NEWS

   1 Bison News
   2 ----------
   3
   4 Changes in version 2.3a+ (????-??-??):
   5
   6 * The -g and --graph options now output graphs in Graphviz DOT format,
   7   not VCG format.
   8
   9 * An experimental directive %language specifies the language of the
  10   generated parser, which can be C (the default) or C++.  This
  11   directive affects the skeleton used, and the names of the generated
  12   files if the grammar file's name ends in ".y".
  13
  14 * The grammar file may now specify the name of the parser header file using
  15   %defines.  For example:
  16
  17     %defines "parser.h"
  18
  19 * The `=' that used to be required in the following declarations is now
  20   deprecated:
  21
  22     %file-prefix "parser"
  23     %name-prefix "c_"
  24     %output "parser.c"
  25
  26 * Revised warning: unset or unused mid-rule values
  27
  28   Since Bison 2.2, Bison has warned about mid-rule values that are set but not
  29   used within any of the actions of the parent rule.  For example, Bison warns
  30   about unused $2 in:
  31
  32     exp: '1' { $$ = 1; } '+' exp { $$ = $1 + $4; };
  33
  34   Now, Bison also warns about mid-rule values that are used but not set.  For
  35   example, Bison warns about unset $$ in the mid-rule action in:
  36
  37     exp: '1' { $1 = 1; } '+' exp { $$ = $2 + $4; };
  38
  39   However, Bison now disables both of these warnings by default since they
  40   sometimes prove to be false alarms in existing grammars employing the Yacc
  41   constructs $0 or $-N (where N is some positive integer).
  42
  43   To enable these warnings, specify the flag `--warnings=midrule-values' or
  44   `-W', which is a synonym for `--warnings=all'.
  45
  46 * Bison now recognizes two separate kinds of default %destructor's and
  47   %printer's:
  48
  49     1. Place `<*>' in a %destructor/%printer symbol list to define a default
  50        %destructor/%printer for all grammar symbols for which you have formally
  51        declared semantic type tags.
  52
  53     2. Place `<>' in a %destructor/%printer symbol list to define a default
  54        %destructor/%printer for all grammar symbols without declared semantic
  55        type tags.
  56
  57   Bison no longer supports the `%symbol-default' notation from Bison 2.3a.
  58   `<*>' and `<>' combined achieve the same effect with one exception: Bison no
  59   longer applies any %destructor to a mid-rule value if that mid-rule value is
  60   not actually ever referenced using either $$ or $n in a semantic action.
  61
  62   The default %destructor's and %printer's are experimental.  More user
  63   feedback will help to determine whether they should become permanent
  64   features.
  65
  66   See the section `Freeing Discarded Symbols' in the Bison manual for further
  67   details.
  68
  69 * The Yacc prologue alternatives from Bison 2.3a have been rewritten as the
  70   following directives:
  71
  72     1. %code {CODE}
  73
  74        Other than semantic actions, this is probably the most common place you
  75        should write verbatim code for the parser implementation.  It replaces
  76        the traditional Yacc prologue, `%{CODE%}', for most purposes.  Compare
  77        with:
  78
  79          - `%{CODE%}' appearing after the first `%union {CODE}' in a grammar
  80            file.  While Bison will continue to support `%{CODE%}' for backward
  81            compatibility, `%code {CODE}' is cleaner as its functionality does
  82            not depend on its position in the grammar file relative to any
  83            `%union {CODE}'.  Specifically, `%code {CODE}' always inserts your
  84            CODE into the parser code file after the usual contents of the
  85            parser header file.
  86          - `%after-header {CODE}', which only Bison 2.3a supported.
  87
  88     2. %requires {CODE}
  89
  90        This is the right place to write dependency code for externally exposed
  91        definitions required by Bison.  Such exposed definitions are those
  92        usually appearing in the parser header file.  Thus, this is the right
  93        place to define types referenced in `%union {CODE}' directives, and it
  94        is the right place to override Bison's default YYSTYPE and YYLTYPE
  95        definitions.  Compare with:
  96
  97          - `%{CODE%}' appearing before the first `%union {CODE}' in a grammar
  98            file.  Unlike `%{CODE%}', `%requires {CODE}' inserts your CODE both
  99            into the parser code file and into the parser header file since
 100            Bison's required definitions should depend on it in both places.
 101          - `%start-header {CODE}', which only Bison 2.3a supported.
 102
 103     3. %provides {CODE}
 104
 105        This is the right place to write additional definitions you would like
 106        Bison to expose externally.  That is, this directive inserts your CODE
 107        both into the parser header file and into the parser code file after
 108        Bison's required definitions.  Compare with:
 109
 110          - `%end-header {CODE}', which only Bison 2.3a supported.
 111
 112     4. %code-top {CODE}
 113
 114        Occasionally it is desirable to insert code near the top of the parser
 115        code file.  For example:
 116
 117          %code-top {
 118            #define _GNU_SOURCE
 119            #include <stdio.h>
 120          }
 121
 122        Compare with:
 123
 124          - `%{CODE%}' appearing before the first `%union {CODE}' in a grammar
 125            file.  `%code-top {CODE}' is cleaner as its functionality does not
 126            depend on its position in the grammar file relative to any
 127            `%union {CODE}'.
 128          - `%before-header {CODE}', which only Bison 2.3a supported.
 129
 130   If you have multiple occurrences of any one of the above four directives,
 131   Bison will concatenate the contents in the order they appear in the grammar
 132   file.
 133
 134   The prologue alternatives are experimental.  More user feedback will help to
 135   determine whether they should become permanent features.
 136
 137   Also see the new section `Prologue Alternatives' in the Bison manual.
 138
 139 Changes in version 2.3a, 2006-09-13:
 140
 141 * Instead of %union, you can define and use your own union type
 142   YYSTYPE if your grammar contains at least one <type> tag.
 143   Your YYSTYPE need not be a macro; it can be a typedef.
 144   This change is for compatibility with other Yacc implementations,
 145   and is required by POSIX.
 146
 147 * Locations columns and lines start at 1.
 148   In accordance with the GNU Coding Standards and Emacs.
 149
 150 * You may now declare per-type and default %destructor's and %printer's:
 151
 152   For example:
 153
 154     %union { char *string; }
 155     %token <string> STRING1
 156     %token <string> STRING2
 157     %type  <string> string1
 158     %type  <string> string2
 159     %union { char character; }
 160     %token <character> CHR
 161     %type  <character> chr
 162     %destructor { free ($$); } %symbol-default
 163     %destructor { free ($$); printf ("%d", @$.first_line); } STRING1 string1
 164     %destructor { } <character>
 165
 166   guarantees that, when the parser discards any user-defined symbol that has a
 167   semantic type tag other than `<character>', it passes its semantic value to
 168   `free'.  However, when the parser discards a `STRING1' or a `string1', it
 169   also prints its line number to `stdout'.  It performs only the second
 170   `%destructor' in this case, so it invokes `free' only once.
 171
 172   [Although we failed to mention this here in the 2.3a release, the default
 173   %destructor's and %printer's were experimental, and they were rewritten in
 174   future versions.]
 175
 176 * Except for LALR(1) parsers in C with POSIX Yacc emulation enabled (with `-y',
 177   `--yacc', or `%yacc'), Bison no longer generates #define statements for
 178   associating token numbers with token names.  Removing the #define statements
 179   helps to sanitize the global namespace during preprocessing, but POSIX Yacc
 180   requires them.  Bison still generates an enum for token names in all cases.
 181
 182 * Handling of traditional Yacc prologue blocks is now more consistent but
 183   potentially incompatible with previous releases of Bison.
 184
 185   As before, you declare prologue blocks in your grammar file with the
 186   `%{ ... %}' syntax.  To generate the pre-prologue, Bison concatenates all
 187   prologue blocks that you've declared before the first %union.  To generate
 188   the post-prologue, Bison concatenates all prologue blocks that you've
 189   declared after the first %union.
 190
 191   Previous releases of Bison inserted the pre-prologue into both the header
 192   file and the code file in all cases except for LALR(1) parsers in C.  In the
 193   latter case, Bison inserted it only into the code file.  For parsers in C++,
 194   the point of insertion was before any token definitions (which associate
 195   token numbers with names).  For parsers in C, the point of insertion was
 196   after the token definitions.
 197
 198   Now, Bison never inserts the pre-prologue into the header file.  In the code
 199   file, it always inserts it before the token definitions.
 200
 201 * Bison now provides a more flexible alternative to the traditional Yacc
 202   prologue blocks: %before-header, %start-header, %end-header, and
 203   %after-header.
 204
 205   For example, the following declaration order in the grammar file reflects the
 206   order in which Bison will output these code blocks.  However, you are free to
 207   declare these code blocks in your grammar file in whatever order is most
 208   convenient for you:
 209
 210     %before-header {
 211       /* Bison treats this block like a pre-prologue block: it inserts it into
 212        * the code file before the contents of the header file.  It does *not*
 213        * insert it into the header file.  This is a good place to put
 214        * #include's that you want at the top of your code file.  A common
 215        * example is `#include "system.h"'.  */
 216     }
 217     %start-header {
 218       /* Bison inserts this block into both the header file and the code file.
 219        * In both files, the point of insertion is before any Bison-generated
 220        * token, semantic type, location type, and class definitions.  This is a
 221        * good place to define %union dependencies, for example.  */
 222     }
 223     %union {
 224       /* Unlike the traditional Yacc prologue blocks, the output order for the
 225        * new %*-header blocks is not affected by their declaration position
 226        * relative to any %union in the grammar file.  */
 227     }
 228     %end-header {
 229       /* Bison inserts this block into both the header file and the code file.
 230        * In both files, the point of insertion is after the Bison-generated
 231        * definitions.  This is a good place to declare or define public
 232        * functions or data structures that depend on the Bison-generated
 233        * definitions.  */
 234     }
 235     %after-header {
 236       /* Bison treats this block like a post-prologue block: it inserts it into
 237        * the code file after the contents of the header file.  It does *not*
 238        * insert it into the header file.  This is a good place to declare or
 239        * define internal functions or data structures that depend on the
 240        * Bison-generated definitions.  */
 241     }
 242
 243   If you have multiple occurrences of any one of the above declarations, Bison
 244   will concatenate the contents in declaration order.
 245
 246   [Although we failed to mention this here in the 2.3a release, the prologue
 247   alternatives were experimental, and they were rewritten in future versions.]
 248
 249 * The option `--report=look-ahead' has been changed to `--report=lookahead'.
 250   The old spelling still works, but is not documented and may be removed
 251   in a future release.
 252
 253 Changes in version 2.3, 2006-06-05:
 254
 255 * GLR grammars should now use `YYRECOVERING ()' instead of `YYRECOVERING',
 256   for compatibility with LALR(1) grammars.
 257
 258 * It is now documented that any definition of YYSTYPE or YYLTYPE should
 259   be to a type name that does not contain parentheses or brackets.
 260
 261 Changes in version 2.2, 2006-05-19:
 262
 263 * The distribution terms for all Bison-generated parsers now permit
 264   using the parsers in nonfree programs.  Previously, this permission
 265   was granted only for Bison-generated LALR(1) parsers in C.
 266
 267 * %name-prefix changes the namespace name in C++ outputs.
 268
 269 * The C++ parsers export their token_type.
 270
 271 * Bison now allows multiple %union declarations, and concatenates
 272   their contents together.
 273
 274 * New warning: unused values
 275   Right-hand side symbols whose values are not used are reported,
 276   if the symbols have destructors.  For instance:
 277
 278      exp: exp "?" exp ":" exp { $1 ? $1 : $3; }
 279         | exp "+" exp
 280         ;
 281
 282   will trigger a warning about $$ and $5 in the first rule, and $3 in
 283   the second ($1 is copied to $$ by the default rule).  This example
 284   most likely contains three errors, and could be rewritten as:
 285
 286      exp: exp "?" exp ":" exp
 287             { $$ = $1 ? $3 : $5; free ($1 ? $5 : $3); free ($1); }
 288         | exp "+" exp
 289             { $$ = $1 ? $1 : $3; if ($1) free ($3); }
 290         ;
 291
 292   However, if the original actions were really intended, memory leaks
 293   and all, the warnings can be suppressed by letting Bison believe the
 294   values are used, e.g.:
 295
 296      exp: exp "?" exp ":" exp { $1 ? $1 : $3; (void) ($$, $5); }
 297         | exp "+" exp         { $$ = $1; (void) $3; }
 298         ;
 299
 300   If there are mid-rule actions, the warning is issued if no action
 301   uses it.  The following triggers no warning: $1 and $3 are used.
 302
 303      exp: exp { push ($1); } '+' exp { push ($3); sum (); };
 304
 305   The warning is intended to help catching lost values and memory leaks.
 306   If a value is ignored, its associated memory typically is not reclaimed.
 307
 308 * %destructor vs. YYABORT, YYACCEPT, and YYERROR.
 309   Destructors are now called when user code invokes YYABORT, YYACCEPT,
 310   and YYERROR, for all objects on the stack, other than objects
 311   corresponding to the right-hand side of the current rule.
 312
 313 * %expect, %expect-rr
 314   Incorrect numbers of expected conflicts are now actual errors,
 315   instead of warnings.
 316
 317 * GLR, YACC parsers.
 318   The %parse-params are available in the destructors (and the
 319   experimental printers) as per the documentation.
 320
 321 * Bison now warns if it finds a stray `$' or `@' in an action.
 322
 323 * %require "VERSION"
 324   This specifies that the grammar file depends on features implemented
 325   in Bison version VERSION or higher.
 326
 327 * lalr1.cc: The token and value types are now class members.
 328   The tokens were defined as free form enums and cpp macros.  YYSTYPE
 329   was defined as a free form union.  They are now class members:
 330   tokens are enumerations of the `yy::parser::token' struct, and the
 331   semantic values have the `yy::parser::semantic_type' type.
 332
 333   If you do not want or can update to this scheme, the directive
 334   `%define "global_tokens_and_yystype" "1"' triggers the global
 335   definition of tokens and YYSTYPE.  This change is suitable both
 336   for previous releases of Bison, and this one.
 337
 338   If you wish to update, then make sure older version of Bison will
 339   fail using `%require "2.2"'.
 340
 341 * DJGPP support added.
 342 \f
 343 Changes in version 2.1, 2005-09-16:
 344
 345 * The C++ lalr1.cc skeleton supports %lex-param.
 346
 347 * Bison-generated parsers now support the translation of diagnostics like
 348   "syntax error" into languages other than English.  The default
 349   language is still English.  For details, please see the new
 350   Internationalization section of the Bison manual.  Software
 351   distributors should also see the new PACKAGING file.  Thanks to
 352   Bruno Haible for this new feature.
 353
 354 * Wording in the Bison-generated parsers has been changed slightly to
 355   simplify translation.  In particular, the message "memory exhausted"
 356   has replaced "parser stack overflow", as the old message was not
 357   always accurate for modern Bison-generated parsers.
 358
 359 * Destructors are now called when the parser aborts, for all symbols left
 360   behind on the stack.  Also, the start symbol is now destroyed after a
 361   successful parse.  In both cases, the behavior was formerly inconsistent.
 362
 363 * When generating verbose diagnostics, Bison-generated parsers no longer
 364   quote the literal strings associated with tokens.  For example, for
 365   a syntax error associated with '%token NUM "number"' they might
 366   print 'syntax error, unexpected number' instead of 'syntax error,
 367   unexpected "number"'.
 368 \f
 369 Changes in version 2.0, 2004-12-25:
 370
 371 * Possibly-incompatible changes
 372
 373   - Bison-generated parsers no longer default to using the alloca function
 374     (when available) to extend the parser stack, due to widespread
 375     problems in unchecked stack-overflow detection.  You can "#define
 376     YYSTACK_USE_ALLOCA 1" to require the use of alloca, but please read
 377     the manual to determine safe values for YYMAXDEPTH in that case.
 378
 379   - Error token location.
 380     During error recovery, the location of the syntax error is updated
 381     to cover the whole sequence covered by the error token: it includes
 382     the shifted symbols thrown away during the first part of the error
 383     recovery, and the lookahead rejected during the second part.
 384
 385   - Semicolon changes:
 386     . Stray semicolons are no longer allowed at the start of a grammar.
 387     . Semicolons are now required after in-grammar declarations.
 388
 389   - Unescaped newlines are no longer allowed in character constants or
 390     string literals.  They were never portable, and GCC 3.4.0 has
 391     dropped support for them.  Better diagnostics are now generated if
 392     forget a closing quote.
 393
 394   - NUL bytes are no longer allowed in Bison string literals, unfortunately.
 395
 396 * New features
 397
 398   - GLR grammars now support locations.
 399
 400   - New directive: %initial-action.
 401     This directive allows the user to run arbitrary code (including
 402     initializing @$) from yyparse before parsing starts.
 403
 404   - A new directive "%expect-rr N" specifies the expected number of
 405     reduce/reduce conflicts in GLR parsers.
 406
 407   - %token numbers can now be hexadecimal integers, e.g., `%token FOO 0x12d'.
 408     This is a GNU extension.
 409
 410   - The option `--report=lookahead' was changed to `--report=look-ahead'.
 411     [However, this was changed back after 2.3.]
 412
 413   - Experimental %destructor support has been added to lalr1.cc.
 414
 415   - New configure option --disable-yacc, to disable installation of the
 416     yacc command and -ly library introduced in 1.875 for POSIX conformance.
 417
 418 * Bug fixes
 419
 420   - For now, %expect-count violations are now just warnings, not errors.
 421     This is for compatibility with Bison 1.75 and earlier (when there are
 422     reduce/reduce conflicts) and with Bison 1.30 and earlier (when there
 423     are too many or too few shift/reduce conflicts).  However, in future
 424     versions of Bison we plan to improve the %expect machinery so that
 425     these violations will become errors again.
 426
 427   - Within Bison itself, numbers (e.g., goto numbers) are no longer
 428     arbitrarily limited to 16-bit counts.
 429
 430   - Semicolons are now allowed before "|" in grammar rules, as POSIX requires.
 431 \f
 432 Changes in version 1.875, 2003-01-01:
 433
 434 * The documentation license has been upgraded to version 1.2
 435   of the GNU Free Documentation License.
 436
 437 * syntax error processing
 438
 439   - In Yacc-style parsers YYLLOC_DEFAULT is now used to compute error
 440     locations too.  This fixes bugs in error-location computation.
 441
 442   - %destructor
 443     It is now possible to reclaim the memory associated to symbols
 444     discarded during error recovery.  This feature is still experimental.
 445
 446   - %error-verbose
 447     This new directive is preferred over YYERROR_VERBOSE.
 448
 449   - #defining yyerror to steal internal variables is discouraged.
 450     It is not guaranteed to work forever.
 451
 452 * POSIX conformance
 453
 454   - Semicolons are once again optional at the end of grammar rules.
 455     This reverts to the behavior of Bison 1.33 and earlier, and improves
 456     compatibility with Yacc.
 457
 458   - `parse error' -> `syntax error'
 459     Bison now uniformly uses the term `syntax error'; formerly, the code
 460     and manual sometimes used the term `parse error' instead.  POSIX
 461     requires `syntax error' in diagnostics, and it was thought better to
 462     be consistent.
 463
 464   - The documentation now emphasizes that yylex and yyerror must be
 465     declared before use.  C99 requires this.
 466
 467   - Bison now parses C99 lexical constructs like UCNs and
 468     backslash-newline within C escape sequences, as POSIX 1003.1-2001 requires.
 469
 470   - File names are properly escaped in C output.  E.g., foo\bar.y is
 471     output as "foo\\bar.y".
 472
 473   - Yacc command and library now available
 474     The Bison distribution now installs a `yacc' command, as POSIX requires.
 475     Also, Bison now installs a small library liby.a containing
 476     implementations of Yacc-compatible yyerror and main functions.
 477     This library is normally not useful, but POSIX requires it.
 478
 479   - Type clashes now generate warnings, not errors.
 480
 481   - If the user does not define YYSTYPE as a macro, Bison now declares it
 482     using typedef instead of defining it as a macro.
 483     For consistency, YYLTYPE is also declared instead of defined.
 484
 485 * Other compatibility issues
 486
 487   - %union directives can now have a tag before the `{', e.g., the
 488     directive `%union foo {...}' now generates the C code
 489     `typedef union foo { ... } YYSTYPE;'; this is for Yacc compatibility.
 490     The default union tag is `YYSTYPE', for compatibility with Solaris 9 Yacc.
 491     For consistency, YYLTYPE's struct tag is now `YYLTYPE' not `yyltype'.
 492     This is for compatibility with both Yacc and Bison 1.35.
 493
 494   - `;' is output before the terminating `}' of an action, for
 495     compatibility with Bison 1.35.
 496
 497   - Bison now uses a Yacc-style format for conflict reports, e.g.,
 498     `conflicts: 2 shift/reduce, 1 reduce/reduce'.
 499
 500   - `yystype' and `yyltype' are now obsolescent macros instead of being
 501     typedefs or tags; they are no longer documented and are planned to be
 502     withdrawn in a future release.
 503
 504 * GLR parser notes
 505
 506   - GLR and inline
 507     Users of Bison have to decide how they handle the portability of the
 508     C keyword `inline'.
 509
 510   - `parsing stack overflow...' -> `parser stack overflow'
 511     GLR parsers now report `parser stack overflow' as per the Bison manual.
 512
 513 * Bison now warns if it detects conflicting outputs to the same file,
 514   e.g., it generates a warning for `bison -d -o foo.h foo.y' since
 515   that command outputs both code and header to foo.h.
 516
 517 * #line in output files
 518   - --no-line works properly.
 519
 520 * Bison can no longer be built by a K&R C compiler; it requires C89 or
 521   later to be built.  This change originally took place a few versions
 522   ago, but nobody noticed until we recently asked someone to try
 523   building Bison with a K&R C compiler.
 524 \f
 525 Changes in version 1.75, 2002-10-14:
 526
 527 * Bison should now work on 64-bit hosts.
 528
 529 * Indonesian translation thanks to Tedi Heriyanto.
 530
 531 * GLR parsers
 532   Fix spurious parse errors.
 533
 534 * Pure parsers
 535   Some people redefine yyerror to steal yyparse' private variables.
 536   Reenable this trick until an official feature replaces it.
 537
 538 * Type Clashes
 539   In agreement with POSIX and with other Yaccs, leaving a default
 540   action is valid when $$ is untyped, and $1 typed:
 541
 542         untyped: ... typed;
 543
 544   but the converse remains an error:
 545
 546         typed: ... untyped;
 547
 548 * Values of mid-rule actions
 549   The following code:
 550
 551         foo: { ... } { $$ = $1; } ...
 552
 553   was incorrectly rejected: $1 is defined in the second mid-rule
 554   action, and is equal to the $$ of the first mid-rule action.
 555 \f
 556 Changes in version 1.50, 2002-10-04:
 557
 558 * GLR parsing
 559   The declaration
 560      %glr-parser
 561   causes Bison to produce a Generalized LR (GLR) parser, capable of handling
 562   almost any context-free grammar, ambiguous or not.  The new declarations
 563   %dprec and %merge on grammar rules allow parse-time resolution of
 564   ambiguities.  Contributed by Paul Hilfinger.
 565
 566   Unfortunately Bison 1.50 does not work properly on 64-bit hosts
 567   like the Alpha, so please stick to 32-bit hosts for now.
 568
 569 * Output Directory
 570   When not in Yacc compatibility mode, when the output file was not
 571   specified, running `bison foo/bar.y' created `foo/bar.c'.  It
 572   now creates `bar.c'.
 573
 574 * Undefined token
 575   The undefined token was systematically mapped to 2 which prevented
 576   the use of 2 by the user.  This is no longer the case.
 577
 578 * Unknown token numbers
 579   If yylex returned an out of range value, yyparse could die.  This is
 580   no longer the case.
 581
 582 * Error token
 583   According to POSIX, the error token must be 256.
 584   Bison extends this requirement by making it a preference: *if* the
 585   user specified that one of her tokens is numbered 256, then error
 586   will be mapped onto another number.
 587
 588 * Verbose error messages
 589   They no longer report `..., expecting error or...' for states where
 590   error recovery is possible.
 591
 592 * End token
 593   Defaults to `$end' instead of `$'.
 594
 595 * Error recovery now conforms to documentation and to POSIX
 596   When a Bison-generated parser encounters a syntax error, it now pops
 597   the stack until it finds a state that allows shifting the error
 598   token.  Formerly, it popped the stack until it found a state that
 599   allowed some non-error action other than a default reduction on the
 600   error token.  The new behavior has long been the documented behavior,
 601   and has long been required by POSIX.  For more details, please see
 602   Paul Eggert, "Reductions during Bison error handling" (2002-05-20)
 603   <http://lists.gnu.org/archive/html/bug-bison/2002-05/msg00038.html>.
 604
 605 * Traces
 606   Popped tokens and nonterminals are now reported.
 607
 608 * Larger grammars
 609   Larger grammars are now supported (larger token numbers, larger grammar
 610   size (= sum of the LHS and RHS lengths), larger LALR tables).
 611   Formerly, many of these numbers ran afoul of 16-bit limits;
 612   now these limits are 32 bits on most hosts.
 613
 614 * Explicit initial rule
 615   Bison used to play hacks with the initial rule, which the user does
 616   not write.  It is now explicit, and visible in the reports and
 617   graphs as rule 0.
 618
 619 * Useless rules
 620   Before, Bison reported the useless rules, but, although not used,
 621   included them in the parsers.  They are now actually removed.
 622
 623 * Useless rules, useless nonterminals
 624   They are now reported, as a warning, with their locations.
 625
 626 * Rules never reduced
 627   Rules that can never be reduced because of conflicts are now
 628   reported.
 629
 630 * Incorrect `Token not used'
 631   On a grammar such as
 632
 633            %token useless useful
 634            %%
 635            exp: '0' %prec useful;
 636
 637   where a token was used to set the precedence of the last rule,
 638   bison reported both `useful' and `useless' as useless tokens.
 639
 640 * Revert the C++ namespace changes introduced in 1.31
 641   as they caused too many portability hassles.
 642
 643 * Default locations
 644   By an accident of design, the default computation of @$ was
 645   performed after another default computation was performed: @$ = @1.
 646   The latter is now removed: YYLLOC_DEFAULT is fully responsible of
 647   the computation of @$.
 648
 649 * Token end-of-file
 650   The token end of file may be specified by the user, in which case,
 651   the user symbol is used in the reports, the graphs, and the verbose
 652   error messages instead of `$end', which remains being the default.
 653   For instance
 654      %token MYEOF 0
 655   or
 656      %token MYEOF 0 "end of file"
 657
 658 * Semantic parser
 659   This old option, which has been broken for ages, is removed.
 660
 661 * New translations
 662   Brazilian Portuguese, thanks to Alexandre Folle de Menezes.
 663   Croatian, thanks to Denis Lackovic.
 664
 665 * Incorrect token definitions
 666   When given `%token 'a' "A"', Bison used to output `#define 'a' 65'.
 667
 668 * Token definitions as enums
 669   Tokens are output both as the traditional #define's, and, provided
 670   the compiler supports ANSI C or is a C++ compiler, as enums.
 671   This lets debuggers display names instead of integers.
 672
 673 * Reports
 674   In addition to --verbose, bison supports --report=THINGS, which
 675   produces additional information:
 676   - itemset
 677     complete the core item sets with their closure
 678   - lookahead [changed to `look-ahead' in 1.875e through 2.3, but changed back]
 679     explicitly associate lookahead tokens to items
 680   - solved
 681     describe shift/reduce conflicts solving.
 682     Bison used to systematically output this information on top of
 683     the report.  Solved conflicts are now attached to their states.
 684
 685 * Type clashes
 686   Previous versions don't complain when there is a type clash on
 687   the default action if the rule has a mid-rule action, such as in:
 688
 689       %type <foo> bar
 690       %%
 691       bar: '0' {} '0';
 692
 693   This is fixed.
 694
 695 * GNU M4 is now required when using Bison.
 696 \f
 697 Changes in version 1.35, 2002-03-25:
 698
 699 * C Skeleton
 700   Some projects use Bison's C parser with C++ compilers, and define
 701   YYSTYPE as a class.  The recent adjustment of C parsers for data
 702   alignment and 64 bit architectures made this impossible.
 703
 704   Because for the time being no real solution for C++ parser
 705   generation exists, kludges were implemented in the parser to
 706   maintain this use.  In the future, when Bison has C++ parsers, this
 707   kludge will be disabled.
 708
 709   This kludge also addresses some C++ problems when the stack was
 710   extended.
 711 \f
 712 Changes in version 1.34, 2002-03-12:
 713
 714 * File name clashes are detected
 715   $ bison foo.y -d -o foo.x
 716   fatal error: header and parser would both be named `foo.x'
 717
 718 * A missing `;' at the end of a rule triggers a warning
 719   In accordance with POSIX, and in agreement with other
 720   Yacc implementations, Bison will mandate this semicolon in the near
 721   future.  This eases the implementation of a Bison parser of Bison
 722   grammars by making this grammar LALR(1) instead of LR(2).  To
 723   facilitate the transition, this release introduces a warning.
 724
 725 * Revert the C++ namespace changes introduced in 1.31, as they caused too
 726   many portability hassles.
 727
 728 * DJGPP support added.
 729
 730 * Fix test suite portability problems.
 731 \f
 732 Changes in version 1.33, 2002-02-07:
 733
 734 * Fix C++ issues
 735   Groff could not be compiled for the definition of size_t was lacking
 736   under some conditions.
 737
 738 * Catch invalid @n
 739   As is done with $n.
 740 \f
 741 Changes in version 1.32, 2002-01-23:
 742
 743 * Fix Yacc output file names
 744
 745 * Portability fixes
 746
 747 * Italian, Dutch translations
 748 \f
 749 Changes in version 1.31, 2002-01-14:
 750
 751 * Many Bug Fixes
 752
 753 * GNU Gettext and %expect
 754   GNU Gettext asserts 10 s/r conflicts, but there are 7.  Now that
 755   Bison dies on incorrect %expectations, we fear there will be
 756   too many bug reports for Gettext, so _for the time being_, %expect
 757   does not trigger an error when the input file is named `plural.y'.
 758
 759 * Use of alloca in parsers
 760   If YYSTACK_USE_ALLOCA is defined to 0, then the parsers will use
 761   malloc exclusively.  Since 1.29, but was not NEWS'ed.
 762
 763   alloca is used only when compiled with GCC, to avoid portability
 764   problems as on AIX.
 765
 766 * yyparse now returns 2 if memory is exhausted; formerly it dumped core.
 767
 768 * When the generated parser lacks debugging code, YYDEBUG is now 0
 769   (as POSIX requires) instead of being undefined.
 770
 771 * User Actions
 772   Bison has always permitted actions such as { $$ = $1 }: it adds the
 773   ending semicolon.  Now if in Yacc compatibility mode, the semicolon
 774   is no longer output: one has to write { $$ = $1; }.
 775
 776 * Better C++ compliance
 777   The output parsers try to respect C++ namespaces.
 778   [This turned out to be a failed experiment, and it was reverted later.]
 779
 780 * Reduced Grammars
 781   Fixed bugs when reporting useless nonterminals.
 782
 783 * 64 bit hosts
 784   The parsers work properly on 64 bit hosts.
 785
 786 * Error messages
 787   Some calls to strerror resulted in scrambled or missing error messages.
 788
 789 * %expect
 790   When the number of shift/reduce conflicts is correct, don't issue
 791   any warning.
 792
 793 * The verbose report includes the rule line numbers.
 794
 795 * Rule line numbers are fixed in traces.
 796
 797 * Swedish translation
 798
 799 * Parse errors
 800   Verbose parse error messages from the parsers are better looking.
 801   Before: parse error: unexpected `'/'', expecting `"number"' or `'-'' or `'(''
 802      Now: parse error: unexpected '/', expecting "number" or '-' or '('
 803
 804 * Fixed parser memory leaks.
 805   When the generated parser was using malloc to extend its stacks, the
 806   previous allocations were not freed.
 807
 808 * Fixed verbose output file.
 809   Some newlines were missing.
 810   Some conflicts in state descriptions were missing.
 811
 812 * Fixed conflict report.
 813   Option -v was needed to get the result.
 814
 815 * %expect
 816   Was not used.
 817   Mismatches are errors, not warnings.
 818
 819 * Fixed incorrect processing of some invalid input.
 820
 821 * Fixed CPP guards: 9foo.h uses BISON_9FOO_H instead of 9FOO_H.
 822
 823 * Fixed some typos in the documentation.
 824
 825 * %token MY_EOF 0 is supported.
 826   Before, MY_EOF was silently renumbered as 257.
 827
 828 * doc/refcard.tex is updated.
 829
 830 * %output, %file-prefix, %name-prefix.
 831   New.
 832
 833 * --output
 834   New, aliasing `--output-file'.
 835 \f
 836 Changes in version 1.30, 2001-10-26:
 837
 838 * `--defines' and `--graph' have now an optional argument which is the
 839   output file name. `-d' and `-g' do not change; they do not take any
 840   argument.
 841
 842 * `%source_extension' and `%header_extension' are removed, failed
 843   experiment.
 844
 845 * Portability fixes.
 846 \f
 847 Changes in version 1.29, 2001-09-07:
 848
 849 * The output file does not define const, as this caused problems when used
 850   with common autoconfiguration schemes.  If you still use ancient compilers
 851   that lack const, compile with the equivalent of the C compiler option
 852   `-Dconst='.  autoconf's AC_C_CONST macro provides one way to do this.
 853
 854 * Added `-g' and `--graph'.
 855
 856 * The Bison manual is now distributed under the terms of the GNU FDL.
 857
 858 * The input and the output files has automatically a similar extension.
 859
 860 * Russian translation added.
 861
 862 * NLS support updated; should hopefully be less troublesome.
 863
 864 * Added the old Bison reference card.
 865
 866 * Added `--locations' and `%locations'.
 867
 868 * Added `-S' and `--skeleton'.
 869
 870 * `%raw', `-r', `--raw' is disabled.
 871
 872 * Special characters are escaped when output.  This solves the problems
 873   of the #line lines with path names including backslashes.
 874
 875 * New directives.
 876   `%yacc', `%fixed_output_files', `%defines', `%no_parser', `%verbose',
 877   `%debug', `%source_extension' and `%header_extension'.
 878
 879 * @$
 880   Automatic location tracking.
 881 \f
 882 Changes in version 1.28, 1999-07-06:
 883
 884 * Should compile better now with K&R compilers.
 885
 886 * Added NLS.
 887
 888 * Fixed a problem with escaping the double quote character.
 889
 890 * There is now a FAQ.
 891 \f
 892 Changes in version 1.27:
 893
 894 * The make rule which prevented bison.simple from being created on
 895   some systems has been fixed.
 896 \f
 897 Changes in version 1.26:
 898
 899 * Bison now uses automake.
 900
 901 * New mailing lists: <bug-bison@gnu.org> and <help-bison@gnu.org>.
 902
 903 * Token numbers now start at 257 as previously documented, not 258.
 904
 905 * Bison honors the TMPDIR environment variable.
 906
 907 * A couple of buffer overruns have been fixed.
 908
 909 * Problems when closing files should now be reported.
 910
 911 * Generated parsers should now work even on operating systems which do
 912   not provide alloca().
 913 \f
 914 Changes in version 1.25, 1995-10-16:
 915
 916 * Errors in the input grammar are not fatal; Bison keeps reading
 917 the grammar file, and reports all the errors found in it.
 918
 919 * Tokens can now be specified as multiple-character strings: for
 920 example, you could use "<=" for a token which looks like <=, instead
 921 of chosing a name like LESSEQ.
 922
 923 * The %token_table declaration says to write a table of tokens (names
 924 and numbers) into the parser file.  The yylex function can use this
 925 table to recognize multiple-character string tokens, or for other
 926 purposes.
 927
 928 * The %no_lines declaration says not to generate any #line preprocessor
 929 directives in the parser file.
 930
 931 * The %raw declaration says to use internal Bison token numbers, not
 932 Yacc-compatible token numbers, when token names are defined as macros.
 933
 934 * The --no-parser option produces the parser tables without including
 935 the parser engine; a project can now use its own parser engine.
 936 The actions go into a separate file called NAME.act, in the form of
 937 a switch statement body.
 938 \f
 939 Changes in version 1.23:
 940
 941 The user can define YYPARSE_PARAM as the name of an argument to be
 942 passed into yyparse.  The argument should have type void *.  It should
 943 actually point to an object.  Grammar actions can access the variable
 944 by casting it to the proper pointer type.
 945
 946 Line numbers in output file corrected.
 947 \f
 948 Changes in version 1.22:
 949
 950 --help option added.
 951 \f
 952 Changes in version 1.20:
 953
 954 Output file does not redefine const for C++.
 955
 956 Local Variables:
 957 mode: outline
 958 End:
 959
 960 -----
 961
 962 Copyright (C) 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003,
 963 2004, 2005, 2006 Free Software Foundation, Inc.
 964
 965 This file is part of Bison, the GNU Compiler Compiler.
 966
 967 Bison is free software; you can redistribute it and/or modify
 968 it under the terms of the GNU General Public License as published by
 969 the Free Software Foundation; either version 2, or (at your option)
 970 any later version.
 971
 972 Bison is distributed in the hope that it will be useful,
 973 but WITHOUT ANY WARRANTY; without even the implied warranty of
 974 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 975 GNU General Public License for more details.
 976
 977 You should have received a copy of the GNU General Public License
 978 along with autoconf; see the file COPYING.  If not, write to
 979 the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
 980 Boston, MA 02110-1301, USA.