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