[PATCH] getdate: don't leak when TZ contains two or more '"'s

FYI, this potential leak was spotted via coverity, as was the one that prompted today's readtokens fix. >From 2e3e4195c0c15b836ec60652a1b3b06d81767a2f Mon Sep 17 00:00:00 2001 From: Jim Meyering Date: Mon, 18 Jan 2010 08:23:24 +0100 Subject: [PATCH] getdate: don't leak when TZ contains two or mor

[PATCH] readtokens: do not leak internal token_lengths buffer

FYI, >From 06758294115f10dd93055a531113c24f8930f992 Mon Sep 17 00:00:00 2001 From: Jim Meyering Date: Mon, 18 Jan 2010 08:10:07 +0100 Subject: [PATCH] readtokens: do not leak internal token_lengths buffer * lib/readtokens.c (readtokens): Free the local, lengths, when the supplied "token_lengths"

Re: [PATCH 0/5] speed up gnulib-tool some more

Hello Ralf, Ralf Wildenhues wrote: > This patch series is a repost and update of the unapplied parts of > > plus a couple of new patches; prompted by > . I'm glad that

split mathl module (was: Add some #undef's to math.in.h, for AIX with xlc.)

Hi Paolo, Ralf Wildenhues wrote: > AIX 5.3 needs a number of #undef's in math.in.h, in order to compile, I don't want to consider Ralf's patch, because the 'mathl' module is broken in so many respects that first a cleanup is needed. The problems with the 'mathl' module are: 1) It provides func

Re: Tru64 5.1 cc #include_next header name expansion

Ralf Wildenhues wrote: > This one is funny: Tru64 5.1 cc, which is > Compaq C V6.5-303 (dtk) on Compaq Tru64 UNIX V5.1 (Rev. 732) > > treats include_next different from include: > > $ echo '#define foo bar > #include ' | cc -E - > # 1 "" > > > cc: Error: , line 2: Cannot find file specified in

Re: AIX xlc gets confused by logic in stdio.h

Ralf Wildenhues wrote: > It comes from test-select.o: > $ nm -B test-select.o | grep print > - U .__printf__ > static void > failed (const char *reason) > { > if (++failures > 1) > __printf__ (" "); > __printf__ ("failed (%s)\n", reason); > } This is as expected. > $ nm -B ../gl

Re: AIX xlc gets confused by logic in stdio.h

* Bruno Haible wrote on Sun, Jan 17, 2010 at 07:34:49PM CET: > Ralf Wildenhues wrote: > > I'm seeing this on AIX 5.3: > > > > /tmp/gl/build-aux/compile cc -qlanglvl=extc89 -qlanglvl=extc99 -g > > -o test-select test-select.o ../gllib/libgnu.a -lm > > ld: 0711-317 ERROR: Undefined s

Re: AIX xlc gets confused by logic in stdio.h

Ralf Wildenhues wrote: > I'm seeing this on AIX 5.3: > > /tmp/gl/build-aux/compile cc -qlanglvl=extc89 -qlanglvl=extc99 -g > -o test-select test-select.o ../gllib/libgnu.a -lm > ld: 0711-317 ERROR: Undefined symbol: .__printf__ > ld: 0711-345 Use the -bloadmap or -bnoquiet option t

Re: Link test-poll against INET_PTON_LIB.

Ralf Wildenhues wrote: > Link test-poll against INET_PTON_LIB. > > * modules/poll-tests (test_poll_LDADD): Add $(INET_PTON_LIB), > for inet_pton on Solaris 10. This is fine too. Thanks! Bruno

Re: Fix a couple of missing LIBTHREAD link failures on AIX.

Hi Ralf, > AIX 5.3 needs -lpthread for git-merge-changelog (through fstrcmp), > and strsignal-tests due to strsignal. The patch below fixes these. Thanks. You're welcome to apply this. Bruno

Re: stat-times LGPL

Hi Paul, On Sun 17 Jan 2010 00:38, Paul Eggert writes: > Andy Wingo writes: > >> I'm writing again to ask that stat-times be relicensed as LGPL, so Guile >> can use it. > > Yes, that's fine. Sorry I didn't reply earlier. Thanks! Andy -- http://wingolog.org/

Re: Remove fts-lgpl?

According to Ralf Wildenhues on 1/17/2010 2:40 AM: >>> I remember the fts-lgpl module already being broken last time I tested >>> gnulib. Does anyone use this module at all, and how come they don't see >>> this issue? >> I think you've nailed it - fts-lgpl isn't really being used anywhere. > > No

AIX xlc gets confused by logic in stdio.h

I'm seeing this on AIX 5.3: /tmp/gl/build-aux/compile cc -qlanglvl=extc89 -qlanglvl=extc99 -g -o test-select test-select.o ../gllib/libgnu.a -lm ld: 0711-317 ERROR: Undefined symbol: .__printf__ ld: 0711-345 Use the -bloadmap or -bnoquiet option to obtain more information. I can o

Re: [PATCH 5/5] Drop self-dependencies in modules/unistdio/*.

Hi Ralf, > * modules/unistdio/u16-sprintf: Drop self-dependency. > * modules/unistdio/u16-u16-sprintf: Likewise. > * modules/unistdio/u32-sprintf: Likewise. > * modules/unistdio/u32-u32-sprintf: Likewise. > * modules/unistdio/u8-sprintf: Likewise. > * modules/unistdio/u8-u8-sprintf: Likewise. > *

Tru64 5.1 cc #include_next header name expansion

This one is funny: Tru64 5.1 cc, which is Compaq C V6.5-303 (dtk) on Compaq Tru64 UNIX V5.1 (Rev. 732) treats include_next different from include: $ echo '#define foo bar #include ' | cc -E - # 1 "" cc: Error: , line 2: Cannot find file specified in #include directive. (noinclfile) #include -

Link test-poll against INET_PTON_LIB.

I need this patch to avoid this failure on Solaris 10: source='../../gltests/test-poll.c' object='test-poll.o' libtool=no \ DEPDIR=.deps depmode=none /bin/bash ../../gltests/../build-aux/depcomp \ cc -xc99=all -DHAVE_CONFIG_H -I. -I../../gltests -I. -I../../gltests -I.. -I../../gltests/.. -I../g

Add some #undef's to math.in.h, for AIX with xlc.

AIX 5.3 needs a number of #undef's in math.in.h, in order to compile, but I'm not sure whether that is the right fix, or some other logic is wrong. What do you need to know to analyze this? Thanks, Ralf Add some #undef's to math.in.h, for AIX with xlc. * lib/math.in.h (acosl, asinl,

Fix a couple of missing LIBTHREAD link failures on AIX.

AIX 5.3 needs -lpthread for git-merge-changelog (through fstrcmp), and strsignal-tests due to strsignal. The patch below fixes these. Thanks, Ralf Fix a couple of missing LIBTHREAD link failures on AIX. * modules/git-merge-changelog (git_merge_changelog_LDADD): Add $(LIBTHREAD).

[PATCH 5/5] Drop self-dependencies in modules/unistdio/*.

* modules/unistdio/u16-sprintf: Drop self-dependency. * modules/unistdio/u16-u16-sprintf: Likewise. * modules/unistdio/u32-sprintf: Likewise. * modules/unistdio/u32-u32-sprintf: Likewise. * modules/unistdio/u8-sprintf: Likewise. * modules/unistdio/u8-u8-sprintf: Likewise. * modules/unistdio/ulc-spr

[PATCH 4/5] Efficient license compatibility checks with --test/--create-testdir.

For licence compatibility testing of requested modules in --test or --create-testdir modes, walk the directed dependency graph; in postorder traversal, annotate the set of potential license incompatibilites in the respective subtrees. Warn about circular dependencies, where the algorithm does not

[PATCH 3/5] gnulib-tool: small sed optimization.

* gnulib-tool (func_get_automake_snippet): Merge multiple sed invocations. Signed-off-by: Ralf Wildenhues --- Noting that these sed scripts can be merged may not be trivial, you may believe me that it is alright in this case. ;-) I have not timed the impact of this. Cheers, Ralf ChangeLog

[PATCH 2/5] gnulib-tool: avoid forks with func_get_* functions.

* gnulib-tool (func_get_description) (func_get_status, func_get_notice, func_get_applicability) (func_get_filelist, func_get_dependencies) (func_get_autoconf_early_snippet, func_get_autoconf_snippet) (func_get_automake_snippet, func_get_include_directive) (func_get_link_directive, func_get_license,

[PATCH 1/5] gnulib-tool: cache module metainformation.

* gnulib-tool (sed_extract_prog): Match newline before each header, and rewrite header to a shell variable suffix. (func_cache_var, func_cache_lookup_module): New functions, to turn a module name into a cache variable prefix, and to look up and cache module metainformation. (func_get_description, f

[PATCH 0/5] speed up gnulib-tool some more

This patch series is a repost and update of the unapplied parts of plus a couple of new patches; prompted by . The first two patches, taken together, have a net effect o

Remove fts-lgpl? (was: fts-lgpl --test failure)

Hi Eric, * Eric Blake wrote on Wed, Jan 06, 2010 at 10:04:08PM CET: > Ralf Wildenhues writes: > > > I remember the fts-lgpl module already being broken last time I tested > > gnulib. Does anyone use this module at all, and how come they don't see > > this issue? > > I think you've nailed it - f