]> git.saurik.com Git - bison.git/blame - NEWS
version 2.6.90
[bison.git] / NEWS
CommitLineData
ed7658fe 1GNU Bison NEWS
3af4feb2 2
a24e6b0d 3* Noteworthy changes in release 2.6.90 (2012-12-07) [beta]
933ec544 4
edf9a06f
AD
5** Bug fixes
6
7 Warnings about uninitialized yylloc in yyparse have been fixed.
8
effd30c0
TR
9** %language is no longer an experimental feature.
10
11 The introduction of this feature, in 2.4, was four years ago. The --language
12 option and the %language directive are no longer experimental.
13
7bada535
TR
14** New format for error reports: carets
15
16 Caret errors have been added to Bison, for example (taken from the
17 documentation):
18
19 input.y:3.20-23: error: ambiguous reference: '$exp'
fb6040f0 20 exp: exp '+' exp { $exp = $1 + $3; };
7bada535 21 ^^^^
fb6040f0
TR
22 input.y:3.1-3: refers to: $exp at $$
23 exp: exp '+' exp { $exp = $1 + $3; };
24 ^^^
25 input.y:3.6-8: refers to: $exp at $1
26 exp: exp '+' exp { $exp = $1 + $3; };
27 ^^^
28 input.y:3.14-16: refers to: $exp at $3
29 exp: exp '+' exp { $exp = $1 + $3; };
30 ^^^
7bada535
TR
31
32 The default behaviour for now is still not to display these unless explictly
33 asked with -fall of -fcaret. However, in a later release, it will be made the
34 default behavior (but may still be deactivated with -fno-caret).
35
1f1bd572
TR
36** New value for %define variable: api.pure full
37
38 The %define variable api.pure requests a pure (reentrant) parser. However,
39 for historical reasons, using it in a location-tracking Yacc parser resulted
fb6040f0 40 in a yyerror function that did not take a location as a parameter. With this
1f1bd572
TR
41 new value, the user may request a better pure parser, where yyerror does take
42 a location as a parameter (in location-tracking parsers).
43
44 The use of "%define api.pure true" is deprecated in favor of this new
45 "%define api.pure full".
46
5a05f42e 47** Changes in the format of error messages
cbaea010
TR
48
49 This used to be the format of many error reports:
50
51 foo.y:5.10-24: result type clash on merge function 'merge': <t3> != <t2>
52 foo.y:4.13-27: previous declaration
53
54 It is now:
55
56 foo.y:5.10-25: result type clash on merge function 'merge': <t3> != <t2>
57 foo.y:4.13-27: previous declaration
58
d3e4409a
AD
59** Exception safety (lalr1.cc)
60
61 The parse function now catches exceptions, uses the %destructors to
62 release memory (the lookahead symbol and the symbols pushed on the stack)
7287be84 63 before re-throwing the exception.
d3e4409a
AD
64
65 This feature is somewhat experimental. User feedback would be
66 appreciated.
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
9c16d399 87** Graph improvements in DOT and XSLT
fc4fdd62
TR
88
89 The graphical presentation of the states is more readable: their shape is
90 now rectangular, the state number is clearly displayed, and the items are
91 numbered and left-justified.
92
93 The reductions are now explicitly represented as transitions to other
94 diamond shaped nodes.
95
9c16d399
TR
96 These changes are present in both --graph output and xml2dot.xsl XSLT
97 processing, with minor (documented) differences.
98
99 Two nodes were added to the documentation: Xml and Graphviz.
100
53e2cd1e
AD
101** Documentation
102
103 The sections about shift/reduce and reduce/reduce conflicts resolution
104 have been fixed and extended.
9d2423f5 105
9d3f7eaf 106* Noteworthy changes in release 2.6.5 (2012-11-07) [stable]
6f1360bd 107
a68b1f23
AD
108 We consider compiler warnings about Bison generated parsers to be bugs.
109 Rather than working around them in your own project, please consider
110 reporting them to us.
111
112** Bug fixes
113
114 Warnings about uninitialized yylval and/or yylloc for push parsers with a
115 pure interface have been fixed for GCC 4.0 up to 4.8, and Clang 2.9 to
116 3.2.
117
118 Other issues in the test suite have been addressed.
6f1360bd 119
95066e92
AD
120 Nul characters are correctly displayed in error messages.
121
a1d1ab50
AD
122 When possible, yylloc is correctly initialized before calling yylex. It
123 is no longer necessary to initialize it in the %initial-action.
124
0ac15849 125* Noteworthy changes in release 2.6.4 (2012-10-23) [stable]
a4eb820f 126
468455e1 127 Bison 2.6.3's --version was incorrect. This release fixes this issue.
a4eb820f 128
6eb8f74f 129* Noteworthy changes in release 2.6.3 (2012-10-22) [stable]
933ec544 130
6b4cb804
AD
131** Bug fixes
132
9a50c55a 133 Bugs and portability issues in the test suite have been fixed.
6b4cb804
AD
134
135 Some errors in translations have been addressed, and --help now directs
136 users to the appropriate place to report them.
137
138 Stray Info files shipped by accident are removed.
139
140 Incorrect definitions of YY_, issued by yacc.c when no parser header is
141 generated, are removed.
933ec544 142
90b1335a
TR
143 All the generated headers are self-contained.
144
6192d2c6
AD
145** Header guards (yacc.c, glr.c, glr.cc)
146
147 In order to avoid collisions, the header guards are now
148 YY_<PREFIX>_<FILE>_INCLUDED, instead of merely <PREFIX>_<FILE>.
149 For instance the header generated from
150
151 %define api.prefix "calc"
152 %defines "lib/parse.h"
153
154 will use YY_CALC_LIB_PARSE_H_INCLUDED as guard.
155
7b70847e 156** Fix compiler warnings in the generated parser (yacc.c, glr.c)
8f8439ce
AD
157
158 The compilation of pure parsers (%define api.pure) can trigger GCC
159 warnings such as:
160
161 input.c: In function 'yyparse':
162 input.c:1503:12: warning: 'yylval' may be used uninitialized in this
163 function [-Wmaybe-uninitialized]
164 *++yyvsp = yylval;
165 ^
166
167 This is now fixed; pragmas to avoid these warnings are no longer needed.
168
7b70847e
AD
169 Warnings from clang ("equality comparison with extraneous parentheses" and
170 "function declared 'noreturn' should not return") have also been
171 addressed.
172
e1eeecd3 173* Noteworthy changes in release 2.6.2 (2012-08-03) [stable]
9c26b8fc 174
43ca8040
AD
175** Bug fixes
176
177 Buffer overruns, complaints from Flex, and portability issues in the test
178 suite have been fixed.
179
c9d546b2
AD
180** Spaces in %lex- and %parse-param (lalr1.cc, glr.cc)
181
182 Trailing end-of-lines in %parse-param or %lex-param would result in
183 invalid C++. This is fixed.
9c26b8fc 184
dcd5344d
AD
185** Spurious spaces and end-of-lines
186
187 The generated files no longer end (nor start) with empty lines.
188
77b214ef 189* Noteworthy changes in release 2.6.1 (2012-07-30) [stable]
a4107f24 190
cbaea010 191 Bison no longer executes user-specified M4 code when processing a grammar.
8617d87e 192
e20e6a50
AD
193** Future Changes
194
195 In addition to the removal of the features announced in Bison 2.6, the
196 next major release will remove the "Temporary hack for adding a semicolon
197 to the user action", as announced in the release 2.5. Instead of:
198
199 exp: exp "+" exp { $$ = $1 + $3 };
200
201 write:
202
203 exp: exp "+" exp { $$ = $1 + $3; };
204
8617d87e
AD
205** Bug fixes
206
0e164d43
AD
207*** Type names are now properly escaped.
208
209*** glr.cc: set_debug_level and debug_level work as expected.
a4107f24 210
26313726
AD
211*** Stray @ or $ in actions
212
213 While Bison used to warn about stray $ or @ in action rules, it did not
214 for other actions such as printers, destructors, or initial actions. It
215 now does.
216
cd735a8c 217** Type names in actions
4982f078
AD
218
219 For consistency with rule actions, it is now possible to qualify $$ by a
cd735a8c 220 type-name in destructors, printers, and initial actions. For instance:
4982f078
AD
221
222 %printer { fprintf (yyo, "(%d, %f)", $<ival>$, $<fval>$); } <*> <>;
223
224 will display two values for each typed and untyped symbol (provided
cd735a8c 225 that YYSTYPE has both "ival" and "fval" fields).
4982f078 226
1505e8bb 227* Noteworthy changes in release 2.6 (2012-07-19) [stable]
ad6b775f 228
c2425191 229** Future Changes
0f11eec2 230
55d1006f
AD
231 The next major release of Bison will drop support for the following
232 deprecated features. Please report disagreements to bug-bison@gnu.org.
0f11eec2 233
aaf61036 234*** K&R C parsers
55d1006f
AD
235
236 Support for generating parsers in K&R C will be removed. Parsers
242cc08e 237 generated for C support ISO C90, and are tested with ISO C99 and ISO C11
55d1006f
AD
238 compilers.
239
258cddbc 240*** Features deprecated since Bison 1.875
55d1006f 241
258cddbc
AD
242 The definitions of yystype and yyltype will be removed; use YYSTYPE and
243 YYLTYPE.
55d1006f 244
258cddbc
AD
245 YYPARSE_PARAM and YYLEX_PARAM, deprecated in favor of %parse-param and
246 %lex-param, will no longer be supported.
247
248 Support for the preprocessor symbol YYERROR_VERBOSE will be removed, use
249 %error-verbose.
55d1006f
AD
250
251*** The generated header will be included (yacc.c)
0f11eec2
AD
252
253 Instead of duplicating the content of the generated header (definition of
55d1006f
AD
254 YYSTYPE, yyparse declaration etc.), the generated parser will include it,
255 as is already the case for GLR or C++ parsers. This change is deferred
256 because existing versions of ylwrap (e.g., Automake 1.12.1) do not support
257 it.
0f11eec2 258
c2425191 259** Generated Parser Headers
56ca3d8f 260
258cddbc 261*** Guards (yacc.c, glr.c, glr.cc)
c3e9f08f
AD
262
263 The generated headers are now guarded, as is already the case for C++
242cc08e 264 parsers (lalr1.cc). For instance, with --defines=foo.h:
c3e9f08f 265
e29f0771
AD
266 #ifndef YY_FOO_H
267 # define YY_FOO_H
268 ...
269 #endif /* !YY_FOO_H */
c3e9f08f 270
258cddbc 271*** New declarations (yacc.c, glr.c)
56ca3d8f
AD
272
273 The generated header now declares yydebug and yyparse. Both honor
274 --name-prefix=bar_, and yield
275
e29f0771 276 int bar_parse (void);
56ca3d8f
AD
277
278 rather than
279
e29f0771
AD
280 #define yyparse bar_parse
281 int yyparse (void);
56ca3d8f
AD
282
283 in order to facilitate the inclusion of several parser headers inside a
284 single compilation unit.
c3e9f08f 285
258cddbc
AD
286*** Exported symbols in C++
287
288 The symbols YYTOKEN_TABLE and YYERROR_VERBOSE, which were defined in the
289 header, are removed, as they prevent the possibility of including several
290 generated headers from a single compilation unit.
291
694af10c
AD
292*** YYLSP_NEEDED
293
294 For the same reasons, the undocumented and unused macro YYLSP_NEEDED is no
295 longer defined.
296
4b3847c3
AD
297** New %define variable: api.prefix
298
299 Now that the generated headers are more complete and properly protected
300 against multiple inclusions, constant names, such as YYSTYPE are a
301 problem. While yyparse and others are properly renamed by %name-prefix,
302 YYSTYPE, YYDEBUG and others have never been affected by it. Because it
303 would introduce backward compatibility issues in projects not expecting
304 YYSTYPE to be renamed, instead of changing the behavior of %name-prefix,
305 it is deprecated in favor of a new %define variable: api.prefix.
306
307 The following examples compares both:
308
309 %name-prefix "bar_" | %define api.prefix "bar_"
310 %token <ival> FOO %token <ival> FOO
311 %union { int ival; } %union { int ival; }
312 %% %%
313 exp: 'a'; exp: 'a';
314
315 bison generates:
316
317 #ifndef BAR_FOO_H #ifndef BAR_FOO_H
318 # define BAR_FOO_H # define BAR_FOO_H
319
320 /* Enabling traces. */ /* Enabling traces. */
5f108727
AD
321 # ifndef YYDEBUG | # ifndef BAR_DEBUG
322 > # if defined YYDEBUG
323 > # if YYDEBUG
324 > # define BAR_DEBUG 1
325 > # else
326 > # define BAR_DEBUG 0
327 > # endif
328 > # else
329 # define YYDEBUG 0 | # define BAR_DEBUG 0
330 > # endif
331 # endif | # endif
332
333 # if YYDEBUG | # if BAR_DEBUG
4b3847c3
AD
334 extern int bar_debug; extern int bar_debug;
335 # endif # endif
336
337 /* Tokens. */ /* Tokens. */
338 # ifndef YYTOKENTYPE | # ifndef BAR_TOKENTYPE
339 # define YYTOKENTYPE | # define BAR_TOKENTYPE
340 enum yytokentype { | enum bar_tokentype {
341 FOO = 258 FOO = 258
342 }; };
343 # endif # endif
344
345 #if ! defined YYSTYPE \ | #if ! defined BAR_STYPE \
346 && ! defined YYSTYPE_IS_DECLARED | && ! defined BAR_STYPE_IS_DECLARED
347 typedef union YYSTYPE | typedef union BAR_STYPE
348 { {
349 int ival; int ival;
350 } YYSTYPE; | } BAR_STYPE;
351 # define YYSTYPE_IS_DECLARED 1 | # define BAR_STYPE_IS_DECLARED 1
352 #endif #endif
353
354 extern YYSTYPE bar_lval; | extern BAR_STYPE bar_lval;
355
356 int bar_parse (void); int bar_parse (void);
357
358 #endif /* !BAR_FOO_H */ #endif /* !BAR_FOO_H */
359
dfaac272 360* Noteworthy changes in release 2.5.1 (2012-06-05) [stable]
90dcd415 361
debe2c03 362** Future changes:
041308d0 363
e4ab1254 364 The next major release will drop support for generating parsers in K&R C.
ef51bfa7 365
466b4cf2 366** yacc.c: YYBACKUP works as expected.
94556574 367
d834eca0
AD
368** glr.c improvements:
369
370*** Location support is eliminated when not requested:
371
e4ab1254
AD
372 GLR parsers used to include location-related code even when locations were
373 not requested, and therefore not even usable.
d834eca0
AD
374
375*** __attribute__ is preserved:
debe2c03 376
e4ab1254
AD
377 __attribute__ is no longer disabled when __STRICT_ANSI__ is defined (i.e.,
378 when -std is passed to GCC).
debe2c03 379
466b4cf2 380** lalr1.java: several fixes:
041308d0 381
e4ab1254
AD
382 The Java parser no longer throws ArrayIndexOutOfBoundsException if the
383 first token leads to a syntax error. Some minor clean ups.
041308d0 384
22172d47
AD
385** Changes for C++:
386
387*** C++11 compatibility:
ef51bfa7 388
e4ab1254
AD
389 C and C++ parsers use "nullptr" instead of "0" when __cplusplus is 201103L
390 or higher.
ef51bfa7 391
22172d47
AD
392*** Header guards
393
394 The header files such as "parser.hh", "location.hh", etc. used a constant
395 name for preprocessor guards, for instance:
396
e29f0771
AD
397 #ifndef BISON_LOCATION_HH
398 # define BISON_LOCATION_HH
399 ...
400 #endif // !BISON_LOCATION_HH
22172d47
AD
401
402 The inclusion guard is now computed from "PREFIX/FILE-NAME", where lower
403 case characters are converted to upper case, and series of
404 non-alphanumerical characters are converted to an underscore.
405
406 With "bison -o lang++/parser.cc", "location.hh" would now include:
407
e29f0771
AD
408 #ifndef YY_LANG_LOCATION_HH
409 # define YY_LANG_LOCATION_HH
410 ...
411 #endif // !YY_LANG_LOCATION_HH
22172d47
AD
412
413*** C++ locations:
936c88d1 414
e4ab1254
AD
415 The position and location constructors (and their initialize methods)
416 accept new arguments for line and column. Several issues in the
417 documentation were fixed.
936c88d1 418
466b4cf2
AD
419** liby is no longer asking for "rpl_fprintf" on some platforms.
420
7e508a2b
AD
421** Changes in the manual:
422
423*** %printer is documented
424
e4ab1254
AD
425 The "%printer" directive, supported since at least Bison 1.50, is finally
426 documented. The "mfcalc" example is extended to demonstrate it.
7e508a2b 427
e4ab1254
AD
428 For consistency with the C skeletons, the C++ parsers now also support
429 "yyoutput" (as an alias to "debug_stream ()").
7e508a2b
AD
430
431*** Several improvements have been made:
466b4cf2 432
e4ab1254
AD
433 The layout for grammar excerpts was changed to a more compact scheme.
434 Named references are motivated. The description of the automaton
435 description file (*.output) is updated to the current format. Incorrect
436 index entries were fixed. Some other errors were fixed.
466b4cf2 437
86b08b49
AD
438** Building bison:
439
440*** Conflicting prototypes with recent/modified Flex.
441
e4ab1254
AD
442 Fixed build problems with the current, unreleased, version of Flex, and
443 some modified versions of 2.5.35, which have modified function prototypes.
466b4cf2 444
8ef26c2a
AD
445*** Warnings during the build procedure have been eliminated.
446
447*** Several portability problems in the test suite have been fixed:
466b4cf2 448
e4ab1254
AD
449 This includes warnings with some compilers, unexpected behavior of tools
450 such as diff, warning messages from the test suite itself, etc.
466b4cf2 451
91aadcc7 452*** The install-pdf target works properly:
8ef26c2a 453
e4ab1254
AD
454 Running "make install-pdf" (or -dvi, -html, -info, and -ps) no longer
455 halts in the middle of its course.
8ef26c2a 456
b2bc8a1b 457* Changes in version 2.5 (2011-05-14):
006faedf 458
eb8c66bb
JD
459** Grammar symbol names can now contain non-initial dashes:
460
461 Consistently with directives (such as %error-verbose) and with
462 %define variables (e.g. push-pull), grammar symbol names may contain
463 dashes in any position except the beginning. This is a GNU
464 extension over POSIX Yacc. Thus, use of this extension is reported
465 by -Wyacc and rejected in Yacc mode (--yacc).
466
3b1977ea 467** Named references:
348f5608
AR
468
469 Historically, Yacc and Bison have supported positional references
470 ($n, $$) to allow access to symbol values from inside of semantic
471 actions code.
472
473 Starting from this version, Bison can also accept named references.
474 When no ambiguity is possible, original symbol names may be used
475 as named references:
476
ac3297d5 477 if_stmt : "if" cond_expr "then" then_stmt ';'
348f5608
AR
478 { $if_stmt = mk_if_stmt($cond_expr, $then_stmt); }
479
480 In the more common case, explicit names may be declared:
481
ac3297d5 482 stmt[res] : "if" expr[cond] "then" stmt[then] "else" stmt[else] ';'
348f5608
AR
483 { $res = mk_if_stmt($cond, $then, $else); }
484
f840c05a 485 Location information is also accessible using @name syntax. When
348f5608
AR
486 accessing symbol names containing dots or dashes, explicit bracketing
487 ($[sym.1]) must be used.
488
f840c05a 489 These features are experimental in this version. More user feedback
348f5608
AR
490 will help to stabilize them.
491
3b1977ea 492** IELR(1) and canonical LR(1):
34a6c2d1
JD
493
494 IELR(1) is a minimal LR(1) parser table generation algorithm. That
495 is, given any context-free grammar, IELR(1) generates parser tables
723206df 496 with the full language-recognition power of canonical LR(1) but with
3b1977ea
JD
497 nearly the same number of parser states as LALR(1). This reduction
498 in parser states is often an order of magnitude. More importantly,
34a6c2d1
JD
499 because canonical LR(1)'s extra parser states may contain duplicate
500 conflicts in the case of non-LR(1) grammars, the number of conflicts
501 for IELR(1) is often an order of magnitude less as well. This can
502 significantly reduce the complexity of developing of a grammar.
503
504 Bison can now generate IELR(1) and canonical LR(1) parser tables in
505 place of its traditional LALR(1) parser tables, which remain the
506 default. You can specify the type of parser tables in the grammar
507 file with these directives:
508
f37495f6
JD
509 %define lr.type lalr
510 %define lr.type ielr
511 %define lr.type canonical-lr
34a6c2d1 512
6f04ee6c 513 The default-reduction optimization in the parser tables can also be
e4ab1254
AD
514 adjusted using "%define lr.default-reductions". For details on both
515 of these features, see the new section "Tuning LR" in the Bison
6f04ee6c 516 manual.
34a6c2d1
JD
517
518 These features are experimental. More user feedback will help to
519 stabilize them.
520
6f04ee6c 521** LAC (Lookahead Correction) for syntax error handling:
4c38b19e
JD
522
523 Canonical LR, IELR, and LALR can suffer from a couple of problems
524 upon encountering a syntax error. First, the parser might perform
525 additional parser stack reductions before discovering the syntax
6f04ee6c 526 error. Such reductions can perform user semantic actions that are
4c38b19e
JD
527 unexpected because they are based on an invalid token, and they
528 cause error recovery to begin in a different syntactic context than
529 the one in which the invalid token was encountered. Second, when
6f04ee6c 530 verbose error messages are enabled (with %error-verbose or the
e4ab1254 531 obsolete "#define YYERROR_VERBOSE"), the expected token list in the
6f04ee6c
JD
532 syntax error message can both contain invalid tokens and omit valid
533 tokens.
4c38b19e
JD
534
535 The culprits for the above problems are %nonassoc, default
536 reductions in inconsistent states, and parser state merging. Thus,
537 IELR and LALR suffer the most. Canonical LR can suffer only if
538 %nonassoc is used or if default reductions are enabled for
539 inconsistent states.
540
6f04ee6c
JD
541 LAC is a new mechanism within the parsing algorithm that solves
542 these problems for canonical LR, IELR, and LALR without sacrificing
543 %nonassoc, default reductions, or state merging. When LAC is in
544 use, canonical LR and IELR behave almost exactly the same for both
545 syntactically acceptable and syntactically unacceptable input.
4c38b19e
JD
546 While LALR still does not support the full language-recognition
547 power of canonical LR and IELR, LAC at least enables LALR's syntax
548 error handling to correctly reflect LALR's language-recognition
549 power.
550
551 Currently, LAC is only supported for deterministic parsers in C.
552 You can enable LAC with the following directive:
553
554 %define parse.lac full
555
e4ab1254 556 See the new section "LAC" in the Bison manual for additional
6f04ee6c 557 details including a few caveats.
4c38b19e
JD
558
559 LAC is an experimental feature. More user feedback will help to
560 stabilize it.
561
6ee881f6 562** %define improvements:
f37495f6 563
3b1977ea 564*** Can now be invoked via the command line:
4ecd3681 565
34d41938 566 Each of these command-line options
4ecd3681 567
34d41938
JD
568 -D NAME[=VALUE]
569 --define=NAME[=VALUE]
570
571 -F NAME[=VALUE]
572 --force-define=NAME[=VALUE]
4ecd3681
JD
573
574 is equivalent to this grammar file declaration
575
34d41938 576 %define NAME ["VALUE"]
4ecd3681 577
34d41938
JD
578 except that the manner in which Bison processes multiple definitions
579 for the same NAME differs. Most importantly, -F and --force-define
580 quietly override %define, but -D and --define do not. For further
e4ab1254 581 details, see the section "Bison Options" in the Bison manual.
4ecd3681 582
3b1977ea 583*** Variables renamed:
812775a0
JD
584
585 The following %define variables
586
587 api.push_pull
588 lr.keep_unreachable_states
589
590 have been renamed to
591
592 api.push-pull
593 lr.keep-unreachable-states
594
595 The old names are now deprecated but will be maintained indefinitely
596 for backward compatibility.
597
723206df 598*** Values no longer need to be quoted in the grammar file:
f37495f6
JD
599
600 If a %define value is an identifier, it no longer needs to be placed
601 within quotations marks. For example,
602
603 %define api.push-pull "push"
604
605 can be rewritten as
606
607 %define api.push-pull push
608
6ee881f6 609*** Unrecognized variables are now errors not warnings.
c046698e 610
6ee881f6
JD
611*** Multiple invocations for any variable is now an error not a warning.
612
613** Unrecognized %code qualifiers are now errors not warnings.
614
615** Character literals not of length one:
616
617 Previously, Bison quietly converted all character literals to length
618 one. For example, without warning, Bison interpreted the operators in
619 the following grammar to be the same token:
620
621 exp: exp '++'
622 | exp '+' exp
623 ;
624
625 Bison now warns when a character literal is not of length one. In
626 some future release, Bison will start reporting an error instead.
627
628** Destructor calls fixed for lookaheads altered in semantic actions:
629
630 Previously for deterministic parsers in C, if a user semantic action
631 altered yychar, the parser in some cases used the old yychar value to
632 determine which destructor to call for the lookahead upon a syntax
633 error or upon parser return. This bug has been fixed.
634
635** C++ parsers use YYRHSLOC:
636
637 Similarly to the C parsers, the C++ parsers now define the YYRHSLOC
638 macro and use it in the default YYLLOC_DEFAULT. You are encouraged
e4ab1254
AD
639 to use it. If, for instance, your location structure has "first"
640 and "last" members, instead of
6ee881f6 641
e29f0771
AD
642 # define YYLLOC_DEFAULT(Current, Rhs, N) \
643 do \
644 if (N) \
645 { \
646 (Current).first = (Rhs)[1].location.first; \
647 (Current).last = (Rhs)[N].location.last; \
648 } \
649 else \
650 { \
651 (Current).first = (Current).last = (Rhs)[0].location.last; \
652 } \
653 while (false)
6ee881f6
JD
654
655 use:
656
e29f0771
AD
657 # define YYLLOC_DEFAULT(Current, Rhs, N) \
658 do \
659 if (N) \
660 { \
661 (Current).first = YYRHSLOC (Rhs, 1).first; \
662 (Current).last = YYRHSLOC (Rhs, N).last; \
663 } \
664 else \
665 { \
666 (Current).first = (Current).last = YYRHSLOC (Rhs, 0).last; \
667 } \
668 while (false)
6ee881f6
JD
669
670** YYLLOC_DEFAULT in C++:
671
672 The default implementation of YYLLOC_DEFAULT used to be issued in
673 the header file. It is now output in the implementation file, after
674 the user %code sections so that its #ifndef guard does not try to
675 override the user's YYLLOC_DEFAULT if provided.
c046698e 676
3b1977ea 677** YYFAIL now produces warnings and Java parsers no longer implement it:
62efdd2a
JD
678
679 YYFAIL has existed for many years as an undocumented feature of
680 deterministic parsers in C generated by Bison. More recently, it was
681 a documented feature of Bison's experimental Java parsers. As
682 promised in Bison 2.4.2's NEWS entry, any appearance of YYFAIL in a
683 semantic action now produces a deprecation warning, and Java parsers
684 no longer implement YYFAIL at all. For further details, including a
685 discussion of how to suppress C preprocessor warnings about YYFAIL
686 being unused, see the Bison 2.4.2 NEWS entry.
687
3b1977ea 688** Temporary hack for adding a semicolon to the user action:
a2d05674
JD
689
690 Previously, Bison appended a semicolon to every user action for
691 reductions when the output language defaulted to C (specifically, when
692 neither %yacc, %language, %skeleton, or equivalent command-line
693 options were specified). This allowed actions such as
694
695 exp: exp "+" exp { $$ = $1 + $3 };
696
697 instead of
698
699 exp: exp "+" exp { $$ = $1 + $3; };
700
701 As a first step in removing this misfeature, Bison now issues a
702 warning when it appends a semicolon. Moreover, in cases where Bison
703 cannot easily determine whether a semicolon is needed (for example, an
704 action ending with a cpp directive or a braced compound initializer),
705 it no longer appends one. Thus, the C compiler might now complain
706 about a missing semicolon where it did not before. Future releases of
707 Bison will cease to append semicolons entirely.
708
095a1d11
JD
709** Verbose syntax error message fixes:
710
e4ab1254 711 When %error-verbose or the obsolete "#define YYERROR_VERBOSE" is
6f04ee6c
JD
712 specified, syntax error messages produced by the generated parser
713 include the unexpected token as well as a list of expected tokens.
714 The effect of %nonassoc on these verbose messages has been corrected
715 in two ways, but a more complete fix requires LAC, described above:
095a1d11
JD
716
717*** When %nonassoc is used, there can exist parser states that accept no
718 tokens, and so the parser does not always require a lookahead token
719 in order to detect a syntax error. Because no unexpected token or
720 expected tokens can then be reported, the verbose syntax error
721 message described above is suppressed, and the parser instead
e4ab1254 722 reports the simpler message, "syntax error". Previously, this
095a1d11
JD
723 suppression was sometimes erroneously triggered by %nonassoc when a
724 lookahead was actually required. Now verbose messages are
725 suppressed only when all previous lookaheads have already been
726 shifted or discarded.
727
728*** Previously, the list of expected tokens erroneously included tokens
729 that would actually induce a syntax error because conflicts for them
730 were resolved with %nonassoc in the current parser state. Such
731 tokens are now properly omitted from the list.
732
733*** Expected token lists are still often wrong due to state merging
4c38b19e
JD
734 (from LALR or IELR) and default reductions, which can both add
735 invalid tokens and subtract valid tokens. Canonical LR almost
736 completely fixes this problem by eliminating state merging and
737 default reductions. However, there is one minor problem left even
738 when using canonical LR and even after the fixes above. That is,
739 if the resolution of a conflict with %nonassoc appears in a later
740 parser state than the one at which some syntax error is
741 discovered, the conflicted token is still erroneously included in
742 the expected token list. Bison's new LAC implementation,
743 described above, eliminates this problem and the need for
744 canonical LR. However, LAC is still experimental and is disabled
745 by default.
1fa30307 746
4c2a6e42
JD
747** Java skeleton fixes:
748
749*** A location handling bug has been fixed.
750
751*** The top element of each of the value stack and location stack is now
752 cleared when popped so that it can be garbage collected.
77768165 753
f0e2c228
JD
754*** Parser traces now print the top element of the stack.
755
248dba53
JD
756** -W/--warnings fixes:
757
e4ab1254 758*** Bison now properly recognizes the "no-" versions of categories:
248dba53
JD
759
760 For example, given the following command line, Bison now enables all
761 warnings except warnings for incompatibilities with POSIX Yacc:
762
763 bison -Wall,no-yacc gram.y
764
6f8bdce2
JD
765*** Bison now treats S/R and R/R conflicts like other warnings:
766
767 Previously, conflict reports were independent of Bison's normal
768 warning system. Now, Bison recognizes the warning categories
e4ab1254 769 "conflicts-sr" and "conflicts-rr". This change has important
6f8bdce2
JD
770 consequences for the -W and --warnings command-line options. For
771 example:
772
773 bison -Wno-conflicts-sr gram.y # S/R conflicts not reported
774 bison -Wno-conflicts-rr gram.y # R/R conflicts not reported
775 bison -Wnone gram.y # no conflicts are reported
776 bison -Werror gram.y # any conflict is an error
777
778 However, as before, if the %expect or %expect-rr directive is
779 specified, an unexpected number of conflicts is an error, and an
780 expected number of conflicts is not reported, so -W and --warning
781 then have no effect on the conflict report.
782
e4ab1254 783*** The "none" category no longer disables a preceding "error":
dab96632
JD
784
785 For example, for the following command line, Bison now reports
786 errors instead of warnings for incompatibilities with POSIX Yacc:
787
788 bison -Werror,none,yacc gram.y
789
e4ab1254 790*** The "none" category now disables all Bison warnings:
8ffd7912 791
e4ab1254 792 Previously, the "none" category disabled only Bison warnings for
8ffd7912
JD
793 which there existed a specific -W/--warning category. However,
794 given the following command line, Bison is now guaranteed to
795 suppress all warnings:
796
797 bison -Wnone gram.y
798
9d6af153
JD
799** Precedence directives can now assign token number 0:
800
801 Since Bison 2.3b, which restored the ability of precedence
802 directives to assign token numbers, doing so for token number 0 has
803 produced an assertion failure. For example:
804
805 %left END 0
806
807 This bug has been fixed.
808
7527c744 809* Changes in version 2.4.3 (2010-08-05):
7a9c3cb3 810
4631c34f
JD
811** Bison now obeys -Werror and --warnings=error for warnings about
812 grammar rules that are useless in the parser due to conflicts.
813
7a9c3cb3
JD
814** Problems with spawning M4 on at least FreeBSD 8 and FreeBSD 9 have
815 been fixed.
816
132247cd
JD
817** Failures in the test suite for GCC 4.5 have been fixed.
818
88bb35d6
JD
819** Failures in the test suite for some versions of Sun Studio C++ have
820 been fixed.
821
43fdc9fd
JD
822** Contrary to Bison 2.4.2's NEWS entry, it has been decided that
823 warnings about undefined %prec identifiers will not be converted to
824 errors in Bison 2.5. They will remain warnings, which should be
825 sufficient for POSIX while avoiding backward compatibility issues.
826
cf22447c
JD
827** Minor documentation fixes.
828
ea66d039 829* Changes in version 2.4.2 (2010-03-20):
8defe11b 830
47fa5747
JD
831** Some portability problems that resulted in failures and livelocks
832 in the test suite on some versions of at least Solaris, AIX, HP-UX,
ea66d039
JD
833 RHEL4, and Tru64 have been addressed. As a result, fatal Bison
834 errors should no longer cause M4 to report a broken pipe on the
47fa5747
JD
835 affected platforms.
836
e4ab1254 837** "%prec IDENTIFIER" requires IDENTIFIER to be defined separately.
2c203528
JD
838
839 POSIX specifies that an error be reported for any identifier that does
840 not appear on the LHS of a grammar rule and that is not defined by
841 %token, %left, %right, or %nonassoc. Bison 2.3b and later lost this
842 error report for the case when an identifier appears only after a
843 %prec directive. It is now restored. However, for backward
844 compatibility with recent Bison releases, it is only a warning for
845 now. In Bison 2.5 and later, it will return to being an error.
43fdc9fd
JD
846 [Between the 2.4.2 and 2.4.3 releases, it was decided that this
847 warning will not be converted to an error in Bison 2.5.]
2c203528 848
c5196098
EB
849** Detection of GNU M4 1.4.6 or newer during configure is improved.
850
34731471
JD
851** Warnings from gcc's -Wundef option about undefined YYENABLE_NLS,
852 YYLTYPE_IS_TRIVIAL, and __STRICT_ANSI__ in C/C++ parsers are now
853 avoided.
af2ffe5c 854
966aba65
JD
855** %code is now a permanent feature.
856
857 A traditional Yacc prologue directive is written in the form:
858
859 %{CODE%}
860
861 To provide a more flexible alternative, Bison 2.3b introduced the
862 %code directive with the following forms for C/C++:
863
864 %code {CODE}
865 %code requires {CODE}
866 %code provides {CODE}
867 %code top {CODE}
868
869 These forms are now considered permanent features of Bison. See the
870 %code entries in the section "Bison Declaration Summary" in the Bison
871 manual for a summary of their functionality. See the section
872 "Prologue Alternatives" for a detailed discussion including the
873 advantages of %code over the traditional Yacc prologue directive.
874
875 Bison's Java feature as a whole including its current usage of %code
876 is still considered experimental.
877
41d35e54
JD
878** YYFAIL is deprecated and will eventually be removed.
879
880 YYFAIL has existed for many years as an undocumented feature of
881 deterministic parsers in C generated by Bison. Previously, it was
882 documented for Bison's experimental Java parsers. YYFAIL is no longer
883 documented for Java parsers and is formally deprecated in both cases.
884 Users are strongly encouraged to migrate to YYERROR, which is
885 specified by POSIX.
886
887 Like YYERROR, you can invoke YYFAIL from a semantic action in order to
888 induce a syntax error. The most obvious difference from YYERROR is
889 that YYFAIL will automatically invoke yyerror to report the syntax
890 error so that you don't have to. However, there are several other
891 subtle differences between YYERROR and YYFAIL, and YYFAIL suffers from
e4ab1254 892 inherent flaws when %error-verbose or "#define YYERROR_VERBOSE" is
41d35e54
JD
893 used. For a more detailed discussion, see:
894
895 http://lists.gnu.org/archive/html/bison-patches/2009-12/msg00024.html
896
897 The upcoming Bison 2.5 will remove YYFAIL from Java parsers, but
898 deterministic parsers in C will continue to implement it. However,
899 because YYFAIL is already flawed, it seems futile to try to make new
900 Bison features compatible with it. Thus, during parser generation,
901 Bison 2.5 will produce a warning whenever it discovers YYFAIL in a
902 rule action. In a later release, YYFAIL will be disabled for
e4ab1254 903 %error-verbose and "#define YYERROR_VERBOSE". Eventually, YYFAIL will
41d35e54
JD
904 be removed altogether.
905
906 There exists at least one case where Bison 2.5's YYFAIL warning will
907 be a false positive. Some projects add phony uses of YYFAIL and other
908 Bison-defined macros for the sole purpose of suppressing C
909 preprocessor warnings (from GCC cpp's -Wunused-macros, for example).
910 To avoid Bison's future warning, such YYFAIL uses can be moved to the
e4ab1254 911 epilogue (that is, after the second "%%") in the Bison input file. In
41d35e54
JD
912 this release (2.4.2), Bison already generates its own code to suppress
913 C preprocessor warnings for YYFAIL, so projects can remove their own
914 phony uses of YYFAIL if compatibility with Bison releases prior to
915 2.4.2 is not necessary.
916
dac8cc0d
AD
917** Internationalization.
918
919 Fix a regression introduced in Bison 2.4: Under some circumstances,
920 message translations were not installed although supported by the
921 host system.
922
41930e7a 923* Changes in version 2.4.1 (2008-12-11):
c9ba9e59 924
a957d06c
JD
925** In the GLR defines file, unexpanded M4 macros in the yylval and yylloc
926 declarations have been fixed.
927
738cde3e
AD
928** Temporary hack for adding a semicolon to the user action.
929
930 Bison used to prepend a trailing semicolon at the end of the user
931 action for reductions. This allowed actions such as
932
933 exp: exp "+" exp { $$ = $1 + $3 };
934
935 instead of
936
937 exp: exp "+" exp { $$ = $1 + $3; };
938
e4ab1254 939 Some grammars still depend on this "feature". Bison 2.4.1 restores
d07932ef
JD
940 the previous behavior in the case of C output (specifically, when
941 neither %language or %skeleton or equivalent command-line options
942 are used) to leave more time for grammars depending on the old
943 behavior to be adjusted. Future releases of Bison will disable this
944 feature.
a957d06c
JD
945
946** A few minor improvements to the Bison manual.
c9ba9e59 947
d6fb461d 948* Changes in version 2.4 (2008-11-02):
7bd1665a 949
d6fb461d 950** %language is an experimental feature.
ed4d67dc
JD
951
952 We first introduced this feature in test release 2.3b as a cleaner
953 alternative to %skeleton. Since then, we have discussed the possibility of
954 modifying its effect on Bison's output file names. Thus, in this release,
955 we consider %language to be an experimental feature that will likely evolve
956 in future releases.
7bd1665a 957
d6fb461d 958** Forward compatibility with GNU M4 has been improved.
241fda7a 959
d6fb461d 960** Several bugs in the C++ skeleton and the experimental Java skeleton have been
241fda7a
JD
961 fixed.
962
d6fb461d 963* Changes in version 2.3b (2008-05-27):
35fe0834 964
d6fb461d 965** The quotes around NAME that used to be required in the following directive
d9df47b6
JD
966 are now deprecated:
967
968 %define NAME "VALUE"
969
e4ab1254 970** The directive "%pure-parser" is now deprecated in favor of:
d9df47b6
JD
971
972 %define api.pure
973
974 which has the same effect except that Bison is more careful to warn about
975 unreasonable usage in the latter case.
976
d6fb461d 977** Push Parsing
c373bf8b
JD
978
979 Bison can now generate an LALR(1) parser in C with a push interface. That
e4ab1254
AD
980 is, instead of invoking "yyparse", which pulls tokens from "yylex", you can
981 push one token at a time to the parser using "yypush_parse", which will
c373bf8b
JD
982 return to the caller after processing each token. By default, the push
983 interface is disabled. Either of the following directives will enable it:
984
985 %define api.push_pull "push" // Just push; does not require yylex.
986 %define api.push_pull "both" // Push and pull; requires yylex.
987
e4ab1254 988 See the new section "A Push Parser" in the Bison manual for details.
c373bf8b 989
59da312b
JD
990 The current push parsing interface is experimental and may evolve. More user
991 feedback will help to stabilize it.
992
d6fb461d 993** The -g and --graph options now output graphs in Graphviz DOT format,
8e55b3aa
JD
994 not VCG format. Like --graph, -g now also takes an optional FILE argument
995 and thus cannot be bundled with other short options.
c373bf8b 996
d6fb461d 997** Java
59da312b
JD
998
999 Bison can now generate an LALR(1) parser in Java. The skeleton is
e4ab1254 1000 "data/lalr1.java". Consider using the new %language directive instead of
59da312b
JD
1001 %skeleton to select it.
1002
e4ab1254 1003 See the new section "Java Parsers" in the Bison manual for details.
59da312b
JD
1004
1005 The current Java interface is experimental and may evolve. More user
1006 feedback will help to stabilize it.
1007
d6fb461d 1008** %language
59da312b
JD
1009
1010 This new directive specifies the programming language of the generated
d43f77e7
PB
1011 parser, which can be C (the default), C++, or Java. Besides the skeleton
1012 that Bison uses, the directive affects the names of the generated files if
1013 the grammar file's name ends in ".y".
59da312b 1014
d6fb461d 1015** XML Automaton Report
59da312b
JD
1016
1017 Bison can now generate an XML report of the LALR(1) automaton using the new
e4ab1254 1018 "--xml" option. The current XML schema is experimental and may evolve. More
59da312b 1019 user feedback will help to stabilize it.
c373bf8b 1020
d6fb461d 1021** The grammar file may now specify the name of the parser header file using
c373bf8b
JD
1022 %defines. For example:
1023
1024 %defines "parser.h"
1025
d6fb461d 1026** When reporting useless rules, useless nonterminals, and unused terminals,
d80fb37a
JD
1027 Bison now employs the terms "useless in grammar" instead of "useless",
1028 "useless in parser" instead of "never reduced", and "unused in grammar"
1029 instead of "unused".
cff03fb2 1030
d6fb461d 1031** Unreachable State Removal
c373bf8b
JD
1032
1033 Previously, Bison sometimes generated parser tables containing unreachable
31984206
JD
1034 states. A state can become unreachable during conflict resolution if Bison
1035 disables a shift action leading to it from a predecessor state. Bison now:
75ad86ee
JD
1036
1037 1. Removes unreachable states.
1038
1039 2. Does not report any conflicts that appeared in unreachable states.
1040 WARNING: As a result, you may need to update %expect and %expect-rr
1041 directives in existing grammar files.
1042
1043 3. For any rule used only in such states, Bison now reports the rule as
cff03fb2 1044 "useless in parser due to conflicts".
75ad86ee 1045
31984206
JD
1046 This feature can be disabled with the following directive:
1047
1048 %define lr.keep_unreachable_states
1049
e4ab1254 1050 See the %define entry in the "Bison Declaration Summary" in the Bison manual
31984206
JD
1051 for further discussion.
1052
e4ab1254 1053** Lookahead Set Correction in the ".output" Report
b1cc23c4 1054
e4ab1254
AD
1055 When instructed to generate a ".output" file including lookahead sets
1056 (using "--report=lookahead", for example), Bison now prints each reduction's
88c78747
JD
1057 lookahead set only next to the associated state's one item that (1) is
1058 associated with the same rule as the reduction and (2) has its dot at the end
1059 of its RHS. Previously, Bison also erroneously printed the lookahead set
1060 next to all of the state's other items associated with the same rule. This
e4ab1254 1061 bug affected only the ".output" file and not the generated parser source
88c78747
JD
1062 code.
1063
e4ab1254 1064** --report-file=FILE is a new option to override the default ".output" file
59da312b 1065 name.
1bb2bd75 1066
e4ab1254 1067** The "=" that used to be required in the following directives is now
02975b9a
JD
1068 deprecated:
1069
1070 %file-prefix "parser"
1071 %name-prefix "c_"
1072 %output "parser.c"
1073
e4ab1254 1074** An Alternative to "%{...%}" -- "%code QUALIFIER {CODE}"
c373bf8b
JD
1075
1076 Bison 2.3a provided a new set of directives as a more flexible alternative to
8e0a5e9e
JD
1077 the traditional Yacc prologue blocks. Those have now been consolidated into
1078 a single %code directive with an optional qualifier field, which identifies
1079 the purpose of the code and thus the location(s) where Bison should generate
1080 it:
1081
e4ab1254
AD
1082 1. "%code {CODE}" replaces "%after-header {CODE}"
1083 2. "%code requires {CODE}" replaces "%start-header {CODE}"
1084 3. "%code provides {CODE}" replaces "%end-header {CODE}"
1085 4. "%code top {CODE}" replaces "%before-header {CODE}"
8e0a5e9e 1086
e4ab1254
AD
1087 See the %code entries in section "Bison Declaration Summary" in the Bison
1088 manual for a summary of the new functionality. See the new section "Prologue
1089 Alternatives" for a detailed discussion including the advantages of %code
8e0a5e9e
JD
1090 over the traditional Yacc prologues.
1091
1092 The prologue alternatives are experimental. More user feedback will help to
1093 determine whether they should become permanent features.
1094
d6fb461d 1095** Revised warning: unset or unused mid-rule values
17bd8a73
JD
1096
1097 Since Bison 2.2, Bison has warned about mid-rule values that are set but not
1098 used within any of the actions of the parent rule. For example, Bison warns
1099 about unused $2 in:
1100
1101 exp: '1' { $$ = 1; } '+' exp { $$ = $1 + $4; };
1102
1103 Now, Bison also warns about mid-rule values that are used but not set. For
1104 example, Bison warns about unset $$ in the mid-rule action in:
1105
1106 exp: '1' { $1 = 1; } '+' exp { $$ = $2 + $4; };
1107
1108 However, Bison now disables both of these warnings by default since they
1109 sometimes prove to be false alarms in existing grammars employing the Yacc
1110 constructs $0 or $-N (where N is some positive integer).
1111
e4ab1254
AD
1112 To enable these warnings, specify the option "--warnings=midrule-values" or
1113 "-W", which is a synonym for "--warnings=all".
17bd8a73 1114
e4ab1254 1115** Default %destructor or %printer with "<*>" or "<>"
c373bf8b
JD
1116
1117 Bison now recognizes two separate kinds of default %destructor's and
12e35840
JD
1118 %printer's:
1119
e4ab1254 1120 1. Place "<*>" in a %destructor/%printer symbol list to define a default
12e35840
JD
1121 %destructor/%printer for all grammar symbols for which you have formally
1122 declared semantic type tags.
1123
e4ab1254 1124 2. Place "<>" in a %destructor/%printer symbol list to define a default
12e35840
JD
1125 %destructor/%printer for all grammar symbols without declared semantic
1126 type tags.
1127
e4ab1254
AD
1128 Bison no longer supports the "%symbol-default" notation from Bison 2.3a.
1129 "<*>" and "<>" combined achieve the same effect with one exception: Bison no
12e35840
JD
1130 longer applies any %destructor to a mid-rule value if that mid-rule value is
1131 not actually ever referenced using either $$ or $n in a semantic action.
1132
85894313
JD
1133 The default %destructor's and %printer's are experimental. More user
1134 feedback will help to determine whether they should become permanent
1135 features.
1136
e4ab1254 1137 See the section "Freeing Discarded Symbols" in the Bison manual for further
12e35840
JD
1138 details.
1139
d6fb461d 1140** %left, %right, and %nonassoc can now declare token numbers. This is required
e4ab1254 1141 by POSIX. However, see the end of section "Operator Precedence" in the Bison
ab7f29f8
JD
1142 manual for a caveat concerning the treatment of literal strings.
1143
d6fb461d 1144** The nonfunctional --no-parser, -n, and %no-parser options have been
b1cc23c4
JD
1145 completely removed from Bison.
1146
d6fb461d 1147* Changes in version 2.3a, 2006-09-13:
742e4900 1148
d6fb461d 1149** Instead of %union, you can define and use your own union type
ddc8ede1
PE
1150 YYSTYPE if your grammar contains at least one <type> tag.
1151 Your YYSTYPE need not be a macro; it can be a typedef.
1152 This change is for compatibility with other Yacc implementations,
1153 and is required by POSIX.
1154
d6fb461d 1155** Locations columns and lines start at 1.
cd48d21d
AD
1156 In accordance with the GNU Coding Standards and Emacs.
1157
d6fb461d 1158** You may now declare per-type and default %destructor's and %printer's:
ec5479ce
JD
1159
1160 For example:
1161
b2a0b7ca
JD
1162 %union { char *string; }
1163 %token <string> STRING1
1164 %token <string> STRING2
1165 %type <string> string1
1166 %type <string> string2
1167 %union { char character; }
1168 %token <character> CHR
1169 %type <character> chr
1170 %destructor { free ($$); } %symbol-default
1171 %destructor { free ($$); printf ("%d", @$.first_line); } STRING1 string1
1172 %destructor { } <character>
1173
1174 guarantees that, when the parser discards any user-defined symbol that has a
e4ab1254
AD
1175 semantic type tag other than "<character>", it passes its semantic value to
1176 "free". However, when the parser discards a "STRING1" or a "string1", it
1177 also prints its line number to "stdout". It performs only the second
1178 "%destructor" in this case, so it invokes "free" only once.
ec5479ce 1179
85894313
JD
1180 [Although we failed to mention this here in the 2.3a release, the default
1181 %destructor's and %printer's were experimental, and they were rewritten in
1182 future versions.]
1183
e4ab1254
AD
1184** Except for LALR(1) parsers in C with POSIX Yacc emulation enabled (with "-y",
1185 "--yacc", or "%yacc"), Bison no longer generates #define statements for
b931235e
JD
1186 associating token numbers with token names. Removing the #define statements
1187 helps to sanitize the global namespace during preprocessing, but POSIX Yacc
1188 requires them. Bison still generates an enum for token names in all cases.
1189
d6fb461d 1190** Handling of traditional Yacc prologue blocks is now more consistent but
34f98f46 1191 potentially incompatible with previous releases of Bison.
9bc0dd67
JD
1192
1193 As before, you declare prologue blocks in your grammar file with the
e4ab1254 1194 "%{ ... %}" syntax. To generate the pre-prologue, Bison concatenates all
34f98f46
JD
1195 prologue blocks that you've declared before the first %union. To generate
1196 the post-prologue, Bison concatenates all prologue blocks that you've
ddc8ede1 1197 declared after the first %union.
9bc0dd67 1198
34f98f46 1199 Previous releases of Bison inserted the pre-prologue into both the header
9bc0dd67
JD
1200 file and the code file in all cases except for LALR(1) parsers in C. In the
1201 latter case, Bison inserted it only into the code file. For parsers in C++,
1202 the point of insertion was before any token definitions (which associate
1203 token numbers with names). For parsers in C, the point of insertion was
1204 after the token definitions.
1205
1206 Now, Bison never inserts the pre-prologue into the header file. In the code
1207 file, it always inserts it before the token definitions.
1208
d6fb461d 1209** Bison now provides a more flexible alternative to the traditional Yacc
34f98f46
JD
1210 prologue blocks: %before-header, %start-header, %end-header, and
1211 %after-header.
1212
1213 For example, the following declaration order in the grammar file reflects the
1214 order in which Bison will output these code blocks. However, you are free to
1215 declare these code blocks in your grammar file in whatever order is most
1216 convenient for you:
1217
1218 %before-header {
1219 /* Bison treats this block like a pre-prologue block: it inserts it into
1220 * the code file before the contents of the header file. It does *not*
1221 * insert it into the header file. This is a good place to put
1222 * #include's that you want at the top of your code file. A common
e4ab1254 1223 * example is '#include "system.h"'. */
34f98f46
JD
1224 }
1225 %start-header {
1226 /* Bison inserts this block into both the header file and the code file.
1227 * In both files, the point of insertion is before any Bison-generated
1228 * token, semantic type, location type, and class definitions. This is a
1229 * good place to define %union dependencies, for example. */
9bc0dd67
JD
1230 }
1231 %union {
34f98f46
JD
1232 /* Unlike the traditional Yacc prologue blocks, the output order for the
1233 * new %*-header blocks is not affected by their declaration position
1234 * relative to any %union in the grammar file. */
9bc0dd67 1235 }
34f98f46
JD
1236 %end-header {
1237 /* Bison inserts this block into both the header file and the code file.
1238 * In both files, the point of insertion is after the Bison-generated
1239 * definitions. This is a good place to declare or define public
1240 * functions or data structures that depend on the Bison-generated
1241 * definitions. */
9bc0dd67 1242 }
34f98f46
JD
1243 %after-header {
1244 /* Bison treats this block like a post-prologue block: it inserts it into
1245 * the code file after the contents of the header file. It does *not*
1246 * insert it into the header file. This is a good place to declare or
1247 * define internal functions or data structures that depend on the
1248 * Bison-generated definitions. */
1249 }
1250
1251 If you have multiple occurrences of any one of the above declarations, Bison
1252 will concatenate the contents in declaration order.
9bc0dd67 1253
85894313
JD
1254 [Although we failed to mention this here in the 2.3a release, the prologue
1255 alternatives were experimental, and they were rewritten in future versions.]
1256
e4ab1254 1257** The option "--report=look-ahead" has been changed to "--report=lookahead".
9e6e7ed2
PE
1258 The old spelling still works, but is not documented and may be removed
1259 in a future release.
742e4900 1260
d6fb461d 1261* Changes in version 2.3, 2006-06-05:
4ad3ed84 1262
e4ab1254 1263** GLR grammars should now use "YYRECOVERING ()" instead of "YYRECOVERING",
4ad3ed84
PE
1264 for compatibility with LALR(1) grammars.
1265
d6fb461d 1266** It is now documented that any definition of YYSTYPE or YYLTYPE should
4ad3ed84
PE
1267 be to a type name that does not contain parentheses or brackets.
1268
d6fb461d 1269* Changes in version 2.2, 2006-05-19:
193d7c70 1270
d6fb461d 1271** The distribution terms for all Bison-generated parsers now permit
193d7c70
PE
1272 using the parsers in nonfree programs. Previously, this permission
1273 was granted only for Bison-generated LALR(1) parsers in C.
5f4236a0 1274
d6fb461d 1275** %name-prefix changes the namespace name in C++ outputs.
aa08666d 1276
d6fb461d 1277** The C++ parsers export their token_type.
5f4236a0 1278
d6fb461d 1279** Bison now allows multiple %union declarations, and concatenates
d6ca7905
PE
1280 their contents together.
1281
d6fb461d 1282** New warning: unused values
4d7bc38c
PE
1283 Right-hand side symbols whose values are not used are reported,
1284 if the symbols have destructors. For instance:
affac613 1285
8f3596a6 1286 exp: exp "?" exp ":" exp { $1 ? $1 : $3; }
e29f0771
AD
1287 | exp "+" exp
1288 ;
affac613 1289
8f3596a6
AD
1290 will trigger a warning about $$ and $5 in the first rule, and $3 in
1291 the second ($1 is copied to $$ by the default rule). This example
4e26c69e 1292 most likely contains three errors, and could be rewritten as:
affac613 1293
4e26c69e 1294 exp: exp "?" exp ":" exp
e29f0771
AD
1295 { $$ = $1 ? $3 : $5; free ($1 ? $5 : $3); free ($1); }
1296 | exp "+" exp
1297 { $$ = $1 ? $1 : $3; if ($1) free ($3); }
1298 ;
affac613 1299
4e26c69e
PE
1300 However, if the original actions were really intended, memory leaks
1301 and all, the warnings can be suppressed by letting Bison believe the
1302 values are used, e.g.:
721be13c 1303
8f3596a6 1304 exp: exp "?" exp ":" exp { $1 ? $1 : $3; (void) ($$, $5); }
e29f0771
AD
1305 | exp "+" exp { $$ = $1; (void) $3; }
1306 ;
721be13c 1307
84866159
AD
1308 If there are mid-rule actions, the warning is issued if no action
1309 uses it. The following triggers no warning: $1 and $3 are used.
1310
1311 exp: exp { push ($1); } '+' exp { push ($3); sum (); };
1312
721be13c
PE
1313 The warning is intended to help catching lost values and memory leaks.
1314 If a value is ignored, its associated memory typically is not reclaimed.
affac613 1315
d6fb461d 1316** %destructor vs. YYABORT, YYACCEPT, and YYERROR.
9d9b8b70
PE
1317 Destructors are now called when user code invokes YYABORT, YYACCEPT,
1318 and YYERROR, for all objects on the stack, other than objects
1319 corresponding to the right-hand side of the current rule.
a85284cf 1320
d6fb461d 1321** %expect, %expect-rr
035aa4a0
PE
1322 Incorrect numbers of expected conflicts are now actual errors,
1323 instead of warnings.
1324
d6fb461d 1325** GLR, YACC parsers.
4e26c69e
PE
1326 The %parse-params are available in the destructors (and the
1327 experimental printers) as per the documentation.
4b367315 1328
e4ab1254 1329** Bison now warns if it finds a stray "$" or "@" in an action.
ad6a9b97 1330
d6fb461d 1331** %require "VERSION"
4e26c69e
PE
1332 This specifies that the grammar file depends on features implemented
1333 in Bison version VERSION or higher.
b50d2359 1334
d6fb461d 1335** lalr1.cc: The token and value types are now class members.
e14d0ab6
AD
1336 The tokens were defined as free form enums and cpp macros. YYSTYPE
1337 was defined as a free form union. They are now class members:
e4ab1254
AD
1338 tokens are enumerations of the "yy::parser::token" struct, and the
1339 semantic values have the "yy::parser::semantic_type" type.
fb9712a9
AD
1340
1341 If you do not want or can update to this scheme, the directive
e4ab1254 1342 '%define "global_tokens_and_yystype" "1"' triggers the global
b50d2359
AD
1343 definition of tokens and YYSTYPE. This change is suitable both
1344 for previous releases of Bison, and this one.
fb9712a9 1345
b50d2359 1346 If you wish to update, then make sure older version of Bison will
e4ab1254 1347 fail using '%require "2.2"'.
fb9712a9 1348
d6fb461d 1349** DJGPP support added.
193d7c70 1350\f
d6fb461d 1351* Changes in version 2.1, 2005-09-16:
1ce59070 1352
d6fb461d 1353** The C++ lalr1.cc skeleton supports %lex-param.
e14d0ab6 1354
d6fb461d 1355** Bison-generated parsers now support the translation of diagnostics like
baf785db
PE
1356 "syntax error" into languages other than English. The default
1357 language is still English. For details, please see the new
0410a6e0
PE
1358 Internationalization section of the Bison manual. Software
1359 distributors should also see the new PACKAGING file. Thanks to
1360 Bruno Haible for this new feature.
1ce59070 1361
d6fb461d 1362** Wording in the Bison-generated parsers has been changed slightly to
1a059451
PE
1363 simplify translation. In particular, the message "memory exhausted"
1364 has replaced "parser stack overflow", as the old message was not
1365 always accurate for modern Bison-generated parsers.
1366
d6fb461d 1367** Destructors are now called when the parser aborts, for all symbols left
258b75ca
PE
1368 behind on the stack. Also, the start symbol is now destroyed after a
1369 successful parse. In both cases, the behavior was formerly inconsistent.
1370
d6fb461d 1371** When generating verbose diagnostics, Bison-generated parsers no longer
72f000b0
PE
1372 quote the literal strings associated with tokens. For example, for
1373 a syntax error associated with '%token NUM "number"' they might
1374 print 'syntax error, unexpected number' instead of 'syntax error,
1375 unexpected "number"'.
193d7c70 1376\f
d6fb461d 1377* Changes in version 2.0, 2004-12-25:
efeed023 1378
d6fb461d 1379** Possibly-incompatible changes
d7e14fc0 1380
82de6b0d
PE
1381 - Bison-generated parsers no longer default to using the alloca function
1382 (when available) to extend the parser stack, due to widespread
1383 problems in unchecked stack-overflow detection. You can "#define
1384 YYSTACK_USE_ALLOCA 1" to require the use of alloca, but please read
1385 the manual to determine safe values for YYMAXDEPTH in that case.
8dd162d3 1386
82de6b0d
PE
1387 - Error token location.
1388 During error recovery, the location of the syntax error is updated
1389 to cover the whole sequence covered by the error token: it includes
1390 the shifted symbols thrown away during the first part of the error
1391 recovery, and the lookahead rejected during the second part.
18d192f0 1392
82de6b0d
PE
1393 - Semicolon changes:
1394 . Stray semicolons are no longer allowed at the start of a grammar.
1395 . Semicolons are now required after in-grammar declarations.
e342c3be 1396
82de6b0d
PE
1397 - Unescaped newlines are no longer allowed in character constants or
1398 string literals. They were never portable, and GCC 3.4.0 has
1399 dropped support for them. Better diagnostics are now generated if
1400 forget a closing quote.
8dd162d3 1401
82de6b0d 1402 - NUL bytes are no longer allowed in Bison string literals, unfortunately.
f74b6f91 1403
d6fb461d 1404** New features
1452af69 1405
82de6b0d 1406 - GLR grammars now support locations.
4febdd96 1407
82de6b0d
PE
1408 - New directive: %initial-action.
1409 This directive allows the user to run arbitrary code (including
1410 initializing @$) from yyparse before parsing starts.
1452af69 1411
82de6b0d
PE
1412 - A new directive "%expect-rr N" specifies the expected number of
1413 reduce/reduce conflicts in GLR parsers.
1452af69 1414
e4ab1254 1415 - %token numbers can now be hexadecimal integers, e.g., "%token FOO 0x12d".
82de6b0d 1416 This is a GNU extension.
4febdd96 1417
e4ab1254 1418 - The option "--report=lookahead" was changed to "--report=look-ahead".
9e6e7ed2 1419 [However, this was changed back after 2.3.]
1452af69 1420
82de6b0d 1421 - Experimental %destructor support has been added to lalr1.cc.
1452af69 1422
82de6b0d
PE
1423 - New configure option --disable-yacc, to disable installation of the
1424 yacc command and -ly library introduced in 1.875 for POSIX conformance.
6040d338 1425
d6fb461d 1426** Bug fixes
d5a3fe37 1427
82de6b0d
PE
1428 - For now, %expect-count violations are now just warnings, not errors.
1429 This is for compatibility with Bison 1.75 and earlier (when there are
1430 reduce/reduce conflicts) and with Bison 1.30 and earlier (when there
1431 are too many or too few shift/reduce conflicts). However, in future
1432 versions of Bison we plan to improve the %expect machinery so that
1433 these violations will become errors again.
3473d0f8 1434
82de6b0d
PE
1435 - Within Bison itself, numbers (e.g., goto numbers) are no longer
1436 arbitrarily limited to 16-bit counts.
d600ee67 1437
82de6b0d 1438 - Semicolons are now allowed before "|" in grammar rules, as POSIX requires.
d600ee67 1439\f
d6fb461d 1440* Changes in version 1.875, 2003-01-01:
963fcc17 1441
d6fb461d 1442** The documentation license has been upgraded to version 1.2
dc546b0f 1443 of the GNU Free Documentation License.
75eb3bc4 1444
d6fb461d 1445** syntax error processing
75eb3bc4 1446
dc546b0f
PE
1447 - In Yacc-style parsers YYLLOC_DEFAULT is now used to compute error
1448 locations too. This fixes bugs in error-location computation.
75eb3bc4 1449
dc546b0f
PE
1450 - %destructor
1451 It is now possible to reclaim the memory associated to symbols
1452 discarded during error recovery. This feature is still experimental.
20daca06 1453
dc546b0f
PE
1454 - %error-verbose
1455 This new directive is preferred over YYERROR_VERBOSE.
74724a70 1456
dc546b0f
PE
1457 - #defining yyerror to steal internal variables is discouraged.
1458 It is not guaranteed to work forever.
d1de5372 1459
d6fb461d 1460** POSIX conformance
d1de5372 1461
dc546b0f
PE
1462 - Semicolons are once again optional at the end of grammar rules.
1463 This reverts to the behavior of Bison 1.33 and earlier, and improves
1464 compatibility with Yacc.
74724a70 1465
e4ab1254
AD
1466 - "parse error" -> "syntax error"
1467 Bison now uniformly uses the term "syntax error"; formerly, the code
1468 and manual sometimes used the term "parse error" instead. POSIX
1469 requires "syntax error" in diagnostics, and it was thought better to
dc546b0f 1470 be consistent.
74724a70 1471
dc546b0f
PE
1472 - The documentation now emphasizes that yylex and yyerror must be
1473 declared before use. C99 requires this.
d1de5372 1474
dc546b0f
PE
1475 - Bison now parses C99 lexical constructs like UCNs and
1476 backslash-newline within C escape sequences, as POSIX 1003.1-2001 requires.
d1de5372 1477
dc546b0f
PE
1478 - File names are properly escaped in C output. E.g., foo\bar.y is
1479 output as "foo\\bar.y".
6780ca7a 1480
dc546b0f 1481 - Yacc command and library now available
e4ab1254 1482 The Bison distribution now installs a "yacc" command, as POSIX requires.
dc546b0f
PE
1483 Also, Bison now installs a small library liby.a containing
1484 implementations of Yacc-compatible yyerror and main functions.
1485 This library is normally not useful, but POSIX requires it.
6e649e65 1486
dc546b0f 1487 - Type clashes now generate warnings, not errors.
6e649e65 1488
dc546b0f
PE
1489 - If the user does not define YYSTYPE as a macro, Bison now declares it
1490 using typedef instead of defining it as a macro.
1491 For consistency, YYLTYPE is also declared instead of defined.
9501dc6e 1492
d6fb461d 1493** Other compatibility issues
886a425c 1494
e4ab1254
AD
1495 - %union directives can now have a tag before the "{", e.g., the
1496 directive "%union foo {...}" now generates the C code
1497 "typedef union foo { ... } YYSTYPE;"; this is for Yacc compatibility.
1498 The default union tag is "YYSTYPE", for compatibility with Solaris 9 Yacc.
1499 For consistency, YYLTYPE's struct tag is now "YYLTYPE" not "yyltype".
dc546b0f 1500 This is for compatibility with both Yacc and Bison 1.35.
72f889cc 1501
e4ab1254 1502 - ";" is output before the terminating "}" of an action, for
dc546b0f 1503 compatibility with Bison 1.35.
886a425c 1504
dc546b0f 1505 - Bison now uses a Yacc-style format for conflict reports, e.g.,
e4ab1254 1506 "conflicts: 2 shift/reduce, 1 reduce/reduce".
437c2d80 1507
e4ab1254 1508 - "yystype" and "yyltype" are now obsolescent macros instead of being
dc546b0f
PE
1509 typedefs or tags; they are no longer documented and are planned to be
1510 withdrawn in a future release.
2a8d363a 1511
d6fb461d 1512** GLR parser notes
2a8d363a 1513
dc546b0f
PE
1514 - GLR and inline
1515 Users of Bison have to decide how they handle the portability of the
e4ab1254 1516 C keyword "inline".
959e5f51 1517
e4ab1254
AD
1518 - "parsing stack overflow..." -> "parser stack overflow"
1519 GLR parsers now report "parser stack overflow" as per the Bison manual.
900c5db5 1520
18ad57b3
AD
1521** %parse-param and %lex-param
1522 The macros YYPARSE_PARAM and YYLEX_PARAM provide a means to pass
1523 additional context to yyparse and yylex. They suffer from several
1524 shortcomings:
1525
1526 - a single argument only can be added,
1527 - their types are weak (void *),
242cc08e 1528 - this context is not passed to ancillary functions such as yyerror,
18ad57b3
AD
1529 - only yacc.c parsers support them.
1530
1531 The new %parse-param/%lex-param directives provide a more precise control.
1532 For instance:
1533
1534 %parse-param {int *nastiness}
1535 %lex-param {int *nastiness}
1536 %parse-param {int *randomness}
1537
1538 results in the following signatures:
1539
1540 int yylex (int *nastiness);
1541 int yyparse (int *nastiness, int *randomness);
1542
1543 or, if both %pure-parser and %locations are used:
1544
1545 int yylex (YYSTYPE *lvalp, YYLTYPE *llocp, int *nastiness);
1546 int yyparse (int *nastiness, int *randomness);
1547
d6fb461d 1548** Bison now warns if it detects conflicting outputs to the same file,
e4ab1254 1549 e.g., it generates a warning for "bison -d -o foo.h foo.y" since
dc546b0f 1550 that command outputs both code and header to foo.h.
6e40b4eb 1551
d6fb461d 1552** #line in output files
dc546b0f 1553 - --no-line works properly.
6e40b4eb 1554
d6fb461d 1555** Bison can no longer be built by a K&R C compiler; it requires C89 or
6e40b4eb
AD
1556 later to be built. This change originally took place a few versions
1557 ago, but nobody noticed until we recently asked someone to try
1558 building Bison with a K&R C compiler.
d600ee67 1559\f
d6fb461d 1560* Changes in version 1.75, 2002-10-14:
7933f2b5 1561
d6fb461d 1562** Bison should now work on 64-bit hosts.
7933f2b5 1563
d6fb461d 1564** Indonesian translation thanks to Tedi Heriyanto.
7933f2b5 1565
d6fb461d 1566** GLR parsers
f50adbbd
AD
1567 Fix spurious parse errors.
1568
d6fb461d 1569** Pure parsers
f50adbbd
AD
1570 Some people redefine yyerror to steal yyparse' private variables.
1571 Reenable this trick until an official feature replaces it.
1572
d6fb461d 1573** Type Clashes
d90c934c
AD
1574 In agreement with POSIX and with other Yaccs, leaving a default
1575 action is valid when $$ is untyped, and $1 typed:
1576
e29f0771 1577 untyped: ... typed;
d90c934c
AD
1578
1579 but the converse remains an error:
1580
e29f0771 1581 typed: ... untyped;
d90c934c 1582
d6fb461d 1583** Values of mid-rule actions
d90c934c
AD
1584 The following code:
1585
e29f0771 1586 foo: { ... } { $$ = $1; } ...
d90c934c
AD
1587
1588 was incorrectly rejected: $1 is defined in the second mid-rule
1589 action, and is equal to the $$ of the first mid-rule action.
d600ee67 1590\f
d6fb461d 1591* Changes in version 1.50, 2002-10-04:
adc8c848 1592
d6fb461d 1593** GLR parsing
676385e2
PH
1594 The declaration
1595 %glr-parser
1596 causes Bison to produce a Generalized LR (GLR) parser, capable of handling
1597 almost any context-free grammar, ambiguous or not. The new declarations
e8832397 1598 %dprec and %merge on grammar rules allow parse-time resolution of
676385e2
PH
1599 ambiguities. Contributed by Paul Hilfinger.
1600
7933f2b5 1601 Unfortunately Bison 1.50 does not work properly on 64-bit hosts
420f93c8
PE
1602 like the Alpha, so please stick to 32-bit hosts for now.
1603
d6fb461d 1604** Output Directory
8c165d89 1605 When not in Yacc compatibility mode, when the output file was not
e4ab1254
AD
1606 specified, running "bison foo/bar.y" created "foo/bar.c". It
1607 now creates "bar.c".
8c165d89 1608
d6fb461d 1609** Undefined token
007a50a4 1610 The undefined token was systematically mapped to 2 which prevented
e88dbdbf 1611 the use of 2 by the user. This is no longer the case.
007a50a4 1612
d6fb461d 1613** Unknown token numbers
e88dbdbf 1614 If yylex returned an out of range value, yyparse could die. This is
007a50a4
AD
1615 no longer the case.
1616
d6fb461d 1617** Error token
e88dbdbf 1618 According to POSIX, the error token must be 256.
23c5a174
AD
1619 Bison extends this requirement by making it a preference: *if* the
1620 user specified that one of her tokens is numbered 256, then error
1621 will be mapped onto another number.
1622
d6fb461d 1623** Verbose error messages
e4ab1254 1624 They no longer report "..., expecting error or..." for states where
217598da
AD
1625 error recovery is possible.
1626
d6fb461d 1627** End token
e4ab1254 1628 Defaults to "$end" instead of "$".
217598da 1629
d6fb461d 1630** Error recovery now conforms to documentation and to POSIX
68cd8af3
PE
1631 When a Bison-generated parser encounters a syntax error, it now pops
1632 the stack until it finds a state that allows shifting the error
1633 token. Formerly, it popped the stack until it found a state that
1634 allowed some non-error action other than a default reduction on the
1635 error token. The new behavior has long been the documented behavior,
1636 and has long been required by POSIX. For more details, please see
337116ba
PE
1637 Paul Eggert, "Reductions during Bison error handling" (2002-05-20)
1638 <http://lists.gnu.org/archive/html/bug-bison/2002-05/msg00038.html>.
68cd8af3 1639
d6fb461d 1640** Traces
5504898e
AD
1641 Popped tokens and nonterminals are now reported.
1642
d6fb461d 1643** Larger grammars
a861a339
PE
1644 Larger grammars are now supported (larger token numbers, larger grammar
1645 size (= sum of the LHS and RHS lengths), larger LALR tables).
1646 Formerly, many of these numbers ran afoul of 16-bit limits;
1647 now these limits are 32 bits on most hosts.
355e7c1c 1648
d6fb461d 1649** Explicit initial rule
643a5994
AD
1650 Bison used to play hacks with the initial rule, which the user does
1651 not write. It is now explicit, and visible in the reports and
1652 graphs as rule 0.
23c5a174 1653
d6fb461d 1654** Useless rules
643a5994 1655 Before, Bison reported the useless rules, but, although not used,
77714df2 1656 included them in the parsers. They are now actually removed.
23c5a174 1657
d6fb461d 1658** Useless rules, useless nonterminals
6b98e4b5
AD
1659 They are now reported, as a warning, with their locations.
1660
d6fb461d 1661** Rules never reduced
e8832397
AD
1662 Rules that can never be reduced because of conflicts are now
1663 reported.
1664
e4ab1254 1665** Incorrect "Token not used"
11652ab3
AD
1666 On a grammar such as
1667
e29f0771
AD
1668 %token useless useful
1669 %%
1670 exp: '0' %prec useful;
11652ab3
AD
1671
1672 where a token was used to set the precedence of the last rule,
e4ab1254 1673 bison reported both "useful" and "useless" as useless tokens.
11652ab3 1674
d6fb461d 1675** Revert the C++ namespace changes introduced in 1.31
77714df2 1676 as they caused too many portability hassles.
0179dd65 1677
d6fb461d 1678** Default locations
b2d52318
AD
1679 By an accident of design, the default computation of @$ was
1680 performed after another default computation was performed: @$ = @1.
1681 The latter is now removed: YYLLOC_DEFAULT is fully responsible of
1682 the computation of @$.
adc8c848 1683
d6fb461d 1684** Token end-of-file
b7c49edf
AD
1685 The token end of file may be specified by the user, in which case,
1686 the user symbol is used in the reports, the graphs, and the verbose
e4ab1254 1687 error messages instead of "$end", which remains being the default.
b7c49edf 1688 For instance
e29f0771 1689 %token MYEOF 0
b7c49edf 1690 or
e29f0771 1691 %token MYEOF 0 "end of file"
fdbcd8e2 1692
d6fb461d 1693** Semantic parser
fdbcd8e2
AD
1694 This old option, which has been broken for ages, is removed.
1695
d6fb461d 1696** New translations
a861a339 1697 Brazilian Portuguese, thanks to Alexandre Folle de Menezes.
84614e13
AD
1698 Croatian, thanks to Denis Lackovic.
1699
d6fb461d 1700** Incorrect token definitions
e4ab1254
AD
1701 When given
1702 %token 'a' "A"
1703 bison used to output
1704 #define 'a' 65
b87f8b21 1705
d6fb461d 1706** Token definitions as enums
77714df2
AD
1707 Tokens are output both as the traditional #define's, and, provided
1708 the compiler supports ANSI C or is a C++ compiler, as enums.
e88dbdbf 1709 This lets debuggers display names instead of integers.
77714df2 1710
d6fb461d 1711** Reports
ec3bc396
AD
1712 In addition to --verbose, bison supports --report=THINGS, which
1713 produces additional information:
b408954b
AD
1714 - itemset
1715 complete the core item sets with their closure
e4ab1254 1716 - lookahead [changed to "look-ahead" in 1.875e through 2.3, but changed back]
9e6e7ed2 1717 explicitly associate lookahead tokens to items
b408954b
AD
1718 - solved
1719 describe shift/reduce conflicts solving.
1720 Bison used to systematically output this information on top of
1721 the report. Solved conflicts are now attached to their states.
ec3bc396 1722
d6fb461d 1723** Type clashes
9af3fbce
AD
1724 Previous versions don't complain when there is a type clash on
1725 the default action if the rule has a mid-rule action, such as in:
1726
e29f0771
AD
1727 %type <foo> bar
1728 %%
1729 bar: '0' {} '0';
9af3fbce
AD
1730
1731 This is fixed.
a861a339 1732
d6fb461d 1733** GNU M4 is now required when using Bison.
f987e9d2 1734\f
d6fb461d 1735* Changes in version 1.35, 2002-03-25:
76551463 1736
d6fb461d 1737** C Skeleton
76551463
AD
1738 Some projects use Bison's C parser with C++ compilers, and define
1739 YYSTYPE as a class. The recent adjustment of C parsers for data
1740 alignment and 64 bit architectures made this impossible.
1741
1742 Because for the time being no real solution for C++ parser
1743 generation exists, kludges were implemented in the parser to
1744 maintain this use. In the future, when Bison has C++ parsers, this
1745 kludge will be disabled.
1746
1747 This kludge also addresses some C++ problems when the stack was
1748 extended.
76551463 1749\f
d6fb461d 1750* Changes in version 1.34, 2002-03-12:
76551463 1751
d6fb461d 1752** File name clashes are detected
76551463 1753 $ bison foo.y -d -o foo.x
e4ab1254 1754 fatal error: header and parser would both be named "foo.x"
76551463 1755
e4ab1254 1756** A missing ";" at the end of a rule triggers a warning
76551463
AD
1757 In accordance with POSIX, and in agreement with other
1758 Yacc implementations, Bison will mandate this semicolon in the near
1759 future. This eases the implementation of a Bison parser of Bison
1760 grammars by making this grammar LALR(1) instead of LR(2). To
1761 facilitate the transition, this release introduces a warning.
1762
d6fb461d 1763** Revert the C++ namespace changes introduced in 1.31, as they caused too
76551463
AD
1764 many portability hassles.
1765
d6fb461d 1766** DJGPP support added.
76551463 1767
d6fb461d 1768** Fix test suite portability problems.
76551463 1769\f
d6fb461d 1770* Changes in version 1.33, 2002-02-07:
76551463 1771
d6fb461d 1772** Fix C++ issues
76551463
AD
1773 Groff could not be compiled for the definition of size_t was lacking
1774 under some conditions.
1775
d6fb461d 1776** Catch invalid @n
76551463
AD
1777 As is done with $n.
1778\f
d6fb461d 1779* Changes in version 1.32, 2002-01-23:
76551463 1780
d6fb461d 1781** Fix Yacc output file names
76551463 1782
d6fb461d 1783** Portability fixes
76551463 1784
d6fb461d 1785** Italian, Dutch translations
76551463 1786\f
d6fb461d 1787* Changes in version 1.31, 2002-01-14:
52d1aeee 1788
d6fb461d 1789** Many Bug Fixes
52d1aeee 1790
d6fb461d 1791** GNU Gettext and %expect
52d1aeee
MA
1792 GNU Gettext asserts 10 s/r conflicts, but there are 7. Now that
1793 Bison dies on incorrect %expectations, we fear there will be
1794 too many bug reports for Gettext, so _for the time being_, %expect
e4ab1254 1795 does not trigger an error when the input file is named "plural.y".
52d1aeee 1796
d6fb461d 1797** Use of alloca in parsers
52d1aeee
MA
1798 If YYSTACK_USE_ALLOCA is defined to 0, then the parsers will use
1799 malloc exclusively. Since 1.29, but was not NEWS'ed.
1800
1801 alloca is used only when compiled with GCC, to avoid portability
1802 problems as on AIX.
1803
d6fb461d 1804** yyparse now returns 2 if memory is exhausted; formerly it dumped core.
b47dbebe 1805
d6fb461d 1806** When the generated parser lacks debugging code, YYDEBUG is now 0
52d1aeee
MA
1807 (as POSIX requires) instead of being undefined.
1808
d6fb461d 1809** User Actions
52d1aeee
MA
1810 Bison has always permitted actions such as { $$ = $1 }: it adds the
1811 ending semicolon. Now if in Yacc compatibility mode, the semicolon
1812 is no longer output: one has to write { $$ = $1; }.
1813
d6fb461d 1814** Better C++ compliance
52d1aeee 1815 The output parsers try to respect C++ namespaces.
76551463 1816 [This turned out to be a failed experiment, and it was reverted later.]
52d1aeee 1817
d6fb461d 1818** Reduced Grammars
52d1aeee
MA
1819 Fixed bugs when reporting useless nonterminals.
1820
d6fb461d 1821** 64 bit hosts
52d1aeee
MA
1822 The parsers work properly on 64 bit hosts.
1823
d6fb461d 1824** Error messages
52d1aeee
MA
1825 Some calls to strerror resulted in scrambled or missing error messages.
1826
d6fb461d 1827** %expect
52d1aeee
MA
1828 When the number of shift/reduce conflicts is correct, don't issue
1829 any warning.
1830
d6fb461d 1831** The verbose report includes the rule line numbers.
52d1aeee 1832
d6fb461d 1833** Rule line numbers are fixed in traces.
52d1aeee 1834
d6fb461d 1835** Swedish translation
52d1aeee 1836
d6fb461d 1837** Parse errors
52d1aeee
MA
1838 Verbose parse error messages from the parsers are better looking.
1839 Before: parse error: unexpected `'/'', expecting `"number"' or `'-'' or `'(''
1840 Now: parse error: unexpected '/', expecting "number" or '-' or '('
1841
d6fb461d 1842** Fixed parser memory leaks.
52d1aeee
MA
1843 When the generated parser was using malloc to extend its stacks, the
1844 previous allocations were not freed.
1845
d6fb461d 1846** Fixed verbose output file.
52d1aeee
MA
1847 Some newlines were missing.
1848 Some conflicts in state descriptions were missing.
1849
d6fb461d 1850** Fixed conflict report.
52d1aeee
MA
1851 Option -v was needed to get the result.
1852
d6fb461d 1853** %expect
52d1aeee
MA
1854 Was not used.
1855 Mismatches are errors, not warnings.
1856
d6fb461d 1857** Fixed incorrect processing of some invalid input.
52d1aeee 1858
d6fb461d 1859** Fixed CPP guards: 9foo.h uses BISON_9FOO_H instead of 9FOO_H.
52d1aeee 1860
d6fb461d 1861** Fixed some typos in the documentation.
52d1aeee 1862
d6fb461d 1863** %token MY_EOF 0 is supported.
52d1aeee
MA
1864 Before, MY_EOF was silently renumbered as 257.
1865
d6fb461d 1866** doc/refcard.tex is updated.
52d1aeee 1867
d6fb461d 1868** %output, %file-prefix, %name-prefix.
52d1aeee
MA
1869 New.
1870
d6fb461d 1871** --output
e4ab1254 1872 New, aliasing "--output-file".
52d1aeee 1873\f
d6fb461d 1874* Changes in version 1.30, 2001-10-26:
342b8b6e 1875
e4ab1254
AD
1876** "--defines" and "--graph" have now an optional argument which is the
1877 output file name. "-d" and "-g" do not change; they do not take any
342b8b6e
AD
1878 argument.
1879
e4ab1254 1880** "%source_extension" and "%header_extension" are removed, failed
342b8b6e
AD
1881 experiment.
1882
d6fb461d 1883** Portability fixes.
f987e9d2 1884\f
d6fb461d 1885* Changes in version 1.29, 2001-09-07:
342b8b6e 1886
d6fb461d 1887** The output file does not define const, as this caused problems when used
342b8b6e
AD
1888 with common autoconfiguration schemes. If you still use ancient compilers
1889 that lack const, compile with the equivalent of the C compiler option
e4ab1254 1890 "-Dconst=". Autoconf's AC_C_CONST macro provides one way to do this.
342b8b6e 1891
e4ab1254 1892** Added "-g" and "--graph".
f87a2205 1893
d6fb461d 1894** The Bison manual is now distributed under the terms of the GNU FDL.
f2b5126e 1895
d6fb461d 1896** The input and the output files has automatically a similar extension.
234a3be3 1897
d6fb461d 1898** Russian translation added.
f87a2205 1899
d6fb461d 1900** NLS support updated; should hopefully be less troublesome.
f87a2205 1901
d6fb461d 1902** Added the old Bison reference card.
c33638bb 1903
e4ab1254 1904** Added "--locations" and "%locations".
6deb4447 1905
e4ab1254 1906** Added "-S" and "--skeleton".
cd5bd6ac 1907
e4ab1254 1908** "%raw", "-r", "--raw" is disabled.
62ab6972 1909
d6fb461d 1910** Special characters are escaped when output. This solves the problems
cd5bd6ac
AD
1911 of the #line lines with path names including backslashes.
1912
d6fb461d 1913** New directives.
e4ab1254
AD
1914 "%yacc", "%fixed_output_files", "%defines", "%no_parser", "%verbose",
1915 "%debug", "%source_extension" and "%header_extension".
f987e9d2 1916
d6fb461d 1917** @$
f987e9d2 1918 Automatic location tracking.
f87a2205 1919\f
d6fb461d 1920* Changes in version 1.28, 1999-07-06:
d2e00347 1921
d6fb461d 1922** Should compile better now with K&R compilers.
d2e00347 1923
d6fb461d 1924** Added NLS.
d2e00347 1925
d6fb461d 1926** Fixed a problem with escaping the double quote character.
d2e00347 1927
d6fb461d 1928** There is now a FAQ.
d2e00347 1929\f
d6fb461d 1930* Changes in version 1.27:
5c31c3c2 1931
d6fb461d 1932** The make rule which prevented bison.simple from being created on
5c31c3c2
JT
1933 some systems has been fixed.
1934\f
d6fb461d 1935* Changes in version 1.26:
4be07551 1936
7e508a2b 1937** Bison now uses Automake.
4be07551 1938
d6fb461d 1939** New mailing lists: <bug-bison@gnu.org> and <help-bison@gnu.org>.
4be07551 1940
d6fb461d 1941** Token numbers now start at 257 as previously documented, not 258.
4be07551 1942
d6fb461d 1943** Bison honors the TMPDIR environment variable.
4be07551 1944
d6fb461d 1945** A couple of buffer overruns have been fixed.
f51dbca1 1946
d6fb461d 1947** Problems when closing files should now be reported.
f51dbca1 1948
d6fb461d 1949** Generated parsers should now work even on operating systems which do
f51dbca1 1950 not provide alloca().
4be07551 1951\f
d6fb461d 1952* Changes in version 1.25, 1995-10-16:
df8878c5 1953
d6fb461d 1954** Errors in the input grammar are not fatal; Bison keeps reading
df8878c5 1955the grammar file, and reports all the errors found in it.
8c44d3ec 1956
d6fb461d 1957** Tokens can now be specified as multiple-character strings: for
df8878c5 1958example, you could use "<=" for a token which looks like <=, instead
7e508a2b 1959of choosing a name like LESSEQ.
df8878c5 1960
d6fb461d 1961** The %token_table declaration says to write a table of tokens (names
df8878c5
RS
1962and numbers) into the parser file. The yylex function can use this
1963table to recognize multiple-character string tokens, or for other
1964purposes.
1965
d6fb461d 1966** The %no_lines declaration says not to generate any #line preprocessor
df8878c5
RS
1967directives in the parser file.
1968
d6fb461d 1969** The %raw declaration says to use internal Bison token numbers, not
df8878c5
RS
1970Yacc-compatible token numbers, when token names are defined as macros.
1971
d6fb461d 1972** The --no-parser option produces the parser tables without including
df8878c5
RS
1973the parser engine; a project can now use its own parser engine.
1974The actions go into a separate file called NAME.act, in the form of
1975a switch statement body.
1976\f
d6fb461d 1977* Changes in version 1.23:
6780ca7a 1978
4d019228
DM
1979The user can define YYPARSE_PARAM as the name of an argument to be
1980passed into yyparse. The argument should have type void *. It should
1981actually point to an object. Grammar actions can access the variable
1982by casting it to the proper pointer type.
6780ca7a 1983
6780ca7a 1984Line numbers in output file corrected.
6780ca7a 1985\f
d6fb461d 1986* Changes in version 1.22:
6780ca7a
DM
1987
1988--help option added.
6780ca7a 1989\f
d6fb461d 1990* Changes in version 1.20:
6780ca7a
DM
1991
1992Output file does not redefine const for C++.
9f4503d6 1993
76551463
AD
1994-----
1995
c932d613 1996Copyright (C) 1995-2012 Free Software Foundation, Inc.
76551463 1997
8defe11b 1998This file is part of Bison, the GNU Parser Generator.
76551463 1999
f16b0819 2000This program is free software: you can redistribute it and/or modify
76551463 2001it under the terms of the GNU General Public License as published by
f16b0819
PE
2002the Free Software Foundation, either version 3 of the License, or
2003(at your option) any later version.
76551463 2004
f16b0819 2005This program is distributed in the hope that it will be useful,
76551463
AD
2006but WITHOUT ANY WARRANTY; without even the implied warranty of
2007MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
2008GNU General Public License for more details.
2009
2010You should have received a copy of the GNU General Public License
f16b0819 2011along with this program. If not, see <http://www.gnu.org/licenses/>.
7e508a2b
AD
2012
2013 LocalWords: yacc YYBACKUP glr GCC lalr ArrayIndexOutOfBoundsException nullptr
2014 LocalWords: cplusplus liby rpl fprintf mfcalc Wyacc stmt cond expr mk sym lr
2015 LocalWords: IELR ielr Lookahead YYERROR nonassoc LALR's api lookaheads yychar
2016 LocalWords: destructor lookahead YYRHSLOC YYLLOC Rhs ifndef YYFAIL cpp sr rr
2017 LocalWords: preprocessor initializer Wno Wnone Werror FreeBSD prec livelocks
2018 LocalWords: Solaris AIX UX RHEL Tru LHS gcc's Wundef YYENABLE NLS YYLTYPE VCG
2019 LocalWords: yyerror cpp's Wunused yylval yylloc prepend yyparse yylex yypush
2020 LocalWords: Graphviz xml nonterminals midrule destructor's YYSTYPE typedef ly
2021 LocalWords: CHR chr printf stdout namespace preprocessing enum pre include's
2022 LocalWords: YYRECOVERING nonfree destructors YYABORT YYACCEPT params enums de
2023 LocalWords: struct yystype DJGPP lex param Haible NUM alloca YYSTACK NUL goto
2024 LocalWords: YYMAXDEPTH Unescaped UCNs YYLTYPE's yyltype typedefs inline Yaccs
2025 LocalWords: Heriyanto Reenable dprec Hilfinger Eggert MYEOF Folle Menezes EOF
242cc08e 2026 LocalWords: Lackovic define's itemset Groff Gettext malloc NEWS'ed YYDEBUG YY
7e508a2b 2027 LocalWords: namespaces strerror const autoconfiguration Dconst Autoconf's FDL
242cc08e
AD
2028 LocalWords: Automake TMPDIR LESSEQ ylwrap endif yydebug YYTOKEN YYLSP ival hh
2029 LocalWords: extern YYTOKENTYPE TOKENTYPE yytokentype tokentype STYPE lval pdf
7287be84
AD
2030 LocalWords: lang yyoutput dvi html ps POSIX lvalp llocp calc yyo fval Wmaybe
2031 LocalWords: yyvsp pragmas noreturn java's
7e508a2b
AD
2032
2033Local Variables:
2034mode: outline
e4ab1254 2035fill-column: 76
7e508a2b 2036End: