3 There is no test with "Parse on stack %ld rejected by rule %d" in it.
6 Clarify that rule numbers in the skeletons are 1-based.
9 There are many macros that should obey api.prefix: YY_CPLUSPLUS, YY_MOVE,
12 ** YYDEBUG etc. in C++
13 Discourage the use of YYDEBUG in C++ (see thread with Jot). Stop supporting
14 #define YYSTYPE by the user.
16 Add value_type as a synonym for semantic_type.
19 In glr_state, we have yysval and yylloc. It should be yyval/yyloc (and
20 yylval/yylloc when referring to the lookahead). glr.c should
25 yystack.yyglrShift (create_state_set_index(0), 0, 0, yylval, &yylloc);
27 Why are yylval and yylloc treated differently?
30 And add tests in calc.at, to prepare work for D.
32 ** YYERROR and yynerrs
33 We are missing some cases. Write a test case, and check all the skeletons.
37 Don't do this (counterexample.c):
39 // This is the fastest way to get the tail node from the gl_list API.
41 list_get_end (gl_list_t list)
43 gl_list_node_t sentinel = gl_list_add_last (list, NULL);
44 gl_list_node_t res = gl_list_previous_node (list, sentinel);
45 gl_list_remove_node (list, sentinel);
49 *** Ambiguous rewriting
50 If the user is stupid enough to have equal rules, then the derivations are
53 Reduce/reduce conflict on tokens $end, "+", "⊕":
57 First derivation exp ::=[ exp "+" exp • ]
59 Second derivation exp ::=[ exp "+" exp • ]
61 Do we care about this? In color, we use twice the same color here, but we
62 could try to use the same color for the same rule.
65 Show the counterexamples. This is going to be really hard and/or painful.
66 Unless we play it dumb (little structure).
69 - How about not evaluating incomplete lines when the text is not finished
74 - Should i18n be part of the Lexer? Currently it's a static method of
77 - is there a migration path that would allow to use TokenKinds in
80 - define the tokens as an enum too.
82 - promote YYEOF rather than EOF.
85 https://git.savannah.gnu.org/gitweb/?p=gettext.git;a=blob;f=gettext-runtime/intl/plural.y;h=a712255af4f2f739c93336d4ff6556d932a426a5;hb=HEAD
87 should be updated to not use YYERRCODE. Returning an undef token is good
92 Stop hard-coding "Calc". Adjust local.at (look for FIXME).
94 ** A dev warning for b4_
95 Maybe we should check for m4_ and b4_ leaking out of the m4 processing, as
96 Autoconf does. It would have caught over-quotation issues.
99 I feel it's ugly to use the GNU style to declare functions in the doc. It
100 generates tons of white space in the page, and may contribute to bad page
107 The YYUNDEFTOK could be assigned a semantic value so that yyerror could be
108 used to report invalid lexemes.
111 Consider deprecating impure push parsers. They add a lot of complexity, for
112 a bad feature. On the other hand, that would make it much harder to sit
113 push parsers on top of pull parser. Which is currently not relevant, since
114 push parsers are measurably slower.
116 ** %define parse.error formatted
117 How about pushing Bistromathic's yyreport_syntax_error as another standard
118 way to generate the error message, and leave to the user the task of
119 providing the message formats? Currently in bistro, it reads:
122 error_format_string (int argc)
126 default: /* Avoid compiler warnings. */
127 case 0: return _("%@: syntax error");
128 case 1: return _("%@: syntax error: unexpected %u");
129 // TRANSLATORS: '%@' is a location in a file, '%u' is an
130 // "unexpected token", and '%0e', '%1e'... are expected tokens
133 // For instance on the expression "1 + * 2", you'd get
135 // 1.5: syntax error: expected - or ( or number or function or variable before *
136 case 2: return _("%@: syntax error: expected %0e before %u");
137 case 3: return _("%@: syntax error: expected %0e or %1e before %u");
138 case 4: return _("%@: syntax error: expected %0e or %1e or %2e before %u");
139 case 5: return _("%@: syntax error: expected %0e or %1e or %2e or %3e before %u");
140 case 6: return _("%@: syntax error: expected %0e or %1e or %2e or %3e or %4e before %u");
141 case 7: return _("%@: syntax error: expected %0e or %1e or %2e or %3e or %4e or %5e before %u");
142 case 8: return _("%@: syntax error: expected %0e or %1e or %2e or %3e or %4e or %5e or %6e before %u");
146 The message would have to be generated in a string, and pushed to yyerror.
147 Which will be a pain in the neck in yacc.c.
149 If we want to do that, we should think very carefully about the syntax of
152 ** yyclearin does not invoke the lookahead token's %destructor
153 https://lists.gnu.org/r/bug-bison/2018-02/msg00000.html
156 > Modifying yyclearin so that it calls yydestruct seems like the simplest
157 > solution to this issue, but it is conceivable that such a change would
158 > break programs which already perform some kind of workaround in order to
159 > destruct the lookahead symbol. So it might be necessary to use some kind of
160 > compatibility %define, or to create a new replacement macro with a
161 > different name such as yydiscardin.
163 > At a minimum, the fact that yyclearin does not invoke the %destructor
164 > should be highlighted in the documentation, since it is not at all obvious.
168 Les catégories d'avertissements incluent :
169 conflicts-sr conflits S/R (activé par défaut)
170 conflicts-rr conflits R/R (activé par défaut)
171 dangling-alias l'alias chaîne n'est pas attaché à un symbole
172 deprecated construction obsolète
173 empty-rule règle vide sans %empty
174 midrule-values valeurs de règle intermédiaire non définies ou inutilisées
175 precedence priorité et associativité inutiles
176 yacc incompatibilités avec POSIX Yacc
177 other tous les autres avertissements (activé par défaut)
178 all tous les avertissements sauf « dangling-alias » et « yacc »
179 no-CATEGORY désactiver les avertissements dans CATEGORIE
180 none désactiver tous les avertissements
181 error[=CATEGORY] traiter les avertissements comme des erreurs
183 Line -1 and -3 should mention CATEGORIE, not CATEGORY.
186 ** Rewrite glr.cc (currently glr2.cc)
188 We can probably replace setjmp/longjmp with exceptions. That would help
189 tremendously other languages such as D and Java that probably have no
190 similar feature. If we remove jumps, we probably no longer need _Noreturn,
191 so simplify `b4_attribute_define([noreturn])` into `b4_attribute_define`.
193 After discussing with Valentin, it was decided that it's better to stay with
194 jumps, since in some places exceptions are ruled out from C++.
197 Move to our coding conventions. In particular names such as yy_glr_stack,
201 It should be a member of the parser object, see lalr1.cc. Let the parser
202 object decide what the debug stream is, rather than open coding std::cerr.
203 And use YYCDEBUG rather than YY_DEBUG_STREAM.
206 There are many places where pointers should be replaced with references.
207 Some occurrences were fixed, but now some have improper names:
209 -yygetToken (int *yycharp, ]b4_namespace_ref[::]b4_parser_class[& yyparser][]b4_pure_if([, glr_stack* yystackp])[]b4_user_formals[)
210 +yygetToken (int& yycharp, ]b4_namespace_ref[::]b4_parser_class[& yyparser][]b4_pure_if([, glr_stack* yystackp])[]b4_user_formals[)
212 yycharp is no longer a Pointer. And yystackp should probably also be a reference.
214 *** Use proper type names
215 Don't use YYSTYPE and YYLTYPE but parser::semantic_type and
216 parser::location_type. Undefine YYSTYPE and YYLTYPE.
219 Currently all the assertions are enabled. Once we are confident in glr2.cc,
220 let parse.assert use the same approach as in lalr1.cc.
223 Stop using std::cerr everywhere.
226 When glr2.cc fully replaces glr.cc, get rid of the glr.cc scaffolding in
230 ** Unit rules / Injection rules (Akim Demaille)
231 Maybe we could expand unit rules (or "injections", see
232 https://homepages.cwi.nl/~daybuild/daily-books/syntax/2-sdf/sdf.html), i.e.,
241 exp: exp '+' exp | exp '&' exp;
243 when there are no actions. This can significantly speed up some grammars.
244 I can't find the papers. In particular the book 'LR parsing: Theory and
245 Practice' is impossible to find, but according to 'Parsing Techniques: a
246 Practical Guide', it includes information about this issue. Does anybody
249 ** clean up (Akim Demaille)
250 Do not work on these items now, as I (Akim) have branches with a lot of
251 changes in this area (hitting several files), and no desire to have to fix
252 conflicts. Addressing these items will happen after my branches have been
256 Introduce a goto struct, and use it in place of from_state/to_state.
257 Rename states1 as path, length as pathlen.
258 Introduce inline functions for things such as nullable[*rp - ntokens]
259 where we need to map from symbol number to nterm number.
261 There are probably a significant part of the relations management that
262 should be migrated on top of a bitsetv.
265 It should probably take a "state*" instead of two arguments.
268 The "automaton" and "set" categories are not so useful. We should probably
269 introduce lr(0) and lalr, just the way we have ielr categories. The
270 "closure" function is too verbose, it should probably have its own category.
272 "set" can still be used for summarizing the important sets. That would make
273 tests easy to maintain.
276 Rename these guys as "diagnostics.*" (or "diagnose.*"), since that's the
277 name they have in GCC, clang, etc. Likewise for the complain_* series of
281 states/nstates, rules/nrules, ..., ritem/nritems
284 *** m4: slot, type, type_tag
285 The meaning of type_tag varies depending on api.value.type. We should avoid
286 that and using clear definitions with stable semantics.
288 * D programming language
289 There's a number of features that are missing, here sorted in _suggested_
290 order of implementation.
292 When copying code from other skeletons, keep the comments exactly as they
293 are. Keep the same variable names. If you change the wording in one place,
294 do it in the others too. In other words: make sure to keep the
295 maintenance *simple* by avoiding any gratuitous difference.
298 Check when gdc and ldc.
300 ** Token Constructors
301 It is possible to mix incorrectly kinds and values, and for instance:
303 return parser.Symbol (TokenKind.NUM, "Hello, World!\n");
305 attaches a string value to NUM kind (wrong, of course). When
306 api.token.constructor is set, in C++, Bison generated "token constructors":
307 parser.make_NUM. parser.make_PLUS, parser.make_STRING, etc. The previous
310 return parser.make_NUM ("Hello, World!\n");
312 which would easily be caught by the type checker.
315 Add support for push parser. Do not start a nice skeleton, just enhance the
316 current one to support push parsers. This is going to be a tougher nut to
319 First, you need to understand well how the push parser is expected to work.
322 - look at examples/c/pushcalc
323 - create an example of a Java push parser.
324 - have a look at the generated parser in Java, which has the advantage of
325 being already based on a parser object, instead of just a function.
327 The C case is harder to read, but it may help too. Keep in mind that
328 because there's no object to maintain state, the C push parser uses some
329 struct (yypstate) to preserve this state. We don't need this in D, the
330 parser object will suffice.
332 I think working directly on the skeleton to add push-parser support is not
333 the simplest path. I suggest that you (1) transform a generated parser into
334 a push parser by hand, and then (2) transform lalr1.d to generate such a
337 Use `git commit` frequently to make sure you keep track of your progress.
339 *** (1.a) Prepare pull parser by hand
340 Copy again one of the D examples into say examples/d/pushcalc. Also
341 check-in the generated parser to facilitate experimentation.
343 - find local variables of yyparse should become members of the parser object
344 (so that we preserve state from one call to the next).
346 - do it in your generated D parser. We don't need an equivalent for
347 yypstate, because we already have it: that the parser object itself.
349 - have your *pull*-parser (i.e., the good old yy::parser::parse()) work
350 properly this way. Write and run tests. That's one of the reasons I
351 suggest using examples/d/calc as a starting point: it already has tests,
352 you can/should add more.
354 At this point you have a pull-parser which you prepared to turn into a
357 *** (1.b) Turn pull parser into push parser by hand
359 - look again at how push parsers are implemented in Java/C to see what needs
360 to change in yyparse so that the control is inverted: parse() will
361 be *given* the tokens, instead of having to call yylex itself. When I say
362 "look at C", I think your best option are (i) yacc.c (look for b4_push_if)
363 and (ii) examples/c/pushcalc.
365 - rename parse() as push_parse(Symbol yyla) (or push_parse(TokenKind, Value,
366 Location)) that takes the symbol as argument. That's the push parser we
369 - define a new parse() function which has the same signature as the usual
370 pull-parser, that repeatedly calls the push_parse function. Something
377 status = this->push_parse (yylex());
378 } while (status == YYPUSH_MORE);
382 - show me that parser, so that we can validate the approach.
384 *** (2) Port that into the skeleton
385 - once we agree on the API of the push parser, implement it into lalr1.d.
386 You will probaby need help on this regard, but imitation, again, should
389 - have example/d/pushcalc work properly and pass tests
391 - add tests in the "real" test suite. Do that in tests/calc.at. I can
397 This is very ambitious. That's the final boss. There are currently no
398 "clean" implementation to get inspiration from.
400 glr.c is very clean but:
402 - is a different skeleton from yacc.c
404 glr.cc is (currently) an ugly hack: a C++ shell around glr.c. Valentin
405 Tolmer is currently rewriting glr.cc to be clean C++, but he is not
406 finished. There will be a lot a common code between lalr1.cc and glr.cc, so
407 eventually I would like them to be fused into a single skeleton, supporting
408 both deterministic and generalized parsing.
410 It would be great for D to also support this.
412 The basic ideas of GLR are explained here:
414 https://www.codeproject.com/Articles/5259825/GLR-Parsing-in-Csharp-How-to-Use-The-Most-Powerful
416 * Better error messages
417 The users are not provided with enough tools to forge their error messages.
418 See for instance "Is there an option to change the message produced by
419 YYERROR_VERBOSE?" by Simon Sobisch, on bison-help.
422 https://www.cs.tufts.edu/~nr/cs257/archive/clinton-jefferey/lr-error-messages.pdf
423 https://research.swtch.com/yyerror
424 http://gallium.inria.fr/~fpottier/publis/fpottier-reachability-cc2016.pdf
427 Fix data/skeletons/yacc.c so that it defines YYPTRDIFF_T properly for modern
428 and older C++ compilers. Currently the code defaults to defining it to
429 'long' for non-GCC compilers, but it should use the proper C++ magic to
430 define it to the same type as the C ptrdiff_t type.
433 Several features are not available in all the back-ends.
435 - push parsers: glr.c, glr.cc, lalr1.cc (not very difficult)
436 - token constructors: Java, C, D (a bit difficult)
437 - glr: D, Java (super difficult)
440 ** Autotest has quotation issues
441 tests/input.at:1730:AT_SETUP([%define errors])
445 $ ./tests/testsuite -l | grep errors | sed q
446 38: input.at:1730 errors
449 ** Get rid of YYPRINT and b4_toknum
450 Besides yytoknum is wrong when api.token.raw is defined.
452 ** Better design for diagnostics
453 The current implementation of diagnostics is ad hoc, it grew organically.
454 It works as a series of calls to several functions, with dependency of the
455 latter calls on the former. For instance:
457 complain (&sym->location,
458 sym->content->status == needed ? complaint : Wother,
459 _("symbol %s is used, but is not defined as a token"
460 " and has no rules; did you mean %s?"),
461 quote_n (0, sym->tag),
462 quote_n (1, best->tag));
463 if (feature_flag & feature_caret)
464 location_caret_suggestion (sym->location, best->tag, stderr);
466 We should rewrite this in a more FP way:
468 1. build a rich structure that denotes the (complete) diagnostic.
469 "Complete" in the sense that it also contains the suggestions, the list
470 of possible matches, etc.
472 2. send this to the pretty-printing routine. The diagnostic structure
473 should be sufficient so that we can generate all the 'format' of
474 diagnostics, including the fixits.
476 If properly done, this diagnostic module can be detached from Bison and be
477 put in gnulib. It could be used, for instance, for errors caught by
480 There's certainly already something alike in GCC. At least that's the
481 impression I get from reading the "-fdiagnostics-format=FORMAT" part of this
484 https://gcc.gnu.org/onlinedocs/gcc/Diagnostic-Message-Formatting-Options.html
486 ** Graphviz display code thoughts
487 The code for the --graph option is over two files: print_graph, and
488 graphviz. This is because Bison used to also produce VCG graphs, but since
489 this is no longer true, maybe we could consider these files for fusion.
491 An other consideration worth noting is that print_graph.c (correct me if I
492 am wrong) should contain generic functions, whereas graphviz.c and other
493 potential files should contain just the specific code for that output
494 format. It will probably prove difficult to tell if the implementation is
495 actually generic whilst only having support for a single format, but it
496 would be nice to keep stuff a bit tidier: right now, the construction of the
497 bitset used to show reductions is in the graphviz-specific code, and on the
498 opposite side we have some use of \l, which is graphviz-specific, in what
499 should be generic code.
501 Little effort seems to have been given to factoring these files and their
502 print{,-xml} counterpart. We would very much like to re-use the pretty format
503 of states from .output for the graphs, etc.
505 Since graphviz dies on medium-to-big grammars, maybe consider an other tool?
508 Check it too when checking the different kinds of parsers. And be
509 sure to check that the initial-action is performed once per parsing.
512 b4_shared_declarations is no longer what it is. Make it
513 b4_parser_declaration for instance.
515 ** yychar in lalr1.cc
516 There is a large difference bw maint and master on the handling of
517 yychar (which was removed in lalr1.cc). See what needs to be
521 /* User semantic actions sometimes alter yychar, and that requires
522 that yytoken be updated with the new translation. We take the
523 approach of translating immediately before every use of yytoken.
524 One alternative is translating here after every semantic action,
525 but that translation would be missed if the semantic action
526 invokes YYABORT, YYACCEPT, or YYERROR immediately after altering
527 yychar. In the case of YYABORT or YYACCEPT, an incorrect
528 destructor might then be invoked immediately. In the case of
529 YYERROR, subsequent parser actions might lead to an incorrect
530 destructor call or verbose syntax error message before the
531 lookahead is translated. */
533 /* Make sure we have latest lookahead translation. See comments at
534 user semantic actions for why this is necessary. */
535 yytoken = yytranslate_ (yychar);
538 ** Get rid of fake #lines [Bison: ...]
539 Possibly as simple as checking whether the column number is nonnegative.
541 I have seen messages like the following from GCC.
543 <built-in>:0: fatal error: opening dependency file .deps/libltdl/argz.Tpo: No such file or directory
546 ** Discuss about %printer/%destroy in the case of C++.
547 It would be very nice to provide the symbol classes with an operator<<
548 and a destructor. Unfortunately the syntax we have chosen for
549 %destroy and %printer make them hard to reuse. For instance, the user
550 is invited to write something like
552 %printer { debug_stream() << $$; } <my_type>;
554 which is hard to reuse elsewhere since it wants to use
555 "debug_stream()" to find the stream to use. The same applies to
556 %destroy: we told the user she could use the members of the Parser
557 class in the printers/destructors, which is not good for an operator<<
558 since it is no longer bound to a particular parser, it's just a
562 ** Rewrite glr.cc in C++ (Valentin Tolmer)
563 As a matter of fact, it would be very interesting to see how much we can
564 share between lalr1.cc and glr.cc. Most of the skeletons should be common.
565 It would be a very nice source of inspiration for the other languages.
567 Valentin Tolmer is working on this.
569 * From lalr1.cc to yacc.c
571 Merging the three stacks in lalr1.cc simplified the code, prompted for
572 other improvements and also made it faster (probably because memory
573 management is performed once instead of three times). I suggest that
574 we do the same in yacc.c.
576 (Some time later): it's also very nice to have three stacks: it's more dense
577 as we don't lose bits to padding. For instance the typical stack for states
578 will use 8 bits, while it is likely to consume 32 bits in a struct.
580 We need trustworthy benchmarks for Bison, for all our backends. Akim has a
581 few things scattered around; we need to put them in the repo, and make them
587 Some statistics about the grammar and the parser would be useful,
588 especially when asking the user to send some information about the
589 grammars she is working on. We should probably also include some
590 information about the variables (I'm not sure for instance we even
591 specify what LR variant was used).
594 How would Paul like to display the conflicted actions? In particular,
595 what when two reductions are possible on a given lookahead token, but one is
596 part of $default. Should we make the two reductions explicit, or just
597 keep $default? See the following point.
599 ** Disabled Reductions
600 See 'tests/conflicts.at (Defaulted Conflicted Reduction)', and decide
604 Extend with error productions. The hard part will probably be finding
605 the right rule so that a single state does not exhibit too many yet
606 undocumented ''features''. Maybe an empty action ought to be
607 presented too. Shall we try to make a single grammar with all these
608 features, or should we have several very small grammars?
612 Well, only if there is really some demand for it.
615 https://github.com/scfc/bison-php/blob/master/data/lalr1.php
618 https://lists.gnu.org/r/bison-patches/2013-09/msg00000.html and following
620 ** Multiple start symbols
621 Would be very useful when parsing closely related languages. The idea is to
622 declare several start symbols, for instance
629 and to generate parse(), parse_stmt() and parse_expr(). Technically, the
630 above grammar would be transformed into
633 %token YY_START_STMT YY_START_EXPR
635 yy_start: YY_START_STMT stmt | YY_START_EXPR expr
637 so that there are no new conflicts in the grammar (as would undoubtedly
638 happen with yy_start: stmt | expr). Then adjust the skeletons so that this
639 initial token (YY_START_STMT, YY_START_EXPR) be shifted first in the
640 corresponding parse function.
642 *** Number of useless symbols
646 [[input.y: warning: 2 nonterminals useless in grammar [-Wother]
647 input.y: warning: 2 rules useless in grammar [-Wother]
648 input.y:2.8-10: error: start symbol exp does not derive any sentence]])
650 We should say "1 nonterminal": the other one is $accept, which should not
651 participate in the count.
654 Do we want to disallow terminal start symbols? The limitation is not
655 technical. Can it be useful to someone to "parse" a token?
658 This is a popular demand. We already made many changes in the parser that
659 should make this reasonably easy to implement.
661 Bruce Mardle <marblypup@yahoo.co.uk>
662 https://lists.gnu.org/archive/html/bison-patches/2015-09/msg00000.html
664 However, there are many other things to do before having such a feature,
665 because I don't want a % equivalent to #include (which we all learned to
666 hate). I want something that builds "modules" of grammars, and assembles
667 them together, paying attention to keep separate bits separated, in pseudo
671 There is demand for push parsers in C++.
673 ** Generate code instead of tables
674 This is certainly quite a lot of work. See
675 http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.50.4539.
678 We should find a means to provide an access to values deep in the
679 stack. For instance, instead of
681 baz: qux { $$ = $<foo>-1 + $<bar>0 + $1; }
683 we should be able to have:
685 foo($foo) bar($bar) baz($bar): qux($qux) { $baz = $foo + $bar + $qux; }
687 Or something like this.
690 It should be possible to have %if/%else/%endif. The implementation is
691 not clear: should it be lexical or syntactic. Vadim Maslow thinks it
692 must be in the scanner: we must not parse what is in a switched off
693 part of %if. Akim Demaille thinks it should be in the parser, so as
694 to avoid falling into another CPP mistake.
696 (Later): I'm sure there's actually good case for this. People who need that
697 feature can use m4/cpp on top of Bison. I don't think it is worth the
698 trouble in Bison itself.
701 There are couple of available extensions of Bison targeting some XML
702 output. Some day we should consider including them. One issue is
703 that they seem to be quite orthogonal to the parsing technique, and
704 seem to depend mostly on the possibility to have some code triggered
705 for each reduction. As a matter of fact, such hooks could also be
706 used to generate the yydebug traces. Some generic scheme probably
709 XML output for GNU Bison and gcc
710 http://www.cs.may.ie/~jpower/Research/bisonXML/
712 XML output for GNU Bison
713 http://yaxx.sourceforge.net/
715 * Coding system independence
718 Currently Bison assumes 8-bit bytes (i.e. that UCHAR_MAX is
719 255). It also assumes that the 8-bit character encoding is
720 the same for the invocation of 'bison' as it is for the
721 invocation of 'cc', but this is not necessarily true when
722 people run bison on an ASCII host and then use cc on an EBCDIC
723 host. I don't think these topics are worth our time
724 addressing (unless we find a gung-ho volunteer for EBCDIC or
725 PDP-10 ports :-) but they should probably be documented
728 More importantly, Bison does not currently allow NUL bytes in
729 tokens, either via escapes (e.g., "x\0y") or via a NUL byte in
730 the source code. This should get fixed.
735 Must we keep %token-table?
740 It is unfortunate that there is a total order for precedence. It
741 makes it impossible to have modular precedence information. We should
742 move to partial orders (sounds like series/parallel orders to me).
744 This is a prerequisite for modules.
746 * Pre and post actions.
747 From: Florian Krohm <florian@edamail.fishkill.ibm.com>
748 Subject: YYACT_EPILOGUE
749 To: bug-bison@gnu.org
750 X-Sent: 1 week, 4 days, 14 hours, 38 minutes, 11 seconds ago
752 The other day I had the need for explicitly building the parse tree. I
753 used %locations for that and defined YYLLOC_DEFAULT to call a function
754 that returns the tree node for the production. Easy. But I also needed
755 to assign the S-attribute to the tree node. That cannot be done in
756 YYLLOC_DEFAULT, because it is invoked before the action is executed.
757 The way I solved this was to define a macro YYACT_EPILOGUE that would
758 be invoked after the action. For reasons of symmetry I also added
759 YYACT_PROLOGUE. Although I had no use for that I can envision how it
760 might come in handy for debugging purposes.
761 All is needed is to add
764 YYACT_EPILOGUE (yyval, (yyvsp - yylen), yylen, yyloc, (yylsp - yylen));
766 YYACT_EPILOGUE (yyval, (yyvsp - yylen), yylen);
769 at the proper place to bison.simple. Ditto for YYACT_PROLOGUE.
771 I was wondering what you think about adding YYACT_PROLOGUE/EPILOGUE
772 to bison. If you're interested, I'll work on a patch.
775 Equip the parser with a means to create the (visual) parse tree.
780 # LocalWords: Cex gnulib gl Bistromathic TokenKinds yylex enum YYEOF EOF
781 # LocalWords: YYerror gettext af hb YYERRCODE undef calc FIXME dev yyerror
782 # LocalWords: Autoconf YYUNDEFTOK lexemes parsers Bistromathic's yyreport
783 # LocalWords: const argc yacc yyclearin lookahead destructor Rici incluent
784 # LocalWords: yydestruct yydiscardin catégories d'avertissements sr activé
785 # LocalWords: conflits défaut rr l'alias chaîne n'est attaché un symbole
786 # LocalWords: obsolète règle vide midrule valeurs de intermédiaire ou avec
787 # LocalWords: définies inutilisées priorité associativité inutiles POSIX
788 # LocalWords: incompatibilités tous les autres avertissements sauf dans rp
789 # LocalWords: désactiver CATEGORIE traiter comme des erreurs glr Akim bool
790 # LocalWords: Demaille arith lalr goto struct pathlen nullable ntokens lr
791 # LocalWords: nterm bitsetv ielr ritem nstates nrules nritems yysymbol EQ
792 # LocalWords: SymbolKind YYEMPTY YYUNDEF YYTNAME NUM yyntokens yytname sed
793 # LocalWords: nonterminals yykind yycode YYNAMES yynames init getName conv
794 # LocalWords: TokenKind semanticVal ival yychar yylval yylexer Tolmer hoc
795 # LocalWords: Sobisch YYPTRDIFF ptrdiff Autotest YYPRINT toknum yytoknum
796 # LocalWords: sym Wother stderr FP fixits xgettext fdiagnostics Graphviz
797 # LocalWords: graphviz VCG bitset xml bw maint yytoken YYABORT deps
798 # LocalWords: YYACCEPT yytranslate nonnegative destructors yyerrlab repo
799 # LocalWords: backends stmt expr yy Mardle baz qux Vadim Maslow CPP cpp
800 # LocalWords: yydebug gcc UCHAR EBCDIC gung PDP NUL Pre Florian Krohm utf
801 # LocalWords: YYACT YYLLOC YYLSP yyval yyvsp yylen yyloc yylsp endif
802 # LocalWords: ispell american
808 ispell-dictionary: "american"
811 Copyright (C) 2001-2004, 2006, 2008-2015, 2018-2020 Free Software
814 This file is part of Bison, the GNU Compiler Compiler.
816 Permission is granted to copy, distribute and/or modify this document
817 under the terms of the GNU Free Documentation License, Version 1.3 or
818 any later version published by the Free Software Foundation; with no
819 Invariant Sections, with no Front-Cover Texts, and with no Back-Cover
820 Texts. A copy of the license is included in the "GNU Free
821 Documentation License" file as part of this distribution.