]> git.saurik.com Git - bison.git/blame - NEWS
output: record what generated files are source or report files
[bison.git] / NEWS
CommitLineData
ed7658fe 1GNU Bison NEWS
3af4feb2 2
6d94eebb
AD
3* Noteworthy changes in release ?.? (????-??-??) [?]
4
7b0ca050
AD
5** Bug fixes
6
184b42c8
AD
7*** Generated source files when errors are reported
8
9 When warnings are issued and -Werror is set, bison would still generate
10 the source files (*.c, *.h...). As a consequence, some runs of "make"
11 could fail the first time, but not the second (as the files were generated
12 anyway).
13
14 This is fixed: bison no longer generates this source files, but, of
15 course, still produces the various reports (*.output, *.xml, etc.).
16
be29c71d
AD
17*** %empty is used in reports
18
19 Empty right-hand sides are denoted by '%empty' in all the reports (text,
20 dot, XML and formats derived from it).
21
7b0ca050
AD
22*** YYERROR and variants
23
24 When C++ variant support is enabled, an error triggered via YYERROR, but
25 not caught via error recovery, resulted in a double deletion.
6d94eebb 26
c2ecada3 27* Noteworthy changes in release 3.0.1 (2013-11-12) [stable]
e386b50f 28
0a244a22
AD
29** Bug fixes
30
39bace5d
AD
31*** Errors in caret diagnostics
32
33 On some platforms, some errors could result in endless diagnostics.
e4678430
AD
34
35*** Fixes of the -Werror option
36
37 Options such as "-Werror -Wno-error=foo" were still turning "foo"
38 diagnostics into errors instead of warnings. This is fixed.
39
40 Actually, for consistency with GCC, "-Wno-error=foo -Werror" now also
41 leaves "foo" diagnostics as warnings. Similarly, with "-Werror=foo
42 -Wno-error", "foo" diagnostics are now errors.
e386b50f 43
071863b3
AD
44*** GLR Predicates
45
46 As demonstrated in the documentation, one can now leave spaces between
47 "%?" and its "{".
48
265640d5
AD
49*** Installation
50
51 The yacc.1 man page is no longer installed if --disable-yacc was
52 specified.
53
39bace5d
AD
54*** Fixes in the test suite
55
56 Bugs and portability issues.
57
534497f5 58* Noteworthy changes in release 3.0 (2013-07-25) [stable]
facb910c 59
8458a411 60** WARNING: Future backward-incompatibilities!
597afd73 61
597afd73
AD
62 Like other GNU packages, Bison will start using some of the C99 features
63 for its own code, especially the definition of variables after statements.
64 The generated C parsers still aim at C90.
65
8458a411 66** Backward incompatible changes
47db7ed1
AD
67
68*** Obsolete features
69
40bb6f78
AD
70 Support for YYFAIL is removed (deprecated in Bison 2.4.2): use YYERROR.
71
72 Support for yystype and yyltype is removed (deprecated in Bison 1.875):
73 use YYSTYPE and YYLTYPE.
74
75 Support for YYLEX_PARAM and YYPARSE_PARAM is removed (deprecated in Bison
76 1.875): use %lex-param, %parse-param, or %param.
47db7ed1 77
05e25f23
AD
78 Missing semicolons at the end of actions are no longer added (as announced
79 in the release 2.5).
80
1fa19a76
AD
81*** Use of YACC='bison -y'
82
83 TL;DR: With Autoconf <= 2.69, pass -Wno-yacc to (AM_)YFLAGS if you use
84 Bison extensions.
85
86 Traditional Yacc generates 'y.tab.c' whatever the name of the input file.
87 Therefore Makefiles written for Yacc expect 'y.tab.c' (and possibly
88 'y.tab.h' and 'y.outout') to be generated from 'foo.y'.
89
90 To this end, for ages, AC_PROG_YACC, Autoconf's macro to look for an
91 implementation of Yacc, was using Bison as 'bison -y'. While it does
92 ensure compatible output file names, it also enables warnings for
93 incompatibilities with POSIX Yacc. In other words, 'bison -y' triggers
94 warnings for Bison extensions.
95
96 Autoconf 2.70+ fixes this incompatibility by using YACC='bison -o y.tab.c'
97 (which also generates 'y.tab.h' and 'y.output' when needed).
98 Alternatively, disable Yacc warnings by passing '-Wno-yacc' to your Yacc
99 flags (YFLAGS, or AM_YFLAGS with Automake).
100
597afd73
AD
101** Bug fixes
102
c21e515e 103*** The epilogue is no longer affected by internal #defines (glr.c)
597afd73
AD
104
105 The glr.c skeleton uses defines such as #define yylval (yystackp->yyval) in
106 generated code. These weren't properly undefined before the inclusion of
107 the user epilogue, so functions such as the following were butchered by the
108 preprocessor expansion:
109
110 int yylex (YYSTYPE *yylval);
111
6c7022f7 112 This is fixed: yylval, yynerrs, yychar, and yylloc are now valid
597afd73
AD
113 identifiers for user-provided variables.
114
f0f95a50
AD
115*** stdio.h is no longer needed when locations are enabled (yacc.c)
116
117 Changes in Bison 2.7 introduced a dependency on FILE and fprintf when
118 locations are enabled. This is fixed.
119
bb4b189b
AD
120*** Warnings about useless %pure-parser/%define api.pure are restored
121
597afd73
AD
122** Diagnostics reported by Bison
123
124 Most of these features were contributed by Théophile Ranquet and Victor
125 Santet.
73370a9d 126
016426c1
TR
127*** Carets
128
129 Version 2.7 introduced caret errors, for a prettier output. These are now
130 activated by default. The old format can still be used by invoking Bison
131 with -fno-caret (or -fnone).
132
fec5f3c0
AD
133 Some error messages that reproduced excerpts of the grammar are now using
134 the caret information only. For instance on:
135
136 %%
137 exp: 'a' | 'a';
138
139 Bison 2.7 reports:
140
141 in.y: warning: 1 reduce/reduce conflict [-Wconflicts-rr]
142 in.y:2.12-14: warning: rule useless in parser due to conflicts: exp: 'a' [-Wother]
143
144 Now bison reports:
145
146 in.y: warning: 1 reduce/reduce conflict [-Wconflicts-rr]
147 in.y:2.12-14: warning: rule useless in parser due to conflicts [-Wother]
148 exp: 'a' | 'a';
149 ^^^
150
151 and "bison -fno-caret" reports:
152
153 in.y: warning: 1 reduce/reduce conflict [-Wconflicts-rr]
154 in.y:2.12-14: warning: rule useless in parser due to conflicts [-Wother]
155
1048a1c9 156*** Enhancements of the -Werror option
518e8830 157
1048a1c9 158 The -Werror=CATEGORY option is now recognized, and will treat specified
d949eefd
AD
159 warnings as errors. The warnings need not have been explicitly activated
160 using the -W option, this is similar to what GCC 4.7 does.
1048a1c9
AD
161
162 For example, given the following command line, Bison will treat both
d949eefd 163 warnings related to POSIX Yacc incompatibilities and S/R conflicts as
1048a1c9
AD
164 errors (and only those):
165
166 $ bison -Werror=yacc,error=conflicts-sr input.y
167
168 If no categories are specified, -Werror will make all active warnings into
169 errors. For example, the following line does the same the previous example:
170
171 $ bison -Werror -Wnone -Wyacc -Wconflicts-sr input.y
172
173 (By default -Wconflicts-sr,conflicts-rr,deprecated,other is enabled.)
174
175 Note that the categories in this -Werror option may not be prefixed with
176 "no-". However, -Wno-error[=CATEGORY] is valid.
177
178 Note that -y enables -Werror=yacc. Therefore it is now possible to require
179 Yacc-like behavior (e.g., always generate y.tab.c), but to report
180 incompatibilities as warnings: "-y -Wno-error=yacc".
518e8830 181
46bdb8ec 182*** The display of warnings is now richer
73370a9d 183
46bdb8ec 184 The option that controls a given warning is now displayed:
73370a9d 185
46bdb8ec 186 foo.y:4.6: warning: type clash on default action: <foo> != <bar> [-Wother]
73370a9d 187
46bdb8ec
TR
188 In the case of warnings treated as errors, the prefix is changed from
189 "warning: " to "error: ", and the suffix is displayed, in a manner similar
d949eefd 190 to GCC, as [-Werror=CATEGORY].
1048a1c9 191
46bdb8ec
TR
192 For instance, where the previous version of Bison would report (and exit
193 with failure):
1048a1c9
AD
194
195 bison: warnings being treated as errors
46bdb8ec 196 input.y:1.1: warning: stray ',' treated as white space
1048a1c9 197
46bdb8ec 198 it now reports:
1048a1c9 199
1048a1c9
AD
200 input.y:1.1: error: stray ',' treated as white space [-Werror=other]
201
202*** Deprecated constructs
203
204 The new 'deprecated' warning category flags obsolete constructs whose
205 support will be discontinued. It is enabled by default. These warnings
206 used to be reported as 'other' warnings.
207
73370a9d 208*** Useless semantic types
9641b918
VS
209
210 Bison now warns about useless (uninhabited) semantic types. Since
211 semantic types are not declared to Bison (they are defined in the opaque
212 %union structure), it is %printer/%destructor directives about useless
213 types that trigger the warning:
214
215 %token <type1> term
216 %type <type2> nterm
217 %printer {} <type1> <type3>
218 %destructor {} <type2> <type4>
219 %%
220 nterm: term { $$ = $1; };
221
222 3.28-34: warning: type <type3> is used, but is not associated to any symbol
223 4.28-34: warning: type <type4> is used, but is not associated to any symbol
224
31557b9e 225*** Undefined but unused symbols
b921d92f 226
31557b9e
AD
227 Bison used to raise an error for undefined symbols that are not used in
228 the grammar. This is now only a warning.
b921d92f
VS
229
230 %printer {} symbol1
231 %destructor {} symbol2
31557b9e 232 %type <type> symbol3
b921d92f
VS
233 %%
234 exp: "a";
235
73370a9d 236*** Useless destructors or printers
ea9a35c6
VS
237
238 Bison now warns about useless destructors or printers. In the following
239 example, the printer for <type1>, and the destructor for <type2> are
240 useless: all symbols of <type1> (token1) already have a printer, and all
241 symbols of type <type2> (token2) already have a destructor.
242
243 %token <type1> token1
244 <type2> token2
245 <type3> token3
246 <type4> token4
247 %printer {} token1 <type1> <type3>
248 %destructor {} token2 <type2> <type4>
249
d87ea54c
AD
250*** Conflicts
251
252 The warnings and error messages about shift/reduce and reduce/reduce
253 conflicts have been normalized. For instance on the following foo.y file:
254
255 %glr-parser
256 %%
257 exp: exp '+' exp | '0' | '0';
258
259 compare the previous version of bison:
260
261 $ bison foo.y
262 foo.y: conflicts: 1 shift/reduce, 2 reduce/reduce
263 $ bison -Werror foo.y
264 bison: warnings being treated as errors
265 foo.y: conflicts: 1 shift/reduce, 2 reduce/reduce
266
267 with the new behavior:
268
269 $ bison foo.y
270 foo.y: warning: 1 shift/reduce conflict [-Wconflicts-sr]
271 foo.y: warning: 2 reduce/reduce conflicts [-Wconflicts-rr]
272 $ bison -Werror foo.y
9503b0a4
TR
273 foo.y: error: 1 shift/reduce conflict [-Werror=conflicts-sr]
274 foo.y: error: 2 reduce/reduce conflicts [-Werror=conflicts-rr]
d87ea54c
AD
275
276 When %expect or %expect-rr is used, such as with bar.y:
277
278 %expect 0
279 %glr-parser
280 %%
281 exp: exp '+' exp | '0' | '0';
282
283 Former behavior:
284
285 $ bison bar.y
286 bar.y: conflicts: 1 shift/reduce, 2 reduce/reduce
287 bar.y: expected 0 shift/reduce conflicts
288 bar.y: expected 0 reduce/reduce conflicts
289
290 New one:
291
292 $ bison bar.y
9503b0a4
TR
293 bar.y: error: shift/reduce conflicts: 1 found, 0 expected
294 bar.y: error: reduce/reduce conflicts: 2 found, 0 expected
d87ea54c 295
f24695ef
AD
296** Incompatibilities with POSIX Yacc
297
298 The 'yacc' category is no longer part of '-Wall', enable it explicitly
299 with '-Wyacc'.
300
2055a44e
AD
301** Additional yylex/yyparse arguments
302
6dc4663d
AD
303 The new directive %param declares additional arguments to both yylex and
304 yyparse. The %lex-param, %parse-param, and %param directives support one
305 or more arguments. Instead of
2055a44e 306
e436fa67
AD
307 %lex-param {arg1_type *arg1}
308 %lex-param {arg2_type *arg2}
309 %parse-param {arg1_type *arg1}
310 %parse-param {arg2_type *arg2}
2055a44e
AD
311
312 one may now declare
313
e436fa67 314 %param {arg1_type *arg1} {arg2_type *arg2}
2055a44e 315
630a0218
AD
316** Types of values for %define variables
317
318 Bison used to make no difference between '%define foo bar' and '%define
319 foo "bar"'. The former is now called a 'keyword value', and the latter a
320 'string value'. A third kind was added: 'code values', such as '%define
321 foo {bar}'.
322
323 Keyword variables are used for fixed value sets, e.g.,
324
325 %define lr.type lalr
326
327 Code variables are used for value in the target language, e.g.,
328
329 %define api.value.type {struct semantic_type}
330
331 String variables are used remaining cases, e.g. file names.
332
2a6b66c5 333** Variable api.token.prefix
99c08fb6 334
2a6b66c5 335 The variable api.token.prefix changes the way tokens are identified in
99c08fb6
AD
336 the generated files. This is especially useful to avoid collisions
337 with identifiers in the target language. For instance
338
e436fa67 339 %token FILE for ERROR
630a0218 340 %define api.token.prefix {TOK_}
e436fa67
AD
341 %%
342 start: FILE for ERROR;
99c08fb6
AD
343
344 will generate the definition of the symbols TOK_FILE, TOK_for, and
345 TOK_ERROR in the generated sources. In particular, the scanner must
346 use these prefixed token names, although the grammar itself still
347 uses the short names (as in the sample rule given above).
348
6574576c
AD
349** Variable api.value.type
350
351 This new %define variable supersedes the #define macro YYSTYPE. The use
352 of YYSTYPE is discouraged. In particular, #defining YYSTYPE *and* either
353 using %union or %defining api.value.type results in undefined behavior.
354
355 Either define api.value.type, or use "%union":
356
357 %union
358 {
359 int ival;
360 char *sval;
361 }
362 %token <ival> INT "integer"
363 %token <sval> STRING "string"
364 %printer { fprintf (yyo, "%d", $$); } <ival>
365 %destructor { free ($$); } <sval>
366
367 /* In yylex(). */
368 yylval.ival = 42; return INT;
369 yylval.sval = "42"; return STRING;
370
630a0218
AD
371 The %define variable api.value.type supports both keyword and code values.
372
373 The keyword value 'union' means that the user provides genuine types, not
435575cb 374 union member names such as "ival" and "sval" above (WARNING: will fail if
1fa19a76 375 -y/--yacc/%yacc is enabled).
6574576c 376
435575cb 377 %define api.value.type union
6574576c
AD
378 %token <int> INT "integer"
379 %token <char *> STRING "string"
380 %printer { fprintf (yyo, "%d", $$); } <int>
381 %destructor { free ($$); } <char *>
382
383 /* In yylex(). */
384 yylval.INT = 42; return INT;
385 yylval.STRING = "42"; return STRING;
386
435575cb
AD
387 The keyword value variant is somewhat equivalent, but for C++ special
388 provision is made to allow classes to be used (more about this below).
6574576c 389
435575cb 390 %define api.value.type variant
6574576c
AD
391 %token <int> INT "integer"
392 %token <std::string> STRING "string"
393
630a0218 394 Code values (in braces) denote user defined types. This is where YYSTYPE
435575cb 395 used to be used.
6574576c
AD
396
397 %code requires
398 {
399 struct my_value
400 {
401 enum
402 {
403 is_int, is_string
404 } kind;
405 union
406 {
407 int ival;
408 char *sval;
409 } u;
410 };
411 }
435575cb 412 %define api.value.type {struct my_value}
6574576c
AD
413 %token <u.ival> INT "integer"
414 %token <u.sval> STRING "string"
415 %printer { fprintf (yyo, "%d", $$); } <u.ival>
416 %destructor { free ($$); } <u.sval>
417
418 /* In yylex(). */
419 yylval.u.ival = 42; return INT;
420 yylval.u.sval = "42"; return STRING;
421
31b850d2
AD
422** Variable parse.error
423
1f77b2e0
AD
424 This variable controls the verbosity of error messages. The use of the
425 %error-verbose directive is deprecated in favor of "%define parse.error
426 verbose".
31b850d2 427
c21e515e
AD
428** Renamed %define variables
429
430 The following variables have been renamed for consistency. Backward
431 compatibility is ensured, but upgrading is recommended.
432
433 lr.default-reductions -> lr.default-reduction
434 lr.keep-unreachable-states -> lr.keep-unreachable-state
435 namespace -> api.namespace
436 stype -> api.value.type
437
ca2a6d15
PH
438** Semantic predicates
439
597afd73
AD
440 Contributed by Paul Hilfinger.
441
1f77b2e0
AD
442 The new, experimental, semantic-predicate feature allows actions of the
443 form "%?{ BOOLEAN-EXPRESSION }", which cause syntax errors (as for
ca2a6d15 444 YYERROR) if the expression evaluates to 0, and are evaluated immediately
1f77b2e0
AD
445 in GLR parsers, rather than being deferred. The result is that they allow
446 the programmer to prune possible parses based on the values of run-time
447 expressions.
ca2a6d15 448
d1400569
AD
449** The directive %expect-rr is now an error in non GLR mode
450
451 It used to be an error only if used in non GLR mode, _and_ if there are
452 reduce/reduce conflicts.
453
5202b6ac 454** Tokens are numbered in their order of appearance
93561c21 455
5202b6ac
VT
456 Contributed by Valentin Tolmer.
457
458 With '%token A B', A had a number less than the one of B. However,
459 precedence declarations used to generate a reversed order. This is now
460 fixed, and introducing tokens with any of %token, %left, %right,
461 %precedence, or %nonassoc yields the same result.
462
463 When mixing declarations of tokens with a litteral character (e.g., 'a')
464 or with an identifier (e.g., B) in a precedence declaration, Bison
465 numbered the litteral characters first. For example
466
467 %right A B 'c' 'd'
468
469 would lead to the tokens declared in this order: 'c' 'd' A B. Again, the
470 input order is now preserved.
471
472 These changes were made so that one can remove useless precedence and
473 associativity declarations (i.e., map %nonassoc, %left or %right to
474 %precedence, or to %token) and get exactly the same output.
93561c21 475
cc2235ac
VT
476** Useless precedence and associativity
477
d2f9ae18
AD
478 Contributed by Valentin Tolmer.
479
1282c124 480 When developing and maintaining a grammar, useless associativity and
cc2235ac
VT
481 precedence directives are common. They can be a nuisance: new ambiguities
482 arising are sometimes masked because their conflicts are resolved due to
483 the extra precedence or associativity information. Furthermore, it can
484 hinder the comprehension of a new grammar: one will wonder about the role
485 of a precedence, where in fact it is useless. The following changes aim
486 at detecting and reporting these extra directives.
487
488*** Precedence warning category
489
490 A new category of warning, -Wprecedence, was introduced. It flags the
491 useless precedence and associativity directives.
492
493*** Useless associativity
494
495 Bison now warns about symbols with a declared associativity that is never
496 used to resolve conflicts. In that case, using %precedence is sufficient;
497 the parsing tables will remain unchanged. Solving these warnings may raise
498 useless precedence warnings, as the symbols no longer have associativity.
499 For example:
500
501 %left '+'
502 %left '*'
503 %%
504 exp:
1282c124
AD
505 "number"
506 | exp '+' "number"
cc2235ac
VT
507 | exp '*' exp
508 ;
509
510 will produce a
511
512 warning: useless associativity for '+', use %precedence [-Wprecedence]
513 %left '+'
514 ^^^
515
516*** Useless precedence
517
518 Bison now warns about symbols with a declared precedence and no declared
519 associativity (i.e., declared with %precedence), and whose precedence is
520 never used. In that case, the symbol can be safely declared with %token
521 instead, without modifying the parsing tables. For example:
522
523 %precedence '='
524 %%
1282c124 525 exp: "var" '=' "number";
cc2235ac
VT
526
527 will produce a
528
529 warning: useless precedence for '=' [-Wprecedence]
530 %precedence '='
531 ^^^
532
533*** Useless precedence and associativity
534
535 In case of both useless precedence and associativity, the issue is flagged
536 as follows:
537
538 %nonassoc '='
539 %%
1282c124 540 exp: "var" '=' "number";
cc2235ac
VT
541
542 The warning is:
543
544 warning: useless precedence and associativity for '=' [-Wprecedence]
545 %nonassoc '='
546 ^^^
09add9c2
AD
547
548** Empty rules
549
6240346a
AD
550 With help from Joel E. Denny and Gabriel Rassoul.
551
09add9c2
AD
552 Empty rules (i.e., with an empty right-hand side) can now be explicitly
553 marked by the new %empty directive. Using %empty on a non-empty rule is
554 an error. The new -Wempty-rule warning reports empty rules without
555 %empty. On the following grammar:
556
557 %%
558 s: a b c;
559 a: ;
560 b: %empty;
561 c: 'a' %empty;
562
563 bison reports:
564
565 3.4-5: warning: empty rule without %empty [-Wempty-rule]
566 a: {}
567 ^^
568 5.8-13: error: %empty on non-empty rule
569 c: 'a' %empty {};
570 ^^^^^^
cc2235ac 571
c21e515e
AD
572** Java skeleton improvements
573
c21e515e
AD
574 The constants for token names were moved to the Lexer interface. Also, it
575 is possible to add code to the parser's constructors using "%code init"
576 and "%define init_throws".
aa94def1
DH
577 Contributed by Paolo Bonzini.
578
de1a2f20
AD
579 The Java skeleton now supports push parsing.
580 Contributed by Dennis Heimbigner.
581
c21e515e
AD
582** C++ skeletons improvements
583
584*** The parser header is no longer mandatory (lalr1.cc, glr.cc)
585
586 Using %defines is now optional. Without it, the needed support classes
587 are defined in the generated parser, instead of additional files (such as
588 location.hh, position.hh and stack.hh).
589
590*** Locations are no longer mandatory (lalr1.cc, glr.cc)
591
592 Both lalr1.cc and glr.cc no longer require %location.
593
594*** syntax_error exception (lalr1.cc)
595
596 The C++ parser features a syntax_error exception, which can be
597 thrown from the scanner or from user rules to raise syntax errors.
598 This facilitates reporting errors caught in sub-functions (e.g.,
599 rejecting too large integral literals from a conversion function
600 used by the scanner, or rejecting invalid combinations from a
601 factory invoked by the user actions).
602
603*** %define api.value.type variant
604
605 This is based on a submission from Michiel De Wilde. With help
606 from Théophile Ranquet.
607
608 In this mode, complex C++ objects can be used as semantic values. For
609 instance:
610
611 %token <::std::string> TEXT;
612 %token <int> NUMBER;
613 %token SEMICOLON ";"
614 %type <::std::string> item;
615 %type <::std::list<std::string>> list;
616 %%
617 result:
618 list { std::cout << $1 << std::endl; }
619 ;
620
621 list:
6240346a 622 %empty { /* Generates an empty string list. */ }
c21e515e
AD
623 | list item ";" { std::swap ($$, $1); $$.push_back ($2); }
624 ;
625
626 item:
627 TEXT { std::swap ($$, $1); }
628 | NUMBER { $$ = string_cast ($1); }
629 ;
630
631*** %define api.token.constructor
632
633 When variants are enabled, Bison can generate functions to build the
634 tokens. This guarantees that the token type (e.g., NUMBER) is consistent
635 with the semantic value (e.g., int):
636
637 parser::symbol_type yylex ()
638 {
639 parser::location_type loc = ...;
640 ...
641 return parser::make_TEXT ("Hello, world!", loc);
642 ...
643 return parser::make_NUMBER (42, loc);
644 ...
645 return parser::make_SEMICOLON (loc);
646 ...
647 }
648
75ae8299
AD
649*** C++ locations
650
651 There are operator- and operator-= for 'location'. Negative line/column
652 increments can no longer underflow the resulting value.
653
1f5542fe 654* Noteworthy changes in release 2.7.1 (2013-04-15) [stable]
cc8962bd 655
80a2826e
AD
656** Bug fixes
657
658*** Fix compiler attribute portability (yacc.c)
659
660 With locations enabled, __attribute__ was used unprotected.
0a7b8559 661
e83be476
AD
662*** Fix some compiler warnings (lalr1.cc)
663
c13bb348 664* Noteworthy changes in release 2.7 (2012-12-12) [stable]
effd30c0 665
edf9a06f 666** Bug fixes
7bada535 667
edf9a06f 668 Warnings about uninitialized yylloc in yyparse have been fixed.
7bada535 669
1127a75a
AD
670 Restored C90 compliance (yet no report was ever made).
671
d4fe9e88 672** Diagnostics are improved
7bada535 673
8458a411
AD
674 Contributed by Théophile Ranquet.
675
d4fe9e88 676*** Changes in the format of error messages
7bada535 677
d4fe9e88 678 This used to be the format of many error reports:
1f1bd572 679
d4fe9e88
AD
680 input.y:2.7-12: %type redeclaration for exp
681 input.y:1.7-12: previous declaration
1f1bd572 682
d4fe9e88 683 It is now:
1f1bd572 684
d4fe9e88
AD
685 input.y:2.7-12: error: %type redeclaration for exp
686 input.y:1.7-12: previous declaration
cbaea010 687
d4fe9e88 688*** New format for error reports: carets
cbaea010 689
d4fe9e88 690 Caret errors have been added to Bison:
cbaea010 691
d4fe9e88
AD
692 input.y:2.7-12: error: %type redeclaration for exp
693 %type <sval> exp
694 ^^^^^^
695 input.y:1.7-12: previous declaration
696 %type <ival> exp
697 ^^^^^^
cbaea010 698
d4fe9e88 699 or
cbaea010 700
7bada535 701 input.y:3.20-23: error: ambiguous reference: '$exp'
fb6040f0 702 exp: exp '+' exp { $exp = $1 + $3; };
7bada535 703 ^^^^
fb6040f0
TR
704 input.y:3.1-3: refers to: $exp at $$
705 exp: exp '+' exp { $exp = $1 + $3; };
706 ^^^
707 input.y:3.6-8: refers to: $exp at $1
708 exp: exp '+' exp { $exp = $1 + $3; };
709 ^^^
710 input.y:3.14-16: refers to: $exp at $3
711 exp: exp '+' exp { $exp = $1 + $3; };
712 ^^^
7bada535 713
1282c124
AD
714 The default behavior for now is still not to display these unless
715 explicitly asked with -fcaret (or -fall). However, in a later release, it
d4fe9e88
AD
716 will be made the default behavior (but may still be deactivated with
717 -fno-caret).
d3e4409a 718
1f1bd572 719** New value for %define variable: api.pure full
d3e4409a 720
1f1bd572 721 The %define variable api.pure requests a pure (reentrant) parser. However,
d4fe9e88
AD
722 for historical reasons, using it in a location-tracking Yacc parser
723 resulted in a yyerror function that did not take a location as a
724 parameter. With this new value, the user may request a better pure parser,
725 where yyerror does take a location as a parameter (in location-tracking
726 parsers).
1f1bd572
TR
727
728 The use of "%define api.pure true" is deprecated in favor of this new
729 "%define api.pure full".
d3e4409a 730
7287be84 731** New %define variable: api.location.type (glr.cc, lalr1.cc, lalr1.java)
db8ab2be
AD
732
733 The %define variable api.location.type defines the name of the type to use
734 for locations. When defined, Bison no longer generates the position.hh
735 and location.hh files, nor does the parser will include them: the user is
736 then responsible to define her type.
737
738 This can be used in programs with several parsers to factor their location
7287be84
AD
739 and position files: let one of them generate them, and the others just use
740 them.
db8ab2be
AD
741
742 This feature was actually introduced, but not documented, in Bison 2.5,
743 under the name "location_type" (which is maintained for backward
744 compatibility).
745
7287be84
AD
746 For consistency, lalr1.java's %define variables location_type and
747 position_type are deprecated in favor of api.location.type and
748 api.position.type.
749
d4fe9e88
AD
750** Exception safety (lalr1.cc)
751
752 The parse function now catches exceptions, uses the %destructors to
753 release memory (the lookahead symbol and the symbols pushed on the stack)
754 before re-throwing the exception.
755
756 This feature is somewhat experimental. User feedback would be
757 appreciated.
758
9c16d399 759** Graph improvements in DOT and XSLT
fc4fdd62 760
8458a411
AD
761 Contributed by Théophile Ranquet.
762
fc4fdd62
TR
763 The graphical presentation of the states is more readable: their shape is
764 now rectangular, the state number is clearly displayed, and the items are
765 numbered and left-justified.
766
767 The reductions are now explicitly represented as transitions to other
768 diamond shaped nodes.
769
9c16d399
TR
770 These changes are present in both --graph output and xml2dot.xsl XSLT
771 processing, with minor (documented) differences.
772
d4fe9e88 773** %language is no longer an experimental feature.
fb4c8a7c 774
d4fe9e88
AD
775 The introduction of this feature, in 2.4, was four years ago. The
776 --language option and the %language directive are no longer experimental.
fb4c8a7c 777
53e2cd1e
AD
778** Documentation
779
780 The sections about shift/reduce and reduce/reduce conflicts resolution
781 have been fixed and extended.
9d2423f5 782
d4fe9e88
AD
783 Although introduced more than four years ago, XML and Graphviz reports
784 were not properly documented.
785
be22823e
AD
786 The translation of mid-rule actions is now described.
787
9d3f7eaf 788* Noteworthy changes in release 2.6.5 (2012-11-07) [stable]
6f1360bd 789
a68b1f23
AD
790 We consider compiler warnings about Bison generated parsers to be bugs.
791 Rather than working around them in your own project, please consider
792 reporting them to us.
793
794** Bug fixes
795
796 Warnings about uninitialized yylval and/or yylloc for push parsers with a
797 pure interface have been fixed for GCC 4.0 up to 4.8, and Clang 2.9 to
798 3.2.
799
800 Other issues in the test suite have been addressed.
6f1360bd 801
1282c124 802 Null characters are correctly displayed in error messages.
95066e92 803
a1d1ab50
AD
804 When possible, yylloc is correctly initialized before calling yylex. It
805 is no longer necessary to initialize it in the %initial-action.
806
0ac15849 807* Noteworthy changes in release 2.6.4 (2012-10-23) [stable]
a4eb820f 808
468455e1 809 Bison 2.6.3's --version was incorrect. This release fixes this issue.
a4eb820f 810
6eb8f74f 811* Noteworthy changes in release 2.6.3 (2012-10-22) [stable]
933ec544 812
6b4cb804
AD
813** Bug fixes
814
a1a77e1f 815 Bugs and portability issues in the test suite have been fixed.
6b4cb804
AD
816
817 Some errors in translations have been addressed, and --help now directs
818 users to the appropriate place to report them.
819
820 Stray Info files shipped by accident are removed.
821
822 Incorrect definitions of YY_, issued by yacc.c when no parser header is
823 generated, are removed.
9c26b8fc 824
a2b3f101
TR
825 All the generated headers are self-contained.
826
c9d5bcc9
AD
827** Header guards (yacc.c, glr.c, glr.cc)
828
829 In order to avoid collisions, the header guards are now
830 YY_<PREFIX>_<FILE>_INCLUDED, instead of merely <PREFIX>_<FILE>.
831 For instance the header generated from
832
833 %define api.prefix "calc"
834 %defines "lib/parse.h"
835
836 will use YY_CALC_LIB_PARSE_H_INCLUDED as guard.
837
c12c4c50 838** Fix compiler warnings in the generated parser (yacc.c, glr.c)
321d3e35
AD
839
840 The compilation of pure parsers (%define api.pure) can trigger GCC
841 warnings such as:
842
843 input.c: In function 'yyparse':
844 input.c:1503:12: warning: 'yylval' may be used uninitialized in this
845 function [-Wmaybe-uninitialized]
846 *++yyvsp = yylval;
847 ^
848
849 This is now fixed; pragmas to avoid these warnings are no longer needed.
850
c12c4c50
AD
851 Warnings from clang ("equality comparison with extraneous parentheses" and
852 "function declared 'noreturn' should not return") have also been
853 addressed.
854
e1eeecd3 855* Noteworthy changes in release 2.6.2 (2012-08-03) [stable]
9c26b8fc 856
43ca8040
AD
857** Bug fixes
858
859 Buffer overruns, complaints from Flex, and portability issues in the test
860 suite have been fixed.
861
c9d546b2
AD
862** Spaces in %lex- and %parse-param (lalr1.cc, glr.cc)
863
864 Trailing end-of-lines in %parse-param or %lex-param would result in
865 invalid C++. This is fixed.
9c26b8fc 866
dcd5344d
AD
867** Spurious spaces and end-of-lines
868
869 The generated files no longer end (nor start) with empty lines.
870
77b214ef 871* Noteworthy changes in release 2.6.1 (2012-07-30) [stable]
a4107f24 872
8617d87e
AD
873 Bison no longer executes user-specified M4 code when processing a grammar.
874
e20e6a50
AD
875** Future Changes
876
877 In addition to the removal of the features announced in Bison 2.6, the
878 next major release will remove the "Temporary hack for adding a semicolon
879 to the user action", as announced in the release 2.5. Instead of:
880
881 exp: exp "+" exp { $$ = $1 + $3 };
882
883 write:
884
885 exp: exp "+" exp { $$ = $1 + $3; };
886
8617d87e
AD
887** Bug fixes
888
0e164d43
AD
889*** Type names are now properly escaped.
890
891*** glr.cc: set_debug_level and debug_level work as expected.
a4107f24 892
26313726
AD
893*** Stray @ or $ in actions
894
895 While Bison used to warn about stray $ or @ in action rules, it did not
896 for other actions such as printers, destructors, or initial actions. It
897 now does.
898
cd735a8c 899** Type names in actions
4982f078
AD
900
901 For consistency with rule actions, it is now possible to qualify $$ by a
cd735a8c 902 type-name in destructors, printers, and initial actions. For instance:
4982f078
AD
903
904 %printer { fprintf (yyo, "(%d, %f)", $<ival>$, $<fval>$); } <*> <>;
905
906 will display two values for each typed and untyped symbol (provided
cd735a8c 907 that YYSTYPE has both "ival" and "fval" fields).
60aa04a2 908
1505e8bb 909* Noteworthy changes in release 2.6 (2012-07-19) [stable]
0f11eec2 910
d0a30438 911** Future changes
9553083c 912
55d1006f
AD
913 The next major release of Bison will drop support for the following
914 deprecated features. Please report disagreements to bug-bison@gnu.org.
0f11eec2 915
aaf61036 916*** K&R C parsers
55d1006f
AD
917
918 Support for generating parsers in K&R C will be removed. Parsers
242cc08e 919 generated for C support ISO C90, and are tested with ISO C99 and ISO C11
55d1006f
AD
920 compilers.
921
258cddbc 922*** Features deprecated since Bison 1.875
0f11eec2 923
258cddbc
AD
924 The definitions of yystype and yyltype will be removed; use YYSTYPE and
925 YYLTYPE.
0f11eec2 926
258cddbc
AD
927 YYPARSE_PARAM and YYLEX_PARAM, deprecated in favor of %parse-param and
928 %lex-param, will no longer be supported.
929
930 Support for the preprocessor symbol YYERROR_VERBOSE will be removed, use
931 %error-verbose.
55d1006f
AD
932
933*** The generated header will be included (yacc.c)
0f11eec2
AD
934
935 Instead of duplicating the content of the generated header (definition of
55d1006f
AD
936 YYSTYPE, yyparse declaration etc.), the generated parser will include it,
937 as is already the case for GLR or C++ parsers. This change is deferred
938 because existing versions of ylwrap (e.g., Automake 1.12.1) do not support
939 it.
0f11eec2 940
c2425191 941** Generated Parser Headers
56ca3d8f 942
258cddbc 943*** Guards (yacc.c, glr.c, glr.cc)
c3e9f08f
AD
944
945 The generated headers are now guarded, as is already the case for C++
242cc08e 946 parsers (lalr1.cc). For instance, with --defines=foo.h:
c3e9f08f 947
e29f0771
AD
948 #ifndef YY_FOO_H
949 # define YY_FOO_H
950 ...
951 #endif /* !YY_FOO_H */
c3e9f08f 952
258cddbc 953*** New declarations (yacc.c, glr.c)
56ca3d8f
AD
954
955 The generated header now declares yydebug and yyparse. Both honor
956 --name-prefix=bar_, and yield
957
e29f0771 958 int bar_parse (void);
56ca3d8f
AD
959
960 rather than
961
e29f0771
AD
962 #define yyparse bar_parse
963 int yyparse (void);
56ca3d8f
AD
964
965 in order to facilitate the inclusion of several parser headers inside a
966 single compilation unit.
c3e9f08f 967
258cddbc
AD
968*** Exported symbols in C++
969
970 The symbols YYTOKEN_TABLE and YYERROR_VERBOSE, which were defined in the
971 header, are removed, as they prevent the possibility of including several
972 generated headers from a single compilation unit.
973
694af10c
AD
974*** YYLSP_NEEDED
975
976 For the same reasons, the undocumented and unused macro YYLSP_NEEDED is no
977 longer defined.
978
4b3847c3
AD
979** New %define variable: api.prefix
980
981 Now that the generated headers are more complete and properly protected
982 against multiple inclusions, constant names, such as YYSTYPE are a
983 problem. While yyparse and others are properly renamed by %name-prefix,
984 YYSTYPE, YYDEBUG and others have never been affected by it. Because it
985 would introduce backward compatibility issues in projects not expecting
986 YYSTYPE to be renamed, instead of changing the behavior of %name-prefix,
987 it is deprecated in favor of a new %define variable: api.prefix.
988
989 The following examples compares both:
990
991 %name-prefix "bar_" | %define api.prefix "bar_"
992 %token <ival> FOO %token <ival> FOO
993 %union { int ival; } %union { int ival; }
994 %% %%
995 exp: 'a'; exp: 'a';
996
997 bison generates:
998
999 #ifndef BAR_FOO_H #ifndef BAR_FOO_H
1000 # define BAR_FOO_H # define BAR_FOO_H
1001
1002 /* Enabling traces. */ /* Enabling traces. */
5f108727
AD
1003 # ifndef YYDEBUG | # ifndef BAR_DEBUG
1004 > # if defined YYDEBUG
1005 > # if YYDEBUG
1006 > # define BAR_DEBUG 1
1007 > # else
1008 > # define BAR_DEBUG 0
1009 > # endif
1010 > # else
1011 # define YYDEBUG 0 | # define BAR_DEBUG 0
1012 > # endif
1013 # endif | # endif
1014
1015 # if YYDEBUG | # if BAR_DEBUG
4b3847c3
AD
1016 extern int bar_debug; extern int bar_debug;
1017 # endif # endif
1018
1019 /* Tokens. */ /* Tokens. */
1020 # ifndef YYTOKENTYPE | # ifndef BAR_TOKENTYPE
1021 # define YYTOKENTYPE | # define BAR_TOKENTYPE
1022 enum yytokentype { | enum bar_tokentype {
1023 FOO = 258 FOO = 258
1024 }; };
1025 # endif # endif
1026
1027 #if ! defined YYSTYPE \ | #if ! defined BAR_STYPE \
1028 && ! defined YYSTYPE_IS_DECLARED | && ! defined BAR_STYPE_IS_DECLARED
1029 typedef union YYSTYPE | typedef union BAR_STYPE
1030 { {
1031 int ival; int ival;
1032 } YYSTYPE; | } BAR_STYPE;
1033 # define YYSTYPE_IS_DECLARED 1 | # define BAR_STYPE_IS_DECLARED 1
1034 #endif #endif
1035
1036 extern YYSTYPE bar_lval; | extern BAR_STYPE bar_lval;
1037
1038 int bar_parse (void); int bar_parse (void);
1039
1040 #endif /* !BAR_FOO_H */ #endif /* !BAR_FOO_H */
1041
dfaac272 1042* Noteworthy changes in release 2.5.1 (2012-06-05) [stable]
df6e3db0 1043
debe2c03 1044** Future changes:
765e1bd4 1045
e4ab1254 1046 The next major release will drop support for generating parsers in K&R C.
041308d0 1047
466b4cf2 1048** yacc.c: YYBACKUP works as expected.
ef51bfa7 1049
d834eca0 1050** glr.c improvements:
041308d0 1051
d834eca0 1052*** Location support is eliminated when not requested:
041308d0 1053
e4ab1254
AD
1054 GLR parsers used to include location-related code even when locations were
1055 not requested, and therefore not even usable.
378e917c 1056
d834eca0 1057*** __attribute__ is preserved:
d115aad9 1058
e4ab1254
AD
1059 __attribute__ is no longer disabled when __STRICT_ANSI__ is defined (i.e.,
1060 when -std is passed to GCC).
041308d0 1061
466b4cf2 1062** lalr1.java: several fixes:
041308d0 1063
e4ab1254
AD
1064 The Java parser no longer throws ArrayIndexOutOfBoundsException if the
1065 first token leads to a syntax error. Some minor clean ups.
041308d0 1066
22172d47 1067** Changes for C++:
ef51bfa7 1068
22172d47 1069*** C++11 compatibility:
ef51bfa7 1070
e4ab1254
AD
1071 C and C++ parsers use "nullptr" instead of "0" when __cplusplus is 201103L
1072 or higher.
936c88d1 1073
22172d47
AD
1074*** Header guards
1075
1076 The header files such as "parser.hh", "location.hh", etc. used a constant
1077 name for preprocessor guards, for instance:
1078
e29f0771
AD
1079 #ifndef BISON_LOCATION_HH
1080 # define BISON_LOCATION_HH
1081 ...
1082 #endif // !BISON_LOCATION_HH
22172d47
AD
1083
1084 The inclusion guard is now computed from "PREFIX/FILE-NAME", where lower
1085 case characters are converted to upper case, and series of
1086 non-alphanumerical characters are converted to an underscore.
1087
1088 With "bison -o lang++/parser.cc", "location.hh" would now include:
1089
e29f0771
AD
1090 #ifndef YY_LANG_LOCATION_HH
1091 # define YY_LANG_LOCATION_HH
1092 ...
1093 #endif // !YY_LANG_LOCATION_HH
22172d47
AD
1094
1095*** C++ locations:
936c88d1 1096
e4ab1254
AD
1097 The position and location constructors (and their initialize methods)
1098 accept new arguments for line and column. Several issues in the
1099 documentation were fixed.
936c88d1 1100
466b4cf2
AD
1101** liby is no longer asking for "rpl_fprintf" on some platforms.
1102
7e508a2b
AD
1103** Changes in the manual:
1104
1105*** %printer is documented
1106
e4ab1254
AD
1107 The "%printer" directive, supported since at least Bison 1.50, is finally
1108 documented. The "mfcalc" example is extended to demonstrate it.
7e508a2b 1109
e4ab1254
AD
1110 For consistency with the C skeletons, the C++ parsers now also support
1111 "yyoutput" (as an alias to "debug_stream ()").
7e508a2b
AD
1112
1113*** Several improvements have been made:
466b4cf2 1114
e4ab1254
AD
1115 The layout for grammar excerpts was changed to a more compact scheme.
1116 Named references are motivated. The description of the automaton
1117 description file (*.output) is updated to the current format. Incorrect
1118 index entries were fixed. Some other errors were fixed.
466b4cf2 1119
86b08b49
AD
1120** Building bison:
1121
1122*** Conflicting prototypes with recent/modified Flex.
466b4cf2 1123
e4ab1254
AD
1124 Fixed build problems with the current, unreleased, version of Flex, and
1125 some modified versions of 2.5.35, which have modified function prototypes.
466b4cf2 1126
8ef26c2a
AD
1127*** Warnings during the build procedure have been eliminated.
1128
1129*** Several portability problems in the test suite have been fixed:
466b4cf2 1130
e4ab1254
AD
1131 This includes warnings with some compilers, unexpected behavior of tools
1132 such as diff, warning messages from the test suite itself, etc.
466b4cf2 1133
91aadcc7 1134*** The install-pdf target works properly:
8ef26c2a 1135
e4ab1254
AD
1136 Running "make install-pdf" (or -dvi, -html, -info, and -ps) no longer
1137 halts in the middle of its course.
8ef26c2a 1138
28801043 1139* Changes in version 2.5 (2011-05-14):
50cca368 1140
82f3355e
JD
1141** Grammar symbol names can now contain non-initial dashes:
1142
1143 Consistently with directives (such as %error-verbose) and with
1144 %define variables (e.g. push-pull), grammar symbol names may contain
1145 dashes in any position except the beginning. This is a GNU
1146 extension over POSIX Yacc. Thus, use of this extension is reported
1147 by -Wyacc and rejected in Yacc mode (--yacc).
1148
f1b238df 1149** Named references:
66381412
AR
1150
1151 Historically, Yacc and Bison have supported positional references
1152 ($n, $$) to allow access to symbol values from inside of semantic
1153 actions code.
1154
1155 Starting from this version, Bison can also accept named references.
1156 When no ambiguity is possible, original symbol names may be used
1157 as named references:
1158
4b568fc0 1159 if_stmt : "if" cond_expr "then" then_stmt ';'
66381412
AR
1160 { $if_stmt = mk_if_stmt($cond_expr, $then_stmt); }
1161
1162 In the more common case, explicit names may be declared:
1163
4b568fc0 1164 stmt[res] : "if" expr[cond] "then" stmt[then] "else" stmt[else] ';'
66381412
AR
1165 { $res = mk_if_stmt($cond, $then, $else); }
1166
5b1ff423 1167 Location information is also accessible using @name syntax. When
66381412
AR
1168 accessing symbol names containing dots or dashes, explicit bracketing
1169 ($[sym.1]) must be used.
1170
5b1ff423 1171 These features are experimental in this version. More user feedback
66381412 1172 will help to stabilize them.
2bd435c3 1173 Contributed by Alex Rozenman.
66381412 1174
f1b238df 1175** IELR(1) and canonical LR(1):
eb45ef3b
JD
1176
1177 IELR(1) is a minimal LR(1) parser table generation algorithm. That
1178 is, given any context-free grammar, IELR(1) generates parser tables
7262f54f 1179 with the full language-recognition power of canonical LR(1) but with
f1b238df
JD
1180 nearly the same number of parser states as LALR(1). This reduction
1181 in parser states is often an order of magnitude. More importantly,
eb45ef3b
JD
1182 because canonical LR(1)'s extra parser states may contain duplicate
1183 conflicts in the case of non-LR(1) grammars, the number of conflicts
1184 for IELR(1) is often an order of magnitude less as well. This can
1185 significantly reduce the complexity of developing of a grammar.
1186
1187 Bison can now generate IELR(1) and canonical LR(1) parser tables in
1188 place of its traditional LALR(1) parser tables, which remain the
1189 default. You can specify the type of parser tables in the grammar
1190 file with these directives:
1191
cf499cff
JD
1192 %define lr.type lalr
1193 %define lr.type ielr
1194 %define lr.type canonical-lr
eb45ef3b 1195
7fceb615 1196 The default-reduction optimization in the parser tables can also be
e4ab1254
AD
1197 adjusted using "%define lr.default-reductions". For details on both
1198 of these features, see the new section "Tuning LR" in the Bison
7fceb615 1199 manual.
eb45ef3b
JD
1200
1201 These features are experimental. More user feedback will help to
1202 stabilize them.
1203
8458a411
AD
1204** LAC (Lookahead Correction) for syntax error handling
1205
1206 Contributed by Joel E. Denny.
fcf834f9
JD
1207
1208 Canonical LR, IELR, and LALR can suffer from a couple of problems
1209 upon encountering a syntax error. First, the parser might perform
1210 additional parser stack reductions before discovering the syntax
7fceb615 1211 error. Such reductions can perform user semantic actions that are
fcf834f9
JD
1212 unexpected because they are based on an invalid token, and they
1213 cause error recovery to begin in a different syntactic context than
1214 the one in which the invalid token was encountered. Second, when
7fceb615 1215 verbose error messages are enabled (with %error-verbose or the
e4ab1254 1216 obsolete "#define YYERROR_VERBOSE"), the expected token list in the
7fceb615
JD
1217 syntax error message can both contain invalid tokens and omit valid
1218 tokens.
fcf834f9
JD
1219
1220 The culprits for the above problems are %nonassoc, default
1221 reductions in inconsistent states, and parser state merging. Thus,
1222 IELR and LALR suffer the most. Canonical LR can suffer only if
1223 %nonassoc is used or if default reductions are enabled for
1224 inconsistent states.
1225
7fceb615
JD
1226 LAC is a new mechanism within the parsing algorithm that solves
1227 these problems for canonical LR, IELR, and LALR without sacrificing
1228 %nonassoc, default reductions, or state merging. When LAC is in
1229 use, canonical LR and IELR behave almost exactly the same for both
1230 syntactically acceptable and syntactically unacceptable input.
fcf834f9
JD
1231 While LALR still does not support the full language-recognition
1232 power of canonical LR and IELR, LAC at least enables LALR's syntax
1233 error handling to correctly reflect LALR's language-recognition
1234 power.
1235
1236 Currently, LAC is only supported for deterministic parsers in C.
1237 You can enable LAC with the following directive:
1238
1239 %define parse.lac full
1240
e4ab1254 1241 See the new section "LAC" in the Bison manual for additional
7fceb615 1242 details including a few caveats.
fcf834f9
JD
1243
1244 LAC is an experimental feature. More user feedback will help to
1245 stabilize it.
1246
d397d9f0 1247** %define improvements:
cf499cff 1248
f1b238df 1249*** Can now be invoked via the command line:
50cca368 1250
de5ab940 1251 Each of these command-line options
50cca368 1252
de5ab940
JD
1253 -D NAME[=VALUE]
1254 --define=NAME[=VALUE]
1255
1256 -F NAME[=VALUE]
1257 --force-define=NAME[=VALUE]
50cca368
JD
1258
1259 is equivalent to this grammar file declaration
1260
de5ab940 1261 %define NAME ["VALUE"]
50cca368 1262
de5ab940
JD
1263 except that the manner in which Bison processes multiple definitions
1264 for the same NAME differs. Most importantly, -F and --force-define
1265 quietly override %define, but -D and --define do not. For further
e4ab1254 1266 details, see the section "Bison Options" in the Bison manual.
50cca368 1267
f1b238df 1268*** Variables renamed:
67212941
JD
1269
1270 The following %define variables
1271
1272 api.push_pull
1273 lr.keep_unreachable_states
1274
1275 have been renamed to
1276
1277 api.push-pull
1278 lr.keep-unreachable-states
1279
1280 The old names are now deprecated but will be maintained indefinitely
1281 for backward compatibility.
1282
7262f54f 1283*** Values no longer need to be quoted in the grammar file:
cf499cff
JD
1284
1285 If a %define value is an identifier, it no longer needs to be placed
1286 within quotations marks. For example,
1287
1288 %define api.push-pull "push"
1289
1290 can be rewritten as
1291
1292 %define api.push-pull push
1293
d397d9f0 1294*** Unrecognized variables are now errors not warnings.
cdf3f113 1295
d397d9f0
JD
1296*** Multiple invocations for any variable is now an error not a warning.
1297
1298** Unrecognized %code qualifiers are now errors not warnings.
1299
1300** Character literals not of length one:
1301
1302 Previously, Bison quietly converted all character literals to length
1303 one. For example, without warning, Bison interpreted the operators in
1304 the following grammar to be the same token:
1305
1306 exp: exp '++'
1307 | exp '+' exp
1308 ;
1309
1310 Bison now warns when a character literal is not of length one. In
1311 some future release, Bison will start reporting an error instead.
1312
1313** Destructor calls fixed for lookaheads altered in semantic actions:
1314
1315 Previously for deterministic parsers in C, if a user semantic action
1316 altered yychar, the parser in some cases used the old yychar value to
1317 determine which destructor to call for the lookahead upon a syntax
1318 error or upon parser return. This bug has been fixed.
1319
1320** C++ parsers use YYRHSLOC:
1321
1322 Similarly to the C parsers, the C++ parsers now define the YYRHSLOC
1323 macro and use it in the default YYLLOC_DEFAULT. You are encouraged
e4ab1254
AD
1324 to use it. If, for instance, your location structure has "first"
1325 and "last" members, instead of
d397d9f0 1326
e29f0771
AD
1327 # define YYLLOC_DEFAULT(Current, Rhs, N) \
1328 do \
1329 if (N) \
1330 { \
1331 (Current).first = (Rhs)[1].location.first; \
1332 (Current).last = (Rhs)[N].location.last; \
1333 } \
1334 else \
1335 { \
1336 (Current).first = (Current).last = (Rhs)[0].location.last; \
1337 } \
1338 while (false)
d397d9f0
JD
1339
1340 use:
1341
e29f0771
AD
1342 # define YYLLOC_DEFAULT(Current, Rhs, N) \
1343 do \
1344 if (N) \
1345 { \
1346 (Current).first = YYRHSLOC (Rhs, 1).first; \
1347 (Current).last = YYRHSLOC (Rhs, N).last; \
1348 } \
1349 else \
1350 { \
1351 (Current).first = (Current).last = YYRHSLOC (Rhs, 0).last; \
1352 } \
1353 while (false)
d397d9f0
JD
1354
1355** YYLLOC_DEFAULT in C++:
1356
1357 The default implementation of YYLLOC_DEFAULT used to be issued in
1358 the header file. It is now output in the implementation file, after
1359 the user %code sections so that its #ifndef guard does not try to
1360 override the user's YYLLOC_DEFAULT if provided.
cdf3f113 1361
f1b238df 1362** YYFAIL now produces warnings and Java parsers no longer implement it:
4395a9ff
JD
1363
1364 YYFAIL has existed for many years as an undocumented feature of
1365 deterministic parsers in C generated by Bison. More recently, it was
1366 a documented feature of Bison's experimental Java parsers. As
1367 promised in Bison 2.4.2's NEWS entry, any appearance of YYFAIL in a
1368 semantic action now produces a deprecation warning, and Java parsers
1369 no longer implement YYFAIL at all. For further details, including a
1370 discussion of how to suppress C preprocessor warnings about YYFAIL
1371 being unused, see the Bison 2.4.2 NEWS entry.
1372
f1b238df 1373** Temporary hack for adding a semicolon to the user action:
197b82ba
JD
1374
1375 Previously, Bison appended a semicolon to every user action for
1376 reductions when the output language defaulted to C (specifically, when
1377 neither %yacc, %language, %skeleton, or equivalent command-line
1378 options were specified). This allowed actions such as
1379
1380 exp: exp "+" exp { $$ = $1 + $3 };
1381
1382 instead of
1383
1384 exp: exp "+" exp { $$ = $1 + $3; };
1385
1386 As a first step in removing this misfeature, Bison now issues a
1387 warning when it appends a semicolon. Moreover, in cases where Bison
1388 cannot easily determine whether a semicolon is needed (for example, an
1389 action ending with a cpp directive or a braced compound initializer),
1390 it no longer appends one. Thus, the C compiler might now complain
1391 about a missing semicolon where it did not before. Future releases of
1392 Bison will cease to append semicolons entirely.
1393
d2060f06
JD
1394** Verbose syntax error message fixes:
1395
e4ab1254 1396 When %error-verbose or the obsolete "#define YYERROR_VERBOSE" is
7fceb615
JD
1397 specified, syntax error messages produced by the generated parser
1398 include the unexpected token as well as a list of expected tokens.
1399 The effect of %nonassoc on these verbose messages has been corrected
1400 in two ways, but a more complete fix requires LAC, described above:
d2060f06
JD
1401
1402*** When %nonassoc is used, there can exist parser states that accept no
1403 tokens, and so the parser does not always require a lookahead token
1404 in order to detect a syntax error. Because no unexpected token or
1405 expected tokens can then be reported, the verbose syntax error
1406 message described above is suppressed, and the parser instead
e4ab1254 1407 reports the simpler message, "syntax error". Previously, this
d2060f06
JD
1408 suppression was sometimes erroneously triggered by %nonassoc when a
1409 lookahead was actually required. Now verbose messages are
1410 suppressed only when all previous lookaheads have already been
1411 shifted or discarded.
1412
1413*** Previously, the list of expected tokens erroneously included tokens
1414 that would actually induce a syntax error because conflicts for them
1415 were resolved with %nonassoc in the current parser state. Such
1416 tokens are now properly omitted from the list.
1417
1418*** Expected token lists are still often wrong due to state merging
fcf834f9
JD
1419 (from LALR or IELR) and default reductions, which can both add
1420 invalid tokens and subtract valid tokens. Canonical LR almost
1421 completely fixes this problem by eliminating state merging and
1422 default reductions. However, there is one minor problem left even
1423 when using canonical LR and even after the fixes above. That is,
1424 if the resolution of a conflict with %nonassoc appears in a later
1425 parser state than the one at which some syntax error is
1426 discovered, the conflicted token is still erroneously included in
1427 the expected token list. Bison's new LAC implementation,
1428 described above, eliminates this problem and the need for
1429 canonical LR. However, LAC is still experimental and is disabled
1430 by default.
53f036ce 1431
1a33f4f6
JD
1432** Java skeleton fixes:
1433
1434*** A location handling bug has been fixed.
1435
1436*** The top element of each of the value stack and location stack is now
1437 cleared when popped so that it can be garbage collected.
6771a463 1438
02803d55
JD
1439*** Parser traces now print the top element of the stack.
1440
86408959
JD
1441** -W/--warnings fixes:
1442
e4ab1254 1443*** Bison now properly recognizes the "no-" versions of categories:
86408959
JD
1444
1445 For example, given the following command line, Bison now enables all
1446 warnings except warnings for incompatibilities with POSIX Yacc:
1447
1448 bison -Wall,no-yacc gram.y
1449
786743d5
JD
1450*** Bison now treats S/R and R/R conflicts like other warnings:
1451
1452 Previously, conflict reports were independent of Bison's normal
1453 warning system. Now, Bison recognizes the warning categories
e4ab1254 1454 "conflicts-sr" and "conflicts-rr". This change has important
786743d5
JD
1455 consequences for the -W and --warnings command-line options. For
1456 example:
1457
1458 bison -Wno-conflicts-sr gram.y # S/R conflicts not reported
1459 bison -Wno-conflicts-rr gram.y # R/R conflicts not reported
1460 bison -Wnone gram.y # no conflicts are reported
1461 bison -Werror gram.y # any conflict is an error
1462
1463 However, as before, if the %expect or %expect-rr directive is
1464 specified, an unexpected number of conflicts is an error, and an
1465 expected number of conflicts is not reported, so -W and --warning
1466 then have no effect on the conflict report.
1467
e4ab1254 1468*** The "none" category no longer disables a preceding "error":
bf0e44e8
JD
1469
1470 For example, for the following command line, Bison now reports
1471 errors instead of warnings for incompatibilities with POSIX Yacc:
1472
1473 bison -Werror,none,yacc gram.y
1474
e4ab1254 1475*** The "none" category now disables all Bison warnings:
c39014ae 1476
e4ab1254 1477 Previously, the "none" category disabled only Bison warnings for
c39014ae
JD
1478 which there existed a specific -W/--warning category. However,
1479 given the following command line, Bison is now guaranteed to
1480 suppress all warnings:
1481
1482 bison -Wnone gram.y
1483
1f36f544
JD
1484** Precedence directives can now assign token number 0:
1485
1486 Since Bison 2.3b, which restored the ability of precedence
1487 directives to assign token numbers, doing so for token number 0 has
1488 produced an assertion failure. For example:
1489
1490 %left END 0
1491
1492 This bug has been fixed.
1493
64877e5e 1494* Changes in version 2.4.3 (2010-08-05):
8b9e021f 1495
2bfcac9a
JD
1496** Bison now obeys -Werror and --warnings=error for warnings about
1497 grammar rules that are useless in the parser due to conflicts.
1498
8b9e021f
JD
1499** Problems with spawning M4 on at least FreeBSD 8 and FreeBSD 9 have
1500 been fixed.
1501
4ad3921d
JD
1502** Failures in the test suite for GCC 4.5 have been fixed.
1503
06cb07d5
JD
1504** Failures in the test suite for some versions of Sun Studio C++ have
1505 been fixed.
1506
9b5049bd
JD
1507** Contrary to Bison 2.4.2's NEWS entry, it has been decided that
1508 warnings about undefined %prec identifiers will not be converted to
1509 errors in Bison 2.5. They will remain warnings, which should be
1510 sufficient for POSIX while avoiding backward compatibility issues.
1511
93d7dde9
JD
1512** Minor documentation fixes.
1513
e19a049c 1514* Changes in version 2.4.2 (2010-03-20):
74553c98 1515
f39ab286
JD
1516** Some portability problems that resulted in failures and livelocks
1517 in the test suite on some versions of at least Solaris, AIX, HP-UX,
e19a049c
JD
1518 RHEL4, and Tru64 have been addressed. As a result, fatal Bison
1519 errors should no longer cause M4 to report a broken pipe on the
f39ab286
JD
1520 affected platforms.
1521
e4ab1254 1522** "%prec IDENTIFIER" requires IDENTIFIER to be defined separately.
8bb3a2e7
JD
1523
1524 POSIX specifies that an error be reported for any identifier that does
1525 not appear on the LHS of a grammar rule and that is not defined by
1526 %token, %left, %right, or %nonassoc. Bison 2.3b and later lost this
1527 error report for the case when an identifier appears only after a
1528 %prec directive. It is now restored. However, for backward
1529 compatibility with recent Bison releases, it is only a warning for
1530 now. In Bison 2.5 and later, it will return to being an error.
9b5049bd
JD
1531 [Between the 2.4.2 and 2.4.3 releases, it was decided that this
1532 warning will not be converted to an error in Bison 2.5.]
8bb3a2e7 1533
d8911864
EB
1534** Detection of GNU M4 1.4.6 or newer during configure is improved.
1535
a603c6e0
JD
1536** Warnings from gcc's -Wundef option about undefined YYENABLE_NLS,
1537 YYLTYPE_IS_TRIVIAL, and __STRICT_ANSI__ in C/C++ parsers are now
1538 avoided.
c938d650 1539
98a345a2
JD
1540** %code is now a permanent feature.
1541
1542 A traditional Yacc prologue directive is written in the form:
1543
1544 %{CODE%}
1545
1546 To provide a more flexible alternative, Bison 2.3b introduced the
1547 %code directive with the following forms for C/C++:
1548
1549 %code {CODE}
1550 %code requires {CODE}
1551 %code provides {CODE}
1552 %code top {CODE}
1553
1554 These forms are now considered permanent features of Bison. See the
1555 %code entries in the section "Bison Declaration Summary" in the Bison
1556 manual for a summary of their functionality. See the section
1557 "Prologue Alternatives" for a detailed discussion including the
1558 advantages of %code over the traditional Yacc prologue directive.
1559
1560 Bison's Java feature as a whole including its current usage of %code
1561 is still considered experimental.
1562
1625df5b
JD
1563** YYFAIL is deprecated and will eventually be removed.
1564
1565 YYFAIL has existed for many years as an undocumented feature of
1566 deterministic parsers in C generated by Bison. Previously, it was
1567 documented for Bison's experimental Java parsers. YYFAIL is no longer
1568 documented for Java parsers and is formally deprecated in both cases.
1569 Users are strongly encouraged to migrate to YYERROR, which is
1570 specified by POSIX.
1571
1572 Like YYERROR, you can invoke YYFAIL from a semantic action in order to
1573 induce a syntax error. The most obvious difference from YYERROR is
1574 that YYFAIL will automatically invoke yyerror to report the syntax
1575 error so that you don't have to. However, there are several other
1576 subtle differences between YYERROR and YYFAIL, and YYFAIL suffers from
e4ab1254 1577 inherent flaws when %error-verbose or "#define YYERROR_VERBOSE" is
1625df5b
JD
1578 used. For a more detailed discussion, see:
1579
1580 http://lists.gnu.org/archive/html/bison-patches/2009-12/msg00024.html
1581
1582 The upcoming Bison 2.5 will remove YYFAIL from Java parsers, but
1583 deterministic parsers in C will continue to implement it. However,
1584 because YYFAIL is already flawed, it seems futile to try to make new
1585 Bison features compatible with it. Thus, during parser generation,
1586 Bison 2.5 will produce a warning whenever it discovers YYFAIL in a
1587 rule action. In a later release, YYFAIL will be disabled for
e4ab1254 1588 %error-verbose and "#define YYERROR_VERBOSE". Eventually, YYFAIL will
1625df5b
JD
1589 be removed altogether.
1590
1591 There exists at least one case where Bison 2.5's YYFAIL warning will
1592 be a false positive. Some projects add phony uses of YYFAIL and other
1593 Bison-defined macros for the sole purpose of suppressing C
1594 preprocessor warnings (from GCC cpp's -Wunused-macros, for example).
1595 To avoid Bison's future warning, such YYFAIL uses can be moved to the
e4ab1254 1596 epilogue (that is, after the second "%%") in the Bison input file. In
1625df5b
JD
1597 this release (2.4.2), Bison already generates its own code to suppress
1598 C preprocessor warnings for YYFAIL, so projects can remove their own
1599 phony uses of YYFAIL if compatibility with Bison releases prior to
1600 2.4.2 is not necessary.
1601
2755de8f
AD
1602** Internationalization.
1603
1604 Fix a regression introduced in Bison 2.4: Under some circumstances,
1605 message translations were not installed although supported by the
1606 host system.
1607
74553c98 1608* Changes in version 2.4.1 (2008-12-11):
c9ba9e59 1609
0ea583d2
AD
1610** In the GLR defines file, unexpanded M4 macros in the yylval and yylloc
1611 declarations have been fixed.
1979121c 1612
0ea583d2
AD
1613** Temporary hack for adding a semicolon to the user action.
1614
1615 Bison used to prepend a trailing semicolon at the end of the user
1616 action for reductions. This allowed actions such as
1617
1618 exp: exp "+" exp { $$ = $1 + $3 };
1619
1620 instead of
1621
1622 exp: exp "+" exp { $$ = $1 + $3; };
1623
e4ab1254 1624 Some grammars still depend on this "feature". Bison 2.4.1 restores
0ea583d2
AD
1625 the previous behavior in the case of C output (specifically, when
1626 neither %language or %skeleton or equivalent command-line options
1627 are used) to leave more time for grammars depending on the old
1628 behavior to be adjusted. Future releases of Bison will disable this
1629 feature.
1630
1631** A few minor improvements to the Bison manual.
c9ba9e59 1632
402b123d 1633* Changes in version 2.4 (2008-11-02):
7bd1665a 1634
402b123d 1635** %language is an experimental feature.
ed4d67dc
JD
1636
1637 We first introduced this feature in test release 2.3b as a cleaner
1638 alternative to %skeleton. Since then, we have discussed the possibility of
1639 modifying its effect on Bison's output file names. Thus, in this release,
1640 we consider %language to be an experimental feature that will likely evolve
1641 in future releases.
7bd1665a 1642
402b123d 1643** Forward compatibility with GNU M4 has been improved.
241fda7a 1644
402b123d 1645** Several bugs in the C++ skeleton and the experimental Java skeleton have been
241fda7a
JD
1646 fixed.
1647
402b123d 1648* Changes in version 2.3b (2008-05-27):
35fe0834 1649
402b123d 1650** The quotes around NAME that used to be required in the following directive
d9df47b6
JD
1651 are now deprecated:
1652
1653 %define NAME "VALUE"
1654
e4ab1254 1655** The directive "%pure-parser" is now deprecated in favor of:
d9df47b6
JD
1656
1657 %define api.pure
1658
1659 which has the same effect except that Bison is more careful to warn about
1660 unreasonable usage in the latter case.
1661
402b123d 1662** Push Parsing
c373bf8b
JD
1663
1664 Bison can now generate an LALR(1) parser in C with a push interface. That
e4ab1254
AD
1665 is, instead of invoking "yyparse", which pulls tokens from "yylex", you can
1666 push one token at a time to the parser using "yypush_parse", which will
c373bf8b
JD
1667 return to the caller after processing each token. By default, the push
1668 interface is disabled. Either of the following directives will enable it:
1669
1670 %define api.push_pull "push" // Just push; does not require yylex.
1671 %define api.push_pull "both" // Push and pull; requires yylex.
1672
e4ab1254 1673 See the new section "A Push Parser" in the Bison manual for details.
c373bf8b 1674
59da312b
JD
1675 The current push parsing interface is experimental and may evolve. More user
1676 feedback will help to stabilize it.
1677
402b123d 1678** The -g and --graph options now output graphs in Graphviz DOT format,
8e55b3aa
JD
1679 not VCG format. Like --graph, -g now also takes an optional FILE argument
1680 and thus cannot be bundled with other short options.
c373bf8b 1681
402b123d 1682** Java
59da312b
JD
1683
1684 Bison can now generate an LALR(1) parser in Java. The skeleton is
e4ab1254 1685 "data/lalr1.java". Consider using the new %language directive instead of
59da312b
JD
1686 %skeleton to select it.
1687
e4ab1254 1688 See the new section "Java Parsers" in the Bison manual for details.
59da312b
JD
1689
1690 The current Java interface is experimental and may evolve. More user
1691 feedback will help to stabilize it.
2bd435c3 1692 Contributed by Paolo Bonzini.
59da312b 1693
402b123d 1694** %language
59da312b
JD
1695
1696 This new directive specifies the programming language of the generated
d43f77e7
PB
1697 parser, which can be C (the default), C++, or Java. Besides the skeleton
1698 that Bison uses, the directive affects the names of the generated files if
1699 the grammar file's name ends in ".y".
59da312b 1700
402b123d 1701** XML Automaton Report
59da312b
JD
1702
1703 Bison can now generate an XML report of the LALR(1) automaton using the new
e4ab1254 1704 "--xml" option. The current XML schema is experimental and may evolve. More
59da312b 1705 user feedback will help to stabilize it.
2bd435c3 1706 Contributed by Wojciech Polak.
c373bf8b 1707
402b123d 1708** The grammar file may now specify the name of the parser header file using
c373bf8b
JD
1709 %defines. For example:
1710
1711 %defines "parser.h"
1712
402b123d 1713** When reporting useless rules, useless nonterminals, and unused terminals,
d80fb37a
JD
1714 Bison now employs the terms "useless in grammar" instead of "useless",
1715 "useless in parser" instead of "never reduced", and "unused in grammar"
1716 instead of "unused".
cff03fb2 1717
402b123d 1718** Unreachable State Removal
c373bf8b
JD
1719
1720 Previously, Bison sometimes generated parser tables containing unreachable
31984206
JD
1721 states. A state can become unreachable during conflict resolution if Bison
1722 disables a shift action leading to it from a predecessor state. Bison now:
75ad86ee
JD
1723
1724 1. Removes unreachable states.
1725
1726 2. Does not report any conflicts that appeared in unreachable states.
1727 WARNING: As a result, you may need to update %expect and %expect-rr
1728 directives in existing grammar files.
1729
1730 3. For any rule used only in such states, Bison now reports the rule as
cff03fb2 1731 "useless in parser due to conflicts".
75ad86ee 1732
31984206
JD
1733 This feature can be disabled with the following directive:
1734
1735 %define lr.keep_unreachable_states
1736
e4ab1254 1737 See the %define entry in the "Bison Declaration Summary" in the Bison manual
31984206
JD
1738 for further discussion.
1739
e4ab1254 1740** Lookahead Set Correction in the ".output" Report
b1cc23c4 1741
e4ab1254
AD
1742 When instructed to generate a ".output" file including lookahead sets
1743 (using "--report=lookahead", for example), Bison now prints each reduction's
88c78747
JD
1744 lookahead set only next to the associated state's one item that (1) is
1745 associated with the same rule as the reduction and (2) has its dot at the end
1746 of its RHS. Previously, Bison also erroneously printed the lookahead set
1747 next to all of the state's other items associated with the same rule. This
e4ab1254 1748 bug affected only the ".output" file and not the generated parser source
88c78747
JD
1749 code.
1750
e4ab1254 1751** --report-file=FILE is a new option to override the default ".output" file
59da312b 1752 name.
1bb2bd75 1753
e4ab1254 1754** The "=" that used to be required in the following directives is now
02975b9a
JD
1755 deprecated:
1756
1757 %file-prefix "parser"
1758 %name-prefix "c_"
1759 %output "parser.c"
1760
e4ab1254 1761** An Alternative to "%{...%}" -- "%code QUALIFIER {CODE}"
c373bf8b
JD
1762
1763 Bison 2.3a provided a new set of directives as a more flexible alternative to
8e0a5e9e
JD
1764 the traditional Yacc prologue blocks. Those have now been consolidated into
1765 a single %code directive with an optional qualifier field, which identifies
1766 the purpose of the code and thus the location(s) where Bison should generate
1767 it:
1768
e4ab1254
AD
1769 1. "%code {CODE}" replaces "%after-header {CODE}"
1770 2. "%code requires {CODE}" replaces "%start-header {CODE}"
1771 3. "%code provides {CODE}" replaces "%end-header {CODE}"
1772 4. "%code top {CODE}" replaces "%before-header {CODE}"
8e0a5e9e 1773
e4ab1254
AD
1774 See the %code entries in section "Bison Declaration Summary" in the Bison
1775 manual for a summary of the new functionality. See the new section "Prologue
1776 Alternatives" for a detailed discussion including the advantages of %code
8e0a5e9e
JD
1777 over the traditional Yacc prologues.
1778
1779 The prologue alternatives are experimental. More user feedback will help to
1780 determine whether they should become permanent features.
1781
402b123d 1782** Revised warning: unset or unused mid-rule values
17bd8a73
JD
1783
1784 Since Bison 2.2, Bison has warned about mid-rule values that are set but not
1785 used within any of the actions of the parent rule. For example, Bison warns
1786 about unused $2 in:
1787
1788 exp: '1' { $$ = 1; } '+' exp { $$ = $1 + $4; };
1789
1790 Now, Bison also warns about mid-rule values that are used but not set. For
1791 example, Bison warns about unset $$ in the mid-rule action in:
1792
1793 exp: '1' { $1 = 1; } '+' exp { $$ = $2 + $4; };
1794
1795 However, Bison now disables both of these warnings by default since they
1796 sometimes prove to be false alarms in existing grammars employing the Yacc
1797 constructs $0 or $-N (where N is some positive integer).
1798
e4ab1254
AD
1799 To enable these warnings, specify the option "--warnings=midrule-values" or
1800 "-W", which is a synonym for "--warnings=all".
17bd8a73 1801
e4ab1254 1802** Default %destructor or %printer with "<*>" or "<>"
c373bf8b
JD
1803
1804 Bison now recognizes two separate kinds of default %destructor's and
12e35840
JD
1805 %printer's:
1806
e4ab1254 1807 1. Place "<*>" in a %destructor/%printer symbol list to define a default
12e35840
JD
1808 %destructor/%printer for all grammar symbols for which you have formally
1809 declared semantic type tags.
1810
e4ab1254 1811 2. Place "<>" in a %destructor/%printer symbol list to define a default
12e35840
JD
1812 %destructor/%printer for all grammar symbols without declared semantic
1813 type tags.
1814
e4ab1254
AD
1815 Bison no longer supports the "%symbol-default" notation from Bison 2.3a.
1816 "<*>" and "<>" combined achieve the same effect with one exception: Bison no
12e35840
JD
1817 longer applies any %destructor to a mid-rule value if that mid-rule value is
1818 not actually ever referenced using either $$ or $n in a semantic action.
1819
85894313
JD
1820 The default %destructor's and %printer's are experimental. More user
1821 feedback will help to determine whether they should become permanent
1822 features.
1823
e4ab1254 1824 See the section "Freeing Discarded Symbols" in the Bison manual for further
12e35840
JD
1825 details.
1826
402b123d 1827** %left, %right, and %nonassoc can now declare token numbers. This is required
e4ab1254 1828 by POSIX. However, see the end of section "Operator Precedence" in the Bison
ab7f29f8
JD
1829 manual for a caveat concerning the treatment of literal strings.
1830
402b123d 1831** The nonfunctional --no-parser, -n, and %no-parser options have been
b1cc23c4
JD
1832 completely removed from Bison.
1833
402b123d 1834* Changes in version 2.3a, 2006-09-13:
742e4900 1835
402b123d 1836** Instead of %union, you can define and use your own union type
ddc8ede1
PE
1837 YYSTYPE if your grammar contains at least one <type> tag.
1838 Your YYSTYPE need not be a macro; it can be a typedef.
1839 This change is for compatibility with other Yacc implementations,
1840 and is required by POSIX.
1841
402b123d 1842** Locations columns and lines start at 1.
cd48d21d
AD
1843 In accordance with the GNU Coding Standards and Emacs.
1844
402b123d 1845** You may now declare per-type and default %destructor's and %printer's:
ec5479ce
JD
1846
1847 For example:
1848
b2a0b7ca
JD
1849 %union { char *string; }
1850 %token <string> STRING1
1851 %token <string> STRING2
1852 %type <string> string1
1853 %type <string> string2
1854 %union { char character; }
1855 %token <character> CHR
1856 %type <character> chr
1857 %destructor { free ($$); } %symbol-default
1858 %destructor { free ($$); printf ("%d", @$.first_line); } STRING1 string1
1859 %destructor { } <character>
1860
1861 guarantees that, when the parser discards any user-defined symbol that has a
e4ab1254
AD
1862 semantic type tag other than "<character>", it passes its semantic value to
1863 "free". However, when the parser discards a "STRING1" or a "string1", it
1864 also prints its line number to "stdout". It performs only the second
1865 "%destructor" in this case, so it invokes "free" only once.
ec5479ce 1866
85894313
JD
1867 [Although we failed to mention this here in the 2.3a release, the default
1868 %destructor's and %printer's were experimental, and they were rewritten in
1869 future versions.]
1870
e4ab1254
AD
1871** Except for LALR(1) parsers in C with POSIX Yacc emulation enabled (with "-y",
1872 "--yacc", or "%yacc"), Bison no longer generates #define statements for
b931235e
JD
1873 associating token numbers with token names. Removing the #define statements
1874 helps to sanitize the global namespace during preprocessing, but POSIX Yacc
1875 requires them. Bison still generates an enum for token names in all cases.
1876
402b123d 1877** Handling of traditional Yacc prologue blocks is now more consistent but
34f98f46 1878 potentially incompatible with previous releases of Bison.
9bc0dd67
JD
1879
1880 As before, you declare prologue blocks in your grammar file with the
e4ab1254 1881 "%{ ... %}" syntax. To generate the pre-prologue, Bison concatenates all
34f98f46
JD
1882 prologue blocks that you've declared before the first %union. To generate
1883 the post-prologue, Bison concatenates all prologue blocks that you've
ddc8ede1 1884 declared after the first %union.
9bc0dd67 1885
34f98f46 1886 Previous releases of Bison inserted the pre-prologue into both the header
9bc0dd67
JD
1887 file and the code file in all cases except for LALR(1) parsers in C. In the
1888 latter case, Bison inserted it only into the code file. For parsers in C++,
1889 the point of insertion was before any token definitions (which associate
1890 token numbers with names). For parsers in C, the point of insertion was
1891 after the token definitions.
1892
1893 Now, Bison never inserts the pre-prologue into the header file. In the code
1894 file, it always inserts it before the token definitions.
1895
402b123d 1896** Bison now provides a more flexible alternative to the traditional Yacc
34f98f46
JD
1897 prologue blocks: %before-header, %start-header, %end-header, and
1898 %after-header.
1899
1900 For example, the following declaration order in the grammar file reflects the
1901 order in which Bison will output these code blocks. However, you are free to
1902 declare these code blocks in your grammar file in whatever order is most
1903 convenient for you:
1904
1905 %before-header {
1906 /* Bison treats this block like a pre-prologue block: it inserts it into
1907 * the code file before the contents of the header file. It does *not*
1908 * insert it into the header file. This is a good place to put
1909 * #include's that you want at the top of your code file. A common
e4ab1254 1910 * example is '#include "system.h"'. */
34f98f46
JD
1911 }
1912 %start-header {
1913 /* Bison inserts this block into both the header file and the code file.
1914 * In both files, the point of insertion is before any Bison-generated
1915 * token, semantic type, location type, and class definitions. This is a
1916 * good place to define %union dependencies, for example. */
9bc0dd67
JD
1917 }
1918 %union {
34f98f46
JD
1919 /* Unlike the traditional Yacc prologue blocks, the output order for the
1920 * new %*-header blocks is not affected by their declaration position
1921 * relative to any %union in the grammar file. */
9bc0dd67 1922 }
34f98f46
JD
1923 %end-header {
1924 /* Bison inserts this block into both the header file and the code file.
1925 * In both files, the point of insertion is after the Bison-generated
1926 * definitions. This is a good place to declare or define public
1927 * functions or data structures that depend on the Bison-generated
1928 * definitions. */
9bc0dd67 1929 }
34f98f46
JD
1930 %after-header {
1931 /* Bison treats this block like a post-prologue block: it inserts it into
1932 * the code file after the contents of the header file. It does *not*
1933 * insert it into the header file. This is a good place to declare or
1934 * define internal functions or data structures that depend on the
1935 * Bison-generated definitions. */
1936 }
1937
1938 If you have multiple occurrences of any one of the above declarations, Bison
1939 will concatenate the contents in declaration order.
9bc0dd67 1940
85894313
JD
1941 [Although we failed to mention this here in the 2.3a release, the prologue
1942 alternatives were experimental, and they were rewritten in future versions.]
1943
e4ab1254 1944** The option "--report=look-ahead" has been changed to "--report=lookahead".
9e6e7ed2
PE
1945 The old spelling still works, but is not documented and may be removed
1946 in a future release.
742e4900 1947
402b123d 1948* Changes in version 2.3, 2006-06-05:
4ad3ed84 1949
e4ab1254 1950** GLR grammars should now use "YYRECOVERING ()" instead of "YYRECOVERING",
4ad3ed84
PE
1951 for compatibility with LALR(1) grammars.
1952
402b123d 1953** It is now documented that any definition of YYSTYPE or YYLTYPE should
4ad3ed84
PE
1954 be to a type name that does not contain parentheses or brackets.
1955
402b123d 1956* Changes in version 2.2, 2006-05-19:
193d7c70 1957
402b123d 1958** The distribution terms for all Bison-generated parsers now permit
193d7c70
PE
1959 using the parsers in nonfree programs. Previously, this permission
1960 was granted only for Bison-generated LALR(1) parsers in C.
5f4236a0 1961
402b123d 1962** %name-prefix changes the namespace name in C++ outputs.
aa08666d 1963
402b123d 1964** The C++ parsers export their token_type.
5f4236a0 1965
402b123d 1966** Bison now allows multiple %union declarations, and concatenates
d6ca7905
PE
1967 their contents together.
1968
402b123d 1969** New warning: unused values
4d7bc38c
PE
1970 Right-hand side symbols whose values are not used are reported,
1971 if the symbols have destructors. For instance:
affac613 1972
8f3596a6 1973 exp: exp "?" exp ":" exp { $1 ? $1 : $3; }
e9690142
JD
1974 | exp "+" exp
1975 ;
affac613 1976
8f3596a6
AD
1977 will trigger a warning about $$ and $5 in the first rule, and $3 in
1978 the second ($1 is copied to $$ by the default rule). This example
4e26c69e 1979 most likely contains three errors, and could be rewritten as:
affac613 1980
4e26c69e 1981 exp: exp "?" exp ":" exp
e9690142
JD
1982 { $$ = $1 ? $3 : $5; free ($1 ? $5 : $3); free ($1); }
1983 | exp "+" exp
1984 { $$ = $1 ? $1 : $3; if ($1) free ($3); }
1985 ;
affac613 1986
4e26c69e
PE
1987 However, if the original actions were really intended, memory leaks
1988 and all, the warnings can be suppressed by letting Bison believe the
1989 values are used, e.g.:
721be13c 1990
8f3596a6 1991 exp: exp "?" exp ":" exp { $1 ? $1 : $3; (void) ($$, $5); }
e9690142
JD
1992 | exp "+" exp { $$ = $1; (void) $3; }
1993 ;
721be13c 1994
84866159
AD
1995 If there are mid-rule actions, the warning is issued if no action
1996 uses it. The following triggers no warning: $1 and $3 are used.
1997
1998 exp: exp { push ($1); } '+' exp { push ($3); sum (); };
1999
721be13c
PE
2000 The warning is intended to help catching lost values and memory leaks.
2001 If a value is ignored, its associated memory typically is not reclaimed.
affac613 2002
402b123d 2003** %destructor vs. YYABORT, YYACCEPT, and YYERROR.
9d9b8b70
PE
2004 Destructors are now called when user code invokes YYABORT, YYACCEPT,
2005 and YYERROR, for all objects on the stack, other than objects
2006 corresponding to the right-hand side of the current rule.
a85284cf 2007
402b123d 2008** %expect, %expect-rr
035aa4a0
PE
2009 Incorrect numbers of expected conflicts are now actual errors,
2010 instead of warnings.
2011
402b123d 2012** GLR, YACC parsers.
4e26c69e
PE
2013 The %parse-params are available in the destructors (and the
2014 experimental printers) as per the documentation.
4b367315 2015
e4ab1254 2016** Bison now warns if it finds a stray "$" or "@" in an action.
ad6a9b97 2017
402b123d 2018** %require "VERSION"
4e26c69e
PE
2019 This specifies that the grammar file depends on features implemented
2020 in Bison version VERSION or higher.
b50d2359 2021
402b123d 2022** lalr1.cc: The token and value types are now class members.
e14d0ab6
AD
2023 The tokens were defined as free form enums and cpp macros. YYSTYPE
2024 was defined as a free form union. They are now class members:
e4ab1254
AD
2025 tokens are enumerations of the "yy::parser::token" struct, and the
2026 semantic values have the "yy::parser::semantic_type" type.
fb9712a9
AD
2027
2028 If you do not want or can update to this scheme, the directive
e4ab1254 2029 '%define "global_tokens_and_yystype" "1"' triggers the global
b50d2359
AD
2030 definition of tokens and YYSTYPE. This change is suitable both
2031 for previous releases of Bison, and this one.
fb9712a9 2032
b50d2359 2033 If you wish to update, then make sure older version of Bison will
e4ab1254 2034 fail using '%require "2.2"'.
fb9712a9 2035
402b123d 2036** DJGPP support added.
193d7c70 2037\f
402b123d 2038* Changes in version 2.1, 2005-09-16:
1ce59070 2039
402b123d 2040** The C++ lalr1.cc skeleton supports %lex-param.
e14d0ab6 2041
402b123d 2042** Bison-generated parsers now support the translation of diagnostics like
baf785db
PE
2043 "syntax error" into languages other than English. The default
2044 language is still English. For details, please see the new
0410a6e0
PE
2045 Internationalization section of the Bison manual. Software
2046 distributors should also see the new PACKAGING file. Thanks to
2047 Bruno Haible for this new feature.
1ce59070 2048
402b123d 2049** Wording in the Bison-generated parsers has been changed slightly to
1a059451
PE
2050 simplify translation. In particular, the message "memory exhausted"
2051 has replaced "parser stack overflow", as the old message was not
2052 always accurate for modern Bison-generated parsers.
2053
402b123d 2054** Destructors are now called when the parser aborts, for all symbols left
258b75ca
PE
2055 behind on the stack. Also, the start symbol is now destroyed after a
2056 successful parse. In both cases, the behavior was formerly inconsistent.
2057
402b123d 2058** When generating verbose diagnostics, Bison-generated parsers no longer
72f000b0
PE
2059 quote the literal strings associated with tokens. For example, for
2060 a syntax error associated with '%token NUM "number"' they might
2061 print 'syntax error, unexpected number' instead of 'syntax error,
2062 unexpected "number"'.
193d7c70 2063\f
402b123d 2064* Changes in version 2.0, 2004-12-25:
efeed023 2065
402b123d 2066** Possibly-incompatible changes
d7e14fc0 2067
82de6b0d
PE
2068 - Bison-generated parsers no longer default to using the alloca function
2069 (when available) to extend the parser stack, due to widespread
2070 problems in unchecked stack-overflow detection. You can "#define
2071 YYSTACK_USE_ALLOCA 1" to require the use of alloca, but please read
2072 the manual to determine safe values for YYMAXDEPTH in that case.
8dd162d3 2073
82de6b0d
PE
2074 - Error token location.
2075 During error recovery, the location of the syntax error is updated
2076 to cover the whole sequence covered by the error token: it includes
2077 the shifted symbols thrown away during the first part of the error
2078 recovery, and the lookahead rejected during the second part.
18d192f0 2079
82de6b0d
PE
2080 - Semicolon changes:
2081 . Stray semicolons are no longer allowed at the start of a grammar.
2082 . Semicolons are now required after in-grammar declarations.
e342c3be 2083
82de6b0d
PE
2084 - Unescaped newlines are no longer allowed in character constants or
2085 string literals. They were never portable, and GCC 3.4.0 has
2086 dropped support for them. Better diagnostics are now generated if
2087 forget a closing quote.
8dd162d3 2088
82de6b0d 2089 - NUL bytes are no longer allowed in Bison string literals, unfortunately.
f74b6f91 2090
402b123d 2091** New features
1452af69 2092
82de6b0d 2093 - GLR grammars now support locations.
4febdd96 2094
82de6b0d
PE
2095 - New directive: %initial-action.
2096 This directive allows the user to run arbitrary code (including
2097 initializing @$) from yyparse before parsing starts.
1452af69 2098
82de6b0d
PE
2099 - A new directive "%expect-rr N" specifies the expected number of
2100 reduce/reduce conflicts in GLR parsers.
1452af69 2101
e4ab1254 2102 - %token numbers can now be hexadecimal integers, e.g., "%token FOO 0x12d".
82de6b0d 2103 This is a GNU extension.
4febdd96 2104
e4ab1254 2105 - The option "--report=lookahead" was changed to "--report=look-ahead".
9e6e7ed2 2106 [However, this was changed back after 2.3.]
1452af69 2107
82de6b0d 2108 - Experimental %destructor support has been added to lalr1.cc.
1452af69 2109
82de6b0d
PE
2110 - New configure option --disable-yacc, to disable installation of the
2111 yacc command and -ly library introduced in 1.875 for POSIX conformance.
6040d338 2112
402b123d 2113** Bug fixes
d5a3fe37 2114
82de6b0d
PE
2115 - For now, %expect-count violations are now just warnings, not errors.
2116 This is for compatibility with Bison 1.75 and earlier (when there are
2117 reduce/reduce conflicts) and with Bison 1.30 and earlier (when there
2118 are too many or too few shift/reduce conflicts). However, in future
2119 versions of Bison we plan to improve the %expect machinery so that
2120 these violations will become errors again.
3473d0f8 2121
82de6b0d
PE
2122 - Within Bison itself, numbers (e.g., goto numbers) are no longer
2123 arbitrarily limited to 16-bit counts.
d600ee67 2124
82de6b0d 2125 - Semicolons are now allowed before "|" in grammar rules, as POSIX requires.
d600ee67 2126\f
402b123d 2127* Changes in version 1.875, 2003-01-01:
963fcc17 2128
402b123d 2129** The documentation license has been upgraded to version 1.2
dc546b0f 2130 of the GNU Free Documentation License.
75eb3bc4 2131
402b123d 2132** syntax error processing
75eb3bc4 2133
dc546b0f
PE
2134 - In Yacc-style parsers YYLLOC_DEFAULT is now used to compute error
2135 locations too. This fixes bugs in error-location computation.
75eb3bc4 2136
dc546b0f
PE
2137 - %destructor
2138 It is now possible to reclaim the memory associated to symbols
2139 discarded during error recovery. This feature is still experimental.
20daca06 2140
dc546b0f
PE
2141 - %error-verbose
2142 This new directive is preferred over YYERROR_VERBOSE.
74724a70 2143
dc546b0f
PE
2144 - #defining yyerror to steal internal variables is discouraged.
2145 It is not guaranteed to work forever.
d1de5372 2146
402b123d 2147** POSIX conformance
d1de5372 2148
dc546b0f
PE
2149 - Semicolons are once again optional at the end of grammar rules.
2150 This reverts to the behavior of Bison 1.33 and earlier, and improves
2151 compatibility with Yacc.
74724a70 2152
e4ab1254
AD
2153 - "parse error" -> "syntax error"
2154 Bison now uniformly uses the term "syntax error"; formerly, the code
2155 and manual sometimes used the term "parse error" instead. POSIX
2156 requires "syntax error" in diagnostics, and it was thought better to
dc546b0f 2157 be consistent.
74724a70 2158
dc546b0f
PE
2159 - The documentation now emphasizes that yylex and yyerror must be
2160 declared before use. C99 requires this.
d1de5372 2161
dc546b0f
PE
2162 - Bison now parses C99 lexical constructs like UCNs and
2163 backslash-newline within C escape sequences, as POSIX 1003.1-2001 requires.
d1de5372 2164
dc546b0f
PE
2165 - File names are properly escaped in C output. E.g., foo\bar.y is
2166 output as "foo\\bar.y".
6780ca7a 2167
dc546b0f 2168 - Yacc command and library now available
e4ab1254 2169 The Bison distribution now installs a "yacc" command, as POSIX requires.
dc546b0f
PE
2170 Also, Bison now installs a small library liby.a containing
2171 implementations of Yacc-compatible yyerror and main functions.
2172 This library is normally not useful, but POSIX requires it.
6e649e65 2173
dc546b0f 2174 - Type clashes now generate warnings, not errors.
6e649e65 2175
dc546b0f
PE
2176 - If the user does not define YYSTYPE as a macro, Bison now declares it
2177 using typedef instead of defining it as a macro.
2178 For consistency, YYLTYPE is also declared instead of defined.
9501dc6e 2179
402b123d 2180** Other compatibility issues
886a425c 2181
e4ab1254
AD
2182 - %union directives can now have a tag before the "{", e.g., the
2183 directive "%union foo {...}" now generates the C code
2184 "typedef union foo { ... } YYSTYPE;"; this is for Yacc compatibility.
2185 The default union tag is "YYSTYPE", for compatibility with Solaris 9 Yacc.
2186 For consistency, YYLTYPE's struct tag is now "YYLTYPE" not "yyltype".
dc546b0f 2187 This is for compatibility with both Yacc and Bison 1.35.
72f889cc 2188
e4ab1254 2189 - ";" is output before the terminating "}" of an action, for
dc546b0f 2190 compatibility with Bison 1.35.
886a425c 2191
dc546b0f 2192 - Bison now uses a Yacc-style format for conflict reports, e.g.,
e4ab1254 2193 "conflicts: 2 shift/reduce, 1 reduce/reduce".
437c2d80 2194
e4ab1254 2195 - "yystype" and "yyltype" are now obsolescent macros instead of being
dc546b0f
PE
2196 typedefs or tags; they are no longer documented and are planned to be
2197 withdrawn in a future release.
2a8d363a 2198
402b123d 2199** GLR parser notes
2a8d363a 2200
dc546b0f
PE
2201 - GLR and inline
2202 Users of Bison have to decide how they handle the portability of the
e4ab1254 2203 C keyword "inline".
959e5f51 2204
e4ab1254
AD
2205 - "parsing stack overflow..." -> "parser stack overflow"
2206 GLR parsers now report "parser stack overflow" as per the Bison manual.
900c5db5 2207
18ad57b3
AD
2208** %parse-param and %lex-param
2209 The macros YYPARSE_PARAM and YYLEX_PARAM provide a means to pass
2210 additional context to yyparse and yylex. They suffer from several
2211 shortcomings:
2212
2213 - a single argument only can be added,
2214 - their types are weak (void *),
242cc08e 2215 - this context is not passed to ancillary functions such as yyerror,
18ad57b3
AD
2216 - only yacc.c parsers support them.
2217
2218 The new %parse-param/%lex-param directives provide a more precise control.
2219 For instance:
2220
2221 %parse-param {int *nastiness}
2222 %lex-param {int *nastiness}
2223 %parse-param {int *randomness}
2224
2225 results in the following signatures:
2226
2227 int yylex (int *nastiness);
2228 int yyparse (int *nastiness, int *randomness);
2229
2230 or, if both %pure-parser and %locations are used:
2231
2232 int yylex (YYSTYPE *lvalp, YYLTYPE *llocp, int *nastiness);
2233 int yyparse (int *nastiness, int *randomness);
2234
402b123d 2235** Bison now warns if it detects conflicting outputs to the same file,
e4ab1254 2236 e.g., it generates a warning for "bison -d -o foo.h foo.y" since
dc546b0f 2237 that command outputs both code and header to foo.h.
6e40b4eb 2238
402b123d 2239** #line in output files
dc546b0f 2240 - --no-line works properly.
6e40b4eb 2241
402b123d 2242** Bison can no longer be built by a K&R C compiler; it requires C89 or
6e40b4eb
AD
2243 later to be built. This change originally took place a few versions
2244 ago, but nobody noticed until we recently asked someone to try
2245 building Bison with a K&R C compiler.
d600ee67 2246\f
402b123d 2247* Changes in version 1.75, 2002-10-14:
7933f2b5 2248
402b123d 2249** Bison should now work on 64-bit hosts.
7933f2b5 2250
402b123d 2251** Indonesian translation thanks to Tedi Heriyanto.
7933f2b5 2252
402b123d 2253** GLR parsers
f50adbbd
AD
2254 Fix spurious parse errors.
2255
402b123d 2256** Pure parsers
f50adbbd
AD
2257 Some people redefine yyerror to steal yyparse' private variables.
2258 Reenable this trick until an official feature replaces it.
2259
402b123d 2260** Type Clashes
d90c934c
AD
2261 In agreement with POSIX and with other Yaccs, leaving a default
2262 action is valid when $$ is untyped, and $1 typed:
2263
e9690142 2264 untyped: ... typed;
d90c934c
AD
2265
2266 but the converse remains an error:
2267
e9690142 2268 typed: ... untyped;
d90c934c 2269
402b123d 2270** Values of mid-rule actions
d90c934c
AD
2271 The following code:
2272
e9690142 2273 foo: { ... } { $$ = $1; } ...
d90c934c
AD
2274
2275 was incorrectly rejected: $1 is defined in the second mid-rule
2276 action, and is equal to the $$ of the first mid-rule action.
d600ee67 2277\f
402b123d 2278* Changes in version 1.50, 2002-10-04:
adc8c848 2279
402b123d 2280** GLR parsing
676385e2
PH
2281 The declaration
2282 %glr-parser
2283 causes Bison to produce a Generalized LR (GLR) parser, capable of handling
2284 almost any context-free grammar, ambiguous or not. The new declarations
e8832397 2285 %dprec and %merge on grammar rules allow parse-time resolution of
676385e2
PH
2286 ambiguities. Contributed by Paul Hilfinger.
2287
7933f2b5 2288 Unfortunately Bison 1.50 does not work properly on 64-bit hosts
420f93c8
PE
2289 like the Alpha, so please stick to 32-bit hosts for now.
2290
402b123d 2291** Output Directory
8c165d89 2292 When not in Yacc compatibility mode, when the output file was not
e4ab1254
AD
2293 specified, running "bison foo/bar.y" created "foo/bar.c". It
2294 now creates "bar.c".
8c165d89 2295
402b123d 2296** Undefined token
007a50a4 2297 The undefined token was systematically mapped to 2 which prevented
e88dbdbf 2298 the use of 2 by the user. This is no longer the case.
007a50a4 2299
402b123d 2300** Unknown token numbers
e88dbdbf 2301 If yylex returned an out of range value, yyparse could die. This is
007a50a4
AD
2302 no longer the case.
2303
402b123d 2304** Error token
e88dbdbf 2305 According to POSIX, the error token must be 256.
23c5a174
AD
2306 Bison extends this requirement by making it a preference: *if* the
2307 user specified that one of her tokens is numbered 256, then error
2308 will be mapped onto another number.
2309
402b123d 2310** Verbose error messages
e4ab1254 2311 They no longer report "..., expecting error or..." for states where
217598da
AD
2312 error recovery is possible.
2313
402b123d 2314** End token
e4ab1254 2315 Defaults to "$end" instead of "$".
217598da 2316
402b123d 2317** Error recovery now conforms to documentation and to POSIX
68cd8af3
PE
2318 When a Bison-generated parser encounters a syntax error, it now pops
2319 the stack until it finds a state that allows shifting the error
2320 token. Formerly, it popped the stack until it found a state that
2321 allowed some non-error action other than a default reduction on the
2322 error token. The new behavior has long been the documented behavior,
2323 and has long been required by POSIX. For more details, please see
337116ba
PE
2324 Paul Eggert, "Reductions during Bison error handling" (2002-05-20)
2325 <http://lists.gnu.org/archive/html/bug-bison/2002-05/msg00038.html>.
68cd8af3 2326
402b123d 2327** Traces
5504898e
AD
2328 Popped tokens and nonterminals are now reported.
2329
402b123d 2330** Larger grammars
a861a339
PE
2331 Larger grammars are now supported (larger token numbers, larger grammar
2332 size (= sum of the LHS and RHS lengths), larger LALR tables).
2333 Formerly, many of these numbers ran afoul of 16-bit limits;
2334 now these limits are 32 bits on most hosts.
355e7c1c 2335
402b123d 2336** Explicit initial rule
643a5994
AD
2337 Bison used to play hacks with the initial rule, which the user does
2338 not write. It is now explicit, and visible in the reports and
2339 graphs as rule 0.
23c5a174 2340
402b123d 2341** Useless rules
643a5994 2342 Before, Bison reported the useless rules, but, although not used,
77714df2 2343 included them in the parsers. They are now actually removed.
23c5a174 2344
402b123d 2345** Useless rules, useless nonterminals
6b98e4b5
AD
2346 They are now reported, as a warning, with their locations.
2347
402b123d 2348** Rules never reduced
e8832397
AD
2349 Rules that can never be reduced because of conflicts are now
2350 reported.
2351
e4ab1254 2352** Incorrect "Token not used"
11652ab3
AD
2353 On a grammar such as
2354
e29f0771
AD
2355 %token useless useful
2356 %%
2357 exp: '0' %prec useful;
11652ab3
AD
2358
2359 where a token was used to set the precedence of the last rule,
e4ab1254 2360 bison reported both "useful" and "useless" as useless tokens.
11652ab3 2361
402b123d 2362** Revert the C++ namespace changes introduced in 1.31
77714df2 2363 as they caused too many portability hassles.
0179dd65 2364
402b123d 2365** Default locations
b2d52318
AD
2366 By an accident of design, the default computation of @$ was
2367 performed after another default computation was performed: @$ = @1.
2368 The latter is now removed: YYLLOC_DEFAULT is fully responsible of
2369 the computation of @$.
adc8c848 2370
402b123d 2371** Token end-of-file
b7c49edf
AD
2372 The token end of file may be specified by the user, in which case,
2373 the user symbol is used in the reports, the graphs, and the verbose
e4ab1254 2374 error messages instead of "$end", which remains being the default.
b7c49edf 2375 For instance
e29f0771 2376 %token MYEOF 0
b7c49edf 2377 or
e29f0771 2378 %token MYEOF 0 "end of file"
fdbcd8e2 2379
402b123d 2380** Semantic parser
fdbcd8e2
AD
2381 This old option, which has been broken for ages, is removed.
2382
402b123d 2383** New translations
a861a339 2384 Brazilian Portuguese, thanks to Alexandre Folle de Menezes.
84614e13
AD
2385 Croatian, thanks to Denis Lackovic.
2386
402b123d 2387** Incorrect token definitions
e4ab1254
AD
2388 When given
2389 %token 'a' "A"
2390 bison used to output
2391 #define 'a' 65
b87f8b21 2392
402b123d 2393** Token definitions as enums
77714df2
AD
2394 Tokens are output both as the traditional #define's, and, provided
2395 the compiler supports ANSI C or is a C++ compiler, as enums.
e88dbdbf 2396 This lets debuggers display names instead of integers.
77714df2 2397
402b123d 2398** Reports
ec3bc396
AD
2399 In addition to --verbose, bison supports --report=THINGS, which
2400 produces additional information:
b408954b
AD
2401 - itemset
2402 complete the core item sets with their closure
e4ab1254 2403 - lookahead [changed to "look-ahead" in 1.875e through 2.3, but changed back]
9e6e7ed2 2404 explicitly associate lookahead tokens to items
b408954b
AD
2405 - solved
2406 describe shift/reduce conflicts solving.
2407 Bison used to systematically output this information on top of
2408 the report. Solved conflicts are now attached to their states.
ec3bc396 2409
402b123d 2410** Type clashes
9af3fbce
AD
2411 Previous versions don't complain when there is a type clash on
2412 the default action if the rule has a mid-rule action, such as in:
2413
e29f0771
AD
2414 %type <foo> bar
2415 %%
2416 bar: '0' {} '0';
9af3fbce
AD
2417
2418 This is fixed.
a861a339 2419
402b123d 2420** GNU M4 is now required when using Bison.
f987e9d2 2421\f
402b123d 2422* Changes in version 1.35, 2002-03-25:
76551463 2423
402b123d 2424** C Skeleton
76551463
AD
2425 Some projects use Bison's C parser with C++ compilers, and define
2426 YYSTYPE as a class. The recent adjustment of C parsers for data
2427 alignment and 64 bit architectures made this impossible.
2428
2429 Because for the time being no real solution for C++ parser
2430 generation exists, kludges were implemented in the parser to
2431 maintain this use. In the future, when Bison has C++ parsers, this
2432 kludge will be disabled.
2433
2434 This kludge also addresses some C++ problems when the stack was
2435 extended.
76551463 2436\f
402b123d 2437* Changes in version 1.34, 2002-03-12:
76551463 2438
402b123d 2439** File name clashes are detected
76551463 2440 $ bison foo.y -d -o foo.x
e4ab1254 2441 fatal error: header and parser would both be named "foo.x"
76551463 2442
e4ab1254 2443** A missing ";" at the end of a rule triggers a warning
76551463
AD
2444 In accordance with POSIX, and in agreement with other
2445 Yacc implementations, Bison will mandate this semicolon in the near
2446 future. This eases the implementation of a Bison parser of Bison
2447 grammars by making this grammar LALR(1) instead of LR(2). To
2448 facilitate the transition, this release introduces a warning.
2449
402b123d 2450** Revert the C++ namespace changes introduced in 1.31, as they caused too
76551463
AD
2451 many portability hassles.
2452
402b123d 2453** DJGPP support added.
76551463 2454
402b123d 2455** Fix test suite portability problems.
76551463 2456\f
402b123d 2457* Changes in version 1.33, 2002-02-07:
76551463 2458
402b123d 2459** Fix C++ issues
76551463
AD
2460 Groff could not be compiled for the definition of size_t was lacking
2461 under some conditions.
2462
402b123d 2463** Catch invalid @n
76551463
AD
2464 As is done with $n.
2465\f
402b123d 2466* Changes in version 1.32, 2002-01-23:
76551463 2467
402b123d 2468** Fix Yacc output file names
76551463 2469
402b123d 2470** Portability fixes
76551463 2471
402b123d 2472** Italian, Dutch translations
76551463 2473\f
402b123d 2474* Changes in version 1.31, 2002-01-14:
52d1aeee 2475
402b123d 2476** Many Bug Fixes
52d1aeee 2477
402b123d 2478** GNU Gettext and %expect
52d1aeee
MA
2479 GNU Gettext asserts 10 s/r conflicts, but there are 7. Now that
2480 Bison dies on incorrect %expectations, we fear there will be
2481 too many bug reports for Gettext, so _for the time being_, %expect
e4ab1254 2482 does not trigger an error when the input file is named "plural.y".
52d1aeee 2483
402b123d 2484** Use of alloca in parsers
52d1aeee
MA
2485 If YYSTACK_USE_ALLOCA is defined to 0, then the parsers will use
2486 malloc exclusively. Since 1.29, but was not NEWS'ed.
2487
2488 alloca is used only when compiled with GCC, to avoid portability
2489 problems as on AIX.
2490
402b123d 2491** yyparse now returns 2 if memory is exhausted; formerly it dumped core.
b47dbebe 2492
402b123d 2493** When the generated parser lacks debugging code, YYDEBUG is now 0
52d1aeee
MA
2494 (as POSIX requires) instead of being undefined.
2495
402b123d 2496** User Actions
52d1aeee
MA
2497 Bison has always permitted actions such as { $$ = $1 }: it adds the
2498 ending semicolon. Now if in Yacc compatibility mode, the semicolon
2499 is no longer output: one has to write { $$ = $1; }.
2500
402b123d 2501** Better C++ compliance
52d1aeee 2502 The output parsers try to respect C++ namespaces.
76551463 2503 [This turned out to be a failed experiment, and it was reverted later.]
52d1aeee 2504
402b123d 2505** Reduced Grammars
52d1aeee
MA
2506 Fixed bugs when reporting useless nonterminals.
2507
402b123d 2508** 64 bit hosts
52d1aeee
MA
2509 The parsers work properly on 64 bit hosts.
2510
402b123d 2511** Error messages
52d1aeee
MA
2512 Some calls to strerror resulted in scrambled or missing error messages.
2513
402b123d 2514** %expect
52d1aeee
MA
2515 When the number of shift/reduce conflicts is correct, don't issue
2516 any warning.
2517
402b123d 2518** The verbose report includes the rule line numbers.
52d1aeee 2519
402b123d 2520** Rule line numbers are fixed in traces.
52d1aeee 2521
402b123d 2522** Swedish translation
52d1aeee 2523
402b123d 2524** Parse errors
52d1aeee
MA
2525 Verbose parse error messages from the parsers are better looking.
2526 Before: parse error: unexpected `'/'', expecting `"number"' or `'-'' or `'(''
2527 Now: parse error: unexpected '/', expecting "number" or '-' or '('
2528
402b123d 2529** Fixed parser memory leaks.
52d1aeee
MA
2530 When the generated parser was using malloc to extend its stacks, the
2531 previous allocations were not freed.
2532
402b123d 2533** Fixed verbose output file.
52d1aeee
MA
2534 Some newlines were missing.
2535 Some conflicts in state descriptions were missing.
2536
402b123d 2537** Fixed conflict report.
52d1aeee
MA
2538 Option -v was needed to get the result.
2539
402b123d 2540** %expect
52d1aeee
MA
2541 Was not used.
2542 Mismatches are errors, not warnings.
2543
402b123d 2544** Fixed incorrect processing of some invalid input.
52d1aeee 2545
402b123d 2546** Fixed CPP guards: 9foo.h uses BISON_9FOO_H instead of 9FOO_H.
52d1aeee 2547
402b123d 2548** Fixed some typos in the documentation.
52d1aeee 2549
402b123d 2550** %token MY_EOF 0 is supported.
52d1aeee
MA
2551 Before, MY_EOF was silently renumbered as 257.
2552
402b123d 2553** doc/refcard.tex is updated.
52d1aeee 2554
402b123d 2555** %output, %file-prefix, %name-prefix.
52d1aeee
MA
2556 New.
2557
402b123d 2558** --output
e4ab1254 2559 New, aliasing "--output-file".
52d1aeee 2560\f
402b123d 2561* Changes in version 1.30, 2001-10-26:
342b8b6e 2562
e4ab1254
AD
2563** "--defines" and "--graph" have now an optional argument which is the
2564 output file name. "-d" and "-g" do not change; they do not take any
342b8b6e
AD
2565 argument.
2566
e4ab1254 2567** "%source_extension" and "%header_extension" are removed, failed
342b8b6e
AD
2568 experiment.
2569
402b123d 2570** Portability fixes.
f987e9d2 2571\f
402b123d 2572* Changes in version 1.29, 2001-09-07:
342b8b6e 2573
402b123d 2574** The output file does not define const, as this caused problems when used
342b8b6e
AD
2575 with common autoconfiguration schemes. If you still use ancient compilers
2576 that lack const, compile with the equivalent of the C compiler option
e4ab1254 2577 "-Dconst=". Autoconf's AC_C_CONST macro provides one way to do this.
342b8b6e 2578
e4ab1254 2579** Added "-g" and "--graph".
f87a2205 2580
402b123d 2581** The Bison manual is now distributed under the terms of the GNU FDL.
f2b5126e 2582
402b123d 2583** The input and the output files has automatically a similar extension.
234a3be3 2584
402b123d 2585** Russian translation added.
f87a2205 2586
402b123d 2587** NLS support updated; should hopefully be less troublesome.
f87a2205 2588
402b123d 2589** Added the old Bison reference card.
c33638bb 2590
e4ab1254 2591** Added "--locations" and "%locations".
6deb4447 2592
e4ab1254 2593** Added "-S" and "--skeleton".
cd5bd6ac 2594
e4ab1254 2595** "%raw", "-r", "--raw" is disabled.
62ab6972 2596
402b123d 2597** Special characters are escaped when output. This solves the problems
cd5bd6ac
AD
2598 of the #line lines with path names including backslashes.
2599
402b123d 2600** New directives.
e4ab1254
AD
2601 "%yacc", "%fixed_output_files", "%defines", "%no_parser", "%verbose",
2602 "%debug", "%source_extension" and "%header_extension".
f987e9d2 2603
402b123d 2604** @$
f987e9d2 2605 Automatic location tracking.
f87a2205 2606\f
402b123d 2607* Changes in version 1.28, 1999-07-06:
d2e00347 2608
402b123d 2609** Should compile better now with K&R compilers.
d2e00347 2610
402b123d 2611** Added NLS.
d2e00347 2612
402b123d 2613** Fixed a problem with escaping the double quote character.
d2e00347 2614
402b123d 2615** There is now a FAQ.
d2e00347 2616\f
402b123d 2617* Changes in version 1.27:
5c31c3c2 2618
402b123d 2619** The make rule which prevented bison.simple from being created on
5c31c3c2
JT
2620 some systems has been fixed.
2621\f
402b123d 2622* Changes in version 1.26:
4be07551 2623
7e508a2b 2624** Bison now uses Automake.
4be07551 2625
402b123d 2626** New mailing lists: <bug-bison@gnu.org> and <help-bison@gnu.org>.
4be07551 2627
402b123d 2628** Token numbers now start at 257 as previously documented, not 258.
4be07551 2629
402b123d 2630** Bison honors the TMPDIR environment variable.
4be07551 2631
402b123d 2632** A couple of buffer overruns have been fixed.
f51dbca1 2633
402b123d 2634** Problems when closing files should now be reported.
f51dbca1 2635
402b123d 2636** Generated parsers should now work even on operating systems which do
f51dbca1 2637 not provide alloca().
4be07551 2638\f
402b123d 2639* Changes in version 1.25, 1995-10-16:
df8878c5 2640
402b123d 2641** Errors in the input grammar are not fatal; Bison keeps reading
df8878c5 2642the grammar file, and reports all the errors found in it.
8c44d3ec 2643
402b123d 2644** Tokens can now be specified as multiple-character strings: for
df8878c5 2645example, you could use "<=" for a token which looks like <=, instead
7e508a2b 2646of choosing a name like LESSEQ.
df8878c5 2647
402b123d 2648** The %token_table declaration says to write a table of tokens (names
df8878c5
RS
2649and numbers) into the parser file. The yylex function can use this
2650table to recognize multiple-character string tokens, or for other
2651purposes.
2652
402b123d 2653** The %no_lines declaration says not to generate any #line preprocessor
df8878c5
RS
2654directives in the parser file.
2655
402b123d 2656** The %raw declaration says to use internal Bison token numbers, not
df8878c5
RS
2657Yacc-compatible token numbers, when token names are defined as macros.
2658
402b123d 2659** The --no-parser option produces the parser tables without including
df8878c5
RS
2660the parser engine; a project can now use its own parser engine.
2661The actions go into a separate file called NAME.act, in the form of
2662a switch statement body.
2663\f
402b123d 2664* Changes in version 1.23:
6780ca7a 2665
4d019228
DM
2666The user can define YYPARSE_PARAM as the name of an argument to be
2667passed into yyparse. The argument should have type void *. It should
2668actually point to an object. Grammar actions can access the variable
2669by casting it to the proper pointer type.
6780ca7a 2670
6780ca7a 2671Line numbers in output file corrected.
6780ca7a 2672\f
402b123d 2673* Changes in version 1.22:
6780ca7a
DM
2674
2675--help option added.
6780ca7a 2676\f
402b123d 2677* Changes in version 1.20:
6780ca7a
DM
2678
2679Output file does not redefine const for C++.
9f4503d6 2680
76551463
AD
2681-----
2682
7d6bad19 2683Copyright (C) 1995-2013 Free Software Foundation, Inc.
76551463 2684
74553c98 2685This file is part of Bison, the GNU Parser Generator.
76551463 2686
f16b0819 2687This program is free software: you can redistribute it and/or modify
76551463 2688it under the terms of the GNU General Public License as published by
f16b0819
PE
2689the Free Software Foundation, either version 3 of the License, or
2690(at your option) any later version.
76551463 2691
f16b0819 2692This program is distributed in the hope that it will be useful,
76551463
AD
2693but WITHOUT ANY WARRANTY; without even the implied warranty of
2694MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
2695GNU General Public License for more details.
2696
2697You should have received a copy of the GNU General Public License
f16b0819 2698along with this program. If not, see <http://www.gnu.org/licenses/>.
7e508a2b
AD
2699
2700 LocalWords: yacc YYBACKUP glr GCC lalr ArrayIndexOutOfBoundsException nullptr
2701 LocalWords: cplusplus liby rpl fprintf mfcalc Wyacc stmt cond expr mk sym lr
2702 LocalWords: IELR ielr Lookahead YYERROR nonassoc LALR's api lookaheads yychar
2703 LocalWords: destructor lookahead YYRHSLOC YYLLOC Rhs ifndef YYFAIL cpp sr rr
2704 LocalWords: preprocessor initializer Wno Wnone Werror FreeBSD prec livelocks
2705 LocalWords: Solaris AIX UX RHEL Tru LHS gcc's Wundef YYENABLE NLS YYLTYPE VCG
2706 LocalWords: yyerror cpp's Wunused yylval yylloc prepend yyparse yylex yypush
2707 LocalWords: Graphviz xml nonterminals midrule destructor's YYSTYPE typedef ly
2708 LocalWords: CHR chr printf stdout namespace preprocessing enum pre include's
2709 LocalWords: YYRECOVERING nonfree destructors YYABORT YYACCEPT params enums de
2710 LocalWords: struct yystype DJGPP lex param Haible NUM alloca YYSTACK NUL goto
2711 LocalWords: YYMAXDEPTH Unescaped UCNs YYLTYPE's yyltype typedefs inline Yaccs
2712 LocalWords: Heriyanto Reenable dprec Hilfinger Eggert MYEOF Folle Menezes EOF
242cc08e 2713 LocalWords: Lackovic define's itemset Groff Gettext malloc NEWS'ed YYDEBUG YY
7e508a2b 2714 LocalWords: namespaces strerror const autoconfiguration Dconst Autoconf's FDL
242cc08e
AD
2715 LocalWords: Automake TMPDIR LESSEQ ylwrap endif yydebug YYTOKEN YYLSP ival hh
2716 LocalWords: extern YYTOKENTYPE TOKENTYPE yytokentype tokentype STYPE lval pdf
dcb366b1 2717 LocalWords: lang yyoutput dvi html ps POSIX lvalp llocp Wother nterm arg init
1282c124
AD
2718 LocalWords: TOK calc yyo fval Wconflicts parsers yystackp yyval yynerrs
2719 LocalWords: Théophile Ranquet Santet fno fnone stype associativity Tolmer
2720 LocalWords: Wprecedence Rassoul Wempty Paolo Bonzini parser's Michiel loc
2721 LocalWords: redeclaration sval fcaret reentrant XSLT xsl Wmaybe yyvsp Tedi
2722 LocalWords: pragmas noreturn untyped Rozenman unexpanded Wojciech Polak
6574576c 2723 LocalWords: Alexandre MERCHANTABILITY yytype
7e508a2b
AD
2724
2725Local Variables:
2726mode: outline
e4ab1254 2727fill-column: 76
7e508a2b 2728End: