]> git.saurik.com Git - bison.git/blame - NEWS
tests: fix invalid assignment when using variants in C++11
[bison.git] / NEWS
CommitLineData
ed7658fe 1GNU Bison NEWS
3af4feb2 2
facb910c
AD
3* Noteworthy changes in release ?.? (????-??-??) [?]
4
5
8faace8d 6* Noteworthy changes in release 2.7.91 (2013-06-24) [beta]
266cdc30 7
9cdcbdf2
AD
8** Java skeleton improvements
9
10 The Java skeleton now supports push parsing.
11 Contributed by Dennis Heimbigner.
266cdc30 12
fea25166 13* Noteworthy changes in release 2.7.90 (2013-05-30) [beta]
74553c98 14
8458a411 15** WARNING: Future backward-incompatibilities!
597afd73 16
597afd73
AD
17 Like other GNU packages, Bison will start using some of the C99 features
18 for its own code, especially the definition of variables after statements.
19 The generated C parsers still aim at C90.
20
8458a411 21** Backward incompatible changes
47db7ed1
AD
22
23*** Obsolete features
24
40bb6f78
AD
25 Support for YYFAIL is removed (deprecated in Bison 2.4.2): use YYERROR.
26
27 Support for yystype and yyltype is removed (deprecated in Bison 1.875):
28 use YYSTYPE and YYLTYPE.
29
30 Support for YYLEX_PARAM and YYPARSE_PARAM is removed (deprecated in Bison
31 1.875): use %lex-param, %parse-param, or %param.
47db7ed1 32
05e25f23
AD
33 Missing semicolons at the end of actions are no longer added (as announced
34 in the release 2.5).
35
1fa19a76
AD
36*** Use of YACC='bison -y'
37
38 TL;DR: With Autoconf <= 2.69, pass -Wno-yacc to (AM_)YFLAGS if you use
39 Bison extensions.
40
41 Traditional Yacc generates 'y.tab.c' whatever the name of the input file.
42 Therefore Makefiles written for Yacc expect 'y.tab.c' (and possibly
43 'y.tab.h' and 'y.outout') to be generated from 'foo.y'.
44
45 To this end, for ages, AC_PROG_YACC, Autoconf's macro to look for an
46 implementation of Yacc, was using Bison as 'bison -y'. While it does
47 ensure compatible output file names, it also enables warnings for
48 incompatibilities with POSIX Yacc. In other words, 'bison -y' triggers
49 warnings for Bison extensions.
50
51 Autoconf 2.70+ fixes this incompatibility by using YACC='bison -o y.tab.c'
52 (which also generates 'y.tab.h' and 'y.output' when needed).
53 Alternatively, disable Yacc warnings by passing '-Wno-yacc' to your Yacc
54 flags (YFLAGS, or AM_YFLAGS with Automake).
55
597afd73
AD
56** Bug fixes
57
c21e515e 58*** The epilogue is no longer affected by internal #defines (glr.c)
597afd73
AD
59
60 The glr.c skeleton uses defines such as #define yylval (yystackp->yyval) in
61 generated code. These weren't properly undefined before the inclusion of
62 the user epilogue, so functions such as the following were butchered by the
63 preprocessor expansion:
64
65 int yylex (YYSTYPE *yylval);
66
6c7022f7 67 This is fixed: yylval, yynerrs, yychar, and yylloc are now valid
597afd73
AD
68 identifiers for user-provided variables.
69
f0f95a50
AD
70*** stdio.h is no longer needed when locations are enabled (yacc.c)
71
72 Changes in Bison 2.7 introduced a dependency on FILE and fprintf when
73 locations are enabled. This is fixed.
74
bb4b189b
AD
75*** Warnings about useless %pure-parser/%define api.pure are restored
76
597afd73
AD
77** Diagnostics reported by Bison
78
79 Most of these features were contributed by Théophile Ranquet and Victor
80 Santet.
73370a9d 81
016426c1
TR
82*** Carets
83
84 Version 2.7 introduced caret errors, for a prettier output. These are now
85 activated by default. The old format can still be used by invoking Bison
86 with -fno-caret (or -fnone).
87
fec5f3c0
AD
88 Some error messages that reproduced excerpts of the grammar are now using
89 the caret information only. For instance on:
90
91 %%
92 exp: 'a' | 'a';
93
94 Bison 2.7 reports:
95
96 in.y: warning: 1 reduce/reduce conflict [-Wconflicts-rr]
97 in.y:2.12-14: warning: rule useless in parser due to conflicts: exp: 'a' [-Wother]
98
99 Now bison reports:
100
101 in.y: warning: 1 reduce/reduce conflict [-Wconflicts-rr]
102 in.y:2.12-14: warning: rule useless in parser due to conflicts [-Wother]
103 exp: 'a' | 'a';
104 ^^^
105
106 and "bison -fno-caret" reports:
107
108 in.y: warning: 1 reduce/reduce conflict [-Wconflicts-rr]
109 in.y:2.12-14: warning: rule useless in parser due to conflicts [-Wother]
110
1048a1c9 111*** Enhancements of the -Werror option
518e8830 112
1048a1c9 113 The -Werror=CATEGORY option is now recognized, and will treat specified
d949eefd
AD
114 warnings as errors. The warnings need not have been explicitly activated
115 using the -W option, this is similar to what GCC 4.7 does.
1048a1c9
AD
116
117 For example, given the following command line, Bison will treat both
d949eefd 118 warnings related to POSIX Yacc incompatibilities and S/R conflicts as
1048a1c9
AD
119 errors (and only those):
120
121 $ bison -Werror=yacc,error=conflicts-sr input.y
122
123 If no categories are specified, -Werror will make all active warnings into
124 errors. For example, the following line does the same the previous example:
125
126 $ bison -Werror -Wnone -Wyacc -Wconflicts-sr input.y
127
128 (By default -Wconflicts-sr,conflicts-rr,deprecated,other is enabled.)
129
130 Note that the categories in this -Werror option may not be prefixed with
131 "no-". However, -Wno-error[=CATEGORY] is valid.
132
133 Note that -y enables -Werror=yacc. Therefore it is now possible to require
134 Yacc-like behavior (e.g., always generate y.tab.c), but to report
135 incompatibilities as warnings: "-y -Wno-error=yacc".
518e8830 136
46bdb8ec 137*** The display of warnings is now richer
73370a9d 138
46bdb8ec 139 The option that controls a given warning is now displayed:
73370a9d 140
46bdb8ec 141 foo.y:4.6: warning: type clash on default action: <foo> != <bar> [-Wother]
73370a9d 142
46bdb8ec
TR
143 In the case of warnings treated as errors, the prefix is changed from
144 "warning: " to "error: ", and the suffix is displayed, in a manner similar
d949eefd 145 to GCC, as [-Werror=CATEGORY].
1048a1c9 146
46bdb8ec
TR
147 For instance, where the previous version of Bison would report (and exit
148 with failure):
1048a1c9
AD
149
150 bison: warnings being treated as errors
46bdb8ec 151 input.y:1.1: warning: stray ',' treated as white space
1048a1c9 152
46bdb8ec 153 it now reports:
1048a1c9 154
1048a1c9
AD
155 input.y:1.1: error: stray ',' treated as white space [-Werror=other]
156
157*** Deprecated constructs
158
159 The new 'deprecated' warning category flags obsolete constructs whose
160 support will be discontinued. It is enabled by default. These warnings
161 used to be reported as 'other' warnings.
162
73370a9d 163*** Useless semantic types
9641b918
VS
164
165 Bison now warns about useless (uninhabited) semantic types. Since
166 semantic types are not declared to Bison (they are defined in the opaque
167 %union structure), it is %printer/%destructor directives about useless
168 types that trigger the warning:
169
170 %token <type1> term
171 %type <type2> nterm
172 %printer {} <type1> <type3>
173 %destructor {} <type2> <type4>
174 %%
175 nterm: term { $$ = $1; };
176
177 3.28-34: warning: type <type3> is used, but is not associated to any symbol
178 4.28-34: warning: type <type4> is used, but is not associated to any symbol
179
31557b9e 180*** Undefined but unused symbols
b921d92f 181
31557b9e
AD
182 Bison used to raise an error for undefined symbols that are not used in
183 the grammar. This is now only a warning.
b921d92f
VS
184
185 %printer {} symbol1
186 %destructor {} symbol2
31557b9e 187 %type <type> symbol3
b921d92f
VS
188 %%
189 exp: "a";
190
73370a9d 191*** Useless destructors or printers
ea9a35c6
VS
192
193 Bison now warns about useless destructors or printers. In the following
194 example, the printer for <type1>, and the destructor for <type2> are
195 useless: all symbols of <type1> (token1) already have a printer, and all
196 symbols of type <type2> (token2) already have a destructor.
197
198 %token <type1> token1
199 <type2> token2
200 <type3> token3
201 <type4> token4
202 %printer {} token1 <type1> <type3>
203 %destructor {} token2 <type2> <type4>
204
d87ea54c
AD
205*** Conflicts
206
207 The warnings and error messages about shift/reduce and reduce/reduce
208 conflicts have been normalized. For instance on the following foo.y file:
209
210 %glr-parser
211 %%
212 exp: exp '+' exp | '0' | '0';
213
214 compare the previous version of bison:
215
216 $ bison foo.y
217 foo.y: conflicts: 1 shift/reduce, 2 reduce/reduce
218 $ bison -Werror foo.y
219 bison: warnings being treated as errors
220 foo.y: conflicts: 1 shift/reduce, 2 reduce/reduce
221
222 with the new behavior:
223
224 $ bison foo.y
225 foo.y: warning: 1 shift/reduce conflict [-Wconflicts-sr]
226 foo.y: warning: 2 reduce/reduce conflicts [-Wconflicts-rr]
227 $ bison -Werror foo.y
9503b0a4
TR
228 foo.y: error: 1 shift/reduce conflict [-Werror=conflicts-sr]
229 foo.y: error: 2 reduce/reduce conflicts [-Werror=conflicts-rr]
d87ea54c
AD
230
231 When %expect or %expect-rr is used, such as with bar.y:
232
233 %expect 0
234 %glr-parser
235 %%
236 exp: exp '+' exp | '0' | '0';
237
238 Former behavior:
239
240 $ bison bar.y
241 bar.y: conflicts: 1 shift/reduce, 2 reduce/reduce
242 bar.y: expected 0 shift/reduce conflicts
243 bar.y: expected 0 reduce/reduce conflicts
244
245 New one:
246
247 $ bison bar.y
9503b0a4
TR
248 bar.y: error: shift/reduce conflicts: 1 found, 0 expected
249 bar.y: error: reduce/reduce conflicts: 2 found, 0 expected
d87ea54c 250
f24695ef
AD
251** Incompatibilities with POSIX Yacc
252
253 The 'yacc' category is no longer part of '-Wall', enable it explicitly
254 with '-Wyacc'.
255
2055a44e
AD
256** Additional yylex/yyparse arguments
257
6dc4663d
AD
258 The new directive %param declares additional arguments to both yylex and
259 yyparse. The %lex-param, %parse-param, and %param directives support one
260 or more arguments. Instead of
2055a44e 261
e436fa67
AD
262 %lex-param {arg1_type *arg1}
263 %lex-param {arg2_type *arg2}
264 %parse-param {arg1_type *arg1}
265 %parse-param {arg2_type *arg2}
2055a44e
AD
266
267 one may now declare
268
e436fa67 269 %param {arg1_type *arg1} {arg2_type *arg2}
2055a44e 270
630a0218
AD
271** Types of values for %define variables
272
273 Bison used to make no difference between '%define foo bar' and '%define
274 foo "bar"'. The former is now called a 'keyword value', and the latter a
275 'string value'. A third kind was added: 'code values', such as '%define
276 foo {bar}'.
277
278 Keyword variables are used for fixed value sets, e.g.,
279
280 %define lr.type lalr
281
282 Code variables are used for value in the target language, e.g.,
283
284 %define api.value.type {struct semantic_type}
285
286 String variables are used remaining cases, e.g. file names.
287
2a6b66c5 288** Variable api.token.prefix
99c08fb6 289
2a6b66c5 290 The variable api.token.prefix changes the way tokens are identified in
99c08fb6
AD
291 the generated files. This is especially useful to avoid collisions
292 with identifiers in the target language. For instance
293
e436fa67 294 %token FILE for ERROR
630a0218 295 %define api.token.prefix {TOK_}
e436fa67
AD
296 %%
297 start: FILE for ERROR;
99c08fb6
AD
298
299 will generate the definition of the symbols TOK_FILE, TOK_for, and
300 TOK_ERROR in the generated sources. In particular, the scanner must
301 use these prefixed token names, although the grammar itself still
302 uses the short names (as in the sample rule given above).
303
6574576c
AD
304** Variable api.value.type
305
306 This new %define variable supersedes the #define macro YYSTYPE. The use
307 of YYSTYPE is discouraged. In particular, #defining YYSTYPE *and* either
308 using %union or %defining api.value.type results in undefined behavior.
309
310 Either define api.value.type, or use "%union":
311
312 %union
313 {
314 int ival;
315 char *sval;
316 }
317 %token <ival> INT "integer"
318 %token <sval> STRING "string"
319 %printer { fprintf (yyo, "%d", $$); } <ival>
320 %destructor { free ($$); } <sval>
321
322 /* In yylex(). */
323 yylval.ival = 42; return INT;
324 yylval.sval = "42"; return STRING;
325
630a0218
AD
326 The %define variable api.value.type supports both keyword and code values.
327
328 The keyword value 'union' means that the user provides genuine types, not
435575cb 329 union member names such as "ival" and "sval" above (WARNING: will fail if
1fa19a76 330 -y/--yacc/%yacc is enabled).
6574576c 331
435575cb 332 %define api.value.type union
6574576c
AD
333 %token <int> INT "integer"
334 %token <char *> STRING "string"
335 %printer { fprintf (yyo, "%d", $$); } <int>
336 %destructor { free ($$); } <char *>
337
338 /* In yylex(). */
339 yylval.INT = 42; return INT;
340 yylval.STRING = "42"; return STRING;
341
435575cb
AD
342 The keyword value variant is somewhat equivalent, but for C++ special
343 provision is made to allow classes to be used (more about this below).
6574576c 344
435575cb 345 %define api.value.type variant
6574576c
AD
346 %token <int> INT "integer"
347 %token <std::string> STRING "string"
348
630a0218 349 Code values (in braces) denote user defined types. This is where YYSTYPE
435575cb 350 used to be used.
6574576c
AD
351
352 %code requires
353 {
354 struct my_value
355 {
356 enum
357 {
358 is_int, is_string
359 } kind;
360 union
361 {
362 int ival;
363 char *sval;
364 } u;
365 };
366 }
435575cb 367 %define api.value.type {struct my_value}
6574576c
AD
368 %token <u.ival> INT "integer"
369 %token <u.sval> STRING "string"
370 %printer { fprintf (yyo, "%d", $$); } <u.ival>
371 %destructor { free ($$); } <u.sval>
372
373 /* In yylex(). */
374 yylval.u.ival = 42; return INT;
375 yylval.u.sval = "42"; return STRING;
376
31b850d2
AD
377** Variable parse.error
378
1f77b2e0
AD
379 This variable controls the verbosity of error messages. The use of the
380 %error-verbose directive is deprecated in favor of "%define parse.error
381 verbose".
31b850d2 382
c21e515e
AD
383** Renamed %define variables
384
385 The following variables have been renamed for consistency. Backward
386 compatibility is ensured, but upgrading is recommended.
387
388 lr.default-reductions -> lr.default-reduction
389 lr.keep-unreachable-states -> lr.keep-unreachable-state
390 namespace -> api.namespace
391 stype -> api.value.type
392
ca2a6d15
PH
393** Semantic predicates
394
597afd73
AD
395 Contributed by Paul Hilfinger.
396
1f77b2e0
AD
397 The new, experimental, semantic-predicate feature allows actions of the
398 form "%?{ BOOLEAN-EXPRESSION }", which cause syntax errors (as for
ca2a6d15 399 YYERROR) if the expression evaluates to 0, and are evaluated immediately
1f77b2e0
AD
400 in GLR parsers, rather than being deferred. The result is that they allow
401 the programmer to prune possible parses based on the values of run-time
402 expressions.
ca2a6d15 403
d1400569
AD
404** The directive %expect-rr is now an error in non GLR mode
405
406 It used to be an error only if used in non GLR mode, _and_ if there are
407 reduce/reduce conflicts.
408
5202b6ac 409** Tokens are numbered in their order of appearance
93561c21 410
5202b6ac
VT
411 Contributed by Valentin Tolmer.
412
413 With '%token A B', A had a number less than the one of B. However,
414 precedence declarations used to generate a reversed order. This is now
415 fixed, and introducing tokens with any of %token, %left, %right,
416 %precedence, or %nonassoc yields the same result.
417
418 When mixing declarations of tokens with a litteral character (e.g., 'a')
419 or with an identifier (e.g., B) in a precedence declaration, Bison
420 numbered the litteral characters first. For example
421
422 %right A B 'c' 'd'
423
424 would lead to the tokens declared in this order: 'c' 'd' A B. Again, the
425 input order is now preserved.
426
427 These changes were made so that one can remove useless precedence and
428 associativity declarations (i.e., map %nonassoc, %left or %right to
429 %precedence, or to %token) and get exactly the same output.
93561c21 430
cc2235ac
VT
431** Useless precedence and associativity
432
d2f9ae18
AD
433 Contributed by Valentin Tolmer.
434
1282c124 435 When developing and maintaining a grammar, useless associativity and
cc2235ac
VT
436 precedence directives are common. They can be a nuisance: new ambiguities
437 arising are sometimes masked because their conflicts are resolved due to
438 the extra precedence or associativity information. Furthermore, it can
439 hinder the comprehension of a new grammar: one will wonder about the role
440 of a precedence, where in fact it is useless. The following changes aim
441 at detecting and reporting these extra directives.
442
443*** Precedence warning category
444
445 A new category of warning, -Wprecedence, was introduced. It flags the
446 useless precedence and associativity directives.
447
448*** Useless associativity
449
450 Bison now warns about symbols with a declared associativity that is never
451 used to resolve conflicts. In that case, using %precedence is sufficient;
452 the parsing tables will remain unchanged. Solving these warnings may raise
453 useless precedence warnings, as the symbols no longer have associativity.
454 For example:
455
456 %left '+'
457 %left '*'
458 %%
459 exp:
1282c124
AD
460 "number"
461 | exp '+' "number"
cc2235ac
VT
462 | exp '*' exp
463 ;
464
465 will produce a
466
467 warning: useless associativity for '+', use %precedence [-Wprecedence]
468 %left '+'
469 ^^^
470
471*** Useless precedence
472
473 Bison now warns about symbols with a declared precedence and no declared
474 associativity (i.e., declared with %precedence), and whose precedence is
475 never used. In that case, the symbol can be safely declared with %token
476 instead, without modifying the parsing tables. For example:
477
478 %precedence '='
479 %%
1282c124 480 exp: "var" '=' "number";
cc2235ac
VT
481
482 will produce a
483
484 warning: useless precedence for '=' [-Wprecedence]
485 %precedence '='
486 ^^^
487
488*** Useless precedence and associativity
489
490 In case of both useless precedence and associativity, the issue is flagged
491 as follows:
492
493 %nonassoc '='
494 %%
1282c124 495 exp: "var" '=' "number";
cc2235ac
VT
496
497 The warning is:
498
499 warning: useless precedence and associativity for '=' [-Wprecedence]
500 %nonassoc '='
501 ^^^
09add9c2
AD
502
503** Empty rules
504
6240346a
AD
505 With help from Joel E. Denny and Gabriel Rassoul.
506
09add9c2
AD
507 Empty rules (i.e., with an empty right-hand side) can now be explicitly
508 marked by the new %empty directive. Using %empty on a non-empty rule is
509 an error. The new -Wempty-rule warning reports empty rules without
510 %empty. On the following grammar:
511
512 %%
513 s: a b c;
514 a: ;
515 b: %empty;
516 c: 'a' %empty;
517
518 bison reports:
519
520 3.4-5: warning: empty rule without %empty [-Wempty-rule]
521 a: {}
522 ^^
523 5.8-13: error: %empty on non-empty rule
524 c: 'a' %empty {};
525 ^^^^^^
cc2235ac 526
c21e515e
AD
527** Java skeleton improvements
528
c21e515e
AD
529 The constants for token names were moved to the Lexer interface. Also, it
530 is possible to add code to the parser's constructors using "%code init"
531 and "%define init_throws".
aa94def1
DH
532 Contributed by Paolo Bonzini.
533
c21e515e
AD
534** C++ skeletons improvements
535
536*** The parser header is no longer mandatory (lalr1.cc, glr.cc)
537
538 Using %defines is now optional. Without it, the needed support classes
539 are defined in the generated parser, instead of additional files (such as
540 location.hh, position.hh and stack.hh).
541
542*** Locations are no longer mandatory (lalr1.cc, glr.cc)
543
544 Both lalr1.cc and glr.cc no longer require %location.
545
546*** syntax_error exception (lalr1.cc)
547
548 The C++ parser features a syntax_error exception, which can be
549 thrown from the scanner or from user rules to raise syntax errors.
550 This facilitates reporting errors caught in sub-functions (e.g.,
551 rejecting too large integral literals from a conversion function
552 used by the scanner, or rejecting invalid combinations from a
553 factory invoked by the user actions).
554
555*** %define api.value.type variant
556
557 This is based on a submission from Michiel De Wilde. With help
558 from Théophile Ranquet.
559
560 In this mode, complex C++ objects can be used as semantic values. For
561 instance:
562
563 %token <::std::string> TEXT;
564 %token <int> NUMBER;
565 %token SEMICOLON ";"
566 %type <::std::string> item;
567 %type <::std::list<std::string>> list;
568 %%
569 result:
570 list { std::cout << $1 << std::endl; }
571 ;
572
573 list:
6240346a 574 %empty { /* Generates an empty string list. */ }
c21e515e
AD
575 | list item ";" { std::swap ($$, $1); $$.push_back ($2); }
576 ;
577
578 item:
579 TEXT { std::swap ($$, $1); }
580 | NUMBER { $$ = string_cast ($1); }
581 ;
582
583*** %define api.token.constructor
584
585 When variants are enabled, Bison can generate functions to build the
586 tokens. This guarantees that the token type (e.g., NUMBER) is consistent
587 with the semantic value (e.g., int):
588
589 parser::symbol_type yylex ()
590 {
591 parser::location_type loc = ...;
592 ...
593 return parser::make_TEXT ("Hello, world!", loc);
594 ...
595 return parser::make_NUMBER (42, loc);
596 ...
597 return parser::make_SEMICOLON (loc);
598 ...
599 }
600
75ae8299
AD
601*** C++ locations
602
603 There are operator- and operator-= for 'location'. Negative line/column
604 increments can no longer underflow the resulting value.
605
1f5542fe 606* Noteworthy changes in release 2.7.1 (2013-04-15) [stable]
cc8962bd 607
80a2826e
AD
608** Bug fixes
609
610*** Fix compiler attribute portability (yacc.c)
611
612 With locations enabled, __attribute__ was used unprotected.
0a7b8559 613
e83be476
AD
614*** Fix some compiler warnings (lalr1.cc)
615
c13bb348 616* Noteworthy changes in release 2.7 (2012-12-12) [stable]
effd30c0 617
edf9a06f 618** Bug fixes
7bada535 619
edf9a06f 620 Warnings about uninitialized yylloc in yyparse have been fixed.
7bada535 621
1127a75a
AD
622 Restored C90 compliance (yet no report was ever made).
623
d4fe9e88 624** Diagnostics are improved
7bada535 625
8458a411
AD
626 Contributed by Théophile Ranquet.
627
d4fe9e88 628*** Changes in the format of error messages
7bada535 629
d4fe9e88 630 This used to be the format of many error reports:
1f1bd572 631
d4fe9e88
AD
632 input.y:2.7-12: %type redeclaration for exp
633 input.y:1.7-12: previous declaration
1f1bd572 634
d4fe9e88 635 It is now:
1f1bd572 636
d4fe9e88
AD
637 input.y:2.7-12: error: %type redeclaration for exp
638 input.y:1.7-12: previous declaration
cbaea010 639
d4fe9e88 640*** New format for error reports: carets
cbaea010 641
d4fe9e88 642 Caret errors have been added to Bison:
cbaea010 643
d4fe9e88
AD
644 input.y:2.7-12: error: %type redeclaration for exp
645 %type <sval> exp
646 ^^^^^^
647 input.y:1.7-12: previous declaration
648 %type <ival> exp
649 ^^^^^^
cbaea010 650
d4fe9e88 651 or
cbaea010 652
7bada535 653 input.y:3.20-23: error: ambiguous reference: '$exp'
fb6040f0 654 exp: exp '+' exp { $exp = $1 + $3; };
7bada535 655 ^^^^
fb6040f0
TR
656 input.y:3.1-3: refers to: $exp at $$
657 exp: exp '+' exp { $exp = $1 + $3; };
658 ^^^
659 input.y:3.6-8: refers to: $exp at $1
660 exp: exp '+' exp { $exp = $1 + $3; };
661 ^^^
662 input.y:3.14-16: refers to: $exp at $3
663 exp: exp '+' exp { $exp = $1 + $3; };
664 ^^^
7bada535 665
1282c124
AD
666 The default behavior for now is still not to display these unless
667 explicitly asked with -fcaret (or -fall). However, in a later release, it
d4fe9e88
AD
668 will be made the default behavior (but may still be deactivated with
669 -fno-caret).
d3e4409a 670
1f1bd572 671** New value for %define variable: api.pure full
d3e4409a 672
1f1bd572 673 The %define variable api.pure requests a pure (reentrant) parser. However,
d4fe9e88
AD
674 for historical reasons, using it in a location-tracking Yacc parser
675 resulted in a yyerror function that did not take a location as a
676 parameter. With this new value, the user may request a better pure parser,
677 where yyerror does take a location as a parameter (in location-tracking
678 parsers).
1f1bd572
TR
679
680 The use of "%define api.pure true" is deprecated in favor of this new
681 "%define api.pure full".
d3e4409a 682
7287be84 683** New %define variable: api.location.type (glr.cc, lalr1.cc, lalr1.java)
db8ab2be
AD
684
685 The %define variable api.location.type defines the name of the type to use
686 for locations. When defined, Bison no longer generates the position.hh
687 and location.hh files, nor does the parser will include them: the user is
688 then responsible to define her type.
689
690 This can be used in programs with several parsers to factor their location
7287be84
AD
691 and position files: let one of them generate them, and the others just use
692 them.
db8ab2be
AD
693
694 This feature was actually introduced, but not documented, in Bison 2.5,
695 under the name "location_type" (which is maintained for backward
696 compatibility).
697
7287be84
AD
698 For consistency, lalr1.java's %define variables location_type and
699 position_type are deprecated in favor of api.location.type and
700 api.position.type.
701
d4fe9e88
AD
702** Exception safety (lalr1.cc)
703
704 The parse function now catches exceptions, uses the %destructors to
705 release memory (the lookahead symbol and the symbols pushed on the stack)
706 before re-throwing the exception.
707
708 This feature is somewhat experimental. User feedback would be
709 appreciated.
710
9c16d399 711** Graph improvements in DOT and XSLT
fc4fdd62 712
8458a411
AD
713 Contributed by Théophile Ranquet.
714
fc4fdd62
TR
715 The graphical presentation of the states is more readable: their shape is
716 now rectangular, the state number is clearly displayed, and the items are
717 numbered and left-justified.
718
719 The reductions are now explicitly represented as transitions to other
720 diamond shaped nodes.
721
9c16d399
TR
722 These changes are present in both --graph output and xml2dot.xsl XSLT
723 processing, with minor (documented) differences.
724
d4fe9e88 725** %language is no longer an experimental feature.
fb4c8a7c 726
d4fe9e88
AD
727 The introduction of this feature, in 2.4, was four years ago. The
728 --language option and the %language directive are no longer experimental.
fb4c8a7c 729
53e2cd1e
AD
730** Documentation
731
732 The sections about shift/reduce and reduce/reduce conflicts resolution
733 have been fixed and extended.
9d2423f5 734
d4fe9e88
AD
735 Although introduced more than four years ago, XML and Graphviz reports
736 were not properly documented.
737
be22823e
AD
738 The translation of mid-rule actions is now described.
739
9d3f7eaf 740* Noteworthy changes in release 2.6.5 (2012-11-07) [stable]
6f1360bd 741
a68b1f23
AD
742 We consider compiler warnings about Bison generated parsers to be bugs.
743 Rather than working around them in your own project, please consider
744 reporting them to us.
745
746** Bug fixes
747
748 Warnings about uninitialized yylval and/or yylloc for push parsers with a
749 pure interface have been fixed for GCC 4.0 up to 4.8, and Clang 2.9 to
750 3.2.
751
752 Other issues in the test suite have been addressed.
6f1360bd 753
1282c124 754 Null characters are correctly displayed in error messages.
95066e92 755
a1d1ab50
AD
756 When possible, yylloc is correctly initialized before calling yylex. It
757 is no longer necessary to initialize it in the %initial-action.
758
0ac15849 759* Noteworthy changes in release 2.6.4 (2012-10-23) [stable]
a4eb820f 760
468455e1 761 Bison 2.6.3's --version was incorrect. This release fixes this issue.
a4eb820f 762
6eb8f74f 763* Noteworthy changes in release 2.6.3 (2012-10-22) [stable]
933ec544 764
6b4cb804
AD
765** Bug fixes
766
a1a77e1f 767 Bugs and portability issues in the test suite have been fixed.
6b4cb804
AD
768
769 Some errors in translations have been addressed, and --help now directs
770 users to the appropriate place to report them.
771
772 Stray Info files shipped by accident are removed.
773
774 Incorrect definitions of YY_, issued by yacc.c when no parser header is
775 generated, are removed.
9c26b8fc 776
a2b3f101
TR
777 All the generated headers are self-contained.
778
c9d5bcc9
AD
779** Header guards (yacc.c, glr.c, glr.cc)
780
781 In order to avoid collisions, the header guards are now
782 YY_<PREFIX>_<FILE>_INCLUDED, instead of merely <PREFIX>_<FILE>.
783 For instance the header generated from
784
785 %define api.prefix "calc"
786 %defines "lib/parse.h"
787
788 will use YY_CALC_LIB_PARSE_H_INCLUDED as guard.
789
c12c4c50 790** Fix compiler warnings in the generated parser (yacc.c, glr.c)
321d3e35
AD
791
792 The compilation of pure parsers (%define api.pure) can trigger GCC
793 warnings such as:
794
795 input.c: In function 'yyparse':
796 input.c:1503:12: warning: 'yylval' may be used uninitialized in this
797 function [-Wmaybe-uninitialized]
798 *++yyvsp = yylval;
799 ^
800
801 This is now fixed; pragmas to avoid these warnings are no longer needed.
802
c12c4c50
AD
803 Warnings from clang ("equality comparison with extraneous parentheses" and
804 "function declared 'noreturn' should not return") have also been
805 addressed.
806
e1eeecd3 807* Noteworthy changes in release 2.6.2 (2012-08-03) [stable]
9c26b8fc 808
43ca8040
AD
809** Bug fixes
810
811 Buffer overruns, complaints from Flex, and portability issues in the test
812 suite have been fixed.
813
c9d546b2
AD
814** Spaces in %lex- and %parse-param (lalr1.cc, glr.cc)
815
816 Trailing end-of-lines in %parse-param or %lex-param would result in
817 invalid C++. This is fixed.
9c26b8fc 818
dcd5344d
AD
819** Spurious spaces and end-of-lines
820
821 The generated files no longer end (nor start) with empty lines.
822
77b214ef 823* Noteworthy changes in release 2.6.1 (2012-07-30) [stable]
a4107f24 824
8617d87e
AD
825 Bison no longer executes user-specified M4 code when processing a grammar.
826
e20e6a50
AD
827** Future Changes
828
829 In addition to the removal of the features announced in Bison 2.6, the
830 next major release will remove the "Temporary hack for adding a semicolon
831 to the user action", as announced in the release 2.5. Instead of:
832
833 exp: exp "+" exp { $$ = $1 + $3 };
834
835 write:
836
837 exp: exp "+" exp { $$ = $1 + $3; };
838
8617d87e
AD
839** Bug fixes
840
0e164d43
AD
841*** Type names are now properly escaped.
842
843*** glr.cc: set_debug_level and debug_level work as expected.
a4107f24 844
26313726
AD
845*** Stray @ or $ in actions
846
847 While Bison used to warn about stray $ or @ in action rules, it did not
848 for other actions such as printers, destructors, or initial actions. It
849 now does.
850
cd735a8c 851** Type names in actions
4982f078
AD
852
853 For consistency with rule actions, it is now possible to qualify $$ by a
cd735a8c 854 type-name in destructors, printers, and initial actions. For instance:
4982f078
AD
855
856 %printer { fprintf (yyo, "(%d, %f)", $<ival>$, $<fval>$); } <*> <>;
857
858 will display two values for each typed and untyped symbol (provided
cd735a8c 859 that YYSTYPE has both "ival" and "fval" fields).
60aa04a2 860
1505e8bb 861* Noteworthy changes in release 2.6 (2012-07-19) [stable]
0f11eec2 862
d0a30438 863** Future changes
9553083c 864
55d1006f
AD
865 The next major release of Bison will drop support for the following
866 deprecated features. Please report disagreements to bug-bison@gnu.org.
0f11eec2 867
aaf61036 868*** K&R C parsers
55d1006f
AD
869
870 Support for generating parsers in K&R C will be removed. Parsers
242cc08e 871 generated for C support ISO C90, and are tested with ISO C99 and ISO C11
55d1006f
AD
872 compilers.
873
258cddbc 874*** Features deprecated since Bison 1.875
0f11eec2 875
258cddbc
AD
876 The definitions of yystype and yyltype will be removed; use YYSTYPE and
877 YYLTYPE.
0f11eec2 878
258cddbc
AD
879 YYPARSE_PARAM and YYLEX_PARAM, deprecated in favor of %parse-param and
880 %lex-param, will no longer be supported.
881
882 Support for the preprocessor symbol YYERROR_VERBOSE will be removed, use
883 %error-verbose.
55d1006f
AD
884
885*** The generated header will be included (yacc.c)
0f11eec2
AD
886
887 Instead of duplicating the content of the generated header (definition of
55d1006f
AD
888 YYSTYPE, yyparse declaration etc.), the generated parser will include it,
889 as is already the case for GLR or C++ parsers. This change is deferred
890 because existing versions of ylwrap (e.g., Automake 1.12.1) do not support
891 it.
0f11eec2 892
c2425191 893** Generated Parser Headers
56ca3d8f 894
258cddbc 895*** Guards (yacc.c, glr.c, glr.cc)
c3e9f08f
AD
896
897 The generated headers are now guarded, as is already the case for C++
242cc08e 898 parsers (lalr1.cc). For instance, with --defines=foo.h:
c3e9f08f 899
e29f0771
AD
900 #ifndef YY_FOO_H
901 # define YY_FOO_H
902 ...
903 #endif /* !YY_FOO_H */
c3e9f08f 904
258cddbc 905*** New declarations (yacc.c, glr.c)
56ca3d8f
AD
906
907 The generated header now declares yydebug and yyparse. Both honor
908 --name-prefix=bar_, and yield
909
e29f0771 910 int bar_parse (void);
56ca3d8f
AD
911
912 rather than
913
e29f0771
AD
914 #define yyparse bar_parse
915 int yyparse (void);
56ca3d8f
AD
916
917 in order to facilitate the inclusion of several parser headers inside a
918 single compilation unit.
c3e9f08f 919
258cddbc
AD
920*** Exported symbols in C++
921
922 The symbols YYTOKEN_TABLE and YYERROR_VERBOSE, which were defined in the
923 header, are removed, as they prevent the possibility of including several
924 generated headers from a single compilation unit.
925
694af10c
AD
926*** YYLSP_NEEDED
927
928 For the same reasons, the undocumented and unused macro YYLSP_NEEDED is no
929 longer defined.
930
4b3847c3
AD
931** New %define variable: api.prefix
932
933 Now that the generated headers are more complete and properly protected
934 against multiple inclusions, constant names, such as YYSTYPE are a
935 problem. While yyparse and others are properly renamed by %name-prefix,
936 YYSTYPE, YYDEBUG and others have never been affected by it. Because it
937 would introduce backward compatibility issues in projects not expecting
938 YYSTYPE to be renamed, instead of changing the behavior of %name-prefix,
939 it is deprecated in favor of a new %define variable: api.prefix.
940
941 The following examples compares both:
942
943 %name-prefix "bar_" | %define api.prefix "bar_"
944 %token <ival> FOO %token <ival> FOO
945 %union { int ival; } %union { int ival; }
946 %% %%
947 exp: 'a'; exp: 'a';
948
949 bison generates:
950
951 #ifndef BAR_FOO_H #ifndef BAR_FOO_H
952 # define BAR_FOO_H # define BAR_FOO_H
953
954 /* Enabling traces. */ /* Enabling traces. */
5f108727
AD
955 # ifndef YYDEBUG | # ifndef BAR_DEBUG
956 > # if defined YYDEBUG
957 > # if YYDEBUG
958 > # define BAR_DEBUG 1
959 > # else
960 > # define BAR_DEBUG 0
961 > # endif
962 > # else
963 # define YYDEBUG 0 | # define BAR_DEBUG 0
964 > # endif
965 # endif | # endif
966
967 # if YYDEBUG | # if BAR_DEBUG
4b3847c3
AD
968 extern int bar_debug; extern int bar_debug;
969 # endif # endif
970
971 /* Tokens. */ /* Tokens. */
972 # ifndef YYTOKENTYPE | # ifndef BAR_TOKENTYPE
973 # define YYTOKENTYPE | # define BAR_TOKENTYPE
974 enum yytokentype { | enum bar_tokentype {
975 FOO = 258 FOO = 258
976 }; };
977 # endif # endif
978
979 #if ! defined YYSTYPE \ | #if ! defined BAR_STYPE \
980 && ! defined YYSTYPE_IS_DECLARED | && ! defined BAR_STYPE_IS_DECLARED
981 typedef union YYSTYPE | typedef union BAR_STYPE
982 { {
983 int ival; int ival;
984 } YYSTYPE; | } BAR_STYPE;
985 # define YYSTYPE_IS_DECLARED 1 | # define BAR_STYPE_IS_DECLARED 1
986 #endif #endif
987
988 extern YYSTYPE bar_lval; | extern BAR_STYPE bar_lval;
989
990 int bar_parse (void); int bar_parse (void);
991
992 #endif /* !BAR_FOO_H */ #endif /* !BAR_FOO_H */
993
dfaac272 994* Noteworthy changes in release 2.5.1 (2012-06-05) [stable]
df6e3db0 995
debe2c03 996** Future changes:
765e1bd4 997
e4ab1254 998 The next major release will drop support for generating parsers in K&R C.
041308d0 999
466b4cf2 1000** yacc.c: YYBACKUP works as expected.
ef51bfa7 1001
d834eca0 1002** glr.c improvements:
041308d0 1003
d834eca0 1004*** Location support is eliminated when not requested:
041308d0 1005
e4ab1254
AD
1006 GLR parsers used to include location-related code even when locations were
1007 not requested, and therefore not even usable.
378e917c 1008
d834eca0 1009*** __attribute__ is preserved:
d115aad9 1010
e4ab1254
AD
1011 __attribute__ is no longer disabled when __STRICT_ANSI__ is defined (i.e.,
1012 when -std is passed to GCC).
041308d0 1013
466b4cf2 1014** lalr1.java: several fixes:
041308d0 1015
e4ab1254
AD
1016 The Java parser no longer throws ArrayIndexOutOfBoundsException if the
1017 first token leads to a syntax error. Some minor clean ups.
041308d0 1018
22172d47 1019** Changes for C++:
ef51bfa7 1020
22172d47 1021*** C++11 compatibility:
ef51bfa7 1022
e4ab1254
AD
1023 C and C++ parsers use "nullptr" instead of "0" when __cplusplus is 201103L
1024 or higher.
936c88d1 1025
22172d47
AD
1026*** Header guards
1027
1028 The header files such as "parser.hh", "location.hh", etc. used a constant
1029 name for preprocessor guards, for instance:
1030
e29f0771
AD
1031 #ifndef BISON_LOCATION_HH
1032 # define BISON_LOCATION_HH
1033 ...
1034 #endif // !BISON_LOCATION_HH
22172d47
AD
1035
1036 The inclusion guard is now computed from "PREFIX/FILE-NAME", where lower
1037 case characters are converted to upper case, and series of
1038 non-alphanumerical characters are converted to an underscore.
1039
1040 With "bison -o lang++/parser.cc", "location.hh" would now include:
1041
e29f0771
AD
1042 #ifndef YY_LANG_LOCATION_HH
1043 # define YY_LANG_LOCATION_HH
1044 ...
1045 #endif // !YY_LANG_LOCATION_HH
22172d47
AD
1046
1047*** C++ locations:
936c88d1 1048
e4ab1254
AD
1049 The position and location constructors (and their initialize methods)
1050 accept new arguments for line and column. Several issues in the
1051 documentation were fixed.
936c88d1 1052
466b4cf2
AD
1053** liby is no longer asking for "rpl_fprintf" on some platforms.
1054
7e508a2b
AD
1055** Changes in the manual:
1056
1057*** %printer is documented
1058
e4ab1254
AD
1059 The "%printer" directive, supported since at least Bison 1.50, is finally
1060 documented. The "mfcalc" example is extended to demonstrate it.
7e508a2b 1061
e4ab1254
AD
1062 For consistency with the C skeletons, the C++ parsers now also support
1063 "yyoutput" (as an alias to "debug_stream ()").
7e508a2b
AD
1064
1065*** Several improvements have been made:
466b4cf2 1066
e4ab1254
AD
1067 The layout for grammar excerpts was changed to a more compact scheme.
1068 Named references are motivated. The description of the automaton
1069 description file (*.output) is updated to the current format. Incorrect
1070 index entries were fixed. Some other errors were fixed.
466b4cf2 1071
86b08b49
AD
1072** Building bison:
1073
1074*** Conflicting prototypes with recent/modified Flex.
466b4cf2 1075
e4ab1254
AD
1076 Fixed build problems with the current, unreleased, version of Flex, and
1077 some modified versions of 2.5.35, which have modified function prototypes.
466b4cf2 1078
8ef26c2a
AD
1079*** Warnings during the build procedure have been eliminated.
1080
1081*** Several portability problems in the test suite have been fixed:
466b4cf2 1082
e4ab1254
AD
1083 This includes warnings with some compilers, unexpected behavior of tools
1084 such as diff, warning messages from the test suite itself, etc.
466b4cf2 1085
91aadcc7 1086*** The install-pdf target works properly:
8ef26c2a 1087
e4ab1254
AD
1088 Running "make install-pdf" (or -dvi, -html, -info, and -ps) no longer
1089 halts in the middle of its course.
8ef26c2a 1090
28801043 1091* Changes in version 2.5 (2011-05-14):
50cca368 1092
82f3355e
JD
1093** Grammar symbol names can now contain non-initial dashes:
1094
1095 Consistently with directives (such as %error-verbose) and with
1096 %define variables (e.g. push-pull), grammar symbol names may contain
1097 dashes in any position except the beginning. This is a GNU
1098 extension over POSIX Yacc. Thus, use of this extension is reported
1099 by -Wyacc and rejected in Yacc mode (--yacc).
1100
f1b238df 1101** Named references:
66381412
AR
1102
1103 Historically, Yacc and Bison have supported positional references
1104 ($n, $$) to allow access to symbol values from inside of semantic
1105 actions code.
1106
1107 Starting from this version, Bison can also accept named references.
1108 When no ambiguity is possible, original symbol names may be used
1109 as named references:
1110
4b568fc0 1111 if_stmt : "if" cond_expr "then" then_stmt ';'
66381412
AR
1112 { $if_stmt = mk_if_stmt($cond_expr, $then_stmt); }
1113
1114 In the more common case, explicit names may be declared:
1115
4b568fc0 1116 stmt[res] : "if" expr[cond] "then" stmt[then] "else" stmt[else] ';'
66381412
AR
1117 { $res = mk_if_stmt($cond, $then, $else); }
1118
5b1ff423 1119 Location information is also accessible using @name syntax. When
66381412
AR
1120 accessing symbol names containing dots or dashes, explicit bracketing
1121 ($[sym.1]) must be used.
1122
5b1ff423 1123 These features are experimental in this version. More user feedback
66381412 1124 will help to stabilize them.
2bd435c3 1125 Contributed by Alex Rozenman.
66381412 1126
f1b238df 1127** IELR(1) and canonical LR(1):
eb45ef3b
JD
1128
1129 IELR(1) is a minimal LR(1) parser table generation algorithm. That
1130 is, given any context-free grammar, IELR(1) generates parser tables
7262f54f 1131 with the full language-recognition power of canonical LR(1) but with
f1b238df
JD
1132 nearly the same number of parser states as LALR(1). This reduction
1133 in parser states is often an order of magnitude. More importantly,
eb45ef3b
JD
1134 because canonical LR(1)'s extra parser states may contain duplicate
1135 conflicts in the case of non-LR(1) grammars, the number of conflicts
1136 for IELR(1) is often an order of magnitude less as well. This can
1137 significantly reduce the complexity of developing of a grammar.
1138
1139 Bison can now generate IELR(1) and canonical LR(1) parser tables in
1140 place of its traditional LALR(1) parser tables, which remain the
1141 default. You can specify the type of parser tables in the grammar
1142 file with these directives:
1143
cf499cff
JD
1144 %define lr.type lalr
1145 %define lr.type ielr
1146 %define lr.type canonical-lr
eb45ef3b 1147
7fceb615 1148 The default-reduction optimization in the parser tables can also be
e4ab1254
AD
1149 adjusted using "%define lr.default-reductions". For details on both
1150 of these features, see the new section "Tuning LR" in the Bison
7fceb615 1151 manual.
eb45ef3b
JD
1152
1153 These features are experimental. More user feedback will help to
1154 stabilize them.
1155
8458a411
AD
1156** LAC (Lookahead Correction) for syntax error handling
1157
1158 Contributed by Joel E. Denny.
fcf834f9
JD
1159
1160 Canonical LR, IELR, and LALR can suffer from a couple of problems
1161 upon encountering a syntax error. First, the parser might perform
1162 additional parser stack reductions before discovering the syntax
7fceb615 1163 error. Such reductions can perform user semantic actions that are
fcf834f9
JD
1164 unexpected because they are based on an invalid token, and they
1165 cause error recovery to begin in a different syntactic context than
1166 the one in which the invalid token was encountered. Second, when
7fceb615 1167 verbose error messages are enabled (with %error-verbose or the
e4ab1254 1168 obsolete "#define YYERROR_VERBOSE"), the expected token list in the
7fceb615
JD
1169 syntax error message can both contain invalid tokens and omit valid
1170 tokens.
fcf834f9
JD
1171
1172 The culprits for the above problems are %nonassoc, default
1173 reductions in inconsistent states, and parser state merging. Thus,
1174 IELR and LALR suffer the most. Canonical LR can suffer only if
1175 %nonassoc is used or if default reductions are enabled for
1176 inconsistent states.
1177
7fceb615
JD
1178 LAC is a new mechanism within the parsing algorithm that solves
1179 these problems for canonical LR, IELR, and LALR without sacrificing
1180 %nonassoc, default reductions, or state merging. When LAC is in
1181 use, canonical LR and IELR behave almost exactly the same for both
1182 syntactically acceptable and syntactically unacceptable input.
fcf834f9
JD
1183 While LALR still does not support the full language-recognition
1184 power of canonical LR and IELR, LAC at least enables LALR's syntax
1185 error handling to correctly reflect LALR's language-recognition
1186 power.
1187
1188 Currently, LAC is only supported for deterministic parsers in C.
1189 You can enable LAC with the following directive:
1190
1191 %define parse.lac full
1192
e4ab1254 1193 See the new section "LAC" in the Bison manual for additional
7fceb615 1194 details including a few caveats.
fcf834f9
JD
1195
1196 LAC is an experimental feature. More user feedback will help to
1197 stabilize it.
1198
d397d9f0 1199** %define improvements:
cf499cff 1200
f1b238df 1201*** Can now be invoked via the command line:
50cca368 1202
de5ab940 1203 Each of these command-line options
50cca368 1204
de5ab940
JD
1205 -D NAME[=VALUE]
1206 --define=NAME[=VALUE]
1207
1208 -F NAME[=VALUE]
1209 --force-define=NAME[=VALUE]
50cca368
JD
1210
1211 is equivalent to this grammar file declaration
1212
de5ab940 1213 %define NAME ["VALUE"]
50cca368 1214
de5ab940
JD
1215 except that the manner in which Bison processes multiple definitions
1216 for the same NAME differs. Most importantly, -F and --force-define
1217 quietly override %define, but -D and --define do not. For further
e4ab1254 1218 details, see the section "Bison Options" in the Bison manual.
50cca368 1219
f1b238df 1220*** Variables renamed:
67212941
JD
1221
1222 The following %define variables
1223
1224 api.push_pull
1225 lr.keep_unreachable_states
1226
1227 have been renamed to
1228
1229 api.push-pull
1230 lr.keep-unreachable-states
1231
1232 The old names are now deprecated but will be maintained indefinitely
1233 for backward compatibility.
1234
7262f54f 1235*** Values no longer need to be quoted in the grammar file:
cf499cff
JD
1236
1237 If a %define value is an identifier, it no longer needs to be placed
1238 within quotations marks. For example,
1239
1240 %define api.push-pull "push"
1241
1242 can be rewritten as
1243
1244 %define api.push-pull push
1245
d397d9f0 1246*** Unrecognized variables are now errors not warnings.
cdf3f113 1247
d397d9f0
JD
1248*** Multiple invocations for any variable is now an error not a warning.
1249
1250** Unrecognized %code qualifiers are now errors not warnings.
1251
1252** Character literals not of length one:
1253
1254 Previously, Bison quietly converted all character literals to length
1255 one. For example, without warning, Bison interpreted the operators in
1256 the following grammar to be the same token:
1257
1258 exp: exp '++'
1259 | exp '+' exp
1260 ;
1261
1262 Bison now warns when a character literal is not of length one. In
1263 some future release, Bison will start reporting an error instead.
1264
1265** Destructor calls fixed for lookaheads altered in semantic actions:
1266
1267 Previously for deterministic parsers in C, if a user semantic action
1268 altered yychar, the parser in some cases used the old yychar value to
1269 determine which destructor to call for the lookahead upon a syntax
1270 error or upon parser return. This bug has been fixed.
1271
1272** C++ parsers use YYRHSLOC:
1273
1274 Similarly to the C parsers, the C++ parsers now define the YYRHSLOC
1275 macro and use it in the default YYLLOC_DEFAULT. You are encouraged
e4ab1254
AD
1276 to use it. If, for instance, your location structure has "first"
1277 and "last" members, instead of
d397d9f0 1278
e29f0771
AD
1279 # define YYLLOC_DEFAULT(Current, Rhs, N) \
1280 do \
1281 if (N) \
1282 { \
1283 (Current).first = (Rhs)[1].location.first; \
1284 (Current).last = (Rhs)[N].location.last; \
1285 } \
1286 else \
1287 { \
1288 (Current).first = (Current).last = (Rhs)[0].location.last; \
1289 } \
1290 while (false)
d397d9f0
JD
1291
1292 use:
1293
e29f0771
AD
1294 # define YYLLOC_DEFAULT(Current, Rhs, N) \
1295 do \
1296 if (N) \
1297 { \
1298 (Current).first = YYRHSLOC (Rhs, 1).first; \
1299 (Current).last = YYRHSLOC (Rhs, N).last; \
1300 } \
1301 else \
1302 { \
1303 (Current).first = (Current).last = YYRHSLOC (Rhs, 0).last; \
1304 } \
1305 while (false)
d397d9f0
JD
1306
1307** YYLLOC_DEFAULT in C++:
1308
1309 The default implementation of YYLLOC_DEFAULT used to be issued in
1310 the header file. It is now output in the implementation file, after
1311 the user %code sections so that its #ifndef guard does not try to
1312 override the user's YYLLOC_DEFAULT if provided.
cdf3f113 1313
f1b238df 1314** YYFAIL now produces warnings and Java parsers no longer implement it:
4395a9ff
JD
1315
1316 YYFAIL has existed for many years as an undocumented feature of
1317 deterministic parsers in C generated by Bison. More recently, it was
1318 a documented feature of Bison's experimental Java parsers. As
1319 promised in Bison 2.4.2's NEWS entry, any appearance of YYFAIL in a
1320 semantic action now produces a deprecation warning, and Java parsers
1321 no longer implement YYFAIL at all. For further details, including a
1322 discussion of how to suppress C preprocessor warnings about YYFAIL
1323 being unused, see the Bison 2.4.2 NEWS entry.
1324
f1b238df 1325** Temporary hack for adding a semicolon to the user action:
197b82ba
JD
1326
1327 Previously, Bison appended a semicolon to every user action for
1328 reductions when the output language defaulted to C (specifically, when
1329 neither %yacc, %language, %skeleton, or equivalent command-line
1330 options were specified). This allowed actions such as
1331
1332 exp: exp "+" exp { $$ = $1 + $3 };
1333
1334 instead of
1335
1336 exp: exp "+" exp { $$ = $1 + $3; };
1337
1338 As a first step in removing this misfeature, Bison now issues a
1339 warning when it appends a semicolon. Moreover, in cases where Bison
1340 cannot easily determine whether a semicolon is needed (for example, an
1341 action ending with a cpp directive or a braced compound initializer),
1342 it no longer appends one. Thus, the C compiler might now complain
1343 about a missing semicolon where it did not before. Future releases of
1344 Bison will cease to append semicolons entirely.
1345
d2060f06
JD
1346** Verbose syntax error message fixes:
1347
e4ab1254 1348 When %error-verbose or the obsolete "#define YYERROR_VERBOSE" is
7fceb615
JD
1349 specified, syntax error messages produced by the generated parser
1350 include the unexpected token as well as a list of expected tokens.
1351 The effect of %nonassoc on these verbose messages has been corrected
1352 in two ways, but a more complete fix requires LAC, described above:
d2060f06
JD
1353
1354*** When %nonassoc is used, there can exist parser states that accept no
1355 tokens, and so the parser does not always require a lookahead token
1356 in order to detect a syntax error. Because no unexpected token or
1357 expected tokens can then be reported, the verbose syntax error
1358 message described above is suppressed, and the parser instead
e4ab1254 1359 reports the simpler message, "syntax error". Previously, this
d2060f06
JD
1360 suppression was sometimes erroneously triggered by %nonassoc when a
1361 lookahead was actually required. Now verbose messages are
1362 suppressed only when all previous lookaheads have already been
1363 shifted or discarded.
1364
1365*** Previously, the list of expected tokens erroneously included tokens
1366 that would actually induce a syntax error because conflicts for them
1367 were resolved with %nonassoc in the current parser state. Such
1368 tokens are now properly omitted from the list.
1369
1370*** Expected token lists are still often wrong due to state merging
fcf834f9
JD
1371 (from LALR or IELR) and default reductions, which can both add
1372 invalid tokens and subtract valid tokens. Canonical LR almost
1373 completely fixes this problem by eliminating state merging and
1374 default reductions. However, there is one minor problem left even
1375 when using canonical LR and even after the fixes above. That is,
1376 if the resolution of a conflict with %nonassoc appears in a later
1377 parser state than the one at which some syntax error is
1378 discovered, the conflicted token is still erroneously included in
1379 the expected token list. Bison's new LAC implementation,
1380 described above, eliminates this problem and the need for
1381 canonical LR. However, LAC is still experimental and is disabled
1382 by default.
53f036ce 1383
1a33f4f6
JD
1384** Java skeleton fixes:
1385
1386*** A location handling bug has been fixed.
1387
1388*** The top element of each of the value stack and location stack is now
1389 cleared when popped so that it can be garbage collected.
6771a463 1390
02803d55
JD
1391*** Parser traces now print the top element of the stack.
1392
86408959
JD
1393** -W/--warnings fixes:
1394
e4ab1254 1395*** Bison now properly recognizes the "no-" versions of categories:
86408959
JD
1396
1397 For example, given the following command line, Bison now enables all
1398 warnings except warnings for incompatibilities with POSIX Yacc:
1399
1400 bison -Wall,no-yacc gram.y
1401
786743d5
JD
1402*** Bison now treats S/R and R/R conflicts like other warnings:
1403
1404 Previously, conflict reports were independent of Bison's normal
1405 warning system. Now, Bison recognizes the warning categories
e4ab1254 1406 "conflicts-sr" and "conflicts-rr". This change has important
786743d5
JD
1407 consequences for the -W and --warnings command-line options. For
1408 example:
1409
1410 bison -Wno-conflicts-sr gram.y # S/R conflicts not reported
1411 bison -Wno-conflicts-rr gram.y # R/R conflicts not reported
1412 bison -Wnone gram.y # no conflicts are reported
1413 bison -Werror gram.y # any conflict is an error
1414
1415 However, as before, if the %expect or %expect-rr directive is
1416 specified, an unexpected number of conflicts is an error, and an
1417 expected number of conflicts is not reported, so -W and --warning
1418 then have no effect on the conflict report.
1419
e4ab1254 1420*** The "none" category no longer disables a preceding "error":
bf0e44e8
JD
1421
1422 For example, for the following command line, Bison now reports
1423 errors instead of warnings for incompatibilities with POSIX Yacc:
1424
1425 bison -Werror,none,yacc gram.y
1426
e4ab1254 1427*** The "none" category now disables all Bison warnings:
c39014ae 1428
e4ab1254 1429 Previously, the "none" category disabled only Bison warnings for
c39014ae
JD
1430 which there existed a specific -W/--warning category. However,
1431 given the following command line, Bison is now guaranteed to
1432 suppress all warnings:
1433
1434 bison -Wnone gram.y
1435
1f36f544
JD
1436** Precedence directives can now assign token number 0:
1437
1438 Since Bison 2.3b, which restored the ability of precedence
1439 directives to assign token numbers, doing so for token number 0 has
1440 produced an assertion failure. For example:
1441
1442 %left END 0
1443
1444 This bug has been fixed.
1445
64877e5e 1446* Changes in version 2.4.3 (2010-08-05):
8b9e021f 1447
2bfcac9a
JD
1448** Bison now obeys -Werror and --warnings=error for warnings about
1449 grammar rules that are useless in the parser due to conflicts.
1450
8b9e021f
JD
1451** Problems with spawning M4 on at least FreeBSD 8 and FreeBSD 9 have
1452 been fixed.
1453
4ad3921d
JD
1454** Failures in the test suite for GCC 4.5 have been fixed.
1455
06cb07d5
JD
1456** Failures in the test suite for some versions of Sun Studio C++ have
1457 been fixed.
1458
9b5049bd
JD
1459** Contrary to Bison 2.4.2's NEWS entry, it has been decided that
1460 warnings about undefined %prec identifiers will not be converted to
1461 errors in Bison 2.5. They will remain warnings, which should be
1462 sufficient for POSIX while avoiding backward compatibility issues.
1463
93d7dde9
JD
1464** Minor documentation fixes.
1465
e19a049c 1466* Changes in version 2.4.2 (2010-03-20):
74553c98 1467
f39ab286
JD
1468** Some portability problems that resulted in failures and livelocks
1469 in the test suite on some versions of at least Solaris, AIX, HP-UX,
e19a049c
JD
1470 RHEL4, and Tru64 have been addressed. As a result, fatal Bison
1471 errors should no longer cause M4 to report a broken pipe on the
f39ab286
JD
1472 affected platforms.
1473
e4ab1254 1474** "%prec IDENTIFIER" requires IDENTIFIER to be defined separately.
8bb3a2e7
JD
1475
1476 POSIX specifies that an error be reported for any identifier that does
1477 not appear on the LHS of a grammar rule and that is not defined by
1478 %token, %left, %right, or %nonassoc. Bison 2.3b and later lost this
1479 error report for the case when an identifier appears only after a
1480 %prec directive. It is now restored. However, for backward
1481 compatibility with recent Bison releases, it is only a warning for
1482 now. In Bison 2.5 and later, it will return to being an error.
9b5049bd
JD
1483 [Between the 2.4.2 and 2.4.3 releases, it was decided that this
1484 warning will not be converted to an error in Bison 2.5.]
8bb3a2e7 1485
d8911864
EB
1486** Detection of GNU M4 1.4.6 or newer during configure is improved.
1487
a603c6e0
JD
1488** Warnings from gcc's -Wundef option about undefined YYENABLE_NLS,
1489 YYLTYPE_IS_TRIVIAL, and __STRICT_ANSI__ in C/C++ parsers are now
1490 avoided.
c938d650 1491
98a345a2
JD
1492** %code is now a permanent feature.
1493
1494 A traditional Yacc prologue directive is written in the form:
1495
1496 %{CODE%}
1497
1498 To provide a more flexible alternative, Bison 2.3b introduced the
1499 %code directive with the following forms for C/C++:
1500
1501 %code {CODE}
1502 %code requires {CODE}
1503 %code provides {CODE}
1504 %code top {CODE}
1505
1506 These forms are now considered permanent features of Bison. See the
1507 %code entries in the section "Bison Declaration Summary" in the Bison
1508 manual for a summary of their functionality. See the section
1509 "Prologue Alternatives" for a detailed discussion including the
1510 advantages of %code over the traditional Yacc prologue directive.
1511
1512 Bison's Java feature as a whole including its current usage of %code
1513 is still considered experimental.
1514
1625df5b
JD
1515** YYFAIL is deprecated and will eventually be removed.
1516
1517 YYFAIL has existed for many years as an undocumented feature of
1518 deterministic parsers in C generated by Bison. Previously, it was
1519 documented for Bison's experimental Java parsers. YYFAIL is no longer
1520 documented for Java parsers and is formally deprecated in both cases.
1521 Users are strongly encouraged to migrate to YYERROR, which is
1522 specified by POSIX.
1523
1524 Like YYERROR, you can invoke YYFAIL from a semantic action in order to
1525 induce a syntax error. The most obvious difference from YYERROR is
1526 that YYFAIL will automatically invoke yyerror to report the syntax
1527 error so that you don't have to. However, there are several other
1528 subtle differences between YYERROR and YYFAIL, and YYFAIL suffers from
e4ab1254 1529 inherent flaws when %error-verbose or "#define YYERROR_VERBOSE" is
1625df5b
JD
1530 used. For a more detailed discussion, see:
1531
1532 http://lists.gnu.org/archive/html/bison-patches/2009-12/msg00024.html
1533
1534 The upcoming Bison 2.5 will remove YYFAIL from Java parsers, but
1535 deterministic parsers in C will continue to implement it. However,
1536 because YYFAIL is already flawed, it seems futile to try to make new
1537 Bison features compatible with it. Thus, during parser generation,
1538 Bison 2.5 will produce a warning whenever it discovers YYFAIL in a
1539 rule action. In a later release, YYFAIL will be disabled for
e4ab1254 1540 %error-verbose and "#define YYERROR_VERBOSE". Eventually, YYFAIL will
1625df5b
JD
1541 be removed altogether.
1542
1543 There exists at least one case where Bison 2.5's YYFAIL warning will
1544 be a false positive. Some projects add phony uses of YYFAIL and other
1545 Bison-defined macros for the sole purpose of suppressing C
1546 preprocessor warnings (from GCC cpp's -Wunused-macros, for example).
1547 To avoid Bison's future warning, such YYFAIL uses can be moved to the
e4ab1254 1548 epilogue (that is, after the second "%%") in the Bison input file. In
1625df5b
JD
1549 this release (2.4.2), Bison already generates its own code to suppress
1550 C preprocessor warnings for YYFAIL, so projects can remove their own
1551 phony uses of YYFAIL if compatibility with Bison releases prior to
1552 2.4.2 is not necessary.
1553
2755de8f
AD
1554** Internationalization.
1555
1556 Fix a regression introduced in Bison 2.4: Under some circumstances,
1557 message translations were not installed although supported by the
1558 host system.
1559
74553c98 1560* Changes in version 2.4.1 (2008-12-11):
c9ba9e59 1561
0ea583d2
AD
1562** In the GLR defines file, unexpanded M4 macros in the yylval and yylloc
1563 declarations have been fixed.
1979121c 1564
0ea583d2
AD
1565** Temporary hack for adding a semicolon to the user action.
1566
1567 Bison used to prepend a trailing semicolon at the end of the user
1568 action for reductions. This allowed actions such as
1569
1570 exp: exp "+" exp { $$ = $1 + $3 };
1571
1572 instead of
1573
1574 exp: exp "+" exp { $$ = $1 + $3; };
1575
e4ab1254 1576 Some grammars still depend on this "feature". Bison 2.4.1 restores
0ea583d2
AD
1577 the previous behavior in the case of C output (specifically, when
1578 neither %language or %skeleton or equivalent command-line options
1579 are used) to leave more time for grammars depending on the old
1580 behavior to be adjusted. Future releases of Bison will disable this
1581 feature.
1582
1583** A few minor improvements to the Bison manual.
c9ba9e59 1584
402b123d 1585* Changes in version 2.4 (2008-11-02):
7bd1665a 1586
402b123d 1587** %language is an experimental feature.
ed4d67dc
JD
1588
1589 We first introduced this feature in test release 2.3b as a cleaner
1590 alternative to %skeleton. Since then, we have discussed the possibility of
1591 modifying its effect on Bison's output file names. Thus, in this release,
1592 we consider %language to be an experimental feature that will likely evolve
1593 in future releases.
7bd1665a 1594
402b123d 1595** Forward compatibility with GNU M4 has been improved.
241fda7a 1596
402b123d 1597** Several bugs in the C++ skeleton and the experimental Java skeleton have been
241fda7a
JD
1598 fixed.
1599
402b123d 1600* Changes in version 2.3b (2008-05-27):
35fe0834 1601
402b123d 1602** The quotes around NAME that used to be required in the following directive
d9df47b6
JD
1603 are now deprecated:
1604
1605 %define NAME "VALUE"
1606
e4ab1254 1607** The directive "%pure-parser" is now deprecated in favor of:
d9df47b6
JD
1608
1609 %define api.pure
1610
1611 which has the same effect except that Bison is more careful to warn about
1612 unreasonable usage in the latter case.
1613
402b123d 1614** Push Parsing
c373bf8b
JD
1615
1616 Bison can now generate an LALR(1) parser in C with a push interface. That
e4ab1254
AD
1617 is, instead of invoking "yyparse", which pulls tokens from "yylex", you can
1618 push one token at a time to the parser using "yypush_parse", which will
c373bf8b
JD
1619 return to the caller after processing each token. By default, the push
1620 interface is disabled. Either of the following directives will enable it:
1621
1622 %define api.push_pull "push" // Just push; does not require yylex.
1623 %define api.push_pull "both" // Push and pull; requires yylex.
1624
e4ab1254 1625 See the new section "A Push Parser" in the Bison manual for details.
c373bf8b 1626
59da312b
JD
1627 The current push parsing interface is experimental and may evolve. More user
1628 feedback will help to stabilize it.
1629
402b123d 1630** The -g and --graph options now output graphs in Graphviz DOT format,
8e55b3aa
JD
1631 not VCG format. Like --graph, -g now also takes an optional FILE argument
1632 and thus cannot be bundled with other short options.
c373bf8b 1633
402b123d 1634** Java
59da312b
JD
1635
1636 Bison can now generate an LALR(1) parser in Java. The skeleton is
e4ab1254 1637 "data/lalr1.java". Consider using the new %language directive instead of
59da312b
JD
1638 %skeleton to select it.
1639
e4ab1254 1640 See the new section "Java Parsers" in the Bison manual for details.
59da312b
JD
1641
1642 The current Java interface is experimental and may evolve. More user
1643 feedback will help to stabilize it.
2bd435c3 1644 Contributed by Paolo Bonzini.
59da312b 1645
402b123d 1646** %language
59da312b
JD
1647
1648 This new directive specifies the programming language of the generated
d43f77e7
PB
1649 parser, which can be C (the default), C++, or Java. Besides the skeleton
1650 that Bison uses, the directive affects the names of the generated files if
1651 the grammar file's name ends in ".y".
59da312b 1652
402b123d 1653** XML Automaton Report
59da312b
JD
1654
1655 Bison can now generate an XML report of the LALR(1) automaton using the new
e4ab1254 1656 "--xml" option. The current XML schema is experimental and may evolve. More
59da312b 1657 user feedback will help to stabilize it.
2bd435c3 1658 Contributed by Wojciech Polak.
c373bf8b 1659
402b123d 1660** The grammar file may now specify the name of the parser header file using
c373bf8b
JD
1661 %defines. For example:
1662
1663 %defines "parser.h"
1664
402b123d 1665** When reporting useless rules, useless nonterminals, and unused terminals,
d80fb37a
JD
1666 Bison now employs the terms "useless in grammar" instead of "useless",
1667 "useless in parser" instead of "never reduced", and "unused in grammar"
1668 instead of "unused".
cff03fb2 1669
402b123d 1670** Unreachable State Removal
c373bf8b
JD
1671
1672 Previously, Bison sometimes generated parser tables containing unreachable
31984206
JD
1673 states. A state can become unreachable during conflict resolution if Bison
1674 disables a shift action leading to it from a predecessor state. Bison now:
75ad86ee
JD
1675
1676 1. Removes unreachable states.
1677
1678 2. Does not report any conflicts that appeared in unreachable states.
1679 WARNING: As a result, you may need to update %expect and %expect-rr
1680 directives in existing grammar files.
1681
1682 3. For any rule used only in such states, Bison now reports the rule as
cff03fb2 1683 "useless in parser due to conflicts".
75ad86ee 1684
31984206
JD
1685 This feature can be disabled with the following directive:
1686
1687 %define lr.keep_unreachable_states
1688
e4ab1254 1689 See the %define entry in the "Bison Declaration Summary" in the Bison manual
31984206
JD
1690 for further discussion.
1691
e4ab1254 1692** Lookahead Set Correction in the ".output" Report
b1cc23c4 1693
e4ab1254
AD
1694 When instructed to generate a ".output" file including lookahead sets
1695 (using "--report=lookahead", for example), Bison now prints each reduction's
88c78747
JD
1696 lookahead set only next to the associated state's one item that (1) is
1697 associated with the same rule as the reduction and (2) has its dot at the end
1698 of its RHS. Previously, Bison also erroneously printed the lookahead set
1699 next to all of the state's other items associated with the same rule. This
e4ab1254 1700 bug affected only the ".output" file and not the generated parser source
88c78747
JD
1701 code.
1702
e4ab1254 1703** --report-file=FILE is a new option to override the default ".output" file
59da312b 1704 name.
1bb2bd75 1705
e4ab1254 1706** The "=" that used to be required in the following directives is now
02975b9a
JD
1707 deprecated:
1708
1709 %file-prefix "parser"
1710 %name-prefix "c_"
1711 %output "parser.c"
1712
e4ab1254 1713** An Alternative to "%{...%}" -- "%code QUALIFIER {CODE}"
c373bf8b
JD
1714
1715 Bison 2.3a provided a new set of directives as a more flexible alternative to
8e0a5e9e
JD
1716 the traditional Yacc prologue blocks. Those have now been consolidated into
1717 a single %code directive with an optional qualifier field, which identifies
1718 the purpose of the code and thus the location(s) where Bison should generate
1719 it:
1720
e4ab1254
AD
1721 1. "%code {CODE}" replaces "%after-header {CODE}"
1722 2. "%code requires {CODE}" replaces "%start-header {CODE}"
1723 3. "%code provides {CODE}" replaces "%end-header {CODE}"
1724 4. "%code top {CODE}" replaces "%before-header {CODE}"
8e0a5e9e 1725
e4ab1254
AD
1726 See the %code entries in section "Bison Declaration Summary" in the Bison
1727 manual for a summary of the new functionality. See the new section "Prologue
1728 Alternatives" for a detailed discussion including the advantages of %code
8e0a5e9e
JD
1729 over the traditional Yacc prologues.
1730
1731 The prologue alternatives are experimental. More user feedback will help to
1732 determine whether they should become permanent features.
1733
402b123d 1734** Revised warning: unset or unused mid-rule values
17bd8a73
JD
1735
1736 Since Bison 2.2, Bison has warned about mid-rule values that are set but not
1737 used within any of the actions of the parent rule. For example, Bison warns
1738 about unused $2 in:
1739
1740 exp: '1' { $$ = 1; } '+' exp { $$ = $1 + $4; };
1741
1742 Now, Bison also warns about mid-rule values that are used but not set. For
1743 example, Bison warns about unset $$ in the mid-rule action in:
1744
1745 exp: '1' { $1 = 1; } '+' exp { $$ = $2 + $4; };
1746
1747 However, Bison now disables both of these warnings by default since they
1748 sometimes prove to be false alarms in existing grammars employing the Yacc
1749 constructs $0 or $-N (where N is some positive integer).
1750
e4ab1254
AD
1751 To enable these warnings, specify the option "--warnings=midrule-values" or
1752 "-W", which is a synonym for "--warnings=all".
17bd8a73 1753
e4ab1254 1754** Default %destructor or %printer with "<*>" or "<>"
c373bf8b
JD
1755
1756 Bison now recognizes two separate kinds of default %destructor's and
12e35840
JD
1757 %printer's:
1758
e4ab1254 1759 1. Place "<*>" in a %destructor/%printer symbol list to define a default
12e35840
JD
1760 %destructor/%printer for all grammar symbols for which you have formally
1761 declared semantic type tags.
1762
e4ab1254 1763 2. Place "<>" in a %destructor/%printer symbol list to define a default
12e35840
JD
1764 %destructor/%printer for all grammar symbols without declared semantic
1765 type tags.
1766
e4ab1254
AD
1767 Bison no longer supports the "%symbol-default" notation from Bison 2.3a.
1768 "<*>" and "<>" combined achieve the same effect with one exception: Bison no
12e35840
JD
1769 longer applies any %destructor to a mid-rule value if that mid-rule value is
1770 not actually ever referenced using either $$ or $n in a semantic action.
1771
85894313
JD
1772 The default %destructor's and %printer's are experimental. More user
1773 feedback will help to determine whether they should become permanent
1774 features.
1775
e4ab1254 1776 See the section "Freeing Discarded Symbols" in the Bison manual for further
12e35840
JD
1777 details.
1778
402b123d 1779** %left, %right, and %nonassoc can now declare token numbers. This is required
e4ab1254 1780 by POSIX. However, see the end of section "Operator Precedence" in the Bison
ab7f29f8
JD
1781 manual for a caveat concerning the treatment of literal strings.
1782
402b123d 1783** The nonfunctional --no-parser, -n, and %no-parser options have been
b1cc23c4
JD
1784 completely removed from Bison.
1785
402b123d 1786* Changes in version 2.3a, 2006-09-13:
742e4900 1787
402b123d 1788** Instead of %union, you can define and use your own union type
ddc8ede1
PE
1789 YYSTYPE if your grammar contains at least one <type> tag.
1790 Your YYSTYPE need not be a macro; it can be a typedef.
1791 This change is for compatibility with other Yacc implementations,
1792 and is required by POSIX.
1793
402b123d 1794** Locations columns and lines start at 1.
cd48d21d
AD
1795 In accordance with the GNU Coding Standards and Emacs.
1796
402b123d 1797** You may now declare per-type and default %destructor's and %printer's:
ec5479ce
JD
1798
1799 For example:
1800
b2a0b7ca
JD
1801 %union { char *string; }
1802 %token <string> STRING1
1803 %token <string> STRING2
1804 %type <string> string1
1805 %type <string> string2
1806 %union { char character; }
1807 %token <character> CHR
1808 %type <character> chr
1809 %destructor { free ($$); } %symbol-default
1810 %destructor { free ($$); printf ("%d", @$.first_line); } STRING1 string1
1811 %destructor { } <character>
1812
1813 guarantees that, when the parser discards any user-defined symbol that has a
e4ab1254
AD
1814 semantic type tag other than "<character>", it passes its semantic value to
1815 "free". However, when the parser discards a "STRING1" or a "string1", it
1816 also prints its line number to "stdout". It performs only the second
1817 "%destructor" in this case, so it invokes "free" only once.
ec5479ce 1818
85894313
JD
1819 [Although we failed to mention this here in the 2.3a release, the default
1820 %destructor's and %printer's were experimental, and they were rewritten in
1821 future versions.]
1822
e4ab1254
AD
1823** Except for LALR(1) parsers in C with POSIX Yacc emulation enabled (with "-y",
1824 "--yacc", or "%yacc"), Bison no longer generates #define statements for
b931235e
JD
1825 associating token numbers with token names. Removing the #define statements
1826 helps to sanitize the global namespace during preprocessing, but POSIX Yacc
1827 requires them. Bison still generates an enum for token names in all cases.
1828
402b123d 1829** Handling of traditional Yacc prologue blocks is now more consistent but
34f98f46 1830 potentially incompatible with previous releases of Bison.
9bc0dd67
JD
1831
1832 As before, you declare prologue blocks in your grammar file with the
e4ab1254 1833 "%{ ... %}" syntax. To generate the pre-prologue, Bison concatenates all
34f98f46
JD
1834 prologue blocks that you've declared before the first %union. To generate
1835 the post-prologue, Bison concatenates all prologue blocks that you've
ddc8ede1 1836 declared after the first %union.
9bc0dd67 1837
34f98f46 1838 Previous releases of Bison inserted the pre-prologue into both the header
9bc0dd67
JD
1839 file and the code file in all cases except for LALR(1) parsers in C. In the
1840 latter case, Bison inserted it only into the code file. For parsers in C++,
1841 the point of insertion was before any token definitions (which associate
1842 token numbers with names). For parsers in C, the point of insertion was
1843 after the token definitions.
1844
1845 Now, Bison never inserts the pre-prologue into the header file. In the code
1846 file, it always inserts it before the token definitions.
1847
402b123d 1848** Bison now provides a more flexible alternative to the traditional Yacc
34f98f46
JD
1849 prologue blocks: %before-header, %start-header, %end-header, and
1850 %after-header.
1851
1852 For example, the following declaration order in the grammar file reflects the
1853 order in which Bison will output these code blocks. However, you are free to
1854 declare these code blocks in your grammar file in whatever order is most
1855 convenient for you:
1856
1857 %before-header {
1858 /* Bison treats this block like a pre-prologue block: it inserts it into
1859 * the code file before the contents of the header file. It does *not*
1860 * insert it into the header file. This is a good place to put
1861 * #include's that you want at the top of your code file. A common
e4ab1254 1862 * example is '#include "system.h"'. */
34f98f46
JD
1863 }
1864 %start-header {
1865 /* Bison inserts this block into both the header file and the code file.
1866 * In both files, the point of insertion is before any Bison-generated
1867 * token, semantic type, location type, and class definitions. This is a
1868 * good place to define %union dependencies, for example. */
9bc0dd67
JD
1869 }
1870 %union {
34f98f46
JD
1871 /* Unlike the traditional Yacc prologue blocks, the output order for the
1872 * new %*-header blocks is not affected by their declaration position
1873 * relative to any %union in the grammar file. */
9bc0dd67 1874 }
34f98f46
JD
1875 %end-header {
1876 /* Bison inserts this block into both the header file and the code file.
1877 * In both files, the point of insertion is after the Bison-generated
1878 * definitions. This is a good place to declare or define public
1879 * functions or data structures that depend on the Bison-generated
1880 * definitions. */
9bc0dd67 1881 }
34f98f46
JD
1882 %after-header {
1883 /* Bison treats this block like a post-prologue block: it inserts it into
1884 * the code file after the contents of the header file. It does *not*
1885 * insert it into the header file. This is a good place to declare or
1886 * define internal functions or data structures that depend on the
1887 * Bison-generated definitions. */
1888 }
1889
1890 If you have multiple occurrences of any one of the above declarations, Bison
1891 will concatenate the contents in declaration order.
9bc0dd67 1892
85894313
JD
1893 [Although we failed to mention this here in the 2.3a release, the prologue
1894 alternatives were experimental, and they were rewritten in future versions.]
1895
e4ab1254 1896** The option "--report=look-ahead" has been changed to "--report=lookahead".
9e6e7ed2
PE
1897 The old spelling still works, but is not documented and may be removed
1898 in a future release.
742e4900 1899
402b123d 1900* Changes in version 2.3, 2006-06-05:
4ad3ed84 1901
e4ab1254 1902** GLR grammars should now use "YYRECOVERING ()" instead of "YYRECOVERING",
4ad3ed84
PE
1903 for compatibility with LALR(1) grammars.
1904
402b123d 1905** It is now documented that any definition of YYSTYPE or YYLTYPE should
4ad3ed84
PE
1906 be to a type name that does not contain parentheses or brackets.
1907
402b123d 1908* Changes in version 2.2, 2006-05-19:
193d7c70 1909
402b123d 1910** The distribution terms for all Bison-generated parsers now permit
193d7c70
PE
1911 using the parsers in nonfree programs. Previously, this permission
1912 was granted only for Bison-generated LALR(1) parsers in C.
5f4236a0 1913
402b123d 1914** %name-prefix changes the namespace name in C++ outputs.
aa08666d 1915
402b123d 1916** The C++ parsers export their token_type.
5f4236a0 1917
402b123d 1918** Bison now allows multiple %union declarations, and concatenates
d6ca7905
PE
1919 their contents together.
1920
402b123d 1921** New warning: unused values
4d7bc38c
PE
1922 Right-hand side symbols whose values are not used are reported,
1923 if the symbols have destructors. For instance:
affac613 1924
8f3596a6 1925 exp: exp "?" exp ":" exp { $1 ? $1 : $3; }
e9690142
JD
1926 | exp "+" exp
1927 ;
affac613 1928
8f3596a6
AD
1929 will trigger a warning about $$ and $5 in the first rule, and $3 in
1930 the second ($1 is copied to $$ by the default rule). This example
4e26c69e 1931 most likely contains three errors, and could be rewritten as:
affac613 1932
4e26c69e 1933 exp: exp "?" exp ":" exp
e9690142
JD
1934 { $$ = $1 ? $3 : $5; free ($1 ? $5 : $3); free ($1); }
1935 | exp "+" exp
1936 { $$ = $1 ? $1 : $3; if ($1) free ($3); }
1937 ;
affac613 1938
4e26c69e
PE
1939 However, if the original actions were really intended, memory leaks
1940 and all, the warnings can be suppressed by letting Bison believe the
1941 values are used, e.g.:
721be13c 1942
8f3596a6 1943 exp: exp "?" exp ":" exp { $1 ? $1 : $3; (void) ($$, $5); }
e9690142
JD
1944 | exp "+" exp { $$ = $1; (void) $3; }
1945 ;
721be13c 1946
84866159
AD
1947 If there are mid-rule actions, the warning is issued if no action
1948 uses it. The following triggers no warning: $1 and $3 are used.
1949
1950 exp: exp { push ($1); } '+' exp { push ($3); sum (); };
1951
721be13c
PE
1952 The warning is intended to help catching lost values and memory leaks.
1953 If a value is ignored, its associated memory typically is not reclaimed.
affac613 1954
402b123d 1955** %destructor vs. YYABORT, YYACCEPT, and YYERROR.
9d9b8b70
PE
1956 Destructors are now called when user code invokes YYABORT, YYACCEPT,
1957 and YYERROR, for all objects on the stack, other than objects
1958 corresponding to the right-hand side of the current rule.
a85284cf 1959
402b123d 1960** %expect, %expect-rr
035aa4a0
PE
1961 Incorrect numbers of expected conflicts are now actual errors,
1962 instead of warnings.
1963
402b123d 1964** GLR, YACC parsers.
4e26c69e
PE
1965 The %parse-params are available in the destructors (and the
1966 experimental printers) as per the documentation.
4b367315 1967
e4ab1254 1968** Bison now warns if it finds a stray "$" or "@" in an action.
ad6a9b97 1969
402b123d 1970** %require "VERSION"
4e26c69e
PE
1971 This specifies that the grammar file depends on features implemented
1972 in Bison version VERSION or higher.
b50d2359 1973
402b123d 1974** lalr1.cc: The token and value types are now class members.
e14d0ab6
AD
1975 The tokens were defined as free form enums and cpp macros. YYSTYPE
1976 was defined as a free form union. They are now class members:
e4ab1254
AD
1977 tokens are enumerations of the "yy::parser::token" struct, and the
1978 semantic values have the "yy::parser::semantic_type" type.
fb9712a9
AD
1979
1980 If you do not want or can update to this scheme, the directive
e4ab1254 1981 '%define "global_tokens_and_yystype" "1"' triggers the global
b50d2359
AD
1982 definition of tokens and YYSTYPE. This change is suitable both
1983 for previous releases of Bison, and this one.
fb9712a9 1984
b50d2359 1985 If you wish to update, then make sure older version of Bison will
e4ab1254 1986 fail using '%require "2.2"'.
fb9712a9 1987
402b123d 1988** DJGPP support added.
193d7c70 1989\f
402b123d 1990* Changes in version 2.1, 2005-09-16:
1ce59070 1991
402b123d 1992** The C++ lalr1.cc skeleton supports %lex-param.
e14d0ab6 1993
402b123d 1994** Bison-generated parsers now support the translation of diagnostics like
baf785db
PE
1995 "syntax error" into languages other than English. The default
1996 language is still English. For details, please see the new
0410a6e0
PE
1997 Internationalization section of the Bison manual. Software
1998 distributors should also see the new PACKAGING file. Thanks to
1999 Bruno Haible for this new feature.
1ce59070 2000
402b123d 2001** Wording in the Bison-generated parsers has been changed slightly to
1a059451
PE
2002 simplify translation. In particular, the message "memory exhausted"
2003 has replaced "parser stack overflow", as the old message was not
2004 always accurate for modern Bison-generated parsers.
2005
402b123d 2006** Destructors are now called when the parser aborts, for all symbols left
258b75ca
PE
2007 behind on the stack. Also, the start symbol is now destroyed after a
2008 successful parse. In both cases, the behavior was formerly inconsistent.
2009
402b123d 2010** When generating verbose diagnostics, Bison-generated parsers no longer
72f000b0
PE
2011 quote the literal strings associated with tokens. For example, for
2012 a syntax error associated with '%token NUM "number"' they might
2013 print 'syntax error, unexpected number' instead of 'syntax error,
2014 unexpected "number"'.
193d7c70 2015\f
402b123d 2016* Changes in version 2.0, 2004-12-25:
efeed023 2017
402b123d 2018** Possibly-incompatible changes
d7e14fc0 2019
82de6b0d
PE
2020 - Bison-generated parsers no longer default to using the alloca function
2021 (when available) to extend the parser stack, due to widespread
2022 problems in unchecked stack-overflow detection. You can "#define
2023 YYSTACK_USE_ALLOCA 1" to require the use of alloca, but please read
2024 the manual to determine safe values for YYMAXDEPTH in that case.
8dd162d3 2025
82de6b0d
PE
2026 - Error token location.
2027 During error recovery, the location of the syntax error is updated
2028 to cover the whole sequence covered by the error token: it includes
2029 the shifted symbols thrown away during the first part of the error
2030 recovery, and the lookahead rejected during the second part.
18d192f0 2031
82de6b0d
PE
2032 - Semicolon changes:
2033 . Stray semicolons are no longer allowed at the start of a grammar.
2034 . Semicolons are now required after in-grammar declarations.
e342c3be 2035
82de6b0d
PE
2036 - Unescaped newlines are no longer allowed in character constants or
2037 string literals. They were never portable, and GCC 3.4.0 has
2038 dropped support for them. Better diagnostics are now generated if
2039 forget a closing quote.
8dd162d3 2040
82de6b0d 2041 - NUL bytes are no longer allowed in Bison string literals, unfortunately.
f74b6f91 2042
402b123d 2043** New features
1452af69 2044
82de6b0d 2045 - GLR grammars now support locations.
4febdd96 2046
82de6b0d
PE
2047 - New directive: %initial-action.
2048 This directive allows the user to run arbitrary code (including
2049 initializing @$) from yyparse before parsing starts.
1452af69 2050
82de6b0d
PE
2051 - A new directive "%expect-rr N" specifies the expected number of
2052 reduce/reduce conflicts in GLR parsers.
1452af69 2053
e4ab1254 2054 - %token numbers can now be hexadecimal integers, e.g., "%token FOO 0x12d".
82de6b0d 2055 This is a GNU extension.
4febdd96 2056
e4ab1254 2057 - The option "--report=lookahead" was changed to "--report=look-ahead".
9e6e7ed2 2058 [However, this was changed back after 2.3.]
1452af69 2059
82de6b0d 2060 - Experimental %destructor support has been added to lalr1.cc.
1452af69 2061
82de6b0d
PE
2062 - New configure option --disable-yacc, to disable installation of the
2063 yacc command and -ly library introduced in 1.875 for POSIX conformance.
6040d338 2064
402b123d 2065** Bug fixes
d5a3fe37 2066
82de6b0d
PE
2067 - For now, %expect-count violations are now just warnings, not errors.
2068 This is for compatibility with Bison 1.75 and earlier (when there are
2069 reduce/reduce conflicts) and with Bison 1.30 and earlier (when there
2070 are too many or too few shift/reduce conflicts). However, in future
2071 versions of Bison we plan to improve the %expect machinery so that
2072 these violations will become errors again.
3473d0f8 2073
82de6b0d
PE
2074 - Within Bison itself, numbers (e.g., goto numbers) are no longer
2075 arbitrarily limited to 16-bit counts.
d600ee67 2076
82de6b0d 2077 - Semicolons are now allowed before "|" in grammar rules, as POSIX requires.
d600ee67 2078\f
402b123d 2079* Changes in version 1.875, 2003-01-01:
963fcc17 2080
402b123d 2081** The documentation license has been upgraded to version 1.2
dc546b0f 2082 of the GNU Free Documentation License.
75eb3bc4 2083
402b123d 2084** syntax error processing
75eb3bc4 2085
dc546b0f
PE
2086 - In Yacc-style parsers YYLLOC_DEFAULT is now used to compute error
2087 locations too. This fixes bugs in error-location computation.
75eb3bc4 2088
dc546b0f
PE
2089 - %destructor
2090 It is now possible to reclaim the memory associated to symbols
2091 discarded during error recovery. This feature is still experimental.
20daca06 2092
dc546b0f
PE
2093 - %error-verbose
2094 This new directive is preferred over YYERROR_VERBOSE.
74724a70 2095
dc546b0f
PE
2096 - #defining yyerror to steal internal variables is discouraged.
2097 It is not guaranteed to work forever.
d1de5372 2098
402b123d 2099** POSIX conformance
d1de5372 2100
dc546b0f
PE
2101 - Semicolons are once again optional at the end of grammar rules.
2102 This reverts to the behavior of Bison 1.33 and earlier, and improves
2103 compatibility with Yacc.
74724a70 2104
e4ab1254
AD
2105 - "parse error" -> "syntax error"
2106 Bison now uniformly uses the term "syntax error"; formerly, the code
2107 and manual sometimes used the term "parse error" instead. POSIX
2108 requires "syntax error" in diagnostics, and it was thought better to
dc546b0f 2109 be consistent.
74724a70 2110
dc546b0f
PE
2111 - The documentation now emphasizes that yylex and yyerror must be
2112 declared before use. C99 requires this.
d1de5372 2113
dc546b0f
PE
2114 - Bison now parses C99 lexical constructs like UCNs and
2115 backslash-newline within C escape sequences, as POSIX 1003.1-2001 requires.
d1de5372 2116
dc546b0f
PE
2117 - File names are properly escaped in C output. E.g., foo\bar.y is
2118 output as "foo\\bar.y".
6780ca7a 2119
dc546b0f 2120 - Yacc command and library now available
e4ab1254 2121 The Bison distribution now installs a "yacc" command, as POSIX requires.
dc546b0f
PE
2122 Also, Bison now installs a small library liby.a containing
2123 implementations of Yacc-compatible yyerror and main functions.
2124 This library is normally not useful, but POSIX requires it.
6e649e65 2125
dc546b0f 2126 - Type clashes now generate warnings, not errors.
6e649e65 2127
dc546b0f
PE
2128 - If the user does not define YYSTYPE as a macro, Bison now declares it
2129 using typedef instead of defining it as a macro.
2130 For consistency, YYLTYPE is also declared instead of defined.
9501dc6e 2131
402b123d 2132** Other compatibility issues
886a425c 2133
e4ab1254
AD
2134 - %union directives can now have a tag before the "{", e.g., the
2135 directive "%union foo {...}" now generates the C code
2136 "typedef union foo { ... } YYSTYPE;"; this is for Yacc compatibility.
2137 The default union tag is "YYSTYPE", for compatibility with Solaris 9 Yacc.
2138 For consistency, YYLTYPE's struct tag is now "YYLTYPE" not "yyltype".
dc546b0f 2139 This is for compatibility with both Yacc and Bison 1.35.
72f889cc 2140
e4ab1254 2141 - ";" is output before the terminating "}" of an action, for
dc546b0f 2142 compatibility with Bison 1.35.
886a425c 2143
dc546b0f 2144 - Bison now uses a Yacc-style format for conflict reports, e.g.,
e4ab1254 2145 "conflicts: 2 shift/reduce, 1 reduce/reduce".
437c2d80 2146
e4ab1254 2147 - "yystype" and "yyltype" are now obsolescent macros instead of being
dc546b0f
PE
2148 typedefs or tags; they are no longer documented and are planned to be
2149 withdrawn in a future release.
2a8d363a 2150
402b123d 2151** GLR parser notes
2a8d363a 2152
dc546b0f
PE
2153 - GLR and inline
2154 Users of Bison have to decide how they handle the portability of the
e4ab1254 2155 C keyword "inline".
959e5f51 2156
e4ab1254
AD
2157 - "parsing stack overflow..." -> "parser stack overflow"
2158 GLR parsers now report "parser stack overflow" as per the Bison manual.
900c5db5 2159
18ad57b3
AD
2160** %parse-param and %lex-param
2161 The macros YYPARSE_PARAM and YYLEX_PARAM provide a means to pass
2162 additional context to yyparse and yylex. They suffer from several
2163 shortcomings:
2164
2165 - a single argument only can be added,
2166 - their types are weak (void *),
242cc08e 2167 - this context is not passed to ancillary functions such as yyerror,
18ad57b3
AD
2168 - only yacc.c parsers support them.
2169
2170 The new %parse-param/%lex-param directives provide a more precise control.
2171 For instance:
2172
2173 %parse-param {int *nastiness}
2174 %lex-param {int *nastiness}
2175 %parse-param {int *randomness}
2176
2177 results in the following signatures:
2178
2179 int yylex (int *nastiness);
2180 int yyparse (int *nastiness, int *randomness);
2181
2182 or, if both %pure-parser and %locations are used:
2183
2184 int yylex (YYSTYPE *lvalp, YYLTYPE *llocp, int *nastiness);
2185 int yyparse (int *nastiness, int *randomness);
2186
402b123d 2187** Bison now warns if it detects conflicting outputs to the same file,
e4ab1254 2188 e.g., it generates a warning for "bison -d -o foo.h foo.y" since
dc546b0f 2189 that command outputs both code and header to foo.h.
6e40b4eb 2190
402b123d 2191** #line in output files
dc546b0f 2192 - --no-line works properly.
6e40b4eb 2193
402b123d 2194** Bison can no longer be built by a K&R C compiler; it requires C89 or
6e40b4eb
AD
2195 later to be built. This change originally took place a few versions
2196 ago, but nobody noticed until we recently asked someone to try
2197 building Bison with a K&R C compiler.
d600ee67 2198\f
402b123d 2199* Changes in version 1.75, 2002-10-14:
7933f2b5 2200
402b123d 2201** Bison should now work on 64-bit hosts.
7933f2b5 2202
402b123d 2203** Indonesian translation thanks to Tedi Heriyanto.
7933f2b5 2204
402b123d 2205** GLR parsers
f50adbbd
AD
2206 Fix spurious parse errors.
2207
402b123d 2208** Pure parsers
f50adbbd
AD
2209 Some people redefine yyerror to steal yyparse' private variables.
2210 Reenable this trick until an official feature replaces it.
2211
402b123d 2212** Type Clashes
d90c934c
AD
2213 In agreement with POSIX and with other Yaccs, leaving a default
2214 action is valid when $$ is untyped, and $1 typed:
2215
e9690142 2216 untyped: ... typed;
d90c934c
AD
2217
2218 but the converse remains an error:
2219
e9690142 2220 typed: ... untyped;
d90c934c 2221
402b123d 2222** Values of mid-rule actions
d90c934c
AD
2223 The following code:
2224
e9690142 2225 foo: { ... } { $$ = $1; } ...
d90c934c
AD
2226
2227 was incorrectly rejected: $1 is defined in the second mid-rule
2228 action, and is equal to the $$ of the first mid-rule action.
d600ee67 2229\f
402b123d 2230* Changes in version 1.50, 2002-10-04:
adc8c848 2231
402b123d 2232** GLR parsing
676385e2
PH
2233 The declaration
2234 %glr-parser
2235 causes Bison to produce a Generalized LR (GLR) parser, capable of handling
2236 almost any context-free grammar, ambiguous or not. The new declarations
e8832397 2237 %dprec and %merge on grammar rules allow parse-time resolution of
676385e2
PH
2238 ambiguities. Contributed by Paul Hilfinger.
2239
7933f2b5 2240 Unfortunately Bison 1.50 does not work properly on 64-bit hosts
420f93c8
PE
2241 like the Alpha, so please stick to 32-bit hosts for now.
2242
402b123d 2243** Output Directory
8c165d89 2244 When not in Yacc compatibility mode, when the output file was not
e4ab1254
AD
2245 specified, running "bison foo/bar.y" created "foo/bar.c". It
2246 now creates "bar.c".
8c165d89 2247
402b123d 2248** Undefined token
007a50a4 2249 The undefined token was systematically mapped to 2 which prevented
e88dbdbf 2250 the use of 2 by the user. This is no longer the case.
007a50a4 2251
402b123d 2252** Unknown token numbers
e88dbdbf 2253 If yylex returned an out of range value, yyparse could die. This is
007a50a4
AD
2254 no longer the case.
2255
402b123d 2256** Error token
e88dbdbf 2257 According to POSIX, the error token must be 256.
23c5a174
AD
2258 Bison extends this requirement by making it a preference: *if* the
2259 user specified that one of her tokens is numbered 256, then error
2260 will be mapped onto another number.
2261
402b123d 2262** Verbose error messages
e4ab1254 2263 They no longer report "..., expecting error or..." for states where
217598da
AD
2264 error recovery is possible.
2265
402b123d 2266** End token
e4ab1254 2267 Defaults to "$end" instead of "$".
217598da 2268
402b123d 2269** Error recovery now conforms to documentation and to POSIX
68cd8af3
PE
2270 When a Bison-generated parser encounters a syntax error, it now pops
2271 the stack until it finds a state that allows shifting the error
2272 token. Formerly, it popped the stack until it found a state that
2273 allowed some non-error action other than a default reduction on the
2274 error token. The new behavior has long been the documented behavior,
2275 and has long been required by POSIX. For more details, please see
337116ba
PE
2276 Paul Eggert, "Reductions during Bison error handling" (2002-05-20)
2277 <http://lists.gnu.org/archive/html/bug-bison/2002-05/msg00038.html>.
68cd8af3 2278
402b123d 2279** Traces
5504898e
AD
2280 Popped tokens and nonterminals are now reported.
2281
402b123d 2282** Larger grammars
a861a339
PE
2283 Larger grammars are now supported (larger token numbers, larger grammar
2284 size (= sum of the LHS and RHS lengths), larger LALR tables).
2285 Formerly, many of these numbers ran afoul of 16-bit limits;
2286 now these limits are 32 bits on most hosts.
355e7c1c 2287
402b123d 2288** Explicit initial rule
643a5994
AD
2289 Bison used to play hacks with the initial rule, which the user does
2290 not write. It is now explicit, and visible in the reports and
2291 graphs as rule 0.
23c5a174 2292
402b123d 2293** Useless rules
643a5994 2294 Before, Bison reported the useless rules, but, although not used,
77714df2 2295 included them in the parsers. They are now actually removed.
23c5a174 2296
402b123d 2297** Useless rules, useless nonterminals
6b98e4b5
AD
2298 They are now reported, as a warning, with their locations.
2299
402b123d 2300** Rules never reduced
e8832397
AD
2301 Rules that can never be reduced because of conflicts are now
2302 reported.
2303
e4ab1254 2304** Incorrect "Token not used"
11652ab3
AD
2305 On a grammar such as
2306
e29f0771
AD
2307 %token useless useful
2308 %%
2309 exp: '0' %prec useful;
11652ab3
AD
2310
2311 where a token was used to set the precedence of the last rule,
e4ab1254 2312 bison reported both "useful" and "useless" as useless tokens.
11652ab3 2313
402b123d 2314** Revert the C++ namespace changes introduced in 1.31
77714df2 2315 as they caused too many portability hassles.
0179dd65 2316
402b123d 2317** Default locations
b2d52318
AD
2318 By an accident of design, the default computation of @$ was
2319 performed after another default computation was performed: @$ = @1.
2320 The latter is now removed: YYLLOC_DEFAULT is fully responsible of
2321 the computation of @$.
adc8c848 2322
402b123d 2323** Token end-of-file
b7c49edf
AD
2324 The token end of file may be specified by the user, in which case,
2325 the user symbol is used in the reports, the graphs, and the verbose
e4ab1254 2326 error messages instead of "$end", which remains being the default.
b7c49edf 2327 For instance
e29f0771 2328 %token MYEOF 0
b7c49edf 2329 or
e29f0771 2330 %token MYEOF 0 "end of file"
fdbcd8e2 2331
402b123d 2332** Semantic parser
fdbcd8e2
AD
2333 This old option, which has been broken for ages, is removed.
2334
402b123d 2335** New translations
a861a339 2336 Brazilian Portuguese, thanks to Alexandre Folle de Menezes.
84614e13
AD
2337 Croatian, thanks to Denis Lackovic.
2338
402b123d 2339** Incorrect token definitions
e4ab1254
AD
2340 When given
2341 %token 'a' "A"
2342 bison used to output
2343 #define 'a' 65
b87f8b21 2344
402b123d 2345** Token definitions as enums
77714df2
AD
2346 Tokens are output both as the traditional #define's, and, provided
2347 the compiler supports ANSI C or is a C++ compiler, as enums.
e88dbdbf 2348 This lets debuggers display names instead of integers.
77714df2 2349
402b123d 2350** Reports
ec3bc396
AD
2351 In addition to --verbose, bison supports --report=THINGS, which
2352 produces additional information:
b408954b
AD
2353 - itemset
2354 complete the core item sets with their closure
e4ab1254 2355 - lookahead [changed to "look-ahead" in 1.875e through 2.3, but changed back]
9e6e7ed2 2356 explicitly associate lookahead tokens to items
b408954b
AD
2357 - solved
2358 describe shift/reduce conflicts solving.
2359 Bison used to systematically output this information on top of
2360 the report. Solved conflicts are now attached to their states.
ec3bc396 2361
402b123d 2362** Type clashes
9af3fbce
AD
2363 Previous versions don't complain when there is a type clash on
2364 the default action if the rule has a mid-rule action, such as in:
2365
e29f0771
AD
2366 %type <foo> bar
2367 %%
2368 bar: '0' {} '0';
9af3fbce
AD
2369
2370 This is fixed.
a861a339 2371
402b123d 2372** GNU M4 is now required when using Bison.
f987e9d2 2373\f
402b123d 2374* Changes in version 1.35, 2002-03-25:
76551463 2375
402b123d 2376** C Skeleton
76551463
AD
2377 Some projects use Bison's C parser with C++ compilers, and define
2378 YYSTYPE as a class. The recent adjustment of C parsers for data
2379 alignment and 64 bit architectures made this impossible.
2380
2381 Because for the time being no real solution for C++ parser
2382 generation exists, kludges were implemented in the parser to
2383 maintain this use. In the future, when Bison has C++ parsers, this
2384 kludge will be disabled.
2385
2386 This kludge also addresses some C++ problems when the stack was
2387 extended.
76551463 2388\f
402b123d 2389* Changes in version 1.34, 2002-03-12:
76551463 2390
402b123d 2391** File name clashes are detected
76551463 2392 $ bison foo.y -d -o foo.x
e4ab1254 2393 fatal error: header and parser would both be named "foo.x"
76551463 2394
e4ab1254 2395** A missing ";" at the end of a rule triggers a warning
76551463
AD
2396 In accordance with POSIX, and in agreement with other
2397 Yacc implementations, Bison will mandate this semicolon in the near
2398 future. This eases the implementation of a Bison parser of Bison
2399 grammars by making this grammar LALR(1) instead of LR(2). To
2400 facilitate the transition, this release introduces a warning.
2401
402b123d 2402** Revert the C++ namespace changes introduced in 1.31, as they caused too
76551463
AD
2403 many portability hassles.
2404
402b123d 2405** DJGPP support added.
76551463 2406
402b123d 2407** Fix test suite portability problems.
76551463 2408\f
402b123d 2409* Changes in version 1.33, 2002-02-07:
76551463 2410
402b123d 2411** Fix C++ issues
76551463
AD
2412 Groff could not be compiled for the definition of size_t was lacking
2413 under some conditions.
2414
402b123d 2415** Catch invalid @n
76551463
AD
2416 As is done with $n.
2417\f
402b123d 2418* Changes in version 1.32, 2002-01-23:
76551463 2419
402b123d 2420** Fix Yacc output file names
76551463 2421
402b123d 2422** Portability fixes
76551463 2423
402b123d 2424** Italian, Dutch translations
76551463 2425\f
402b123d 2426* Changes in version 1.31, 2002-01-14:
52d1aeee 2427
402b123d 2428** Many Bug Fixes
52d1aeee 2429
402b123d 2430** GNU Gettext and %expect
52d1aeee
MA
2431 GNU Gettext asserts 10 s/r conflicts, but there are 7. Now that
2432 Bison dies on incorrect %expectations, we fear there will be
2433 too many bug reports for Gettext, so _for the time being_, %expect
e4ab1254 2434 does not trigger an error when the input file is named "plural.y".
52d1aeee 2435
402b123d 2436** Use of alloca in parsers
52d1aeee
MA
2437 If YYSTACK_USE_ALLOCA is defined to 0, then the parsers will use
2438 malloc exclusively. Since 1.29, but was not NEWS'ed.
2439
2440 alloca is used only when compiled with GCC, to avoid portability
2441 problems as on AIX.
2442
402b123d 2443** yyparse now returns 2 if memory is exhausted; formerly it dumped core.
b47dbebe 2444
402b123d 2445** When the generated parser lacks debugging code, YYDEBUG is now 0
52d1aeee
MA
2446 (as POSIX requires) instead of being undefined.
2447
402b123d 2448** User Actions
52d1aeee
MA
2449 Bison has always permitted actions such as { $$ = $1 }: it adds the
2450 ending semicolon. Now if in Yacc compatibility mode, the semicolon
2451 is no longer output: one has to write { $$ = $1; }.
2452
402b123d 2453** Better C++ compliance
52d1aeee 2454 The output parsers try to respect C++ namespaces.
76551463 2455 [This turned out to be a failed experiment, and it was reverted later.]
52d1aeee 2456
402b123d 2457** Reduced Grammars
52d1aeee
MA
2458 Fixed bugs when reporting useless nonterminals.
2459
402b123d 2460** 64 bit hosts
52d1aeee
MA
2461 The parsers work properly on 64 bit hosts.
2462
402b123d 2463** Error messages
52d1aeee
MA
2464 Some calls to strerror resulted in scrambled or missing error messages.
2465
402b123d 2466** %expect
52d1aeee
MA
2467 When the number of shift/reduce conflicts is correct, don't issue
2468 any warning.
2469
402b123d 2470** The verbose report includes the rule line numbers.
52d1aeee 2471
402b123d 2472** Rule line numbers are fixed in traces.
52d1aeee 2473
402b123d 2474** Swedish translation
52d1aeee 2475
402b123d 2476** Parse errors
52d1aeee
MA
2477 Verbose parse error messages from the parsers are better looking.
2478 Before: parse error: unexpected `'/'', expecting `"number"' or `'-'' or `'(''
2479 Now: parse error: unexpected '/', expecting "number" or '-' or '('
2480
402b123d 2481** Fixed parser memory leaks.
52d1aeee
MA
2482 When the generated parser was using malloc to extend its stacks, the
2483 previous allocations were not freed.
2484
402b123d 2485** Fixed verbose output file.
52d1aeee
MA
2486 Some newlines were missing.
2487 Some conflicts in state descriptions were missing.
2488
402b123d 2489** Fixed conflict report.
52d1aeee
MA
2490 Option -v was needed to get the result.
2491
402b123d 2492** %expect
52d1aeee
MA
2493 Was not used.
2494 Mismatches are errors, not warnings.
2495
402b123d 2496** Fixed incorrect processing of some invalid input.
52d1aeee 2497
402b123d 2498** Fixed CPP guards: 9foo.h uses BISON_9FOO_H instead of 9FOO_H.
52d1aeee 2499
402b123d 2500** Fixed some typos in the documentation.
52d1aeee 2501
402b123d 2502** %token MY_EOF 0 is supported.
52d1aeee
MA
2503 Before, MY_EOF was silently renumbered as 257.
2504
402b123d 2505** doc/refcard.tex is updated.
52d1aeee 2506
402b123d 2507** %output, %file-prefix, %name-prefix.
52d1aeee
MA
2508 New.
2509
402b123d 2510** --output
e4ab1254 2511 New, aliasing "--output-file".
52d1aeee 2512\f
402b123d 2513* Changes in version 1.30, 2001-10-26:
342b8b6e 2514
e4ab1254
AD
2515** "--defines" and "--graph" have now an optional argument which is the
2516 output file name. "-d" and "-g" do not change; they do not take any
342b8b6e
AD
2517 argument.
2518
e4ab1254 2519** "%source_extension" and "%header_extension" are removed, failed
342b8b6e
AD
2520 experiment.
2521
402b123d 2522** Portability fixes.
f987e9d2 2523\f
402b123d 2524* Changes in version 1.29, 2001-09-07:
342b8b6e 2525
402b123d 2526** The output file does not define const, as this caused problems when used
342b8b6e
AD
2527 with common autoconfiguration schemes. If you still use ancient compilers
2528 that lack const, compile with the equivalent of the C compiler option
e4ab1254 2529 "-Dconst=". Autoconf's AC_C_CONST macro provides one way to do this.
342b8b6e 2530
e4ab1254 2531** Added "-g" and "--graph".
f87a2205 2532
402b123d 2533** The Bison manual is now distributed under the terms of the GNU FDL.
f2b5126e 2534
402b123d 2535** The input and the output files has automatically a similar extension.
234a3be3 2536
402b123d 2537** Russian translation added.
f87a2205 2538
402b123d 2539** NLS support updated; should hopefully be less troublesome.
f87a2205 2540
402b123d 2541** Added the old Bison reference card.
c33638bb 2542
e4ab1254 2543** Added "--locations" and "%locations".
6deb4447 2544
e4ab1254 2545** Added "-S" and "--skeleton".
cd5bd6ac 2546
e4ab1254 2547** "%raw", "-r", "--raw" is disabled.
62ab6972 2548
402b123d 2549** Special characters are escaped when output. This solves the problems
cd5bd6ac
AD
2550 of the #line lines with path names including backslashes.
2551
402b123d 2552** New directives.
e4ab1254
AD
2553 "%yacc", "%fixed_output_files", "%defines", "%no_parser", "%verbose",
2554 "%debug", "%source_extension" and "%header_extension".
f987e9d2 2555
402b123d 2556** @$
f987e9d2 2557 Automatic location tracking.
f87a2205 2558\f
402b123d 2559* Changes in version 1.28, 1999-07-06:
d2e00347 2560
402b123d 2561** Should compile better now with K&R compilers.
d2e00347 2562
402b123d 2563** Added NLS.
d2e00347 2564
402b123d 2565** Fixed a problem with escaping the double quote character.
d2e00347 2566
402b123d 2567** There is now a FAQ.
d2e00347 2568\f
402b123d 2569* Changes in version 1.27:
5c31c3c2 2570
402b123d 2571** The make rule which prevented bison.simple from being created on
5c31c3c2
JT
2572 some systems has been fixed.
2573\f
402b123d 2574* Changes in version 1.26:
4be07551 2575
7e508a2b 2576** Bison now uses Automake.
4be07551 2577
402b123d 2578** New mailing lists: <bug-bison@gnu.org> and <help-bison@gnu.org>.
4be07551 2579
402b123d 2580** Token numbers now start at 257 as previously documented, not 258.
4be07551 2581
402b123d 2582** Bison honors the TMPDIR environment variable.
4be07551 2583
402b123d 2584** A couple of buffer overruns have been fixed.
f51dbca1 2585
402b123d 2586** Problems when closing files should now be reported.
f51dbca1 2587
402b123d 2588** Generated parsers should now work even on operating systems which do
f51dbca1 2589 not provide alloca().
4be07551 2590\f
402b123d 2591* Changes in version 1.25, 1995-10-16:
df8878c5 2592
402b123d 2593** Errors in the input grammar are not fatal; Bison keeps reading
df8878c5 2594the grammar file, and reports all the errors found in it.
8c44d3ec 2595
402b123d 2596** Tokens can now be specified as multiple-character strings: for
df8878c5 2597example, you could use "<=" for a token which looks like <=, instead
7e508a2b 2598of choosing a name like LESSEQ.
df8878c5 2599
402b123d 2600** The %token_table declaration says to write a table of tokens (names
df8878c5
RS
2601and numbers) into the parser file. The yylex function can use this
2602table to recognize multiple-character string tokens, or for other
2603purposes.
2604
402b123d 2605** The %no_lines declaration says not to generate any #line preprocessor
df8878c5
RS
2606directives in the parser file.
2607
402b123d 2608** The %raw declaration says to use internal Bison token numbers, not
df8878c5
RS
2609Yacc-compatible token numbers, when token names are defined as macros.
2610
402b123d 2611** The --no-parser option produces the parser tables without including
df8878c5
RS
2612the parser engine; a project can now use its own parser engine.
2613The actions go into a separate file called NAME.act, in the form of
2614a switch statement body.
2615\f
402b123d 2616* Changes in version 1.23:
6780ca7a 2617
4d019228
DM
2618The user can define YYPARSE_PARAM as the name of an argument to be
2619passed into yyparse. The argument should have type void *. It should
2620actually point to an object. Grammar actions can access the variable
2621by casting it to the proper pointer type.
6780ca7a 2622
6780ca7a 2623Line numbers in output file corrected.
6780ca7a 2624\f
402b123d 2625* Changes in version 1.22:
6780ca7a
DM
2626
2627--help option added.
6780ca7a 2628\f
402b123d 2629* Changes in version 1.20:
6780ca7a
DM
2630
2631Output file does not redefine const for C++.
9f4503d6 2632
76551463
AD
2633-----
2634
7d6bad19 2635Copyright (C) 1995-2013 Free Software Foundation, Inc.
76551463 2636
74553c98 2637This file is part of Bison, the GNU Parser Generator.
76551463 2638
f16b0819 2639This program is free software: you can redistribute it and/or modify
76551463 2640it under the terms of the GNU General Public License as published by
f16b0819
PE
2641the Free Software Foundation, either version 3 of the License, or
2642(at your option) any later version.
76551463 2643
f16b0819 2644This program is distributed in the hope that it will be useful,
76551463
AD
2645but WITHOUT ANY WARRANTY; without even the implied warranty of
2646MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
2647GNU General Public License for more details.
2648
2649You should have received a copy of the GNU General Public License
f16b0819 2650along with this program. If not, see <http://www.gnu.org/licenses/>.
7e508a2b
AD
2651
2652 LocalWords: yacc YYBACKUP glr GCC lalr ArrayIndexOutOfBoundsException nullptr
2653 LocalWords: cplusplus liby rpl fprintf mfcalc Wyacc stmt cond expr mk sym lr
2654 LocalWords: IELR ielr Lookahead YYERROR nonassoc LALR's api lookaheads yychar
2655 LocalWords: destructor lookahead YYRHSLOC YYLLOC Rhs ifndef YYFAIL cpp sr rr
2656 LocalWords: preprocessor initializer Wno Wnone Werror FreeBSD prec livelocks
2657 LocalWords: Solaris AIX UX RHEL Tru LHS gcc's Wundef YYENABLE NLS YYLTYPE VCG
2658 LocalWords: yyerror cpp's Wunused yylval yylloc prepend yyparse yylex yypush
2659 LocalWords: Graphviz xml nonterminals midrule destructor's YYSTYPE typedef ly
2660 LocalWords: CHR chr printf stdout namespace preprocessing enum pre include's
2661 LocalWords: YYRECOVERING nonfree destructors YYABORT YYACCEPT params enums de
2662 LocalWords: struct yystype DJGPP lex param Haible NUM alloca YYSTACK NUL goto
2663 LocalWords: YYMAXDEPTH Unescaped UCNs YYLTYPE's yyltype typedefs inline Yaccs
2664 LocalWords: Heriyanto Reenable dprec Hilfinger Eggert MYEOF Folle Menezes EOF
242cc08e 2665 LocalWords: Lackovic define's itemset Groff Gettext malloc NEWS'ed YYDEBUG YY
7e508a2b 2666 LocalWords: namespaces strerror const autoconfiguration Dconst Autoconf's FDL
242cc08e
AD
2667 LocalWords: Automake TMPDIR LESSEQ ylwrap endif yydebug YYTOKEN YYLSP ival hh
2668 LocalWords: extern YYTOKENTYPE TOKENTYPE yytokentype tokentype STYPE lval pdf
dcb366b1 2669 LocalWords: lang yyoutput dvi html ps POSIX lvalp llocp Wother nterm arg init
1282c124
AD
2670 LocalWords: TOK calc yyo fval Wconflicts parsers yystackp yyval yynerrs
2671 LocalWords: Théophile Ranquet Santet fno fnone stype associativity Tolmer
2672 LocalWords: Wprecedence Rassoul Wempty Paolo Bonzini parser's Michiel loc
2673 LocalWords: redeclaration sval fcaret reentrant XSLT xsl Wmaybe yyvsp Tedi
2674 LocalWords: pragmas noreturn untyped Rozenman unexpanded Wojciech Polak
6574576c 2675 LocalWords: Alexandre MERCHANTABILITY yytype
7e508a2b
AD
2676
2677Local Variables:
2678mode: outline
e4ab1254 2679fill-column: 76
7e508a2b 2680End: