]> git.saurik.com Git - bison.git/blob - NEWS
tests: link lib/libbison.a for gnulib.
[bison.git] / NEWS
1 Bison News
2 ----------
3
4 * Changes in version 2.4.2 (????-??-??):
5
6 ** Some portability problems in the testsuite that resulted in failures
7 on at least Solaris 2.7 have been fixed.
8
9 ** `%prec IDENTIFIER' requires IDENTIFIER to be defined separately.
10
11 POSIX specifies that an error be reported for any identifier that does
12 not appear on the LHS of a grammar rule and that is not defined by
13 %token, %left, %right, or %nonassoc. Bison 2.3b and later lost this
14 error report for the case when an identifier appears only after a
15 %prec directive. It is now restored. However, for backward
16 compatibility with recent Bison releases, it is only a warning for
17 now. In Bison 2.5 and later, it will return to being an error.
18
19 ** Detection of GNU M4 1.4.6 or newer during configure is improved.
20
21 ** Warnings from gcc's -Wundef option about undefined YYENABLE_NLS,
22 YYLTYPE_IS_TRIVIAL, and __STRICT_ANSI__ in C/C++ parsers are now
23 avoided.
24
25 ** %code is now a permanent feature.
26
27 A traditional Yacc prologue directive is written in the form:
28
29 %{CODE%}
30
31 To provide a more flexible alternative, Bison 2.3b introduced the
32 %code directive with the following forms for C/C++:
33
34 %code {CODE}
35 %code requires {CODE}
36 %code provides {CODE}
37 %code top {CODE}
38
39 These forms are now considered permanent features of Bison. See the
40 %code entries in the section "Bison Declaration Summary" in the Bison
41 manual for a summary of their functionality. See the section
42 "Prologue Alternatives" for a detailed discussion including the
43 advantages of %code over the traditional Yacc prologue directive.
44
45 Bison's Java feature as a whole including its current usage of %code
46 is still considered experimental.
47
48 ** YYFAIL is deprecated and will eventually be removed.
49
50 YYFAIL has existed for many years as an undocumented feature of
51 deterministic parsers in C generated by Bison. Previously, it was
52 documented for Bison's experimental Java parsers. YYFAIL is no longer
53 documented for Java parsers and is formally deprecated in both cases.
54 Users are strongly encouraged to migrate to YYERROR, which is
55 specified by POSIX.
56
57 Like YYERROR, you can invoke YYFAIL from a semantic action in order to
58 induce a syntax error. The most obvious difference from YYERROR is
59 that YYFAIL will automatically invoke yyerror to report the syntax
60 error so that you don't have to. However, there are several other
61 subtle differences between YYERROR and YYFAIL, and YYFAIL suffers from
62 inherent flaws when %error-verbose or `#define YYERROR_VERBOSE' is
63 used. For a more detailed discussion, see:
64
65 http://lists.gnu.org/archive/html/bison-patches/2009-12/msg00024.html
66
67 The upcoming Bison 2.5 will remove YYFAIL from Java parsers, but
68 deterministic parsers in C will continue to implement it. However,
69 because YYFAIL is already flawed, it seems futile to try to make new
70 Bison features compatible with it. Thus, during parser generation,
71 Bison 2.5 will produce a warning whenever it discovers YYFAIL in a
72 rule action. In a later release, YYFAIL will be disabled for
73 %error-verbose and `#define YYERROR_VERBOSE'. Eventually, YYFAIL will
74 be removed altogether.
75
76 There exists at least one case where Bison 2.5's YYFAIL warning will
77 be a false positive. Some projects add phony uses of YYFAIL and other
78 Bison-defined macros for the sole purpose of suppressing C
79 preprocessor warnings (from GCC cpp's -Wunused-macros, for example).
80 To avoid Bison's future warning, such YYFAIL uses can be moved to the
81 epilogue (that is, after the second `%%') in the Bison input file. In
82 this release (2.4.2), Bison already generates its own code to suppress
83 C preprocessor warnings for YYFAIL, so projects can remove their own
84 phony uses of YYFAIL if compatibility with Bison releases prior to
85 2.4.2 is not necessary.
86
87 ** Internationalization.
88
89 Fix a regression introduced in Bison 2.4: Under some circumstances,
90 message translations were not installed although supported by the
91 host system.
92
93 * Changes in version 2.4.1 (2008-12-11):
94
95 ** In the GLR defines file, unexpanded M4 macros in the yylval and yylloc
96 declarations have been fixed.
97
98 ** Temporary hack for adding a semicolon to the user action.
99
100 Bison used to prepend a trailing semicolon at the end of the user
101 action for reductions. This allowed actions such as
102
103 exp: exp "+" exp { $$ = $1 + $3 };
104
105 instead of
106
107 exp: exp "+" exp { $$ = $1 + $3; };
108
109 Some grammars still depend on this `feature'. Bison 2.4.1 restores
110 the previous behavior in the case of C output (specifically, when
111 neither %language or %skeleton or equivalent command-line options
112 are used) to leave more time for grammars depending on the old
113 behavior to be adjusted. Future releases of Bison will disable this
114 feature.
115
116 ** A few minor improvements to the Bison manual.
117
118 * Changes in version 2.4 (2008-11-02):
119
120 ** %language is an experimental feature.
121
122 We first introduced this feature in test release 2.3b as a cleaner
123 alternative to %skeleton. Since then, we have discussed the possibility of
124 modifying its effect on Bison's output file names. Thus, in this release,
125 we consider %language to be an experimental feature that will likely evolve
126 in future releases.
127
128 ** Forward compatibility with GNU M4 has been improved.
129
130 ** Several bugs in the C++ skeleton and the experimental Java skeleton have been
131 fixed.
132
133 * Changes in version 2.3b (2008-05-27):
134
135 ** The quotes around NAME that used to be required in the following directive
136 are now deprecated:
137
138 %define NAME "VALUE"
139
140 ** The directive `%pure-parser' is now deprecated in favor of:
141
142 %define api.pure
143
144 which has the same effect except that Bison is more careful to warn about
145 unreasonable usage in the latter case.
146
147 ** Push Parsing
148
149 Bison can now generate an LALR(1) parser in C with a push interface. That
150 is, instead of invoking `yyparse', which pulls tokens from `yylex', you can
151 push one token at a time to the parser using `yypush_parse', which will
152 return to the caller after processing each token. By default, the push
153 interface is disabled. Either of the following directives will enable it:
154
155 %define api.push_pull "push" // Just push; does not require yylex.
156 %define api.push_pull "both" // Push and pull; requires yylex.
157
158 See the new section `A Push Parser' in the Bison manual for details.
159
160 The current push parsing interface is experimental and may evolve. More user
161 feedback will help to stabilize it.
162
163 ** The -g and --graph options now output graphs in Graphviz DOT format,
164 not VCG format. Like --graph, -g now also takes an optional FILE argument
165 and thus cannot be bundled with other short options.
166
167 ** Java
168
169 Bison can now generate an LALR(1) parser in Java. The skeleton is
170 `data/lalr1.java'. Consider using the new %language directive instead of
171 %skeleton to select it.
172
173 See the new section `Java Parsers' in the Bison manual for details.
174
175 The current Java interface is experimental and may evolve. More user
176 feedback will help to stabilize it.
177
178 ** %language
179
180 This new directive specifies the programming language of the generated
181 parser, which can be C (the default), C++, or Java. Besides the skeleton
182 that Bison uses, the directive affects the names of the generated files if
183 the grammar file's name ends in ".y".
184
185 ** XML Automaton Report
186
187 Bison can now generate an XML report of the LALR(1) automaton using the new
188 `--xml' option. The current XML schema is experimental and may evolve. More
189 user feedback will help to stabilize it.
190
191 ** The grammar file may now specify the name of the parser header file using
192 %defines. For example:
193
194 %defines "parser.h"
195
196 ** When reporting useless rules, useless nonterminals, and unused terminals,
197 Bison now employs the terms "useless in grammar" instead of "useless",
198 "useless in parser" instead of "never reduced", and "unused in grammar"
199 instead of "unused".
200
201 ** Unreachable State Removal
202
203 Previously, Bison sometimes generated parser tables containing unreachable
204 states. A state can become unreachable during conflict resolution if Bison
205 disables a shift action leading to it from a predecessor state. Bison now:
206
207 1. Removes unreachable states.
208
209 2. Does not report any conflicts that appeared in unreachable states.
210 WARNING: As a result, you may need to update %expect and %expect-rr
211 directives in existing grammar files.
212
213 3. For any rule used only in such states, Bison now reports the rule as
214 "useless in parser due to conflicts".
215
216 This feature can be disabled with the following directive:
217
218 %define lr.keep_unreachable_states
219
220 See the %define entry in the `Bison Declaration Summary' in the Bison manual
221 for further discussion.
222
223 ** Lookahead Set Correction in the `.output' Report
224
225 When instructed to generate a `.output' file including lookahead sets
226 (using `--report=lookahead', for example), Bison now prints each reduction's
227 lookahead set only next to the associated state's one item that (1) is
228 associated with the same rule as the reduction and (2) has its dot at the end
229 of its RHS. Previously, Bison also erroneously printed the lookahead set
230 next to all of the state's other items associated with the same rule. This
231 bug affected only the `.output' file and not the generated parser source
232 code.
233
234 ** --report-file=FILE is a new option to override the default `.output' file
235 name.
236
237 ** The `=' that used to be required in the following directives is now
238 deprecated:
239
240 %file-prefix "parser"
241 %name-prefix "c_"
242 %output "parser.c"
243
244 ** An Alternative to `%{...%}' -- `%code QUALIFIER {CODE}'
245
246 Bison 2.3a provided a new set of directives as a more flexible alternative to
247 the traditional Yacc prologue blocks. Those have now been consolidated into
248 a single %code directive with an optional qualifier field, which identifies
249 the purpose of the code and thus the location(s) where Bison should generate
250 it:
251
252 1. `%code {CODE}' replaces `%after-header {CODE}'
253 2. `%code requires {CODE}' replaces `%start-header {CODE}'
254 3. `%code provides {CODE}' replaces `%end-header {CODE}'
255 4. `%code top {CODE}' replaces `%before-header {CODE}'
256
257 See the %code entries in section `Bison Declaration Summary' in the Bison
258 manual for a summary of the new functionality. See the new section `Prologue
259 Alternatives' for a detailed discussion including the advantages of %code
260 over the traditional Yacc prologues.
261
262 The prologue alternatives are experimental. More user feedback will help to
263 determine whether they should become permanent features.
264
265 ** Revised warning: unset or unused mid-rule values
266
267 Since Bison 2.2, Bison has warned about mid-rule values that are set but not
268 used within any of the actions of the parent rule. For example, Bison warns
269 about unused $2 in:
270
271 exp: '1' { $$ = 1; } '+' exp { $$ = $1 + $4; };
272
273 Now, Bison also warns about mid-rule values that are used but not set. For
274 example, Bison warns about unset $$ in the mid-rule action in:
275
276 exp: '1' { $1 = 1; } '+' exp { $$ = $2 + $4; };
277
278 However, Bison now disables both of these warnings by default since they
279 sometimes prove to be false alarms in existing grammars employing the Yacc
280 constructs $0 or $-N (where N is some positive integer).
281
282 To enable these warnings, specify the option `--warnings=midrule-values' or
283 `-W', which is a synonym for `--warnings=all'.
284
285 ** Default %destructor or %printer with `<*>' or `<>'
286
287 Bison now recognizes two separate kinds of default %destructor's and
288 %printer's:
289
290 1. Place `<*>' in a %destructor/%printer symbol list to define a default
291 %destructor/%printer for all grammar symbols for which you have formally
292 declared semantic type tags.
293
294 2. Place `<>' in a %destructor/%printer symbol list to define a default
295 %destructor/%printer for all grammar symbols without declared semantic
296 type tags.
297
298 Bison no longer supports the `%symbol-default' notation from Bison 2.3a.
299 `<*>' and `<>' combined achieve the same effect with one exception: Bison no
300 longer applies any %destructor to a mid-rule value if that mid-rule value is
301 not actually ever referenced using either $$ or $n in a semantic action.
302
303 The default %destructor's and %printer's are experimental. More user
304 feedback will help to determine whether they should become permanent
305 features.
306
307 See the section `Freeing Discarded Symbols' in the Bison manual for further
308 details.
309
310 ** %left, %right, and %nonassoc can now declare token numbers. This is required
311 by POSIX. However, see the end of section `Operator Precedence' in the Bison
312 manual for a caveat concerning the treatment of literal strings.
313
314 ** The nonfunctional --no-parser, -n, and %no-parser options have been
315 completely removed from Bison.
316
317 * Changes in version 2.3a, 2006-09-13:
318
319 ** Instead of %union, you can define and use your own union type
320 YYSTYPE if your grammar contains at least one <type> tag.
321 Your YYSTYPE need not be a macro; it can be a typedef.
322 This change is for compatibility with other Yacc implementations,
323 and is required by POSIX.
324
325 ** Locations columns and lines start at 1.
326 In accordance with the GNU Coding Standards and Emacs.
327
328 ** You may now declare per-type and default %destructor's and %printer's:
329
330 For example:
331
332 %union { char *string; }
333 %token <string> STRING1
334 %token <string> STRING2
335 %type <string> string1
336 %type <string> string2
337 %union { char character; }
338 %token <character> CHR
339 %type <character> chr
340 %destructor { free ($$); } %symbol-default
341 %destructor { free ($$); printf ("%d", @$.first_line); } STRING1 string1
342 %destructor { } <character>
343
344 guarantees that, when the parser discards any user-defined symbol that has a
345 semantic type tag other than `<character>', it passes its semantic value to
346 `free'. However, when the parser discards a `STRING1' or a `string1', it
347 also prints its line number to `stdout'. It performs only the second
348 `%destructor' in this case, so it invokes `free' only once.
349
350 [Although we failed to mention this here in the 2.3a release, the default
351 %destructor's and %printer's were experimental, and they were rewritten in
352 future versions.]
353
354 ** Except for LALR(1) parsers in C with POSIX Yacc emulation enabled (with `-y',
355 `--yacc', or `%yacc'), Bison no longer generates #define statements for
356 associating token numbers with token names. Removing the #define statements
357 helps to sanitize the global namespace during preprocessing, but POSIX Yacc
358 requires them. Bison still generates an enum for token names in all cases.
359
360 ** Handling of traditional Yacc prologue blocks is now more consistent but
361 potentially incompatible with previous releases of Bison.
362
363 As before, you declare prologue blocks in your grammar file with the
364 `%{ ... %}' syntax. To generate the pre-prologue, Bison concatenates all
365 prologue blocks that you've declared before the first %union. To generate
366 the post-prologue, Bison concatenates all prologue blocks that you've
367 declared after the first %union.
368
369 Previous releases of Bison inserted the pre-prologue into both the header
370 file and the code file in all cases except for LALR(1) parsers in C. In the
371 latter case, Bison inserted it only into the code file. For parsers in C++,
372 the point of insertion was before any token definitions (which associate
373 token numbers with names). For parsers in C, the point of insertion was
374 after the token definitions.
375
376 Now, Bison never inserts the pre-prologue into the header file. In the code
377 file, it always inserts it before the token definitions.
378
379 ** Bison now provides a more flexible alternative to the traditional Yacc
380 prologue blocks: %before-header, %start-header, %end-header, and
381 %after-header.
382
383 For example, the following declaration order in the grammar file reflects the
384 order in which Bison will output these code blocks. However, you are free to
385 declare these code blocks in your grammar file in whatever order is most
386 convenient for you:
387
388 %before-header {
389 /* Bison treats this block like a pre-prologue block: it inserts it into
390 * the code file before the contents of the header file. It does *not*
391 * insert it into the header file. This is a good place to put
392 * #include's that you want at the top of your code file. A common
393 * example is `#include "system.h"'. */
394 }
395 %start-header {
396 /* Bison inserts this block into both the header file and the code file.
397 * In both files, the point of insertion is before any Bison-generated
398 * token, semantic type, location type, and class definitions. This is a
399 * good place to define %union dependencies, for example. */
400 }
401 %union {
402 /* Unlike the traditional Yacc prologue blocks, the output order for the
403 * new %*-header blocks is not affected by their declaration position
404 * relative to any %union in the grammar file. */
405 }
406 %end-header {
407 /* Bison inserts this block into both the header file and the code file.
408 * In both files, the point of insertion is after the Bison-generated
409 * definitions. This is a good place to declare or define public
410 * functions or data structures that depend on the Bison-generated
411 * definitions. */
412 }
413 %after-header {
414 /* Bison treats this block like a post-prologue block: it inserts it into
415 * the code file after the contents of the header file. It does *not*
416 * insert it into the header file. This is a good place to declare or
417 * define internal functions or data structures that depend on the
418 * Bison-generated definitions. */
419 }
420
421 If you have multiple occurrences of any one of the above declarations, Bison
422 will concatenate the contents in declaration order.
423
424 [Although we failed to mention this here in the 2.3a release, the prologue
425 alternatives were experimental, and they were rewritten in future versions.]
426
427 ** The option `--report=look-ahead' has been changed to `--report=lookahead'.
428 The old spelling still works, but is not documented and may be removed
429 in a future release.
430
431 * Changes in version 2.3, 2006-06-05:
432
433 ** GLR grammars should now use `YYRECOVERING ()' instead of `YYRECOVERING',
434 for compatibility with LALR(1) grammars.
435
436 ** It is now documented that any definition of YYSTYPE or YYLTYPE should
437 be to a type name that does not contain parentheses or brackets.
438
439 * Changes in version 2.2, 2006-05-19:
440
441 ** The distribution terms for all Bison-generated parsers now permit
442 using the parsers in nonfree programs. Previously, this permission
443 was granted only for Bison-generated LALR(1) parsers in C.
444
445 ** %name-prefix changes the namespace name in C++ outputs.
446
447 ** The C++ parsers export their token_type.
448
449 ** Bison now allows multiple %union declarations, and concatenates
450 their contents together.
451
452 ** New warning: unused values
453 Right-hand side symbols whose values are not used are reported,
454 if the symbols have destructors. For instance:
455
456 exp: exp "?" exp ":" exp { $1 ? $1 : $3; }
457 | exp "+" exp
458 ;
459
460 will trigger a warning about $$ and $5 in the first rule, and $3 in
461 the second ($1 is copied to $$ by the default rule). This example
462 most likely contains three errors, and could be rewritten as:
463
464 exp: exp "?" exp ":" exp
465 { $$ = $1 ? $3 : $5; free ($1 ? $5 : $3); free ($1); }
466 | exp "+" exp
467 { $$ = $1 ? $1 : $3; if ($1) free ($3); }
468 ;
469
470 However, if the original actions were really intended, memory leaks
471 and all, the warnings can be suppressed by letting Bison believe the
472 values are used, e.g.:
473
474 exp: exp "?" exp ":" exp { $1 ? $1 : $3; (void) ($$, $5); }
475 | exp "+" exp { $$ = $1; (void) $3; }
476 ;
477
478 If there are mid-rule actions, the warning is issued if no action
479 uses it. The following triggers no warning: $1 and $3 are used.
480
481 exp: exp { push ($1); } '+' exp { push ($3); sum (); };
482
483 The warning is intended to help catching lost values and memory leaks.
484 If a value is ignored, its associated memory typically is not reclaimed.
485
486 ** %destructor vs. YYABORT, YYACCEPT, and YYERROR.
487 Destructors are now called when user code invokes YYABORT, YYACCEPT,
488 and YYERROR, for all objects on the stack, other than objects
489 corresponding to the right-hand side of the current rule.
490
491 ** %expect, %expect-rr
492 Incorrect numbers of expected conflicts are now actual errors,
493 instead of warnings.
494
495 ** GLR, YACC parsers.
496 The %parse-params are available in the destructors (and the
497 experimental printers) as per the documentation.
498
499 ** Bison now warns if it finds a stray `$' or `@' in an action.
500
501 ** %require "VERSION"
502 This specifies that the grammar file depends on features implemented
503 in Bison version VERSION or higher.
504
505 ** lalr1.cc: The token and value types are now class members.
506 The tokens were defined as free form enums and cpp macros. YYSTYPE
507 was defined as a free form union. They are now class members:
508 tokens are enumerations of the `yy::parser::token' struct, and the
509 semantic values have the `yy::parser::semantic_type' type.
510
511 If you do not want or can update to this scheme, the directive
512 `%define "global_tokens_and_yystype" "1"' triggers the global
513 definition of tokens and YYSTYPE. This change is suitable both
514 for previous releases of Bison, and this one.
515
516 If you wish to update, then make sure older version of Bison will
517 fail using `%require "2.2"'.
518
519 ** DJGPP support added.
520 \f
521 * Changes in version 2.1, 2005-09-16:
522
523 ** The C++ lalr1.cc skeleton supports %lex-param.
524
525 ** Bison-generated parsers now support the translation of diagnostics like
526 "syntax error" into languages other than English. The default
527 language is still English. For details, please see the new
528 Internationalization section of the Bison manual. Software
529 distributors should also see the new PACKAGING file. Thanks to
530 Bruno Haible for this new feature.
531
532 ** Wording in the Bison-generated parsers has been changed slightly to
533 simplify translation. In particular, the message "memory exhausted"
534 has replaced "parser stack overflow", as the old message was not
535 always accurate for modern Bison-generated parsers.
536
537 ** Destructors are now called when the parser aborts, for all symbols left
538 behind on the stack. Also, the start symbol is now destroyed after a
539 successful parse. In both cases, the behavior was formerly inconsistent.
540
541 ** When generating verbose diagnostics, Bison-generated parsers no longer
542 quote the literal strings associated with tokens. For example, for
543 a syntax error associated with '%token NUM "number"' they might
544 print 'syntax error, unexpected number' instead of 'syntax error,
545 unexpected "number"'.
546 \f
547 * Changes in version 2.0, 2004-12-25:
548
549 ** Possibly-incompatible changes
550
551 - Bison-generated parsers no longer default to using the alloca function
552 (when available) to extend the parser stack, due to widespread
553 problems in unchecked stack-overflow detection. You can "#define
554 YYSTACK_USE_ALLOCA 1" to require the use of alloca, but please read
555 the manual to determine safe values for YYMAXDEPTH in that case.
556
557 - Error token location.
558 During error recovery, the location of the syntax error is updated
559 to cover the whole sequence covered by the error token: it includes
560 the shifted symbols thrown away during the first part of the error
561 recovery, and the lookahead rejected during the second part.
562
563 - Semicolon changes:
564 . Stray semicolons are no longer allowed at the start of a grammar.
565 . Semicolons are now required after in-grammar declarations.
566
567 - Unescaped newlines are no longer allowed in character constants or
568 string literals. They were never portable, and GCC 3.4.0 has
569 dropped support for them. Better diagnostics are now generated if
570 forget a closing quote.
571
572 - NUL bytes are no longer allowed in Bison string literals, unfortunately.
573
574 ** New features
575
576 - GLR grammars now support locations.
577
578 - New directive: %initial-action.
579 This directive allows the user to run arbitrary code (including
580 initializing @$) from yyparse before parsing starts.
581
582 - A new directive "%expect-rr N" specifies the expected number of
583 reduce/reduce conflicts in GLR parsers.
584
585 - %token numbers can now be hexadecimal integers, e.g., `%token FOO 0x12d'.
586 This is a GNU extension.
587
588 - The option `--report=lookahead' was changed to `--report=look-ahead'.
589 [However, this was changed back after 2.3.]
590
591 - Experimental %destructor support has been added to lalr1.cc.
592
593 - New configure option --disable-yacc, to disable installation of the
594 yacc command and -ly library introduced in 1.875 for POSIX conformance.
595
596 ** Bug fixes
597
598 - For now, %expect-count violations are now just warnings, not errors.
599 This is for compatibility with Bison 1.75 and earlier (when there are
600 reduce/reduce conflicts) and with Bison 1.30 and earlier (when there
601 are too many or too few shift/reduce conflicts). However, in future
602 versions of Bison we plan to improve the %expect machinery so that
603 these violations will become errors again.
604
605 - Within Bison itself, numbers (e.g., goto numbers) are no longer
606 arbitrarily limited to 16-bit counts.
607
608 - Semicolons are now allowed before "|" in grammar rules, as POSIX requires.
609 \f
610 * Changes in version 1.875, 2003-01-01:
611
612 ** The documentation license has been upgraded to version 1.2
613 of the GNU Free Documentation License.
614
615 ** syntax error processing
616
617 - In Yacc-style parsers YYLLOC_DEFAULT is now used to compute error
618 locations too. This fixes bugs in error-location computation.
619
620 - %destructor
621 It is now possible to reclaim the memory associated to symbols
622 discarded during error recovery. This feature is still experimental.
623
624 - %error-verbose
625 This new directive is preferred over YYERROR_VERBOSE.
626
627 - #defining yyerror to steal internal variables is discouraged.
628 It is not guaranteed to work forever.
629
630 ** POSIX conformance
631
632 - Semicolons are once again optional at the end of grammar rules.
633 This reverts to the behavior of Bison 1.33 and earlier, and improves
634 compatibility with Yacc.
635
636 - `parse error' -> `syntax error'
637 Bison now uniformly uses the term `syntax error'; formerly, the code
638 and manual sometimes used the term `parse error' instead. POSIX
639 requires `syntax error' in diagnostics, and it was thought better to
640 be consistent.
641
642 - The documentation now emphasizes that yylex and yyerror must be
643 declared before use. C99 requires this.
644
645 - Bison now parses C99 lexical constructs like UCNs and
646 backslash-newline within C escape sequences, as POSIX 1003.1-2001 requires.
647
648 - File names are properly escaped in C output. E.g., foo\bar.y is
649 output as "foo\\bar.y".
650
651 - Yacc command and library now available
652 The Bison distribution now installs a `yacc' command, as POSIX requires.
653 Also, Bison now installs a small library liby.a containing
654 implementations of Yacc-compatible yyerror and main functions.
655 This library is normally not useful, but POSIX requires it.
656
657 - Type clashes now generate warnings, not errors.
658
659 - If the user does not define YYSTYPE as a macro, Bison now declares it
660 using typedef instead of defining it as a macro.
661 For consistency, YYLTYPE is also declared instead of defined.
662
663 ** Other compatibility issues
664
665 - %union directives can now have a tag before the `{', e.g., the
666 directive `%union foo {...}' now generates the C code
667 `typedef union foo { ... } YYSTYPE;'; this is for Yacc compatibility.
668 The default union tag is `YYSTYPE', for compatibility with Solaris 9 Yacc.
669 For consistency, YYLTYPE's struct tag is now `YYLTYPE' not `yyltype'.
670 This is for compatibility with both Yacc and Bison 1.35.
671
672 - `;' is output before the terminating `}' of an action, for
673 compatibility with Bison 1.35.
674
675 - Bison now uses a Yacc-style format for conflict reports, e.g.,
676 `conflicts: 2 shift/reduce, 1 reduce/reduce'.
677
678 - `yystype' and `yyltype' are now obsolescent macros instead of being
679 typedefs or tags; they are no longer documented and are planned to be
680 withdrawn in a future release.
681
682 ** GLR parser notes
683
684 - GLR and inline
685 Users of Bison have to decide how they handle the portability of the
686 C keyword `inline'.
687
688 - `parsing stack overflow...' -> `parser stack overflow'
689 GLR parsers now report `parser stack overflow' as per the Bison manual.
690
691 ** Bison now warns if it detects conflicting outputs to the same file,
692 e.g., it generates a warning for `bison -d -o foo.h foo.y' since
693 that command outputs both code and header to foo.h.
694
695 ** #line in output files
696 - --no-line works properly.
697
698 ** Bison can no longer be built by a K&R C compiler; it requires C89 or
699 later to be built. This change originally took place a few versions
700 ago, but nobody noticed until we recently asked someone to try
701 building Bison with a K&R C compiler.
702 \f
703 * Changes in version 1.75, 2002-10-14:
704
705 ** Bison should now work on 64-bit hosts.
706
707 ** Indonesian translation thanks to Tedi Heriyanto.
708
709 ** GLR parsers
710 Fix spurious parse errors.
711
712 ** Pure parsers
713 Some people redefine yyerror to steal yyparse' private variables.
714 Reenable this trick until an official feature replaces it.
715
716 ** Type Clashes
717 In agreement with POSIX and with other Yaccs, leaving a default
718 action is valid when $$ is untyped, and $1 typed:
719
720 untyped: ... typed;
721
722 but the converse remains an error:
723
724 typed: ... untyped;
725
726 ** Values of mid-rule actions
727 The following code:
728
729 foo: { ... } { $$ = $1; } ...
730
731 was incorrectly rejected: $1 is defined in the second mid-rule
732 action, and is equal to the $$ of the first mid-rule action.
733 \f
734 * Changes in version 1.50, 2002-10-04:
735
736 ** GLR parsing
737 The declaration
738 %glr-parser
739 causes Bison to produce a Generalized LR (GLR) parser, capable of handling
740 almost any context-free grammar, ambiguous or not. The new declarations
741 %dprec and %merge on grammar rules allow parse-time resolution of
742 ambiguities. Contributed by Paul Hilfinger.
743
744 Unfortunately Bison 1.50 does not work properly on 64-bit hosts
745 like the Alpha, so please stick to 32-bit hosts for now.
746
747 ** Output Directory
748 When not in Yacc compatibility mode, when the output file was not
749 specified, running `bison foo/bar.y' created `foo/bar.c'. It
750 now creates `bar.c'.
751
752 ** Undefined token
753 The undefined token was systematically mapped to 2 which prevented
754 the use of 2 by the user. This is no longer the case.
755
756 ** Unknown token numbers
757 If yylex returned an out of range value, yyparse could die. This is
758 no longer the case.
759
760 ** Error token
761 According to POSIX, the error token must be 256.
762 Bison extends this requirement by making it a preference: *if* the
763 user specified that one of her tokens is numbered 256, then error
764 will be mapped onto another number.
765
766 ** Verbose error messages
767 They no longer report `..., expecting error or...' for states where
768 error recovery is possible.
769
770 ** End token
771 Defaults to `$end' instead of `$'.
772
773 ** Error recovery now conforms to documentation and to POSIX
774 When a Bison-generated parser encounters a syntax error, it now pops
775 the stack until it finds a state that allows shifting the error
776 token. Formerly, it popped the stack until it found a state that
777 allowed some non-error action other than a default reduction on the
778 error token. The new behavior has long been the documented behavior,
779 and has long been required by POSIX. For more details, please see
780 Paul Eggert, "Reductions during Bison error handling" (2002-05-20)
781 <http://lists.gnu.org/archive/html/bug-bison/2002-05/msg00038.html>.
782
783 ** Traces
784 Popped tokens and nonterminals are now reported.
785
786 ** Larger grammars
787 Larger grammars are now supported (larger token numbers, larger grammar
788 size (= sum of the LHS and RHS lengths), larger LALR tables).
789 Formerly, many of these numbers ran afoul of 16-bit limits;
790 now these limits are 32 bits on most hosts.
791
792 ** Explicit initial rule
793 Bison used to play hacks with the initial rule, which the user does
794 not write. It is now explicit, and visible in the reports and
795 graphs as rule 0.
796
797 ** Useless rules
798 Before, Bison reported the useless rules, but, although not used,
799 included them in the parsers. They are now actually removed.
800
801 ** Useless rules, useless nonterminals
802 They are now reported, as a warning, with their locations.
803
804 ** Rules never reduced
805 Rules that can never be reduced because of conflicts are now
806 reported.
807
808 ** Incorrect `Token not used'
809 On a grammar such as
810
811 %token useless useful
812 %%
813 exp: '0' %prec useful;
814
815 where a token was used to set the precedence of the last rule,
816 bison reported both `useful' and `useless' as useless tokens.
817
818 ** Revert the C++ namespace changes introduced in 1.31
819 as they caused too many portability hassles.
820
821 ** Default locations
822 By an accident of design, the default computation of @$ was
823 performed after another default computation was performed: @$ = @1.
824 The latter is now removed: YYLLOC_DEFAULT is fully responsible of
825 the computation of @$.
826
827 ** Token end-of-file
828 The token end of file may be specified by the user, in which case,
829 the user symbol is used in the reports, the graphs, and the verbose
830 error messages instead of `$end', which remains being the default.
831 For instance
832 %token MYEOF 0
833 or
834 %token MYEOF 0 "end of file"
835
836 ** Semantic parser
837 This old option, which has been broken for ages, is removed.
838
839 ** New translations
840 Brazilian Portuguese, thanks to Alexandre Folle de Menezes.
841 Croatian, thanks to Denis Lackovic.
842
843 ** Incorrect token definitions
844 When given `%token 'a' "A"', Bison used to output `#define 'a' 65'.
845
846 ** Token definitions as enums
847 Tokens are output both as the traditional #define's, and, provided
848 the compiler supports ANSI C or is a C++ compiler, as enums.
849 This lets debuggers display names instead of integers.
850
851 ** Reports
852 In addition to --verbose, bison supports --report=THINGS, which
853 produces additional information:
854 - itemset
855 complete the core item sets with their closure
856 - lookahead [changed to `look-ahead' in 1.875e through 2.3, but changed back]
857 explicitly associate lookahead tokens to items
858 - solved
859 describe shift/reduce conflicts solving.
860 Bison used to systematically output this information on top of
861 the report. Solved conflicts are now attached to their states.
862
863 ** Type clashes
864 Previous versions don't complain when there is a type clash on
865 the default action if the rule has a mid-rule action, such as in:
866
867 %type <foo> bar
868 %%
869 bar: '0' {} '0';
870
871 This is fixed.
872
873 ** GNU M4 is now required when using Bison.
874 \f
875 * Changes in version 1.35, 2002-03-25:
876
877 ** C Skeleton
878 Some projects use Bison's C parser with C++ compilers, and define
879 YYSTYPE as a class. The recent adjustment of C parsers for data
880 alignment and 64 bit architectures made this impossible.
881
882 Because for the time being no real solution for C++ parser
883 generation exists, kludges were implemented in the parser to
884 maintain this use. In the future, when Bison has C++ parsers, this
885 kludge will be disabled.
886
887 This kludge also addresses some C++ problems when the stack was
888 extended.
889 \f
890 * Changes in version 1.34, 2002-03-12:
891
892 ** File name clashes are detected
893 $ bison foo.y -d -o foo.x
894 fatal error: header and parser would both be named `foo.x'
895
896 ** A missing `;' at the end of a rule triggers a warning
897 In accordance with POSIX, and in agreement with other
898 Yacc implementations, Bison will mandate this semicolon in the near
899 future. This eases the implementation of a Bison parser of Bison
900 grammars by making this grammar LALR(1) instead of LR(2). To
901 facilitate the transition, this release introduces a warning.
902
903 ** Revert the C++ namespace changes introduced in 1.31, as they caused too
904 many portability hassles.
905
906 ** DJGPP support added.
907
908 ** Fix test suite portability problems.
909 \f
910 * Changes in version 1.33, 2002-02-07:
911
912 ** Fix C++ issues
913 Groff could not be compiled for the definition of size_t was lacking
914 under some conditions.
915
916 ** Catch invalid @n
917 As is done with $n.
918 \f
919 * Changes in version 1.32, 2002-01-23:
920
921 ** Fix Yacc output file names
922
923 ** Portability fixes
924
925 ** Italian, Dutch translations
926 \f
927 * Changes in version 1.31, 2002-01-14:
928
929 ** Many Bug Fixes
930
931 ** GNU Gettext and %expect
932 GNU Gettext asserts 10 s/r conflicts, but there are 7. Now that
933 Bison dies on incorrect %expectations, we fear there will be
934 too many bug reports for Gettext, so _for the time being_, %expect
935 does not trigger an error when the input file is named `plural.y'.
936
937 ** Use of alloca in parsers
938 If YYSTACK_USE_ALLOCA is defined to 0, then the parsers will use
939 malloc exclusively. Since 1.29, but was not NEWS'ed.
940
941 alloca is used only when compiled with GCC, to avoid portability
942 problems as on AIX.
943
944 ** yyparse now returns 2 if memory is exhausted; formerly it dumped core.
945
946 ** When the generated parser lacks debugging code, YYDEBUG is now 0
947 (as POSIX requires) instead of being undefined.
948
949 ** User Actions
950 Bison has always permitted actions such as { $$ = $1 }: it adds the
951 ending semicolon. Now if in Yacc compatibility mode, the semicolon
952 is no longer output: one has to write { $$ = $1; }.
953
954 ** Better C++ compliance
955 The output parsers try to respect C++ namespaces.
956 [This turned out to be a failed experiment, and it was reverted later.]
957
958 ** Reduced Grammars
959 Fixed bugs when reporting useless nonterminals.
960
961 ** 64 bit hosts
962 The parsers work properly on 64 bit hosts.
963
964 ** Error messages
965 Some calls to strerror resulted in scrambled or missing error messages.
966
967 ** %expect
968 When the number of shift/reduce conflicts is correct, don't issue
969 any warning.
970
971 ** The verbose report includes the rule line numbers.
972
973 ** Rule line numbers are fixed in traces.
974
975 ** Swedish translation
976
977 ** Parse errors
978 Verbose parse error messages from the parsers are better looking.
979 Before: parse error: unexpected `'/'', expecting `"number"' or `'-'' or `'(''
980 Now: parse error: unexpected '/', expecting "number" or '-' or '('
981
982 ** Fixed parser memory leaks.
983 When the generated parser was using malloc to extend its stacks, the
984 previous allocations were not freed.
985
986 ** Fixed verbose output file.
987 Some newlines were missing.
988 Some conflicts in state descriptions were missing.
989
990 ** Fixed conflict report.
991 Option -v was needed to get the result.
992
993 ** %expect
994 Was not used.
995 Mismatches are errors, not warnings.
996
997 ** Fixed incorrect processing of some invalid input.
998
999 ** Fixed CPP guards: 9foo.h uses BISON_9FOO_H instead of 9FOO_H.
1000
1001 ** Fixed some typos in the documentation.
1002
1003 ** %token MY_EOF 0 is supported.
1004 Before, MY_EOF was silently renumbered as 257.
1005
1006 ** doc/refcard.tex is updated.
1007
1008 ** %output, %file-prefix, %name-prefix.
1009 New.
1010
1011 ** --output
1012 New, aliasing `--output-file'.
1013 \f
1014 * Changes in version 1.30, 2001-10-26:
1015
1016 ** `--defines' and `--graph' have now an optional argument which is the
1017 output file name. `-d' and `-g' do not change; they do not take any
1018 argument.
1019
1020 ** `%source_extension' and `%header_extension' are removed, failed
1021 experiment.
1022
1023 ** Portability fixes.
1024 \f
1025 * Changes in version 1.29, 2001-09-07:
1026
1027 ** The output file does not define const, as this caused problems when used
1028 with common autoconfiguration schemes. If you still use ancient compilers
1029 that lack const, compile with the equivalent of the C compiler option
1030 `-Dconst='. autoconf's AC_C_CONST macro provides one way to do this.
1031
1032 ** Added `-g' and `--graph'.
1033
1034 ** The Bison manual is now distributed under the terms of the GNU FDL.
1035
1036 ** The input and the output files has automatically a similar extension.
1037
1038 ** Russian translation added.
1039
1040 ** NLS support updated; should hopefully be less troublesome.
1041
1042 ** Added the old Bison reference card.
1043
1044 ** Added `--locations' and `%locations'.
1045
1046 ** Added `-S' and `--skeleton'.
1047
1048 ** `%raw', `-r', `--raw' is disabled.
1049
1050 ** Special characters are escaped when output. This solves the problems
1051 of the #line lines with path names including backslashes.
1052
1053 ** New directives.
1054 `%yacc', `%fixed_output_files', `%defines', `%no_parser', `%verbose',
1055 `%debug', `%source_extension' and `%header_extension'.
1056
1057 ** @$
1058 Automatic location tracking.
1059 \f
1060 * Changes in version 1.28, 1999-07-06:
1061
1062 ** Should compile better now with K&R compilers.
1063
1064 ** Added NLS.
1065
1066 ** Fixed a problem with escaping the double quote character.
1067
1068 ** There is now a FAQ.
1069 \f
1070 * Changes in version 1.27:
1071
1072 ** The make rule which prevented bison.simple from being created on
1073 some systems has been fixed.
1074 \f
1075 * Changes in version 1.26:
1076
1077 ** Bison now uses automake.
1078
1079 ** New mailing lists: <bug-bison@gnu.org> and <help-bison@gnu.org>.
1080
1081 ** Token numbers now start at 257 as previously documented, not 258.
1082
1083 ** Bison honors the TMPDIR environment variable.
1084
1085 ** A couple of buffer overruns have been fixed.
1086
1087 ** Problems when closing files should now be reported.
1088
1089 ** Generated parsers should now work even on operating systems which do
1090 not provide alloca().
1091 \f
1092 * Changes in version 1.25, 1995-10-16:
1093
1094 ** Errors in the input grammar are not fatal; Bison keeps reading
1095 the grammar file, and reports all the errors found in it.
1096
1097 ** Tokens can now be specified as multiple-character strings: for
1098 example, you could use "<=" for a token which looks like <=, instead
1099 of chosing a name like LESSEQ.
1100
1101 ** The %token_table declaration says to write a table of tokens (names
1102 and numbers) into the parser file. The yylex function can use this
1103 table to recognize multiple-character string tokens, or for other
1104 purposes.
1105
1106 ** The %no_lines declaration says not to generate any #line preprocessor
1107 directives in the parser file.
1108
1109 ** The %raw declaration says to use internal Bison token numbers, not
1110 Yacc-compatible token numbers, when token names are defined as macros.
1111
1112 ** The --no-parser option produces the parser tables without including
1113 the parser engine; a project can now use its own parser engine.
1114 The actions go into a separate file called NAME.act, in the form of
1115 a switch statement body.
1116 \f
1117 * Changes in version 1.23:
1118
1119 The user can define YYPARSE_PARAM as the name of an argument to be
1120 passed into yyparse. The argument should have type void *. It should
1121 actually point to an object. Grammar actions can access the variable
1122 by casting it to the proper pointer type.
1123
1124 Line numbers in output file corrected.
1125 \f
1126 * Changes in version 1.22:
1127
1128 --help option added.
1129 \f
1130 * Changes in version 1.20:
1131
1132 Output file does not redefine const for C++.
1133
1134 Local Variables:
1135 mode: outline
1136 End:
1137
1138 -----
1139
1140 Copyright (C) 1995-2010 Free Software Foundation, Inc.
1141
1142 This file is part of Bison, the GNU Parser Generator.
1143
1144 This program is free software: you can redistribute it and/or modify
1145 it under the terms of the GNU General Public License as published by
1146 the Free Software Foundation, either version 3 of the License, or
1147 (at your option) any later version.
1148
1149 This program is distributed in the hope that it will be useful,
1150 but WITHOUT ANY WARRANTY; without even the implied warranty of
1151 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
1152 GNU General Public License for more details.
1153
1154 You should have received a copy of the GNU General Public License
1155 along with this program. If not, see <http://www.gnu.org/licenses/>.