Paul Eggert [Sat, 23 Jul 2005 17:06:41 +0000 (17:06 +0000)]
(message): Don't print a message more than once,
and omit line-number decoration that makes Emacs compile think
that informative messages are worth worrying about.
Paul Eggert [Fri, 22 Jul 2005 18:35:27 +0000 (18:35 +0000)]
For translations that have not yet been upgraded to
the new runtime-po domain, prime the pump by extracting the
relevant strings from the obsolete translations. This code can be
removed once the bison-runtime domain has been translated by each
team.
Paul Eggert [Fri, 22 Jul 2005 17:07:31 +0000 (17:07 +0000)]
* data/glr.c (YYTOKEN_TABLE): New macro.
(yytname): Define if YYTOKEN_TABLE.
* data/yacc.c (YYTOKEN_TABLE, yytname): Likewise.
* data/lalr1.cc (YYTOKEN_TABLE, yytname_): Likewise.
(YYERROR_VERBOSE): Define the same way the other skeletons do.
* src/output.c (prepare_symbols): Output token_table_flag.
Paul Eggert [Thu, 21 Jul 2005 20:53:03 +0000 (20:53 +0000)]
* data/glr.c (yytnamerr): New function.
(yyreportSyntaxError): Use it to dequote most string literals.
* data/lalr1.c (yytname_): Renamed from yyname_, for compatibility
with other skeletons. All uses changed.
(yytnameerr_): New function.
(yyreport_syntax_error): Use it to dequote most string literals.
* data/yacc.c (yytnamerr): New function.
(yyerrlab): Use it to decode most string literals.
* doc/bison.texinfo (Decl Summary, Calling Convention):
Clarify quoting convention of yytname.
* src/output.c (prepare_symbols): Quote all names. This undoes
the 2005-04-17 change, which is now accomplished (mostly) via
changes in the parsers as described above.
* tests/regression.at (Token definitions, Web2c Actions):
Undo most 2005-04-17 change here, too.
Paul Eggert [Wed, 20 Jul 2005 22:01:47 +0000 (22:01 +0000)]
Don't pipe output of ./types through sed to
remove trailing spaces. This loses the exit status of ./types,
and isn't needed since ./types shouldn't be emitting trailing
spaces.
Paul Eggert [Tue, 19 Jul 2005 06:56:44 +0000 (06:56 +0000)]
Destructor cleanups and regularization among the three skeletons.
* NEWS: Document the behavior changes.
* data/glr.c (yyrecoverSyntaxError): Don't bother to pop the
stack before failing, as the cleanup code will do it for us now.
* data/lalr1.cc (yyerrlab): Likewise.
* data/glr.c (yyparse): Pop everything off the stack before
freeing it, so that destructors get called properly.
* data/lalr1.cc (yyreturn): Likewise.
* data/yacc.c (yyreturn): Pop and destroy the start symbol, too.
This is more consistent.
* doc/bison.texinfo (Destructor Decl): Mention more reasons
why destructors might be called. 1.875 -> 2.1.
(Destructor Decl, Decl Summary, Table of Symbols):
Some English-language cleanups for %destructor.
* tests/actions.at (_AT_CHECK_PRINTER_AND_DESTRUCTOR):
Add output line for destructor of start symbol.
* tests/calc.at (AT_CHECK_CALC): Add one to line counts,
because of that same extra output line.
Paul Eggert [Tue, 19 Jul 2005 00:05:57 +0000 (00:05 +0000)]
* NEWS: Document minor wording changes in diagnostics of
Bison-generated parsers.
* data/glr.c (yyMemoryExhausted): Renamed from yyStackOverflow.
Remove unused formals. All uses changed.
(yyreportAmbiguity): "ambiguity detected" -> "syntax is ambiguous".
(yyparse): Rename yyoverflowlab to yyexhaustedlab.
* data/yacc.c (yyparse): "parser stack overflow" -> "memory exhausted".
Rename yyoverflowab to yyexhaustedlab.
When memory exhaustion occurs during syntax-error reporting,
report it separately rather than in a single diagnostic; this
eases translation.
* doc/bison.texinfo (Memory Management): Renamed from Stack Overflow.
(Memory Exhausted): Renamed from Parser Stack Overflow.
Revamp wording slightly to prefer "memory exhaustion".
* tests/actions.at: "parser stack overflow" -> "memory exhausted".
Paul Eggert [Mon, 18 Jul 2005 22:10:15 +0000 (22:10 +0000)]
Add i18n support to the GLR skeleton. Partially fix the C++
skeleton; a C++ expert needs to finish this. Remove debugging
msgids; there's little point to having them translated, since they
can be understood only by someone who can read the
(English-language) source code.
Generate runtime-po/bison-runtime.pot automatically, so that we
don't have to worry about garbage getting in that file. We'll
make sure after the next official release that old msgids don't
get lost. See
<http://lists.gnu.org/archive/html/bison-patches/2005-07/msg00119.html>.
* runtime-po/Makefile.in.in, runtime-po/bison-runtime.pot: Remove.
Now auto-generated.
* PACKAGING: Don't claim that Gawk, GCC, Perl use this method yet.
Fix typos in explanations of the runtime file.
* bootstrap: Change gettext keyword from YYI18N to YY_.
Use standard Makefile.in.in in runtime-po, since we'll arrange
for backward-compatible bison-runtime.po files in a different way.
* data/glr.c (YY_): New macro, from yacc.c.
(yyuserAction, yyreportAmbiguity, yyreportSyntaxError, yyparse):
Translate messages intended for users.
(yyreportSyntaxError): Change "virtual memory" to "memory" to match
the wording in the other skeletons. We don't know that the memory
is virtual.
* data/lalr1.cc (YY_): Renamed from _. All uses changed.
Use same method that yacc.c uses.
Don't translate debugging messages.
(yy::yyreport_syntax_error): Put in a FIXME for the i18n stuff;
it doesn't work (yet), and requires C++ expertise to fix.
* data/yacc.c (YY_): Renamed from YY18N. All uses changed.
Move defn to a more logical place, to be consistent with other
skeletons.
Don't translate debugging messages.
Don't assume line numbers fit in unsigned int; use unsigned long fmts.
* doc/bison.texinfo: Mention <libintl.h>. Change glibc cross reference
to gettext cross reference. Add indexing terms. Mention YYENABLE_NLS.
* runtime-po/POTFILES.in: Add data/glr.c, data/lalr1.cc.
Paul Eggert [Mon, 18 Jul 2005 18:39:01 +0000 (18:39 +0000)]
Fix yyerror / yylex test glitches noted by twlevo@xs4all.nl.
* tests/cxx-type.at (_AT_TEST_GLR_CXXTYPES): Have yyerror return
void, not int.
* tests/glr-regression.at (Badly Collapsed GLR States):
Likewise.
(Improper handling of embedded actions and dollar(-N) in GLR parsers):
yylex should return 0 at EOF rather than aborting.
Paul Eggert [Mon, 18 Jul 2005 18:09:40 +0000 (18:09 +0000)]
Improve tests for stack overflow in GLR parser.
Problem reported by twlevo@xs4all.nl.
* data/glr.c (struct yyGLRStack): Remove yyerrflag member.
All uses removed.
(yyStackOverflow): Just longjmp, but with value 2 so that caller
can handle the problem.
(YYCHK1): Use goto (a la yacc.c) rather than setting a flag.
(yyparse): New local variable yyresult to record the result.
Use result of setjmp to set it, rather than storing itinto
struct.
(yyDone): Remove label.
(yyacceptlab, yyabortlab, yyoverflowlab, yyreturn): New labels,
to mimic yacc.c. Do not discard lookahead if it's EOF (possible
if YYABORT is used).
* tests/actions.at (_AT_CHECK_PRINTER_AND_DESTRUCTOR): Exit with
yyparse status; put status > 1 into diagnostic.
Check that status==2 works.
* tests/calc.at, tests/cxx-type.at, tests/glr-regression.at:
Use exit status 3 for failure to open (which shouldn't happen).
Paul Eggert [Mon, 18 Jul 2005 06:14:16 +0000 (06:14 +0000)]
* tests/conflicts.at (%nonassoc and eof): Don't exit with status
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.
Paul Eggert [Sun, 17 Jul 2005 07:30:38 +0000 (07:30 +0000)]
* data/glr.c (yyparse): Don't assume that the initial calls
to YYMALLOC succeed; in that case, yyparse incorrectly returned 0.
Print a stack-overflow message and fail instead.
Initialize the line-number information before creating the stack,
so that the stack-overflow message can report line zero safely.
Paul Eggert [Thu, 14 Jul 2005 23:15:47 +0000 (23:15 +0000)]
* data/glr.c (YYSTACKEXPANDABLE): Don't define if already defined.
(yyuserMerge): Provide a default case if b4_mergers is empty.
* tests/cxx-type.at (_AT_TEST_GLR_CXXTYPES): Define YYSTACKEXPANDABLE.
* tests/glr-regression.at
(Improper handling of embedded actions and dollar(-N) in GLR parsers):
YYSTYPE is char *, not char const *, so that strcpy ($$, ...) works.
Paul Eggert [Thu, 14 Jul 2005 17:27:01 +0000 (17:27 +0000)]
* tests/Makefile.am (maintainer-check-g++): Remove bogus
test for valgrind; valgrind is independent of g++.
(maintainer-check-posix): Add _POSIX2_VERSION=200112, to check
for compatibility with POSIX 1003.1-2001 (if running coreutils).
* tests/cxx-type.at (_AT_TEST_GLR_CXXTYPES): Port to stricter C++.
Use a destructor, so that we can expand the stack. Change
YYSTYPE to char * so that we can free it. Cast result of malloc.
Paul Eggert [Wed, 13 Jul 2005 18:16:30 +0000 (18:16 +0000)]
* PACKAGING: New file, suggested by Bruno Haible and taken from
similar wording in gettext's PACKAGING file.
* NEWS: Mention PACKAGING.
* Makefile.am (EXTRA_DIST): Add PACKAGING.
Paul Eggert [Tue, 12 Jul 2005 23:28:54 +0000 (23:28 +0000)]
* bootstrap: Get runtime translations into runtime-po.
Create runtime-po files automatically, if possible.
* configure.ac: Invoke BISON_I18N, so that we eat our own dog food.
* data/yacc.c: Rewrite inclusion of <libintl.h> so that ENABLE_NLS
does not infringe on the user's name space.
(YYENABLE_NLS): Renamed from ENABLE_BISON_NLS.
* doc/bison.texinfo (Internationalization): Revamp the English
and Texinfo syntax a bit, to try to make it clearer.
(Bison Options, Option Cross Key): Mention --print-localedir.
* m4/bison-i18n.m4 (BISON_I18N): Rename ENABLE_BISON_NLS to
YYENABLE_NLS. Quote a bit more.
* runtime-po/.cvsignore: New file.
* runtime-po/Makefile.in.in (mostlyclean): Remove *.old.pot.
* runtime-po/Rules-quot: Remove; now created by bootstrap.
* runtime-po/quot.sed: Likewise.
* runtime-po/boldquot.sed: Likewise.
* runtime-po/en@quot.header: Likewise.
* runtime-po/en@boldquot.header: Likewise.
* runtime-po/insert-header.sin: Likewise.
* runtime-po/remove-potcdate.sin: Likewise.
* runtime-po/Makevars: Likewise.
* runtime-po/LINGUAS: Likewise.
* runtime-po/de.po: Likewise; we will rely on the translation project
to maintain this, so "bootstrap" should get it.
* src/getarg.s (PRINT_LOCALEDIR_OPTION): Let the C compiler determine
its value.
* src/main.c (main): Bind the bison-runtime domain, too.
* data/yacc.c: Include <libintl.h> when NLS is enabled.
(YYI18N): Renamed from _. Use dgettext when NLS is enabled.
* po/POTFILES.in: Remove autogenerated file src/parse-gram.c.
* runtime-po: New directory.
* runtime-po/Makefile.in.in: New file, copied from po/, with modified
$(DOMAIN).pot-update rule, so that old messages are never dropped.
* runtime-po/Rules-quot: New file, copied from po/.
* runtime-po/quot.sed: Likewise.
* runtime-po/boldquot.sed: Likewise.
* runtime-po/en@quot.header: Likewise.
* runtime-po/en@boldquot.header: Likewise.
* runtime-po/insert-header.sin: Likewise.
* runtime-po/remove-potcdate.sin: Likewise.
* runtime-po/Makevars: New file.
* runtime-po/POTFILES.in: New file.
* runtime-po/LINGUAS: New file.
* runtime-po/bison-runtime.pot: New file.
* runtime-po/de.po: New file.
* m4/bison.m4: New file.
* Makefile.am (SUBDIRS): Add runtime-po.
(aclocaldir, aclocal_DATA): New variables.
* configure.ac: Add AC_CONFIG_FILES of runtime-po/Makefile.in.
Define aclocaldir.
* src/getargs.c (usage): Document --print-localedir option.
(PRINT_LOCALEDIR_OPTION): New enum item.
(long_options): Add --print-localedir option.
(getargs): Handle --print-localedir option.
* doc/bison.texinfo (Bison Parser): Remove paragraph about _().
(Internationalization): New section.
* src/symtab.h, src/symtab.c (symbol_print): Swap the arguments,
for consistency with the rest of the code.
* src/symlist.h, src/symlist.c (symbol_list_print): Ditto.
Add separators.
Paul Eggert [Sat, 9 Jul 2005 07:14:18 +0000 (07:14 +0000)]
* data/yacc.c (yyparse): In the initial action, set
yylsp[0] and yyvsp[0] rather than yylloc and yylval.
This avoids the use of undefined variables if the initial
action does not set yylloc and/or yylval.
Paul Eggert [Thu, 7 Jul 2005 23:16:46 +0000 (23:16 +0000)]
* examples/calc++/calc++-driver.cc, examples/calc++/calc++-driver.hh:
* examples/calc++/calc++-scanner.ll, examples/calc++/calc++.cc:
Remove from CVS. These files are automatically generated.
* examples/extexi: Clarify that this file is now part of Bison,
not GNU M4, and that it works with any POSIX-compatible Awk.
* examples/calc++/Makefile.am (run_extexi): Remove; not used.
($(calc_extracted)): Renamed from $(calc_sources_extracted),
so that we also get calc++-parser.yy. Geneate it.
Use $(AWK), not gawk, since any conforming Awk will do.
Put comment before action, since older 'make' can't handle comment
in action.
$(BUILT_SOURCES): List all built sources, not just some of them.
$(MAINTAINERCLEANFILES): Remove *.stamp, and all built sources.
$($(srcdir)/calc++-parser.stamp): Work even if POSIXLY_CORRECT.
$($(calc_sources_generated)): Remove unnecessary test for existence
of target. (This had a shell syntax error anyway; a stray "x".)
(calc_extracted): List $(srcdir)/calc++-parser.yy, not
calc++-parser.yy.
* examples/.cvsignore, examples/calc++/.cvsignore: New files.
Paul Eggert [Wed, 6 Jul 2005 04:12:23 +0000 (04:12 +0000)]
* data/glr.c (yyFail): Drastically simplify; since the format argument
never had any % directives, we can simply pass it to yyerror.
(yyparse): Use "t a; a=b;" rather than "t a = b;" when a will
be modified later, as that is the usual style in glr.c.
Paul Eggert [Tue, 5 Jul 2005 21:58:37 +0000 (21:58 +0000)]
Rewrite GLR parser to catch more buffer overrun, storage exhaustion,
and size overflow errors.
* data/glr.c: Include <stdio.h> etc. after user prolog, not before,
in case the user prolog sets feature-test macros like _GNU_SOURCE.
(YYSIZEMAX): New macro.
(yystpcpy): New function, taken from yacc.c.
(struct yyGLRStack.yyspaceLeft): Now size_t, not int.
(yyinitGLRStack, yyfreeGLRstack): Remove unnecessary forward decls,
so that we don't have to maintain their signatures.
(yyFail): Check for buffer overflow, by using vsnprintf rather
than vsprintf. Allocate a bigger buffer if possible.
Report an error if buffer allocation fails.
(yyStackOverflow): New function.
(yyinitStateSet, yyinitGLRStack): Return a boolean indicating whether
the initialization was successful. It might fail if storage was
exhausted.
(yyexpandGLRStack): Add more checks for storage allocation failure.
Use yyStackOverflow to report failures.
(yymarkStackDeleted, yyglrShift, yyglrShiftDefer, yydoAction):
(yysplitStack, yyprocessOneStack, yyparse, yypstack):
Don't assume stack number fits in int.
(yysplitStack): Check for storage allocation failure.
(yysplitStack, yyprocessOneStack): Add pure_formals, so that we
can print diagnostics on storage allocation failure. All callers
changed.
(yyresolveValue): Use yybool for boolean.
(yyreportSyntaxError): Check for size-calculation overflow.
This code is taken from yacc.c.
(yyparse): Check for storage allocation errors when allocating
the initial stack.
Extract calc++ from the documentation.
* doc/bison.texinfo (Calc++): Add the extraction marks.
* examples/extexi: New, from the aborted GNU Programming 2E.
Separate the different paragraph of a file with empty lines.
* examples/Makefile: Use it to extract the whole calc++ example.
Paul Eggert [Fri, 27 May 2005 23:22:10 +0000 (23:22 +0000)]
Fix infringement on user name space reported by Bruno Haible.
* data/yacc.c (YYSIZE_T): Define first, so that later decls can use it.
Prefer GCC's __SIZE_TYPE__ if available, so that we don't infringe on
the user's name space.
(alloca): Include <stdlib.h> to get it, if it's not built in.
(YYMALLOC, YYFREE): Define only if needed.
(malloc, free): Declare, but only if needed, as this infringes on
the user name space.
Paul Eggert [Wed, 25 May 2005 19:48:00 +0000 (19:48 +0000)]
Fix BeOS, FreeBSD, MacOS porting problems reported by Bruno Haible.
* lib/bitset.c (bitset_print): Don't assume size_t can be printed
with %d format.
* lib/ebitset.c (min, max): Undef before defining.
* lib/vbitset.c (min, max): Likewise.
* lib/subpipe.c (create_subpipe): Save local variables in case
vfork clobbers them.
* tests/synclines.at (AT_SYNCLINES_COMPILE): Add support for the
error message syntax used by gcc-4.0.
Paul Eggert [Sun, 22 May 2005 17:45:28 +0000 (17:45 +0000)]
Various maintainer cleanups.
* .cvsignore: Add a.exe, a.out, b.out,, conf[0-9]*, confdefs*,
conftest*, for benefit of CVS commands run at the same time as
"configure". Add build-aux, since "bootstrap" now creates it and
its subfiles.
* Makefile.cfg (move_if_change): Remove.
* Makefile.maint: Remove the update stuff; we now use "bootstrap".
(ftp-gnu, www-gnu, move_if_change, local_updates, update):
(po_repo, do-po-update, po-update, wget_files, get-targets):
(config.guess-url_prefix, config.sub-url_prefix):
(ansi2knr.c-url_prefix, texinfo.tex-url_prefix):
(standards.texi-url_prefix, make-stds.texi-url_prefix, taget, url):
($(get-targets), cvs-files, automake_repo, wget-update, cvs-update):
Remove.
* configure.ac (AC_CONFIG_AUX_DIR): Change from config to build-aux;
this is now the recommended name.
* config/.cvsignore: Remove config.guess, config.rpath, config.sub,
depcomp, install-sh, mdate-sh, missing, mkinstalldirs, texinfo.tex,
ylwrap. These files now go into build-aux.
* config/move-if-change: Remove.
* config/prev-version.txt: Bump from 1.75 to 2.0.
Paul Eggert [Sun, 22 May 2005 08:04:06 +0000 (08:04 +0000)]
* bootstrap: Add stdio-safer, unistd-safer modules.
Remove m4/glibc2.m4 (introduced by latest gnulib, but
we don't need it).
* lib/.cvsignore: Add dup-safer.c, fd-safer.c,
fopen-safer.c, stdio-safer.h, unistd-safer.h.
* lib/subpipe.c: Include "unistd-safer.h".
(create_subpipe): Make sure all the newly-created
file descriptors are > 2, so that diagnostics don't
get sent down them (which might cause Bison to hang, in theory).
* m4/.cvsignore: Add stdio-safer.m4, unistd-safer.m4.
* src/files.c (xfopen): Use fopen_safer, not fopen.
Paul Eggert [Sun, 22 May 2005 05:20:01 +0000 (05:20 +0000)]
Fix a destructor bug reported by Wolfgang Spraul in
<http://lists.gnu.org/archive/html/bug-bison/2005-05/msg00042.html>.
* data/yacc.c (yyabortlab): Don't call destructor, and
don't set yychar to EMPTY.
(yyoverflowlab): Don't call destructor.
(yyreturn): Call destructor, if yychar is neither YYEOF nor YYEMPTY.
* tests/calc.at (AT_CHECK_CALC): Expect one fewer output lines,
since we no longer output the message "discarding lookahead token
end of input ()".
Paul Hilfinger [Sat, 21 May 2005 08:35:51 +0000 (08:35 +0000)]
* data/glr.c (YY_SYMBOL_PRINT): Don't print newline at end to
fix a small glitch in debugging output.
(yyprocessOneStack, yyrecoverSyntaxError, yyparse): Print newline
after YY_SYMBOL_PRINT where needed.
(struct yyGLRState): Add some comments.
(struct yySemanticOption): Add some comments.
(union yyGLRStackItem): Add comment.
(yymergeOptionSets): Correct this to properly perform the union,
avoiding infinite reported by Michael Rosien.
Update comment.
* tests/glr-regression.at: Add test for GLR merging error reported
by M. Rosien.
Paul Eggert [Mon, 2 May 2005 04:52:33 +0000 (04:52 +0000)]
* tests/actions.at: Test that stack overflow invokes destructors.
From Marcus Holland-Moritz.
* data/yacc.c (yyerrlab): Move the code that destroys the stack
from here....
(yyreturn): to here. That way, destructors are called properly
even if the stack overflows, or the user calls YYACCEPT or
YYABORT. Stack-overflow problem reported by Marcus Holland-Moritz.
(yyoverflowlab): Destroy the lookahead.