The problem was that yacc.c didn't always update the yychar
translation afterwards. However, other skeletons appear to be
fine. glr.c appears to already translate yychar before every
use. lalr1.cc does not define yychar and does not document its
replacement, yyla, for users. It does provide yyclearin, but
that does not manipulate yyla and thus requires no translation
update. In lalr1.java, yychar is out of scope during semantic
actions.
* NEWS (2.5): Document.
* data/yacc.c (YYBACKUP): Don't bother translating yychar into
yytoken here.
(yyparse, yypush_parse): Instead, translate before every use of
yytoken, and add comments explaining this approach.
* tests/actions.at (Destroying lookahead assigned by semantic
action): New test group checking that translation happens before
lookahead destructor calls at parser return. Previously,
incorrect destructors were called.
* tests/conflicts.at (%error-verbose and consistent
errors): New test group checking that translation happens at
syntax error detection before the associated verbose error
message and the associated lookahead destructor calls. While
the destructor call is fixed by this patch, the verbose error
message is currently incorrect due to another bug (see
comments in test group), so this is an expected failure for now.
(cherry picked from commit df222dfaf4)
Conflicts:
src/parse-gram.c
src/parse-gram.h
The problem was that yacc.c didn't always update the yychar
translation afterwards. However, other skeletons appear to be
fine. glr.c appears to already translate yychar before every
use. lalr1.cc does not define yychar and does not document its
replacement, yyla, for users. It does provide yyclearin, but
that does not manipulate yyla and thus requires no translation
update. In lalr1.java, yychar is out of scope during semantic
actions.
* NEWS (2.5): Document.
* data/yacc.c (YYBACKUP): Don't bother translating yychar into
yytoken here.
(yyparse, yypush_parse): Instead, translate before every use of
yytoken, and add comments explaining this approach.
* tests/actions.at (Destroying lookahead assigned by semantic
action): New test group checking that translation happens before
lookahead destructor calls at parser return. Previously,
incorrect destructors were called.
* tests/conflicts.at (parse.error=verbose and consistent
errors): New test group checking that translation happens at
syntax error detection before the associated verbose error
message and the associated lookahead destructor calls. While
the destructor call is fixed by this patch, the verbose error
message is currently incorrect due to another bug (see
comments in test group), so this is an expected failure for now.
* tests/conflicts.at (%nonassoc and eof): Extend to test that it
prevents the omission of expected tokens for %error-verbose.
(cherry picked from commit d1cc31c5f0)
For now, just api.push-pull and lr.keep-unreachable-states.
Maintain old names for backward compatibility.
* NEWS (2.5): Document.
* data/c.m4 (b4_identification): Update comment.
* data/yacc.c: Update access.
* doc/bison.texinfo: Update.
* etc/bench.pl.in (bench_push_parser): Update use.
* src/files.c (tr): Move to...
* src/getargs.c, src/getargs.h (tr): ... here because I can't
think of a better place to expose it. My logic is that, for all
uses of tr so far, command-line arguments can be involved, and
getargs.h is already included.
* src/main.c (main): Update access.
* src/muscle_tab.c (muscle_percent_define_insert): Convert old
variable names to new variable names before assigning value.
* src/reader.c (reader): Update setting default.
* tests/calc.at: Update uses.
* tests/conflicts.at (Unreachable States After Conflict
Resolution): Update use.
* tests/input.at (%define enum variables): Update use.
(%define backward compatibility): New test group.
* tests/push.at: Update uses.
* tests/reduce.at: Update uses.
* tests/torture.at: Update uses.
(cherry picked from commit 812775a039)
Conflicts:
data/c.m4
etc/bench.pl.in
src/parse-gram.c
src/parse-gram.h
tests/conflicts.at
For now, just api.push-pull and lr.keep-unreachable-states.
Maintain old names for backward compatibility.
* NEWS (2.5): Document.
* data/c.m4 (b4_identification): Update comment.
* data/yacc.c: Update access.
* doc/bison.texinfo: Update.
* etc/bench.pl.in (bench_grammar): Update use.
* src/files.c (tr): Move to...
* src/getargs.c, src/getargs.h (tr): ... here because I can't
think of a better place to expose it. My logic is that, for all
uses of tr so far, command-line arguments can be involved, and
getargs.h is already included.
* src/main.c (main): Update access.
* src/muscle_tab.c (muscle_percent_define_insert): Convert old
variable names to new variable names before assigning value.
* src/reader.c (reader): Update setting default.
* tests/calc.at: Update uses.
* tests/conflicts.at (Unreachable States After Conflict
Resolution): Update use.
* tests/input.at (%define enum variables): Update use.
(%define backward compatibility): New test group.
* tests/push.at: Update uses.
* tests/reduce.at: Update uses.
* tests/torture.at: Update uses.
Unfortunately it is not possible to reuse the %prec directive. This
is because to please POSIX, we do not require to end the rules with a
semicolon. As a result,
foo: bar %prec baz
is ambiguous: either a rule which precedence is that of baz, or a rule,
and then a declaration of the precedence of the token baz.
* doc/bison.texinfo: Document %precedence.
(Precedence Only): New.
* src/assoc.h, src/assoc.c (precedence_assoc): New.
* src/conflicts.c (resolve_sr_conflict): Support it.
* src/scan-gram.l, src/parse-gram.y (%precedence): New token.
Parse it.
* tests/calc.at: Use %precedence for NEG.
* tests/conflicts.at (%precedence does not suffice)
(%precedence suffices): New tests.
starting at
<http://lists.gnu.org/archive/html/bison-patches/2007-11/msg00021.html>.
* configure.ac (XSLTPROC): New substitution.
* Makefile.am (maintainer-xml-check): New phony target invoking...
* tests/Makefile.am (maintainer-xml-check): ... this new phony target
invoking make maintainer-check with BISON_TEST_XML=1.
* tests/atlocal.in (XSLTPROC): New.
* tests/local.at (AT_BISON_CHECK): New macro to (1) instruct Valgrind
not to report reachable memory when Bison is expected to have a
non-zero exit status and (2) to compare XML/XSLT output with --graph
and --report=all output for every working grammar when
BISON_TEST_XML=1.
(AT_BISON_CHECK_NO_XML): Likewise, but skip XML checks.
(AT_BISON_CHECK_XML): New.
(AT_QUELL_VALGRIND): New.
* tests/testsuite.at (ORIGINAL_AT_CHECK): Remove this and...
(AT_CHECK): ... don't redefine this since this was the old way to
quell Valgrind.
* tests/actions.at: Rewrite all AT_CHECK invocations for bison as
AT_BISON_CHECK invocations.
* tests/c++.at: Likewise.
* tests/calc.at: Likewise.
* tests/conflicts.at: Likewise.
* tests/cxx-type.at: Likewise.
* tests/existing.at: Likewise.
* tests/glr-regression.at: Likewise.
* tests/headers.at: Likewise.
* tests/input.at: Likewise.
* tests/java.at: Likewise.
* tests/output.at: Likewise.
* tests/push.at: Likewise.
* tests/reduce.at: Likewise.
* tests/regression.at: Likewise.
* tests/sets.at: Likewise.
* tests/skeletons.at: Likewise.
* tests/synclines.at: Likewise.
* tests/torture.at: Likewise.
(Big triangle): Use AT_BISON_CHECK_NO_XML instead since this grammar
tends to hang xsltproc.
(Big horizontal): Likewise.
When reporting useless rules and nonterminals, say "useless in grammar"
instead of "useless", and say "useless in parser" instead of "never
reduced". Discussed starting at
<http://lists.gnu.org/archive/html/bison-patches/2007-10/msg00033.html>.
* NEWS (2.3a+): Mention this change.
* data/xslt/xml2text.xsl: Update output text and expected input XML
element names to match changes below.
* data/xslt/xml2xhtml.xsl: Likewise.
(xsl:template match="bison-xml-report"): Add missing entry in Table of
Contents: "Rules useless in parser due to conflicts".
* doc/bison.texinfo (Decl Summary): Reword a little.
(Understanding): Update example output for changes below.
* src/gram.c: (rule_useful_p): Rename to...
(rule_useful_in_grammar_p): ... this.
(rule_useless_p): Rename to...
(rule_useless_in_grammar_p): ... this.
(rule_never_reduced_p): Rename to...
(rule_useless_in_parser_p): ... this.
(grammar_rules_print): Update for renames.
(grammar_rules_print_xml): Update for renames.
(grammar_rules_never_reduced_report): Rename to...
(grammar_rules_useless_report): ... this since it is used for either
kind of useless rule.
* src/gram.h: Reword comments and update function names in prototypes.
* src/main.c (main): Say "rule useless in parser due to conflicts".
* src/print-xml.c (print_rules_never_reduced): Rename to...
(print_rules_useless_in_parser): ... this, and rename output XML
element "rules-never-reduced" to "rules-useless-in-parser".
(print_xml): Update for rename.
* src/print.c (print_results): Say "Rules useless in parser due to
conflicts".
* src/reduce.c (reduce_grammar_tables): Say "rule useless in grammar".
(nonterminals_reduce): Say "nonterminal useless in grammar".
(reduce_output): Say "Nonterminals useless in grammar".
Say "Rules useless in grammar".
(reduce_xml): Rename output XML element "useless" to
"useless-in-grammar".
(reduce_print): Don't report the count of grammatically useless rules
as "rules never reduced" just because %yacc is specified.
In the correct report of this count, say nonterminal(s) and rule(s)
"useless in grammar".
* tests/conflicts.at (S/R in initial): Update expected output.
(Defaulted Conflicted Reduction): Likewise.
(Unreachable States After Conflict Resolution): Likewise.
* tests/existing.at (GNU pic Grammar): Likewise.
* tests/reduce.at (Useless Nonterminals): Likewise.
(Useless Rules): Likewise.
(Reduced Automaton): Likewise.
(Underivable Rules): Likewise.
(Empty Language): Likewise.
* NEWS (2.3a+): Mention it in the entry for unreachable state removal.
* doc/bison.texinfo (Decl Summary): Mention it in the %define entry.
* src/main.c (main): Implement it.
* tests/conflicts.at (Unreachable States After Conflict Resolution):
Extend to test it, and fix a typo.
(Option Cross Key): Likewise.
* src/print-xml.c (print_core): Don't print a reduction's lookahead set
next to an item whose dot is not at the end of the RHS even if it
happens to be associated with the same rule.
* src/print.c (print_core): Likewise.
* tests/conflicts.at (Unresolved SR Conflicts): Update output.
(Resolved SR Conflicts): Update output.
* tests/regression.at (Extra lookahead sets in report): New test case.
conflicts involving multiple reductions.
* src/conflicts.c (resolve_sr_conflict): Don't assign the error action
set for a state here or Bison will abort if it is reassigned on a
later conflicted reduction in the same state.
Similarly, don't finalize and assign the solved conflicts report here
or it will be lost if it is reassigned on a later conflicted reduction
in the same state.
(set_conflicts): Instead, assign them both here after all S/R conflicts
in the state have been fully examined.
* src/print.c (shift_set): Rename to...
(no_reduce_set): ... this.
(print_reductions): Update for rename, and add %nonassoc error action
tokens to no_reduce_set so that, when printing the first remaining
reduction on an error action token, the reduction is enclosed in
brackets.
(print_results): Update for rename.
* tests/conflicts.at (Solved conflicts report for multiple reductions
in a state): New test case.
(%nonassoc error actions for multiple reductions in a state): New test
case.
* src/main.c (main): Don't depend on C99 features.
report rules that are then unused, and don't report conflicts in those
states.
* src/conflicts.c, src/conflicts.h (conflicts_update_state_numbers):
New global function.
* src/lalr.c, src/lalr.h (lalr_update_state_numbers): New global
function.
* src/main.c (main): After conflict resolution, remove the unreachable
states and update all data structures that reference states by number.
* src/state.c (state_new): Initialize each state's reachable member to
false.
(state_mark_reachable_states): New static function.
(state_remove_unreachable_states): New global function.
* src/state.h (struct state): Add member bool reachable.
(state_remove_unreachable_states): Prototype.
* tests/conflicts.at (Unreachable States After Conflict Resolution):
New test case.
* tests/existing.at (GNU pic Grammar): Update test case output now that
an unused rule is discovered.
<http://lists.gnu.org/archive/html/bug-bison/2006-10/msg00000.html>
and try to make such failures easier to catch in the future.
* data/glr.c (YYTRANSLATE): Don't check for nonpositive arg;
that's now the caller's responsibility.
(yyprocessOneStack, yyrecoverSyntaxError, yyparse):
Set yychar = YYEOF if it's negative.
* tests/actions.at (yylex): Abort if asked to read past EOF.
* tests/conflicts.at (yylex): Likewise.
* tests/cxx-type.at (yylex): Likewise.
* tests/glr-regression.at (yylex): Likewise.
* tests/input.at (yylex): Likewise.
* tests/regression.at (yylex): Likewise.
* tests/torture.at (yylex): Likewise.
when %expect fails because there were no conflicts.
* doc/bison.texinfo (Expect Decl): Tighten up wording in previous
change.
* tests/conflicts.at (%expect not enough, %expect too much):
(%expect with reduce conflicts): Adjust to new behavior.
* src/conflicts.c (conflicts_print): Unsatisfied %expectation are
errors.
* NEWS: Document this.
* doc/bison.texinfo (Expect Decl): Likewise.
1 on syntax error; just let yyparse do its thing.
* tests/glr-regression.at (Badly Collapsed GLR States): Likewise.
* tests/torture.at (AT_DATA_STACK_TORTURE): Likewise.
(Exploding the Stack Size with Alloca):
(Exploding the Stack Size with Malloc):
Expect exit status 2, not 1, since the parser is supposed to blow
its stack. Problem reported by twlevo@xs4all.nl.
to Bison 1.30 and 1.75 behavior. This fixes the GCC 3.2
bootstrapping problem reported by Matthias Klose; see
<http://mail.gnu.org/archive/html/bug-bison/2003-01/msg00053.html>.
* src/conflicts.c (conflicts_print): Likewise.
* tests/conflicts.at (%expect not enough, %expect too much,
%expect with reduce conflicts): Likewise.
* doc/bison.texinfo (Expect Decl): Document this. Also mention
that the warning is enabled if the number of conflicts changes
(not necessarily increases).
* tests/testsuite.at (AT_DATA_GRAMMAR_PROLOGUE, AT_DATA_GRAMMAR):
New.
* tests/actions.at, tests/calc.at, tests/conflicts.at,
* tests/cxx-type.at, tests/glr-regr1.at, tests/headers.at,
* tests/regression.at, tests/torture.at: Use them for all the
grammars that are to be compiled.
* tests/cxx-type.at (_AT_TEST_GLR_CALC): Rename as...
* tests/cxx-type.at (_AT_TEST_GLR_CXXTYPES): this.
* doc/bison.texinfo (GLR Parsers): Document `inline'.
reduced here, since anyway that computation doesn't work.
* src/gram.h, src/gram.h (rule_filter_t, rule_useful_p)
(rule_useless_p, rule_never_reduced_p): New.
(grammar_rules_partial_print): Use a filter instead of a range.
Display the title only if needed.
(grammar_rules_print): Adjust.
(grammar_rules_never_reduced_report): New.
* src/tables.c (action_row): Move the computation of rules never
reduced to...
(token_actions): here.
* src/main.c (main): Make the parser before making the report, so
that rules never reduced are computed.
Call grammar_rules_never_reduced_report.
* src/print.c (print_results): Report rules never reduced.
* tests/conflicts.at, tests/reduce.at: Adjust.
by conflicts.
* src/LR0.c (save_reductions): Don't make the final state too
different: save its reduction (accept) instead of having a state
without any action (no shift or goto, no reduce).
Note: the final state is now a ``regular'' state, i.e., the
parsers now contain `reduce 0' as default reduction.
Nevertheless, since they decide to `accept' when yystate =
final_state, they still will not reduce rule 0.
* src/print.c (print_actions, print_reduction): Adjust.
* src/output.c (action_row): Track reduced rules.
(token_actions): Report rules never reduced.
* tests/conflicts.at, tests/regression.at: Adjust.
* src/symtab.h, src/symtab.c (eoftoken, axiom): Rename as...
(endtoken, accept): these.
* src/reader.c (reader): Set endtoken's default tag to "$end".
Set undeftoken's tag to "$undefined" instead of "$undefined.".
* doc/bison.texinfo (Table of Symbols): Mention $accept and $end.
Adjust.
* src/LR0.c, src/closure.c, src/derives.c, src/gram.c, src/lalr.c
* src/nullable.c, src/output.c, src/print.c, src/reader.c
* src/reduce.c: Allocate and free from &rules[0], not &rules[1].
Iterate from 0 to nrules.
Use rule_number_as_item_number and item_number_as_rule_number.
Adjust to `derive' now containing possibly 0.
* src/gram.h (rule_number_as_item_number, item_number_as_rule_number):
Handle the `- 1' part in rule numbers from/to item numbers.
* src/conflicts.c (log_resolution): Fix the message which reversed
shift and reduce.
* src/output.c (action_row): Initialize default_rule to -1.
(token_actions): Adjust.
* tests/sets.at (Nullable, Firsts): Fix the previously bogus
expected output.
* tests/conflicts.at (Resolved SR Conflicts): Likewise.
* data/glr.c (YYFPRINTF): Always define it, not only when YYDEBUG
(yyuserAction, yyreportAmbiguity): `Use' all the arguments.
(yyremoveDeletes): `yyi' and `yyj' are size_t.
Use YYFPRINTF when under if (YYDEBUG) to avoid empty `if' bodies.
(yyaddDeferredAction): static.
(yyglrReduce): yyi, yyk, amd yyposn are size_t.
(yyreportParseError): yyprefix is const.
yytokenp is used only when verbose.
(yy__GNUC__): Replace with __GNUC__.
(yypdumpstack): yyi is size_t.
(yypreference): Un-yy local variables and arguments, to avoid
clashes with `yyr1'. Anyway, we are not in the user name space.
(yytname_size): be an int, as is compared with ints.
* tests/testsuite.at (AT_COMPILE, AT_PARSER_CHECK): New.
Use them.
* tests/cxx-gram.at: Use quotation to protect $1.
Use AT_COMPILE to enable warnings hunts.
Prototype yylex and yyerror.
`Use' argc.
Include `string.h', not `strings.h'.
Produce and prototype stmtMerge only when used.
yylex takes a location.
