]> git.saurik.com Git - bison.git/blame - NEWS
Merge remote-tracking branch 'origin/maint'
[bison.git] / NEWS
CommitLineData
ed7658fe 1GNU Bison NEWS
3af4feb2 2
52718713 3* Noteworthy changes in release ?.? (????-??-??) [?]
74553c98 4
47db7ed1
AD
5** Incompatible changes
6
7*** Obsolete features
8
de62ae86 9 Support for YYFAIL is removed (deprecated in Bison 2.4.2).
5f819b2f 10 Support for yystype and yyltype (instead of YYSTYPE and YYLTYPE)
de62ae86
AD
11 is removed (deprecated in Bison 1.875).
12 Support for YYPARSE_PARAM is removed (deprecated in Bison 1.875).
47db7ed1 13
73370a9d
VS
14** Warnings
15
518e8830
AD
16*** Deprecated constructs
17
1f77b2e0
AD
18 The new 'deprecated' warning category flags obsolete constructs whose
19 support will be discontinued. It is enabled by default. These warnings
20 used to be reported as 'other' warnings.
518e8830 21
34904c57 22*** Warning categories are now displayed
73370a9d
VS
23
24 For instance:
25
26 foo.y:4.6: warning: type clash on default action: <foo> != <bar> [-Wother]
27
28*** Useless semantic types
9641b918
VS
29
30 Bison now warns about useless (uninhabited) semantic types. Since
31 semantic types are not declared to Bison (they are defined in the opaque
32 %union structure), it is %printer/%destructor directives about useless
33 types that trigger the warning:
34
35 %token <type1> term
36 %type <type2> nterm
37 %printer {} <type1> <type3>
38 %destructor {} <type2> <type4>
39 %%
40 nterm: term { $$ = $1; };
41
42 3.28-34: warning: type <type3> is used, but is not associated to any symbol
43 4.28-34: warning: type <type4> is used, but is not associated to any symbol
44
31557b9e 45*** Undefined but unused symbols
b921d92f 46
31557b9e
AD
47 Bison used to raise an error for undefined symbols that are not used in
48 the grammar. This is now only a warning.
b921d92f
VS
49
50 %printer {} symbol1
51 %destructor {} symbol2
31557b9e 52 %type <type> symbol3
b921d92f
VS
53 %%
54 exp: "a";
55
73370a9d 56*** Useless destructors or printers
ea9a35c6
VS
57
58 Bison now warns about useless destructors or printers. In the following
59 example, the printer for <type1>, and the destructor for <type2> are
60 useless: all symbols of <type1> (token1) already have a printer, and all
61 symbols of type <type2> (token2) already have a destructor.
62
63 %token <type1> token1
64 <type2> token2
65 <type3> token3
66 <type4> token4
67 %printer {} token1 <type1> <type3>
68 %destructor {} token2 <type2> <type4>
69
d87ea54c
AD
70*** Conflicts
71
72 The warnings and error messages about shift/reduce and reduce/reduce
73 conflicts have been normalized. For instance on the following foo.y file:
74
75 %glr-parser
76 %%
77 exp: exp '+' exp | '0' | '0';
78
79 compare the previous version of bison:
80
81 $ bison foo.y
82 foo.y: conflicts: 1 shift/reduce, 2 reduce/reduce
83 $ bison -Werror foo.y
84 bison: warnings being treated as errors
85 foo.y: conflicts: 1 shift/reduce, 2 reduce/reduce
86
87 with the new behavior:
88
89 $ bison foo.y
90 foo.y: warning: 1 shift/reduce conflict [-Wconflicts-sr]
91 foo.y: warning: 2 reduce/reduce conflicts [-Wconflicts-rr]
92 $ bison -Werror foo.y
93 bison: warnings being treated as errors
94 foo.y: warning: 1 shift/reduce conflict [-Wconflicts-sr]
95 foo.y: warning: 2 reduce/reduce conflicts [-Wconflicts-rr]
96
97 When %expect or %expect-rr is used, such as with bar.y:
98
99 %expect 0
100 %glr-parser
101 %%
102 exp: exp '+' exp | '0' | '0';
103
104 Former behavior:
105
106 $ bison bar.y
107 bar.y: conflicts: 1 shift/reduce, 2 reduce/reduce
108 bar.y: expected 0 shift/reduce conflicts
109 bar.y: expected 0 reduce/reduce conflicts
110
111 New one:
112
113 $ bison bar.y
114 bar.y: shift/reduce conflicts: 1 found, 0 expected
115 bar.y: reduce/reduce conflicts: 2 found, 0 expected
116
2055a44e
AD
117** Additional yylex/yyparse arguments
118
6dc4663d
AD
119 The new directive %param declares additional arguments to both yylex and
120 yyparse. The %lex-param, %parse-param, and %param directives support one
121 or more arguments. Instead of
2055a44e 122
e436fa67
AD
123 %lex-param {arg1_type *arg1}
124 %lex-param {arg2_type *arg2}
125 %parse-param {arg1_type *arg1}
126 %parse-param {arg2_type *arg2}
2055a44e
AD
127
128 one may now declare
129
e436fa67 130 %param {arg1_type *arg1} {arg2_type *arg2}
2055a44e
AD
131
132** Java skeleton improvements
0ea583d2 133
1f77b2e0
AD
134 The constants for token names were moved to the Lexer interface. Also, it
135 is possible to add code to the parser's constructors using "%code init"
136 and "%define init_throws".
0ea583d2 137
34904c57
AD
138** C++ skeletons improvements
139
e4c0985b 140*** The parser header is no longer mandatory (lalr1.cc, glr.cc)
93549bcd 141
e4c0985b
AD
142 Using %defines is now optional. Without it, the needed support classes
143 are defined in the generated parser, instead of additional files (such as
144 location.hh, position.hh and stack.hh).
93549bcd 145
e4c0985b 146*** Locations are no longer mandatory (lalr1.cc, glr.cc)
34904c57
AD
147
148 Both lalr1.cc and glr.cc no longer require %location.
149
150*** syntax_error exception (lalr1.cc)
a6552c5d
AD
151
152 The C++ parser features a syntax_error exception, which can be
153 thrown from the scanner or from user rules to raise syntax errors.
154 This facilitates reporting errors caught in sub-functions (e.g.,
155 rejecting too large integral literals from a conversion function
156 used by the scanner, or rejecting invalid combinations from a
157 factory invoked by the user actions).
158
4c6622c2 159** Variable api.tokens.prefix
99c08fb6 160
4c6622c2 161 The variable api.tokens.prefix changes the way tokens are identified in
99c08fb6
AD
162 the generated files. This is especially useful to avoid collisions
163 with identifiers in the target language. For instance
164
e436fa67
AD
165 %token FILE for ERROR
166 %define api.tokens.prefix "TOK_"
167 %%
168 start: FILE for ERROR;
99c08fb6
AD
169
170 will generate the definition of the symbols TOK_FILE, TOK_for, and
171 TOK_ERROR in the generated sources. In particular, the scanner must
172 use these prefixed token names, although the grammar itself still
173 uses the short names (as in the sample rule given above).
174
67501061
AD
175** Variable api.namespace
176
1f77b2e0 177 The 'namespace' variable is renamed 'api.namespace'. Backward
67501061
AD
178 compatibility is ensured, but upgrading is recommended.
179
31b850d2
AD
180** Variable parse.error
181
1f77b2e0
AD
182 This variable controls the verbosity of error messages. The use of the
183 %error-verbose directive is deprecated in favor of "%define parse.error
184 verbose".
31b850d2 185
ca2a6d15
PH
186** Semantic predicates
187
1f77b2e0
AD
188 The new, experimental, semantic-predicate feature allows actions of the
189 form "%?{ BOOLEAN-EXPRESSION }", which cause syntax errors (as for
ca2a6d15 190 YYERROR) if the expression evaluates to 0, and are evaluated immediately
1f77b2e0
AD
191 in GLR parsers, rather than being deferred. The result is that they allow
192 the programmer to prune possible parses based on the values of run-time
193 expressions.
ca2a6d15 194
d1400569
AD
195** The directive %expect-rr is now an error in non GLR mode
196
197 It used to be an error only if used in non GLR mode, _and_ if there are
198 reduce/reduce conflicts.
199
e436fa67
AD
200* Noteworthy changes in release ?.? (????-??-??) [?]
201
6b4cb804
AD
202** Bug fixes
203
204 Bugs in the test suite have been fixed.
205
206 Some errors in translations have been addressed, and --help now directs
207 users to the appropriate place to report them.
208
209 Stray Info files shipped by accident are removed.
210
211 Incorrect definitions of YY_, issued by yacc.c when no parser header is
212 generated, are removed.
9c26b8fc 213
e1eeecd3 214* Noteworthy changes in release 2.6.2 (2012-08-03) [stable]
9c26b8fc 215
43ca8040
AD
216** Bug fixes
217
218 Buffer overruns, complaints from Flex, and portability issues in the test
219 suite have been fixed.
220
c9d546b2
AD
221** Spaces in %lex- and %parse-param (lalr1.cc, glr.cc)
222
223 Trailing end-of-lines in %parse-param or %lex-param would result in
224 invalid C++. This is fixed.
9c26b8fc 225
dcd5344d
AD
226** Spurious spaces and end-of-lines
227
228 The generated files no longer end (nor start) with empty lines.
229
77b214ef 230* Noteworthy changes in release 2.6.1 (2012-07-30) [stable]
a4107f24 231
8617d87e
AD
232 Bison no longer executes user-specified M4 code when processing a grammar.
233
e20e6a50
AD
234** Future Changes
235
236 In addition to the removal of the features announced in Bison 2.6, the
237 next major release will remove the "Temporary hack for adding a semicolon
238 to the user action", as announced in the release 2.5. Instead of:
239
240 exp: exp "+" exp { $$ = $1 + $3 };
241
242 write:
243
244 exp: exp "+" exp { $$ = $1 + $3; };
245
8617d87e
AD
246** Bug fixes
247
0e164d43
AD
248*** Type names are now properly escaped.
249
250*** glr.cc: set_debug_level and debug_level work as expected.
a4107f24 251
26313726
AD
252*** Stray @ or $ in actions
253
254 While Bison used to warn about stray $ or @ in action rules, it did not
255 for other actions such as printers, destructors, or initial actions. It
256 now does.
257
cd735a8c 258** Type names in actions
4982f078
AD
259
260 For consistency with rule actions, it is now possible to qualify $$ by a
cd735a8c 261 type-name in destructors, printers, and initial actions. For instance:
4982f078
AD
262
263 %printer { fprintf (yyo, "(%d, %f)", $<ival>$, $<fval>$); } <*> <>;
264
265 will display two values for each typed and untyped symbol (provided
cd735a8c 266 that YYSTYPE has both "ival" and "fval" fields).
60aa04a2 267
1505e8bb 268* Noteworthy changes in release 2.6 (2012-07-19) [stable]
0f11eec2 269
9553083c
AD
270** Future changes:
271
55d1006f
AD
272 The next major release of Bison will drop support for the following
273 deprecated features. Please report disagreements to bug-bison@gnu.org.
0f11eec2 274
aaf61036 275*** K&R C parsers
55d1006f
AD
276
277 Support for generating parsers in K&R C will be removed. Parsers
242cc08e 278 generated for C support ISO C90, and are tested with ISO C99 and ISO C11
55d1006f
AD
279 compilers.
280
258cddbc 281*** Features deprecated since Bison 1.875
0f11eec2 282
258cddbc
AD
283 The definitions of yystype and yyltype will be removed; use YYSTYPE and
284 YYLTYPE.
0f11eec2 285
258cddbc
AD
286 YYPARSE_PARAM and YYLEX_PARAM, deprecated in favor of %parse-param and
287 %lex-param, will no longer be supported.
288
289 Support for the preprocessor symbol YYERROR_VERBOSE will be removed, use
290 %error-verbose.
55d1006f
AD
291
292*** The generated header will be included (yacc.c)
0f11eec2
AD
293
294 Instead of duplicating the content of the generated header (definition of
55d1006f
AD
295 YYSTYPE, yyparse declaration etc.), the generated parser will include it,
296 as is already the case for GLR or C++ parsers. This change is deferred
297 because existing versions of ylwrap (e.g., Automake 1.12.1) do not support
298 it.
0f11eec2 299
c2425191 300** Generated Parser Headers
56ca3d8f 301
258cddbc 302*** Guards (yacc.c, glr.c, glr.cc)
c3e9f08f
AD
303
304 The generated headers are now guarded, as is already the case for C++
242cc08e 305 parsers (lalr1.cc). For instance, with --defines=foo.h:
c3e9f08f 306
e29f0771
AD
307 #ifndef YY_FOO_H
308 # define YY_FOO_H
309 ...
310 #endif /* !YY_FOO_H */
c3e9f08f 311
258cddbc 312*** New declarations (yacc.c, glr.c)
56ca3d8f
AD
313
314 The generated header now declares yydebug and yyparse. Both honor
315 --name-prefix=bar_, and yield
316
e29f0771 317 int bar_parse (void);
56ca3d8f
AD
318
319 rather than
320
e29f0771
AD
321 #define yyparse bar_parse
322 int yyparse (void);
56ca3d8f
AD
323
324 in order to facilitate the inclusion of several parser headers inside a
325 single compilation unit.
c3e9f08f 326
258cddbc
AD
327*** Exported symbols in C++
328
329 The symbols YYTOKEN_TABLE and YYERROR_VERBOSE, which were defined in the
330 header, are removed, as they prevent the possibility of including several
331 generated headers from a single compilation unit.
332
694af10c
AD
333*** YYLSP_NEEDED
334
335 For the same reasons, the undocumented and unused macro YYLSP_NEEDED is no
336 longer defined.
337
4b3847c3
AD
338** New %define variable: api.prefix
339
340 Now that the generated headers are more complete and properly protected
341 against multiple inclusions, constant names, such as YYSTYPE are a
342 problem. While yyparse and others are properly renamed by %name-prefix,
343 YYSTYPE, YYDEBUG and others have never been affected by it. Because it
344 would introduce backward compatibility issues in projects not expecting
345 YYSTYPE to be renamed, instead of changing the behavior of %name-prefix,
346 it is deprecated in favor of a new %define variable: api.prefix.
347
348 The following examples compares both:
349
350 %name-prefix "bar_" | %define api.prefix "bar_"
351 %token <ival> FOO %token <ival> FOO
352 %union { int ival; } %union { int ival; }
353 %% %%
354 exp: 'a'; exp: 'a';
355
356 bison generates:
357
358 #ifndef BAR_FOO_H #ifndef BAR_FOO_H
359 # define BAR_FOO_H # define BAR_FOO_H
360
361 /* Enabling traces. */ /* Enabling traces. */
5f108727
AD
362 # ifndef YYDEBUG | # ifndef BAR_DEBUG
363 > # if defined YYDEBUG
364 > # if YYDEBUG
365 > # define BAR_DEBUG 1
366 > # else
367 > # define BAR_DEBUG 0
368 > # endif
369 > # else
370 # define YYDEBUG 0 | # define BAR_DEBUG 0
371 > # endif
372 # endif | # endif
373
374 # if YYDEBUG | # if BAR_DEBUG
4b3847c3
AD
375 extern int bar_debug; extern int bar_debug;
376 # endif # endif
377
378 /* Tokens. */ /* Tokens. */
379 # ifndef YYTOKENTYPE | # ifndef BAR_TOKENTYPE
380 # define YYTOKENTYPE | # define BAR_TOKENTYPE
381 enum yytokentype { | enum bar_tokentype {
382 FOO = 258 FOO = 258
383 }; };
384 # endif # endif
385
386 #if ! defined YYSTYPE \ | #if ! defined BAR_STYPE \
387 && ! defined YYSTYPE_IS_DECLARED | && ! defined BAR_STYPE_IS_DECLARED
388 typedef union YYSTYPE | typedef union BAR_STYPE
389 { {
390 int ival; int ival;
391 } YYSTYPE; | } BAR_STYPE;
392 # define YYSTYPE_IS_DECLARED 1 | # define BAR_STYPE_IS_DECLARED 1
393 #endif #endif
394
395 extern YYSTYPE bar_lval; | extern BAR_STYPE bar_lval;
396
397 int bar_parse (void); int bar_parse (void);
398
399 #endif /* !BAR_FOO_H */ #endif /* !BAR_FOO_H */
400
dfaac272 401* Noteworthy changes in release 2.5.1 (2012-06-05) [stable]
df6e3db0 402
debe2c03 403** Future changes:
765e1bd4 404
e4ab1254 405 The next major release will drop support for generating parsers in K&R C.
041308d0 406
466b4cf2 407** yacc.c: YYBACKUP works as expected.
ef51bfa7 408
d834eca0 409** glr.c improvements:
041308d0 410
d834eca0 411*** Location support is eliminated when not requested:
041308d0 412
e4ab1254
AD
413 GLR parsers used to include location-related code even when locations were
414 not requested, and therefore not even usable.
378e917c 415
d834eca0 416*** __attribute__ is preserved:
d115aad9 417
e4ab1254
AD
418 __attribute__ is no longer disabled when __STRICT_ANSI__ is defined (i.e.,
419 when -std is passed to GCC).
041308d0 420
466b4cf2 421** lalr1.java: several fixes:
041308d0 422
e4ab1254
AD
423 The Java parser no longer throws ArrayIndexOutOfBoundsException if the
424 first token leads to a syntax error. Some minor clean ups.
041308d0 425
22172d47 426** Changes for C++:
ef51bfa7 427
22172d47 428*** C++11 compatibility:
ef51bfa7 429
e4ab1254
AD
430 C and C++ parsers use "nullptr" instead of "0" when __cplusplus is 201103L
431 or higher.
936c88d1 432
22172d47
AD
433*** Header guards
434
435 The header files such as "parser.hh", "location.hh", etc. used a constant
436 name for preprocessor guards, for instance:
437
e29f0771
AD
438 #ifndef BISON_LOCATION_HH
439 # define BISON_LOCATION_HH
440 ...
441 #endif // !BISON_LOCATION_HH
22172d47
AD
442
443 The inclusion guard is now computed from "PREFIX/FILE-NAME", where lower
444 case characters are converted to upper case, and series of
445 non-alphanumerical characters are converted to an underscore.
446
447 With "bison -o lang++/parser.cc", "location.hh" would now include:
448
e29f0771
AD
449 #ifndef YY_LANG_LOCATION_HH
450 # define YY_LANG_LOCATION_HH
451 ...
452 #endif // !YY_LANG_LOCATION_HH
22172d47
AD
453
454*** C++ locations:
936c88d1 455
e4ab1254
AD
456 The position and location constructors (and their initialize methods)
457 accept new arguments for line and column. Several issues in the
458 documentation were fixed.
936c88d1 459
466b4cf2
AD
460** liby is no longer asking for "rpl_fprintf" on some platforms.
461
7e508a2b
AD
462** Changes in the manual:
463
464*** %printer is documented
465
e4ab1254
AD
466 The "%printer" directive, supported since at least Bison 1.50, is finally
467 documented. The "mfcalc" example is extended to demonstrate it.
7e508a2b 468
e4ab1254
AD
469 For consistency with the C skeletons, the C++ parsers now also support
470 "yyoutput" (as an alias to "debug_stream ()").
7e508a2b
AD
471
472*** Several improvements have been made:
466b4cf2 473
e4ab1254
AD
474 The layout for grammar excerpts was changed to a more compact scheme.
475 Named references are motivated. The description of the automaton
476 description file (*.output) is updated to the current format. Incorrect
477 index entries were fixed. Some other errors were fixed.
466b4cf2 478
86b08b49
AD
479** Building bison:
480
481*** Conflicting prototypes with recent/modified Flex.
466b4cf2 482
e4ab1254
AD
483 Fixed build problems with the current, unreleased, version of Flex, and
484 some modified versions of 2.5.35, which have modified function prototypes.
466b4cf2 485
8ef26c2a
AD
486*** Warnings during the build procedure have been eliminated.
487
488*** Several portability problems in the test suite have been fixed:
466b4cf2 489
e4ab1254
AD
490 This includes warnings with some compilers, unexpected behavior of tools
491 such as diff, warning messages from the test suite itself, etc.
466b4cf2 492
91aadcc7 493*** The install-pdf target works properly:
8ef26c2a 494
e4ab1254
AD
495 Running "make install-pdf" (or -dvi, -html, -info, and -ps) no longer
496 halts in the middle of its course.
8ef26c2a 497
28801043 498* Changes in version 2.5 (2011-05-14):
50cca368 499
82f3355e
JD
500** Grammar symbol names can now contain non-initial dashes:
501
502 Consistently with directives (such as %error-verbose) and with
503 %define variables (e.g. push-pull), grammar symbol names may contain
504 dashes in any position except the beginning. This is a GNU
505 extension over POSIX Yacc. Thus, use of this extension is reported
506 by -Wyacc and rejected in Yacc mode (--yacc).
507
f1b238df 508** Named references:
66381412
AR
509
510 Historically, Yacc and Bison have supported positional references
511 ($n, $$) to allow access to symbol values from inside of semantic
512 actions code.
513
514 Starting from this version, Bison can also accept named references.
515 When no ambiguity is possible, original symbol names may be used
516 as named references:
517
4b568fc0 518 if_stmt : "if" cond_expr "then" then_stmt ';'
66381412
AR
519 { $if_stmt = mk_if_stmt($cond_expr, $then_stmt); }
520
521 In the more common case, explicit names may be declared:
522
4b568fc0 523 stmt[res] : "if" expr[cond] "then" stmt[then] "else" stmt[else] ';'
66381412
AR
524 { $res = mk_if_stmt($cond, $then, $else); }
525
5b1ff423 526 Location information is also accessible using @name syntax. When
66381412
AR
527 accessing symbol names containing dots or dashes, explicit bracketing
528 ($[sym.1]) must be used.
529
5b1ff423 530 These features are experimental in this version. More user feedback
66381412
AR
531 will help to stabilize them.
532
f1b238df 533** IELR(1) and canonical LR(1):
eb45ef3b
JD
534
535 IELR(1) is a minimal LR(1) parser table generation algorithm. That
536 is, given any context-free grammar, IELR(1) generates parser tables
7262f54f 537 with the full language-recognition power of canonical LR(1) but with
f1b238df
JD
538 nearly the same number of parser states as LALR(1). This reduction
539 in parser states is often an order of magnitude. More importantly,
eb45ef3b
JD
540 because canonical LR(1)'s extra parser states may contain duplicate
541 conflicts in the case of non-LR(1) grammars, the number of conflicts
542 for IELR(1) is often an order of magnitude less as well. This can
543 significantly reduce the complexity of developing of a grammar.
544
545 Bison can now generate IELR(1) and canonical LR(1) parser tables in
546 place of its traditional LALR(1) parser tables, which remain the
547 default. You can specify the type of parser tables in the grammar
548 file with these directives:
549
cf499cff
JD
550 %define lr.type lalr
551 %define lr.type ielr
552 %define lr.type canonical-lr
eb45ef3b 553
7fceb615 554 The default-reduction optimization in the parser tables can also be
e4ab1254
AD
555 adjusted using "%define lr.default-reductions". For details on both
556 of these features, see the new section "Tuning LR" in the Bison
7fceb615 557 manual.
eb45ef3b
JD
558
559 These features are experimental. More user feedback will help to
560 stabilize them.
561
7fceb615 562** LAC (Lookahead Correction) for syntax error handling:
fcf834f9
JD
563
564 Canonical LR, IELR, and LALR can suffer from a couple of problems
565 upon encountering a syntax error. First, the parser might perform
566 additional parser stack reductions before discovering the syntax
7fceb615 567 error. Such reductions can perform user semantic actions that are
fcf834f9
JD
568 unexpected because they are based on an invalid token, and they
569 cause error recovery to begin in a different syntactic context than
570 the one in which the invalid token was encountered. Second, when
7fceb615 571 verbose error messages are enabled (with %error-verbose or the
e4ab1254 572 obsolete "#define YYERROR_VERBOSE"), the expected token list in the
7fceb615
JD
573 syntax error message can both contain invalid tokens and omit valid
574 tokens.
fcf834f9
JD
575
576 The culprits for the above problems are %nonassoc, default
577 reductions in inconsistent states, and parser state merging. Thus,
578 IELR and LALR suffer the most. Canonical LR can suffer only if
579 %nonassoc is used or if default reductions are enabled for
580 inconsistent states.
581
7fceb615
JD
582 LAC is a new mechanism within the parsing algorithm that solves
583 these problems for canonical LR, IELR, and LALR without sacrificing
584 %nonassoc, default reductions, or state merging. When LAC is in
585 use, canonical LR and IELR behave almost exactly the same for both
586 syntactically acceptable and syntactically unacceptable input.
fcf834f9
JD
587 While LALR still does not support the full language-recognition
588 power of canonical LR and IELR, LAC at least enables LALR's syntax
589 error handling to correctly reflect LALR's language-recognition
590 power.
591
592 Currently, LAC is only supported for deterministic parsers in C.
593 You can enable LAC with the following directive:
594
595 %define parse.lac full
596
e4ab1254 597 See the new section "LAC" in the Bison manual for additional
7fceb615 598 details including a few caveats.
fcf834f9
JD
599
600 LAC is an experimental feature. More user feedback will help to
601 stabilize it.
602
d397d9f0 603** %define improvements:
cf499cff 604
f1b238df 605*** Can now be invoked via the command line:
50cca368 606
de5ab940 607 Each of these command-line options
50cca368 608
de5ab940
JD
609 -D NAME[=VALUE]
610 --define=NAME[=VALUE]
611
612 -F NAME[=VALUE]
613 --force-define=NAME[=VALUE]
50cca368
JD
614
615 is equivalent to this grammar file declaration
616
de5ab940 617 %define NAME ["VALUE"]
50cca368 618
de5ab940
JD
619 except that the manner in which Bison processes multiple definitions
620 for the same NAME differs. Most importantly, -F and --force-define
621 quietly override %define, but -D and --define do not. For further
e4ab1254 622 details, see the section "Bison Options" in the Bison manual.
50cca368 623
f1b238df 624*** Variables renamed:
67212941
JD
625
626 The following %define variables
627
628 api.push_pull
629 lr.keep_unreachable_states
630
631 have been renamed to
632
633 api.push-pull
634 lr.keep-unreachable-states
635
636 The old names are now deprecated but will be maintained indefinitely
637 for backward compatibility.
638
7262f54f 639*** Values no longer need to be quoted in the grammar file:
cf499cff
JD
640
641 If a %define value is an identifier, it no longer needs to be placed
642 within quotations marks. For example,
643
644 %define api.push-pull "push"
645
646 can be rewritten as
647
648 %define api.push-pull push
649
d397d9f0 650*** Unrecognized variables are now errors not warnings.
cdf3f113 651
d397d9f0
JD
652*** Multiple invocations for any variable is now an error not a warning.
653
654** Unrecognized %code qualifiers are now errors not warnings.
655
656** Character literals not of length one:
657
658 Previously, Bison quietly converted all character literals to length
659 one. For example, without warning, Bison interpreted the operators in
660 the following grammar to be the same token:
661
662 exp: exp '++'
663 | exp '+' exp
664 ;
665
666 Bison now warns when a character literal is not of length one. In
667 some future release, Bison will start reporting an error instead.
668
669** Destructor calls fixed for lookaheads altered in semantic actions:
670
671 Previously for deterministic parsers in C, if a user semantic action
672 altered yychar, the parser in some cases used the old yychar value to
673 determine which destructor to call for the lookahead upon a syntax
674 error or upon parser return. This bug has been fixed.
675
676** C++ parsers use YYRHSLOC:
677
678 Similarly to the C parsers, the C++ parsers now define the YYRHSLOC
679 macro and use it in the default YYLLOC_DEFAULT. You are encouraged
e4ab1254
AD
680 to use it. If, for instance, your location structure has "first"
681 and "last" members, instead of
d397d9f0 682
e29f0771
AD
683 # define YYLLOC_DEFAULT(Current, Rhs, N) \
684 do \
685 if (N) \
686 { \
687 (Current).first = (Rhs)[1].location.first; \
688 (Current).last = (Rhs)[N].location.last; \
689 } \
690 else \
691 { \
692 (Current).first = (Current).last = (Rhs)[0].location.last; \
693 } \
694 while (false)
d397d9f0
JD
695
696 use:
697
e29f0771
AD
698 # define YYLLOC_DEFAULT(Current, Rhs, N) \
699 do \
700 if (N) \
701 { \
702 (Current).first = YYRHSLOC (Rhs, 1).first; \
703 (Current).last = YYRHSLOC (Rhs, N).last; \
704 } \
705 else \
706 { \
707 (Current).first = (Current).last = YYRHSLOC (Rhs, 0).last; \
708 } \
709 while (false)
d397d9f0
JD
710
711** YYLLOC_DEFAULT in C++:
712
713 The default implementation of YYLLOC_DEFAULT used to be issued in
714 the header file. It is now output in the implementation file, after
715 the user %code sections so that its #ifndef guard does not try to
716 override the user's YYLLOC_DEFAULT if provided.
cdf3f113 717
f1b238df 718** YYFAIL now produces warnings and Java parsers no longer implement it:
4395a9ff
JD
719
720 YYFAIL has existed for many years as an undocumented feature of
721 deterministic parsers in C generated by Bison. More recently, it was
722 a documented feature of Bison's experimental Java parsers. As
723 promised in Bison 2.4.2's NEWS entry, any appearance of YYFAIL in a
724 semantic action now produces a deprecation warning, and Java parsers
725 no longer implement YYFAIL at all. For further details, including a
726 discussion of how to suppress C preprocessor warnings about YYFAIL
727 being unused, see the Bison 2.4.2 NEWS entry.
728
f1b238df 729** Temporary hack for adding a semicolon to the user action:
197b82ba
JD
730
731 Previously, Bison appended a semicolon to every user action for
732 reductions when the output language defaulted to C (specifically, when
733 neither %yacc, %language, %skeleton, or equivalent command-line
734 options were specified). This allowed actions such as
735
736 exp: exp "+" exp { $$ = $1 + $3 };
737
738 instead of
739
740 exp: exp "+" exp { $$ = $1 + $3; };
741
742 As a first step in removing this misfeature, Bison now issues a
743 warning when it appends a semicolon. Moreover, in cases where Bison
744 cannot easily determine whether a semicolon is needed (for example, an
745 action ending with a cpp directive or a braced compound initializer),
746 it no longer appends one. Thus, the C compiler might now complain
747 about a missing semicolon where it did not before. Future releases of
748 Bison will cease to append semicolons entirely.
749
d2060f06
JD
750** Verbose syntax error message fixes:
751
e4ab1254 752 When %error-verbose or the obsolete "#define YYERROR_VERBOSE" is
7fceb615
JD
753 specified, syntax error messages produced by the generated parser
754 include the unexpected token as well as a list of expected tokens.
755 The effect of %nonassoc on these verbose messages has been corrected
756 in two ways, but a more complete fix requires LAC, described above:
d2060f06
JD
757
758*** When %nonassoc is used, there can exist parser states that accept no
759 tokens, and so the parser does not always require a lookahead token
760 in order to detect a syntax error. Because no unexpected token or
761 expected tokens can then be reported, the verbose syntax error
762 message described above is suppressed, and the parser instead
e4ab1254 763 reports the simpler message, "syntax error". Previously, this
d2060f06
JD
764 suppression was sometimes erroneously triggered by %nonassoc when a
765 lookahead was actually required. Now verbose messages are
766 suppressed only when all previous lookaheads have already been
767 shifted or discarded.
768
769*** Previously, the list of expected tokens erroneously included tokens
770 that would actually induce a syntax error because conflicts for them
771 were resolved with %nonassoc in the current parser state. Such
772 tokens are now properly omitted from the list.
773
774*** Expected token lists are still often wrong due to state merging
fcf834f9
JD
775 (from LALR or IELR) and default reductions, which can both add
776 invalid tokens and subtract valid tokens. Canonical LR almost
777 completely fixes this problem by eliminating state merging and
778 default reductions. However, there is one minor problem left even
779 when using canonical LR and even after the fixes above. That is,
780 if the resolution of a conflict with %nonassoc appears in a later
781 parser state than the one at which some syntax error is
782 discovered, the conflicted token is still erroneously included in
783 the expected token list. Bison's new LAC implementation,
784 described above, eliminates this problem and the need for
785 canonical LR. However, LAC is still experimental and is disabled
786 by default.
53f036ce 787
1a33f4f6
JD
788** Java skeleton fixes:
789
790*** A location handling bug has been fixed.
791
792*** The top element of each of the value stack and location stack is now
793 cleared when popped so that it can be garbage collected.
6771a463 794
02803d55
JD
795*** Parser traces now print the top element of the stack.
796
86408959
JD
797** -W/--warnings fixes:
798
e4ab1254 799*** Bison now properly recognizes the "no-" versions of categories:
86408959
JD
800
801 For example, given the following command line, Bison now enables all
802 warnings except warnings for incompatibilities with POSIX Yacc:
803
804 bison -Wall,no-yacc gram.y
805
786743d5
JD
806*** Bison now treats S/R and R/R conflicts like other warnings:
807
808 Previously, conflict reports were independent of Bison's normal
809 warning system. Now, Bison recognizes the warning categories
e4ab1254 810 "conflicts-sr" and "conflicts-rr". This change has important
786743d5
JD
811 consequences for the -W and --warnings command-line options. For
812 example:
813
814 bison -Wno-conflicts-sr gram.y # S/R conflicts not reported
815 bison -Wno-conflicts-rr gram.y # R/R conflicts not reported
816 bison -Wnone gram.y # no conflicts are reported
817 bison -Werror gram.y # any conflict is an error
818
819 However, as before, if the %expect or %expect-rr directive is
820 specified, an unexpected number of conflicts is an error, and an
821 expected number of conflicts is not reported, so -W and --warning
822 then have no effect on the conflict report.
823
e4ab1254 824*** The "none" category no longer disables a preceding "error":
bf0e44e8
JD
825
826 For example, for the following command line, Bison now reports
827 errors instead of warnings for incompatibilities with POSIX Yacc:
828
829 bison -Werror,none,yacc gram.y
830
e4ab1254 831*** The "none" category now disables all Bison warnings:
c39014ae 832
e4ab1254 833 Previously, the "none" category disabled only Bison warnings for
c39014ae
JD
834 which there existed a specific -W/--warning category. However,
835 given the following command line, Bison is now guaranteed to
836 suppress all warnings:
837
838 bison -Wnone gram.y
839
1f36f544
JD
840** Precedence directives can now assign token number 0:
841
842 Since Bison 2.3b, which restored the ability of precedence
843 directives to assign token numbers, doing so for token number 0 has
844 produced an assertion failure. For example:
845
846 %left END 0
847
848 This bug has been fixed.
849
64877e5e 850* Changes in version 2.4.3 (2010-08-05):
8b9e021f 851
2bfcac9a
JD
852** Bison now obeys -Werror and --warnings=error for warnings about
853 grammar rules that are useless in the parser due to conflicts.
854
8b9e021f
JD
855** Problems with spawning M4 on at least FreeBSD 8 and FreeBSD 9 have
856 been fixed.
857
4ad3921d
JD
858** Failures in the test suite for GCC 4.5 have been fixed.
859
06cb07d5
JD
860** Failures in the test suite for some versions of Sun Studio C++ have
861 been fixed.
862
9b5049bd
JD
863** Contrary to Bison 2.4.2's NEWS entry, it has been decided that
864 warnings about undefined %prec identifiers will not be converted to
865 errors in Bison 2.5. They will remain warnings, which should be
866 sufficient for POSIX while avoiding backward compatibility issues.
867
93d7dde9
JD
868** Minor documentation fixes.
869
e19a049c 870* Changes in version 2.4.2 (2010-03-20):
74553c98 871
f39ab286
JD
872** Some portability problems that resulted in failures and livelocks
873 in the test suite on some versions of at least Solaris, AIX, HP-UX,
e19a049c
JD
874 RHEL4, and Tru64 have been addressed. As a result, fatal Bison
875 errors should no longer cause M4 to report a broken pipe on the
f39ab286
JD
876 affected platforms.
877
e4ab1254 878** "%prec IDENTIFIER" requires IDENTIFIER to be defined separately.
8bb3a2e7
JD
879
880 POSIX specifies that an error be reported for any identifier that does
881 not appear on the LHS of a grammar rule and that is not defined by
882 %token, %left, %right, or %nonassoc. Bison 2.3b and later lost this
883 error report for the case when an identifier appears only after a
884 %prec directive. It is now restored. However, for backward
885 compatibility with recent Bison releases, it is only a warning for
886 now. In Bison 2.5 and later, it will return to being an error.
9b5049bd
JD
887 [Between the 2.4.2 and 2.4.3 releases, it was decided that this
888 warning will not be converted to an error in Bison 2.5.]
8bb3a2e7 889
d8911864
EB
890** Detection of GNU M4 1.4.6 or newer during configure is improved.
891
a603c6e0
JD
892** Warnings from gcc's -Wundef option about undefined YYENABLE_NLS,
893 YYLTYPE_IS_TRIVIAL, and __STRICT_ANSI__ in C/C++ parsers are now
894 avoided.
c938d650 895
98a345a2
JD
896** %code is now a permanent feature.
897
898 A traditional Yacc prologue directive is written in the form:
899
900 %{CODE%}
901
902 To provide a more flexible alternative, Bison 2.3b introduced the
903 %code directive with the following forms for C/C++:
904
905 %code {CODE}
906 %code requires {CODE}
907 %code provides {CODE}
908 %code top {CODE}
909
910 These forms are now considered permanent features of Bison. See the
911 %code entries in the section "Bison Declaration Summary" in the Bison
912 manual for a summary of their functionality. See the section
913 "Prologue Alternatives" for a detailed discussion including the
914 advantages of %code over the traditional Yacc prologue directive.
915
916 Bison's Java feature as a whole including its current usage of %code
917 is still considered experimental.
918
1625df5b
JD
919** YYFAIL is deprecated and will eventually be removed.
920
921 YYFAIL has existed for many years as an undocumented feature of
922 deterministic parsers in C generated by Bison. Previously, it was
923 documented for Bison's experimental Java parsers. YYFAIL is no longer
924 documented for Java parsers and is formally deprecated in both cases.
925 Users are strongly encouraged to migrate to YYERROR, which is
926 specified by POSIX.
927
928 Like YYERROR, you can invoke YYFAIL from a semantic action in order to
929 induce a syntax error. The most obvious difference from YYERROR is
930 that YYFAIL will automatically invoke yyerror to report the syntax
931 error so that you don't have to. However, there are several other
932 subtle differences between YYERROR and YYFAIL, and YYFAIL suffers from
e4ab1254 933 inherent flaws when %error-verbose or "#define YYERROR_VERBOSE" is
1625df5b
JD
934 used. For a more detailed discussion, see:
935
936 http://lists.gnu.org/archive/html/bison-patches/2009-12/msg00024.html
937
938 The upcoming Bison 2.5 will remove YYFAIL from Java parsers, but
939 deterministic parsers in C will continue to implement it. However,
940 because YYFAIL is already flawed, it seems futile to try to make new
941 Bison features compatible with it. Thus, during parser generation,
942 Bison 2.5 will produce a warning whenever it discovers YYFAIL in a
943 rule action. In a later release, YYFAIL will be disabled for
e4ab1254 944 %error-verbose and "#define YYERROR_VERBOSE". Eventually, YYFAIL will
1625df5b
JD
945 be removed altogether.
946
947 There exists at least one case where Bison 2.5's YYFAIL warning will
948 be a false positive. Some projects add phony uses of YYFAIL and other
949 Bison-defined macros for the sole purpose of suppressing C
950 preprocessor warnings (from GCC cpp's -Wunused-macros, for example).
951 To avoid Bison's future warning, such YYFAIL uses can be moved to the
e4ab1254 952 epilogue (that is, after the second "%%") in the Bison input file. In
1625df5b
JD
953 this release (2.4.2), Bison already generates its own code to suppress
954 C preprocessor warnings for YYFAIL, so projects can remove their own
955 phony uses of YYFAIL if compatibility with Bison releases prior to
956 2.4.2 is not necessary.
957
2755de8f
AD
958** Internationalization.
959
960 Fix a regression introduced in Bison 2.4: Under some circumstances,
961 message translations were not installed although supported by the
962 host system.
963
74553c98 964* Changes in version 2.4.1 (2008-12-11):
c9ba9e59 965
0ea583d2
AD
966** In the GLR defines file, unexpanded M4 macros in the yylval and yylloc
967 declarations have been fixed.
1979121c 968
0ea583d2
AD
969** Temporary hack for adding a semicolon to the user action.
970
971 Bison used to prepend a trailing semicolon at the end of the user
972 action for reductions. This allowed actions such as
973
974 exp: exp "+" exp { $$ = $1 + $3 };
975
976 instead of
977
978 exp: exp "+" exp { $$ = $1 + $3; };
979
e4ab1254 980 Some grammars still depend on this "feature". Bison 2.4.1 restores
0ea583d2
AD
981 the previous behavior in the case of C output (specifically, when
982 neither %language or %skeleton or equivalent command-line options
983 are used) to leave more time for grammars depending on the old
984 behavior to be adjusted. Future releases of Bison will disable this
985 feature.
986
987** A few minor improvements to the Bison manual.
c9ba9e59 988
402b123d 989* Changes in version 2.4 (2008-11-02):
7bd1665a 990
402b123d 991** %language is an experimental feature.
ed4d67dc
JD
992
993 We first introduced this feature in test release 2.3b as a cleaner
994 alternative to %skeleton. Since then, we have discussed the possibility of
995 modifying its effect on Bison's output file names. Thus, in this release,
996 we consider %language to be an experimental feature that will likely evolve
997 in future releases.
7bd1665a 998
402b123d 999** Forward compatibility with GNU M4 has been improved.
241fda7a 1000
402b123d 1001** Several bugs in the C++ skeleton and the experimental Java skeleton have been
241fda7a
JD
1002 fixed.
1003
402b123d 1004* Changes in version 2.3b (2008-05-27):
35fe0834 1005
402b123d 1006** The quotes around NAME that used to be required in the following directive
d9df47b6
JD
1007 are now deprecated:
1008
1009 %define NAME "VALUE"
1010
e4ab1254 1011** The directive "%pure-parser" is now deprecated in favor of:
d9df47b6
JD
1012
1013 %define api.pure
1014
1015 which has the same effect except that Bison is more careful to warn about
1016 unreasonable usage in the latter case.
1017
402b123d 1018** Push Parsing
c373bf8b
JD
1019
1020 Bison can now generate an LALR(1) parser in C with a push interface. That
e4ab1254
AD
1021 is, instead of invoking "yyparse", which pulls tokens from "yylex", you can
1022 push one token at a time to the parser using "yypush_parse", which will
c373bf8b
JD
1023 return to the caller after processing each token. By default, the push
1024 interface is disabled. Either of the following directives will enable it:
1025
1026 %define api.push_pull "push" // Just push; does not require yylex.
1027 %define api.push_pull "both" // Push and pull; requires yylex.
1028
e4ab1254 1029 See the new section "A Push Parser" in the Bison manual for details.
c373bf8b 1030
59da312b
JD
1031 The current push parsing interface is experimental and may evolve. More user
1032 feedback will help to stabilize it.
1033
402b123d 1034** The -g and --graph options now output graphs in Graphviz DOT format,
8e55b3aa
JD
1035 not VCG format. Like --graph, -g now also takes an optional FILE argument
1036 and thus cannot be bundled with other short options.
c373bf8b 1037
402b123d 1038** Java
59da312b
JD
1039
1040 Bison can now generate an LALR(1) parser in Java. The skeleton is
e4ab1254 1041 "data/lalr1.java". Consider using the new %language directive instead of
59da312b
JD
1042 %skeleton to select it.
1043
e4ab1254 1044 See the new section "Java Parsers" in the Bison manual for details.
59da312b
JD
1045
1046 The current Java interface is experimental and may evolve. More user
1047 feedback will help to stabilize it.
1048
402b123d 1049** %language
59da312b
JD
1050
1051 This new directive specifies the programming language of the generated
d43f77e7
PB
1052 parser, which can be C (the default), C++, or Java. Besides the skeleton
1053 that Bison uses, the directive affects the names of the generated files if
1054 the grammar file's name ends in ".y".
59da312b 1055
402b123d 1056** XML Automaton Report
59da312b
JD
1057
1058 Bison can now generate an XML report of the LALR(1) automaton using the new
e4ab1254 1059 "--xml" option. The current XML schema is experimental and may evolve. More
59da312b 1060 user feedback will help to stabilize it.
c373bf8b 1061
402b123d 1062** The grammar file may now specify the name of the parser header file using
c373bf8b
JD
1063 %defines. For example:
1064
1065 %defines "parser.h"
1066
402b123d 1067** When reporting useless rules, useless nonterminals, and unused terminals,
d80fb37a
JD
1068 Bison now employs the terms "useless in grammar" instead of "useless",
1069 "useless in parser" instead of "never reduced", and "unused in grammar"
1070 instead of "unused".
cff03fb2 1071
402b123d 1072** Unreachable State Removal
c373bf8b
JD
1073
1074 Previously, Bison sometimes generated parser tables containing unreachable
31984206
JD
1075 states. A state can become unreachable during conflict resolution if Bison
1076 disables a shift action leading to it from a predecessor state. Bison now:
75ad86ee
JD
1077
1078 1. Removes unreachable states.
1079
1080 2. Does not report any conflicts that appeared in unreachable states.
1081 WARNING: As a result, you may need to update %expect and %expect-rr
1082 directives in existing grammar files.
1083
1084 3. For any rule used only in such states, Bison now reports the rule as
cff03fb2 1085 "useless in parser due to conflicts".
75ad86ee 1086
31984206
JD
1087 This feature can be disabled with the following directive:
1088
1089 %define lr.keep_unreachable_states
1090
e4ab1254 1091 See the %define entry in the "Bison Declaration Summary" in the Bison manual
31984206
JD
1092 for further discussion.
1093
e4ab1254 1094** Lookahead Set Correction in the ".output" Report
b1cc23c4 1095
e4ab1254
AD
1096 When instructed to generate a ".output" file including lookahead sets
1097 (using "--report=lookahead", for example), Bison now prints each reduction's
88c78747
JD
1098 lookahead set only next to the associated state's one item that (1) is
1099 associated with the same rule as the reduction and (2) has its dot at the end
1100 of its RHS. Previously, Bison also erroneously printed the lookahead set
1101 next to all of the state's other items associated with the same rule. This
e4ab1254 1102 bug affected only the ".output" file and not the generated parser source
88c78747
JD
1103 code.
1104
e4ab1254 1105** --report-file=FILE is a new option to override the default ".output" file
59da312b 1106 name.
1bb2bd75 1107
e4ab1254 1108** The "=" that used to be required in the following directives is now
02975b9a
JD
1109 deprecated:
1110
1111 %file-prefix "parser"
1112 %name-prefix "c_"
1113 %output "parser.c"
1114
e4ab1254 1115** An Alternative to "%{...%}" -- "%code QUALIFIER {CODE}"
c373bf8b
JD
1116
1117 Bison 2.3a provided a new set of directives as a more flexible alternative to
8e0a5e9e
JD
1118 the traditional Yacc prologue blocks. Those have now been consolidated into
1119 a single %code directive with an optional qualifier field, which identifies
1120 the purpose of the code and thus the location(s) where Bison should generate
1121 it:
1122
e4ab1254
AD
1123 1. "%code {CODE}" replaces "%after-header {CODE}"
1124 2. "%code requires {CODE}" replaces "%start-header {CODE}"
1125 3. "%code provides {CODE}" replaces "%end-header {CODE}"
1126 4. "%code top {CODE}" replaces "%before-header {CODE}"
8e0a5e9e 1127
e4ab1254
AD
1128 See the %code entries in section "Bison Declaration Summary" in the Bison
1129 manual for a summary of the new functionality. See the new section "Prologue
1130 Alternatives" for a detailed discussion including the advantages of %code
8e0a5e9e
JD
1131 over the traditional Yacc prologues.
1132
1133 The prologue alternatives are experimental. More user feedback will help to
1134 determine whether they should become permanent features.
1135
402b123d 1136** Revised warning: unset or unused mid-rule values
17bd8a73
JD
1137
1138 Since Bison 2.2, Bison has warned about mid-rule values that are set but not
1139 used within any of the actions of the parent rule. For example, Bison warns
1140 about unused $2 in:
1141
1142 exp: '1' { $$ = 1; } '+' exp { $$ = $1 + $4; };
1143
1144 Now, Bison also warns about mid-rule values that are used but not set. For
1145 example, Bison warns about unset $$ in the mid-rule action in:
1146
1147 exp: '1' { $1 = 1; } '+' exp { $$ = $2 + $4; };
1148
1149 However, Bison now disables both of these warnings by default since they
1150 sometimes prove to be false alarms in existing grammars employing the Yacc
1151 constructs $0 or $-N (where N is some positive integer).
1152
e4ab1254
AD
1153 To enable these warnings, specify the option "--warnings=midrule-values" or
1154 "-W", which is a synonym for "--warnings=all".
17bd8a73 1155
e4ab1254 1156** Default %destructor or %printer with "<*>" or "<>"
c373bf8b
JD
1157
1158 Bison now recognizes two separate kinds of default %destructor's and
12e35840
JD
1159 %printer's:
1160
e4ab1254 1161 1. Place "<*>" in a %destructor/%printer symbol list to define a default
12e35840
JD
1162 %destructor/%printer for all grammar symbols for which you have formally
1163 declared semantic type tags.
1164
e4ab1254 1165 2. Place "<>" in a %destructor/%printer symbol list to define a default
12e35840
JD
1166 %destructor/%printer for all grammar symbols without declared semantic
1167 type tags.
1168
e4ab1254
AD
1169 Bison no longer supports the "%symbol-default" notation from Bison 2.3a.
1170 "<*>" and "<>" combined achieve the same effect with one exception: Bison no
12e35840
JD
1171 longer applies any %destructor to a mid-rule value if that mid-rule value is
1172 not actually ever referenced using either $$ or $n in a semantic action.
1173
85894313
JD
1174 The default %destructor's and %printer's are experimental. More user
1175 feedback will help to determine whether they should become permanent
1176 features.
1177
e4ab1254 1178 See the section "Freeing Discarded Symbols" in the Bison manual for further
12e35840
JD
1179 details.
1180
402b123d 1181** %left, %right, and %nonassoc can now declare token numbers. This is required
e4ab1254 1182 by POSIX. However, see the end of section "Operator Precedence" in the Bison
ab7f29f8
JD
1183 manual for a caveat concerning the treatment of literal strings.
1184
402b123d 1185** The nonfunctional --no-parser, -n, and %no-parser options have been
b1cc23c4
JD
1186 completely removed from Bison.
1187
402b123d 1188* Changes in version 2.3a, 2006-09-13:
742e4900 1189
402b123d 1190** Instead of %union, you can define and use your own union type
ddc8ede1
PE
1191 YYSTYPE if your grammar contains at least one <type> tag.
1192 Your YYSTYPE need not be a macro; it can be a typedef.
1193 This change is for compatibility with other Yacc implementations,
1194 and is required by POSIX.
1195
402b123d 1196** Locations columns and lines start at 1.
cd48d21d
AD
1197 In accordance with the GNU Coding Standards and Emacs.
1198
402b123d 1199** You may now declare per-type and default %destructor's and %printer's:
ec5479ce
JD
1200
1201 For example:
1202
b2a0b7ca
JD
1203 %union { char *string; }
1204 %token <string> STRING1
1205 %token <string> STRING2
1206 %type <string> string1
1207 %type <string> string2
1208 %union { char character; }
1209 %token <character> CHR
1210 %type <character> chr
1211 %destructor { free ($$); } %symbol-default
1212 %destructor { free ($$); printf ("%d", @$.first_line); } STRING1 string1
1213 %destructor { } <character>
1214
1215 guarantees that, when the parser discards any user-defined symbol that has a
e4ab1254
AD
1216 semantic type tag other than "<character>", it passes its semantic value to
1217 "free". However, when the parser discards a "STRING1" or a "string1", it
1218 also prints its line number to "stdout". It performs only the second
1219 "%destructor" in this case, so it invokes "free" only once.
ec5479ce 1220
85894313
JD
1221 [Although we failed to mention this here in the 2.3a release, the default
1222 %destructor's and %printer's were experimental, and they were rewritten in
1223 future versions.]
1224
e4ab1254
AD
1225** Except for LALR(1) parsers in C with POSIX Yacc emulation enabled (with "-y",
1226 "--yacc", or "%yacc"), Bison no longer generates #define statements for
b931235e
JD
1227 associating token numbers with token names. Removing the #define statements
1228 helps to sanitize the global namespace during preprocessing, but POSIX Yacc
1229 requires them. Bison still generates an enum for token names in all cases.
1230
402b123d 1231** Handling of traditional Yacc prologue blocks is now more consistent but
34f98f46 1232 potentially incompatible with previous releases of Bison.
9bc0dd67
JD
1233
1234 As before, you declare prologue blocks in your grammar file with the
e4ab1254 1235 "%{ ... %}" syntax. To generate the pre-prologue, Bison concatenates all
34f98f46
JD
1236 prologue blocks that you've declared before the first %union. To generate
1237 the post-prologue, Bison concatenates all prologue blocks that you've
ddc8ede1 1238 declared after the first %union.
9bc0dd67 1239
34f98f46 1240 Previous releases of Bison inserted the pre-prologue into both the header
9bc0dd67
JD
1241 file and the code file in all cases except for LALR(1) parsers in C. In the
1242 latter case, Bison inserted it only into the code file. For parsers in C++,
1243 the point of insertion was before any token definitions (which associate
1244 token numbers with names). For parsers in C, the point of insertion was
1245 after the token definitions.
1246
1247 Now, Bison never inserts the pre-prologue into the header file. In the code
1248 file, it always inserts it before the token definitions.
1249
402b123d 1250** Bison now provides a more flexible alternative to the traditional Yacc
34f98f46
JD
1251 prologue blocks: %before-header, %start-header, %end-header, and
1252 %after-header.
1253
1254 For example, the following declaration order in the grammar file reflects the
1255 order in which Bison will output these code blocks. However, you are free to
1256 declare these code blocks in your grammar file in whatever order is most
1257 convenient for you:
1258
1259 %before-header {
1260 /* Bison treats this block like a pre-prologue block: it inserts it into
1261 * the code file before the contents of the header file. It does *not*
1262 * insert it into the header file. This is a good place to put
1263 * #include's that you want at the top of your code file. A common
e4ab1254 1264 * example is '#include "system.h"'. */
34f98f46
JD
1265 }
1266 %start-header {
1267 /* Bison inserts this block into both the header file and the code file.
1268 * In both files, the point of insertion is before any Bison-generated
1269 * token, semantic type, location type, and class definitions. This is a
1270 * good place to define %union dependencies, for example. */
9bc0dd67
JD
1271 }
1272 %union {
34f98f46
JD
1273 /* Unlike the traditional Yacc prologue blocks, the output order for the
1274 * new %*-header blocks is not affected by their declaration position
1275 * relative to any %union in the grammar file. */
9bc0dd67 1276 }
34f98f46
JD
1277 %end-header {
1278 /* Bison inserts this block into both the header file and the code file.
1279 * In both files, the point of insertion is after the Bison-generated
1280 * definitions. This is a good place to declare or define public
1281 * functions or data structures that depend on the Bison-generated
1282 * definitions. */
9bc0dd67 1283 }
34f98f46
JD
1284 %after-header {
1285 /* Bison treats this block like a post-prologue block: it inserts it into
1286 * the code file after the contents of the header file. It does *not*
1287 * insert it into the header file. This is a good place to declare or
1288 * define internal functions or data structures that depend on the
1289 * Bison-generated definitions. */
1290 }
1291
1292 If you have multiple occurrences of any one of the above declarations, Bison
1293 will concatenate the contents in declaration order.
9bc0dd67 1294
85894313
JD
1295 [Although we failed to mention this here in the 2.3a release, the prologue
1296 alternatives were experimental, and they were rewritten in future versions.]
1297
e4ab1254 1298** The option "--report=look-ahead" has been changed to "--report=lookahead".
9e6e7ed2
PE
1299 The old spelling still works, but is not documented and may be removed
1300 in a future release.
742e4900 1301
402b123d 1302* Changes in version 2.3, 2006-06-05:
4ad3ed84 1303
e4ab1254 1304** GLR grammars should now use "YYRECOVERING ()" instead of "YYRECOVERING",
4ad3ed84
PE
1305 for compatibility with LALR(1) grammars.
1306
402b123d 1307** It is now documented that any definition of YYSTYPE or YYLTYPE should
4ad3ed84
PE
1308 be to a type name that does not contain parentheses or brackets.
1309
402b123d 1310* Changes in version 2.2, 2006-05-19:
193d7c70 1311
402b123d 1312** The distribution terms for all Bison-generated parsers now permit
193d7c70
PE
1313 using the parsers in nonfree programs. Previously, this permission
1314 was granted only for Bison-generated LALR(1) parsers in C.
5f4236a0 1315
402b123d 1316** %name-prefix changes the namespace name in C++ outputs.
aa08666d 1317
402b123d 1318** The C++ parsers export their token_type.
5f4236a0 1319
402b123d 1320** Bison now allows multiple %union declarations, and concatenates
d6ca7905
PE
1321 their contents together.
1322
402b123d 1323** New warning: unused values
4d7bc38c
PE
1324 Right-hand side symbols whose values are not used are reported,
1325 if the symbols have destructors. For instance:
affac613 1326
8f3596a6 1327 exp: exp "?" exp ":" exp { $1 ? $1 : $3; }
e9690142
JD
1328 | exp "+" exp
1329 ;
affac613 1330
8f3596a6
AD
1331 will trigger a warning about $$ and $5 in the first rule, and $3 in
1332 the second ($1 is copied to $$ by the default rule). This example
4e26c69e 1333 most likely contains three errors, and could be rewritten as:
affac613 1334
4e26c69e 1335 exp: exp "?" exp ":" exp
e9690142
JD
1336 { $$ = $1 ? $3 : $5; free ($1 ? $5 : $3); free ($1); }
1337 | exp "+" exp
1338 { $$ = $1 ? $1 : $3; if ($1) free ($3); }
1339 ;
affac613 1340
4e26c69e
PE
1341 However, if the original actions were really intended, memory leaks
1342 and all, the warnings can be suppressed by letting Bison believe the
1343 values are used, e.g.:
721be13c 1344
8f3596a6 1345 exp: exp "?" exp ":" exp { $1 ? $1 : $3; (void) ($$, $5); }
e9690142
JD
1346 | exp "+" exp { $$ = $1; (void) $3; }
1347 ;
721be13c 1348
84866159
AD
1349 If there are mid-rule actions, the warning is issued if no action
1350 uses it. The following triggers no warning: $1 and $3 are used.
1351
1352 exp: exp { push ($1); } '+' exp { push ($3); sum (); };
1353
721be13c
PE
1354 The warning is intended to help catching lost values and memory leaks.
1355 If a value is ignored, its associated memory typically is not reclaimed.
affac613 1356
402b123d 1357** %destructor vs. YYABORT, YYACCEPT, and YYERROR.
9d9b8b70
PE
1358 Destructors are now called when user code invokes YYABORT, YYACCEPT,
1359 and YYERROR, for all objects on the stack, other than objects
1360 corresponding to the right-hand side of the current rule.
a85284cf 1361
402b123d 1362** %expect, %expect-rr
035aa4a0
PE
1363 Incorrect numbers of expected conflicts are now actual errors,
1364 instead of warnings.
1365
402b123d 1366** GLR, YACC parsers.
4e26c69e
PE
1367 The %parse-params are available in the destructors (and the
1368 experimental printers) as per the documentation.
4b367315 1369
e4ab1254 1370** Bison now warns if it finds a stray "$" or "@" in an action.
ad6a9b97 1371
402b123d 1372** %require "VERSION"
4e26c69e
PE
1373 This specifies that the grammar file depends on features implemented
1374 in Bison version VERSION or higher.
b50d2359 1375
402b123d 1376** lalr1.cc: The token and value types are now class members.
e14d0ab6
AD
1377 The tokens were defined as free form enums and cpp macros. YYSTYPE
1378 was defined as a free form union. They are now class members:
e4ab1254
AD
1379 tokens are enumerations of the "yy::parser::token" struct, and the
1380 semantic values have the "yy::parser::semantic_type" type.
fb9712a9
AD
1381
1382 If you do not want or can update to this scheme, the directive
e4ab1254 1383 '%define "global_tokens_and_yystype" "1"' triggers the global
b50d2359
AD
1384 definition of tokens and YYSTYPE. This change is suitable both
1385 for previous releases of Bison, and this one.
fb9712a9 1386
b50d2359 1387 If you wish to update, then make sure older version of Bison will
e4ab1254 1388 fail using '%require "2.2"'.
fb9712a9 1389
402b123d 1390** DJGPP support added.
193d7c70 1391\f
402b123d 1392* Changes in version 2.1, 2005-09-16:
1ce59070 1393
402b123d 1394** The C++ lalr1.cc skeleton supports %lex-param.
e14d0ab6 1395
402b123d 1396** Bison-generated parsers now support the translation of diagnostics like
baf785db
PE
1397 "syntax error" into languages other than English. The default
1398 language is still English. For details, please see the new
0410a6e0
PE
1399 Internationalization section of the Bison manual. Software
1400 distributors should also see the new PACKAGING file. Thanks to
1401 Bruno Haible for this new feature.
1ce59070 1402
402b123d 1403** Wording in the Bison-generated parsers has been changed slightly to
1a059451
PE
1404 simplify translation. In particular, the message "memory exhausted"
1405 has replaced "parser stack overflow", as the old message was not
1406 always accurate for modern Bison-generated parsers.
1407
402b123d 1408** Destructors are now called when the parser aborts, for all symbols left
258b75ca
PE
1409 behind on the stack. Also, the start symbol is now destroyed after a
1410 successful parse. In both cases, the behavior was formerly inconsistent.
1411
402b123d 1412** When generating verbose diagnostics, Bison-generated parsers no longer
72f000b0
PE
1413 quote the literal strings associated with tokens. For example, for
1414 a syntax error associated with '%token NUM "number"' they might
1415 print 'syntax error, unexpected number' instead of 'syntax error,
1416 unexpected "number"'.
193d7c70 1417\f
402b123d 1418* Changes in version 2.0, 2004-12-25:
efeed023 1419
402b123d 1420** Possibly-incompatible changes
d7e14fc0 1421
82de6b0d
PE
1422 - Bison-generated parsers no longer default to using the alloca function
1423 (when available) to extend the parser stack, due to widespread
1424 problems in unchecked stack-overflow detection. You can "#define
1425 YYSTACK_USE_ALLOCA 1" to require the use of alloca, but please read
1426 the manual to determine safe values for YYMAXDEPTH in that case.
8dd162d3 1427
82de6b0d
PE
1428 - Error token location.
1429 During error recovery, the location of the syntax error is updated
1430 to cover the whole sequence covered by the error token: it includes
1431 the shifted symbols thrown away during the first part of the error
1432 recovery, and the lookahead rejected during the second part.
18d192f0 1433
82de6b0d
PE
1434 - Semicolon changes:
1435 . Stray semicolons are no longer allowed at the start of a grammar.
1436 . Semicolons are now required after in-grammar declarations.
e342c3be 1437
82de6b0d
PE
1438 - Unescaped newlines are no longer allowed in character constants or
1439 string literals. They were never portable, and GCC 3.4.0 has
1440 dropped support for them. Better diagnostics are now generated if
1441 forget a closing quote.
8dd162d3 1442
82de6b0d 1443 - NUL bytes are no longer allowed in Bison string literals, unfortunately.
f74b6f91 1444
402b123d 1445** New features
1452af69 1446
82de6b0d 1447 - GLR grammars now support locations.
4febdd96 1448
82de6b0d
PE
1449 - New directive: %initial-action.
1450 This directive allows the user to run arbitrary code (including
1451 initializing @$) from yyparse before parsing starts.
1452af69 1452
82de6b0d
PE
1453 - A new directive "%expect-rr N" specifies the expected number of
1454 reduce/reduce conflicts in GLR parsers.
1452af69 1455
e4ab1254 1456 - %token numbers can now be hexadecimal integers, e.g., "%token FOO 0x12d".
82de6b0d 1457 This is a GNU extension.
4febdd96 1458
e4ab1254 1459 - The option "--report=lookahead" was changed to "--report=look-ahead".
9e6e7ed2 1460 [However, this was changed back after 2.3.]
1452af69 1461
82de6b0d 1462 - Experimental %destructor support has been added to lalr1.cc.
1452af69 1463
82de6b0d
PE
1464 - New configure option --disable-yacc, to disable installation of the
1465 yacc command and -ly library introduced in 1.875 for POSIX conformance.
6040d338 1466
402b123d 1467** Bug fixes
d5a3fe37 1468
82de6b0d
PE
1469 - For now, %expect-count violations are now just warnings, not errors.
1470 This is for compatibility with Bison 1.75 and earlier (when there are
1471 reduce/reduce conflicts) and with Bison 1.30 and earlier (when there
1472 are too many or too few shift/reduce conflicts). However, in future
1473 versions of Bison we plan to improve the %expect machinery so that
1474 these violations will become errors again.
3473d0f8 1475
82de6b0d
PE
1476 - Within Bison itself, numbers (e.g., goto numbers) are no longer
1477 arbitrarily limited to 16-bit counts.
d600ee67 1478
82de6b0d 1479 - Semicolons are now allowed before "|" in grammar rules, as POSIX requires.
d600ee67 1480\f
402b123d 1481* Changes in version 1.875, 2003-01-01:
963fcc17 1482
402b123d 1483** The documentation license has been upgraded to version 1.2
dc546b0f 1484 of the GNU Free Documentation License.
75eb3bc4 1485
402b123d 1486** syntax error processing
75eb3bc4 1487
dc546b0f
PE
1488 - In Yacc-style parsers YYLLOC_DEFAULT is now used to compute error
1489 locations too. This fixes bugs in error-location computation.
75eb3bc4 1490
dc546b0f
PE
1491 - %destructor
1492 It is now possible to reclaim the memory associated to symbols
1493 discarded during error recovery. This feature is still experimental.
20daca06 1494
dc546b0f
PE
1495 - %error-verbose
1496 This new directive is preferred over YYERROR_VERBOSE.
74724a70 1497
dc546b0f
PE
1498 - #defining yyerror to steal internal variables is discouraged.
1499 It is not guaranteed to work forever.
d1de5372 1500
402b123d 1501** POSIX conformance
d1de5372 1502
dc546b0f
PE
1503 - Semicolons are once again optional at the end of grammar rules.
1504 This reverts to the behavior of Bison 1.33 and earlier, and improves
1505 compatibility with Yacc.
74724a70 1506
e4ab1254
AD
1507 - "parse error" -> "syntax error"
1508 Bison now uniformly uses the term "syntax error"; formerly, the code
1509 and manual sometimes used the term "parse error" instead. POSIX
1510 requires "syntax error" in diagnostics, and it was thought better to
dc546b0f 1511 be consistent.
74724a70 1512
dc546b0f
PE
1513 - The documentation now emphasizes that yylex and yyerror must be
1514 declared before use. C99 requires this.
d1de5372 1515
dc546b0f
PE
1516 - Bison now parses C99 lexical constructs like UCNs and
1517 backslash-newline within C escape sequences, as POSIX 1003.1-2001 requires.
d1de5372 1518
dc546b0f
PE
1519 - File names are properly escaped in C output. E.g., foo\bar.y is
1520 output as "foo\\bar.y".
6780ca7a 1521
dc546b0f 1522 - Yacc command and library now available
e4ab1254 1523 The Bison distribution now installs a "yacc" command, as POSIX requires.
dc546b0f
PE
1524 Also, Bison now installs a small library liby.a containing
1525 implementations of Yacc-compatible yyerror and main functions.
1526 This library is normally not useful, but POSIX requires it.
6e649e65 1527
dc546b0f 1528 - Type clashes now generate warnings, not errors.
6e649e65 1529
dc546b0f
PE
1530 - If the user does not define YYSTYPE as a macro, Bison now declares it
1531 using typedef instead of defining it as a macro.
1532 For consistency, YYLTYPE is also declared instead of defined.
9501dc6e 1533
402b123d 1534** Other compatibility issues
886a425c 1535
e4ab1254
AD
1536 - %union directives can now have a tag before the "{", e.g., the
1537 directive "%union foo {...}" now generates the C code
1538 "typedef union foo { ... } YYSTYPE;"; this is for Yacc compatibility.
1539 The default union tag is "YYSTYPE", for compatibility with Solaris 9 Yacc.
1540 For consistency, YYLTYPE's struct tag is now "YYLTYPE" not "yyltype".
dc546b0f 1541 This is for compatibility with both Yacc and Bison 1.35.
72f889cc 1542
e4ab1254 1543 - ";" is output before the terminating "}" of an action, for
dc546b0f 1544 compatibility with Bison 1.35.
886a425c 1545
dc546b0f 1546 - Bison now uses a Yacc-style format for conflict reports, e.g.,
e4ab1254 1547 "conflicts: 2 shift/reduce, 1 reduce/reduce".
437c2d80 1548
e4ab1254 1549 - "yystype" and "yyltype" are now obsolescent macros instead of being
dc546b0f
PE
1550 typedefs or tags; they are no longer documented and are planned to be
1551 withdrawn in a future release.
2a8d363a 1552
402b123d 1553** GLR parser notes
2a8d363a 1554
dc546b0f
PE
1555 - GLR and inline
1556 Users of Bison have to decide how they handle the portability of the
e4ab1254 1557 C keyword "inline".
959e5f51 1558
e4ab1254
AD
1559 - "parsing stack overflow..." -> "parser stack overflow"
1560 GLR parsers now report "parser stack overflow" as per the Bison manual.
900c5db5 1561
18ad57b3
AD
1562** %parse-param and %lex-param
1563 The macros YYPARSE_PARAM and YYLEX_PARAM provide a means to pass
1564 additional context to yyparse and yylex. They suffer from several
1565 shortcomings:
1566
1567 - a single argument only can be added,
1568 - their types are weak (void *),
242cc08e 1569 - this context is not passed to ancillary functions such as yyerror,
18ad57b3
AD
1570 - only yacc.c parsers support them.
1571
1572 The new %parse-param/%lex-param directives provide a more precise control.
1573 For instance:
1574
1575 %parse-param {int *nastiness}
1576 %lex-param {int *nastiness}
1577 %parse-param {int *randomness}
1578
1579 results in the following signatures:
1580
1581 int yylex (int *nastiness);
1582 int yyparse (int *nastiness, int *randomness);
1583
1584 or, if both %pure-parser and %locations are used:
1585
1586 int yylex (YYSTYPE *lvalp, YYLTYPE *llocp, int *nastiness);
1587 int yyparse (int *nastiness, int *randomness);
1588
402b123d 1589** Bison now warns if it detects conflicting outputs to the same file,
e4ab1254 1590 e.g., it generates a warning for "bison -d -o foo.h foo.y" since
dc546b0f 1591 that command outputs both code and header to foo.h.
6e40b4eb 1592
402b123d 1593** #line in output files
dc546b0f 1594 - --no-line works properly.
6e40b4eb 1595
402b123d 1596** Bison can no longer be built by a K&R C compiler; it requires C89 or
6e40b4eb
AD
1597 later to be built. This change originally took place a few versions
1598 ago, but nobody noticed until we recently asked someone to try
1599 building Bison with a K&R C compiler.
d600ee67 1600\f
402b123d 1601* Changes in version 1.75, 2002-10-14:
7933f2b5 1602
402b123d 1603** Bison should now work on 64-bit hosts.
7933f2b5 1604
402b123d 1605** Indonesian translation thanks to Tedi Heriyanto.
7933f2b5 1606
402b123d 1607** GLR parsers
f50adbbd
AD
1608 Fix spurious parse errors.
1609
402b123d 1610** Pure parsers
f50adbbd
AD
1611 Some people redefine yyerror to steal yyparse' private variables.
1612 Reenable this trick until an official feature replaces it.
1613
402b123d 1614** Type Clashes
d90c934c
AD
1615 In agreement with POSIX and with other Yaccs, leaving a default
1616 action is valid when $$ is untyped, and $1 typed:
1617
e9690142 1618 untyped: ... typed;
d90c934c
AD
1619
1620 but the converse remains an error:
1621
e9690142 1622 typed: ... untyped;
d90c934c 1623
402b123d 1624** Values of mid-rule actions
d90c934c
AD
1625 The following code:
1626
e9690142 1627 foo: { ... } { $$ = $1; } ...
d90c934c
AD
1628
1629 was incorrectly rejected: $1 is defined in the second mid-rule
1630 action, and is equal to the $$ of the first mid-rule action.
d600ee67 1631\f
402b123d 1632* Changes in version 1.50, 2002-10-04:
adc8c848 1633
402b123d 1634** GLR parsing
676385e2
PH
1635 The declaration
1636 %glr-parser
1637 causes Bison to produce a Generalized LR (GLR) parser, capable of handling
1638 almost any context-free grammar, ambiguous or not. The new declarations
e8832397 1639 %dprec and %merge on grammar rules allow parse-time resolution of
676385e2
PH
1640 ambiguities. Contributed by Paul Hilfinger.
1641
7933f2b5 1642 Unfortunately Bison 1.50 does not work properly on 64-bit hosts
420f93c8
PE
1643 like the Alpha, so please stick to 32-bit hosts for now.
1644
402b123d 1645** Output Directory
8c165d89 1646 When not in Yacc compatibility mode, when the output file was not
e4ab1254
AD
1647 specified, running "bison foo/bar.y" created "foo/bar.c". It
1648 now creates "bar.c".
8c165d89 1649
402b123d 1650** Undefined token
007a50a4 1651 The undefined token was systematically mapped to 2 which prevented
e88dbdbf 1652 the use of 2 by the user. This is no longer the case.
007a50a4 1653
402b123d 1654** Unknown token numbers
e88dbdbf 1655 If yylex returned an out of range value, yyparse could die. This is
007a50a4
AD
1656 no longer the case.
1657
402b123d 1658** Error token
e88dbdbf 1659 According to POSIX, the error token must be 256.
23c5a174
AD
1660 Bison extends this requirement by making it a preference: *if* the
1661 user specified that one of her tokens is numbered 256, then error
1662 will be mapped onto another number.
1663
402b123d 1664** Verbose error messages
e4ab1254 1665 They no longer report "..., expecting error or..." for states where
217598da
AD
1666 error recovery is possible.
1667
402b123d 1668** End token
e4ab1254 1669 Defaults to "$end" instead of "$".
217598da 1670
402b123d 1671** Error recovery now conforms to documentation and to POSIX
68cd8af3
PE
1672 When a Bison-generated parser encounters a syntax error, it now pops
1673 the stack until it finds a state that allows shifting the error
1674 token. Formerly, it popped the stack until it found a state that
1675 allowed some non-error action other than a default reduction on the
1676 error token. The new behavior has long been the documented behavior,
1677 and has long been required by POSIX. For more details, please see
337116ba
PE
1678 Paul Eggert, "Reductions during Bison error handling" (2002-05-20)
1679 <http://lists.gnu.org/archive/html/bug-bison/2002-05/msg00038.html>.
68cd8af3 1680
402b123d 1681** Traces
5504898e
AD
1682 Popped tokens and nonterminals are now reported.
1683
402b123d 1684** Larger grammars
a861a339
PE
1685 Larger grammars are now supported (larger token numbers, larger grammar
1686 size (= sum of the LHS and RHS lengths), larger LALR tables).
1687 Formerly, many of these numbers ran afoul of 16-bit limits;
1688 now these limits are 32 bits on most hosts.
355e7c1c 1689
402b123d 1690** Explicit initial rule
643a5994
AD
1691 Bison used to play hacks with the initial rule, which the user does
1692 not write. It is now explicit, and visible in the reports and
1693 graphs as rule 0.
23c5a174 1694
402b123d 1695** Useless rules
643a5994 1696 Before, Bison reported the useless rules, but, although not used,
77714df2 1697 included them in the parsers. They are now actually removed.
23c5a174 1698
402b123d 1699** Useless rules, useless nonterminals
6b98e4b5
AD
1700 They are now reported, as a warning, with their locations.
1701
402b123d 1702** Rules never reduced
e8832397
AD
1703 Rules that can never be reduced because of conflicts are now
1704 reported.
1705
e4ab1254 1706** Incorrect "Token not used"
11652ab3
AD
1707 On a grammar such as
1708
e29f0771
AD
1709 %token useless useful
1710 %%
1711 exp: '0' %prec useful;
11652ab3
AD
1712
1713 where a token was used to set the precedence of the last rule,
e4ab1254 1714 bison reported both "useful" and "useless" as useless tokens.
11652ab3 1715
402b123d 1716** Revert the C++ namespace changes introduced in 1.31
77714df2 1717 as they caused too many portability hassles.
0179dd65 1718
402b123d 1719** Default locations
b2d52318
AD
1720 By an accident of design, the default computation of @$ was
1721 performed after another default computation was performed: @$ = @1.
1722 The latter is now removed: YYLLOC_DEFAULT is fully responsible of
1723 the computation of @$.
adc8c848 1724
402b123d 1725** Token end-of-file
b7c49edf
AD
1726 The token end of file may be specified by the user, in which case,
1727 the user symbol is used in the reports, the graphs, and the verbose
e4ab1254 1728 error messages instead of "$end", which remains being the default.
b7c49edf 1729 For instance
e29f0771 1730 %token MYEOF 0
b7c49edf 1731 or
e29f0771 1732 %token MYEOF 0 "end of file"
fdbcd8e2 1733
402b123d 1734** Semantic parser
fdbcd8e2
AD
1735 This old option, which has been broken for ages, is removed.
1736
402b123d 1737** New translations
a861a339 1738 Brazilian Portuguese, thanks to Alexandre Folle de Menezes.
84614e13
AD
1739 Croatian, thanks to Denis Lackovic.
1740
402b123d 1741** Incorrect token definitions
e4ab1254
AD
1742 When given
1743 %token 'a' "A"
1744 bison used to output
1745 #define 'a' 65
b87f8b21 1746
402b123d 1747** Token definitions as enums
77714df2
AD
1748 Tokens are output both as the traditional #define's, and, provided
1749 the compiler supports ANSI C or is a C++ compiler, as enums.
e88dbdbf 1750 This lets debuggers display names instead of integers.
77714df2 1751
402b123d 1752** Reports
ec3bc396
AD
1753 In addition to --verbose, bison supports --report=THINGS, which
1754 produces additional information:
b408954b
AD
1755 - itemset
1756 complete the core item sets with their closure
e4ab1254 1757 - lookahead [changed to "look-ahead" in 1.875e through 2.3, but changed back]
9e6e7ed2 1758 explicitly associate lookahead tokens to items
b408954b
AD
1759 - solved
1760 describe shift/reduce conflicts solving.
1761 Bison used to systematically output this information on top of
1762 the report. Solved conflicts are now attached to their states.
ec3bc396 1763
402b123d 1764** Type clashes
9af3fbce
AD
1765 Previous versions don't complain when there is a type clash on
1766 the default action if the rule has a mid-rule action, such as in:
1767
e29f0771
AD
1768 %type <foo> bar
1769 %%
1770 bar: '0' {} '0';
9af3fbce
AD
1771
1772 This is fixed.
a861a339 1773
402b123d 1774** GNU M4 is now required when using Bison.
f987e9d2 1775\f
402b123d 1776* Changes in version 1.35, 2002-03-25:
76551463 1777
402b123d 1778** C Skeleton
76551463
AD
1779 Some projects use Bison's C parser with C++ compilers, and define
1780 YYSTYPE as a class. The recent adjustment of C parsers for data
1781 alignment and 64 bit architectures made this impossible.
1782
1783 Because for the time being no real solution for C++ parser
1784 generation exists, kludges were implemented in the parser to
1785 maintain this use. In the future, when Bison has C++ parsers, this
1786 kludge will be disabled.
1787
1788 This kludge also addresses some C++ problems when the stack was
1789 extended.
76551463 1790\f
402b123d 1791* Changes in version 1.34, 2002-03-12:
76551463 1792
402b123d 1793** File name clashes are detected
76551463 1794 $ bison foo.y -d -o foo.x
e4ab1254 1795 fatal error: header and parser would both be named "foo.x"
76551463 1796
e4ab1254 1797** A missing ";" at the end of a rule triggers a warning
76551463
AD
1798 In accordance with POSIX, and in agreement with other
1799 Yacc implementations, Bison will mandate this semicolon in the near
1800 future. This eases the implementation of a Bison parser of Bison
1801 grammars by making this grammar LALR(1) instead of LR(2). To
1802 facilitate the transition, this release introduces a warning.
1803
402b123d 1804** Revert the C++ namespace changes introduced in 1.31, as they caused too
76551463
AD
1805 many portability hassles.
1806
402b123d 1807** DJGPP support added.
76551463 1808
402b123d 1809** Fix test suite portability problems.
76551463 1810\f
402b123d 1811* Changes in version 1.33, 2002-02-07:
76551463 1812
402b123d 1813** Fix C++ issues
76551463
AD
1814 Groff could not be compiled for the definition of size_t was lacking
1815 under some conditions.
1816
402b123d 1817** Catch invalid @n
76551463
AD
1818 As is done with $n.
1819\f
402b123d 1820* Changes in version 1.32, 2002-01-23:
76551463 1821
402b123d 1822** Fix Yacc output file names
76551463 1823
402b123d 1824** Portability fixes
76551463 1825
402b123d 1826** Italian, Dutch translations
76551463 1827\f
402b123d 1828* Changes in version 1.31, 2002-01-14:
52d1aeee 1829
402b123d 1830** Many Bug Fixes
52d1aeee 1831
402b123d 1832** GNU Gettext and %expect
52d1aeee
MA
1833 GNU Gettext asserts 10 s/r conflicts, but there are 7. Now that
1834 Bison dies on incorrect %expectations, we fear there will be
1835 too many bug reports for Gettext, so _for the time being_, %expect
e4ab1254 1836 does not trigger an error when the input file is named "plural.y".
52d1aeee 1837
402b123d 1838** Use of alloca in parsers
52d1aeee
MA
1839 If YYSTACK_USE_ALLOCA is defined to 0, then the parsers will use
1840 malloc exclusively. Since 1.29, but was not NEWS'ed.
1841
1842 alloca is used only when compiled with GCC, to avoid portability
1843 problems as on AIX.
1844
402b123d 1845** yyparse now returns 2 if memory is exhausted; formerly it dumped core.
b47dbebe 1846
402b123d 1847** When the generated parser lacks debugging code, YYDEBUG is now 0
52d1aeee
MA
1848 (as POSIX requires) instead of being undefined.
1849
402b123d 1850** User Actions
52d1aeee
MA
1851 Bison has always permitted actions such as { $$ = $1 }: it adds the
1852 ending semicolon. Now if in Yacc compatibility mode, the semicolon
1853 is no longer output: one has to write { $$ = $1; }.
1854
402b123d 1855** Better C++ compliance
52d1aeee 1856 The output parsers try to respect C++ namespaces.
76551463 1857 [This turned out to be a failed experiment, and it was reverted later.]
52d1aeee 1858
402b123d 1859** Reduced Grammars
52d1aeee
MA
1860 Fixed bugs when reporting useless nonterminals.
1861
402b123d 1862** 64 bit hosts
52d1aeee
MA
1863 The parsers work properly on 64 bit hosts.
1864
402b123d 1865** Error messages
52d1aeee
MA
1866 Some calls to strerror resulted in scrambled or missing error messages.
1867
402b123d 1868** %expect
52d1aeee
MA
1869 When the number of shift/reduce conflicts is correct, don't issue
1870 any warning.
1871
402b123d 1872** The verbose report includes the rule line numbers.
52d1aeee 1873
402b123d 1874** Rule line numbers are fixed in traces.
52d1aeee 1875
402b123d 1876** Swedish translation
52d1aeee 1877
402b123d 1878** Parse errors
52d1aeee
MA
1879 Verbose parse error messages from the parsers are better looking.
1880 Before: parse error: unexpected `'/'', expecting `"number"' or `'-'' or `'(''
1881 Now: parse error: unexpected '/', expecting "number" or '-' or '('
1882
402b123d 1883** Fixed parser memory leaks.
52d1aeee
MA
1884 When the generated parser was using malloc to extend its stacks, the
1885 previous allocations were not freed.
1886
402b123d 1887** Fixed verbose output file.
52d1aeee
MA
1888 Some newlines were missing.
1889 Some conflicts in state descriptions were missing.
1890
402b123d 1891** Fixed conflict report.
52d1aeee
MA
1892 Option -v was needed to get the result.
1893
402b123d 1894** %expect
52d1aeee
MA
1895 Was not used.
1896 Mismatches are errors, not warnings.
1897
402b123d 1898** Fixed incorrect processing of some invalid input.
52d1aeee 1899
402b123d 1900** Fixed CPP guards: 9foo.h uses BISON_9FOO_H instead of 9FOO_H.
52d1aeee 1901
402b123d 1902** Fixed some typos in the documentation.
52d1aeee 1903
402b123d 1904** %token MY_EOF 0 is supported.
52d1aeee
MA
1905 Before, MY_EOF was silently renumbered as 257.
1906
402b123d 1907** doc/refcard.tex is updated.
52d1aeee 1908
402b123d 1909** %output, %file-prefix, %name-prefix.
52d1aeee
MA
1910 New.
1911
402b123d 1912** --output
e4ab1254 1913 New, aliasing "--output-file".
52d1aeee 1914\f
402b123d 1915* Changes in version 1.30, 2001-10-26:
342b8b6e 1916
e4ab1254
AD
1917** "--defines" and "--graph" have now an optional argument which is the
1918 output file name. "-d" and "-g" do not change; they do not take any
342b8b6e
AD
1919 argument.
1920
e4ab1254 1921** "%source_extension" and "%header_extension" are removed, failed
342b8b6e
AD
1922 experiment.
1923
402b123d 1924** Portability fixes.
f987e9d2 1925\f
402b123d 1926* Changes in version 1.29, 2001-09-07:
342b8b6e 1927
402b123d 1928** The output file does not define const, as this caused problems when used
342b8b6e
AD
1929 with common autoconfiguration schemes. If you still use ancient compilers
1930 that lack const, compile with the equivalent of the C compiler option
e4ab1254 1931 "-Dconst=". Autoconf's AC_C_CONST macro provides one way to do this.
342b8b6e 1932
e4ab1254 1933** Added "-g" and "--graph".
f87a2205 1934
402b123d 1935** The Bison manual is now distributed under the terms of the GNU FDL.
f2b5126e 1936
402b123d 1937** The input and the output files has automatically a similar extension.
234a3be3 1938
402b123d 1939** Russian translation added.
f87a2205 1940
402b123d 1941** NLS support updated; should hopefully be less troublesome.
f87a2205 1942
402b123d 1943** Added the old Bison reference card.
c33638bb 1944
e4ab1254 1945** Added "--locations" and "%locations".
6deb4447 1946
e4ab1254 1947** Added "-S" and "--skeleton".
cd5bd6ac 1948
e4ab1254 1949** "%raw", "-r", "--raw" is disabled.
62ab6972 1950
402b123d 1951** Special characters are escaped when output. This solves the problems
cd5bd6ac
AD
1952 of the #line lines with path names including backslashes.
1953
402b123d 1954** New directives.
e4ab1254
AD
1955 "%yacc", "%fixed_output_files", "%defines", "%no_parser", "%verbose",
1956 "%debug", "%source_extension" and "%header_extension".
f987e9d2 1957
402b123d 1958** @$
f987e9d2 1959 Automatic location tracking.
f87a2205 1960\f
402b123d 1961* Changes in version 1.28, 1999-07-06:
d2e00347 1962
402b123d 1963** Should compile better now with K&R compilers.
d2e00347 1964
402b123d 1965** Added NLS.
d2e00347 1966
402b123d 1967** Fixed a problem with escaping the double quote character.
d2e00347 1968
402b123d 1969** There is now a FAQ.
d2e00347 1970\f
402b123d 1971* Changes in version 1.27:
5c31c3c2 1972
402b123d 1973** The make rule which prevented bison.simple from being created on
5c31c3c2
JT
1974 some systems has been fixed.
1975\f
402b123d 1976* Changes in version 1.26:
4be07551 1977
7e508a2b 1978** Bison now uses Automake.
4be07551 1979
402b123d 1980** New mailing lists: <bug-bison@gnu.org> and <help-bison@gnu.org>.
4be07551 1981
402b123d 1982** Token numbers now start at 257 as previously documented, not 258.
4be07551 1983
402b123d 1984** Bison honors the TMPDIR environment variable.
4be07551 1985
402b123d 1986** A couple of buffer overruns have been fixed.
f51dbca1 1987
402b123d 1988** Problems when closing files should now be reported.
f51dbca1 1989
402b123d 1990** Generated parsers should now work even on operating systems which do
f51dbca1 1991 not provide alloca().
4be07551 1992\f
402b123d 1993* Changes in version 1.25, 1995-10-16:
df8878c5 1994
402b123d 1995** Errors in the input grammar are not fatal; Bison keeps reading
df8878c5 1996the grammar file, and reports all the errors found in it.
8c44d3ec 1997
402b123d 1998** Tokens can now be specified as multiple-character strings: for
df8878c5 1999example, you could use "<=" for a token which looks like <=, instead
7e508a2b 2000of choosing a name like LESSEQ.
df8878c5 2001
402b123d 2002** The %token_table declaration says to write a table of tokens (names
df8878c5
RS
2003and numbers) into the parser file. The yylex function can use this
2004table to recognize multiple-character string tokens, or for other
2005purposes.
2006
402b123d 2007** The %no_lines declaration says not to generate any #line preprocessor
df8878c5
RS
2008directives in the parser file.
2009
402b123d 2010** The %raw declaration says to use internal Bison token numbers, not
df8878c5
RS
2011Yacc-compatible token numbers, when token names are defined as macros.
2012
402b123d 2013** The --no-parser option produces the parser tables without including
df8878c5
RS
2014the parser engine; a project can now use its own parser engine.
2015The actions go into a separate file called NAME.act, in the form of
2016a switch statement body.
2017\f
402b123d 2018* Changes in version 1.23:
6780ca7a 2019
4d019228
DM
2020The user can define YYPARSE_PARAM as the name of an argument to be
2021passed into yyparse. The argument should have type void *. It should
2022actually point to an object. Grammar actions can access the variable
2023by casting it to the proper pointer type.
6780ca7a 2024
6780ca7a 2025Line numbers in output file corrected.
6780ca7a 2026\f
402b123d 2027* Changes in version 1.22:
6780ca7a
DM
2028
2029--help option added.
6780ca7a 2030\f
402b123d 2031* Changes in version 1.20:
6780ca7a
DM
2032
2033Output file does not redefine const for C++.
9f4503d6 2034
76551463
AD
2035-----
2036
34136e65 2037Copyright (C) 1995-2012 Free Software Foundation, Inc.
76551463 2038
74553c98 2039This file is part of Bison, the GNU Parser Generator.
76551463 2040
f16b0819 2041This program is free software: you can redistribute it and/or modify
76551463 2042it under the terms of the GNU General Public License as published by
f16b0819
PE
2043the Free Software Foundation, either version 3 of the License, or
2044(at your option) any later version.
76551463 2045
f16b0819 2046This program is distributed in the hope that it will be useful,
76551463
AD
2047but WITHOUT ANY WARRANTY; without even the implied warranty of
2048MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
2049GNU General Public License for more details.
2050
2051You should have received a copy of the GNU General Public License
f16b0819 2052along with this program. If not, see <http://www.gnu.org/licenses/>.
7e508a2b
AD
2053
2054 LocalWords: yacc YYBACKUP glr GCC lalr ArrayIndexOutOfBoundsException nullptr
2055 LocalWords: cplusplus liby rpl fprintf mfcalc Wyacc stmt cond expr mk sym lr
2056 LocalWords: IELR ielr Lookahead YYERROR nonassoc LALR's api lookaheads yychar
2057 LocalWords: destructor lookahead YYRHSLOC YYLLOC Rhs ifndef YYFAIL cpp sr rr
2058 LocalWords: preprocessor initializer Wno Wnone Werror FreeBSD prec livelocks
2059 LocalWords: Solaris AIX UX RHEL Tru LHS gcc's Wundef YYENABLE NLS YYLTYPE VCG
2060 LocalWords: yyerror cpp's Wunused yylval yylloc prepend yyparse yylex yypush
2061 LocalWords: Graphviz xml nonterminals midrule destructor's YYSTYPE typedef ly
2062 LocalWords: CHR chr printf stdout namespace preprocessing enum pre include's
2063 LocalWords: YYRECOVERING nonfree destructors YYABORT YYACCEPT params enums de
2064 LocalWords: struct yystype DJGPP lex param Haible NUM alloca YYSTACK NUL goto
2065 LocalWords: YYMAXDEPTH Unescaped UCNs YYLTYPE's yyltype typedefs inline Yaccs
2066 LocalWords: Heriyanto Reenable dprec Hilfinger Eggert MYEOF Folle Menezes EOF
242cc08e 2067 LocalWords: Lackovic define's itemset Groff Gettext malloc NEWS'ed YYDEBUG YY
7e508a2b 2068 LocalWords: namespaces strerror const autoconfiguration Dconst Autoconf's FDL
242cc08e
AD
2069 LocalWords: Automake TMPDIR LESSEQ ylwrap endif yydebug YYTOKEN YYLSP ival hh
2070 LocalWords: extern YYTOKENTYPE TOKENTYPE yytokentype tokentype STYPE lval pdf
dcb366b1
AD
2071 LocalWords: lang yyoutput dvi html ps POSIX lvalp llocp Wother nterm arg init
2072 LocalWords: TOK
7e508a2b
AD
2073
2074Local Variables:
2075mode: outline
e4ab1254 2076fill-column: 76
7e508a2b 2077End: