]> git.saurik.com Git - bison.git/blame - NEWS
yacc.c: scope reduction
[bison.git] / NEWS
CommitLineData
ed7658fe 1GNU Bison NEWS
3af4feb2 2
09a1cd5e
AD
3* Noteworthy changes in release ?.? (????-??-??) [?]
4
edf9a06f
AD
5** Bug fixes
6
7 Warnings about uninitialized yylloc in yyparse have been fixed.
8
1127a75a
AD
9 Restored C90 compliance (yet no report was ever made).
10
d4fe9e88
AD
11** Diagnostics are improved
12
13*** Changes in the format of error messages
14
15 This used to be the format of many error reports:
16
17 input.y:2.7-12: %type redeclaration for exp
18 input.y:1.7-12: previous declaration
19
20 It is now:
21
22 input.y:2.7-12: error: %type redeclaration for exp
23 input.y:1.7-12: previous declaration
effd30c0 24
d4fe9e88 25*** New format for error reports: carets
effd30c0 26
d4fe9e88 27 Caret errors have been added to Bison:
7bada535 28
d4fe9e88
AD
29 input.y:2.7-12: error: %type redeclaration for exp
30 %type <sval> exp
31 ^^^^^^
32 input.y:1.7-12: previous declaration
33 %type <ival> exp
34 ^^^^^^
35
36 or
7bada535
TR
37
38 input.y:3.20-23: error: ambiguous reference: '$exp'
fb6040f0 39 exp: exp '+' exp { $exp = $1 + $3; };
7bada535 40 ^^^^
fb6040f0
TR
41 input.y:3.1-3: refers to: $exp at $$
42 exp: exp '+' exp { $exp = $1 + $3; };
43 ^^^
44 input.y:3.6-8: refers to: $exp at $1
45 exp: exp '+' exp { $exp = $1 + $3; };
46 ^^^
47 input.y:3.14-16: refers to: $exp at $3
48 exp: exp '+' exp { $exp = $1 + $3; };
49 ^^^
7bada535 50
d4fe9e88
AD
51 The default behaviour for now is still not to display these unless
52 explictly asked with -fcaret (or -fall). However, in a later release, it
53 will be made the default behavior (but may still be deactivated with
54 -fno-caret).
7bada535 55
1f1bd572
TR
56** New value for %define variable: api.pure full
57
58 The %define variable api.pure requests a pure (reentrant) parser. However,
d4fe9e88
AD
59 for historical reasons, using it in a location-tracking Yacc parser
60 resulted in a yyerror function that did not take a location as a
61 parameter. With this new value, the user may request a better pure parser,
62 where yyerror does take a location as a parameter (in location-tracking
63 parsers).
1f1bd572
TR
64
65 The use of "%define api.pure true" is deprecated in favor of this new
66 "%define api.pure full".
67
7287be84 68** New %define variable: api.location.type (glr.cc, lalr1.cc, lalr1.java)
db8ab2be
AD
69
70 The %define variable api.location.type defines the name of the type to use
71 for locations. When defined, Bison no longer generates the position.hh
72 and location.hh files, nor does the parser will include them: the user is
73 then responsible to define her type.
74
75 This can be used in programs with several parsers to factor their location
7287be84
AD
76 and position files: let one of them generate them, and the others just use
77 them.
db8ab2be
AD
78
79 This feature was actually introduced, but not documented, in Bison 2.5,
80 under the name "location_type" (which is maintained for backward
81 compatibility).
82
7287be84
AD
83 For consistency, lalr1.java's %define variables location_type and
84 position_type are deprecated in favor of api.location.type and
85 api.position.type.
86
d4fe9e88
AD
87** Exception safety (lalr1.cc)
88
89 The parse function now catches exceptions, uses the %destructors to
90 release memory (the lookahead symbol and the symbols pushed on the stack)
91 before re-throwing the exception.
92
93 This feature is somewhat experimental. User feedback would be
94 appreciated.
95
9c16d399 96** Graph improvements in DOT and XSLT
fc4fdd62
TR
97
98 The graphical presentation of the states is more readable: their shape is
99 now rectangular, the state number is clearly displayed, and the items are
100 numbered and left-justified.
101
102 The reductions are now explicitly represented as transitions to other
103 diamond shaped nodes.
104
9c16d399
TR
105 These changes are present in both --graph output and xml2dot.xsl XSLT
106 processing, with minor (documented) differences.
107
d4fe9e88
AD
108** %language is no longer an experimental feature.
109
110 The introduction of this feature, in 2.4, was four years ago. The
111 --language option and the %language directive are no longer experimental.
9c16d399 112
53e2cd1e
AD
113** Documentation
114
115 The sections about shift/reduce and reduce/reduce conflicts resolution
116 have been fixed and extended.
9d2423f5 117
d4fe9e88
AD
118 Although introduced more than four years ago, XML and Graphviz reports
119 were not properly documented.
120
be22823e
AD
121 The translation of mid-rule actions is now described.
122
9d3f7eaf 123* Noteworthy changes in release 2.6.5 (2012-11-07) [stable]
6f1360bd 124
a68b1f23
AD
125 We consider compiler warnings about Bison generated parsers to be bugs.
126 Rather than working around them in your own project, please consider
127 reporting them to us.
128
129** Bug fixes
130
131 Warnings about uninitialized yylval and/or yylloc for push parsers with a
132 pure interface have been fixed for GCC 4.0 up to 4.8, and Clang 2.9 to
133 3.2.
134
135 Other issues in the test suite have been addressed.
6f1360bd 136
95066e92
AD
137 Nul characters are correctly displayed in error messages.
138
a1d1ab50
AD
139 When possible, yylloc is correctly initialized before calling yylex. It
140 is no longer necessary to initialize it in the %initial-action.
141
0ac15849 142* Noteworthy changes in release 2.6.4 (2012-10-23) [stable]
a4eb820f 143
468455e1 144 Bison 2.6.3's --version was incorrect. This release fixes this issue.
a4eb820f 145
6eb8f74f 146* Noteworthy changes in release 2.6.3 (2012-10-22) [stable]
933ec544 147
6b4cb804
AD
148** Bug fixes
149
9a50c55a 150 Bugs and portability issues in the test suite have been fixed.
6b4cb804
AD
151
152 Some errors in translations have been addressed, and --help now directs
153 users to the appropriate place to report them.
154
155 Stray Info files shipped by accident are removed.
156
157 Incorrect definitions of YY_, issued by yacc.c when no parser header is
158 generated, are removed.
933ec544 159
90b1335a
TR
160 All the generated headers are self-contained.
161
6192d2c6
AD
162** Header guards (yacc.c, glr.c, glr.cc)
163
164 In order to avoid collisions, the header guards are now
165 YY_<PREFIX>_<FILE>_INCLUDED, instead of merely <PREFIX>_<FILE>.
166 For instance the header generated from
167
168 %define api.prefix "calc"
169 %defines "lib/parse.h"
170
171 will use YY_CALC_LIB_PARSE_H_INCLUDED as guard.
172
7b70847e 173** Fix compiler warnings in the generated parser (yacc.c, glr.c)
8f8439ce
AD
174
175 The compilation of pure parsers (%define api.pure) can trigger GCC
176 warnings such as:
177
178 input.c: In function 'yyparse':
179 input.c:1503:12: warning: 'yylval' may be used uninitialized in this
180 function [-Wmaybe-uninitialized]
181 *++yyvsp = yylval;
182 ^
183
184 This is now fixed; pragmas to avoid these warnings are no longer needed.
185
7b70847e
AD
186 Warnings from clang ("equality comparison with extraneous parentheses" and
187 "function declared 'noreturn' should not return") have also been
188 addressed.
189
e1eeecd3 190* Noteworthy changes in release 2.6.2 (2012-08-03) [stable]
9c26b8fc 191
43ca8040
AD
192** Bug fixes
193
194 Buffer overruns, complaints from Flex, and portability issues in the test
195 suite have been fixed.
196
c9d546b2
AD
197** Spaces in %lex- and %parse-param (lalr1.cc, glr.cc)
198
199 Trailing end-of-lines in %parse-param or %lex-param would result in
200 invalid C++. This is fixed.
9c26b8fc 201
dcd5344d
AD
202** Spurious spaces and end-of-lines
203
204 The generated files no longer end (nor start) with empty lines.
205
77b214ef 206* Noteworthy changes in release 2.6.1 (2012-07-30) [stable]
a4107f24 207
cbaea010 208 Bison no longer executes user-specified M4 code when processing a grammar.
8617d87e 209
e20e6a50
AD
210** Future Changes
211
212 In addition to the removal of the features announced in Bison 2.6, the
213 next major release will remove the "Temporary hack for adding a semicolon
214 to the user action", as announced in the release 2.5. Instead of:
215
216 exp: exp "+" exp { $$ = $1 + $3 };
217
218 write:
219
220 exp: exp "+" exp { $$ = $1 + $3; };
221
8617d87e
AD
222** Bug fixes
223
0e164d43
AD
224*** Type names are now properly escaped.
225
226*** glr.cc: set_debug_level and debug_level work as expected.
a4107f24 227
26313726
AD
228*** Stray @ or $ in actions
229
230 While Bison used to warn about stray $ or @ in action rules, it did not
231 for other actions such as printers, destructors, or initial actions. It
232 now does.
233
cd735a8c 234** Type names in actions
4982f078
AD
235
236 For consistency with rule actions, it is now possible to qualify $$ by a
cd735a8c 237 type-name in destructors, printers, and initial actions. For instance:
4982f078
AD
238
239 %printer { fprintf (yyo, "(%d, %f)", $<ival>$, $<fval>$); } <*> <>;
240
241 will display two values for each typed and untyped symbol (provided
cd735a8c 242 that YYSTYPE has both "ival" and "fval" fields).
4982f078 243
1505e8bb 244* Noteworthy changes in release 2.6 (2012-07-19) [stable]
ad6b775f 245
c2425191 246** Future Changes
0f11eec2 247
55d1006f
AD
248 The next major release of Bison will drop support for the following
249 deprecated features. Please report disagreements to bug-bison@gnu.org.
0f11eec2 250
aaf61036 251*** K&R C parsers
55d1006f
AD
252
253 Support for generating parsers in K&R C will be removed. Parsers
242cc08e 254 generated for C support ISO C90, and are tested with ISO C99 and ISO C11
55d1006f
AD
255 compilers.
256
258cddbc 257*** Features deprecated since Bison 1.875
55d1006f 258
258cddbc
AD
259 The definitions of yystype and yyltype will be removed; use YYSTYPE and
260 YYLTYPE.
55d1006f 261
258cddbc
AD
262 YYPARSE_PARAM and YYLEX_PARAM, deprecated in favor of %parse-param and
263 %lex-param, will no longer be supported.
264
265 Support for the preprocessor symbol YYERROR_VERBOSE will be removed, use
266 %error-verbose.
55d1006f
AD
267
268*** The generated header will be included (yacc.c)
0f11eec2
AD
269
270 Instead of duplicating the content of the generated header (definition of
55d1006f
AD
271 YYSTYPE, yyparse declaration etc.), the generated parser will include it,
272 as is already the case for GLR or C++ parsers. This change is deferred
273 because existing versions of ylwrap (e.g., Automake 1.12.1) do not support
274 it.
0f11eec2 275
c2425191 276** Generated Parser Headers
56ca3d8f 277
258cddbc 278*** Guards (yacc.c, glr.c, glr.cc)
c3e9f08f
AD
279
280 The generated headers are now guarded, as is already the case for C++
242cc08e 281 parsers (lalr1.cc). For instance, with --defines=foo.h:
c3e9f08f 282
e29f0771
AD
283 #ifndef YY_FOO_H
284 # define YY_FOO_H
285 ...
286 #endif /* !YY_FOO_H */
c3e9f08f 287
258cddbc 288*** New declarations (yacc.c, glr.c)
56ca3d8f
AD
289
290 The generated header now declares yydebug and yyparse. Both honor
291 --name-prefix=bar_, and yield
292
e29f0771 293 int bar_parse (void);
56ca3d8f
AD
294
295 rather than
296
e29f0771
AD
297 #define yyparse bar_parse
298 int yyparse (void);
56ca3d8f
AD
299
300 in order to facilitate the inclusion of several parser headers inside a
301 single compilation unit.
c3e9f08f 302
258cddbc
AD
303*** Exported symbols in C++
304
305 The symbols YYTOKEN_TABLE and YYERROR_VERBOSE, which were defined in the
306 header, are removed, as they prevent the possibility of including several
307 generated headers from a single compilation unit.
308
694af10c
AD
309*** YYLSP_NEEDED
310
311 For the same reasons, the undocumented and unused macro YYLSP_NEEDED is no
312 longer defined.
313
4b3847c3
AD
314** New %define variable: api.prefix
315
316 Now that the generated headers are more complete and properly protected
317 against multiple inclusions, constant names, such as YYSTYPE are a
318 problem. While yyparse and others are properly renamed by %name-prefix,
319 YYSTYPE, YYDEBUG and others have never been affected by it. Because it
320 would introduce backward compatibility issues in projects not expecting
321 YYSTYPE to be renamed, instead of changing the behavior of %name-prefix,
322 it is deprecated in favor of a new %define variable: api.prefix.
323
324 The following examples compares both:
325
326 %name-prefix "bar_" | %define api.prefix "bar_"
327 %token <ival> FOO %token <ival> FOO
328 %union { int ival; } %union { int ival; }
329 %% %%
330 exp: 'a'; exp: 'a';
331
332 bison generates:
333
334 #ifndef BAR_FOO_H #ifndef BAR_FOO_H
335 # define BAR_FOO_H # define BAR_FOO_H
336
337 /* Enabling traces. */ /* Enabling traces. */
5f108727
AD
338 # ifndef YYDEBUG | # ifndef BAR_DEBUG
339 > # if defined YYDEBUG
340 > # if YYDEBUG
341 > # define BAR_DEBUG 1
342 > # else
343 > # define BAR_DEBUG 0
344 > # endif
345 > # else
346 # define YYDEBUG 0 | # define BAR_DEBUG 0
347 > # endif
348 # endif | # endif
349
350 # if YYDEBUG | # if BAR_DEBUG
4b3847c3
AD
351 extern int bar_debug; extern int bar_debug;
352 # endif # endif
353
354 /* Tokens. */ /* Tokens. */
355 # ifndef YYTOKENTYPE | # ifndef BAR_TOKENTYPE
356 # define YYTOKENTYPE | # define BAR_TOKENTYPE
357 enum yytokentype { | enum bar_tokentype {
358 FOO = 258 FOO = 258
359 }; };
360 # endif # endif
361
362 #if ! defined YYSTYPE \ | #if ! defined BAR_STYPE \
363 && ! defined YYSTYPE_IS_DECLARED | && ! defined BAR_STYPE_IS_DECLARED
364 typedef union YYSTYPE | typedef union BAR_STYPE
365 { {
366 int ival; int ival;
367 } YYSTYPE; | } BAR_STYPE;
368 # define YYSTYPE_IS_DECLARED 1 | # define BAR_STYPE_IS_DECLARED 1
369 #endif #endif
370
371 extern YYSTYPE bar_lval; | extern BAR_STYPE bar_lval;
372
373 int bar_parse (void); int bar_parse (void);
374
375 #endif /* !BAR_FOO_H */ #endif /* !BAR_FOO_H */
376
dfaac272 377* Noteworthy changes in release 2.5.1 (2012-06-05) [stable]
90dcd415 378
debe2c03 379** Future changes:
041308d0 380
e4ab1254 381 The next major release will drop support for generating parsers in K&R C.
ef51bfa7 382
466b4cf2 383** yacc.c: YYBACKUP works as expected.
94556574 384
d834eca0
AD
385** glr.c improvements:
386
387*** Location support is eliminated when not requested:
388
e4ab1254
AD
389 GLR parsers used to include location-related code even when locations were
390 not requested, and therefore not even usable.
d834eca0
AD
391
392*** __attribute__ is preserved:
debe2c03 393
e4ab1254
AD
394 __attribute__ is no longer disabled when __STRICT_ANSI__ is defined (i.e.,
395 when -std is passed to GCC).
debe2c03 396
466b4cf2 397** lalr1.java: several fixes:
041308d0 398
e4ab1254
AD
399 The Java parser no longer throws ArrayIndexOutOfBoundsException if the
400 first token leads to a syntax error. Some minor clean ups.
041308d0 401
22172d47
AD
402** Changes for C++:
403
404*** C++11 compatibility:
ef51bfa7 405
e4ab1254
AD
406 C and C++ parsers use "nullptr" instead of "0" when __cplusplus is 201103L
407 or higher.
ef51bfa7 408
22172d47
AD
409*** Header guards
410
411 The header files such as "parser.hh", "location.hh", etc. used a constant
412 name for preprocessor guards, for instance:
413
e29f0771
AD
414 #ifndef BISON_LOCATION_HH
415 # define BISON_LOCATION_HH
416 ...
417 #endif // !BISON_LOCATION_HH
22172d47
AD
418
419 The inclusion guard is now computed from "PREFIX/FILE-NAME", where lower
420 case characters are converted to upper case, and series of
421 non-alphanumerical characters are converted to an underscore.
422
423 With "bison -o lang++/parser.cc", "location.hh" would now include:
424
e29f0771
AD
425 #ifndef YY_LANG_LOCATION_HH
426 # define YY_LANG_LOCATION_HH
427 ...
428 #endif // !YY_LANG_LOCATION_HH
22172d47
AD
429
430*** C++ locations:
936c88d1 431
e4ab1254
AD
432 The position and location constructors (and their initialize methods)
433 accept new arguments for line and column. Several issues in the
434 documentation were fixed.
936c88d1 435
466b4cf2
AD
436** liby is no longer asking for "rpl_fprintf" on some platforms.
437
7e508a2b
AD
438** Changes in the manual:
439
440*** %printer is documented
441
e4ab1254
AD
442 The "%printer" directive, supported since at least Bison 1.50, is finally
443 documented. The "mfcalc" example is extended to demonstrate it.
7e508a2b 444
e4ab1254
AD
445 For consistency with the C skeletons, the C++ parsers now also support
446 "yyoutput" (as an alias to "debug_stream ()").
7e508a2b
AD
447
448*** Several improvements have been made:
466b4cf2 449
e4ab1254
AD
450 The layout for grammar excerpts was changed to a more compact scheme.
451 Named references are motivated. The description of the automaton
452 description file (*.output) is updated to the current format. Incorrect
453 index entries were fixed. Some other errors were fixed.
466b4cf2 454
86b08b49
AD
455** Building bison:
456
457*** Conflicting prototypes with recent/modified Flex.
458
e4ab1254
AD
459 Fixed build problems with the current, unreleased, version of Flex, and
460 some modified versions of 2.5.35, which have modified function prototypes.
466b4cf2 461
8ef26c2a
AD
462*** Warnings during the build procedure have been eliminated.
463
464*** Several portability problems in the test suite have been fixed:
466b4cf2 465
e4ab1254
AD
466 This includes warnings with some compilers, unexpected behavior of tools
467 such as diff, warning messages from the test suite itself, etc.
466b4cf2 468
91aadcc7 469*** The install-pdf target works properly:
8ef26c2a 470
e4ab1254
AD
471 Running "make install-pdf" (or -dvi, -html, -info, and -ps) no longer
472 halts in the middle of its course.
8ef26c2a 473
b2bc8a1b 474* Changes in version 2.5 (2011-05-14):
006faedf 475
eb8c66bb
JD
476** Grammar symbol names can now contain non-initial dashes:
477
478 Consistently with directives (such as %error-verbose) and with
479 %define variables (e.g. push-pull), grammar symbol names may contain
480 dashes in any position except the beginning. This is a GNU
481 extension over POSIX Yacc. Thus, use of this extension is reported
482 by -Wyacc and rejected in Yacc mode (--yacc).
483
3b1977ea 484** Named references:
348f5608
AR
485
486 Historically, Yacc and Bison have supported positional references
487 ($n, $$) to allow access to symbol values from inside of semantic
488 actions code.
489
490 Starting from this version, Bison can also accept named references.
491 When no ambiguity is possible, original symbol names may be used
492 as named references:
493
ac3297d5 494 if_stmt : "if" cond_expr "then" then_stmt ';'
348f5608
AR
495 { $if_stmt = mk_if_stmt($cond_expr, $then_stmt); }
496
497 In the more common case, explicit names may be declared:
498
ac3297d5 499 stmt[res] : "if" expr[cond] "then" stmt[then] "else" stmt[else] ';'
348f5608
AR
500 { $res = mk_if_stmt($cond, $then, $else); }
501
f840c05a 502 Location information is also accessible using @name syntax. When
348f5608
AR
503 accessing symbol names containing dots or dashes, explicit bracketing
504 ($[sym.1]) must be used.
505
f840c05a 506 These features are experimental in this version. More user feedback
348f5608
AR
507 will help to stabilize them.
508
3b1977ea 509** IELR(1) and canonical LR(1):
34a6c2d1
JD
510
511 IELR(1) is a minimal LR(1) parser table generation algorithm. That
512 is, given any context-free grammar, IELR(1) generates parser tables
723206df 513 with the full language-recognition power of canonical LR(1) but with
3b1977ea
JD
514 nearly the same number of parser states as LALR(1). This reduction
515 in parser states is often an order of magnitude. More importantly,
34a6c2d1
JD
516 because canonical LR(1)'s extra parser states may contain duplicate
517 conflicts in the case of non-LR(1) grammars, the number of conflicts
518 for IELR(1) is often an order of magnitude less as well. This can
519 significantly reduce the complexity of developing of a grammar.
520
521 Bison can now generate IELR(1) and canonical LR(1) parser tables in
522 place of its traditional LALR(1) parser tables, which remain the
523 default. You can specify the type of parser tables in the grammar
524 file with these directives:
525
f37495f6
JD
526 %define lr.type lalr
527 %define lr.type ielr
528 %define lr.type canonical-lr
34a6c2d1 529
6f04ee6c 530 The default-reduction optimization in the parser tables can also be
e4ab1254
AD
531 adjusted using "%define lr.default-reductions". For details on both
532 of these features, see the new section "Tuning LR" in the Bison
6f04ee6c 533 manual.
34a6c2d1
JD
534
535 These features are experimental. More user feedback will help to
536 stabilize them.
537
6f04ee6c 538** LAC (Lookahead Correction) for syntax error handling:
4c38b19e
JD
539
540 Canonical LR, IELR, and LALR can suffer from a couple of problems
541 upon encountering a syntax error. First, the parser might perform
542 additional parser stack reductions before discovering the syntax
6f04ee6c 543 error. Such reductions can perform user semantic actions that are
4c38b19e
JD
544 unexpected because they are based on an invalid token, and they
545 cause error recovery to begin in a different syntactic context than
546 the one in which the invalid token was encountered. Second, when
6f04ee6c 547 verbose error messages are enabled (with %error-verbose or the
e4ab1254 548 obsolete "#define YYERROR_VERBOSE"), the expected token list in the
6f04ee6c
JD
549 syntax error message can both contain invalid tokens and omit valid
550 tokens.
4c38b19e
JD
551
552 The culprits for the above problems are %nonassoc, default
553 reductions in inconsistent states, and parser state merging. Thus,
554 IELR and LALR suffer the most. Canonical LR can suffer only if
555 %nonassoc is used or if default reductions are enabled for
556 inconsistent states.
557
6f04ee6c
JD
558 LAC is a new mechanism within the parsing algorithm that solves
559 these problems for canonical LR, IELR, and LALR without sacrificing
560 %nonassoc, default reductions, or state merging. When LAC is in
561 use, canonical LR and IELR behave almost exactly the same for both
562 syntactically acceptable and syntactically unacceptable input.
4c38b19e
JD
563 While LALR still does not support the full language-recognition
564 power of canonical LR and IELR, LAC at least enables LALR's syntax
565 error handling to correctly reflect LALR's language-recognition
566 power.
567
568 Currently, LAC is only supported for deterministic parsers in C.
569 You can enable LAC with the following directive:
570
571 %define parse.lac full
572
e4ab1254 573 See the new section "LAC" in the Bison manual for additional
6f04ee6c 574 details including a few caveats.
4c38b19e
JD
575
576 LAC is an experimental feature. More user feedback will help to
577 stabilize it.
578
6ee881f6 579** %define improvements:
f37495f6 580
3b1977ea 581*** Can now be invoked via the command line:
4ecd3681 582
34d41938 583 Each of these command-line options
4ecd3681 584
34d41938
JD
585 -D NAME[=VALUE]
586 --define=NAME[=VALUE]
587
588 -F NAME[=VALUE]
589 --force-define=NAME[=VALUE]
4ecd3681
JD
590
591 is equivalent to this grammar file declaration
592
34d41938 593 %define NAME ["VALUE"]
4ecd3681 594
34d41938
JD
595 except that the manner in which Bison processes multiple definitions
596 for the same NAME differs. Most importantly, -F and --force-define
597 quietly override %define, but -D and --define do not. For further
e4ab1254 598 details, see the section "Bison Options" in the Bison manual.
4ecd3681 599
3b1977ea 600*** Variables renamed:
812775a0
JD
601
602 The following %define variables
603
604 api.push_pull
605 lr.keep_unreachable_states
606
607 have been renamed to
608
609 api.push-pull
610 lr.keep-unreachable-states
611
612 The old names are now deprecated but will be maintained indefinitely
613 for backward compatibility.
614
723206df 615*** Values no longer need to be quoted in the grammar file:
f37495f6
JD
616
617 If a %define value is an identifier, it no longer needs to be placed
618 within quotations marks. For example,
619
620 %define api.push-pull "push"
621
622 can be rewritten as
623
624 %define api.push-pull push
625
6ee881f6 626*** Unrecognized variables are now errors not warnings.
c046698e 627
6ee881f6
JD
628*** Multiple invocations for any variable is now an error not a warning.
629
630** Unrecognized %code qualifiers are now errors not warnings.
631
632** Character literals not of length one:
633
634 Previously, Bison quietly converted all character literals to length
635 one. For example, without warning, Bison interpreted the operators in
636 the following grammar to be the same token:
637
638 exp: exp '++'
639 | exp '+' exp
640 ;
641
642 Bison now warns when a character literal is not of length one. In
643 some future release, Bison will start reporting an error instead.
644
645** Destructor calls fixed for lookaheads altered in semantic actions:
646
647 Previously for deterministic parsers in C, if a user semantic action
648 altered yychar, the parser in some cases used the old yychar value to
649 determine which destructor to call for the lookahead upon a syntax
650 error or upon parser return. This bug has been fixed.
651
652** C++ parsers use YYRHSLOC:
653
654 Similarly to the C parsers, the C++ parsers now define the YYRHSLOC
655 macro and use it in the default YYLLOC_DEFAULT. You are encouraged
e4ab1254
AD
656 to use it. If, for instance, your location structure has "first"
657 and "last" members, instead of
6ee881f6 658
e29f0771
AD
659 # define YYLLOC_DEFAULT(Current, Rhs, N) \
660 do \
661 if (N) \
662 { \
663 (Current).first = (Rhs)[1].location.first; \
664 (Current).last = (Rhs)[N].location.last; \
665 } \
666 else \
667 { \
668 (Current).first = (Current).last = (Rhs)[0].location.last; \
669 } \
670 while (false)
6ee881f6
JD
671
672 use:
673
e29f0771
AD
674 # define YYLLOC_DEFAULT(Current, Rhs, N) \
675 do \
676 if (N) \
677 { \
678 (Current).first = YYRHSLOC (Rhs, 1).first; \
679 (Current).last = YYRHSLOC (Rhs, N).last; \
680 } \
681 else \
682 { \
683 (Current).first = (Current).last = YYRHSLOC (Rhs, 0).last; \
684 } \
685 while (false)
6ee881f6
JD
686
687** YYLLOC_DEFAULT in C++:
688
689 The default implementation of YYLLOC_DEFAULT used to be issued in
690 the header file. It is now output in the implementation file, after
691 the user %code sections so that its #ifndef guard does not try to
692 override the user's YYLLOC_DEFAULT if provided.
c046698e 693
3b1977ea 694** YYFAIL now produces warnings and Java parsers no longer implement it:
62efdd2a
JD
695
696 YYFAIL has existed for many years as an undocumented feature of
697 deterministic parsers in C generated by Bison. More recently, it was
698 a documented feature of Bison's experimental Java parsers. As
699 promised in Bison 2.4.2's NEWS entry, any appearance of YYFAIL in a
700 semantic action now produces a deprecation warning, and Java parsers
701 no longer implement YYFAIL at all. For further details, including a
702 discussion of how to suppress C preprocessor warnings about YYFAIL
703 being unused, see the Bison 2.4.2 NEWS entry.
704
3b1977ea 705** Temporary hack for adding a semicolon to the user action:
a2d05674
JD
706
707 Previously, Bison appended a semicolon to every user action for
708 reductions when the output language defaulted to C (specifically, when
709 neither %yacc, %language, %skeleton, or equivalent command-line
710 options were specified). This allowed actions such as
711
712 exp: exp "+" exp { $$ = $1 + $3 };
713
714 instead of
715
716 exp: exp "+" exp { $$ = $1 + $3; };
717
718 As a first step in removing this misfeature, Bison now issues a
719 warning when it appends a semicolon. Moreover, in cases where Bison
720 cannot easily determine whether a semicolon is needed (for example, an
721 action ending with a cpp directive or a braced compound initializer),
722 it no longer appends one. Thus, the C compiler might now complain
723 about a missing semicolon where it did not before. Future releases of
724 Bison will cease to append semicolons entirely.
725
095a1d11
JD
726** Verbose syntax error message fixes:
727
e4ab1254 728 When %error-verbose or the obsolete "#define YYERROR_VERBOSE" is
6f04ee6c
JD
729 specified, syntax error messages produced by the generated parser
730 include the unexpected token as well as a list of expected tokens.
731 The effect of %nonassoc on these verbose messages has been corrected
732 in two ways, but a more complete fix requires LAC, described above:
095a1d11
JD
733
734*** When %nonassoc is used, there can exist parser states that accept no
735 tokens, and so the parser does not always require a lookahead token
736 in order to detect a syntax error. Because no unexpected token or
737 expected tokens can then be reported, the verbose syntax error
738 message described above is suppressed, and the parser instead
e4ab1254 739 reports the simpler message, "syntax error". Previously, this
095a1d11
JD
740 suppression was sometimes erroneously triggered by %nonassoc when a
741 lookahead was actually required. Now verbose messages are
742 suppressed only when all previous lookaheads have already been
743 shifted or discarded.
744
745*** Previously, the list of expected tokens erroneously included tokens
746 that would actually induce a syntax error because conflicts for them
747 were resolved with %nonassoc in the current parser state. Such
748 tokens are now properly omitted from the list.
749
750*** Expected token lists are still often wrong due to state merging
4c38b19e
JD
751 (from LALR or IELR) and default reductions, which can both add
752 invalid tokens and subtract valid tokens. Canonical LR almost
753 completely fixes this problem by eliminating state merging and
754 default reductions. However, there is one minor problem left even
755 when using canonical LR and even after the fixes above. That is,
756 if the resolution of a conflict with %nonassoc appears in a later
757 parser state than the one at which some syntax error is
758 discovered, the conflicted token is still erroneously included in
759 the expected token list. Bison's new LAC implementation,
760 described above, eliminates this problem and the need for
761 canonical LR. However, LAC is still experimental and is disabled
762 by default.
1fa30307 763
4c2a6e42
JD
764** Java skeleton fixes:
765
766*** A location handling bug has been fixed.
767
768*** The top element of each of the value stack and location stack is now
769 cleared when popped so that it can be garbage collected.
77768165 770
f0e2c228
JD
771*** Parser traces now print the top element of the stack.
772
248dba53
JD
773** -W/--warnings fixes:
774
e4ab1254 775*** Bison now properly recognizes the "no-" versions of categories:
248dba53
JD
776
777 For example, given the following command line, Bison now enables all
778 warnings except warnings for incompatibilities with POSIX Yacc:
779
780 bison -Wall,no-yacc gram.y
781
6f8bdce2
JD
782*** Bison now treats S/R and R/R conflicts like other warnings:
783
784 Previously, conflict reports were independent of Bison's normal
785 warning system. Now, Bison recognizes the warning categories
e4ab1254 786 "conflicts-sr" and "conflicts-rr". This change has important
6f8bdce2
JD
787 consequences for the -W and --warnings command-line options. For
788 example:
789
790 bison -Wno-conflicts-sr gram.y # S/R conflicts not reported
791 bison -Wno-conflicts-rr gram.y # R/R conflicts not reported
792 bison -Wnone gram.y # no conflicts are reported
793 bison -Werror gram.y # any conflict is an error
794
795 However, as before, if the %expect or %expect-rr directive is
796 specified, an unexpected number of conflicts is an error, and an
797 expected number of conflicts is not reported, so -W and --warning
798 then have no effect on the conflict report.
799
e4ab1254 800*** The "none" category no longer disables a preceding "error":
dab96632
JD
801
802 For example, for the following command line, Bison now reports
803 errors instead of warnings for incompatibilities with POSIX Yacc:
804
805 bison -Werror,none,yacc gram.y
806
e4ab1254 807*** The "none" category now disables all Bison warnings:
8ffd7912 808
e4ab1254 809 Previously, the "none" category disabled only Bison warnings for
8ffd7912
JD
810 which there existed a specific -W/--warning category. However,
811 given the following command line, Bison is now guaranteed to
812 suppress all warnings:
813
814 bison -Wnone gram.y
815
9d6af153
JD
816** Precedence directives can now assign token number 0:
817
818 Since Bison 2.3b, which restored the ability of precedence
819 directives to assign token numbers, doing so for token number 0 has
820 produced an assertion failure. For example:
821
822 %left END 0
823
824 This bug has been fixed.
825
7527c744 826* Changes in version 2.4.3 (2010-08-05):
7a9c3cb3 827
4631c34f
JD
828** Bison now obeys -Werror and --warnings=error for warnings about
829 grammar rules that are useless in the parser due to conflicts.
830
7a9c3cb3
JD
831** Problems with spawning M4 on at least FreeBSD 8 and FreeBSD 9 have
832 been fixed.
833
132247cd
JD
834** Failures in the test suite for GCC 4.5 have been fixed.
835
88bb35d6
JD
836** Failures in the test suite for some versions of Sun Studio C++ have
837 been fixed.
838
43fdc9fd
JD
839** Contrary to Bison 2.4.2's NEWS entry, it has been decided that
840 warnings about undefined %prec identifiers will not be converted to
841 errors in Bison 2.5. They will remain warnings, which should be
842 sufficient for POSIX while avoiding backward compatibility issues.
843
cf22447c
JD
844** Minor documentation fixes.
845
ea66d039 846* Changes in version 2.4.2 (2010-03-20):
8defe11b 847
47fa5747
JD
848** Some portability problems that resulted in failures and livelocks
849 in the test suite on some versions of at least Solaris, AIX, HP-UX,
ea66d039
JD
850 RHEL4, and Tru64 have been addressed. As a result, fatal Bison
851 errors should no longer cause M4 to report a broken pipe on the
47fa5747
JD
852 affected platforms.
853
e4ab1254 854** "%prec IDENTIFIER" requires IDENTIFIER to be defined separately.
2c203528
JD
855
856 POSIX specifies that an error be reported for any identifier that does
857 not appear on the LHS of a grammar rule and that is not defined by
858 %token, %left, %right, or %nonassoc. Bison 2.3b and later lost this
859 error report for the case when an identifier appears only after a
860 %prec directive. It is now restored. However, for backward
861 compatibility with recent Bison releases, it is only a warning for
862 now. In Bison 2.5 and later, it will return to being an error.
43fdc9fd
JD
863 [Between the 2.4.2 and 2.4.3 releases, it was decided that this
864 warning will not be converted to an error in Bison 2.5.]
2c203528 865
c5196098
EB
866** Detection of GNU M4 1.4.6 or newer during configure is improved.
867
34731471
JD
868** Warnings from gcc's -Wundef option about undefined YYENABLE_NLS,
869 YYLTYPE_IS_TRIVIAL, and __STRICT_ANSI__ in C/C++ parsers are now
870 avoided.
af2ffe5c 871
966aba65
JD
872** %code is now a permanent feature.
873
874 A traditional Yacc prologue directive is written in the form:
875
876 %{CODE%}
877
878 To provide a more flexible alternative, Bison 2.3b introduced the
879 %code directive with the following forms for C/C++:
880
881 %code {CODE}
882 %code requires {CODE}
883 %code provides {CODE}
884 %code top {CODE}
885
886 These forms are now considered permanent features of Bison. See the
887 %code entries in the section "Bison Declaration Summary" in the Bison
888 manual for a summary of their functionality. See the section
889 "Prologue Alternatives" for a detailed discussion including the
890 advantages of %code over the traditional Yacc prologue directive.
891
892 Bison's Java feature as a whole including its current usage of %code
893 is still considered experimental.
894
41d35e54
JD
895** YYFAIL is deprecated and will eventually be removed.
896
897 YYFAIL has existed for many years as an undocumented feature of
898 deterministic parsers in C generated by Bison. Previously, it was
899 documented for Bison's experimental Java parsers. YYFAIL is no longer
900 documented for Java parsers and is formally deprecated in both cases.
901 Users are strongly encouraged to migrate to YYERROR, which is
902 specified by POSIX.
903
904 Like YYERROR, you can invoke YYFAIL from a semantic action in order to
905 induce a syntax error. The most obvious difference from YYERROR is
906 that YYFAIL will automatically invoke yyerror to report the syntax
907 error so that you don't have to. However, there are several other
908 subtle differences between YYERROR and YYFAIL, and YYFAIL suffers from
e4ab1254 909 inherent flaws when %error-verbose or "#define YYERROR_VERBOSE" is
41d35e54
JD
910 used. For a more detailed discussion, see:
911
912 http://lists.gnu.org/archive/html/bison-patches/2009-12/msg00024.html
913
914 The upcoming Bison 2.5 will remove YYFAIL from Java parsers, but
915 deterministic parsers in C will continue to implement it. However,
916 because YYFAIL is already flawed, it seems futile to try to make new
917 Bison features compatible with it. Thus, during parser generation,
918 Bison 2.5 will produce a warning whenever it discovers YYFAIL in a
919 rule action. In a later release, YYFAIL will be disabled for
e4ab1254 920 %error-verbose and "#define YYERROR_VERBOSE". Eventually, YYFAIL will
41d35e54
JD
921 be removed altogether.
922
923 There exists at least one case where Bison 2.5's YYFAIL warning will
924 be a false positive. Some projects add phony uses of YYFAIL and other
925 Bison-defined macros for the sole purpose of suppressing C
926 preprocessor warnings (from GCC cpp's -Wunused-macros, for example).
927 To avoid Bison's future warning, such YYFAIL uses can be moved to the
e4ab1254 928 epilogue (that is, after the second "%%") in the Bison input file. In
41d35e54
JD
929 this release (2.4.2), Bison already generates its own code to suppress
930 C preprocessor warnings for YYFAIL, so projects can remove their own
931 phony uses of YYFAIL if compatibility with Bison releases prior to
932 2.4.2 is not necessary.
933
dac8cc0d
AD
934** Internationalization.
935
936 Fix a regression introduced in Bison 2.4: Under some circumstances,
937 message translations were not installed although supported by the
938 host system.
939
41930e7a 940* Changes in version 2.4.1 (2008-12-11):
c9ba9e59 941
a957d06c
JD
942** In the GLR defines file, unexpanded M4 macros in the yylval and yylloc
943 declarations have been fixed.
944
738cde3e
AD
945** Temporary hack for adding a semicolon to the user action.
946
947 Bison used to prepend a trailing semicolon at the end of the user
948 action for reductions. This allowed actions such as
949
950 exp: exp "+" exp { $$ = $1 + $3 };
951
952 instead of
953
954 exp: exp "+" exp { $$ = $1 + $3; };
955
e4ab1254 956 Some grammars still depend on this "feature". Bison 2.4.1 restores
d07932ef
JD
957 the previous behavior in the case of C output (specifically, when
958 neither %language or %skeleton or equivalent command-line options
959 are used) to leave more time for grammars depending on the old
960 behavior to be adjusted. Future releases of Bison will disable this
961 feature.
a957d06c
JD
962
963** A few minor improvements to the Bison manual.
c9ba9e59 964
d6fb461d 965* Changes in version 2.4 (2008-11-02):
7bd1665a 966
d6fb461d 967** %language is an experimental feature.
ed4d67dc
JD
968
969 We first introduced this feature in test release 2.3b as a cleaner
970 alternative to %skeleton. Since then, we have discussed the possibility of
971 modifying its effect on Bison's output file names. Thus, in this release,
972 we consider %language to be an experimental feature that will likely evolve
973 in future releases.
7bd1665a 974
d6fb461d 975** Forward compatibility with GNU M4 has been improved.
241fda7a 976
d6fb461d 977** Several bugs in the C++ skeleton and the experimental Java skeleton have been
241fda7a
JD
978 fixed.
979
d6fb461d 980* Changes in version 2.3b (2008-05-27):
35fe0834 981
d6fb461d 982** The quotes around NAME that used to be required in the following directive
d9df47b6
JD
983 are now deprecated:
984
985 %define NAME "VALUE"
986
e4ab1254 987** The directive "%pure-parser" is now deprecated in favor of:
d9df47b6
JD
988
989 %define api.pure
990
991 which has the same effect except that Bison is more careful to warn about
992 unreasonable usage in the latter case.
993
d6fb461d 994** Push Parsing
c373bf8b
JD
995
996 Bison can now generate an LALR(1) parser in C with a push interface. That
e4ab1254
AD
997 is, instead of invoking "yyparse", which pulls tokens from "yylex", you can
998 push one token at a time to the parser using "yypush_parse", which will
c373bf8b
JD
999 return to the caller after processing each token. By default, the push
1000 interface is disabled. Either of the following directives will enable it:
1001
1002 %define api.push_pull "push" // Just push; does not require yylex.
1003 %define api.push_pull "both" // Push and pull; requires yylex.
1004
e4ab1254 1005 See the new section "A Push Parser" in the Bison manual for details.
c373bf8b 1006
59da312b
JD
1007 The current push parsing interface is experimental and may evolve. More user
1008 feedback will help to stabilize it.
1009
d6fb461d 1010** The -g and --graph options now output graphs in Graphviz DOT format,
8e55b3aa
JD
1011 not VCG format. Like --graph, -g now also takes an optional FILE argument
1012 and thus cannot be bundled with other short options.
c373bf8b 1013
d6fb461d 1014** Java
59da312b
JD
1015
1016 Bison can now generate an LALR(1) parser in Java. The skeleton is
e4ab1254 1017 "data/lalr1.java". Consider using the new %language directive instead of
59da312b
JD
1018 %skeleton to select it.
1019
e4ab1254 1020 See the new section "Java Parsers" in the Bison manual for details.
59da312b
JD
1021
1022 The current Java interface is experimental and may evolve. More user
1023 feedback will help to stabilize it.
1024
d6fb461d 1025** %language
59da312b
JD
1026
1027 This new directive specifies the programming language of the generated
d43f77e7
PB
1028 parser, which can be C (the default), C++, or Java. Besides the skeleton
1029 that Bison uses, the directive affects the names of the generated files if
1030 the grammar file's name ends in ".y".
59da312b 1031
d6fb461d 1032** XML Automaton Report
59da312b
JD
1033
1034 Bison can now generate an XML report of the LALR(1) automaton using the new
e4ab1254 1035 "--xml" option. The current XML schema is experimental and may evolve. More
59da312b 1036 user feedback will help to stabilize it.
c373bf8b 1037
d6fb461d 1038** The grammar file may now specify the name of the parser header file using
c373bf8b
JD
1039 %defines. For example:
1040
1041 %defines "parser.h"
1042
d6fb461d 1043** When reporting useless rules, useless nonterminals, and unused terminals,
d80fb37a
JD
1044 Bison now employs the terms "useless in grammar" instead of "useless",
1045 "useless in parser" instead of "never reduced", and "unused in grammar"
1046 instead of "unused".
cff03fb2 1047
d6fb461d 1048** Unreachable State Removal
c373bf8b
JD
1049
1050 Previously, Bison sometimes generated parser tables containing unreachable
31984206
JD
1051 states. A state can become unreachable during conflict resolution if Bison
1052 disables a shift action leading to it from a predecessor state. Bison now:
75ad86ee
JD
1053
1054 1. Removes unreachable states.
1055
1056 2. Does not report any conflicts that appeared in unreachable states.
1057 WARNING: As a result, you may need to update %expect and %expect-rr
1058 directives in existing grammar files.
1059
1060 3. For any rule used only in such states, Bison now reports the rule as
cff03fb2 1061 "useless in parser due to conflicts".
75ad86ee 1062
31984206
JD
1063 This feature can be disabled with the following directive:
1064
1065 %define lr.keep_unreachable_states
1066
e4ab1254 1067 See the %define entry in the "Bison Declaration Summary" in the Bison manual
31984206
JD
1068 for further discussion.
1069
e4ab1254 1070** Lookahead Set Correction in the ".output" Report
b1cc23c4 1071
e4ab1254
AD
1072 When instructed to generate a ".output" file including lookahead sets
1073 (using "--report=lookahead", for example), Bison now prints each reduction's
88c78747
JD
1074 lookahead set only next to the associated state's one item that (1) is
1075 associated with the same rule as the reduction and (2) has its dot at the end
1076 of its RHS. Previously, Bison also erroneously printed the lookahead set
1077 next to all of the state's other items associated with the same rule. This
e4ab1254 1078 bug affected only the ".output" file and not the generated parser source
88c78747
JD
1079 code.
1080
e4ab1254 1081** --report-file=FILE is a new option to override the default ".output" file
59da312b 1082 name.
1bb2bd75 1083
e4ab1254 1084** The "=" that used to be required in the following directives is now
02975b9a
JD
1085 deprecated:
1086
1087 %file-prefix "parser"
1088 %name-prefix "c_"
1089 %output "parser.c"
1090
e4ab1254 1091** An Alternative to "%{...%}" -- "%code QUALIFIER {CODE}"
c373bf8b
JD
1092
1093 Bison 2.3a provided a new set of directives as a more flexible alternative to
8e0a5e9e
JD
1094 the traditional Yacc prologue blocks. Those have now been consolidated into
1095 a single %code directive with an optional qualifier field, which identifies
1096 the purpose of the code and thus the location(s) where Bison should generate
1097 it:
1098
e4ab1254
AD
1099 1. "%code {CODE}" replaces "%after-header {CODE}"
1100 2. "%code requires {CODE}" replaces "%start-header {CODE}"
1101 3. "%code provides {CODE}" replaces "%end-header {CODE}"
1102 4. "%code top {CODE}" replaces "%before-header {CODE}"
8e0a5e9e 1103
e4ab1254
AD
1104 See the %code entries in section "Bison Declaration Summary" in the Bison
1105 manual for a summary of the new functionality. See the new section "Prologue
1106 Alternatives" for a detailed discussion including the advantages of %code
8e0a5e9e
JD
1107 over the traditional Yacc prologues.
1108
1109 The prologue alternatives are experimental. More user feedback will help to
1110 determine whether they should become permanent features.
1111
d6fb461d 1112** Revised warning: unset or unused mid-rule values
17bd8a73
JD
1113
1114 Since Bison 2.2, Bison has warned about mid-rule values that are set but not
1115 used within any of the actions of the parent rule. For example, Bison warns
1116 about unused $2 in:
1117
1118 exp: '1' { $$ = 1; } '+' exp { $$ = $1 + $4; };
1119
1120 Now, Bison also warns about mid-rule values that are used but not set. For
1121 example, Bison warns about unset $$ in the mid-rule action in:
1122
1123 exp: '1' { $1 = 1; } '+' exp { $$ = $2 + $4; };
1124
1125 However, Bison now disables both of these warnings by default since they
1126 sometimes prove to be false alarms in existing grammars employing the Yacc
1127 constructs $0 or $-N (where N is some positive integer).
1128
e4ab1254
AD
1129 To enable these warnings, specify the option "--warnings=midrule-values" or
1130 "-W", which is a synonym for "--warnings=all".
17bd8a73 1131
e4ab1254 1132** Default %destructor or %printer with "<*>" or "<>"
c373bf8b
JD
1133
1134 Bison now recognizes two separate kinds of default %destructor's and
12e35840
JD
1135 %printer's:
1136
e4ab1254 1137 1. Place "<*>" in a %destructor/%printer symbol list to define a default
12e35840
JD
1138 %destructor/%printer for all grammar symbols for which you have formally
1139 declared semantic type tags.
1140
e4ab1254 1141 2. Place "<>" in a %destructor/%printer symbol list to define a default
12e35840
JD
1142 %destructor/%printer for all grammar symbols without declared semantic
1143 type tags.
1144
e4ab1254
AD
1145 Bison no longer supports the "%symbol-default" notation from Bison 2.3a.
1146 "<*>" and "<>" combined achieve the same effect with one exception: Bison no
12e35840
JD
1147 longer applies any %destructor to a mid-rule value if that mid-rule value is
1148 not actually ever referenced using either $$ or $n in a semantic action.
1149
85894313
JD
1150 The default %destructor's and %printer's are experimental. More user
1151 feedback will help to determine whether they should become permanent
1152 features.
1153
e4ab1254 1154 See the section "Freeing Discarded Symbols" in the Bison manual for further
12e35840
JD
1155 details.
1156
d6fb461d 1157** %left, %right, and %nonassoc can now declare token numbers. This is required
e4ab1254 1158 by POSIX. However, see the end of section "Operator Precedence" in the Bison
ab7f29f8
JD
1159 manual for a caveat concerning the treatment of literal strings.
1160
d6fb461d 1161** The nonfunctional --no-parser, -n, and %no-parser options have been
b1cc23c4
JD
1162 completely removed from Bison.
1163
d6fb461d 1164* Changes in version 2.3a, 2006-09-13:
742e4900 1165
d6fb461d 1166** Instead of %union, you can define and use your own union type
ddc8ede1
PE
1167 YYSTYPE if your grammar contains at least one <type> tag.
1168 Your YYSTYPE need not be a macro; it can be a typedef.
1169 This change is for compatibility with other Yacc implementations,
1170 and is required by POSIX.
1171
d6fb461d 1172** Locations columns and lines start at 1.
cd48d21d
AD
1173 In accordance with the GNU Coding Standards and Emacs.
1174
d6fb461d 1175** You may now declare per-type and default %destructor's and %printer's:
ec5479ce
JD
1176
1177 For example:
1178
b2a0b7ca
JD
1179 %union { char *string; }
1180 %token <string> STRING1
1181 %token <string> STRING2
1182 %type <string> string1
1183 %type <string> string2
1184 %union { char character; }
1185 %token <character> CHR
1186 %type <character> chr
1187 %destructor { free ($$); } %symbol-default
1188 %destructor { free ($$); printf ("%d", @$.first_line); } STRING1 string1
1189 %destructor { } <character>
1190
1191 guarantees that, when the parser discards any user-defined symbol that has a
e4ab1254
AD
1192 semantic type tag other than "<character>", it passes its semantic value to
1193 "free". However, when the parser discards a "STRING1" or a "string1", it
1194 also prints its line number to "stdout". It performs only the second
1195 "%destructor" in this case, so it invokes "free" only once.
ec5479ce 1196
85894313
JD
1197 [Although we failed to mention this here in the 2.3a release, the default
1198 %destructor's and %printer's were experimental, and they were rewritten in
1199 future versions.]
1200
e4ab1254
AD
1201** Except for LALR(1) parsers in C with POSIX Yacc emulation enabled (with "-y",
1202 "--yacc", or "%yacc"), Bison no longer generates #define statements for
b931235e
JD
1203 associating token numbers with token names. Removing the #define statements
1204 helps to sanitize the global namespace during preprocessing, but POSIX Yacc
1205 requires them. Bison still generates an enum for token names in all cases.
1206
d6fb461d 1207** Handling of traditional Yacc prologue blocks is now more consistent but
34f98f46 1208 potentially incompatible with previous releases of Bison.
9bc0dd67
JD
1209
1210 As before, you declare prologue blocks in your grammar file with the
e4ab1254 1211 "%{ ... %}" syntax. To generate the pre-prologue, Bison concatenates all
34f98f46
JD
1212 prologue blocks that you've declared before the first %union. To generate
1213 the post-prologue, Bison concatenates all prologue blocks that you've
ddc8ede1 1214 declared after the first %union.
9bc0dd67 1215
34f98f46 1216 Previous releases of Bison inserted the pre-prologue into both the header
9bc0dd67
JD
1217 file and the code file in all cases except for LALR(1) parsers in C. In the
1218 latter case, Bison inserted it only into the code file. For parsers in C++,
1219 the point of insertion was before any token definitions (which associate
1220 token numbers with names). For parsers in C, the point of insertion was
1221 after the token definitions.
1222
1223 Now, Bison never inserts the pre-prologue into the header file. In the code
1224 file, it always inserts it before the token definitions.
1225
d6fb461d 1226** Bison now provides a more flexible alternative to the traditional Yacc
34f98f46
JD
1227 prologue blocks: %before-header, %start-header, %end-header, and
1228 %after-header.
1229
1230 For example, the following declaration order in the grammar file reflects the
1231 order in which Bison will output these code blocks. However, you are free to
1232 declare these code blocks in your grammar file in whatever order is most
1233 convenient for you:
1234
1235 %before-header {
1236 /* Bison treats this block like a pre-prologue block: it inserts it into
1237 * the code file before the contents of the header file. It does *not*
1238 * insert it into the header file. This is a good place to put
1239 * #include's that you want at the top of your code file. A common
e4ab1254 1240 * example is '#include "system.h"'. */
34f98f46
JD
1241 }
1242 %start-header {
1243 /* Bison inserts this block into both the header file and the code file.
1244 * In both files, the point of insertion is before any Bison-generated
1245 * token, semantic type, location type, and class definitions. This is a
1246 * good place to define %union dependencies, for example. */
9bc0dd67
JD
1247 }
1248 %union {
34f98f46
JD
1249 /* Unlike the traditional Yacc prologue blocks, the output order for the
1250 * new %*-header blocks is not affected by their declaration position
1251 * relative to any %union in the grammar file. */
9bc0dd67 1252 }
34f98f46
JD
1253 %end-header {
1254 /* Bison inserts this block into both the header file and the code file.
1255 * In both files, the point of insertion is after the Bison-generated
1256 * definitions. This is a good place to declare or define public
1257 * functions or data structures that depend on the Bison-generated
1258 * definitions. */
9bc0dd67 1259 }
34f98f46
JD
1260 %after-header {
1261 /* Bison treats this block like a post-prologue block: it inserts it into
1262 * the code file after the contents of the header file. It does *not*
1263 * insert it into the header file. This is a good place to declare or
1264 * define internal functions or data structures that depend on the
1265 * Bison-generated definitions. */
1266 }
1267
1268 If you have multiple occurrences of any one of the above declarations, Bison
1269 will concatenate the contents in declaration order.
9bc0dd67 1270
85894313
JD
1271 [Although we failed to mention this here in the 2.3a release, the prologue
1272 alternatives were experimental, and they were rewritten in future versions.]
1273
e4ab1254 1274** The option "--report=look-ahead" has been changed to "--report=lookahead".
9e6e7ed2
PE
1275 The old spelling still works, but is not documented and may be removed
1276 in a future release.
742e4900 1277
d6fb461d 1278* Changes in version 2.3, 2006-06-05:
4ad3ed84 1279
e4ab1254 1280** GLR grammars should now use "YYRECOVERING ()" instead of "YYRECOVERING",
4ad3ed84
PE
1281 for compatibility with LALR(1) grammars.
1282
d6fb461d 1283** It is now documented that any definition of YYSTYPE or YYLTYPE should
4ad3ed84
PE
1284 be to a type name that does not contain parentheses or brackets.
1285
d6fb461d 1286* Changes in version 2.2, 2006-05-19:
193d7c70 1287
d6fb461d 1288** The distribution terms for all Bison-generated parsers now permit
193d7c70
PE
1289 using the parsers in nonfree programs. Previously, this permission
1290 was granted only for Bison-generated LALR(1) parsers in C.
5f4236a0 1291
d6fb461d 1292** %name-prefix changes the namespace name in C++ outputs.
aa08666d 1293
d6fb461d 1294** The C++ parsers export their token_type.
5f4236a0 1295
d6fb461d 1296** Bison now allows multiple %union declarations, and concatenates
d6ca7905
PE
1297 their contents together.
1298
d6fb461d 1299** New warning: unused values
4d7bc38c
PE
1300 Right-hand side symbols whose values are not used are reported,
1301 if the symbols have destructors. For instance:
affac613 1302
8f3596a6 1303 exp: exp "?" exp ":" exp { $1 ? $1 : $3; }
e29f0771
AD
1304 | exp "+" exp
1305 ;
affac613 1306
8f3596a6
AD
1307 will trigger a warning about $$ and $5 in the first rule, and $3 in
1308 the second ($1 is copied to $$ by the default rule). This example
4e26c69e 1309 most likely contains three errors, and could be rewritten as:
affac613 1310
4e26c69e 1311 exp: exp "?" exp ":" exp
e29f0771
AD
1312 { $$ = $1 ? $3 : $5; free ($1 ? $5 : $3); free ($1); }
1313 | exp "+" exp
1314 { $$ = $1 ? $1 : $3; if ($1) free ($3); }
1315 ;
affac613 1316
4e26c69e
PE
1317 However, if the original actions were really intended, memory leaks
1318 and all, the warnings can be suppressed by letting Bison believe the
1319 values are used, e.g.:
721be13c 1320
8f3596a6 1321 exp: exp "?" exp ":" exp { $1 ? $1 : $3; (void) ($$, $5); }
e29f0771
AD
1322 | exp "+" exp { $$ = $1; (void) $3; }
1323 ;
721be13c 1324
84866159
AD
1325 If there are mid-rule actions, the warning is issued if no action
1326 uses it. The following triggers no warning: $1 and $3 are used.
1327
1328 exp: exp { push ($1); } '+' exp { push ($3); sum (); };
1329
721be13c
PE
1330 The warning is intended to help catching lost values and memory leaks.
1331 If a value is ignored, its associated memory typically is not reclaimed.
affac613 1332
d6fb461d 1333** %destructor vs. YYABORT, YYACCEPT, and YYERROR.
9d9b8b70
PE
1334 Destructors are now called when user code invokes YYABORT, YYACCEPT,
1335 and YYERROR, for all objects on the stack, other than objects
1336 corresponding to the right-hand side of the current rule.
a85284cf 1337
d6fb461d 1338** %expect, %expect-rr
035aa4a0
PE
1339 Incorrect numbers of expected conflicts are now actual errors,
1340 instead of warnings.
1341
d6fb461d 1342** GLR, YACC parsers.
4e26c69e
PE
1343 The %parse-params are available in the destructors (and the
1344 experimental printers) as per the documentation.
4b367315 1345
e4ab1254 1346** Bison now warns if it finds a stray "$" or "@" in an action.
ad6a9b97 1347
d6fb461d 1348** %require "VERSION"
4e26c69e
PE
1349 This specifies that the grammar file depends on features implemented
1350 in Bison version VERSION or higher.
b50d2359 1351
d6fb461d 1352** lalr1.cc: The token and value types are now class members.
e14d0ab6
AD
1353 The tokens were defined as free form enums and cpp macros. YYSTYPE
1354 was defined as a free form union. They are now class members:
e4ab1254
AD
1355 tokens are enumerations of the "yy::parser::token" struct, and the
1356 semantic values have the "yy::parser::semantic_type" type.
fb9712a9
AD
1357
1358 If you do not want or can update to this scheme, the directive
e4ab1254 1359 '%define "global_tokens_and_yystype" "1"' triggers the global
b50d2359
AD
1360 definition of tokens and YYSTYPE. This change is suitable both
1361 for previous releases of Bison, and this one.
fb9712a9 1362
b50d2359 1363 If you wish to update, then make sure older version of Bison will
e4ab1254 1364 fail using '%require "2.2"'.
fb9712a9 1365
d6fb461d 1366** DJGPP support added.
193d7c70 1367\f
d6fb461d 1368* Changes in version 2.1, 2005-09-16:
1ce59070 1369
d6fb461d 1370** The C++ lalr1.cc skeleton supports %lex-param.
e14d0ab6 1371
d6fb461d 1372** Bison-generated parsers now support the translation of diagnostics like
baf785db
PE
1373 "syntax error" into languages other than English. The default
1374 language is still English. For details, please see the new
0410a6e0
PE
1375 Internationalization section of the Bison manual. Software
1376 distributors should also see the new PACKAGING file. Thanks to
1377 Bruno Haible for this new feature.
1ce59070 1378
d6fb461d 1379** Wording in the Bison-generated parsers has been changed slightly to
1a059451
PE
1380 simplify translation. In particular, the message "memory exhausted"
1381 has replaced "parser stack overflow", as the old message was not
1382 always accurate for modern Bison-generated parsers.
1383
d6fb461d 1384** Destructors are now called when the parser aborts, for all symbols left
258b75ca
PE
1385 behind on the stack. Also, the start symbol is now destroyed after a
1386 successful parse. In both cases, the behavior was formerly inconsistent.
1387
d6fb461d 1388** When generating verbose diagnostics, Bison-generated parsers no longer
72f000b0
PE
1389 quote the literal strings associated with tokens. For example, for
1390 a syntax error associated with '%token NUM "number"' they might
1391 print 'syntax error, unexpected number' instead of 'syntax error,
1392 unexpected "number"'.
193d7c70 1393\f
d6fb461d 1394* Changes in version 2.0, 2004-12-25:
efeed023 1395
d6fb461d 1396** Possibly-incompatible changes
d7e14fc0 1397
82de6b0d
PE
1398 - Bison-generated parsers no longer default to using the alloca function
1399 (when available) to extend the parser stack, due to widespread
1400 problems in unchecked stack-overflow detection. You can "#define
1401 YYSTACK_USE_ALLOCA 1" to require the use of alloca, but please read
1402 the manual to determine safe values for YYMAXDEPTH in that case.
8dd162d3 1403
82de6b0d
PE
1404 - Error token location.
1405 During error recovery, the location of the syntax error is updated
1406 to cover the whole sequence covered by the error token: it includes
1407 the shifted symbols thrown away during the first part of the error
1408 recovery, and the lookahead rejected during the second part.
18d192f0 1409
82de6b0d
PE
1410 - Semicolon changes:
1411 . Stray semicolons are no longer allowed at the start of a grammar.
1412 . Semicolons are now required after in-grammar declarations.
e342c3be 1413
82de6b0d
PE
1414 - Unescaped newlines are no longer allowed in character constants or
1415 string literals. They were never portable, and GCC 3.4.0 has
1416 dropped support for them. Better diagnostics are now generated if
1417 forget a closing quote.
8dd162d3 1418
82de6b0d 1419 - NUL bytes are no longer allowed in Bison string literals, unfortunately.
f74b6f91 1420
d6fb461d 1421** New features
1452af69 1422
82de6b0d 1423 - GLR grammars now support locations.
4febdd96 1424
82de6b0d
PE
1425 - New directive: %initial-action.
1426 This directive allows the user to run arbitrary code (including
1427 initializing @$) from yyparse before parsing starts.
1452af69 1428
82de6b0d
PE
1429 - A new directive "%expect-rr N" specifies the expected number of
1430 reduce/reduce conflicts in GLR parsers.
1452af69 1431
e4ab1254 1432 - %token numbers can now be hexadecimal integers, e.g., "%token FOO 0x12d".
82de6b0d 1433 This is a GNU extension.
4febdd96 1434
e4ab1254 1435 - The option "--report=lookahead" was changed to "--report=look-ahead".
9e6e7ed2 1436 [However, this was changed back after 2.3.]
1452af69 1437
82de6b0d 1438 - Experimental %destructor support has been added to lalr1.cc.
1452af69 1439
82de6b0d
PE
1440 - New configure option --disable-yacc, to disable installation of the
1441 yacc command and -ly library introduced in 1.875 for POSIX conformance.
6040d338 1442
d6fb461d 1443** Bug fixes
d5a3fe37 1444
82de6b0d
PE
1445 - For now, %expect-count violations are now just warnings, not errors.
1446 This is for compatibility with Bison 1.75 and earlier (when there are
1447 reduce/reduce conflicts) and with Bison 1.30 and earlier (when there
1448 are too many or too few shift/reduce conflicts). However, in future
1449 versions of Bison we plan to improve the %expect machinery so that
1450 these violations will become errors again.
3473d0f8 1451
82de6b0d
PE
1452 - Within Bison itself, numbers (e.g., goto numbers) are no longer
1453 arbitrarily limited to 16-bit counts.
d600ee67 1454
82de6b0d 1455 - Semicolons are now allowed before "|" in grammar rules, as POSIX requires.
d600ee67 1456\f
d6fb461d 1457* Changes in version 1.875, 2003-01-01:
963fcc17 1458
d6fb461d 1459** The documentation license has been upgraded to version 1.2
dc546b0f 1460 of the GNU Free Documentation License.
75eb3bc4 1461
d6fb461d 1462** syntax error processing
75eb3bc4 1463
dc546b0f
PE
1464 - In Yacc-style parsers YYLLOC_DEFAULT is now used to compute error
1465 locations too. This fixes bugs in error-location computation.
75eb3bc4 1466
dc546b0f
PE
1467 - %destructor
1468 It is now possible to reclaim the memory associated to symbols
1469 discarded during error recovery. This feature is still experimental.
20daca06 1470
dc546b0f
PE
1471 - %error-verbose
1472 This new directive is preferred over YYERROR_VERBOSE.
74724a70 1473
dc546b0f
PE
1474 - #defining yyerror to steal internal variables is discouraged.
1475 It is not guaranteed to work forever.
d1de5372 1476
d6fb461d 1477** POSIX conformance
d1de5372 1478
dc546b0f
PE
1479 - Semicolons are once again optional at the end of grammar rules.
1480 This reverts to the behavior of Bison 1.33 and earlier, and improves
1481 compatibility with Yacc.
74724a70 1482
e4ab1254
AD
1483 - "parse error" -> "syntax error"
1484 Bison now uniformly uses the term "syntax error"; formerly, the code
1485 and manual sometimes used the term "parse error" instead. POSIX
1486 requires "syntax error" in diagnostics, and it was thought better to
dc546b0f 1487 be consistent.
74724a70 1488
dc546b0f
PE
1489 - The documentation now emphasizes that yylex and yyerror must be
1490 declared before use. C99 requires this.
d1de5372 1491
dc546b0f
PE
1492 - Bison now parses C99 lexical constructs like UCNs and
1493 backslash-newline within C escape sequences, as POSIX 1003.1-2001 requires.
d1de5372 1494
dc546b0f
PE
1495 - File names are properly escaped in C output. E.g., foo\bar.y is
1496 output as "foo\\bar.y".
6780ca7a 1497
dc546b0f 1498 - Yacc command and library now available
e4ab1254 1499 The Bison distribution now installs a "yacc" command, as POSIX requires.
dc546b0f
PE
1500 Also, Bison now installs a small library liby.a containing
1501 implementations of Yacc-compatible yyerror and main functions.
1502 This library is normally not useful, but POSIX requires it.
6e649e65 1503
dc546b0f 1504 - Type clashes now generate warnings, not errors.
6e649e65 1505
dc546b0f
PE
1506 - If the user does not define YYSTYPE as a macro, Bison now declares it
1507 using typedef instead of defining it as a macro.
1508 For consistency, YYLTYPE is also declared instead of defined.
9501dc6e 1509
d6fb461d 1510** Other compatibility issues
886a425c 1511
e4ab1254
AD
1512 - %union directives can now have a tag before the "{", e.g., the
1513 directive "%union foo {...}" now generates the C code
1514 "typedef union foo { ... } YYSTYPE;"; this is for Yacc compatibility.
1515 The default union tag is "YYSTYPE", for compatibility with Solaris 9 Yacc.
1516 For consistency, YYLTYPE's struct tag is now "YYLTYPE" not "yyltype".
dc546b0f 1517 This is for compatibility with both Yacc and Bison 1.35.
72f889cc 1518
e4ab1254 1519 - ";" is output before the terminating "}" of an action, for
dc546b0f 1520 compatibility with Bison 1.35.
886a425c 1521
dc546b0f 1522 - Bison now uses a Yacc-style format for conflict reports, e.g.,
e4ab1254 1523 "conflicts: 2 shift/reduce, 1 reduce/reduce".
437c2d80 1524
e4ab1254 1525 - "yystype" and "yyltype" are now obsolescent macros instead of being
dc546b0f
PE
1526 typedefs or tags; they are no longer documented and are planned to be
1527 withdrawn in a future release.
2a8d363a 1528
d6fb461d 1529** GLR parser notes
2a8d363a 1530
dc546b0f
PE
1531 - GLR and inline
1532 Users of Bison have to decide how they handle the portability of the
e4ab1254 1533 C keyword "inline".
959e5f51 1534
e4ab1254
AD
1535 - "parsing stack overflow..." -> "parser stack overflow"
1536 GLR parsers now report "parser stack overflow" as per the Bison manual.
900c5db5 1537
18ad57b3
AD
1538** %parse-param and %lex-param
1539 The macros YYPARSE_PARAM and YYLEX_PARAM provide a means to pass
1540 additional context to yyparse and yylex. They suffer from several
1541 shortcomings:
1542
1543 - a single argument only can be added,
1544 - their types are weak (void *),
242cc08e 1545 - this context is not passed to ancillary functions such as yyerror,
18ad57b3
AD
1546 - only yacc.c parsers support them.
1547
1548 The new %parse-param/%lex-param directives provide a more precise control.
1549 For instance:
1550
1551 %parse-param {int *nastiness}
1552 %lex-param {int *nastiness}
1553 %parse-param {int *randomness}
1554
1555 results in the following signatures:
1556
1557 int yylex (int *nastiness);
1558 int yyparse (int *nastiness, int *randomness);
1559
1560 or, if both %pure-parser and %locations are used:
1561
1562 int yylex (YYSTYPE *lvalp, YYLTYPE *llocp, int *nastiness);
1563 int yyparse (int *nastiness, int *randomness);
1564
d6fb461d 1565** Bison now warns if it detects conflicting outputs to the same file,
e4ab1254 1566 e.g., it generates a warning for "bison -d -o foo.h foo.y" since
dc546b0f 1567 that command outputs both code and header to foo.h.
6e40b4eb 1568
d6fb461d 1569** #line in output files
dc546b0f 1570 - --no-line works properly.
6e40b4eb 1571
d6fb461d 1572** Bison can no longer be built by a K&R C compiler; it requires C89 or
6e40b4eb
AD
1573 later to be built. This change originally took place a few versions
1574 ago, but nobody noticed until we recently asked someone to try
1575 building Bison with a K&R C compiler.
d600ee67 1576\f
d6fb461d 1577* Changes in version 1.75, 2002-10-14:
7933f2b5 1578
d6fb461d 1579** Bison should now work on 64-bit hosts.
7933f2b5 1580
d6fb461d 1581** Indonesian translation thanks to Tedi Heriyanto.
7933f2b5 1582
d6fb461d 1583** GLR parsers
f50adbbd
AD
1584 Fix spurious parse errors.
1585
d6fb461d 1586** Pure parsers
f50adbbd
AD
1587 Some people redefine yyerror to steal yyparse' private variables.
1588 Reenable this trick until an official feature replaces it.
1589
d6fb461d 1590** Type Clashes
d90c934c
AD
1591 In agreement with POSIX and with other Yaccs, leaving a default
1592 action is valid when $$ is untyped, and $1 typed:
1593
e29f0771 1594 untyped: ... typed;
d90c934c
AD
1595
1596 but the converse remains an error:
1597
e29f0771 1598 typed: ... untyped;
d90c934c 1599
d6fb461d 1600** Values of mid-rule actions
d90c934c
AD
1601 The following code:
1602
e29f0771 1603 foo: { ... } { $$ = $1; } ...
d90c934c
AD
1604
1605 was incorrectly rejected: $1 is defined in the second mid-rule
1606 action, and is equal to the $$ of the first mid-rule action.
d600ee67 1607\f
d6fb461d 1608* Changes in version 1.50, 2002-10-04:
adc8c848 1609
d6fb461d 1610** GLR parsing
676385e2
PH
1611 The declaration
1612 %glr-parser
1613 causes Bison to produce a Generalized LR (GLR) parser, capable of handling
1614 almost any context-free grammar, ambiguous or not. The new declarations
e8832397 1615 %dprec and %merge on grammar rules allow parse-time resolution of
676385e2
PH
1616 ambiguities. Contributed by Paul Hilfinger.
1617
7933f2b5 1618 Unfortunately Bison 1.50 does not work properly on 64-bit hosts
420f93c8
PE
1619 like the Alpha, so please stick to 32-bit hosts for now.
1620
d6fb461d 1621** Output Directory
8c165d89 1622 When not in Yacc compatibility mode, when the output file was not
e4ab1254
AD
1623 specified, running "bison foo/bar.y" created "foo/bar.c". It
1624 now creates "bar.c".
8c165d89 1625
d6fb461d 1626** Undefined token
007a50a4 1627 The undefined token was systematically mapped to 2 which prevented
e88dbdbf 1628 the use of 2 by the user. This is no longer the case.
007a50a4 1629
d6fb461d 1630** Unknown token numbers
e88dbdbf 1631 If yylex returned an out of range value, yyparse could die. This is
007a50a4
AD
1632 no longer the case.
1633
d6fb461d 1634** Error token
e88dbdbf 1635 According to POSIX, the error token must be 256.
23c5a174
AD
1636 Bison extends this requirement by making it a preference: *if* the
1637 user specified that one of her tokens is numbered 256, then error
1638 will be mapped onto another number.
1639
d6fb461d 1640** Verbose error messages
e4ab1254 1641 They no longer report "..., expecting error or..." for states where
217598da
AD
1642 error recovery is possible.
1643
d6fb461d 1644** End token
e4ab1254 1645 Defaults to "$end" instead of "$".
217598da 1646
d6fb461d 1647** Error recovery now conforms to documentation and to POSIX
68cd8af3
PE
1648 When a Bison-generated parser encounters a syntax error, it now pops
1649 the stack until it finds a state that allows shifting the error
1650 token. Formerly, it popped the stack until it found a state that
1651 allowed some non-error action other than a default reduction on the
1652 error token. The new behavior has long been the documented behavior,
1653 and has long been required by POSIX. For more details, please see
337116ba
PE
1654 Paul Eggert, "Reductions during Bison error handling" (2002-05-20)
1655 <http://lists.gnu.org/archive/html/bug-bison/2002-05/msg00038.html>.
68cd8af3 1656
d6fb461d 1657** Traces
5504898e
AD
1658 Popped tokens and nonterminals are now reported.
1659
d6fb461d 1660** Larger grammars
a861a339
PE
1661 Larger grammars are now supported (larger token numbers, larger grammar
1662 size (= sum of the LHS and RHS lengths), larger LALR tables).
1663 Formerly, many of these numbers ran afoul of 16-bit limits;
1664 now these limits are 32 bits on most hosts.
355e7c1c 1665
d6fb461d 1666** Explicit initial rule
643a5994
AD
1667 Bison used to play hacks with the initial rule, which the user does
1668 not write. It is now explicit, and visible in the reports and
1669 graphs as rule 0.
23c5a174 1670
d6fb461d 1671** Useless rules
643a5994 1672 Before, Bison reported the useless rules, but, although not used,
77714df2 1673 included them in the parsers. They are now actually removed.
23c5a174 1674
d6fb461d 1675** Useless rules, useless nonterminals
6b98e4b5
AD
1676 They are now reported, as a warning, with their locations.
1677
d6fb461d 1678** Rules never reduced
e8832397
AD
1679 Rules that can never be reduced because of conflicts are now
1680 reported.
1681
e4ab1254 1682** Incorrect "Token not used"
11652ab3
AD
1683 On a grammar such as
1684
e29f0771
AD
1685 %token useless useful
1686 %%
1687 exp: '0' %prec useful;
11652ab3
AD
1688
1689 where a token was used to set the precedence of the last rule,
e4ab1254 1690 bison reported both "useful" and "useless" as useless tokens.
11652ab3 1691
d6fb461d 1692** Revert the C++ namespace changes introduced in 1.31
77714df2 1693 as they caused too many portability hassles.
0179dd65 1694
d6fb461d 1695** Default locations
b2d52318
AD
1696 By an accident of design, the default computation of @$ was
1697 performed after another default computation was performed: @$ = @1.
1698 The latter is now removed: YYLLOC_DEFAULT is fully responsible of
1699 the computation of @$.
adc8c848 1700
d6fb461d 1701** Token end-of-file
b7c49edf
AD
1702 The token end of file may be specified by the user, in which case,
1703 the user symbol is used in the reports, the graphs, and the verbose
e4ab1254 1704 error messages instead of "$end", which remains being the default.
b7c49edf 1705 For instance
e29f0771 1706 %token MYEOF 0
b7c49edf 1707 or
e29f0771 1708 %token MYEOF 0 "end of file"
fdbcd8e2 1709
d6fb461d 1710** Semantic parser
fdbcd8e2
AD
1711 This old option, which has been broken for ages, is removed.
1712
d6fb461d 1713** New translations
a861a339 1714 Brazilian Portuguese, thanks to Alexandre Folle de Menezes.
84614e13
AD
1715 Croatian, thanks to Denis Lackovic.
1716
d6fb461d 1717** Incorrect token definitions
e4ab1254
AD
1718 When given
1719 %token 'a' "A"
1720 bison used to output
1721 #define 'a' 65
b87f8b21 1722
d6fb461d 1723** Token definitions as enums
77714df2
AD
1724 Tokens are output both as the traditional #define's, and, provided
1725 the compiler supports ANSI C or is a C++ compiler, as enums.
e88dbdbf 1726 This lets debuggers display names instead of integers.
77714df2 1727
d6fb461d 1728** Reports
ec3bc396
AD
1729 In addition to --verbose, bison supports --report=THINGS, which
1730 produces additional information:
b408954b
AD
1731 - itemset
1732 complete the core item sets with their closure
e4ab1254 1733 - lookahead [changed to "look-ahead" in 1.875e through 2.3, but changed back]
9e6e7ed2 1734 explicitly associate lookahead tokens to items
b408954b
AD
1735 - solved
1736 describe shift/reduce conflicts solving.
1737 Bison used to systematically output this information on top of
1738 the report. Solved conflicts are now attached to their states.
ec3bc396 1739
d6fb461d 1740** Type clashes
9af3fbce
AD
1741 Previous versions don't complain when there is a type clash on
1742 the default action if the rule has a mid-rule action, such as in:
1743
e29f0771
AD
1744 %type <foo> bar
1745 %%
1746 bar: '0' {} '0';
9af3fbce
AD
1747
1748 This is fixed.
a861a339 1749
d6fb461d 1750** GNU M4 is now required when using Bison.
f987e9d2 1751\f
d6fb461d 1752* Changes in version 1.35, 2002-03-25:
76551463 1753
d6fb461d 1754** C Skeleton
76551463
AD
1755 Some projects use Bison's C parser with C++ compilers, and define
1756 YYSTYPE as a class. The recent adjustment of C parsers for data
1757 alignment and 64 bit architectures made this impossible.
1758
1759 Because for the time being no real solution for C++ parser
1760 generation exists, kludges were implemented in the parser to
1761 maintain this use. In the future, when Bison has C++ parsers, this
1762 kludge will be disabled.
1763
1764 This kludge also addresses some C++ problems when the stack was
1765 extended.
76551463 1766\f
d6fb461d 1767* Changes in version 1.34, 2002-03-12:
76551463 1768
d6fb461d 1769** File name clashes are detected
76551463 1770 $ bison foo.y -d -o foo.x
e4ab1254 1771 fatal error: header and parser would both be named "foo.x"
76551463 1772
e4ab1254 1773** A missing ";" at the end of a rule triggers a warning
76551463
AD
1774 In accordance with POSIX, and in agreement with other
1775 Yacc implementations, Bison will mandate this semicolon in the near
1776 future. This eases the implementation of a Bison parser of Bison
1777 grammars by making this grammar LALR(1) instead of LR(2). To
1778 facilitate the transition, this release introduces a warning.
1779
d6fb461d 1780** Revert the C++ namespace changes introduced in 1.31, as they caused too
76551463
AD
1781 many portability hassles.
1782
d6fb461d 1783** DJGPP support added.
76551463 1784
d6fb461d 1785** Fix test suite portability problems.
76551463 1786\f
d6fb461d 1787* Changes in version 1.33, 2002-02-07:
76551463 1788
d6fb461d 1789** Fix C++ issues
76551463
AD
1790 Groff could not be compiled for the definition of size_t was lacking
1791 under some conditions.
1792
d6fb461d 1793** Catch invalid @n
76551463
AD
1794 As is done with $n.
1795\f
d6fb461d 1796* Changes in version 1.32, 2002-01-23:
76551463 1797
d6fb461d 1798** Fix Yacc output file names
76551463 1799
d6fb461d 1800** Portability fixes
76551463 1801
d6fb461d 1802** Italian, Dutch translations
76551463 1803\f
d6fb461d 1804* Changes in version 1.31, 2002-01-14:
52d1aeee 1805
d6fb461d 1806** Many Bug Fixes
52d1aeee 1807
d6fb461d 1808** GNU Gettext and %expect
52d1aeee
MA
1809 GNU Gettext asserts 10 s/r conflicts, but there are 7. Now that
1810 Bison dies on incorrect %expectations, we fear there will be
1811 too many bug reports for Gettext, so _for the time being_, %expect
e4ab1254 1812 does not trigger an error when the input file is named "plural.y".
52d1aeee 1813
d6fb461d 1814** Use of alloca in parsers
52d1aeee
MA
1815 If YYSTACK_USE_ALLOCA is defined to 0, then the parsers will use
1816 malloc exclusively. Since 1.29, but was not NEWS'ed.
1817
1818 alloca is used only when compiled with GCC, to avoid portability
1819 problems as on AIX.
1820
d6fb461d 1821** yyparse now returns 2 if memory is exhausted; formerly it dumped core.
b47dbebe 1822
d6fb461d 1823** When the generated parser lacks debugging code, YYDEBUG is now 0
52d1aeee
MA
1824 (as POSIX requires) instead of being undefined.
1825
d6fb461d 1826** User Actions
52d1aeee
MA
1827 Bison has always permitted actions such as { $$ = $1 }: it adds the
1828 ending semicolon. Now if in Yacc compatibility mode, the semicolon
1829 is no longer output: one has to write { $$ = $1; }.
1830
d6fb461d 1831** Better C++ compliance
52d1aeee 1832 The output parsers try to respect C++ namespaces.
76551463 1833 [This turned out to be a failed experiment, and it was reverted later.]
52d1aeee 1834
d6fb461d 1835** Reduced Grammars
52d1aeee
MA
1836 Fixed bugs when reporting useless nonterminals.
1837
d6fb461d 1838** 64 bit hosts
52d1aeee
MA
1839 The parsers work properly on 64 bit hosts.
1840
d6fb461d 1841** Error messages
52d1aeee
MA
1842 Some calls to strerror resulted in scrambled or missing error messages.
1843
d6fb461d 1844** %expect
52d1aeee
MA
1845 When the number of shift/reduce conflicts is correct, don't issue
1846 any warning.
1847
d6fb461d 1848** The verbose report includes the rule line numbers.
52d1aeee 1849
d6fb461d 1850** Rule line numbers are fixed in traces.
52d1aeee 1851
d6fb461d 1852** Swedish translation
52d1aeee 1853
d6fb461d 1854** Parse errors
52d1aeee
MA
1855 Verbose parse error messages from the parsers are better looking.
1856 Before: parse error: unexpected `'/'', expecting `"number"' or `'-'' or `'(''
1857 Now: parse error: unexpected '/', expecting "number" or '-' or '('
1858
d6fb461d 1859** Fixed parser memory leaks.
52d1aeee
MA
1860 When the generated parser was using malloc to extend its stacks, the
1861 previous allocations were not freed.
1862
d6fb461d 1863** Fixed verbose output file.
52d1aeee
MA
1864 Some newlines were missing.
1865 Some conflicts in state descriptions were missing.
1866
d6fb461d 1867** Fixed conflict report.
52d1aeee
MA
1868 Option -v was needed to get the result.
1869
d6fb461d 1870** %expect
52d1aeee
MA
1871 Was not used.
1872 Mismatches are errors, not warnings.
1873
d6fb461d 1874** Fixed incorrect processing of some invalid input.
52d1aeee 1875
d6fb461d 1876** Fixed CPP guards: 9foo.h uses BISON_9FOO_H instead of 9FOO_H.
52d1aeee 1877
d6fb461d 1878** Fixed some typos in the documentation.
52d1aeee 1879
d6fb461d 1880** %token MY_EOF 0 is supported.
52d1aeee
MA
1881 Before, MY_EOF was silently renumbered as 257.
1882
d6fb461d 1883** doc/refcard.tex is updated.
52d1aeee 1884
d6fb461d 1885** %output, %file-prefix, %name-prefix.
52d1aeee
MA
1886 New.
1887
d6fb461d 1888** --output
e4ab1254 1889 New, aliasing "--output-file".
52d1aeee 1890\f
d6fb461d 1891* Changes in version 1.30, 2001-10-26:
342b8b6e 1892
e4ab1254
AD
1893** "--defines" and "--graph" have now an optional argument which is the
1894 output file name. "-d" and "-g" do not change; they do not take any
342b8b6e
AD
1895 argument.
1896
e4ab1254 1897** "%source_extension" and "%header_extension" are removed, failed
342b8b6e
AD
1898 experiment.
1899
d6fb461d 1900** Portability fixes.
f987e9d2 1901\f
d6fb461d 1902* Changes in version 1.29, 2001-09-07:
342b8b6e 1903
d6fb461d 1904** The output file does not define const, as this caused problems when used
342b8b6e
AD
1905 with common autoconfiguration schemes. If you still use ancient compilers
1906 that lack const, compile with the equivalent of the C compiler option
e4ab1254 1907 "-Dconst=". Autoconf's AC_C_CONST macro provides one way to do this.
342b8b6e 1908
e4ab1254 1909** Added "-g" and "--graph".
f87a2205 1910
d6fb461d 1911** The Bison manual is now distributed under the terms of the GNU FDL.
f2b5126e 1912
d6fb461d 1913** The input and the output files has automatically a similar extension.
234a3be3 1914
d6fb461d 1915** Russian translation added.
f87a2205 1916
d6fb461d 1917** NLS support updated; should hopefully be less troublesome.
f87a2205 1918
d6fb461d 1919** Added the old Bison reference card.
c33638bb 1920
e4ab1254 1921** Added "--locations" and "%locations".
6deb4447 1922
e4ab1254 1923** Added "-S" and "--skeleton".
cd5bd6ac 1924
e4ab1254 1925** "%raw", "-r", "--raw" is disabled.
62ab6972 1926
d6fb461d 1927** Special characters are escaped when output. This solves the problems
cd5bd6ac
AD
1928 of the #line lines with path names including backslashes.
1929
d6fb461d 1930** New directives.
e4ab1254
AD
1931 "%yacc", "%fixed_output_files", "%defines", "%no_parser", "%verbose",
1932 "%debug", "%source_extension" and "%header_extension".
f987e9d2 1933
d6fb461d 1934** @$
f987e9d2 1935 Automatic location tracking.
f87a2205 1936\f
d6fb461d 1937* Changes in version 1.28, 1999-07-06:
d2e00347 1938
d6fb461d 1939** Should compile better now with K&R compilers.
d2e00347 1940
d6fb461d 1941** Added NLS.
d2e00347 1942
d6fb461d 1943** Fixed a problem with escaping the double quote character.
d2e00347 1944
d6fb461d 1945** There is now a FAQ.
d2e00347 1946\f
d6fb461d 1947* Changes in version 1.27:
5c31c3c2 1948
d6fb461d 1949** The make rule which prevented bison.simple from being created on
5c31c3c2
JT
1950 some systems has been fixed.
1951\f
d6fb461d 1952* Changes in version 1.26:
4be07551 1953
7e508a2b 1954** Bison now uses Automake.
4be07551 1955
d6fb461d 1956** New mailing lists: <bug-bison@gnu.org> and <help-bison@gnu.org>.
4be07551 1957
d6fb461d 1958** Token numbers now start at 257 as previously documented, not 258.
4be07551 1959
d6fb461d 1960** Bison honors the TMPDIR environment variable.
4be07551 1961
d6fb461d 1962** A couple of buffer overruns have been fixed.
f51dbca1 1963
d6fb461d 1964** Problems when closing files should now be reported.
f51dbca1 1965
d6fb461d 1966** Generated parsers should now work even on operating systems which do
f51dbca1 1967 not provide alloca().
4be07551 1968\f
d6fb461d 1969* Changes in version 1.25, 1995-10-16:
df8878c5 1970
d6fb461d 1971** Errors in the input grammar are not fatal; Bison keeps reading
df8878c5 1972the grammar file, and reports all the errors found in it.
8c44d3ec 1973
d6fb461d 1974** Tokens can now be specified as multiple-character strings: for
df8878c5 1975example, you could use "<=" for a token which looks like <=, instead
7e508a2b 1976of choosing a name like LESSEQ.
df8878c5 1977
d6fb461d 1978** The %token_table declaration says to write a table of tokens (names
df8878c5
RS
1979and numbers) into the parser file. The yylex function can use this
1980table to recognize multiple-character string tokens, or for other
1981purposes.
1982
d6fb461d 1983** The %no_lines declaration says not to generate any #line preprocessor
df8878c5
RS
1984directives in the parser file.
1985
d6fb461d 1986** The %raw declaration says to use internal Bison token numbers, not
df8878c5
RS
1987Yacc-compatible token numbers, when token names are defined as macros.
1988
d6fb461d 1989** The --no-parser option produces the parser tables without including
df8878c5
RS
1990the parser engine; a project can now use its own parser engine.
1991The actions go into a separate file called NAME.act, in the form of
1992a switch statement body.
1993\f
d6fb461d 1994* Changes in version 1.23:
6780ca7a 1995
4d019228
DM
1996The user can define YYPARSE_PARAM as the name of an argument to be
1997passed into yyparse. The argument should have type void *. It should
1998actually point to an object. Grammar actions can access the variable
1999by casting it to the proper pointer type.
6780ca7a 2000
6780ca7a 2001Line numbers in output file corrected.
6780ca7a 2002\f
d6fb461d 2003* Changes in version 1.22:
6780ca7a
DM
2004
2005--help option added.
6780ca7a 2006\f
d6fb461d 2007* Changes in version 1.20:
6780ca7a
DM
2008
2009Output file does not redefine const for C++.
9f4503d6 2010
76551463
AD
2011-----
2012
c932d613 2013Copyright (C) 1995-2012 Free Software Foundation, Inc.
76551463 2014
8defe11b 2015This file is part of Bison, the GNU Parser Generator.
76551463 2016
f16b0819 2017This program is free software: you can redistribute it and/or modify
76551463 2018it under the terms of the GNU General Public License as published by
f16b0819
PE
2019the Free Software Foundation, either version 3 of the License, or
2020(at your option) any later version.
76551463 2021
f16b0819 2022This program is distributed in the hope that it will be useful,
76551463
AD
2023but WITHOUT ANY WARRANTY; without even the implied warranty of
2024MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
2025GNU General Public License for more details.
2026
2027You should have received a copy of the GNU General Public License
f16b0819 2028along with this program. If not, see <http://www.gnu.org/licenses/>.
7e508a2b
AD
2029
2030 LocalWords: yacc YYBACKUP glr GCC lalr ArrayIndexOutOfBoundsException nullptr
2031 LocalWords: cplusplus liby rpl fprintf mfcalc Wyacc stmt cond expr mk sym lr
2032 LocalWords: IELR ielr Lookahead YYERROR nonassoc LALR's api lookaheads yychar
2033 LocalWords: destructor lookahead YYRHSLOC YYLLOC Rhs ifndef YYFAIL cpp sr rr
2034 LocalWords: preprocessor initializer Wno Wnone Werror FreeBSD prec livelocks
2035 LocalWords: Solaris AIX UX RHEL Tru LHS gcc's Wundef YYENABLE NLS YYLTYPE VCG
2036 LocalWords: yyerror cpp's Wunused yylval yylloc prepend yyparse yylex yypush
2037 LocalWords: Graphviz xml nonterminals midrule destructor's YYSTYPE typedef ly
2038 LocalWords: CHR chr printf stdout namespace preprocessing enum pre include's
2039 LocalWords: YYRECOVERING nonfree destructors YYABORT YYACCEPT params enums de
2040 LocalWords: struct yystype DJGPP lex param Haible NUM alloca YYSTACK NUL goto
2041 LocalWords: YYMAXDEPTH Unescaped UCNs YYLTYPE's yyltype typedefs inline Yaccs
2042 LocalWords: Heriyanto Reenable dprec Hilfinger Eggert MYEOF Folle Menezes EOF
242cc08e 2043 LocalWords: Lackovic define's itemset Groff Gettext malloc NEWS'ed YYDEBUG YY
7e508a2b 2044 LocalWords: namespaces strerror const autoconfiguration Dconst Autoconf's FDL
242cc08e
AD
2045 LocalWords: Automake TMPDIR LESSEQ ylwrap endif yydebug YYTOKEN YYLSP ival hh
2046 LocalWords: extern YYTOKENTYPE TOKENTYPE yytokentype tokentype STYPE lval pdf
7287be84
AD
2047 LocalWords: lang yyoutput dvi html ps POSIX lvalp llocp calc yyo fval Wmaybe
2048 LocalWords: yyvsp pragmas noreturn java's
7e508a2b
AD
2049
2050Local Variables:
2051mode: outline
e4ab1254 2052fill-column: 76
7e508a2b 2053End: