]> git.saurik.com Git - bison.git/blame - NEWS
package: bump to 2014
[bison.git] / NEWS
CommitLineData
ed7658fe 1GNU Bison NEWS
3af4feb2 2
0bd5ee5f
AD
3* Noteworthy changes in release ?.? (????-??-??) [?]
4
461f1516
AD
5** Bug fixes
6
7*** Redeclarations are reported in proper order
8
9 On
10
11 %token FOO "foo"
12 %printer {} "foo"
13 %printer {} FOO
14
15 bison used to report:
16
17 /tmp/foo.yy:2.10-11: error: %printer redeclaration for FOO
18 %printer {} "foo"
19 ^^
20 /tmp/foo.yy:3.10-11: previous declaration
21 %printer {} FOO
22 ^^
23
24 Now, the "previous" declaration is always the first one.
25
0bd5ee5f 26
b167e7ba
AD
27** Documentation
28
29 Bison now installs various files in its docdir (which defaults to
30 '/usr/local/share/doc/bison'), including the three fully blown examples
31 extracted from the documentation:
32
33 - rpcalc
34 Reverse polish calculator, a simple introductory example.
35 - mfcalc
36 Multi-function Calc, a calculator with memory and functions and located
37 error messages.
38 - calc++
39 a calculator in C++ using variant support and token constructors.
40
a3efd13e 41* Noteworthy changes in release 3.0.2 (2013-12-05) [stable]
6d94eebb 42
7b0ca050
AD
43** Bug fixes
44
184b42c8
AD
45*** Generated source files when errors are reported
46
47 When warnings are issued and -Werror is set, bison would still generate
48 the source files (*.c, *.h...). As a consequence, some runs of "make"
49 could fail the first time, but not the second (as the files were generated
50 anyway).
51
52 This is fixed: bison no longer generates this source files, but, of
53 course, still produces the various reports (*.output, *.xml, etc.).
54
be29c71d
AD
55*** %empty is used in reports
56
57 Empty right-hand sides are denoted by '%empty' in all the reports (text,
58 dot, XML and formats derived from it).
59
7b0ca050
AD
60*** YYERROR and variants
61
62 When C++ variant support is enabled, an error triggered via YYERROR, but
63 not caught via error recovery, resulted in a double deletion.
6d94eebb 64
c2ecada3 65* Noteworthy changes in release 3.0.1 (2013-11-12) [stable]
e386b50f 66
0a244a22
AD
67** Bug fixes
68
39bace5d
AD
69*** Errors in caret diagnostics
70
71 On some platforms, some errors could result in endless diagnostics.
e4678430
AD
72
73*** Fixes of the -Werror option
74
75 Options such as "-Werror -Wno-error=foo" were still turning "foo"
76 diagnostics into errors instead of warnings. This is fixed.
77
78 Actually, for consistency with GCC, "-Wno-error=foo -Werror" now also
79 leaves "foo" diagnostics as warnings. Similarly, with "-Werror=foo
80 -Wno-error", "foo" diagnostics are now errors.
e386b50f 81
071863b3
AD
82*** GLR Predicates
83
84 As demonstrated in the documentation, one can now leave spaces between
85 "%?" and its "{".
86
265640d5
AD
87*** Installation
88
89 The yacc.1 man page is no longer installed if --disable-yacc was
90 specified.
91
39bace5d
AD
92*** Fixes in the test suite
93
94 Bugs and portability issues.
95
534497f5 96* Noteworthy changes in release 3.0 (2013-07-25) [stable]
facb910c 97
8458a411 98** WARNING: Future backward-incompatibilities!
597afd73 99
597afd73
AD
100 Like other GNU packages, Bison will start using some of the C99 features
101 for its own code, especially the definition of variables after statements.
102 The generated C parsers still aim at C90.
103
8458a411 104** Backward incompatible changes
47db7ed1
AD
105
106*** Obsolete features
107
40bb6f78
AD
108 Support for YYFAIL is removed (deprecated in Bison 2.4.2): use YYERROR.
109
110 Support for yystype and yyltype is removed (deprecated in Bison 1.875):
111 use YYSTYPE and YYLTYPE.
112
113 Support for YYLEX_PARAM and YYPARSE_PARAM is removed (deprecated in Bison
114 1.875): use %lex-param, %parse-param, or %param.
47db7ed1 115
05e25f23
AD
116 Missing semicolons at the end of actions are no longer added (as announced
117 in the release 2.5).
118
1fa19a76
AD
119*** Use of YACC='bison -y'
120
121 TL;DR: With Autoconf <= 2.69, pass -Wno-yacc to (AM_)YFLAGS if you use
122 Bison extensions.
123
124 Traditional Yacc generates 'y.tab.c' whatever the name of the input file.
125 Therefore Makefiles written for Yacc expect 'y.tab.c' (and possibly
126 'y.tab.h' and 'y.outout') to be generated from 'foo.y'.
127
128 To this end, for ages, AC_PROG_YACC, Autoconf's macro to look for an
129 implementation of Yacc, was using Bison as 'bison -y'. While it does
130 ensure compatible output file names, it also enables warnings for
131 incompatibilities with POSIX Yacc. In other words, 'bison -y' triggers
132 warnings for Bison extensions.
133
134 Autoconf 2.70+ fixes this incompatibility by using YACC='bison -o y.tab.c'
135 (which also generates 'y.tab.h' and 'y.output' when needed).
136 Alternatively, disable Yacc warnings by passing '-Wno-yacc' to your Yacc
137 flags (YFLAGS, or AM_YFLAGS with Automake).
138
597afd73
AD
139** Bug fixes
140
c21e515e 141*** The epilogue is no longer affected by internal #defines (glr.c)
597afd73
AD
142
143 The glr.c skeleton uses defines such as #define yylval (yystackp->yyval) in
144 generated code. These weren't properly undefined before the inclusion of
145 the user epilogue, so functions such as the following were butchered by the
146 preprocessor expansion:
147
148 int yylex (YYSTYPE *yylval);
149
6c7022f7 150 This is fixed: yylval, yynerrs, yychar, and yylloc are now valid
597afd73
AD
151 identifiers for user-provided variables.
152
f0f95a50
AD
153*** stdio.h is no longer needed when locations are enabled (yacc.c)
154
155 Changes in Bison 2.7 introduced a dependency on FILE and fprintf when
156 locations are enabled. This is fixed.
157
bb4b189b
AD
158*** Warnings about useless %pure-parser/%define api.pure are restored
159
597afd73
AD
160** Diagnostics reported by Bison
161
162 Most of these features were contributed by Théophile Ranquet and Victor
163 Santet.
73370a9d 164
016426c1
TR
165*** Carets
166
167 Version 2.7 introduced caret errors, for a prettier output. These are now
168 activated by default. The old format can still be used by invoking Bison
169 with -fno-caret (or -fnone).
170
fec5f3c0
AD
171 Some error messages that reproduced excerpts of the grammar are now using
172 the caret information only. For instance on:
173
174 %%
175 exp: 'a' | 'a';
176
177 Bison 2.7 reports:
178
179 in.y: warning: 1 reduce/reduce conflict [-Wconflicts-rr]
180 in.y:2.12-14: warning: rule useless in parser due to conflicts: exp: 'a' [-Wother]
181
182 Now bison reports:
183
184 in.y: warning: 1 reduce/reduce conflict [-Wconflicts-rr]
185 in.y:2.12-14: warning: rule useless in parser due to conflicts [-Wother]
186 exp: 'a' | 'a';
187 ^^^
188
189 and "bison -fno-caret" reports:
190
191 in.y: warning: 1 reduce/reduce conflict [-Wconflicts-rr]
192 in.y:2.12-14: warning: rule useless in parser due to conflicts [-Wother]
193
1048a1c9 194*** Enhancements of the -Werror option
518e8830 195
1048a1c9 196 The -Werror=CATEGORY option is now recognized, and will treat specified
d949eefd
AD
197 warnings as errors. The warnings need not have been explicitly activated
198 using the -W option, this is similar to what GCC 4.7 does.
1048a1c9
AD
199
200 For example, given the following command line, Bison will treat both
d949eefd 201 warnings related to POSIX Yacc incompatibilities and S/R conflicts as
1048a1c9
AD
202 errors (and only those):
203
204 $ bison -Werror=yacc,error=conflicts-sr input.y
205
206 If no categories are specified, -Werror will make all active warnings into
207 errors. For example, the following line does the same the previous example:
208
209 $ bison -Werror -Wnone -Wyacc -Wconflicts-sr input.y
210
211 (By default -Wconflicts-sr,conflicts-rr,deprecated,other is enabled.)
212
213 Note that the categories in this -Werror option may not be prefixed with
214 "no-". However, -Wno-error[=CATEGORY] is valid.
215
216 Note that -y enables -Werror=yacc. Therefore it is now possible to require
217 Yacc-like behavior (e.g., always generate y.tab.c), but to report
218 incompatibilities as warnings: "-y -Wno-error=yacc".
518e8830 219
46bdb8ec 220*** The display of warnings is now richer
73370a9d 221
46bdb8ec 222 The option that controls a given warning is now displayed:
73370a9d 223
46bdb8ec 224 foo.y:4.6: warning: type clash on default action: <foo> != <bar> [-Wother]
73370a9d 225
46bdb8ec
TR
226 In the case of warnings treated as errors, the prefix is changed from
227 "warning: " to "error: ", and the suffix is displayed, in a manner similar
d949eefd 228 to GCC, as [-Werror=CATEGORY].
1048a1c9 229
46bdb8ec
TR
230 For instance, where the previous version of Bison would report (and exit
231 with failure):
1048a1c9
AD
232
233 bison: warnings being treated as errors
46bdb8ec 234 input.y:1.1: warning: stray ',' treated as white space
1048a1c9 235
46bdb8ec 236 it now reports:
1048a1c9 237
1048a1c9
AD
238 input.y:1.1: error: stray ',' treated as white space [-Werror=other]
239
240*** Deprecated constructs
241
242 The new 'deprecated' warning category flags obsolete constructs whose
243 support will be discontinued. It is enabled by default. These warnings
244 used to be reported as 'other' warnings.
245
73370a9d 246*** Useless semantic types
9641b918
VS
247
248 Bison now warns about useless (uninhabited) semantic types. Since
249 semantic types are not declared to Bison (they are defined in the opaque
250 %union structure), it is %printer/%destructor directives about useless
251 types that trigger the warning:
252
253 %token <type1> term
254 %type <type2> nterm
255 %printer {} <type1> <type3>
256 %destructor {} <type2> <type4>
257 %%
258 nterm: term { $$ = $1; };
259
260 3.28-34: warning: type <type3> is used, but is not associated to any symbol
261 4.28-34: warning: type <type4> is used, but is not associated to any symbol
262
31557b9e 263*** Undefined but unused symbols
b921d92f 264
31557b9e
AD
265 Bison used to raise an error for undefined symbols that are not used in
266 the grammar. This is now only a warning.
b921d92f
VS
267
268 %printer {} symbol1
269 %destructor {} symbol2
31557b9e 270 %type <type> symbol3
b921d92f
VS
271 %%
272 exp: "a";
273
73370a9d 274*** Useless destructors or printers
ea9a35c6
VS
275
276 Bison now warns about useless destructors or printers. In the following
277 example, the printer for <type1>, and the destructor for <type2> are
278 useless: all symbols of <type1> (token1) already have a printer, and all
279 symbols of type <type2> (token2) already have a destructor.
280
281 %token <type1> token1
282 <type2> token2
283 <type3> token3
284 <type4> token4
285 %printer {} token1 <type1> <type3>
286 %destructor {} token2 <type2> <type4>
287
d87ea54c
AD
288*** Conflicts
289
290 The warnings and error messages about shift/reduce and reduce/reduce
291 conflicts have been normalized. For instance on the following foo.y file:
292
293 %glr-parser
294 %%
295 exp: exp '+' exp | '0' | '0';
296
297 compare the previous version of bison:
298
299 $ bison foo.y
300 foo.y: conflicts: 1 shift/reduce, 2 reduce/reduce
301 $ bison -Werror foo.y
302 bison: warnings being treated as errors
303 foo.y: conflicts: 1 shift/reduce, 2 reduce/reduce
304
305 with the new behavior:
306
307 $ bison foo.y
308 foo.y: warning: 1 shift/reduce conflict [-Wconflicts-sr]
309 foo.y: warning: 2 reduce/reduce conflicts [-Wconflicts-rr]
310 $ bison -Werror foo.y
9503b0a4
TR
311 foo.y: error: 1 shift/reduce conflict [-Werror=conflicts-sr]
312 foo.y: error: 2 reduce/reduce conflicts [-Werror=conflicts-rr]
d87ea54c
AD
313
314 When %expect or %expect-rr is used, such as with bar.y:
315
316 %expect 0
317 %glr-parser
318 %%
319 exp: exp '+' exp | '0' | '0';
320
321 Former behavior:
322
323 $ bison bar.y
324 bar.y: conflicts: 1 shift/reduce, 2 reduce/reduce
325 bar.y: expected 0 shift/reduce conflicts
326 bar.y: expected 0 reduce/reduce conflicts
327
328 New one:
329
330 $ bison bar.y
9503b0a4
TR
331 bar.y: error: shift/reduce conflicts: 1 found, 0 expected
332 bar.y: error: reduce/reduce conflicts: 2 found, 0 expected
d87ea54c 333
f24695ef
AD
334** Incompatibilities with POSIX Yacc
335
336 The 'yacc' category is no longer part of '-Wall', enable it explicitly
337 with '-Wyacc'.
338
2055a44e
AD
339** Additional yylex/yyparse arguments
340
6dc4663d
AD
341 The new directive %param declares additional arguments to both yylex and
342 yyparse. The %lex-param, %parse-param, and %param directives support one
343 or more arguments. Instead of
2055a44e 344
e436fa67
AD
345 %lex-param {arg1_type *arg1}
346 %lex-param {arg2_type *arg2}
347 %parse-param {arg1_type *arg1}
348 %parse-param {arg2_type *arg2}
2055a44e
AD
349
350 one may now declare
351
e436fa67 352 %param {arg1_type *arg1} {arg2_type *arg2}
2055a44e 353
630a0218
AD
354** Types of values for %define variables
355
356 Bison used to make no difference between '%define foo bar' and '%define
357 foo "bar"'. The former is now called a 'keyword value', and the latter a
358 'string value'. A third kind was added: 'code values', such as '%define
359 foo {bar}'.
360
361 Keyword variables are used for fixed value sets, e.g.,
362
363 %define lr.type lalr
364
365 Code variables are used for value in the target language, e.g.,
366
367 %define api.value.type {struct semantic_type}
368
369 String variables are used remaining cases, e.g. file names.
370
2a6b66c5 371** Variable api.token.prefix
99c08fb6 372
2a6b66c5 373 The variable api.token.prefix changes the way tokens are identified in
99c08fb6
AD
374 the generated files. This is especially useful to avoid collisions
375 with identifiers in the target language. For instance
376
e436fa67 377 %token FILE for ERROR
630a0218 378 %define api.token.prefix {TOK_}
e436fa67
AD
379 %%
380 start: FILE for ERROR;
99c08fb6
AD
381
382 will generate the definition of the symbols TOK_FILE, TOK_for, and
383 TOK_ERROR in the generated sources. In particular, the scanner must
384 use these prefixed token names, although the grammar itself still
385 uses the short names (as in the sample rule given above).
386
6574576c
AD
387** Variable api.value.type
388
389 This new %define variable supersedes the #define macro YYSTYPE. The use
390 of YYSTYPE is discouraged. In particular, #defining YYSTYPE *and* either
391 using %union or %defining api.value.type results in undefined behavior.
392
393 Either define api.value.type, or use "%union":
394
395 %union
396 {
397 int ival;
398 char *sval;
399 }
400 %token <ival> INT "integer"
401 %token <sval> STRING "string"
402 %printer { fprintf (yyo, "%d", $$); } <ival>
403 %destructor { free ($$); } <sval>
404
405 /* In yylex(). */
406 yylval.ival = 42; return INT;
407 yylval.sval = "42"; return STRING;
408
630a0218
AD
409 The %define variable api.value.type supports both keyword and code values.
410
411 The keyword value 'union' means that the user provides genuine types, not
435575cb 412 union member names such as "ival" and "sval" above (WARNING: will fail if
1fa19a76 413 -y/--yacc/%yacc is enabled).
6574576c 414
435575cb 415 %define api.value.type union
6574576c
AD
416 %token <int> INT "integer"
417 %token <char *> STRING "string"
418 %printer { fprintf (yyo, "%d", $$); } <int>
419 %destructor { free ($$); } <char *>
420
421 /* In yylex(). */
422 yylval.INT = 42; return INT;
423 yylval.STRING = "42"; return STRING;
424
435575cb
AD
425 The keyword value variant is somewhat equivalent, but for C++ special
426 provision is made to allow classes to be used (more about this below).
6574576c 427
435575cb 428 %define api.value.type variant
6574576c
AD
429 %token <int> INT "integer"
430 %token <std::string> STRING "string"
431
630a0218 432 Code values (in braces) denote user defined types. This is where YYSTYPE
435575cb 433 used to be used.
6574576c
AD
434
435 %code requires
436 {
437 struct my_value
438 {
439 enum
440 {
441 is_int, is_string
442 } kind;
443 union
444 {
445 int ival;
446 char *sval;
447 } u;
448 };
449 }
435575cb 450 %define api.value.type {struct my_value}
6574576c
AD
451 %token <u.ival> INT "integer"
452 %token <u.sval> STRING "string"
453 %printer { fprintf (yyo, "%d", $$); } <u.ival>
454 %destructor { free ($$); } <u.sval>
455
456 /* In yylex(). */
457 yylval.u.ival = 42; return INT;
458 yylval.u.sval = "42"; return STRING;
459
31b850d2
AD
460** Variable parse.error
461
1f77b2e0
AD
462 This variable controls the verbosity of error messages. The use of the
463 %error-verbose directive is deprecated in favor of "%define parse.error
464 verbose".
31b850d2 465
c21e515e
AD
466** Renamed %define variables
467
468 The following variables have been renamed for consistency. Backward
469 compatibility is ensured, but upgrading is recommended.
470
471 lr.default-reductions -> lr.default-reduction
472 lr.keep-unreachable-states -> lr.keep-unreachable-state
473 namespace -> api.namespace
474 stype -> api.value.type
475
ca2a6d15
PH
476** Semantic predicates
477
597afd73
AD
478 Contributed by Paul Hilfinger.
479
1f77b2e0
AD
480 The new, experimental, semantic-predicate feature allows actions of the
481 form "%?{ BOOLEAN-EXPRESSION }", which cause syntax errors (as for
ca2a6d15 482 YYERROR) if the expression evaluates to 0, and are evaluated immediately
1f77b2e0
AD
483 in GLR parsers, rather than being deferred. The result is that they allow
484 the programmer to prune possible parses based on the values of run-time
485 expressions.
ca2a6d15 486
d1400569
AD
487** The directive %expect-rr is now an error in non GLR mode
488
489 It used to be an error only if used in non GLR mode, _and_ if there are
490 reduce/reduce conflicts.
491
5202b6ac 492** Tokens are numbered in their order of appearance
93561c21 493
5202b6ac
VT
494 Contributed by Valentin Tolmer.
495
496 With '%token A B', A had a number less than the one of B. However,
497 precedence declarations used to generate a reversed order. This is now
498 fixed, and introducing tokens with any of %token, %left, %right,
499 %precedence, or %nonassoc yields the same result.
500
501 When mixing declarations of tokens with a litteral character (e.g., 'a')
502 or with an identifier (e.g., B) in a precedence declaration, Bison
503 numbered the litteral characters first. For example
504
505 %right A B 'c' 'd'
506
507 would lead to the tokens declared in this order: 'c' 'd' A B. Again, the
508 input order is now preserved.
509
510 These changes were made so that one can remove useless precedence and
511 associativity declarations (i.e., map %nonassoc, %left or %right to
512 %precedence, or to %token) and get exactly the same output.
93561c21 513
cc2235ac
VT
514** Useless precedence and associativity
515
d2f9ae18
AD
516 Contributed by Valentin Tolmer.
517
1282c124 518 When developing and maintaining a grammar, useless associativity and
cc2235ac
VT
519 precedence directives are common. They can be a nuisance: new ambiguities
520 arising are sometimes masked because their conflicts are resolved due to
521 the extra precedence or associativity information. Furthermore, it can
522 hinder the comprehension of a new grammar: one will wonder about the role
523 of a precedence, where in fact it is useless. The following changes aim
524 at detecting and reporting these extra directives.
525
526*** Precedence warning category
527
528 A new category of warning, -Wprecedence, was introduced. It flags the
529 useless precedence and associativity directives.
530
531*** Useless associativity
532
533 Bison now warns about symbols with a declared associativity that is never
534 used to resolve conflicts. In that case, using %precedence is sufficient;
535 the parsing tables will remain unchanged. Solving these warnings may raise
536 useless precedence warnings, as the symbols no longer have associativity.
537 For example:
538
539 %left '+'
540 %left '*'
541 %%
542 exp:
1282c124
AD
543 "number"
544 | exp '+' "number"
cc2235ac
VT
545 | exp '*' exp
546 ;
547
548 will produce a
549
550 warning: useless associativity for '+', use %precedence [-Wprecedence]
551 %left '+'
552 ^^^
553
554*** Useless precedence
555
556 Bison now warns about symbols with a declared precedence and no declared
557 associativity (i.e., declared with %precedence), and whose precedence is
558 never used. In that case, the symbol can be safely declared with %token
559 instead, without modifying the parsing tables. For example:
560
561 %precedence '='
562 %%
1282c124 563 exp: "var" '=' "number";
cc2235ac
VT
564
565 will produce a
566
567 warning: useless precedence for '=' [-Wprecedence]
568 %precedence '='
569 ^^^
570
571*** Useless precedence and associativity
572
573 In case of both useless precedence and associativity, the issue is flagged
574 as follows:
575
576 %nonassoc '='
577 %%
1282c124 578 exp: "var" '=' "number";
cc2235ac
VT
579
580 The warning is:
581
582 warning: useless precedence and associativity for '=' [-Wprecedence]
583 %nonassoc '='
584 ^^^
09add9c2
AD
585
586** Empty rules
587
6240346a
AD
588 With help from Joel E. Denny and Gabriel Rassoul.
589
09add9c2
AD
590 Empty rules (i.e., with an empty right-hand side) can now be explicitly
591 marked by the new %empty directive. Using %empty on a non-empty rule is
592 an error. The new -Wempty-rule warning reports empty rules without
593 %empty. On the following grammar:
594
595 %%
596 s: a b c;
597 a: ;
598 b: %empty;
599 c: 'a' %empty;
600
601 bison reports:
602
603 3.4-5: warning: empty rule without %empty [-Wempty-rule]
604 a: {}
605 ^^
606 5.8-13: error: %empty on non-empty rule
607 c: 'a' %empty {};
608 ^^^^^^
cc2235ac 609
c21e515e
AD
610** Java skeleton improvements
611
c21e515e
AD
612 The constants for token names were moved to the Lexer interface. Also, it
613 is possible to add code to the parser's constructors using "%code init"
614 and "%define init_throws".
aa94def1
DH
615 Contributed by Paolo Bonzini.
616
de1a2f20
AD
617 The Java skeleton now supports push parsing.
618 Contributed by Dennis Heimbigner.
619
c21e515e
AD
620** C++ skeletons improvements
621
622*** The parser header is no longer mandatory (lalr1.cc, glr.cc)
623
624 Using %defines is now optional. Without it, the needed support classes
625 are defined in the generated parser, instead of additional files (such as
626 location.hh, position.hh and stack.hh).
627
628*** Locations are no longer mandatory (lalr1.cc, glr.cc)
629
630 Both lalr1.cc and glr.cc no longer require %location.
631
632*** syntax_error exception (lalr1.cc)
633
634 The C++ parser features a syntax_error exception, which can be
635 thrown from the scanner or from user rules to raise syntax errors.
636 This facilitates reporting errors caught in sub-functions (e.g.,
637 rejecting too large integral literals from a conversion function
638 used by the scanner, or rejecting invalid combinations from a
639 factory invoked by the user actions).
640
641*** %define api.value.type variant
642
643 This is based on a submission from Michiel De Wilde. With help
644 from Théophile Ranquet.
645
646 In this mode, complex C++ objects can be used as semantic values. For
647 instance:
648
649 %token <::std::string> TEXT;
650 %token <int> NUMBER;
651 %token SEMICOLON ";"
652 %type <::std::string> item;
653 %type <::std::list<std::string>> list;
654 %%
655 result:
656 list { std::cout << $1 << std::endl; }
657 ;
658
659 list:
6240346a 660 %empty { /* Generates an empty string list. */ }
c21e515e
AD
661 | list item ";" { std::swap ($$, $1); $$.push_back ($2); }
662 ;
663
664 item:
665 TEXT { std::swap ($$, $1); }
666 | NUMBER { $$ = string_cast ($1); }
667 ;
668
669*** %define api.token.constructor
670
671 When variants are enabled, Bison can generate functions to build the
672 tokens. This guarantees that the token type (e.g., NUMBER) is consistent
673 with the semantic value (e.g., int):
674
675 parser::symbol_type yylex ()
676 {
677 parser::location_type loc = ...;
678 ...
679 return parser::make_TEXT ("Hello, world!", loc);
680 ...
681 return parser::make_NUMBER (42, loc);
682 ...
683 return parser::make_SEMICOLON (loc);
684 ...
685 }
686
75ae8299
AD
687*** C++ locations
688
689 There are operator- and operator-= for 'location'. Negative line/column
690 increments can no longer underflow the resulting value.
691
1f5542fe 692* Noteworthy changes in release 2.7.1 (2013-04-15) [stable]
cc8962bd 693
80a2826e
AD
694** Bug fixes
695
696*** Fix compiler attribute portability (yacc.c)
697
698 With locations enabled, __attribute__ was used unprotected.
0a7b8559 699
e83be476
AD
700*** Fix some compiler warnings (lalr1.cc)
701
c13bb348 702* Noteworthy changes in release 2.7 (2012-12-12) [stable]
effd30c0 703
edf9a06f 704** Bug fixes
7bada535 705
edf9a06f 706 Warnings about uninitialized yylloc in yyparse have been fixed.
7bada535 707
1127a75a
AD
708 Restored C90 compliance (yet no report was ever made).
709
d4fe9e88 710** Diagnostics are improved
7bada535 711
8458a411
AD
712 Contributed by Théophile Ranquet.
713
d4fe9e88 714*** Changes in the format of error messages
7bada535 715
d4fe9e88 716 This used to be the format of many error reports:
1f1bd572 717
d4fe9e88
AD
718 input.y:2.7-12: %type redeclaration for exp
719 input.y:1.7-12: previous declaration
1f1bd572 720
d4fe9e88 721 It is now:
1f1bd572 722
d4fe9e88
AD
723 input.y:2.7-12: error: %type redeclaration for exp
724 input.y:1.7-12: previous declaration
cbaea010 725
d4fe9e88 726*** New format for error reports: carets
cbaea010 727
d4fe9e88 728 Caret errors have been added to Bison:
cbaea010 729
d4fe9e88
AD
730 input.y:2.7-12: error: %type redeclaration for exp
731 %type <sval> exp
732 ^^^^^^
733 input.y:1.7-12: previous declaration
734 %type <ival> exp
735 ^^^^^^
cbaea010 736
d4fe9e88 737 or
cbaea010 738
7bada535 739 input.y:3.20-23: error: ambiguous reference: '$exp'
fb6040f0 740 exp: exp '+' exp { $exp = $1 + $3; };
7bada535 741 ^^^^
fb6040f0
TR
742 input.y:3.1-3: refers to: $exp at $$
743 exp: exp '+' exp { $exp = $1 + $3; };
744 ^^^
745 input.y:3.6-8: refers to: $exp at $1
746 exp: exp '+' exp { $exp = $1 + $3; };
747 ^^^
748 input.y:3.14-16: refers to: $exp at $3
749 exp: exp '+' exp { $exp = $1 + $3; };
750 ^^^
7bada535 751
1282c124
AD
752 The default behavior for now is still not to display these unless
753 explicitly asked with -fcaret (or -fall). However, in a later release, it
d4fe9e88
AD
754 will be made the default behavior (but may still be deactivated with
755 -fno-caret).
d3e4409a 756
1f1bd572 757** New value for %define variable: api.pure full
d3e4409a 758
1f1bd572 759 The %define variable api.pure requests a pure (reentrant) parser. However,
d4fe9e88
AD
760 for historical reasons, using it in a location-tracking Yacc parser
761 resulted in a yyerror function that did not take a location as a
762 parameter. With this new value, the user may request a better pure parser,
763 where yyerror does take a location as a parameter (in location-tracking
764 parsers).
1f1bd572
TR
765
766 The use of "%define api.pure true" is deprecated in favor of this new
767 "%define api.pure full".
d3e4409a 768
7287be84 769** New %define variable: api.location.type (glr.cc, lalr1.cc, lalr1.java)
db8ab2be
AD
770
771 The %define variable api.location.type defines the name of the type to use
772 for locations. When defined, Bison no longer generates the position.hh
773 and location.hh files, nor does the parser will include them: the user is
774 then responsible to define her type.
775
776 This can be used in programs with several parsers to factor their location
7287be84
AD
777 and position files: let one of them generate them, and the others just use
778 them.
db8ab2be
AD
779
780 This feature was actually introduced, but not documented, in Bison 2.5,
781 under the name "location_type" (which is maintained for backward
782 compatibility).
783
7287be84
AD
784 For consistency, lalr1.java's %define variables location_type and
785 position_type are deprecated in favor of api.location.type and
786 api.position.type.
787
d4fe9e88
AD
788** Exception safety (lalr1.cc)
789
790 The parse function now catches exceptions, uses the %destructors to
791 release memory (the lookahead symbol and the symbols pushed on the stack)
792 before re-throwing the exception.
793
794 This feature is somewhat experimental. User feedback would be
795 appreciated.
796
9c16d399 797** Graph improvements in DOT and XSLT
fc4fdd62 798
8458a411
AD
799 Contributed by Théophile Ranquet.
800
fc4fdd62
TR
801 The graphical presentation of the states is more readable: their shape is
802 now rectangular, the state number is clearly displayed, and the items are
803 numbered and left-justified.
804
805 The reductions are now explicitly represented as transitions to other
806 diamond shaped nodes.
807
9c16d399
TR
808 These changes are present in both --graph output and xml2dot.xsl XSLT
809 processing, with minor (documented) differences.
810
d4fe9e88 811** %language is no longer an experimental feature.
fb4c8a7c 812
d4fe9e88
AD
813 The introduction of this feature, in 2.4, was four years ago. The
814 --language option and the %language directive are no longer experimental.
fb4c8a7c 815
53e2cd1e
AD
816** Documentation
817
818 The sections about shift/reduce and reduce/reduce conflicts resolution
819 have been fixed and extended.
9d2423f5 820
d4fe9e88
AD
821 Although introduced more than four years ago, XML and Graphviz reports
822 were not properly documented.
823
be22823e
AD
824 The translation of mid-rule actions is now described.
825
9d3f7eaf 826* Noteworthy changes in release 2.6.5 (2012-11-07) [stable]
6f1360bd 827
a68b1f23
AD
828 We consider compiler warnings about Bison generated parsers to be bugs.
829 Rather than working around them in your own project, please consider
830 reporting them to us.
831
832** Bug fixes
833
834 Warnings about uninitialized yylval and/or yylloc for push parsers with a
835 pure interface have been fixed for GCC 4.0 up to 4.8, and Clang 2.9 to
836 3.2.
837
838 Other issues in the test suite have been addressed.
6f1360bd 839
1282c124 840 Null characters are correctly displayed in error messages.
95066e92 841
a1d1ab50
AD
842 When possible, yylloc is correctly initialized before calling yylex. It
843 is no longer necessary to initialize it in the %initial-action.
844
0ac15849 845* Noteworthy changes in release 2.6.4 (2012-10-23) [stable]
a4eb820f 846
468455e1 847 Bison 2.6.3's --version was incorrect. This release fixes this issue.
a4eb820f 848
6eb8f74f 849* Noteworthy changes in release 2.6.3 (2012-10-22) [stable]
933ec544 850
6b4cb804
AD
851** Bug fixes
852
a1a77e1f 853 Bugs and portability issues in the test suite have been fixed.
6b4cb804
AD
854
855 Some errors in translations have been addressed, and --help now directs
856 users to the appropriate place to report them.
857
858 Stray Info files shipped by accident are removed.
859
860 Incorrect definitions of YY_, issued by yacc.c when no parser header is
861 generated, are removed.
9c26b8fc 862
a2b3f101
TR
863 All the generated headers are self-contained.
864
c9d5bcc9
AD
865** Header guards (yacc.c, glr.c, glr.cc)
866
867 In order to avoid collisions, the header guards are now
868 YY_<PREFIX>_<FILE>_INCLUDED, instead of merely <PREFIX>_<FILE>.
869 For instance the header generated from
870
871 %define api.prefix "calc"
872 %defines "lib/parse.h"
873
874 will use YY_CALC_LIB_PARSE_H_INCLUDED as guard.
875
c12c4c50 876** Fix compiler warnings in the generated parser (yacc.c, glr.c)
321d3e35
AD
877
878 The compilation of pure parsers (%define api.pure) can trigger GCC
879 warnings such as:
880
881 input.c: In function 'yyparse':
882 input.c:1503:12: warning: 'yylval' may be used uninitialized in this
883 function [-Wmaybe-uninitialized]
884 *++yyvsp = yylval;
885 ^
886
887 This is now fixed; pragmas to avoid these warnings are no longer needed.
888
c12c4c50
AD
889 Warnings from clang ("equality comparison with extraneous parentheses" and
890 "function declared 'noreturn' should not return") have also been
891 addressed.
892
e1eeecd3 893* Noteworthy changes in release 2.6.2 (2012-08-03) [stable]
9c26b8fc 894
43ca8040
AD
895** Bug fixes
896
897 Buffer overruns, complaints from Flex, and portability issues in the test
898 suite have been fixed.
899
c9d546b2
AD
900** Spaces in %lex- and %parse-param (lalr1.cc, glr.cc)
901
902 Trailing end-of-lines in %parse-param or %lex-param would result in
903 invalid C++. This is fixed.
9c26b8fc 904
dcd5344d
AD
905** Spurious spaces and end-of-lines
906
907 The generated files no longer end (nor start) with empty lines.
908
77b214ef 909* Noteworthy changes in release 2.6.1 (2012-07-30) [stable]
a4107f24 910
8617d87e
AD
911 Bison no longer executes user-specified M4 code when processing a grammar.
912
e20e6a50
AD
913** Future Changes
914
915 In addition to the removal of the features announced in Bison 2.6, the
916 next major release will remove the "Temporary hack for adding a semicolon
917 to the user action", as announced in the release 2.5. Instead of:
918
919 exp: exp "+" exp { $$ = $1 + $3 };
920
921 write:
922
923 exp: exp "+" exp { $$ = $1 + $3; };
924
8617d87e
AD
925** Bug fixes
926
0e164d43
AD
927*** Type names are now properly escaped.
928
929*** glr.cc: set_debug_level and debug_level work as expected.
a4107f24 930
26313726
AD
931*** Stray @ or $ in actions
932
933 While Bison used to warn about stray $ or @ in action rules, it did not
934 for other actions such as printers, destructors, or initial actions. It
935 now does.
936
cd735a8c 937** Type names in actions
4982f078
AD
938
939 For consistency with rule actions, it is now possible to qualify $$ by a
cd735a8c 940 type-name in destructors, printers, and initial actions. For instance:
4982f078
AD
941
942 %printer { fprintf (yyo, "(%d, %f)", $<ival>$, $<fval>$); } <*> <>;
943
944 will display two values for each typed and untyped symbol (provided
cd735a8c 945 that YYSTYPE has both "ival" and "fval" fields).
60aa04a2 946
1505e8bb 947* Noteworthy changes in release 2.6 (2012-07-19) [stable]
0f11eec2 948
d0a30438 949** Future changes
9553083c 950
55d1006f
AD
951 The next major release of Bison will drop support for the following
952 deprecated features. Please report disagreements to bug-bison@gnu.org.
0f11eec2 953
aaf61036 954*** K&R C parsers
55d1006f
AD
955
956 Support for generating parsers in K&R C will be removed. Parsers
242cc08e 957 generated for C support ISO C90, and are tested with ISO C99 and ISO C11
55d1006f
AD
958 compilers.
959
258cddbc 960*** Features deprecated since Bison 1.875
0f11eec2 961
258cddbc
AD
962 The definitions of yystype and yyltype will be removed; use YYSTYPE and
963 YYLTYPE.
0f11eec2 964
258cddbc
AD
965 YYPARSE_PARAM and YYLEX_PARAM, deprecated in favor of %parse-param and
966 %lex-param, will no longer be supported.
967
968 Support for the preprocessor symbol YYERROR_VERBOSE will be removed, use
969 %error-verbose.
55d1006f
AD
970
971*** The generated header will be included (yacc.c)
0f11eec2
AD
972
973 Instead of duplicating the content of the generated header (definition of
55d1006f
AD
974 YYSTYPE, yyparse declaration etc.), the generated parser will include it,
975 as is already the case for GLR or C++ parsers. This change is deferred
976 because existing versions of ylwrap (e.g., Automake 1.12.1) do not support
977 it.
0f11eec2 978
c2425191 979** Generated Parser Headers
56ca3d8f 980
258cddbc 981*** Guards (yacc.c, glr.c, glr.cc)
c3e9f08f
AD
982
983 The generated headers are now guarded, as is already the case for C++
242cc08e 984 parsers (lalr1.cc). For instance, with --defines=foo.h:
c3e9f08f 985
e29f0771
AD
986 #ifndef YY_FOO_H
987 # define YY_FOO_H
988 ...
989 #endif /* !YY_FOO_H */
c3e9f08f 990
258cddbc 991*** New declarations (yacc.c, glr.c)
56ca3d8f
AD
992
993 The generated header now declares yydebug and yyparse. Both honor
994 --name-prefix=bar_, and yield
995
e29f0771 996 int bar_parse (void);
56ca3d8f
AD
997
998 rather than
999
e29f0771
AD
1000 #define yyparse bar_parse
1001 int yyparse (void);
56ca3d8f
AD
1002
1003 in order to facilitate the inclusion of several parser headers inside a
1004 single compilation unit.
c3e9f08f 1005
258cddbc
AD
1006*** Exported symbols in C++
1007
1008 The symbols YYTOKEN_TABLE and YYERROR_VERBOSE, which were defined in the
1009 header, are removed, as they prevent the possibility of including several
1010 generated headers from a single compilation unit.
1011
694af10c
AD
1012*** YYLSP_NEEDED
1013
1014 For the same reasons, the undocumented and unused macro YYLSP_NEEDED is no
1015 longer defined.
1016
4b3847c3
AD
1017** New %define variable: api.prefix
1018
1019 Now that the generated headers are more complete and properly protected
1020 against multiple inclusions, constant names, such as YYSTYPE are a
1021 problem. While yyparse and others are properly renamed by %name-prefix,
1022 YYSTYPE, YYDEBUG and others have never been affected by it. Because it
1023 would introduce backward compatibility issues in projects not expecting
1024 YYSTYPE to be renamed, instead of changing the behavior of %name-prefix,
1025 it is deprecated in favor of a new %define variable: api.prefix.
1026
1027 The following examples compares both:
1028
1029 %name-prefix "bar_" | %define api.prefix "bar_"
1030 %token <ival> FOO %token <ival> FOO
1031 %union { int ival; } %union { int ival; }
1032 %% %%
1033 exp: 'a'; exp: 'a';
1034
1035 bison generates:
1036
1037 #ifndef BAR_FOO_H #ifndef BAR_FOO_H
1038 # define BAR_FOO_H # define BAR_FOO_H
1039
1040 /* Enabling traces. */ /* Enabling traces. */
5f108727
AD
1041 # ifndef YYDEBUG | # ifndef BAR_DEBUG
1042 > # if defined YYDEBUG
1043 > # if YYDEBUG
1044 > # define BAR_DEBUG 1
1045 > # else
1046 > # define BAR_DEBUG 0
1047 > # endif
1048 > # else
1049 # define YYDEBUG 0 | # define BAR_DEBUG 0
1050 > # endif
1051 # endif | # endif
1052
1053 # if YYDEBUG | # if BAR_DEBUG
4b3847c3
AD
1054 extern int bar_debug; extern int bar_debug;
1055 # endif # endif
1056
1057 /* Tokens. */ /* Tokens. */
1058 # ifndef YYTOKENTYPE | # ifndef BAR_TOKENTYPE
1059 # define YYTOKENTYPE | # define BAR_TOKENTYPE
1060 enum yytokentype { | enum bar_tokentype {
1061 FOO = 258 FOO = 258
1062 }; };
1063 # endif # endif
1064
1065 #if ! defined YYSTYPE \ | #if ! defined BAR_STYPE \
1066 && ! defined YYSTYPE_IS_DECLARED | && ! defined BAR_STYPE_IS_DECLARED
1067 typedef union YYSTYPE | typedef union BAR_STYPE
1068 { {
1069 int ival; int ival;
1070 } YYSTYPE; | } BAR_STYPE;
1071 # define YYSTYPE_IS_DECLARED 1 | # define BAR_STYPE_IS_DECLARED 1
1072 #endif #endif
1073
1074 extern YYSTYPE bar_lval; | extern BAR_STYPE bar_lval;
1075
1076 int bar_parse (void); int bar_parse (void);
1077
1078 #endif /* !BAR_FOO_H */ #endif /* !BAR_FOO_H */
1079
dfaac272 1080* Noteworthy changes in release 2.5.1 (2012-06-05) [stable]
df6e3db0 1081
debe2c03 1082** Future changes:
765e1bd4 1083
e4ab1254 1084 The next major release will drop support for generating parsers in K&R C.
041308d0 1085
466b4cf2 1086** yacc.c: YYBACKUP works as expected.
ef51bfa7 1087
d834eca0 1088** glr.c improvements:
041308d0 1089
d834eca0 1090*** Location support is eliminated when not requested:
041308d0 1091
e4ab1254
AD
1092 GLR parsers used to include location-related code even when locations were
1093 not requested, and therefore not even usable.
378e917c 1094
d834eca0 1095*** __attribute__ is preserved:
d115aad9 1096
e4ab1254
AD
1097 __attribute__ is no longer disabled when __STRICT_ANSI__ is defined (i.e.,
1098 when -std is passed to GCC).
041308d0 1099
466b4cf2 1100** lalr1.java: several fixes:
041308d0 1101
e4ab1254
AD
1102 The Java parser no longer throws ArrayIndexOutOfBoundsException if the
1103 first token leads to a syntax error. Some minor clean ups.
041308d0 1104
22172d47 1105** Changes for C++:
ef51bfa7 1106
22172d47 1107*** C++11 compatibility:
ef51bfa7 1108
e4ab1254
AD
1109 C and C++ parsers use "nullptr" instead of "0" when __cplusplus is 201103L
1110 or higher.
936c88d1 1111
22172d47
AD
1112*** Header guards
1113
1114 The header files such as "parser.hh", "location.hh", etc. used a constant
1115 name for preprocessor guards, for instance:
1116
e29f0771
AD
1117 #ifndef BISON_LOCATION_HH
1118 # define BISON_LOCATION_HH
1119 ...
1120 #endif // !BISON_LOCATION_HH
22172d47
AD
1121
1122 The inclusion guard is now computed from "PREFIX/FILE-NAME", where lower
1123 case characters are converted to upper case, and series of
1124 non-alphanumerical characters are converted to an underscore.
1125
1126 With "bison -o lang++/parser.cc", "location.hh" would now include:
1127
e29f0771
AD
1128 #ifndef YY_LANG_LOCATION_HH
1129 # define YY_LANG_LOCATION_HH
1130 ...
1131 #endif // !YY_LANG_LOCATION_HH
22172d47
AD
1132
1133*** C++ locations:
936c88d1 1134
e4ab1254
AD
1135 The position and location constructors (and their initialize methods)
1136 accept new arguments for line and column. Several issues in the
1137 documentation were fixed.
936c88d1 1138
466b4cf2
AD
1139** liby is no longer asking for "rpl_fprintf" on some platforms.
1140
7e508a2b
AD
1141** Changes in the manual:
1142
1143*** %printer is documented
1144
e4ab1254
AD
1145 The "%printer" directive, supported since at least Bison 1.50, is finally
1146 documented. The "mfcalc" example is extended to demonstrate it.
7e508a2b 1147
e4ab1254
AD
1148 For consistency with the C skeletons, the C++ parsers now also support
1149 "yyoutput" (as an alias to "debug_stream ()").
7e508a2b
AD
1150
1151*** Several improvements have been made:
466b4cf2 1152
e4ab1254
AD
1153 The layout for grammar excerpts was changed to a more compact scheme.
1154 Named references are motivated. The description of the automaton
1155 description file (*.output) is updated to the current format. Incorrect
1156 index entries were fixed. Some other errors were fixed.
466b4cf2 1157
86b08b49
AD
1158** Building bison:
1159
1160*** Conflicting prototypes with recent/modified Flex.
466b4cf2 1161
e4ab1254
AD
1162 Fixed build problems with the current, unreleased, version of Flex, and
1163 some modified versions of 2.5.35, which have modified function prototypes.
466b4cf2 1164
8ef26c2a
AD
1165*** Warnings during the build procedure have been eliminated.
1166
1167*** Several portability problems in the test suite have been fixed:
466b4cf2 1168
e4ab1254
AD
1169 This includes warnings with some compilers, unexpected behavior of tools
1170 such as diff, warning messages from the test suite itself, etc.
466b4cf2 1171
91aadcc7 1172*** The install-pdf target works properly:
8ef26c2a 1173
e4ab1254
AD
1174 Running "make install-pdf" (or -dvi, -html, -info, and -ps) no longer
1175 halts in the middle of its course.
8ef26c2a 1176
28801043 1177* Changes in version 2.5 (2011-05-14):
50cca368 1178
82f3355e
JD
1179** Grammar symbol names can now contain non-initial dashes:
1180
1181 Consistently with directives (such as %error-verbose) and with
1182 %define variables (e.g. push-pull), grammar symbol names may contain
1183 dashes in any position except the beginning. This is a GNU
1184 extension over POSIX Yacc. Thus, use of this extension is reported
1185 by -Wyacc and rejected in Yacc mode (--yacc).
1186
f1b238df 1187** Named references:
66381412
AR
1188
1189 Historically, Yacc and Bison have supported positional references
1190 ($n, $$) to allow access to symbol values from inside of semantic
1191 actions code.
1192
1193 Starting from this version, Bison can also accept named references.
1194 When no ambiguity is possible, original symbol names may be used
1195 as named references:
1196
4b568fc0 1197 if_stmt : "if" cond_expr "then" then_stmt ';'
66381412
AR
1198 { $if_stmt = mk_if_stmt($cond_expr, $then_stmt); }
1199
1200 In the more common case, explicit names may be declared:
1201
4b568fc0 1202 stmt[res] : "if" expr[cond] "then" stmt[then] "else" stmt[else] ';'
66381412
AR
1203 { $res = mk_if_stmt($cond, $then, $else); }
1204
5b1ff423 1205 Location information is also accessible using @name syntax. When
66381412
AR
1206 accessing symbol names containing dots or dashes, explicit bracketing
1207 ($[sym.1]) must be used.
1208
5b1ff423 1209 These features are experimental in this version. More user feedback
66381412 1210 will help to stabilize them.
2bd435c3 1211 Contributed by Alex Rozenman.
66381412 1212
f1b238df 1213** IELR(1) and canonical LR(1):
eb45ef3b
JD
1214
1215 IELR(1) is a minimal LR(1) parser table generation algorithm. That
1216 is, given any context-free grammar, IELR(1) generates parser tables
7262f54f 1217 with the full language-recognition power of canonical LR(1) but with
f1b238df
JD
1218 nearly the same number of parser states as LALR(1). This reduction
1219 in parser states is often an order of magnitude. More importantly,
eb45ef3b
JD
1220 because canonical LR(1)'s extra parser states may contain duplicate
1221 conflicts in the case of non-LR(1) grammars, the number of conflicts
1222 for IELR(1) is often an order of magnitude less as well. This can
1223 significantly reduce the complexity of developing of a grammar.
1224
1225 Bison can now generate IELR(1) and canonical LR(1) parser tables in
1226 place of its traditional LALR(1) parser tables, which remain the
1227 default. You can specify the type of parser tables in the grammar
1228 file with these directives:
1229
cf499cff
JD
1230 %define lr.type lalr
1231 %define lr.type ielr
1232 %define lr.type canonical-lr
eb45ef3b 1233
7fceb615 1234 The default-reduction optimization in the parser tables can also be
e4ab1254
AD
1235 adjusted using "%define lr.default-reductions". For details on both
1236 of these features, see the new section "Tuning LR" in the Bison
7fceb615 1237 manual.
eb45ef3b
JD
1238
1239 These features are experimental. More user feedback will help to
1240 stabilize them.
1241
8458a411
AD
1242** LAC (Lookahead Correction) for syntax error handling
1243
1244 Contributed by Joel E. Denny.
fcf834f9
JD
1245
1246 Canonical LR, IELR, and LALR can suffer from a couple of problems
1247 upon encountering a syntax error. First, the parser might perform
1248 additional parser stack reductions before discovering the syntax
7fceb615 1249 error. Such reductions can perform user semantic actions that are
fcf834f9
JD
1250 unexpected because they are based on an invalid token, and they
1251 cause error recovery to begin in a different syntactic context than
1252 the one in which the invalid token was encountered. Second, when
7fceb615 1253 verbose error messages are enabled (with %error-verbose or the
e4ab1254 1254 obsolete "#define YYERROR_VERBOSE"), the expected token list in the
7fceb615
JD
1255 syntax error message can both contain invalid tokens and omit valid
1256 tokens.
fcf834f9
JD
1257
1258 The culprits for the above problems are %nonassoc, default
1259 reductions in inconsistent states, and parser state merging. Thus,
1260 IELR and LALR suffer the most. Canonical LR can suffer only if
1261 %nonassoc is used or if default reductions are enabled for
1262 inconsistent states.
1263
7fceb615
JD
1264 LAC is a new mechanism within the parsing algorithm that solves
1265 these problems for canonical LR, IELR, and LALR without sacrificing
1266 %nonassoc, default reductions, or state merging. When LAC is in
1267 use, canonical LR and IELR behave almost exactly the same for both
1268 syntactically acceptable and syntactically unacceptable input.
fcf834f9
JD
1269 While LALR still does not support the full language-recognition
1270 power of canonical LR and IELR, LAC at least enables LALR's syntax
1271 error handling to correctly reflect LALR's language-recognition
1272 power.
1273
1274 Currently, LAC is only supported for deterministic parsers in C.
1275 You can enable LAC with the following directive:
1276
1277 %define parse.lac full
1278
e4ab1254 1279 See the new section "LAC" in the Bison manual for additional
7fceb615 1280 details including a few caveats.
fcf834f9
JD
1281
1282 LAC is an experimental feature. More user feedback will help to
1283 stabilize it.
1284
d397d9f0 1285** %define improvements:
cf499cff 1286
f1b238df 1287*** Can now be invoked via the command line:
50cca368 1288
de5ab940 1289 Each of these command-line options
50cca368 1290
de5ab940
JD
1291 -D NAME[=VALUE]
1292 --define=NAME[=VALUE]
1293
1294 -F NAME[=VALUE]
1295 --force-define=NAME[=VALUE]
50cca368
JD
1296
1297 is equivalent to this grammar file declaration
1298
de5ab940 1299 %define NAME ["VALUE"]
50cca368 1300
de5ab940
JD
1301 except that the manner in which Bison processes multiple definitions
1302 for the same NAME differs. Most importantly, -F and --force-define
1303 quietly override %define, but -D and --define do not. For further
e4ab1254 1304 details, see the section "Bison Options" in the Bison manual.
50cca368 1305
f1b238df 1306*** Variables renamed:
67212941
JD
1307
1308 The following %define variables
1309
1310 api.push_pull
1311 lr.keep_unreachable_states
1312
1313 have been renamed to
1314
1315 api.push-pull
1316 lr.keep-unreachable-states
1317
1318 The old names are now deprecated but will be maintained indefinitely
1319 for backward compatibility.
1320
7262f54f 1321*** Values no longer need to be quoted in the grammar file:
cf499cff
JD
1322
1323 If a %define value is an identifier, it no longer needs to be placed
1324 within quotations marks. For example,
1325
1326 %define api.push-pull "push"
1327
1328 can be rewritten as
1329
1330 %define api.push-pull push
1331
d397d9f0 1332*** Unrecognized variables are now errors not warnings.
cdf3f113 1333
d397d9f0
JD
1334*** Multiple invocations for any variable is now an error not a warning.
1335
1336** Unrecognized %code qualifiers are now errors not warnings.
1337
1338** Character literals not of length one:
1339
1340 Previously, Bison quietly converted all character literals to length
1341 one. For example, without warning, Bison interpreted the operators in
1342 the following grammar to be the same token:
1343
1344 exp: exp '++'
1345 | exp '+' exp
1346 ;
1347
1348 Bison now warns when a character literal is not of length one. In
1349 some future release, Bison will start reporting an error instead.
1350
1351** Destructor calls fixed for lookaheads altered in semantic actions:
1352
1353 Previously for deterministic parsers in C, if a user semantic action
1354 altered yychar, the parser in some cases used the old yychar value to
1355 determine which destructor to call for the lookahead upon a syntax
1356 error or upon parser return. This bug has been fixed.
1357
1358** C++ parsers use YYRHSLOC:
1359
1360 Similarly to the C parsers, the C++ parsers now define the YYRHSLOC
1361 macro and use it in the default YYLLOC_DEFAULT. You are encouraged
e4ab1254
AD
1362 to use it. If, for instance, your location structure has "first"
1363 and "last" members, instead of
d397d9f0 1364
e29f0771
AD
1365 # define YYLLOC_DEFAULT(Current, Rhs, N) \
1366 do \
1367 if (N) \
1368 { \
1369 (Current).first = (Rhs)[1].location.first; \
1370 (Current).last = (Rhs)[N].location.last; \
1371 } \
1372 else \
1373 { \
1374 (Current).first = (Current).last = (Rhs)[0].location.last; \
1375 } \
1376 while (false)
d397d9f0
JD
1377
1378 use:
1379
e29f0771
AD
1380 # define YYLLOC_DEFAULT(Current, Rhs, N) \
1381 do \
1382 if (N) \
1383 { \
1384 (Current).first = YYRHSLOC (Rhs, 1).first; \
1385 (Current).last = YYRHSLOC (Rhs, N).last; \
1386 } \
1387 else \
1388 { \
1389 (Current).first = (Current).last = YYRHSLOC (Rhs, 0).last; \
1390 } \
1391 while (false)
d397d9f0
JD
1392
1393** YYLLOC_DEFAULT in C++:
1394
1395 The default implementation of YYLLOC_DEFAULT used to be issued in
1396 the header file. It is now output in the implementation file, after
1397 the user %code sections so that its #ifndef guard does not try to
1398 override the user's YYLLOC_DEFAULT if provided.
cdf3f113 1399
f1b238df 1400** YYFAIL now produces warnings and Java parsers no longer implement it:
4395a9ff
JD
1401
1402 YYFAIL has existed for many years as an undocumented feature of
1403 deterministic parsers in C generated by Bison. More recently, it was
1404 a documented feature of Bison's experimental Java parsers. As
1405 promised in Bison 2.4.2's NEWS entry, any appearance of YYFAIL in a
1406 semantic action now produces a deprecation warning, and Java parsers
1407 no longer implement YYFAIL at all. For further details, including a
1408 discussion of how to suppress C preprocessor warnings about YYFAIL
1409 being unused, see the Bison 2.4.2 NEWS entry.
1410
f1b238df 1411** Temporary hack for adding a semicolon to the user action:
197b82ba
JD
1412
1413 Previously, Bison appended a semicolon to every user action for
1414 reductions when the output language defaulted to C (specifically, when
1415 neither %yacc, %language, %skeleton, or equivalent command-line
1416 options were specified). This allowed actions such as
1417
1418 exp: exp "+" exp { $$ = $1 + $3 };
1419
1420 instead of
1421
1422 exp: exp "+" exp { $$ = $1 + $3; };
1423
1424 As a first step in removing this misfeature, Bison now issues a
1425 warning when it appends a semicolon. Moreover, in cases where Bison
1426 cannot easily determine whether a semicolon is needed (for example, an
1427 action ending with a cpp directive or a braced compound initializer),
1428 it no longer appends one. Thus, the C compiler might now complain
1429 about a missing semicolon where it did not before. Future releases of
1430 Bison will cease to append semicolons entirely.
1431
d2060f06
JD
1432** Verbose syntax error message fixes:
1433
e4ab1254 1434 When %error-verbose or the obsolete "#define YYERROR_VERBOSE" is
7fceb615
JD
1435 specified, syntax error messages produced by the generated parser
1436 include the unexpected token as well as a list of expected tokens.
1437 The effect of %nonassoc on these verbose messages has been corrected
1438 in two ways, but a more complete fix requires LAC, described above:
d2060f06
JD
1439
1440*** When %nonassoc is used, there can exist parser states that accept no
1441 tokens, and so the parser does not always require a lookahead token
1442 in order to detect a syntax error. Because no unexpected token or
1443 expected tokens can then be reported, the verbose syntax error
1444 message described above is suppressed, and the parser instead
e4ab1254 1445 reports the simpler message, "syntax error". Previously, this
d2060f06
JD
1446 suppression was sometimes erroneously triggered by %nonassoc when a
1447 lookahead was actually required. Now verbose messages are
1448 suppressed only when all previous lookaheads have already been
1449 shifted or discarded.
1450
1451*** Previously, the list of expected tokens erroneously included tokens
1452 that would actually induce a syntax error because conflicts for them
1453 were resolved with %nonassoc in the current parser state. Such
1454 tokens are now properly omitted from the list.
1455
1456*** Expected token lists are still often wrong due to state merging
fcf834f9
JD
1457 (from LALR or IELR) and default reductions, which can both add
1458 invalid tokens and subtract valid tokens. Canonical LR almost
1459 completely fixes this problem by eliminating state merging and
1460 default reductions. However, there is one minor problem left even
1461 when using canonical LR and even after the fixes above. That is,
1462 if the resolution of a conflict with %nonassoc appears in a later
1463 parser state than the one at which some syntax error is
1464 discovered, the conflicted token is still erroneously included in
1465 the expected token list. Bison's new LAC implementation,
1466 described above, eliminates this problem and the need for
1467 canonical LR. However, LAC is still experimental and is disabled
1468 by default.
53f036ce 1469
1a33f4f6
JD
1470** Java skeleton fixes:
1471
1472*** A location handling bug has been fixed.
1473
1474*** The top element of each of the value stack and location stack is now
1475 cleared when popped so that it can be garbage collected.
6771a463 1476
02803d55
JD
1477*** Parser traces now print the top element of the stack.
1478
86408959
JD
1479** -W/--warnings fixes:
1480
e4ab1254 1481*** Bison now properly recognizes the "no-" versions of categories:
86408959
JD
1482
1483 For example, given the following command line, Bison now enables all
1484 warnings except warnings for incompatibilities with POSIX Yacc:
1485
1486 bison -Wall,no-yacc gram.y
1487
786743d5
JD
1488*** Bison now treats S/R and R/R conflicts like other warnings:
1489
1490 Previously, conflict reports were independent of Bison's normal
1491 warning system. Now, Bison recognizes the warning categories
e4ab1254 1492 "conflicts-sr" and "conflicts-rr". This change has important
786743d5
JD
1493 consequences for the -W and --warnings command-line options. For
1494 example:
1495
1496 bison -Wno-conflicts-sr gram.y # S/R conflicts not reported
1497 bison -Wno-conflicts-rr gram.y # R/R conflicts not reported
1498 bison -Wnone gram.y # no conflicts are reported
1499 bison -Werror gram.y # any conflict is an error
1500
1501 However, as before, if the %expect or %expect-rr directive is
1502 specified, an unexpected number of conflicts is an error, and an
1503 expected number of conflicts is not reported, so -W and --warning
1504 then have no effect on the conflict report.
1505
e4ab1254 1506*** The "none" category no longer disables a preceding "error":
bf0e44e8
JD
1507
1508 For example, for the following command line, Bison now reports
1509 errors instead of warnings for incompatibilities with POSIX Yacc:
1510
1511 bison -Werror,none,yacc gram.y
1512
e4ab1254 1513*** The "none" category now disables all Bison warnings:
c39014ae 1514
e4ab1254 1515 Previously, the "none" category disabled only Bison warnings for
c39014ae
JD
1516 which there existed a specific -W/--warning category. However,
1517 given the following command line, Bison is now guaranteed to
1518 suppress all warnings:
1519
1520 bison -Wnone gram.y
1521
1f36f544
JD
1522** Precedence directives can now assign token number 0:
1523
1524 Since Bison 2.3b, which restored the ability of precedence
1525 directives to assign token numbers, doing so for token number 0 has
1526 produced an assertion failure. For example:
1527
1528 %left END 0
1529
1530 This bug has been fixed.
1531
64877e5e 1532* Changes in version 2.4.3 (2010-08-05):
8b9e021f 1533
2bfcac9a
JD
1534** Bison now obeys -Werror and --warnings=error for warnings about
1535 grammar rules that are useless in the parser due to conflicts.
1536
8b9e021f
JD
1537** Problems with spawning M4 on at least FreeBSD 8 and FreeBSD 9 have
1538 been fixed.
1539
4ad3921d
JD
1540** Failures in the test suite for GCC 4.5 have been fixed.
1541
06cb07d5
JD
1542** Failures in the test suite for some versions of Sun Studio C++ have
1543 been fixed.
1544
9b5049bd
JD
1545** Contrary to Bison 2.4.2's NEWS entry, it has been decided that
1546 warnings about undefined %prec identifiers will not be converted to
1547 errors in Bison 2.5. They will remain warnings, which should be
1548 sufficient for POSIX while avoiding backward compatibility issues.
1549
93d7dde9
JD
1550** Minor documentation fixes.
1551
e19a049c 1552* Changes in version 2.4.2 (2010-03-20):
74553c98 1553
f39ab286
JD
1554** Some portability problems that resulted in failures and livelocks
1555 in the test suite on some versions of at least Solaris, AIX, HP-UX,
e19a049c
JD
1556 RHEL4, and Tru64 have been addressed. As a result, fatal Bison
1557 errors should no longer cause M4 to report a broken pipe on the
f39ab286
JD
1558 affected platforms.
1559
e4ab1254 1560** "%prec IDENTIFIER" requires IDENTIFIER to be defined separately.
8bb3a2e7
JD
1561
1562 POSIX specifies that an error be reported for any identifier that does
1563 not appear on the LHS of a grammar rule and that is not defined by
1564 %token, %left, %right, or %nonassoc. Bison 2.3b and later lost this
1565 error report for the case when an identifier appears only after a
1566 %prec directive. It is now restored. However, for backward
1567 compatibility with recent Bison releases, it is only a warning for
1568 now. In Bison 2.5 and later, it will return to being an error.
9b5049bd
JD
1569 [Between the 2.4.2 and 2.4.3 releases, it was decided that this
1570 warning will not be converted to an error in Bison 2.5.]
8bb3a2e7 1571
d8911864
EB
1572** Detection of GNU M4 1.4.6 or newer during configure is improved.
1573
a603c6e0
JD
1574** Warnings from gcc's -Wundef option about undefined YYENABLE_NLS,
1575 YYLTYPE_IS_TRIVIAL, and __STRICT_ANSI__ in C/C++ parsers are now
1576 avoided.
c938d650 1577
98a345a2
JD
1578** %code is now a permanent feature.
1579
1580 A traditional Yacc prologue directive is written in the form:
1581
1582 %{CODE%}
1583
1584 To provide a more flexible alternative, Bison 2.3b introduced the
1585 %code directive with the following forms for C/C++:
1586
1587 %code {CODE}
1588 %code requires {CODE}
1589 %code provides {CODE}
1590 %code top {CODE}
1591
1592 These forms are now considered permanent features of Bison. See the
1593 %code entries in the section "Bison Declaration Summary" in the Bison
1594 manual for a summary of their functionality. See the section
1595 "Prologue Alternatives" for a detailed discussion including the
1596 advantages of %code over the traditional Yacc prologue directive.
1597
1598 Bison's Java feature as a whole including its current usage of %code
1599 is still considered experimental.
1600
1625df5b
JD
1601** YYFAIL is deprecated and will eventually be removed.
1602
1603 YYFAIL has existed for many years as an undocumented feature of
1604 deterministic parsers in C generated by Bison. Previously, it was
1605 documented for Bison's experimental Java parsers. YYFAIL is no longer
1606 documented for Java parsers and is formally deprecated in both cases.
1607 Users are strongly encouraged to migrate to YYERROR, which is
1608 specified by POSIX.
1609
1610 Like YYERROR, you can invoke YYFAIL from a semantic action in order to
1611 induce a syntax error. The most obvious difference from YYERROR is
1612 that YYFAIL will automatically invoke yyerror to report the syntax
1613 error so that you don't have to. However, there are several other
1614 subtle differences between YYERROR and YYFAIL, and YYFAIL suffers from
e4ab1254 1615 inherent flaws when %error-verbose or "#define YYERROR_VERBOSE" is
1625df5b
JD
1616 used. For a more detailed discussion, see:
1617
1618 http://lists.gnu.org/archive/html/bison-patches/2009-12/msg00024.html
1619
1620 The upcoming Bison 2.5 will remove YYFAIL from Java parsers, but
1621 deterministic parsers in C will continue to implement it. However,
1622 because YYFAIL is already flawed, it seems futile to try to make new
1623 Bison features compatible with it. Thus, during parser generation,
1624 Bison 2.5 will produce a warning whenever it discovers YYFAIL in a
1625 rule action. In a later release, YYFAIL will be disabled for
e4ab1254 1626 %error-verbose and "#define YYERROR_VERBOSE". Eventually, YYFAIL will
1625df5b
JD
1627 be removed altogether.
1628
1629 There exists at least one case where Bison 2.5's YYFAIL warning will
1630 be a false positive. Some projects add phony uses of YYFAIL and other
1631 Bison-defined macros for the sole purpose of suppressing C
1632 preprocessor warnings (from GCC cpp's -Wunused-macros, for example).
1633 To avoid Bison's future warning, such YYFAIL uses can be moved to the
e4ab1254 1634 epilogue (that is, after the second "%%") in the Bison input file. In
1625df5b
JD
1635 this release (2.4.2), Bison already generates its own code to suppress
1636 C preprocessor warnings for YYFAIL, so projects can remove their own
1637 phony uses of YYFAIL if compatibility with Bison releases prior to
1638 2.4.2 is not necessary.
1639
2755de8f
AD
1640** Internationalization.
1641
1642 Fix a regression introduced in Bison 2.4: Under some circumstances,
1643 message translations were not installed although supported by the
1644 host system.
1645
74553c98 1646* Changes in version 2.4.1 (2008-12-11):
c9ba9e59 1647
0ea583d2
AD
1648** In the GLR defines file, unexpanded M4 macros in the yylval and yylloc
1649 declarations have been fixed.
1979121c 1650
0ea583d2
AD
1651** Temporary hack for adding a semicolon to the user action.
1652
1653 Bison used to prepend a trailing semicolon at the end of the user
1654 action for reductions. This allowed actions such as
1655
1656 exp: exp "+" exp { $$ = $1 + $3 };
1657
1658 instead of
1659
1660 exp: exp "+" exp { $$ = $1 + $3; };
1661
e4ab1254 1662 Some grammars still depend on this "feature". Bison 2.4.1 restores
0ea583d2
AD
1663 the previous behavior in the case of C output (specifically, when
1664 neither %language or %skeleton or equivalent command-line options
1665 are used) to leave more time for grammars depending on the old
1666 behavior to be adjusted. Future releases of Bison will disable this
1667 feature.
1668
1669** A few minor improvements to the Bison manual.
c9ba9e59 1670
402b123d 1671* Changes in version 2.4 (2008-11-02):
7bd1665a 1672
402b123d 1673** %language is an experimental feature.
ed4d67dc
JD
1674
1675 We first introduced this feature in test release 2.3b as a cleaner
1676 alternative to %skeleton. Since then, we have discussed the possibility of
1677 modifying its effect on Bison's output file names. Thus, in this release,
1678 we consider %language to be an experimental feature that will likely evolve
1679 in future releases.
7bd1665a 1680
402b123d 1681** Forward compatibility with GNU M4 has been improved.
241fda7a 1682
402b123d 1683** Several bugs in the C++ skeleton and the experimental Java skeleton have been
241fda7a
JD
1684 fixed.
1685
402b123d 1686* Changes in version 2.3b (2008-05-27):
35fe0834 1687
402b123d 1688** The quotes around NAME that used to be required in the following directive
d9df47b6
JD
1689 are now deprecated:
1690
1691 %define NAME "VALUE"
1692
e4ab1254 1693** The directive "%pure-parser" is now deprecated in favor of:
d9df47b6
JD
1694
1695 %define api.pure
1696
1697 which has the same effect except that Bison is more careful to warn about
1698 unreasonable usage in the latter case.
1699
402b123d 1700** Push Parsing
c373bf8b
JD
1701
1702 Bison can now generate an LALR(1) parser in C with a push interface. That
e4ab1254
AD
1703 is, instead of invoking "yyparse", which pulls tokens from "yylex", you can
1704 push one token at a time to the parser using "yypush_parse", which will
c373bf8b
JD
1705 return to the caller after processing each token. By default, the push
1706 interface is disabled. Either of the following directives will enable it:
1707
1708 %define api.push_pull "push" // Just push; does not require yylex.
1709 %define api.push_pull "both" // Push and pull; requires yylex.
1710
e4ab1254 1711 See the new section "A Push Parser" in the Bison manual for details.
c373bf8b 1712
59da312b
JD
1713 The current push parsing interface is experimental and may evolve. More user
1714 feedback will help to stabilize it.
1715
402b123d 1716** The -g and --graph options now output graphs in Graphviz DOT format,
8e55b3aa
JD
1717 not VCG format. Like --graph, -g now also takes an optional FILE argument
1718 and thus cannot be bundled with other short options.
c373bf8b 1719
402b123d 1720** Java
59da312b
JD
1721
1722 Bison can now generate an LALR(1) parser in Java. The skeleton is
e4ab1254 1723 "data/lalr1.java". Consider using the new %language directive instead of
59da312b
JD
1724 %skeleton to select it.
1725
e4ab1254 1726 See the new section "Java Parsers" in the Bison manual for details.
59da312b
JD
1727
1728 The current Java interface is experimental and may evolve. More user
1729 feedback will help to stabilize it.
2bd435c3 1730 Contributed by Paolo Bonzini.
59da312b 1731
402b123d 1732** %language
59da312b
JD
1733
1734 This new directive specifies the programming language of the generated
d43f77e7
PB
1735 parser, which can be C (the default), C++, or Java. Besides the skeleton
1736 that Bison uses, the directive affects the names of the generated files if
1737 the grammar file's name ends in ".y".
59da312b 1738
402b123d 1739** XML Automaton Report
59da312b
JD
1740
1741 Bison can now generate an XML report of the LALR(1) automaton using the new
e4ab1254 1742 "--xml" option. The current XML schema is experimental and may evolve. More
59da312b 1743 user feedback will help to stabilize it.
2bd435c3 1744 Contributed by Wojciech Polak.
c373bf8b 1745
402b123d 1746** The grammar file may now specify the name of the parser header file using
c373bf8b
JD
1747 %defines. For example:
1748
1749 %defines "parser.h"
1750
402b123d 1751** When reporting useless rules, useless nonterminals, and unused terminals,
d80fb37a
JD
1752 Bison now employs the terms "useless in grammar" instead of "useless",
1753 "useless in parser" instead of "never reduced", and "unused in grammar"
1754 instead of "unused".
cff03fb2 1755
402b123d 1756** Unreachable State Removal
c373bf8b
JD
1757
1758 Previously, Bison sometimes generated parser tables containing unreachable
31984206
JD
1759 states. A state can become unreachable during conflict resolution if Bison
1760 disables a shift action leading to it from a predecessor state. Bison now:
75ad86ee
JD
1761
1762 1. Removes unreachable states.
1763
1764 2. Does not report any conflicts that appeared in unreachable states.
1765 WARNING: As a result, you may need to update %expect and %expect-rr
1766 directives in existing grammar files.
1767
1768 3. For any rule used only in such states, Bison now reports the rule as
cff03fb2 1769 "useless in parser due to conflicts".
75ad86ee 1770
31984206
JD
1771 This feature can be disabled with the following directive:
1772
1773 %define lr.keep_unreachable_states
1774
e4ab1254 1775 See the %define entry in the "Bison Declaration Summary" in the Bison manual
31984206
JD
1776 for further discussion.
1777
e4ab1254 1778** Lookahead Set Correction in the ".output" Report
b1cc23c4 1779
e4ab1254
AD
1780 When instructed to generate a ".output" file including lookahead sets
1781 (using "--report=lookahead", for example), Bison now prints each reduction's
88c78747
JD
1782 lookahead set only next to the associated state's one item that (1) is
1783 associated with the same rule as the reduction and (2) has its dot at the end
1784 of its RHS. Previously, Bison also erroneously printed the lookahead set
1785 next to all of the state's other items associated with the same rule. This
e4ab1254 1786 bug affected only the ".output" file and not the generated parser source
88c78747
JD
1787 code.
1788
e4ab1254 1789** --report-file=FILE is a new option to override the default ".output" file
59da312b 1790 name.
1bb2bd75 1791
e4ab1254 1792** The "=" that used to be required in the following directives is now
02975b9a
JD
1793 deprecated:
1794
1795 %file-prefix "parser"
1796 %name-prefix "c_"
1797 %output "parser.c"
1798
e4ab1254 1799** An Alternative to "%{...%}" -- "%code QUALIFIER {CODE}"
c373bf8b
JD
1800
1801 Bison 2.3a provided a new set of directives as a more flexible alternative to
8e0a5e9e
JD
1802 the traditional Yacc prologue blocks. Those have now been consolidated into
1803 a single %code directive with an optional qualifier field, which identifies
1804 the purpose of the code and thus the location(s) where Bison should generate
1805 it:
1806
e4ab1254
AD
1807 1. "%code {CODE}" replaces "%after-header {CODE}"
1808 2. "%code requires {CODE}" replaces "%start-header {CODE}"
1809 3. "%code provides {CODE}" replaces "%end-header {CODE}"
1810 4. "%code top {CODE}" replaces "%before-header {CODE}"
8e0a5e9e 1811
e4ab1254
AD
1812 See the %code entries in section "Bison Declaration Summary" in the Bison
1813 manual for a summary of the new functionality. See the new section "Prologue
1814 Alternatives" for a detailed discussion including the advantages of %code
8e0a5e9e
JD
1815 over the traditional Yacc prologues.
1816
1817 The prologue alternatives are experimental. More user feedback will help to
1818 determine whether they should become permanent features.
1819
402b123d 1820** Revised warning: unset or unused mid-rule values
17bd8a73
JD
1821
1822 Since Bison 2.2, Bison has warned about mid-rule values that are set but not
1823 used within any of the actions of the parent rule. For example, Bison warns
1824 about unused $2 in:
1825
1826 exp: '1' { $$ = 1; } '+' exp { $$ = $1 + $4; };
1827
1828 Now, Bison also warns about mid-rule values that are used but not set. For
1829 example, Bison warns about unset $$ in the mid-rule action in:
1830
1831 exp: '1' { $1 = 1; } '+' exp { $$ = $2 + $4; };
1832
1833 However, Bison now disables both of these warnings by default since they
1834 sometimes prove to be false alarms in existing grammars employing the Yacc
1835 constructs $0 or $-N (where N is some positive integer).
1836
e4ab1254
AD
1837 To enable these warnings, specify the option "--warnings=midrule-values" or
1838 "-W", which is a synonym for "--warnings=all".
17bd8a73 1839
e4ab1254 1840** Default %destructor or %printer with "<*>" or "<>"
c373bf8b
JD
1841
1842 Bison now recognizes two separate kinds of default %destructor's and
12e35840
JD
1843 %printer's:
1844
e4ab1254 1845 1. Place "<*>" in a %destructor/%printer symbol list to define a default
12e35840
JD
1846 %destructor/%printer for all grammar symbols for which you have formally
1847 declared semantic type tags.
1848
e4ab1254 1849 2. Place "<>" in a %destructor/%printer symbol list to define a default
12e35840
JD
1850 %destructor/%printer for all grammar symbols without declared semantic
1851 type tags.
1852
e4ab1254
AD
1853 Bison no longer supports the "%symbol-default" notation from Bison 2.3a.
1854 "<*>" and "<>" combined achieve the same effect with one exception: Bison no
12e35840
JD
1855 longer applies any %destructor to a mid-rule value if that mid-rule value is
1856 not actually ever referenced using either $$ or $n in a semantic action.
1857
85894313
JD
1858 The default %destructor's and %printer's are experimental. More user
1859 feedback will help to determine whether they should become permanent
1860 features.
1861
e4ab1254 1862 See the section "Freeing Discarded Symbols" in the Bison manual for further
12e35840
JD
1863 details.
1864
402b123d 1865** %left, %right, and %nonassoc can now declare token numbers. This is required
e4ab1254 1866 by POSIX. However, see the end of section "Operator Precedence" in the Bison
ab7f29f8
JD
1867 manual for a caveat concerning the treatment of literal strings.
1868
402b123d 1869** The nonfunctional --no-parser, -n, and %no-parser options have been
b1cc23c4
JD
1870 completely removed from Bison.
1871
402b123d 1872* Changes in version 2.3a, 2006-09-13:
742e4900 1873
402b123d 1874** Instead of %union, you can define and use your own union type
ddc8ede1
PE
1875 YYSTYPE if your grammar contains at least one <type> tag.
1876 Your YYSTYPE need not be a macro; it can be a typedef.
1877 This change is for compatibility with other Yacc implementations,
1878 and is required by POSIX.
1879
402b123d 1880** Locations columns and lines start at 1.
cd48d21d
AD
1881 In accordance with the GNU Coding Standards and Emacs.
1882
402b123d 1883** You may now declare per-type and default %destructor's and %printer's:
ec5479ce
JD
1884
1885 For example:
1886
b2a0b7ca
JD
1887 %union { char *string; }
1888 %token <string> STRING1
1889 %token <string> STRING2
1890 %type <string> string1
1891 %type <string> string2
1892 %union { char character; }
1893 %token <character> CHR
1894 %type <character> chr
1895 %destructor { free ($$); } %symbol-default
1896 %destructor { free ($$); printf ("%d", @$.first_line); } STRING1 string1
1897 %destructor { } <character>
1898
1899 guarantees that, when the parser discards any user-defined symbol that has a
e4ab1254
AD
1900 semantic type tag other than "<character>", it passes its semantic value to
1901 "free". However, when the parser discards a "STRING1" or a "string1", it
1902 also prints its line number to "stdout". It performs only the second
1903 "%destructor" in this case, so it invokes "free" only once.
ec5479ce 1904
85894313
JD
1905 [Although we failed to mention this here in the 2.3a release, the default
1906 %destructor's and %printer's were experimental, and they were rewritten in
1907 future versions.]
1908
e4ab1254
AD
1909** Except for LALR(1) parsers in C with POSIX Yacc emulation enabled (with "-y",
1910 "--yacc", or "%yacc"), Bison no longer generates #define statements for
b931235e
JD
1911 associating token numbers with token names. Removing the #define statements
1912 helps to sanitize the global namespace during preprocessing, but POSIX Yacc
1913 requires them. Bison still generates an enum for token names in all cases.
1914
402b123d 1915** Handling of traditional Yacc prologue blocks is now more consistent but
34f98f46 1916 potentially incompatible with previous releases of Bison.
9bc0dd67
JD
1917
1918 As before, you declare prologue blocks in your grammar file with the
e4ab1254 1919 "%{ ... %}" syntax. To generate the pre-prologue, Bison concatenates all
34f98f46
JD
1920 prologue blocks that you've declared before the first %union. To generate
1921 the post-prologue, Bison concatenates all prologue blocks that you've
ddc8ede1 1922 declared after the first %union.
9bc0dd67 1923
34f98f46 1924 Previous releases of Bison inserted the pre-prologue into both the header
9bc0dd67
JD
1925 file and the code file in all cases except for LALR(1) parsers in C. In the
1926 latter case, Bison inserted it only into the code file. For parsers in C++,
1927 the point of insertion was before any token definitions (which associate
1928 token numbers with names). For parsers in C, the point of insertion was
1929 after the token definitions.
1930
1931 Now, Bison never inserts the pre-prologue into the header file. In the code
1932 file, it always inserts it before the token definitions.
1933
402b123d 1934** Bison now provides a more flexible alternative to the traditional Yacc
34f98f46
JD
1935 prologue blocks: %before-header, %start-header, %end-header, and
1936 %after-header.
1937
1938 For example, the following declaration order in the grammar file reflects the
1939 order in which Bison will output these code blocks. However, you are free to
1940 declare these code blocks in your grammar file in whatever order is most
1941 convenient for you:
1942
1943 %before-header {
1944 /* Bison treats this block like a pre-prologue block: it inserts it into
1945 * the code file before the contents of the header file. It does *not*
1946 * insert it into the header file. This is a good place to put
1947 * #include's that you want at the top of your code file. A common
e4ab1254 1948 * example is '#include "system.h"'. */
34f98f46
JD
1949 }
1950 %start-header {
1951 /* Bison inserts this block into both the header file and the code file.
1952 * In both files, the point of insertion is before any Bison-generated
1953 * token, semantic type, location type, and class definitions. This is a
1954 * good place to define %union dependencies, for example. */
9bc0dd67
JD
1955 }
1956 %union {
34f98f46
JD
1957 /* Unlike the traditional Yacc prologue blocks, the output order for the
1958 * new %*-header blocks is not affected by their declaration position
1959 * relative to any %union in the grammar file. */
9bc0dd67 1960 }
34f98f46
JD
1961 %end-header {
1962 /* Bison inserts this block into both the header file and the code file.
1963 * In both files, the point of insertion is after the Bison-generated
1964 * definitions. This is a good place to declare or define public
1965 * functions or data structures that depend on the Bison-generated
1966 * definitions. */
9bc0dd67 1967 }
34f98f46
JD
1968 %after-header {
1969 /* Bison treats this block like a post-prologue block: it inserts it into
1970 * the code file after the contents of the header file. It does *not*
1971 * insert it into the header file. This is a good place to declare or
1972 * define internal functions or data structures that depend on the
1973 * Bison-generated definitions. */
1974 }
1975
1976 If you have multiple occurrences of any one of the above declarations, Bison
1977 will concatenate the contents in declaration order.
9bc0dd67 1978
85894313
JD
1979 [Although we failed to mention this here in the 2.3a release, the prologue
1980 alternatives were experimental, and they were rewritten in future versions.]
1981
e4ab1254 1982** The option "--report=look-ahead" has been changed to "--report=lookahead".
9e6e7ed2
PE
1983 The old spelling still works, but is not documented and may be removed
1984 in a future release.
742e4900 1985
402b123d 1986* Changes in version 2.3, 2006-06-05:
4ad3ed84 1987
e4ab1254 1988** GLR grammars should now use "YYRECOVERING ()" instead of "YYRECOVERING",
4ad3ed84
PE
1989 for compatibility with LALR(1) grammars.
1990
402b123d 1991** It is now documented that any definition of YYSTYPE or YYLTYPE should
4ad3ed84
PE
1992 be to a type name that does not contain parentheses or brackets.
1993
402b123d 1994* Changes in version 2.2, 2006-05-19:
193d7c70 1995
402b123d 1996** The distribution terms for all Bison-generated parsers now permit
193d7c70
PE
1997 using the parsers in nonfree programs. Previously, this permission
1998 was granted only for Bison-generated LALR(1) parsers in C.
5f4236a0 1999
402b123d 2000** %name-prefix changes the namespace name in C++ outputs.
aa08666d 2001
402b123d 2002** The C++ parsers export their token_type.
5f4236a0 2003
402b123d 2004** Bison now allows multiple %union declarations, and concatenates
d6ca7905
PE
2005 their contents together.
2006
402b123d 2007** New warning: unused values
4d7bc38c
PE
2008 Right-hand side symbols whose values are not used are reported,
2009 if the symbols have destructors. For instance:
affac613 2010
8f3596a6 2011 exp: exp "?" exp ":" exp { $1 ? $1 : $3; }
e9690142
JD
2012 | exp "+" exp
2013 ;
affac613 2014
8f3596a6
AD
2015 will trigger a warning about $$ and $5 in the first rule, and $3 in
2016 the second ($1 is copied to $$ by the default rule). This example
4e26c69e 2017 most likely contains three errors, and could be rewritten as:
affac613 2018
4e26c69e 2019 exp: exp "?" exp ":" exp
e9690142
JD
2020 { $$ = $1 ? $3 : $5; free ($1 ? $5 : $3); free ($1); }
2021 | exp "+" exp
2022 { $$ = $1 ? $1 : $3; if ($1) free ($3); }
2023 ;
affac613 2024
4e26c69e
PE
2025 However, if the original actions were really intended, memory leaks
2026 and all, the warnings can be suppressed by letting Bison believe the
2027 values are used, e.g.:
721be13c 2028
8f3596a6 2029 exp: exp "?" exp ":" exp { $1 ? $1 : $3; (void) ($$, $5); }
e9690142
JD
2030 | exp "+" exp { $$ = $1; (void) $3; }
2031 ;
721be13c 2032
84866159
AD
2033 If there are mid-rule actions, the warning is issued if no action
2034 uses it. The following triggers no warning: $1 and $3 are used.
2035
2036 exp: exp { push ($1); } '+' exp { push ($3); sum (); };
2037
721be13c
PE
2038 The warning is intended to help catching lost values and memory leaks.
2039 If a value is ignored, its associated memory typically is not reclaimed.
affac613 2040
402b123d 2041** %destructor vs. YYABORT, YYACCEPT, and YYERROR.
9d9b8b70
PE
2042 Destructors are now called when user code invokes YYABORT, YYACCEPT,
2043 and YYERROR, for all objects on the stack, other than objects
2044 corresponding to the right-hand side of the current rule.
a85284cf 2045
402b123d 2046** %expect, %expect-rr
035aa4a0
PE
2047 Incorrect numbers of expected conflicts are now actual errors,
2048 instead of warnings.
2049
402b123d 2050** GLR, YACC parsers.
4e26c69e
PE
2051 The %parse-params are available in the destructors (and the
2052 experimental printers) as per the documentation.
4b367315 2053
e4ab1254 2054** Bison now warns if it finds a stray "$" or "@" in an action.
ad6a9b97 2055
402b123d 2056** %require "VERSION"
4e26c69e
PE
2057 This specifies that the grammar file depends on features implemented
2058 in Bison version VERSION or higher.
b50d2359 2059
402b123d 2060** lalr1.cc: The token and value types are now class members.
e14d0ab6
AD
2061 The tokens were defined as free form enums and cpp macros. YYSTYPE
2062 was defined as a free form union. They are now class members:
e4ab1254
AD
2063 tokens are enumerations of the "yy::parser::token" struct, and the
2064 semantic values have the "yy::parser::semantic_type" type.
fb9712a9
AD
2065
2066 If you do not want or can update to this scheme, the directive
e4ab1254 2067 '%define "global_tokens_and_yystype" "1"' triggers the global
b50d2359
AD
2068 definition of tokens and YYSTYPE. This change is suitable both
2069 for previous releases of Bison, and this one.
fb9712a9 2070
b50d2359 2071 If you wish to update, then make sure older version of Bison will
e4ab1254 2072 fail using '%require "2.2"'.
fb9712a9 2073
402b123d 2074** DJGPP support added.
193d7c70 2075\f
402b123d 2076* Changes in version 2.1, 2005-09-16:
1ce59070 2077
402b123d 2078** The C++ lalr1.cc skeleton supports %lex-param.
e14d0ab6 2079
402b123d 2080** Bison-generated parsers now support the translation of diagnostics like
baf785db
PE
2081 "syntax error" into languages other than English. The default
2082 language is still English. For details, please see the new
0410a6e0
PE
2083 Internationalization section of the Bison manual. Software
2084 distributors should also see the new PACKAGING file. Thanks to
2085 Bruno Haible for this new feature.
1ce59070 2086
402b123d 2087** Wording in the Bison-generated parsers has been changed slightly to
1a059451
PE
2088 simplify translation. In particular, the message "memory exhausted"
2089 has replaced "parser stack overflow", as the old message was not
2090 always accurate for modern Bison-generated parsers.
2091
402b123d 2092** Destructors are now called when the parser aborts, for all symbols left
258b75ca
PE
2093 behind on the stack. Also, the start symbol is now destroyed after a
2094 successful parse. In both cases, the behavior was formerly inconsistent.
2095
402b123d 2096** When generating verbose diagnostics, Bison-generated parsers no longer
72f000b0
PE
2097 quote the literal strings associated with tokens. For example, for
2098 a syntax error associated with '%token NUM "number"' they might
2099 print 'syntax error, unexpected number' instead of 'syntax error,
2100 unexpected "number"'.
193d7c70 2101\f
402b123d 2102* Changes in version 2.0, 2004-12-25:
efeed023 2103
402b123d 2104** Possibly-incompatible changes
d7e14fc0 2105
82de6b0d
PE
2106 - Bison-generated parsers no longer default to using the alloca function
2107 (when available) to extend the parser stack, due to widespread
2108 problems in unchecked stack-overflow detection. You can "#define
2109 YYSTACK_USE_ALLOCA 1" to require the use of alloca, but please read
2110 the manual to determine safe values for YYMAXDEPTH in that case.
8dd162d3 2111
82de6b0d
PE
2112 - Error token location.
2113 During error recovery, the location of the syntax error is updated
2114 to cover the whole sequence covered by the error token: it includes
2115 the shifted symbols thrown away during the first part of the error
2116 recovery, and the lookahead rejected during the second part.
18d192f0 2117
82de6b0d
PE
2118 - Semicolon changes:
2119 . Stray semicolons are no longer allowed at the start of a grammar.
2120 . Semicolons are now required after in-grammar declarations.
e342c3be 2121
82de6b0d
PE
2122 - Unescaped newlines are no longer allowed in character constants or
2123 string literals. They were never portable, and GCC 3.4.0 has
2124 dropped support for them. Better diagnostics are now generated if
2125 forget a closing quote.
8dd162d3 2126
82de6b0d 2127 - NUL bytes are no longer allowed in Bison string literals, unfortunately.
f74b6f91 2128
402b123d 2129** New features
1452af69 2130
82de6b0d 2131 - GLR grammars now support locations.
4febdd96 2132
82de6b0d
PE
2133 - New directive: %initial-action.
2134 This directive allows the user to run arbitrary code (including
2135 initializing @$) from yyparse before parsing starts.
1452af69 2136
82de6b0d
PE
2137 - A new directive "%expect-rr N" specifies the expected number of
2138 reduce/reduce conflicts in GLR parsers.
1452af69 2139
e4ab1254 2140 - %token numbers can now be hexadecimal integers, e.g., "%token FOO 0x12d".
82de6b0d 2141 This is a GNU extension.
4febdd96 2142
e4ab1254 2143 - The option "--report=lookahead" was changed to "--report=look-ahead".
9e6e7ed2 2144 [However, this was changed back after 2.3.]
1452af69 2145
82de6b0d 2146 - Experimental %destructor support has been added to lalr1.cc.
1452af69 2147
82de6b0d
PE
2148 - New configure option --disable-yacc, to disable installation of the
2149 yacc command and -ly library introduced in 1.875 for POSIX conformance.
6040d338 2150
402b123d 2151** Bug fixes
d5a3fe37 2152
82de6b0d
PE
2153 - For now, %expect-count violations are now just warnings, not errors.
2154 This is for compatibility with Bison 1.75 and earlier (when there are
2155 reduce/reduce conflicts) and with Bison 1.30 and earlier (when there
2156 are too many or too few shift/reduce conflicts). However, in future
2157 versions of Bison we plan to improve the %expect machinery so that
2158 these violations will become errors again.
3473d0f8 2159
82de6b0d
PE
2160 - Within Bison itself, numbers (e.g., goto numbers) are no longer
2161 arbitrarily limited to 16-bit counts.
d600ee67 2162
82de6b0d 2163 - Semicolons are now allowed before "|" in grammar rules, as POSIX requires.
d600ee67 2164\f
402b123d 2165* Changes in version 1.875, 2003-01-01:
963fcc17 2166
402b123d 2167** The documentation license has been upgraded to version 1.2
dc546b0f 2168 of the GNU Free Documentation License.
75eb3bc4 2169
402b123d 2170** syntax error processing
75eb3bc4 2171
dc546b0f
PE
2172 - In Yacc-style parsers YYLLOC_DEFAULT is now used to compute error
2173 locations too. This fixes bugs in error-location computation.
75eb3bc4 2174
dc546b0f
PE
2175 - %destructor
2176 It is now possible to reclaim the memory associated to symbols
2177 discarded during error recovery. This feature is still experimental.
20daca06 2178
dc546b0f
PE
2179 - %error-verbose
2180 This new directive is preferred over YYERROR_VERBOSE.
74724a70 2181
dc546b0f
PE
2182 - #defining yyerror to steal internal variables is discouraged.
2183 It is not guaranteed to work forever.
d1de5372 2184
402b123d 2185** POSIX conformance
d1de5372 2186
dc546b0f
PE
2187 - Semicolons are once again optional at the end of grammar rules.
2188 This reverts to the behavior of Bison 1.33 and earlier, and improves
2189 compatibility with Yacc.
74724a70 2190
e4ab1254
AD
2191 - "parse error" -> "syntax error"
2192 Bison now uniformly uses the term "syntax error"; formerly, the code
2193 and manual sometimes used the term "parse error" instead. POSIX
2194 requires "syntax error" in diagnostics, and it was thought better to
dc546b0f 2195 be consistent.
74724a70 2196
dc546b0f
PE
2197 - The documentation now emphasizes that yylex and yyerror must be
2198 declared before use. C99 requires this.
d1de5372 2199
dc546b0f
PE
2200 - Bison now parses C99 lexical constructs like UCNs and
2201 backslash-newline within C escape sequences, as POSIX 1003.1-2001 requires.
d1de5372 2202
dc546b0f
PE
2203 - File names are properly escaped in C output. E.g., foo\bar.y is
2204 output as "foo\\bar.y".
6780ca7a 2205
dc546b0f 2206 - Yacc command and library now available
e4ab1254 2207 The Bison distribution now installs a "yacc" command, as POSIX requires.
dc546b0f
PE
2208 Also, Bison now installs a small library liby.a containing
2209 implementations of Yacc-compatible yyerror and main functions.
2210 This library is normally not useful, but POSIX requires it.
6e649e65 2211
dc546b0f 2212 - Type clashes now generate warnings, not errors.
6e649e65 2213
dc546b0f
PE
2214 - If the user does not define YYSTYPE as a macro, Bison now declares it
2215 using typedef instead of defining it as a macro.
2216 For consistency, YYLTYPE is also declared instead of defined.
9501dc6e 2217
402b123d 2218** Other compatibility issues
886a425c 2219
e4ab1254
AD
2220 - %union directives can now have a tag before the "{", e.g., the
2221 directive "%union foo {...}" now generates the C code
2222 "typedef union foo { ... } YYSTYPE;"; this is for Yacc compatibility.
2223 The default union tag is "YYSTYPE", for compatibility with Solaris 9 Yacc.
2224 For consistency, YYLTYPE's struct tag is now "YYLTYPE" not "yyltype".
dc546b0f 2225 This is for compatibility with both Yacc and Bison 1.35.
72f889cc 2226
e4ab1254 2227 - ";" is output before the terminating "}" of an action, for
dc546b0f 2228 compatibility with Bison 1.35.
886a425c 2229
dc546b0f 2230 - Bison now uses a Yacc-style format for conflict reports, e.g.,
e4ab1254 2231 "conflicts: 2 shift/reduce, 1 reduce/reduce".
437c2d80 2232
e4ab1254 2233 - "yystype" and "yyltype" are now obsolescent macros instead of being
dc546b0f
PE
2234 typedefs or tags; they are no longer documented and are planned to be
2235 withdrawn in a future release.
2a8d363a 2236
402b123d 2237** GLR parser notes
2a8d363a 2238
dc546b0f
PE
2239 - GLR and inline
2240 Users of Bison have to decide how they handle the portability of the
e4ab1254 2241 C keyword "inline".
959e5f51 2242
e4ab1254
AD
2243 - "parsing stack overflow..." -> "parser stack overflow"
2244 GLR parsers now report "parser stack overflow" as per the Bison manual.
900c5db5 2245
18ad57b3
AD
2246** %parse-param and %lex-param
2247 The macros YYPARSE_PARAM and YYLEX_PARAM provide a means to pass
2248 additional context to yyparse and yylex. They suffer from several
2249 shortcomings:
2250
2251 - a single argument only can be added,
2252 - their types are weak (void *),
242cc08e 2253 - this context is not passed to ancillary functions such as yyerror,
18ad57b3
AD
2254 - only yacc.c parsers support them.
2255
2256 The new %parse-param/%lex-param directives provide a more precise control.
2257 For instance:
2258
2259 %parse-param {int *nastiness}
2260 %lex-param {int *nastiness}
2261 %parse-param {int *randomness}
2262
2263 results in the following signatures:
2264
2265 int yylex (int *nastiness);
2266 int yyparse (int *nastiness, int *randomness);
2267
2268 or, if both %pure-parser and %locations are used:
2269
2270 int yylex (YYSTYPE *lvalp, YYLTYPE *llocp, int *nastiness);
2271 int yyparse (int *nastiness, int *randomness);
2272
402b123d 2273** Bison now warns if it detects conflicting outputs to the same file,
e4ab1254 2274 e.g., it generates a warning for "bison -d -o foo.h foo.y" since
dc546b0f 2275 that command outputs both code and header to foo.h.
6e40b4eb 2276
402b123d 2277** #line in output files
dc546b0f 2278 - --no-line works properly.
6e40b4eb 2279
402b123d 2280** Bison can no longer be built by a K&R C compiler; it requires C89 or
6e40b4eb
AD
2281 later to be built. This change originally took place a few versions
2282 ago, but nobody noticed until we recently asked someone to try
2283 building Bison with a K&R C compiler.
d600ee67 2284\f
402b123d 2285* Changes in version 1.75, 2002-10-14:
7933f2b5 2286
402b123d 2287** Bison should now work on 64-bit hosts.
7933f2b5 2288
402b123d 2289** Indonesian translation thanks to Tedi Heriyanto.
7933f2b5 2290
402b123d 2291** GLR parsers
f50adbbd
AD
2292 Fix spurious parse errors.
2293
402b123d 2294** Pure parsers
f50adbbd
AD
2295 Some people redefine yyerror to steal yyparse' private variables.
2296 Reenable this trick until an official feature replaces it.
2297
402b123d 2298** Type Clashes
d90c934c
AD
2299 In agreement with POSIX and with other Yaccs, leaving a default
2300 action is valid when $$ is untyped, and $1 typed:
2301
e9690142 2302 untyped: ... typed;
d90c934c
AD
2303
2304 but the converse remains an error:
2305
e9690142 2306 typed: ... untyped;
d90c934c 2307
402b123d 2308** Values of mid-rule actions
d90c934c
AD
2309 The following code:
2310
e9690142 2311 foo: { ... } { $$ = $1; } ...
d90c934c
AD
2312
2313 was incorrectly rejected: $1 is defined in the second mid-rule
2314 action, and is equal to the $$ of the first mid-rule action.
d600ee67 2315\f
402b123d 2316* Changes in version 1.50, 2002-10-04:
adc8c848 2317
402b123d 2318** GLR parsing
676385e2
PH
2319 The declaration
2320 %glr-parser
2321 causes Bison to produce a Generalized LR (GLR) parser, capable of handling
2322 almost any context-free grammar, ambiguous or not. The new declarations
e8832397 2323 %dprec and %merge on grammar rules allow parse-time resolution of
676385e2
PH
2324 ambiguities. Contributed by Paul Hilfinger.
2325
7933f2b5 2326 Unfortunately Bison 1.50 does not work properly on 64-bit hosts
420f93c8
PE
2327 like the Alpha, so please stick to 32-bit hosts for now.
2328
402b123d 2329** Output Directory
8c165d89 2330 When not in Yacc compatibility mode, when the output file was not
e4ab1254
AD
2331 specified, running "bison foo/bar.y" created "foo/bar.c". It
2332 now creates "bar.c".
8c165d89 2333
402b123d 2334** Undefined token
007a50a4 2335 The undefined token was systematically mapped to 2 which prevented
e88dbdbf 2336 the use of 2 by the user. This is no longer the case.
007a50a4 2337
402b123d 2338** Unknown token numbers
e88dbdbf 2339 If yylex returned an out of range value, yyparse could die. This is
007a50a4
AD
2340 no longer the case.
2341
402b123d 2342** Error token
e88dbdbf 2343 According to POSIX, the error token must be 256.
23c5a174
AD
2344 Bison extends this requirement by making it a preference: *if* the
2345 user specified that one of her tokens is numbered 256, then error
2346 will be mapped onto another number.
2347
402b123d 2348** Verbose error messages
e4ab1254 2349 They no longer report "..., expecting error or..." for states where
217598da
AD
2350 error recovery is possible.
2351
402b123d 2352** End token
e4ab1254 2353 Defaults to "$end" instead of "$".
217598da 2354
402b123d 2355** Error recovery now conforms to documentation and to POSIX
68cd8af3
PE
2356 When a Bison-generated parser encounters a syntax error, it now pops
2357 the stack until it finds a state that allows shifting the error
2358 token. Formerly, it popped the stack until it found a state that
2359 allowed some non-error action other than a default reduction on the
2360 error token. The new behavior has long been the documented behavior,
2361 and has long been required by POSIX. For more details, please see
337116ba
PE
2362 Paul Eggert, "Reductions during Bison error handling" (2002-05-20)
2363 <http://lists.gnu.org/archive/html/bug-bison/2002-05/msg00038.html>.
68cd8af3 2364
402b123d 2365** Traces
5504898e
AD
2366 Popped tokens and nonterminals are now reported.
2367
402b123d 2368** Larger grammars
a861a339
PE
2369 Larger grammars are now supported (larger token numbers, larger grammar
2370 size (= sum of the LHS and RHS lengths), larger LALR tables).
2371 Formerly, many of these numbers ran afoul of 16-bit limits;
2372 now these limits are 32 bits on most hosts.
355e7c1c 2373
402b123d 2374** Explicit initial rule
643a5994
AD
2375 Bison used to play hacks with the initial rule, which the user does
2376 not write. It is now explicit, and visible in the reports and
2377 graphs as rule 0.
23c5a174 2378
402b123d 2379** Useless rules
643a5994 2380 Before, Bison reported the useless rules, but, although not used,
77714df2 2381 included them in the parsers. They are now actually removed.
23c5a174 2382
402b123d 2383** Useless rules, useless nonterminals
6b98e4b5
AD
2384 They are now reported, as a warning, with their locations.
2385
402b123d 2386** Rules never reduced
e8832397
AD
2387 Rules that can never be reduced because of conflicts are now
2388 reported.
2389
e4ab1254 2390** Incorrect "Token not used"
11652ab3
AD
2391 On a grammar such as
2392
e29f0771
AD
2393 %token useless useful
2394 %%
2395 exp: '0' %prec useful;
11652ab3
AD
2396
2397 where a token was used to set the precedence of the last rule,
e4ab1254 2398 bison reported both "useful" and "useless" as useless tokens.
11652ab3 2399
402b123d 2400** Revert the C++ namespace changes introduced in 1.31
77714df2 2401 as they caused too many portability hassles.
0179dd65 2402
402b123d 2403** Default locations
b2d52318
AD
2404 By an accident of design, the default computation of @$ was
2405 performed after another default computation was performed: @$ = @1.
2406 The latter is now removed: YYLLOC_DEFAULT is fully responsible of
2407 the computation of @$.
adc8c848 2408
402b123d 2409** Token end-of-file
b7c49edf
AD
2410 The token end of file may be specified by the user, in which case,
2411 the user symbol is used in the reports, the graphs, and the verbose
e4ab1254 2412 error messages instead of "$end", which remains being the default.
b7c49edf 2413 For instance
e29f0771 2414 %token MYEOF 0
b7c49edf 2415 or
e29f0771 2416 %token MYEOF 0 "end of file"
fdbcd8e2 2417
402b123d 2418** Semantic parser
fdbcd8e2
AD
2419 This old option, which has been broken for ages, is removed.
2420
402b123d 2421** New translations
a861a339 2422 Brazilian Portuguese, thanks to Alexandre Folle de Menezes.
84614e13
AD
2423 Croatian, thanks to Denis Lackovic.
2424
402b123d 2425** Incorrect token definitions
e4ab1254
AD
2426 When given
2427 %token 'a' "A"
2428 bison used to output
2429 #define 'a' 65
b87f8b21 2430
402b123d 2431** Token definitions as enums
77714df2
AD
2432 Tokens are output both as the traditional #define's, and, provided
2433 the compiler supports ANSI C or is a C++ compiler, as enums.
e88dbdbf 2434 This lets debuggers display names instead of integers.
77714df2 2435
402b123d 2436** Reports
ec3bc396
AD
2437 In addition to --verbose, bison supports --report=THINGS, which
2438 produces additional information:
b408954b
AD
2439 - itemset
2440 complete the core item sets with their closure
e4ab1254 2441 - lookahead [changed to "look-ahead" in 1.875e through 2.3, but changed back]
9e6e7ed2 2442 explicitly associate lookahead tokens to items
b408954b
AD
2443 - solved
2444 describe shift/reduce conflicts solving.
2445 Bison used to systematically output this information on top of
2446 the report. Solved conflicts are now attached to their states.
ec3bc396 2447
402b123d 2448** Type clashes
9af3fbce
AD
2449 Previous versions don't complain when there is a type clash on
2450 the default action if the rule has a mid-rule action, such as in:
2451
e29f0771
AD
2452 %type <foo> bar
2453 %%
2454 bar: '0' {} '0';
9af3fbce
AD
2455
2456 This is fixed.
a861a339 2457
402b123d 2458** GNU M4 is now required when using Bison.
f987e9d2 2459\f
402b123d 2460* Changes in version 1.35, 2002-03-25:
76551463 2461
402b123d 2462** C Skeleton
76551463
AD
2463 Some projects use Bison's C parser with C++ compilers, and define
2464 YYSTYPE as a class. The recent adjustment of C parsers for data
2465 alignment and 64 bit architectures made this impossible.
2466
2467 Because for the time being no real solution for C++ parser
2468 generation exists, kludges were implemented in the parser to
2469 maintain this use. In the future, when Bison has C++ parsers, this
2470 kludge will be disabled.
2471
2472 This kludge also addresses some C++ problems when the stack was
2473 extended.
76551463 2474\f
402b123d 2475* Changes in version 1.34, 2002-03-12:
76551463 2476
402b123d 2477** File name clashes are detected
76551463 2478 $ bison foo.y -d -o foo.x
e4ab1254 2479 fatal error: header and parser would both be named "foo.x"
76551463 2480
e4ab1254 2481** A missing ";" at the end of a rule triggers a warning
76551463
AD
2482 In accordance with POSIX, and in agreement with other
2483 Yacc implementations, Bison will mandate this semicolon in the near
2484 future. This eases the implementation of a Bison parser of Bison
2485 grammars by making this grammar LALR(1) instead of LR(2). To
2486 facilitate the transition, this release introduces a warning.
2487
402b123d 2488** Revert the C++ namespace changes introduced in 1.31, as they caused too
76551463
AD
2489 many portability hassles.
2490
402b123d 2491** DJGPP support added.
76551463 2492
402b123d 2493** Fix test suite portability problems.
76551463 2494\f
402b123d 2495* Changes in version 1.33, 2002-02-07:
76551463 2496
402b123d 2497** Fix C++ issues
76551463
AD
2498 Groff could not be compiled for the definition of size_t was lacking
2499 under some conditions.
2500
402b123d 2501** Catch invalid @n
76551463
AD
2502 As is done with $n.
2503\f
402b123d 2504* Changes in version 1.32, 2002-01-23:
76551463 2505
402b123d 2506** Fix Yacc output file names
76551463 2507
402b123d 2508** Portability fixes
76551463 2509
402b123d 2510** Italian, Dutch translations
76551463 2511\f
402b123d 2512* Changes in version 1.31, 2002-01-14:
52d1aeee 2513
402b123d 2514** Many Bug Fixes
52d1aeee 2515
402b123d 2516** GNU Gettext and %expect
52d1aeee
MA
2517 GNU Gettext asserts 10 s/r conflicts, but there are 7. Now that
2518 Bison dies on incorrect %expectations, we fear there will be
2519 too many bug reports for Gettext, so _for the time being_, %expect
e4ab1254 2520 does not trigger an error when the input file is named "plural.y".
52d1aeee 2521
402b123d 2522** Use of alloca in parsers
52d1aeee
MA
2523 If YYSTACK_USE_ALLOCA is defined to 0, then the parsers will use
2524 malloc exclusively. Since 1.29, but was not NEWS'ed.
2525
2526 alloca is used only when compiled with GCC, to avoid portability
2527 problems as on AIX.
2528
402b123d 2529** yyparse now returns 2 if memory is exhausted; formerly it dumped core.
b47dbebe 2530
402b123d 2531** When the generated parser lacks debugging code, YYDEBUG is now 0
52d1aeee
MA
2532 (as POSIX requires) instead of being undefined.
2533
402b123d 2534** User Actions
52d1aeee
MA
2535 Bison has always permitted actions such as { $$ = $1 }: it adds the
2536 ending semicolon. Now if in Yacc compatibility mode, the semicolon
2537 is no longer output: one has to write { $$ = $1; }.
2538
402b123d 2539** Better C++ compliance
52d1aeee 2540 The output parsers try to respect C++ namespaces.
76551463 2541 [This turned out to be a failed experiment, and it was reverted later.]
52d1aeee 2542
402b123d 2543** Reduced Grammars
52d1aeee
MA
2544 Fixed bugs when reporting useless nonterminals.
2545
402b123d 2546** 64 bit hosts
52d1aeee
MA
2547 The parsers work properly on 64 bit hosts.
2548
402b123d 2549** Error messages
52d1aeee
MA
2550 Some calls to strerror resulted in scrambled or missing error messages.
2551
402b123d 2552** %expect
52d1aeee
MA
2553 When the number of shift/reduce conflicts is correct, don't issue
2554 any warning.
2555
402b123d 2556** The verbose report includes the rule line numbers.
52d1aeee 2557
402b123d 2558** Rule line numbers are fixed in traces.
52d1aeee 2559
402b123d 2560** Swedish translation
52d1aeee 2561
402b123d 2562** Parse errors
52d1aeee
MA
2563 Verbose parse error messages from the parsers are better looking.
2564 Before: parse error: unexpected `'/'', expecting `"number"' or `'-'' or `'(''
2565 Now: parse error: unexpected '/', expecting "number" or '-' or '('
2566
402b123d 2567** Fixed parser memory leaks.
52d1aeee
MA
2568 When the generated parser was using malloc to extend its stacks, the
2569 previous allocations were not freed.
2570
402b123d 2571** Fixed verbose output file.
52d1aeee
MA
2572 Some newlines were missing.
2573 Some conflicts in state descriptions were missing.
2574
402b123d 2575** Fixed conflict report.
52d1aeee
MA
2576 Option -v was needed to get the result.
2577
402b123d 2578** %expect
52d1aeee
MA
2579 Was not used.
2580 Mismatches are errors, not warnings.
2581
402b123d 2582** Fixed incorrect processing of some invalid input.
52d1aeee 2583
402b123d 2584** Fixed CPP guards: 9foo.h uses BISON_9FOO_H instead of 9FOO_H.
52d1aeee 2585
402b123d 2586** Fixed some typos in the documentation.
52d1aeee 2587
402b123d 2588** %token MY_EOF 0 is supported.
52d1aeee
MA
2589 Before, MY_EOF was silently renumbered as 257.
2590
402b123d 2591** doc/refcard.tex is updated.
52d1aeee 2592
402b123d 2593** %output, %file-prefix, %name-prefix.
52d1aeee
MA
2594 New.
2595
402b123d 2596** --output
e4ab1254 2597 New, aliasing "--output-file".
52d1aeee 2598\f
402b123d 2599* Changes in version 1.30, 2001-10-26:
342b8b6e 2600
e4ab1254
AD
2601** "--defines" and "--graph" have now an optional argument which is the
2602 output file name. "-d" and "-g" do not change; they do not take any
342b8b6e
AD
2603 argument.
2604
e4ab1254 2605** "%source_extension" and "%header_extension" are removed, failed
342b8b6e
AD
2606 experiment.
2607
402b123d 2608** Portability fixes.
f987e9d2 2609\f
402b123d 2610* Changes in version 1.29, 2001-09-07:
342b8b6e 2611
402b123d 2612** The output file does not define const, as this caused problems when used
342b8b6e
AD
2613 with common autoconfiguration schemes. If you still use ancient compilers
2614 that lack const, compile with the equivalent of the C compiler option
e4ab1254 2615 "-Dconst=". Autoconf's AC_C_CONST macro provides one way to do this.
342b8b6e 2616
e4ab1254 2617** Added "-g" and "--graph".
f87a2205 2618
402b123d 2619** The Bison manual is now distributed under the terms of the GNU FDL.
f2b5126e 2620
402b123d 2621** The input and the output files has automatically a similar extension.
234a3be3 2622
402b123d 2623** Russian translation added.
f87a2205 2624
402b123d 2625** NLS support updated; should hopefully be less troublesome.
f87a2205 2626
402b123d 2627** Added the old Bison reference card.
c33638bb 2628
e4ab1254 2629** Added "--locations" and "%locations".
6deb4447 2630
e4ab1254 2631** Added "-S" and "--skeleton".
cd5bd6ac 2632
e4ab1254 2633** "%raw", "-r", "--raw" is disabled.
62ab6972 2634
402b123d 2635** Special characters are escaped when output. This solves the problems
cd5bd6ac
AD
2636 of the #line lines with path names including backslashes.
2637
402b123d 2638** New directives.
e4ab1254
AD
2639 "%yacc", "%fixed_output_files", "%defines", "%no_parser", "%verbose",
2640 "%debug", "%source_extension" and "%header_extension".
f987e9d2 2641
402b123d 2642** @$
f987e9d2 2643 Automatic location tracking.
f87a2205 2644\f
402b123d 2645* Changes in version 1.28, 1999-07-06:
d2e00347 2646
402b123d 2647** Should compile better now with K&R compilers.
d2e00347 2648
402b123d 2649** Added NLS.
d2e00347 2650
402b123d 2651** Fixed a problem with escaping the double quote character.
d2e00347 2652
402b123d 2653** There is now a FAQ.
d2e00347 2654\f
402b123d 2655* Changes in version 1.27:
5c31c3c2 2656
402b123d 2657** The make rule which prevented bison.simple from being created on
5c31c3c2
JT
2658 some systems has been fixed.
2659\f
402b123d 2660* Changes in version 1.26:
4be07551 2661
7e508a2b 2662** Bison now uses Automake.
4be07551 2663
402b123d 2664** New mailing lists: <bug-bison@gnu.org> and <help-bison@gnu.org>.
4be07551 2665
402b123d 2666** Token numbers now start at 257 as previously documented, not 258.
4be07551 2667
402b123d 2668** Bison honors the TMPDIR environment variable.
4be07551 2669
402b123d 2670** A couple of buffer overruns have been fixed.
f51dbca1 2671
402b123d 2672** Problems when closing files should now be reported.
f51dbca1 2673
402b123d 2674** Generated parsers should now work even on operating systems which do
f51dbca1 2675 not provide alloca().
4be07551 2676\f
402b123d 2677* Changes in version 1.25, 1995-10-16:
df8878c5 2678
402b123d 2679** Errors in the input grammar are not fatal; Bison keeps reading
df8878c5 2680the grammar file, and reports all the errors found in it.
8c44d3ec 2681
402b123d 2682** Tokens can now be specified as multiple-character strings: for
df8878c5 2683example, you could use "<=" for a token which looks like <=, instead
7e508a2b 2684of choosing a name like LESSEQ.
df8878c5 2685
402b123d 2686** The %token_table declaration says to write a table of tokens (names
df8878c5
RS
2687and numbers) into the parser file. The yylex function can use this
2688table to recognize multiple-character string tokens, or for other
2689purposes.
2690
402b123d 2691** The %no_lines declaration says not to generate any #line preprocessor
df8878c5
RS
2692directives in the parser file.
2693
402b123d 2694** The %raw declaration says to use internal Bison token numbers, not
df8878c5
RS
2695Yacc-compatible token numbers, when token names are defined as macros.
2696
402b123d 2697** The --no-parser option produces the parser tables without including
df8878c5
RS
2698the parser engine; a project can now use its own parser engine.
2699The actions go into a separate file called NAME.act, in the form of
2700a switch statement body.
2701\f
402b123d 2702* Changes in version 1.23:
6780ca7a 2703
4d019228
DM
2704The user can define YYPARSE_PARAM as the name of an argument to be
2705passed into yyparse. The argument should have type void *. It should
2706actually point to an object. Grammar actions can access the variable
2707by casting it to the proper pointer type.
6780ca7a 2708
6780ca7a 2709Line numbers in output file corrected.
6780ca7a 2710\f
402b123d 2711* Changes in version 1.22:
6780ca7a
DM
2712
2713--help option added.
6780ca7a 2714\f
402b123d 2715* Changes in version 1.20:
6780ca7a
DM
2716
2717Output file does not redefine const for C++.
9f4503d6 2718
76551463
AD
2719-----
2720
fc51acdd 2721Copyright (C) 1995-2014 Free Software Foundation, Inc.
76551463 2722
74553c98 2723This file is part of Bison, the GNU Parser Generator.
76551463 2724
f16b0819 2725This program is free software: you can redistribute it and/or modify
76551463 2726it under the terms of the GNU General Public License as published by
f16b0819
PE
2727the Free Software Foundation, either version 3 of the License, or
2728(at your option) any later version.
76551463 2729
f16b0819 2730This program is distributed in the hope that it will be useful,
76551463
AD
2731but WITHOUT ANY WARRANTY; without even the implied warranty of
2732MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
2733GNU General Public License for more details.
2734
2735You should have received a copy of the GNU General Public License
f16b0819 2736along with this program. If not, see <http://www.gnu.org/licenses/>.
7e508a2b
AD
2737
2738 LocalWords: yacc YYBACKUP glr GCC lalr ArrayIndexOutOfBoundsException nullptr
2739 LocalWords: cplusplus liby rpl fprintf mfcalc Wyacc stmt cond expr mk sym lr
2740 LocalWords: IELR ielr Lookahead YYERROR nonassoc LALR's api lookaheads yychar
2741 LocalWords: destructor lookahead YYRHSLOC YYLLOC Rhs ifndef YYFAIL cpp sr rr
2742 LocalWords: preprocessor initializer Wno Wnone Werror FreeBSD prec livelocks
2743 LocalWords: Solaris AIX UX RHEL Tru LHS gcc's Wundef YYENABLE NLS YYLTYPE VCG
2744 LocalWords: yyerror cpp's Wunused yylval yylloc prepend yyparse yylex yypush
2745 LocalWords: Graphviz xml nonterminals midrule destructor's YYSTYPE typedef ly
2746 LocalWords: CHR chr printf stdout namespace preprocessing enum pre include's
2747 LocalWords: YYRECOVERING nonfree destructors YYABORT YYACCEPT params enums de
2748 LocalWords: struct yystype DJGPP lex param Haible NUM alloca YYSTACK NUL goto
2749 LocalWords: YYMAXDEPTH Unescaped UCNs YYLTYPE's yyltype typedefs inline Yaccs
2750 LocalWords: Heriyanto Reenable dprec Hilfinger Eggert MYEOF Folle Menezes EOF
242cc08e 2751 LocalWords: Lackovic define's itemset Groff Gettext malloc NEWS'ed YYDEBUG YY
7e508a2b 2752 LocalWords: namespaces strerror const autoconfiguration Dconst Autoconf's FDL
242cc08e
AD
2753 LocalWords: Automake TMPDIR LESSEQ ylwrap endif yydebug YYTOKEN YYLSP ival hh
2754 LocalWords: extern YYTOKENTYPE TOKENTYPE yytokentype tokentype STYPE lval pdf
dcb366b1 2755 LocalWords: lang yyoutput dvi html ps POSIX lvalp llocp Wother nterm arg init
1282c124
AD
2756 LocalWords: TOK calc yyo fval Wconflicts parsers yystackp yyval yynerrs
2757 LocalWords: Théophile Ranquet Santet fno fnone stype associativity Tolmer
2758 LocalWords: Wprecedence Rassoul Wempty Paolo Bonzini parser's Michiel loc
2759 LocalWords: redeclaration sval fcaret reentrant XSLT xsl Wmaybe yyvsp Tedi
2760 LocalWords: pragmas noreturn untyped Rozenman unexpanded Wojciech Polak
6574576c 2761 LocalWords: Alexandre MERCHANTABILITY yytype
7e508a2b
AD
2762
2763Local Variables:
2764mode: outline
e4ab1254 2765fill-column: 76
7e508a2b 2766End: