Paul Eggert [Mon, 12 Aug 2002 15:27:31 +0000 (15:27 +0000)]
* Makefile.am (SUBDIRS): Remove intl.
(DISTCLEANFILES): Remove.
* NEWS: Mention that GNU M4 is now required. Clarify what is
meant by "larger grammars". Mention the pt_BR translation.
* configure.in (AC_CHECK_DECLS): Add getenv, getopt.
(AM_GNU_GETTEXT_VERSION): New macro, replacing GETTEXT_VERSION var.
Bump version from 0.11.2 to 0.11.5.
(BISON_PREREQ_STAGE): Remove.
(AM_GNU_GETTEXT): Use external gettext.
(AC_OUTPUT): Remove intl/Makefile.
* config/depcomp, config/install-sh: Sync with Automake 1.6.3.
* data/glr.c: Include string.h, for strlen.
(yyreportParseError): Use size_t for yysize.
(yy_yypstack): No longer nested inside yypstates, as nested
functions are not portable. Do not assume size_t is the
same width as int.
(yypstates): Do not assume that ptrdiff_t is the same width
as int, and similarly for yyposn and YYINDEX.
* data/yacc.c: Fix comment about `$$ = $1': it can copy garbage.
* lib/Makefile.am (INCLUDES): Do not include from the intl
directory, which has been removed.
* src/Makefile.am (INCLUDES): Likewise.
* lib/Makefile.am (libbison_a_SOURCES): Add unlocked-io.h.
(bitsets_sources, additional_bitsets_sources, timevars_sources):
New vars.
* lib/Makefile.am (libbison_a_SOURCES): Avoid +=, a GNU make extension.
* tests/Makefile.am (EXTRA_DIST): Likewise.
* lib/abitset.c (abitset_reverse_list, ebitset_reverse_list):
Do not assume that bitset_windex is the same width as unsigned.
* lib/abitset.c (abitset_unused_clear): Do not assume that
bitset_word is the same width as int.
* lib/bbitset.h (BITSET_INDEX_MAX, BITSET_MSB): Likewise.
* lib/bitset.h (bitset_set, bitset_reset): Likewise.
* lib/bitset_stats.c (bitset_stats_set, bitset_stats_reset): Likewise.
* lib/ebitset.c (ebitset_set, ebitset_reset): Likewise.
* lib/lbitset.c (lbitset_set, lbitset_reset): Likewise.
* lib/abitset.c (abitset_op1): Use -1, not ~0, as memset arg (for
portability to one's complement hosts!).
* lib/ebitset.c (ebitset_op1): Likewise.
* lib/lbitset.c (lbitset_op1): Likewise.
* lib/argmatch.c, lib/quotearg.c, quotearg.h: Sync with GNU tar.
* lib/argmatch.h, lib/basename.c, lib/dirname.c, lib/dirname.h,
lib/hash.c, lib/hash.h, lib/strnlen.c, lib/xmalloc.c:
Sync with fileutils.
* lib/error.c, lib/getopt.c, lib/getopt.h, lib/getopt1.c,
lib/gettext.h: Sync with diffutils.
* lib/memrchr.c, lib/mkstemp.c, lib/strchr.c, lib/strnlen.c,
lib/strspn.c, lib/tempname.c: Use GPL, not LGPL.
* lib/obstack.c, lib/obstack.h: Sync with fileutils, except use
PROTOTYPES to check for prototypes, and "defined __STDC__" to
check for void *.
* lib/bbitset.h (BITSET_WORD_BITS): Now of type unsigned, not
size_t; the old version tried to do this but casted improperly.
(bitset_bindex, bitset_windex): Now size_t, not unsigned long.
(bitset_test): Now returns int, not unsigned long.
* lib/bitset_stats.c: Include "gettext.h".
(_): New macro.
(bitset_stats_set, bitset_stats_reset, bitset_stats_test): Don't
name locals "index", as it generates unnecessary warnings on some
hosts that have an "index" function.
* src/gram.c (grammar_dump): Do not assume ptrdiff_t fits in int.
* src/main.c (main): Cast result of bindtextdomain and textdomain
to void, to avoid a GCC warning when --disable-nls is in effect.
* src/scan-gram.l: Use strings rather than escapes when possible,
to minimize the number of warnings from xgettext.
(handle_action_dollar, handle_action_at): Don't use isdigit,
as it mishandles negative chars and it may not work as expected
outside the C locale.
* src/symtab.c (symbol_get): Don't cast LHS of an assignment;
this is a GCC extension and is not portable to other compilers.
* src/system.h (alloca): Use same pattern as ../lib/error.c.
Do not include <ctype.h>; no longer needed.
Do not include <malloc.h>; no longer needed (and generates
warnings on OpenBSD 3.0).
* tests/cxx-type.at (yylex): Do not pass signed char to isupper;
it's not portable.
* tests/regression.at: Do not use 'cc -c input.c -o input';
Sun C rejects this. Instead, use 'cc -c input.c -o input.o'.
* tests/synclines.at (AC_SYNCLINES_COMPILE): Accept any nonzero
exit status as failure, not just exit status 1. Sun C exits
with status 2 sometimes.
* tests/torture.at (AT_INCREASE_DATA_SIZE): New macro.
Use it for the two large tests.
Paul Eggert [Mon, 12 Aug 2002 14:54:27 +0000 (14:54 +0000)]
(alloca): Use same pattern as ../lib/error.c.
Do not include <ctype.h>; no longer needed.
Do not include <malloc.h>; no longer needed (and generates
warnings on OpenBSD 3.0).
Paul Eggert [Mon, 12 Aug 2002 14:28:29 +0000 (14:28 +0000)]
Add src/assoc.c, src/closure.c, src/gram.c,
src/main.c, src/relation.c, src/state.c, lib/bitset_stats.c,
lib/timevar.c.
Use src/parse-gram.y instead of src/parse-gram.c, as the gettext
manual recommends.
Similarly, use src/scan-gram.l instead of src/scan-gram.c.
Paul Eggert [Mon, 12 Aug 2002 14:22:08 +0000 (14:22 +0000)]
(lbitset_set, lbitset_reset): Do not assume that
bitset_word is the same width as int.
(lbitset_op1): Use -1, not ~0, as memset arg (for
portability to one's complement hosts!).
Paul Eggert [Mon, 12 Aug 2002 14:20:35 +0000 (14:20 +0000)]
(ebitset_set, ebitset_reset): Do not assume that bitset_word is the
same width as int.
(ebitset_op1): Use -1, not ~0, as memset arg (for portability to one's
complement hosts!).
Paul Eggert [Mon, 12 Aug 2002 14:16:49 +0000 (14:16 +0000)]
Include "gettext.h".
(_): New macro.
(bitset_stats_set, bitset_stats_reset, bitset_stats_test): Don't
name locals "index", as it generates unnecessary warnings on some
hosts that have an "index" function.
(bitset_stats_print_1, bitset_stats_print, bitset_stats_read,
bitset_stats_write): Wrap strings in _() if they need translation.
(bitset_stats_set, bitset_stats_reset): Do not assume that bitset_word
is the same width as int.
Paul Eggert [Mon, 12 Aug 2002 14:13:56 +0000 (14:13 +0000)]
(BITSET_WORD_BITS): Now of type unsigned, not
size_t; the old version tried to do this but casted improperly.
(bitset_bindex, bitset_windex): Now size_t, not unsigned long.
(bitset_test): Now returns int, not unsigned long.
Paul Eggert [Mon, 12 Aug 2002 14:09:24 +0000 (14:09 +0000)]
(abitset_reverse_list, ebitset_reverse_list):
Do not assume that bitset_windex is the same width as unsigned.
(abitset_unused_clear): Do not assume that bitset_word is the same
width as int.
(abitset_op1): Use -1, not ~0, as memset arg (for portability to one's
complement hosts!).
Paul Eggert [Mon, 12 Aug 2002 14:08:04 +0000 (14:08 +0000)]
(INCLUDES): Do not include from the intl directory, which has been
removed.
(libbison_a_SOURCES): Add unlocked-io.h. Avoid +=, a GNU make extension.
(bitsets_sources, additional_bitsets_sources, timevars_sources): New vars.
Paul Eggert [Mon, 12 Aug 2002 14:02:32 +0000 (14:02 +0000)]
Include string.h, for strlen.
(yyreportParseError): Use size_t for yysize.
(yy_yypstack): No longer nested inside yypstates, as nested
functions are not portable. Do not assume size_t is the
same width as int.
(yypstates): Do not assume that ptrdiff_t is the same width
as int, and similarly for yyposn and YYINDEX.
Paul Eggert [Mon, 12 Aug 2002 14:00:04 +0000 (14:00 +0000)]
(AC_CHECK_DECLS): Add getenv, getopt.
(AM_GNU_GETTEXT_VERSION): New macro, replacing GETTEXT_VERSION var.
Bump version from 0.11.2 to 0.11.5.
(BISON_PREREQ_STAGE): Remove.
(AM_GNU_GETTEXT): Use external gettext.
(AC_OUTPUT): Remove intl/Makefile.
Akim Demaille [Fri, 2 Aug 2002 08:05:01 +0000 (08:05 +0000)]
* src/conflicts.c (conflicts_output): Don't output rules never
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.
Akim Demaille [Thu, 1 Aug 2002 18:14:30 +0000 (18:14 +0000)]
Instead of attaching lookaheads and duplicating the rules being
reduced by a state, attach the lookaheads to the reductions.
* src/state.h (state_t): Remove the `lookaheads',
`lookaheads_rule' member.
(reductions_t): Add a `lookaheads' member.
Use a regular array for the `rules'.
* src/state.c (reductions_new): Initialize the lookaheads member
to 0.
(state_rule_lookaheads_print): Adjust.
* src/state.h, src/state.c (state_reductions_find): New.
* src/conflicts.c (resolve_sr_conflict, set_conflicts)
(count_rr_conflicts): Adjust.
* src/lalr.c (LArule): Remove.
(add_lookback_edge): Adjust.
(state_lookaheads_count): New.
(states_lookaheads_initialize): Merge into...
(initialize_LA): this.
(lalr_free): Adjust.
* src/main.c (main): Don't free nullable and derives too early: it
is used by --verbose.
* src/print.c, src/print_graph.c, src/tables.c: Adjust.
Akim Demaille [Thu, 1 Aug 2002 18:13:29 +0000 (18:13 +0000)]
* src/derives.h, src/derives.c (derives): A `rule_t***' instead of
`rule_number_t**'.
(set_derives, free_derives): Rename as...
(derives_compute, derives_free): this.
Adjust all dependencies.
* src/nullable.c (set_nullable, free_nullable): Rename as...
(nullable_compute, nullable_free): these.
(rule_list_t): Store rule_t *, not rule_number_t.
* src/state.c (state_rule_lookaheads_print): Directly compare rule
pointers, instead of their numbers.
* src/main.c (main): Call nullable_free, and derives_free earlier,
as they were lo longer used.
Akim Demaille [Thu, 1 Aug 2002 18:12:11 +0000 (18:12 +0000)]
* lib/timevar.c (get_time): Include children time.
* src/lalr.h (LA, LArule): Don't export them: used with the
state_t.
* src/lalr.c (LA, LArule): Static.
* src/lalr.h, src/lalr.c (lalr_free): New.
* src/main.c (main): Call it.
* src/tables.c (pack_vector): Check whether loc is >= to the
table_size, not >.
(pack_tables): Don't free froms, tos, conflict_tos, and pos...
(tables_generate): do it, since that's also it which allocates
them.
Don't free LA and LArule, main does.
Let --trace have arguments.
* src/getargs.h (enum trace_e): New.
* src/getargs.c (trace_args, trace_types, trace_argmatch): New.
(long_options, short_options): --trace/-T takes an optional
argument.
Change all the uses of trace_flag to reflect the new flags.
* tests/sets.at (Firsts, Nullable, Broken Closure): Use --trace=sets.
Strengthen `stage' portability.
* m4/stage.m4 (BISON_PREREQ_STAGE): New.
* configure.in: Use it.
Don't check for malloc.h and sys/times.h.
* src/system.h: Include them when appropriate.
* src/main.c (stage): Compile only when mallinfo, struct mallinfo,
times and struct tms are available.
In verbose parse error message, don't report `error' as an
expected token.
* tests/actions.at (Printers and Destructors): Adjust.
* tests/calc.at (Calculator $1): Adjust.
* data/yacc.c, data/glr.c, data/lalr1.c: When making the verbose
error message, do not report the parser accepts the error token in
that state.
Report rules which are never reduced by the parser: those hidden
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.
`stage' was accidently included in a previous patch.
Initiate its autoconfiscation.
* configure.in: Look for malloc.h and sys/times.h.
* src/main.c (stage): Adjust.
Report only when trace_flag.
Use $accept and $end, as BYacc and BTYacc do, instead of $axiom and $.
* src/symtab.h, src/symtab.c (eoftoken, axiom): Rename as...
(endtoken, accept): these.
* src/reader.c (reader): Set endtoken's default tag to "$end".
Set undeftoken's tag to "$undefined" instead of "$undefined.".
* doc/bison.texinfo (Table of Symbols): Mention $accept and $end.
Adjust.
* src/reduce.c (reduce_grammar): When the language is empty,
complain about the start symbol, not the axiom.
Use its location.
* tests/reduce.at (Empty Language): New.
* src/reader.h, src/reader.c (gram_error): ... can't get
yycontrol without making too strong assumptions on the parser
itself.
* src/output.c (prepare_tokens): Use the real 0th value of
token_translations instead of `0'.
* src/parse-gram.y (yyerror): Don't rely on yycontrol being
visible here.
* data/yacc.c (yyreport_parse_error): Rename yylocation as yylloc
for the time being: %locations ought to provide it to yyerror.
Stop storing rules from 1 to nrules + 1.
* 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/c.m4 (b4_c_function, b4_c_ansi_args, b4_c_ansi_arg)
(b4_c_knr_arg_names, b4_c_knr_arg_name, b4_c_knr_arg_decls)
(b4_c_knr_arg_decl): New.
* data/yacc.c: Use it to define yysymprint, yydestruct, and
yyreport_parse_error.
* src/parse-gram.y (rules_or_grammar_declaration): Add an error
recovery rule, and forbid extensions when --yacc.
(gram_error): Use complain_at.
* src/reader.c (reader): Exit if there were parse errors.
* src/output.c (froms, tos): Are state_number_t.
(save_column): sp, sp1, and sp2 are state_number_t.
(prepare): Rename `final' as `final_state_number', `nnts' as
`nterms_number', `nrules' as `rules_number', `nstates' as
`states_number', and `ntokens' as `tokens_number'. Remove `nsym',
unused.
* data/yacc.c, data/glr.c, data/lalr1.cc: Adjust.
* data/lalr1.cc (nsym_): Remove, unused.
Fix some memory leaks, and fix a bug: state 0 was examined twice.
* src/LR0.c (new_state): Merge into...
(state_list_append): this.
(new_states): Merge into...
(generate_states): here.
(set_states): Don't ensure a proper `errs' state member here, do it...
* src/conflicts.c (conflicts_solve): here.
* src/state.h, src/state.c: Comment changes.
(state_t): Rename member `shifts' as `transitions'.
Adjust all dependencies.
(errs_new): For consistency, also take the values as argument.
(errs_dup): Remove.
(state_errs_set): New.
(state_reductions_set, state_transitions_set): Assert that no
previous value was assigned.
(state_free): New.
(states_free): Use it.
* src/conflicts.c (resolve_sr_conflict): Don't use an `errs_t' as
temporary storage: use `errs' and `nerrs' as elsewhere.
(set_conflicts): Allocate and free this `errs'.
projects / bison.git / log
