been corrected
266 in two ways, but a more complete fix requires LAC, described above:
095a1d11
JD
267
268*** When %nonassoc is used, there can exist parser states that accept no
269 tokens, and so the parser does not always require a lookahead token
270 in order to detect a syntax error. Because no unexpected token or
271 expected tokens can then be reported, the verbose syntax error
272 message described above is suppressed, and the parser instead
ac3297d5 273 reports the simpler message, `syntax error'. Previously, this
095a1d11
JD
274 suppression was sometimes erroneously triggered by %nonassoc when a
275 lookahead was actually required. Now verbose messages are
276 suppressed only when all previous lookaheads have already been
277 shifted or discarded.
278
279*** Previously, the list of expected tokens erroneously included tokens
280 that would actually induce a syntax error because conflicts for them
281 were resolved with %nonassoc in the current parser state. Such
282 tokens are now properly omitted from the list.
283
284*** Expected token lists are still often wrong due to state merging
4c38b19e
JD
285 (from LALR or IELR) and default reductions, which can both add
286 invalid tokens and subtract valid tokens. Canonical LR almost
287 completely fixes this problem by eliminating state merging and
288 default reductions. However, there is one minor problem left even
289 when using canonical LR and even after the fixes above. That is,
290 if the resolution of a conflict with %nonassoc appears in a later
291 parser state than the one at which some syntax error is
292 discovered, the conflicted token is still erroneously included in
293 the expected token list. Bison's new LAC implementation,
294 described above, eliminates this problem and the need for
295 canonical LR. However, LAC is still experimental and is disabled
296 by default.
1fa30307 297
4c2a6e42
JD
298** Java skeleton fixes:
299
300*** A location handling bug has been fixed.
301
302*** The top element of each of the value stack and location stack is now
303 cleared when popped so that it can be garbage collected.
77768165 304
f0e2c228
JD
305*** Parser traces now print the top element of the stack.
306
248dba53
JD
307** -W/--warnings fixes:
308
ac3297d5 309*** Bison now properly recognizes the `no-' versions of categories:
248dba53
JD
310
311 For example, given the following command line, Bison now enables all
312 warnings except warnings for incompatibilities with POSIX Yacc:
313
314 bison -Wall,no-yacc gram.y
315
6f8bdce2
JD
316*** Bison now treats S/R and R/R conflicts like other warnings:
317
318 Previously, conflict reports were independent of Bison's normal
319 warning system. Now, Bison recognizes the warning categories
ac3297d5 320 `conflicts-sr' and `conflicts-rr'. This change has important
6f8bdce2
JD
321 consequences for the -W and --warnings command-line options. For
322 example:
323
324 bison -Wno-conflicts-sr gram.y # S/R conflicts not reported
325 bison -Wno-conflicts-rr gram.y # R/R conflicts not reported
326 bison -Wnone gram.y # no conflicts are reported
327 bison -Werror gram.y # any conflict is an error
328
329 However, as before, if the %expect or %expect-rr directive is
330 specified, an unexpected number of conflicts is an error, and an
331 expected number of conflicts is not reported, so -W and --warning
332 then have no effect on the conflict report.
333
ac3297d5 334*** The `none' category no longer disables a preceding `error':
dab96632
JD
335
336 For example, for the following command line, Bison now reports
337 errors instead of warnings for incompatibilities with POSIX Yacc:
338
339 bison -Werror,none,yacc gram.y
340
b2bc8a1b 341*** The `none' category now disables all Bison warnings:
8ffd7912 342
ac3297d5 343 Previously, the `none' category disabled only Bison warnings for
8ffd7912
JD
344 which there existed a specific -W/--warning category. However,
345 given the following command line, Bison is now guaranteed to
346 suppress all warnings:
347
348 bison -Wnone gram.y
349
9d6af153
JD
350** Precedence directives can now assign token number 0:
351
352 Since Bison 2.3b, which restored the ability of precedence
353 directives to assign token numbers, doing so for token number 0 has
354 produced an assertion failure. For example:
355
356 %left END 0
357
358 This bug has been fixed.
359
7527c744 360* Changes in version 2.4.3 (2010-08-05):
7a9c3cb3 361
4631c34f
JD
362** Bison now obeys -Werror and --warnings=error for warnings about
363 grammar rules that are useless in the parser due to conflicts.
364
7a9c3cb3
JD
365** Problems with spawning M4 on at least FreeBSD 8 and FreeBSD 9 have
366 been fixed.
367
132247cd
JD
368** Failures in the test suite for GCC 4.5 have been fixed.
369
88bb35d6
JD
370** Failures in the test suite for some versions of Sun Studio C++ have
371 been fixed.
372
43fdc9fd
JD
373** Contrary to Bison 2.4.2's NEWS entry, it has been decided that
374 warnings about undefined %prec identifiers will not be converted to
375 errors in Bison 2.5. They will remain warnings, which should be
376 sufficient for POSIX while avoiding backward compatibility issues.
377
cf22447c
JD
378** Minor documentation fixes.
379
ea66d039 380* Changes in version 2.4.2 (2010-03-20):
8defe11b 381
47fa5747
JD
382** Some portability problems that resulted in failures and livelocks
383 in the test suite on some versions of at least Solaris, AIX, HP-UX,
ea66d039
JD
384 RHEL4, and Tru64 have been addressed. As a result, fatal Bison
385 errors should no longer cause M4 to report a broken pipe on the
47fa5747
JD
386 affected platforms.
387
2c203528
JD
388** `%prec IDENTIFIER' requires IDENTIFIER to be defined separately.
389
390 POSIX specifies that an error be reported for any identifier that does
391 not appear on the LHS of a grammar rule and that is not defined by
392 %token, %left, %right, or %nonassoc. Bison 2.3b and later lost this
393 error report for the case when an identifier appears only after a
394 %prec directive. It is now restored. However, for backward
395 compatibility with recent Bison releases, it is only a warning for
396 now. In Bison 2.5 and later, it will return to being an error.
43fdc9fd
JD
397 [Between the 2.4.2 and 2.4.3 releases, it was decided that this
398 warning will not be converted to an error in Bison 2.5.]
2c203528 399
c5196098
EB
400** Detection of GNU M4 1.4.6 or newer during configure is improved.
401
34731471
JD
402** Warnings from gcc's -Wundef option about undefined YYENABLE_NLS,
403 YYLTYPE_IS_TRIVIAL, and __STRICT_ANSI__ in C/C++ parsers are now
404 avoided.
af2ffe5c 405
966aba65
JD
406** %code is now a permanent feature.
407
408 A traditional Yacc prologue directive is written in the form:
409
410 %{CODE%}
411
412 To provide a more flexible alternative, Bison 2.3b introduced the
413 %code directive with the following forms for C/C++:
414
415 %code {CODE}
416 %code requires {CODE}
417 %code provides {CODE}
418 %code top {CODE}
419
420 These forms are now considered permanent features of Bison. See the
421 %code entries in the section "Bison Declaration Summary" in the Bison
422 manual for a summary of their functionality. See the section
423 "Prologue Alternatives" for a detailed discussion including the
424 advantages of %code over the traditional Yacc prologue directive.
425
426 Bison's Java feature as a whole including its current usage of %code
427 is still considered experimental.
428
41d35e54
JD
429** YYFAIL is deprecated and will eventually be removed.
430
431 YYFAIL has existed for many years as an undocumented feature of
432 deterministic parsers in C generated by Bison. Previously, it was
433 documented for Bison's experimental Java parsers. YYFAIL is no longer
434 documented for Java parsers and is formally deprecated in both cases.
435 Users are strongly encouraged to migrate to YYERROR, which is
436 specified by POSIX.
437
438 Like YYERROR, you can invoke YYFAIL from a semantic action in order to
439 induce a syntax error. The most obvious difference from YYERROR is
440 that YYFAIL will automatically invoke yyerror to report the syntax
441 error so that you don't have to. However, there are several other
442 subtle differences between YYERROR and YYFAIL, and YYFAIL suffers from
443 inherent flaws when %error-verbose or `#define YYERROR_VERBOSE' is
444 used. For a more detailed discussion, see:
445
446 http://lists.gnu.org/archive/html/bison-patches/2009-12/msg00024.html
447
448 The upcoming Bison 2.5 will remove YYFAIL from Java parsers, but
449 deterministic parsers in C will continue to implement it. However,
450 because YYFAIL is already flawed, it seems futile to try to make new
451 Bison features compatible with it. Thus, during parser generation,
452 Bison 2.5 will produce a warning whenever it discovers YYFAIL in a
453 rule action. In a later release, YYFAIL will be disabled for
454 %error-verbose and `#define YYERROR_VERBOSE'. Eventually, YYFAIL will
455 be removed altogether.
456
457 There exists at least one case where Bison 2.5's YYFAIL warning will
458 be a false positive. Some projects add phony uses of YYFAIL and other
459 Bison-defined macros for the sole purpose of suppressing C
460 preprocessor warnings (from GCC cpp's -Wunused-macros, for example).
461 To avoid Bison's future warning, such YYFAIL uses can be moved to the
462 epilogue (that is, after the second `%%') in the Bison input file. In
463 this release (2.4.2), Bison already generates its own code to suppress
464 C preprocessor warnings for YYFAIL, so projects can remove their own
465 phony uses of YYFAIL if compatibility with Bison releases prior to
466 2.4.2 is not necessary.
467
dac8cc0d
AD
468** Internationalization.
469
470 Fix a regression introduced in Bison 2.4: Under some circumstances,
471 message translations were not installed although supported by the
472 host system.
473
41930e7a 474* Changes in version 2.4.1 (2008-12-11):
c9ba9e59 475
a957d06c
JD
476** In the GLR defines file, unexpanded M4 macros in the yylval and yylloc
477 declarations have been fixed.
478
738cde3e
AD
479** Temporary hack for adding a semicolon to the user action.
480
481 Bison used to prepend a trailing semicolon at the end of the user
482 action for reductions. This allowed actions such as
483
484 exp: exp "+" exp { $$ = $1 + $3 };
485
486 instead of
487
488 exp: exp "+" exp { $$ = $1 + $3; };
489
d07932ef
JD
490 Some grammars still depend on this `feature'. Bison 2.4.1 restores
491 the previous behavior in the case of C output (specifically, when
492 neither %language or %skeleton or equivalent command-line options
493 are used) to leave more time for grammars depending on the old
494 behavior to be adjusted. Future releases of Bison will disable this
495 feature.
a957d06c
JD
496
497** A few minor improvements to the Bison manual.
c9ba9e59 498
d6fb461d 499* Changes in version 2.4 (2008-11-02):
7bd1665a 500
d6fb461d 501** %language is an experimental feature.
ed4d67dc
JD
502
503 We first introduced this feature in test release 2.3b as a cleaner
504 alternative to %skeleton. Since then, we have discussed the possibility of
505 modifying its effect on Bison's output file names. Thus, in this release,
506 we consider %language to be an experimental feature that will likely evolve
507 in future releases.
7bd1665a 508
d6fb461d 509** Forward compatibility with GNU M4 has been improved.
241fda7a 510
d6fb461d 511** Several bugs in the C++ skeleton and the experimental Java skeleton have been
241fda7a
JD
512 fixed.
513
d6fb461d 514* Changes in version 2.3b (2008-05-27):
35fe0834 515
d6fb461d 516** The quotes around NAME that used to be required in the following directive
d9df47b6
JD
517 are now deprecated:
518
519 %define NAME "VALUE"
520
d6fb461d 521** The directive `%pure-parser' is now deprecated in favor of:
d9df47b6
JD
522
523 %define api.pure
524
525 which has the same effect except that Bison is more careful to warn about
526 unreasonable usage in the latter case.
527
d6fb461d 528** Push Parsing
c373bf8b
JD
529
530 Bison can now generate an LALR(1) parser in C with a push interface. That
ef1b4273 531 is, instead of invoking `yyparse', which pulls tokens from `yylex', you can
c373bf8b
JD
532 push one token at a time to the parser using `yypush_parse', which will
533 return to the caller after processing each token. By default, the push
534 interface is disabled. Either of the following directives will enable it:
535
536 %define api.push_pull "push" // Just push; does not require yylex.
537 %define api.push_pull "both" // Push and pull; requires yylex.
538
539 See the new section `A Push Parser' in the Bison manual for details.
540
59da312b
JD
541 The current push parsing interface is experimental and may evolve. More user
542 feedback will help to stabilize it.
543
d6fb461d 544** The -g and --graph options now output graphs in Graphviz DOT format,
8e55b3aa
JD
545 not VCG format. Like --graph, -g now also takes an optional FILE argument
546 and thus cannot be bundled with other short options.
c373bf8b 547
d6fb461d 548** Java
59da312b
JD
549
550 Bison can now generate an LALR(1) parser in Java. The skeleton is
551 `data/lalr1.java'. Consider using the new %language directive instead of
552 %skeleton to select it.
553
554 See the new section `Java Parsers' in the Bison manual for details.
555
556 The current Java interface is experimental and may evolve. More user
557 feedback will help to stabilize it.
558
d6fb461d 559** %language
59da312b
JD
560
561 This new directive specifies the programming language of the generated
d43f77e7
PB
562 parser, which can be C (the default), C++, or Java. Besides the skeleton
563 that Bison uses, the directive affects the names of the generated files if
564 the grammar file's name ends in ".y".
59da312b 565
d6fb461d 566** XML Automaton Report
59da312b
JD
567
568 Bison can now generate an XML report of the LALR(1) automaton using the new
569 `--xml' option. The current XML schema is experimental and may evolve. More
570 user feedback will help to stabilize it.
c373bf8b 571
d6fb461d 572** The grammar file may now specify the name of the parser header file using
c373bf8b
JD
573 %defines. For example:
574
575 %defines "parser.h"
576
d6fb461d 577** When reporting useless rules, useless nonterminals, and unused terminals,
d80fb37a
JD
578 Bison now employs the terms "useless in grammar" instead of "useless",
579 "useless in parser" instead of "never reduced", and "unused in grammar"
580 instead of "unused".
cff03fb2 581
d6fb461d 582** Unreachable State Removal
c373bf8b
JD
583
584 Previously, Bison sometimes generated parser tables containing unreachable
31984206
JD
585 states. A state can become unreachable during conflict resolution if Bison
586 disables a shift action leading to it from a predecessor state. Bison now:
75ad86ee
JD
587
588 1. Removes unreachable states.
589
590 2. Does not report any conflicts that appeared in unreachable states.
591 WARNING: As a result, you may need to update %expect and %expect-rr
592 directives in existing grammar files.
593
594 3. For any rule used only in such states, Bison now reports the rule as
cff03fb2 595 "useless in parser due to conflicts".
75ad86ee 596
31984206
JD
597 This feature can be disabled with the following directive:
598
599 %define lr.keep_unreachable_states
600
601 See the %define entry in the `Bison Declaration Summary' in the Bison manual
602 for further discussion.
603
d6fb461d 604** Lookahead Set Correction in the `.output' Report
b1cc23c4 605
c373bf8b 606 When instructed to generate a `.output' file including lookahead sets
88c78747
JD
607 (using `--report=lookahead', for example), Bison now prints each reduction's
608 lookahead set only next to the associated state's one item that (1) is
609 associated with the same rule as the reduction and (2) has its dot at the end
610 of its RHS. Previously, Bison also erroneously printed the lookahead set
611 next to all of the state's other items associated with the same rule. This
612 bug affected only the `.output' file and not the generated parser source
613 code.
614
d6fb461d 615** --report-file=FILE is a new option to override the default `.output' file
59da312b 616 name.
1bb2bd75 617
d6fb461d 618** The `=' that used to be required in the following directives is now
02975b9a
JD
619 deprecated:
620
621 %file-prefix "parser"
622 %name-prefix "c_"
623 %output "parser.c"
624
d6fb461d 625** An Alternative to `%{...%}' -- `%code QUALIFIER {CODE}'
c373bf8b
JD
626
627 Bison 2.3a provided a new set of directives as a more flexible alternative to
8e0a5e9e
JD
628 the traditional Yacc prologue blocks. Those have now been consolidated into
629 a single %code directive with an optional qualifier field, which identifies
630 the purpose of the code and thus the location(s) where Bison should generate
631 it:
632
16dc6a9e
JD
633 1. `%code {CODE}' replaces `%after-header {CODE}'
634 2. `%code requires {CODE}' replaces `%start-header {CODE}'
635 3. `%code provides {CODE}' replaces `%end-header {CODE}'
636 4. `%code top {CODE}' replaces `%before-header {CODE}'
8e0a5e9e 637
61fee93e
JD
638 See the %code entries in section `Bison Declaration Summary' in the Bison
639 manual for a summary of the new functionality. See the new section `Prologue
8e0a5e9e
JD
640 Alternatives' for a detailed discussion including the advantages of %code
641 over the traditional Yacc prologues.
642
643 The prologue alternatives are experimental. More user feedback will help to
644 determine whether they should become permanent features.
645
d6fb461d 646** Revised warning: unset or unused mid-rule values
17bd8a73
JD
647
648 Since Bison 2.2, Bison has warned about mid-rule values that are set but not
649 used within any of the actions of the parent rule. For example, Bison warns
650 about unused $2 in:
651
652 exp: '1' { $$ = 1; } '+' exp { $$ = $1 + $4; };
653
654 Now, Bison also warns about mid-rule values that are used but not set. For
655 example, Bison warns about unset $$ in the mid-rule action in:
656
657 exp: '1' { $1 = 1; } '+' exp { $$ = $2 + $4; };
658
659 However, Bison now disables both of these warnings by default since they
660 sometimes prove to be false alarms in existing grammars employing the Yacc
661 constructs $0 or $-N (where N is some positive integer).
662
59da312b 663 To enable these warnings, specify the option `--warnings=midrule-values' or
17bd8a73
JD
664 `-W', which is a synonym for `--warnings=all'.
665
d6fb461d 666** Default %destructor or %printer with `<*>' or `<>'
c373bf8b
JD
667
668 Bison now recognizes two separate kinds of default %destructor's and
12e35840
JD
669 %printer's:
670
671 1. Place `<*>' in a %destructor/%printer symbol list to define a default
672 %destructor/%printer for all grammar symbols for which you have formally
673 declared semantic type tags.
674
3ebecc24 675 2. Place `<>' in a %destructor/%printer symbol list to define a default
12e35840
JD
676 %destructor/%printer for all grammar symbols without declared semantic
677 type tags.
678
679 Bison no longer supports the `%symbol-default' notation from Bison 2.3a.
3ebecc24 680 `<*>' and `<>' combined achieve the same effect with one exception: Bison no
12e35840
JD
681 longer applies any %destructor to a mid-rule value if that mid-rule value is
682 not actually ever referenced using either $$ or $n in a semantic action.
683
85894313
JD
684 The default %destructor's and %printer's are experimental. More user
685 feedback will help to determine whether they should become permanent
686 features.
687
12e35840
JD
688 See the section `Freeing Discarded Symbols' in the Bison manual for further
689 details.
690
d6fb461d 691** %left, %right, and %nonassoc can now declare token numbers. This is required
ab7f29f8
JD
692 by POSIX. However, see the end of section `Operator Precedence' in the Bison
693 manual for a caveat concerning the treatment of literal strings.
694
d6fb461d 695** The nonfunctional --no-parser, -n, and %no-parser options have been
b1cc23c4
JD
696 completely removed from Bison.
697
d6fb461d 698* Changes in version 2.3a, 2006-09-13:
742e4900 699
d6fb461d 700** Instead of %union, you can define and use your own union type
ddc8ede1
PE
701 YYSTYPE if your grammar contains at least one <type> tag.
702 Your YYSTYPE need not be a macro; it can be a typedef.
703 This change is for compatibility with other Yacc implementations,
704 and is required by POSIX.
705
d6fb461d 706** Locations columns and lines start at 1.
cd48d21d
AD
707 In accordance with the GNU Coding Standards and Emacs.
708
d6fb461d 709** You may now declare per-type and default %destructor's and %printer's:
ec5479ce
JD
710
711 For example:
712
b2a0b7ca
JD
713 %union { char *string; }
714 %token <string> STRING1
715 %token <string> STRING2
716 %type <string> string1
717 %type <string> string2
718 %union { char character; }
719 %token <character> CHR
720 %type <character> chr
721 %destructor { free ($$); } %symbol-default
722 %destructor { free ($$); printf ("%d", @$.first_line); } STRING1 string1
723 %destructor { } <character>
724
725 guarantees that, when the parser discards any user-defined symbol that has a
726 semantic type tag other than `<character>', it passes its semantic value to
727 `free'. However, when the parser discards a `STRING1' or a `string1', it
728 also prints its line number to `stdout'. It performs only the second
729 `%destructor' in this case, so it invokes `free' only once.
ec5479ce 730
85894313
JD
731 [Although we failed to mention this here in the 2.3a release, the default
732 %destructor's and %printer's were experimental, and they were rewritten in
733 future versions.]
734
d6fb461d 735** Except for LALR(1) parsers in C with POSIX Yacc emulation enabled (with `-y',
b931235e
JD
736 `--yacc', or `%yacc'), Bison no longer generates #define statements for
737 associating token numbers with token names. Removing the #define statements
738 helps to sanitize the global namespace during preprocessing, but POSIX Yacc
739 requires them. Bison still generates an enum for token names in all cases.
740
d6fb461d 741** Handling of traditional Yacc prologue blocks is now more consistent but
34f98f46 742 potentially incompatible with previous releases of Bison.
9bc0dd67
JD
743
744 As before, you declare prologue blocks in your grammar file with the
745 `%{ ... %}' syntax. To generate the pre-prologue, Bison concatenates all
34f98f46
JD
746 prologue blocks that you've declared before the first %union. To generate
747 the post-prologue, Bison concatenates all prologue blocks that you've
ddc8ede1 748 declared after the first %union.
9bc0dd67 749
34f98f46 750 Previous releases of Bison inserted the pre-prologue into both the header
9bc0dd67
JD
751 file and the code file in all cases except for LALR(1) parsers in C. In the
752 latter case, Bison inserted it only into the code file. For parsers in C++,
753 the point of insertion was before any token definitions (which associate
754 token numbers with names). For parsers in C, the point of insertion was
755 after the token definitions.
756
757 Now, Bison never inserts the pre-prologue into the header file. In the code
758 file, it always inserts it before the token definitions.
759
d6fb461d 760** Bison now provides a more flexible alternative to the traditional Yacc
34f98f46
JD
761 prologue blocks: %before-header, %start-header, %end-header, and
762 %after-header.
763
764 For example, the following declaration order in the grammar file reflects the
765 order in which Bison will output these code blocks. However, you are free to
766 declare these code blocks in your grammar file in whatever order is most
767 convenient for you:
768
769 %before-header {
770 /* Bison treats this block like a pre-prologue block: it inserts it into
771 * the code file before the contents of the header file. It does *not*
772 * insert it into the header file. This is a good place to put
773 * #include's that you want at the top of your code file. A common
774 * example is `#include "system.h"'. */
775 }
776 %start-header {
777 /* Bison inserts this block into both the header file and the code file.
778 * In both files, the point of insertion is before any Bison-generated
779 * token, semantic type, location type, and class definitions. This is a
780 * good place to define %union dependencies, for example. */
9bc0dd67
JD
781 }
782 %union {
34f98f46
JD
783 /* Unlike the traditional Yacc prologue blocks, the output order for the
784 * new %*-header blocks is not affected by their declaration position
785 * relative to any %union in the grammar file. */
9bc0dd67 786 }
34f98f46
JD
787 %end-header {
788 /* Bison inserts this block into both the header file and the code file.
789 * In both files, the point of insertion is after the Bison-generated
790 * definitions. This is a good place to declare or define public
791 * functions or data structures that depend on the Bison-generated
792 * definitions. */
9bc0dd67 793 }
34f98f46
JD
794 %after-header {
795 /* Bison treats this block like a post-prologue block: it inserts it into
796 * the code file after the contents of the header file. It does *not*
797 * insert it into the header file. This is a good place to declare or
798 * define internal functions or data structures that depend on the
799 * Bison-generated definitions. */
800 }
801
802 If you have multiple occurrences of any one of the above declarations, Bison
803 will concatenate the contents in declaration order.
9bc0dd67 804
85894313
JD
805 [Although we failed to mention this here in the 2.3a release, the prologue
806 alternatives were experimental, and they were rewritten in future versions.]
807
d6fb461d 808** The option `--report=look-ahead' has been changed to `--report=lookahead'.
9e6e7ed2
PE
809 The old spelling still works, but is not documented and may be removed
810 in a future release.
742e4900 811
d6fb461d 812* Changes in version 2.3, 2006-06-05:
4ad3ed84 813
d6fb461d 814** GLR grammars should now use `YYRECOVERING ()' instead of `YYRECOVERING',
4ad3ed84
PE
815 for compatibility with LALR(1) grammars.
816
d6fb461d 817** It is now documented that any definition of YYSTYPE or YYLTYPE should
4ad3ed84
PE
818 be to a type name that does not contain parentheses or brackets.
819
d6fb461d 820* Changes in version 2.2, 2006-05-19:
193d7c70 821
d6fb461d 822** The distribution terms for all Bison-generated parsers now permit
193d7c70
PE
823 using the parsers in nonfree programs. Previously, this permission
824 was granted only for Bison-generated LALR(1) parsers in C.
5f4236a0 825
d6fb461d 826** %name-prefix changes the namespace name in C++ outputs.
aa08666d 827
d6fb461d 828** The C++ parsers export their token_type.
5f4236a0 829
d6fb461d 830** Bison now allows multiple %union declarations, and concatenates
d6ca7905
PE
831 their contents together.
832
d6fb461d 833** New warning: unused values
4d7bc38c
PE
834 Right-hand side symbols whose values are not used are reported,
835 if the symbols have destructors. For instance:
affac613 836
8f3596a6 837 exp: exp "?" exp ":" exp { $1 ? $1 : $3; }
721be13c
PE
838 | exp "+" exp
839 ;
affac613 840
8f3596a6
AD
841 will trigger a warning about $$ and $5 in the first rule, and $3 in
842 the second ($1 is copied to $$ by the default rule). This example
4e26c69e 843 most likely contains three errors, and could be rewritten as:
affac613 844
4e26c69e
PE
845 exp: exp "?" exp ":" exp
846 { $$ = $1 ? $3 : $5; free ($1 ? $5 : $3); free ($1); }
847 | exp "+" exp
848 { $$ = $1 ? $1 : $3; if ($1) free ($3); }
721be13c 849 ;
affac613 850
4e26c69e
PE
851 However, if the original actions were really intended, memory leaks
852 and all, the warnings can be suppressed by letting Bison believe the
853 values are used, e.g.:
721be13c 854
8f3596a6 855 exp: exp "?" exp ":" exp { $1 ? $1 : $3; (void) ($$, $5); }
721be13c
PE
856 | exp "+" exp { $$ = $1; (void) $3; }
857 ;
858
84866159
AD
859 If there are mid-rule actions, the warning is issued if no action
860 uses it. The following triggers no warning: $1 and $3 are used.
861
862 exp: exp { push ($1); } '+' exp { push ($3); sum (); };
863
721be13c
PE
864 The warning is intended to help catching lost values and memory leaks.
865 If a value is ignored, its associated memory typically is not reclaimed.
affac613 866
d6fb461d 867** %destructor vs. YYABORT, YYACCEPT, and YYERROR.
9d9b8b70
PE
868 Destructors are now called when user code invokes YYABORT, YYACCEPT,
869 and YYERROR, for all objects on the stack, other than objects
870 corresponding to the right-hand side of the current rule.
a85284cf 871
d6fb461d 872** %expect, %expect-rr
035aa4a0
PE
873 Incorrect numbers of expected conflicts are now actual errors,
874 instead of warnings.
875
d6fb461d 876** GLR, YACC parsers.
4e26c69e
PE
877 The %parse-params are available in the destructors (and the
878 experimental printers) as per the documentation.
4b367315 879
d6fb461d 880** Bison now warns if it finds a stray `$' or `@' in an action.
ad6a9b97 881
d6fb461d 882** %require "VERSION"
4e26c69e
PE
883 This specifies that the grammar file depends on features implemented
884 in Bison version VERSION or higher.
b50d2359 885
d6fb461d 886** lalr1.cc: The token and value types are now class members.
e14d0ab6
AD
887 The tokens were defined as free form enums and cpp macros. YYSTYPE
888 was defined as a free form union. They are now class members:
fb9712a9
AD
889 tokens are enumerations of the `yy::parser::token' struct, and the
890 semantic values have the `yy::parser::semantic_type' type.
891
892 If you do not want or can update to this scheme, the directive
893 `%define "global_tokens_and_yystype" "1"' triggers the global
b50d2359
AD
894 definition of tokens and YYSTYPE. This change is suitable both
895 for previous releases of Bison, and this one.
fb9712a9 896
b50d2359 897 If you wish to update, then make sure older version of Bison will
ab8d9dc5 898 fail using `%require "2.2"'.
fb9712a9 899
d6fb461d 900** DJGPP support added.
193d7c70 901\f
d6fb461d 902* Changes in version 2.1, 2005-09-16:
1ce59070 903
d6fb461d 904** The C++ lalr1.cc skeleton supports %lex-param.
e14d0ab6 905
d6fb461d 906** Bison-generated parsers now support the translation of diagnostics like
baf785db
PE
907 "syntax error" into languages other than English. The default
908 language is still English. For details, please see the new
0410a6e0
PE
909 Internationalization section of the Bison manual. Software
910 distributors should also see the new PACKAGING file. Thanks to
911 Bruno Haible for this new feature.
1ce59070 912
d6fb461d 913** Wording in the Bison-generated parsers has been changed slightly to
1a059451
PE
914 simplify translation. In particular, the message "memory exhausted"
915 has replaced "parser stack overflow", as the old message was not
916 always accurate for modern Bison-generated parsers.
917
d6fb461d 918** Destructors are now called when the parser aborts, for all symbols left
258b75ca
PE
919 behind on the stack. Also, the start symbol is now destroyed after a
920 successful parse. In both cases, the behavior was formerly inconsistent.
921
d6fb461d 922** When generating verbose diagnostics, Bison-generated parsers no longer
72f000b0
PE
923 quote the literal strings associated with tokens. For example, for
924 a syntax error associated with '%token NUM "number"' they might
925 print 'syntax error, unexpected number' instead of 'syntax error,
926 unexpected "number"'.
193d7c70 927\f
d6fb461d 928* Changes in version 2.0, 2004-12-25:
efeed023 929
d6fb461d 930** Possibly-incompatible changes
d7e14fc0 931
82de6b0d
PE
932 - Bison-generated parsers no longer default to using the alloca function
933 (when available) to extend the parser stack, due to widespread
934 problems in unchecked stack-overflow detection. You can "#define
935 YYSTACK_USE_ALLOCA 1" to require the use of alloca, but please read
936 the manual to determine safe values for YYMAXDEPTH in that case.
8dd162d3 937
82de6b0d
PE
938 - Error token location.
939 During error recovery, the location of the syntax error is updated
940 to cover the whole sequence covered by the error token: it includes
941 the shifted symbols thrown away during the first part of the error
942 recovery, and the lookahead rejected during the second part.
18d192f0 943
82de6b0d
PE
944 - Semicolon changes:
945 . Stray semicolons are no longer allowed at the start of a grammar.
946 . Semicolons are now required after in-grammar declarations.
e342c3be 947
82de6b0d
PE
948 - Unescaped newlines are no longer allowed in character constants or
949 string literals. They were never portable, and GCC 3.4.0 has
950 dropped support for them. Better diagnostics are now generated if
951 forget a closing quote.
8dd162d3 952
82de6b0d 953 - NUL bytes are no longer allowed in Bison string literals, unfortunately.
f74b6f91 954
d6fb461d 955** New features
1452af69 956
82de6b0d 957 - GLR grammars now support locations.
4febdd96 958
82de6b0d
PE
959 - New directive: %initial-action.
960 This directive allows the user to run arbitrary code (including
961 initializing @$) from yyparse before parsing starts.
1452af69 962
82de6b0d
PE
963 - A new directive "%expect-rr N" specifies the expected number of
964 reduce/reduce conflicts in GLR parsers.
1452af69 965
82de6b0d
PE
966 - %token numbers can now be hexadecimal integers, e.g., `%token FOO 0x12d'.
967 This is a GNU extension.
4febdd96 968
82de6b0d 969 - The option `--report=lookahead' was changed to `--report=look-ahead'.
9e6e7ed2 970 [However, this was changed back after 2.3.]
1452af69 971
82de6b0d 972 - Experimental %destructor support has been added to lalr1.cc.
1452af69 973
82de6b0d
PE
974 - New configure option --disable-yacc, to disable installation of the
975 yacc command and -ly library introduced in 1.875 for POSIX conformance.
6040d338 976
d6fb461d 977** Bug fixes
d5a3fe37 978
82de6b0d
PE
979 - For now, %expect-count violations are now just warnings, not errors.
980 This is for compatibility with Bison 1.75 and earlier (when there are
981 reduce/reduce conflicts) and with Bison 1.30 and earlier (when there
982 are too many or too few shift/reduce conflicts). However, in future
983 versions of Bison we plan to improve the %expect machinery so that
984 these violations will become errors again.
3473d0f8 985
82de6b0d
PE
986 - Within Bison itself, numbers (e.g., goto numbers) are no longer
987 arbitrarily limited to 16-bit counts.
d600ee67 988
82de6b0d 989 - Semicolons are now allowed before "|" in grammar rules, as POSIX requires.
d600ee67 990\f
d6fb461d 991* Changes in version 1.875, 2003-01-01:
963fcc17 992
d6fb461d 993** The documentation license has been upgraded to version 1.2
dc546b0f 994 of the GNU Free Documentation License.
75eb3bc4 995
d6fb461d 996** syntax error processing
75eb3bc4 997
dc546b0f
PE
998 - In Yacc-style parsers YYLLOC_DEFAULT is now used to compute error
999 locations too. This fixes bugs in error-location computation.
75eb3bc4 1000
dc546b0f
PE
1001 - %destructor
1002 It is now possible to reclaim the memory associated to symbols
1003 discarded during error recovery. This feature is still experimental.
20daca06 1004
dc546b0f
PE
1005 - %error-verbose
1006 This new directive is preferred over YYERROR_VERBOSE.
74724a70 1007
dc546b0f
PE
1008 - #defining yyerror to steal internal variables is discouraged.
1009 It is not guaranteed to work forever.
d1de5372 1010
d6fb461d 1011** POSIX conformance
d1de5372 1012
dc546b0f
PE
1013 - Semicolons are once again optional at the end of grammar rules.
1014 This reverts to the behavior of Bison 1.33 and earlier, and improves
1015 compatibility with Yacc.
74724a70 1016
dc546b0f
PE
1017 - `parse error' -> `syntax error'
1018 Bison now uniformly uses the term `syntax error'; formerly, the code
1019 and manual sometimes used the term `parse error' instead. POSIX
1020 requires `syntax error' in diagnostics, and it was thought better to
1021 be consistent.
74724a70 1022
dc546b0f
PE
1023 - The documentation now emphasizes that yylex and yyerror must be
1024 declared before use. C99 requires this.
d1de5372 1025
dc546b0f
PE
1026 - Bison now parses C99 lexical constructs like UCNs and
1027 backslash-newline within C escape sequences, as POSIX 1003.1-2001 requires.
d1de5372 1028
dc546b0f
PE
1029 - File names are properly escaped in C output. E.g., foo\bar.y is
1030 output as "foo\\bar.y".
6780ca7a 1031
dc546b0f
PE
1032 - Yacc command and library now available
1033 The Bison distribution now installs a `yacc' command, as POSIX requires.
1034 Also, Bison now installs a small library liby.a containing
1035 implementations of Yacc-compatible yyerror and main functions.
1036 This library is normally not useful, but POSIX requires it.
6e649e65 1037
dc546b0f 1038 - Type clashes now generate warnings, not errors.
6e649e65 1039
dc546b0f
PE
1040 - If the user does not define YYSTYPE as a macro, Bison now declares it
1041 using typedef instead of defining it as a macro.
1042 For consistency, YYLTYPE is also declared instead of defined.
9501dc6e 1043
d6fb461d 1044** Other compatibility issues
886a425c 1045
dc546b0f
PE
1046 - %union directives can now have a tag before the `{', e.g., the
1047 directive `%union foo {...}' now generates the C code
1048 `typedef union foo { ... } YYSTYPE;'; this is for Yacc compatibility.
1049 The default union tag is `YYSTYPE', for compatibility with Solaris 9 Yacc.
1050 For consistency, YYLTYPE's struct tag is now `YYLTYPE' not `yyltype'.
1051 This is for compatibility with both Yacc and Bison 1.35.
72f889cc 1052
dc546b0f
PE
1053 - `;' is output before the terminating `}' of an action, for
1054 compatibility with Bison 1.35.
886a425c 1055
dc546b0f
PE
1056 - Bison now uses a Yacc-style format for conflict reports, e.g.,
1057 `conflicts: 2 shift/reduce, 1 reduce/reduce'.
437c2d80 1058
dc546b0f
PE
1059 - `yystype' and `yyltype' are now obsolescent macros instead of being
1060 typedefs or tags; they are no longer documented and are planned to be
1061 withdrawn in a future release.
2a8d363a 1062
d6fb461d 1063** GLR parser notes
2a8d363a 1064
dc546b0f
PE
1065 - GLR and inline
1066 Users of Bison have to decide how they handle the portability of the
1067 C keyword `inline'.
959e5f51 1068
dc546b0f
PE
1069 - `parsing stack overflow...' -> `parser stack overflow'
1070 GLR parsers now report `parser stack overflow' as per the Bison manual.
900c5db5 1071
d6fb461d 1072** Bison now warns if it detects conflicting outputs to the same file,
dc546b0f
PE
1073 e.g., it generates a warning for `bison -d -o foo.h foo.y' since
1074 that command outputs both code and header to foo.h.
6e40b4eb 1075
d6fb461d 1076** #line in output files
dc546b0f 1077 - --no-line works properly.
6e40b4eb 1078
d6fb461d 1079** Bison can no longer be built by a K&R C compiler; it requires C89 or
6e40b4eb
AD
1080 later to be built. This change originally took place a few versions
1081 ago, but nobody noticed until we recently asked someone to try
1082 building Bison with a K&R C compiler.
d600ee67 1083\f
d6fb461d 1084* Changes in version 1.75, 2002-10-14:
7933f2b5 1085
d6fb461d 1086** Bison should now work on 64-bit hosts.
7933f2b5 1087
d6fb461d 1088** Indonesian translation thanks to Tedi Heriyanto.
7933f2b5 1089
d6fb461d 1090** GLR parsers
f50adbbd
AD
1091 Fix spurious parse errors.
1092
d6fb461d 1093** Pure parsers
f50adbbd
AD
1094 Some people redefine yyerror to steal yyparse' private variables.
1095 Reenable this trick until an official feature replaces it.
1096
d6fb461d 1097** Type Clashes
d90c934c
AD
1098 In agreement with POSIX and with other Yaccs, leaving a default
1099 action is valid when $$ is untyped, and $1 typed:
1100
1101 untyped: ... typed;
1102
1103 but the converse remains an error:
1104
1105 typed: ... untyped;
1106
d6fb461d 1107** Values of mid-rule actions
d90c934c
AD
1108 The following code:
1109
1110 foo: { ... } { $$ = $1; } ...
1111
1112 was incorrectly rejected: $1 is defined in the second mid-rule
1113 action, and is equal to the $$ of the first mid-rule action.
d600ee67 1114\f
d6fb461d 1115* Changes in version 1.50, 2002-10-04:
adc8c848 1116
d6fb461d 1117** GLR parsing
676385e2
PH
1118 The declaration
1119 %glr-parser
1120 causes Bison to produce a Generalized LR (GLR) parser, capable of handling
1121 almost any context-free grammar, ambiguous or not. The new declarations
e8832397 1122 %dprec and %merge on grammar rules allow parse-time resolution of
676385e2
PH
1123 ambiguities. Contributed by Paul Hilfinger.
1124
7933f2b5 1125 Unfortunately Bison 1.50 does not work properly on 64-bit hosts
420f93c8
PE
1126 like the Alpha, so please stick to 32-bit hosts for now.
1127
d6fb461d 1128** Output Directory
8c165d89 1129 When not in Yacc compatibility mode, when the output file was not
e88dbdbf 1130 specified, running `bison foo/bar.y' created `foo/bar.c'. It
8c165d89
AD
1131 now creates `bar.c'.
1132
d6fb461d 1133** Undefined token
007a50a4 1134 The undefined token was systematically mapped to 2 which prevented
e88dbdbf 1135 the use of 2 by the user. This is no longer the case.
007a50a4 1136
d6fb461d 1137** Unknown token numbers
e88dbdbf 1138 If yylex returned an out of range value, yyparse could die. This is
007a50a4
AD
1139 no longer the case.
1140
d6fb461d 1141** Error token
e88dbdbf 1142 According to POSIX, the error token must be 256.
23c5a174
AD
1143 Bison extends this requirement by making it a preference: *if* the
1144 user specified that one of her tokens is numbered 256, then error
1145 will be mapped onto another number.
1146
d6fb461d 1147** Verbose error messages
e88dbdbf 1148 They no longer report `..., expecting error or...' for states where
217598da
AD
1149 error recovery is possible.
1150
d6fb461d 1151** End token
217598da
AD
1152 Defaults to `$end' instead of `$'.
1153
d6fb461d 1154** Error recovery now conforms to documentation and to POSIX
68cd8af3
PE
1155 When a Bison-generated parser encounters a syntax error, it now pops
1156 the stack until it finds a state that allows shifting the error
1157 token. Formerly, it popped the stack until it found a state that
1158 allowed some non-error action other than a default reduction on the
1159 error token. The new behavior has long been the documented behavior,
1160 and has long been required by POSIX. For more details, please see
337116ba
PE
1161 Paul Eggert, "Reductions during Bison error handling" (2002-05-20)
1162 <http://lists.gnu.org/archive/html/bug-bison/2002-05/msg00038.html>.
68cd8af3 1163
d6fb461d 1164** Traces
5504898e
AD
1165 Popped tokens and nonterminals are now reported.
1166
d6fb461d 1167** Larger grammars
a861a339
PE
1168 Larger grammars are now supported (larger token numbers, larger grammar
1169 size (= sum of the LHS and RHS lengths), larger LALR tables).
1170 Formerly, many of these numbers ran afoul of 16-bit limits;
1171 now these limits are 32 bits on most hosts.
355e7c1c 1172
d6fb461d 1173** Explicit initial rule
643a5994
AD
1174 Bison used to play hacks with the initial rule, which the user does
1175 not write. It is now explicit, and visible in the reports and
1176 graphs as rule 0.
23c5a174 1177
d6fb461d 1178** Useless rules
643a5994 1179 Before, Bison reported the useless rules, but, although not used,
77714df2 1180 included them in the parsers. They are now actually removed.
23c5a174 1181
d6fb461d 1182** Useless rules, useless nonterminals
6b98e4b5
AD
1183 They are now reported, as a warning, with their locations.
1184
d6fb461d 1185** Rules never reduced
e8832397
AD
1186 Rules that can never be reduced because of conflicts are now
1187 reported.
1188
d6fb461d 1189** Incorrect `Token not used'
11652ab3
AD
1190 On a grammar such as
1191
1192 %token useless useful
1193 %%
1194 exp: '0' %prec useful;
1195
1196 where a token was used to set the precedence of the last rule,
1197 bison reported both `useful' and `useless' as useless tokens.
1198
d6fb461d 1199** Revert the C++ namespace changes introduced in 1.31
77714df2 1200 as they caused too many portability hassles.
0179dd65 1201
d6fb461d 1202** Default locations
b2d52318
AD
1203 By an accident of design, the default computation of @$ was
1204 performed after another default computation was performed: @$ = @1.
1205 The latter is now removed: YYLLOC_DEFAULT is fully responsible of
1206 the computation of @$.
adc8c848 1207
d6fb461d 1208** Token end-of-file
b7c49edf
AD
1209 The token end of file may be specified by the user, in which case,
1210 the user symbol is used in the reports, the graphs, and the verbose
a861a339 1211 error messages instead of `$end', which remains being the default.
b7c49edf 1212 For instance
7bd6c77e 1213 %token MYEOF 0
b7c49edf 1214 or
7bd6c77e 1215 %token MYEOF 0 "end of file"
fdbcd8e2 1216
d6fb461d 1217** Semantic parser
fdbcd8e2
AD
1218 This old option, which has been broken for ages, is removed.
1219
d6fb461d 1220** New translations
a861a339 1221 Brazilian Portuguese, thanks to Alexandre Folle de Menezes.
84614e13
AD
1222 Croatian, thanks to Denis Lackovic.
1223
d6fb461d 1224** Incorrect token definitions
e88dbdbf 1225 When given `%token 'a' "A"', Bison used to output `#define 'a' 65'.
b87f8b21 1226
d6fb461d 1227** Token definitions as enums
77714df2
AD
1228 Tokens are output both as the traditional #define's, and, provided
1229 the compiler supports ANSI C or is a C++ compiler, as enums.
e88dbdbf 1230 This lets debuggers display names instead of integers.
77714df2 1231
d6fb461d 1232** Reports
ec3bc396
AD
1233 In addition to --verbose, bison supports --report=THINGS, which
1234 produces additional information:
b408954b
AD
1235 - itemset
1236 complete the core item sets with their closure
9e6e7ed2
PE
1237 - lookahead [changed to `look-ahead' in 1.875e through 2.3, but changed back]
1238 explicitly associate lookahead tokens to items
b408954b
AD
1239 - solved
1240 describe shift/reduce conflicts solving.
1241 Bison used to systematically output this information on top of
1242 the report. Solved conflicts are now attached to their states.
ec3bc396 1243
d6fb461d 1244** Type clashes
9af3fbce
AD
1245 Previous versions don't complain when there is a type clash on
1246 the default action if the rule has a mid-rule action, such as in:
1247
1248 %type <foo> bar
1249 %%
1250 bar: '0' {} '0';
1251
1252 This is fixed.
a861a339 1253
d6fb461d 1254** GNU M4 is now required when using Bison.
f987e9d2 1255\f
d6fb461d 1256* Changes in version 1.35, 2002-03-25:
76551463 1257
d6fb461d 1258** C Skeleton
76551463
AD
1259 Some projects use Bison's C parser with C++ compilers, and define
1260 YYSTYPE as a class. The recent adjustment of C parsers for data
1261 alignment and 64 bit architectures made this impossible.
1262
1263 Because for the time being no real solution for C++ parser
1264 generation exists, kludges were implemented in the parser to
1265 maintain this use. In the future, when Bison has C++ parsers, this
1266 kludge will be disabled.
1267
1268 This kludge also addresses some C++ problems when the stack was
1269 extended.
76551463 1270\f
d6fb461d 1271* Changes in version 1.34, 2002-03-12:
76551463 1272
d6fb461d 1273** File name clashes are detected
76551463
AD
1274 $ bison foo.y -d -o foo.x
1275 fatal error: header and parser would both be named `foo.x'
1276
d6fb461d 1277** A missing `;' at the end of a rule triggers a warning
76551463
AD
1278 In accordance with POSIX, and in agreement with other
1279 Yacc implementations, Bison will mandate this semicolon in the near
1280 future. This eases the implementation of a Bison parser of Bison
1281 grammars by making this grammar LALR(1) instead of LR(2). To
1282 facilitate the transition, this release introduces a warning.
1283
d6fb461d 1284** Revert the C++ namespace changes introduced in 1.31, as they caused too
76551463
AD
1285 many portability hassles.
1286
d6fb461d 1287** DJGPP support added.
76551463 1288
d6fb461d 1289** Fix test suite portability problems.
76551463 1290\f
d6fb461d 1291* Changes in version 1.33, 2002-02-07:
76551463 1292
d6fb461d 1293** Fix C++ issues
76551463
AD
1294 Groff could not be compiled for the definition of size_t was lacking
1295 under some conditions.
1296
d6fb461d 1297** Catch invalid @n
76551463
AD
1298 As is done with $n.
1299\f
d6fb461d 1300* Changes in version 1.32, 2002-01-23:
76551463 1301
d6fb461d 1302** Fix Yacc output file names
76551463 1303
d6fb461d 1304** Portability fixes
76551463 1305
d6fb461d 1306** Italian, Dutch translations
76551463 1307\f
d6fb461d 1308* Changes in version 1.31, 2002-01-14:
52d1aeee 1309
d6fb461d 1310** Many Bug Fixes
52d1aeee 1311
d6fb461d 1312** GNU Gettext and %expect
52d1aeee
MA
1313 GNU Gettext asserts 10 s/r conflicts, but there are 7. Now that
1314 Bison dies on incorrect %expectations, we fear there will be
1315 too many bug reports for Gettext, so _for the time being_, %expect
1316 does not trigger an error when the input file is named `plural.y'.
1317
d6fb461d 1318** Use of alloca in parsers
52d1aeee
MA
1319 If YYSTACK_USE_ALLOCA is defined to 0, then the parsers will use
1320 malloc exclusively. Since 1.29, but was not NEWS'ed.
1321
1322 alloca is used only when compiled with GCC, to avoid portability
1323 problems as on AIX.
1324
d6fb461d 1325** yyparse now returns 2 if memory is exhausted; formerly it dumped core.
b47dbebe 1326
d6fb461d 1327** When the generated parser lacks debugging code, YYDEBUG is now 0
52d1aeee
MA
1328 (as POSIX requires) instead of being undefined.
1329
d6fb461d 1330** User Actions
52d1aeee
MA
1331 Bison has always permitted actions such as { $$ = $1 }: it adds the
1332 ending semicolon. Now if in Yacc compatibility mode, the semicolon
1333 is no longer output: one has to write { $$ = $1; }.
1334
d6fb461d 1335** Better C++ compliance
52d1aeee 1336 The output parsers try to respect C++ namespaces.
76551463 1337 [This turned out to be a failed experiment, and it was reverted later.]
52d1aeee 1338
d6fb461d 1339** Reduced Grammars
52d1aeee
MA
1340 Fixed bugs when reporting useless nonterminals.
1341
d6fb461d 1342** 64 bit hosts
52d1aeee
MA
1343 The parsers work properly on 64 bit hosts.
1344
d6fb461d 1345** Error messages
52d1aeee
MA
1346 Some calls to strerror resulted in scrambled or missing error messages.
1347
d6fb461d 1348** %expect
52d1aeee
MA
1349 When the number of shift/reduce conflicts is correct, don't issue
1350 any warning.
1351
d6fb461d 1352** The verbose report includes the rule line numbers.
52d1aeee 1353
d6fb461d 1354** Rule line numbers are fixed in traces.
52d1aeee 1355
d6fb461d 1356** Swedish translation
52d1aeee 1357
d6fb461d 1358** Parse errors
52d1aeee
MA
1359 Verbose parse error messages from the parsers are better looking.
1360 Before: parse error: unexpected `'/'', expecting `"number"' or `'-'' or `'(''
1361 Now: parse error: unexpected '/', expecting "number" or '-' or '('
1362
d6fb461d 1363** Fixed parser memory leaks.
52d1aeee
MA
1364 When the generated parser was using malloc to extend its stacks, the
1365 previous allocations were not freed.
1366
d6fb461d 1367** Fixed verbose output file.
52d1aeee
MA
1368 Some newlines were missing.
1369 Some conflicts in state descriptions were missing.
1370
d6fb461d 1371** Fixed conflict report.
52d1aeee
MA
1372 Option -v was needed to get the result.
1373
d6fb461d 1374** %expect
52d1aeee
MA
1375 Was not used.
1376 Mismatches are errors, not warnings.
1377
d6fb461d 1378** Fixed incorrect processing of some invalid input.
52d1aeee 1379
d6fb461d 1380** Fixed CPP guards: 9foo.h uses BISON_9FOO_H instead of 9FOO_H.
52d1aeee 1381
d6fb461d 1382** Fixed some typos in the documentation.
52d1aeee 1383
d6fb461d 1384** %token MY_EOF 0 is supported.
52d1aeee
MA
1385 Before, MY_EOF was silently renumbered as 257.
1386
d6fb461d 1387** doc/refcard.tex is updated.
52d1aeee 1388
d6fb461d 1389** %output, %file-prefix, %name-prefix.
52d1aeee
MA
1390 New.
1391
d6fb461d 1392** --output
52d1aeee
MA
1393 New, aliasing `--output-file'.
1394\f
d6fb461d 1395* Changes in version 1.30, 2001-10-26:
342b8b6e 1396
d6fb461d 1397** `--defines' and `--graph' have now an optional argument which is the
fdac0091 1398 output file name. `-d' and `-g' do not change; they do not take any
342b8b6e
AD
1399 argument.
1400
d6fb461d 1401** `%source_extension' and `%header_extension' are removed, failed
342b8b6e
AD
1402 experiment.
1403
d6fb461d 1404** Portability fixes.
f987e9d2 1405\f
d6fb461d 1406* Changes in version 1.29, 2001-09-07:
342b8b6e 1407
d6fb461d 1408** The output file does not define const, as this caused problems when used
342b8b6e
AD
1409 with common autoconfiguration schemes. If you still use ancient compilers
1410 that lack const, compile with the equivalent of the C compiler option
1411 `-Dconst='. autoconf's AC_C_CONST macro provides one way to do this.
1412
d6fb461d 1413** Added `-g' and `--graph'.
f87a2205 1414
d6fb461d 1415** The Bison manual is now distributed under the terms of the GNU FDL.
f2b5126e 1416
d6fb461d 1417** The input and the output files has automatically a similar extension.
234a3be3 1418
d6fb461d 1419** Russian translation added.
f87a2205 1420
d6fb461d 1421** NLS support updated; should hopefully be less troublesome.
f87a2205 1422
d6fb461d 1423** Added the old Bison reference card.
c33638bb 1424
d6fb461d 1425** Added `--locations' and `%locations'.
6deb4447 1426
d6fb461d 1427** Added `-S' and `--skeleton'.
cd5bd6ac 1428
d6fb461d 1429** `%raw', `-r', `--raw' is disabled.
62ab6972 1430
d6fb461d 1431** Special characters are escaped when output. This solves the problems
cd5bd6ac
AD
1432 of the #line lines with path names including backslashes.
1433
d6fb461d 1434** New directives.
4ecbf796
MA
1435 `%yacc', `%fixed_output_files', `%defines', `%no_parser', `%verbose',
1436 `%debug', `%source_extension' and `%header_extension'.
f987e9d2 1437
d6fb461d 1438** @$
f987e9d2 1439 Automatic location tracking.
f87a2205 1440\f
d6fb461d 1441* Changes in version 1.28, 1999-07-06:
d2e00347 1442
d6fb461d 1443** Should compile better now with K&R compilers.
d2e00347 1444
d6fb461d 1445** Added NLS.
d2e00347 1446
d6fb461d 1447** Fixed a problem with escaping the double quote character.
d2e00347 1448
d6fb461d 1449** There is now a FAQ.
d2e00347 1450\f
d6fb461d 1451* Changes in version 1.27:
5c31c3c2 1452
d6fb461d 1453** The make rule which prevented bison.simple from being created on
5c31c3c2
JT
1454 some systems has been fixed.
1455\f
d6fb461d 1456* Changes in version 1.26:
4be07551 1457
d6fb461d 1458** Bison now uses automake.
4be07551 1459
d6fb461d 1460** New mailing lists: <bug-bison@gnu.org> and <help-bison@gnu.org>.
4be07551 1461
d6fb461d 1462** Token numbers now start at 257 as previously documented, not 258.
4be07551 1463
d6fb461d 1464** Bison honors the TMPDIR environment variable.
4be07551 1465
d6fb461d 1466** A couple of buffer overruns have been fixed.
f51dbca1 1467
d6fb461d 1468** Problems when closing files should now be reported.
f51dbca1 1469
d6fb461d 1470** Generated parsers should now work even on operating systems which do
f51dbca1 1471 not provide alloca().
4be07551 1472\f
d6fb461d 1473* Changes in version 1.25, 1995-10-16:
df8878c5 1474
d6fb461d 1475** Errors in the input grammar are not fatal; Bison keeps reading
df8878c5 1476the grammar file, and reports all the errors found in it.
8c44d3ec 1477
d6fb461d 1478** Tokens can now be specified as multiple-character strings: for
df8878c5
RS
1479example, you could use "<=" for a token which looks like <=, instead
1480of chosing a name like LESSEQ.
1481
d6fb461d 1482** The %token_table declaration says to write a table of tokens (names
df8878c5
RS
1483and numbers) into the parser file. The yylex function can use this
1484table to recognize multiple-character string tokens, or for other
1485purposes.
1486
d6fb461d 1487** The %no_lines declaration says not to generate any #line preprocessor
df8878c5
RS
1488directives in the parser file.
1489
d6fb461d 1490** The %raw declaration says to use internal Bison token numbers, not
df8878c5
RS
1491Yacc-compatible token numbers, when token names are defined as macros.
1492
d6fb461d 1493** The --no-parser option produces the parser tables without including
df8878c5
RS
1494the parser engine; a project can now use its own parser engine.
1495The actions go into a separate file called NAME.act, in the form of
1496a switch statement body.
1497\f
d6fb461d 1498* Changes in version 1.23:
6780ca7a 1499
4d019228
DM
1500The user can define YYPARSE_PARAM as the name of an argument to be
1501passed into yyparse. The argument should have type void *. It should
1502actually point to an object. Grammar actions can access the variable
1503by casting it to the proper pointer type.
6780ca7a 1504
6780ca7a 1505Line numbers in output file corrected.
6780ca7a 1506\f
d6fb461d 1507* Changes in version 1.22:
6780ca7a
DM
1508
1509--help option added.
6780ca7a 1510\f
d6fb461d 1511* Changes in version 1.20:
6780ca7a
DM
1512
1513Output file does not redefine const for C++.
9f4503d6
AD
1514
1515Local Variables:
1516mode: outline
1517End:
76551463
AD
1518
1519-----
1520
ea0a7676 1521Copyright (C) 1995-2011 Free Software Foundation, Inc.
76551463 1522
8defe11b 1523This file is part of Bison, the GNU Parser Generator.
76551463 1524
f16b0819 1525This program is free software: you can redistribute it and/or modify
76551463 1526it under the terms of the GNU General Public License as published by
f16b0819
PE
1527the Free Software Foundation, either version 3 of the License, or
1528(at your option) any later version.
76551463 1529
f16b0819 1530This program is distributed in the hope that it will be useful,
76551463
AD
1531but WITHOUT ANY WARRANTY; without even the implied warranty of
1532MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
1533GNU General Public License for more details.
1534
1535You should have received a copy of the GNU General Public License
f16b0819 1536along with this program. If not, see <http://www.gnu.org/licenses/>.