gnulib.git
12 years agoverify: fix bug when gnulib <assert.h> is also included
Paul Eggert [Mon, 23 May 2011 01:36:04 +0000 (18:36 -0700)]
verify: fix bug when gnulib <assert.h> is also included

* lib/verify.h (verify, verify_true): Define if _GL_VERIFY_H
is defined, not if _GL_STATIC_ASSERT_H is not defined.
Perhaps there's a better way, but this fixes the immediate problem.
Problem reported by Bruno Haible in
<http://lists.gnu.org/archive/html/bug-gnulib/2011-05/msg00478.html>.

12 years ago* doc/intprops.texi: fix typo in copyright date
Paul Eggert [Sun, 22 May 2011 20:31:38 +0000 (13:31 -0700)]
* doc/intprops.texi: fix typo in copyright date

12 years agoxgetcwd: Simplify autoconf macro.
Bruno Haible [Sun, 22 May 2011 15:30:36 +0000 (17:30 +0200)]
xgetcwd: Simplify autoconf macro.

* m4/xgetcwd.m4 (gl_XGETCWD): Don't require gl_FUNC_GETCWD.

12 years agoNew module 'mktime-internal'.
Bruno Haible [Sun, 22 May 2011 14:39:23 +0000 (16:39 +0200)]
New module 'mktime-internal'.

* modules/mktime-internal: New file.
* m4/timegm.m4 (gl_PREREQ_TIMEGM): Move contents to ...
* m4/mktime.m4 (gl_FUNC_MKTIME_INTERNAL): New macro. Define
mktime_internal as a C macro if libc has __mktime_internal.
* modules/timegm (Depends-on): Add mktime-internal. Remove mktime. Add
conditions.
* MODULES.html.sh (Date and time <time.h>): Add mktime-internal.

12 years agotimegm: Correct mktime replacement statements.
Bruno Haible [Sun, 22 May 2011 13:54:36 +0000 (15:54 +0200)]
timegm: Correct mktime replacement statements.

* m4/timegm.m4 (gl_PREREQ_TIMEGM): Set REPLACE_MKTIME, instead of
defining mktime as a C macro. This completes a 2009-07-28 commit.

12 years agotimegm: Simplify autoconf macro.
Bruno Haible [Sun, 22 May 2011 13:17:29 +0000 (15:17 +0200)]
timegm: Simplify autoconf macro.

* m4/timegm.m4 (gl_PREREQ_TIMEGM): Don't require gl_TIME_R.

13 years agoclock-time: change to LGPLv2+.
Paul Eggert [Sun, 22 May 2011 03:25:03 +0000 (20:25 -0700)]
clock-time: change to LGPLv2+.

* modules/clock-time: Change from GPL to LGPLv2+.  Actually, it's
BSD-like but we have no mark for that; this is good enough for now.

13 years agostrerror_r: Fix comments.
Bruno Haible [Sun, 22 May 2011 02:23:33 +0000 (04:23 +0200)]
strerror_r: Fix comments.

* lib/strerror_r.c (strerror_r): Fix comment about Cygwin and sys_nerr.

13 years agorelocatable-prog-wrapper: Fix possible link error.
Bruno Haible [Sat, 21 May 2011 21:41:37 +0000 (23:41 +0200)]
relocatable-prog-wrapper: Fix possible link error.

* m4/setenv.m4 (gl_FUNC_SETENV_SEPARATE): Move determination of
HAVE_SETENV and REPLACE_SETENV and AC_LIBOBJ invocation from here...
(gl_FUNC_SETENV): ... to here.
* m4/canonicalize.m4 (gl_CANONICALIZE_LGPL_SEPARATE): Update comment.
* m4/readlink.m4 (gl_FUNC_READLINK_SEPARATE): Likewise.

13 years agorelocatable-prog-wrapper: Assume strerror() exists.
Bruno Haible [Sat, 21 May 2011 21:35:46 +0000 (23:35 +0200)]
relocatable-prog-wrapper: Assume strerror() exists.

* modules/relocatable-prog-wrapper (Files): Remove lib/strerror.c,
m4/strerror.m4.
(configure.ac): Don't invoke gl_FUNC_STRERROR_SEPARATE.
* lib/relocwrapper.c: Remove mention of strerror module.
* lib/strerror.c: Assume REPLACE_STRERROR is 1.
* m4/strerror.m4 (gl_FUNC_STRERROR_SEPARATE): Remove macro.
(gl_FUNC_STRERROR): Inline it here. Don't define REPLACE_STRERROR as a
C macro.

13 years agoselect: Simplify replacement idiom.
Bruno Haible [Sat, 21 May 2011 20:28:43 +0000 (22:28 +0200)]
select: Simplify replacement idiom.

* m4/select.m4 (gl_FUNC_SELECT): Set REPLACE_SELECT also on native
Win32 platforms.
* lib/sys_select.in.h (select): Simplify accordingly.
* modules/select (Depends-on): Likewise.

13 years agomkdir-p: Simplify autoconf macro.
Bruno Haible [Sat, 21 May 2011 18:13:03 +0000 (20:13 +0200)]
mkdir-p: Simplify autoconf macro.

* m4/mkdir-p.m4 (gl_MKDIR_PARENTS): Don't require gl_FUNC_LCHMOD,
gl_FUNC_LCHOWN.

13 years agoFix some mistakes in ChangeLog entries.
Bruno Haible [Sat, 21 May 2011 16:53:58 +0000 (18:53 +0200)]
Fix some mistakes in ChangeLog entries.

13 years agostrerror_r: avoid clobbering strerror on cygwin
Eric Blake [Sat, 21 May 2011 16:08:28 +0000 (10:08 -0600)]
strerror_r: avoid clobbering strerror on cygwin

Avoid cygwin strerror_r, since it clobbers strerror buffer through
cygwin 1.7.9.  __xpg_strerror_r is okay, but if a program is
compiled on cygwin 1.7.8 or earlier, it is not available.

* lib/strerror_r.c (strerror_r): Don't use cygwin's strerror_r;
fall back instead to sys_errlist.
* modules/strerror (configure.ac): Add witness.
* tests/test-strerror_r.c (main): Enhance test.
* doc/posix-functions/strerror_r.texi (strerror_r): Document it.
* tests/test-perror2.c (main): Free memory before exit.

Signed-off-by: Eric Blake <eblake@redhat.com>
13 years agomkdtemp: Use gnulib naming conventions.
Bruno Haible [Sat, 21 May 2011 13:38:22 +0000 (15:38 +0200)]
mkdtemp: Use gnulib naming conventions.

* m4/mkdtemp.m4 (gl_FUNC_MKDTEMP): Renamed from gt_FUNC_MKDTEMP.
* modules/mkdtemp (configure.ac): Update.

13 years agostrerror_r: avoid corrupting errno on Solaris
Eric Blake [Fri, 20 May 2011 23:45:13 +0000 (17:45 -0600)]
strerror_r: avoid corrupting errno on Solaris

On Solaris, a non-zero return was also reflected into errno.
Leaving errno unchanged is a useful feature worth guaranteeing.

* m4/strerror_r.m4 (gl_FUNC_STRERROR_R): Check for Solaris behavior.
* doc/posix-functions/strerror_r.texi (strerror_r): Document it.

Signed-off-by: Eric Blake <eblake@redhat.com>
13 years agostrerror_r: avoid compiler warning
Eric Blake [Fri, 20 May 2011 23:33:16 +0000 (17:33 -0600)]
strerror_r: avoid compiler warning

A couple of copy-and-paste issues led to a compiler warning
during configure, detected on at least Solaris.

* m4/strerror_r.m4 (gl_FUNC_STRERROR_R): Don't return a char*.

Signed-off-by: Eric Blake <eblake@redhat.com>
13 years agostrerror_r: simplify AIX code.
Eric Blake [Fri, 20 May 2011 21:19:05 +0000 (15:19 -0600)]
strerror_r: simplify AIX code.

* lib/strerror_r.c (strerror_r): Filter out buflen of 1 up front.

Signed-off-by: Eric Blake <eblake@redhat.com>
13 years agotest-perror: avoid spurious failure on FreeBSD
Eric Blake [Fri, 20 May 2011 19:02:16 +0000 (13:02 -0600)]
test-perror: avoid spurious failure on FreeBSD

* modules/perror-tests (Depends-on): Add strerror, now that
strerror_r no longer pulls it in.

Signed-off-by: Eric Blake <eblake@redhat.com>
13 years agostrerror_r-posix: Remove unused dependencies.
Bruno Haible [Fri, 20 May 2011 18:29:44 +0000 (20:29 +0200)]
strerror_r-posix: Remove unused dependencies.

* modules/strerror_r-posix (Depends-on): Remove strerror.
Reported by Eric Blake.

13 years agointprops: remove assumption about A|B representation
Paul Eggert [Fri, 20 May 2011 18:30:16 +0000 (11:30 -0700)]
intprops: remove assumption about A|B representation

* lib/intprops.h (_GL_BINARY_OP_OVERFLOW): Do not assume that A|B
is a valid integer if both A and B are.  Although this is true for
all known practical hosts, the C standard doesn't guarantee it,
and the code need not assume it.  Also, this change may work around
HP-UX 11.23 and IRIX 6.5 cc bugs reported by Bruno Haible in
<http://lists.gnu.org/archive/html/bug-gnulib/2011-05/msg00426.html>.

13 years agoperror: work around FreeBSD bug
Eric Blake [Fri, 20 May 2011 17:09:53 +0000 (11:09 -0600)]
perror: work around FreeBSD bug

POSIX requires that 'errno = 0; perror ("")' print the same message
as strerror(0), but this failed if we were replacing strerror to work
around the FreeBSD bug of treating 0 as a failure.

The goal here is to _not_ replace perror on glibc, even though
strerror_r has to be replaced, because the strerror_r replacement is
only for the sake of correcting the signature rather than working
around bugs in the handling of any particular errnum value.  Recall
that $gl_cv_func_strerror_r_works is only set if the POSIX signature
was detected in the first place.

* m4/perror.m4 (gl_FUNC_PERROR): Also replace perror if strerror_r
is broken.  Move AC_LIBOBJ...
* modules/perror (configure.ac): Here.
* doc/posix-functions/perror.texi (perror): Document this.
* tests/test-perror2.c (main): Enhance test.

Signed-off-by: Eric Blake <eblake@redhat.com>
13 years agotest-perror: check for strerror interactions
Eric Blake [Fri, 20 May 2011 16:42:26 +0000 (10:42 -0600)]
test-perror: check for strerror interactions

This uncovered a glibc bug, although not many people check for
perror failures, so for now I'm not working around it.
http://sourceware.org/bugzilla/show_bug.cgi?id=12792

* tests/macros.h (STREQ) Add macro.
* modules/perror-tests (Files): Add second test.
* tests/test-perror2.c (main): New file.
* doc/posix-functions/perror.texi (perror): Document glibc bug.

Signed-off-by: Eric Blake <eblake@redhat.com>
13 years agotest-perror: rewrite to use init script
Eric Blake [Fri, 20 May 2011 14:57:54 +0000 (08:57 -0600)]
test-perror: rewrite to use init script

Make the test simpler, and also check for correct exit status.

* modules/perror-tests (Files): Add init.sh.
* tests/test-perror.sh: Use temporary directory.

Signed-off-by: Eric Blake <eblake@redhat.com>
13 years agomaint: replace misused "a" with "an"
Jim Meyering [Fri, 20 May 2011 09:53:34 +0000 (11:53 +0200)]
maint: replace misused "a" with "an"

* doc/intprops.texi: "a integer"
* doc/regex.texi: "a explanation"
* lib/alignof.h: "a object"
* lib/argmatch.h: "a explanation"
* lib/argp-help.c: "a option" and "a OPTION_DOC"
* lib/stdint.in.h: "a integer"
* lib/userspec.c: "a owner"
* doc/gnulib.texi: Fix "a idea", and reword.

13 years agomaint: correct misuse of "a" and "an"
Jim Meyering [Thu, 19 May 2011 20:18:05 +0000 (22:18 +0200)]
maint: correct misuse of "a" and "an"

* doc/regex.texi (Collating Symbol Operators): s/an close.../a close/
* lib/argp-help.c: "an docum...": s/an/a/
* lib/argp-parse.c: "An vector": s/An/A/
* lib/execute.c: "an native": s/an/a/
* lib/spawn-pipe.c: Likewise.
* lib/gc.h: "an Gc_rc": s/an/a/
* lib/unigbrk.in.h: "an grapheme": s/an/a/
* lib/fts.c: "an stat.st_dev": s/an/a/

13 years agointprops-tests: work around HP-UX 11.23 cc bug with constants
Paul Eggert [Fri, 20 May 2011 02:16:43 +0000 (19:16 -0700)]
intprops-tests: work around HP-UX 11.23 cc bug with constants

* tests/test-intprops.c (VERIFY): New macro.
(main): Use it, instead of verify, to work around the compiler bug; see
<http://lists.gnu.org/archive/html/bug-gnulib/2011-05/msg00401.html>.

13 years agointprops: work around IRIX 6.5 cc bug with 0u - 0u + -1
Paul Eggert [Fri, 20 May 2011 02:16:04 +0000 (19:16 -0700)]
intprops: work around IRIX 6.5 cc bug with 0u - 0u + -1

See http://lists.gnu.org/archive/html/bug-gnulib/2011-05/msg00406.html
* lib/intprops.h (_GL_INT_NEGATE_CONVERT): New macro.
(_GL_INT_SIGNED, _GL_INT_MAXIMUM, _GL_DIVIDE_OVERFLOW):
(_GL_REMAINDER_OVERFLOW): Use it.

13 years agointprops-tests: revert unsigned part of previous change
Paul Eggert [Fri, 20 May 2011 02:15:19 +0000 (19:15 -0700)]
intprops-tests: revert unsigned part of previous change

* tests/test-intprops.c (UINT_MAX, ULONG_MAX, UINTMAX_MAX, U0, U1):
Remove; they weren't actually needed.  All uses of U0 and U1 removed,
and other casts to 'unsigned int' reverted to 'u' suffixes.  See
<http://lists.gnu.org/archive/html/bug-gnulib/2011-05/msg00406.html>.

13 years agostrerror_r: Work around strerror_r() change in Cygwin 1.7.8.
Bruno Haible [Fri, 20 May 2011 02:01:43 +0000 (04:01 +0200)]
strerror_r: Work around strerror_r() change in Cygwin 1.7.8.

* lib/strerror_r.c (strerror_r) [CYGWIN]: Recognize when the system's
strerror_r() returned without filling the buffer.
Reported by Eric Blake.

13 years agostrerror_r: guarantee unchanged errno
Eric Blake [Thu, 19 May 2011 19:35:39 +0000 (13:35 -0600)]
strerror_r: guarantee unchanged errno

POSIX guarantees that strerror doesn't change errno on success,
and since we implement strerror by using strerror_r, it makes
sense to make the same guarantee for strerror_r (rather, going
one step further to say that sterror_r does not corrupt errno,
since it returns an error value explicitly).

See also http://austingroupbugs.net/view.php?id=447

* lib/strerror_r.c (strerror_r): Guarantee unchanged errno.
* lib/strerror-impl.h (strerror): Set errno to match strerror_r
failure.
* tests/test-strerror_r.c (main): Enhance test.

Signed-off-by: Eric Blake <eblake@redhat.com>
13 years agostrerror_r: Reorder #if blocks.
Bruno Haible [Thu, 19 May 2011 19:59:46 +0000 (21:59 +0200)]
strerror_r: Reorder #if blocks.

* lib/strerror_r.c (strerror_r): Reorder conditionals in the function
for consistency with the previous commit.

13 years agoperror: Avoid clobbering the strerror buffer when possible.
Bruno Haible [Thu, 19 May 2011 19:49:01 +0000 (21:49 +0200)]
perror: Avoid clobbering the strerror buffer when possible.

* lib/strerror-impl.h: New file, extracted from lib/strerror.c.
* lib/strerror.c: Include it.
* modules/strerror (Files): Add lib/strerror-impl.h.
* lib/perror.c: Include <stdlib.h>, intprops.h, verify.h.
(my_strerror): New function, defined through lib/strerror-impl.h.
(perror): Use it instead of strerror.
* modules/perror (Files): Add lib/strerror-impl.h.
(Depends-on): Remove strerror. Add intprops, verify, strerror_r-posix.

13 years agostrerror_r: fix on newer cygwin
Eric Blake [Thu, 19 May 2011 18:51:36 +0000 (12:51 -0600)]
strerror_r: fix on newer cygwin

* lib/strerror_r.c (rpl_strerror_r): Cygwin now has
__xpg_strerror_r, use it.

Signed-off-by: Eric Blake <eblake@redhat.com>
13 years agostrerror_r: Avoid clobbering the strerror buffer when possible.
Bruno Haible [Thu, 19 May 2011 18:57:21 +0000 (20:57 +0200)]
strerror_r: Avoid clobbering the strerror buffer when possible.

* lib/strerror.c: Define _NETBSD_SOURCE. Include <nl_types.h>.
(sys_nerr, sys_errlist): New declarations.
(strerror_r): Be careful not to clobber the strerror buffer on NetBSD,
HP-UX, native Win32, IRIX, and 32-bit Solaris.
* m4/strerror_r.m4 (gl_PREREQ_STRERROR_R): Test whether catgets exists.

13 years agostrerror_r: Fix test failure on mingw.
Bruno Haible [Thu, 19 May 2011 18:45:59 +0000 (20:45 +0200)]
strerror_r: Fix test failure on mingw.

* m4/strerror_r.m4 (gl_FUNC_STRERROR_R): Don't define
EXTEND_STRERROR_R.
* lib/strerror_r.c (strerror_r): Test the various GNULIB_defined_*
macros from errno.in.h instead.

13 years agostrerror: relax test for Solaris
Eric Blake [Thu, 19 May 2011 16:17:18 +0000 (10:17 -0600)]
strerror: relax test for Solaris

Solaris returns "Error 0" for 0 vs. "Unknown error" for -1; while that
does not quite have the positive connotation that POSIX recommends, it
is distinct enough that we do not need to replace the native strerror
just to pick a better string.

The test is really trying to reject implementations that use the same
string for both 0 and 1 (modulo any %d output of the error number).

* tests/test-strerror.c (main): Permit Solaris behavior.
* tests/test-strerror_r.c (main): Likewise.

Signed-off-by: Eric Blake <eblake@redhat.com>
13 years agostrerror: enforce POSIX ruling on strerror(0)
Eric Blake [Thu, 19 May 2011 00:16:59 +0000 (18:16 -0600)]
strerror: enforce POSIX ruling on strerror(0)

http://austingroupbugs.net/view.php?id=382 requires that strerror(0)
succeed, but FreeBSD reports "Unknown error: 0" and fails with EINVAL.

* m4/strerror.m4 (gl_FUNC_STRERROR_SEPARATE): Expose BSD bug.
* m4/strerror_r.m4 (gl_FUNC_STRERROR_R): Likewise.
* lib/strerror_r.c (rpl_strerror_r): Work around it.
* doc/posix-functions/strerror.texi (strerror): Document it.
* doc/posix-functions/strerror_r.texi (strerror_r): Likewise.
* tests/test-strerror.c (main): Strengthen test.
* tests/test-strerror_r.c (main): Likewise.

Signed-off-by: Eric Blake <eblake@redhat.com>
13 years agointprop-tests: port to older and more-pedantic compilers
Paul Eggert [Thu, 19 May 2011 08:43:17 +0000 (01:43 -0700)]
intprop-tests: port to older and more-pedantic compilers

* modules/intprops-tests (Files): Add tests/macros.h.
* tests/test-intprops.c: Include macros.h.
(TYPE_IS_INTEGER): Use ASSERT, not verify, to test this macro, as
it's no longer documented to expand to an integer constant expression.
(TYPE_SIGNED): Use ASSERT, not verify, to test this macro when the
argument is floating point, as it's no longer documented to expand
to an integer constant expression in that case.
(UINT_MAX, ULONG_MAX, UINTMAX_MAX): Redefine to work around
compiler bugs reported by Bruno Haible.  See
<http://lists.gnu.org/archive/html/bug-gnulib/2011-05/msg00401.html>.
(U0, U1): New constants, to work around the same bugs.  Also,
in tests, use e.g., "(unsigned int) 39" rather than "39u".

13 years agointprops: work around C compiler bugs
Paul Eggert [Thu, 19 May 2011 08:36:25 +0000 (01:36 -0700)]
intprops: work around C compiler bugs

* lib/intprops.h (INT_MULTIPLY_RANGE_OVERFLOW): Work around compiler
bug in Sun C 5.11 2010/08/13 and other compilers; see
<http://lists.gnu.org/archive/html/bug-gnulib/2011-05/msg00401.html>.

13 years agointprops: TYPE_IS_INTEGER, TYPE_SIGNED not integer constant exprs
Paul Eggert [Thu, 19 May 2011 08:34:14 +0000 (01:34 -0700)]
intprops: TYPE_IS_INTEGER, TYPE_SIGNED not integer constant exprs

* doc/intprops.texi (Integer Type Determination): Fix
documentation for TYPE_IS_INTEGER: it returns an constant
expression, not an integer constant expression.  Fix doc for
TYPE_SIGNED: it returns an integer constant expression only if its
argument is an integer type.  (TYPE_IS_INTEGER is the same, but is
hardly worth documented that way....)

13 years agostrerror_r: Avoid clobbering the strerror buffer when possible.
Bruno Haible [Thu, 19 May 2011 03:24:33 +0000 (05:24 +0200)]
strerror_r: Avoid clobbering the strerror buffer when possible.

* lib/strerror_r.c (strerror_r): Merge the three implementations.
Handle gnulib defined errno values here. When strerror() returns NULL
or an empty string, return EINVAL.
* lib/strerror.c (strerror): Always call strerror_r. Don't handle
gnulib defined errno values here.
* modules/strerror (Depends-on): Add verify, strerror_r-posix.

13 years agofnmatch: avoid compiler warning
Eric Blake [Wed, 18 May 2011 19:13:03 +0000 (13:13 -0600)]
fnmatch: avoid compiler warning

Detected on Ubuntu 10.04, where the glibc fnmatch fix is not yet
present; also reproduced via:
$ gl_cv_func_fnmatch_posix=no CFLAGS=-Wall \
    ./gnulib-tool --with-tests --test fnmatch

In file included from gllib/fnmatch.c:172:0:
gllib/fnmatch_loop.c: In function ‘internal_fnmatch’:
gllib/fnmatch_loop.c:203:28: error: initialization discards qualifiers from pointer target type
gllib/fnmatch_loop.c:204:28: error: initialization discards qualifiers from pointer target type

* lib/fnmatch_loop.c (FCT): Use correct type.
Reported by Matthias Bolte.

Signed-off-by: Eric Blake <eblake@redhat.com>
13 years agomaint.mk: three new prohibit_<HDR>_without_use rules
Jim Meyering [Fri, 13 May 2011 21:35:48 +0000 (23:35 +0200)]
maint.mk: three new prohibit_<HDR>_without_use rules

* top/maint.mk (sc_prohibit_stdio--_without_use): New rule.
(sc_prohibit_stdio-safer_without_use): Likewise.
(sc_prohibit_xfreopen_without_use): Likewise.

13 years agoannounce-gen: fail if the NEWS delta is empty
Jim Meyering [Tue, 17 May 2011 20:24:59 +0000 (22:24 +0200)]
announce-gen: fail if the NEWS delta is empty

If there's nothing noteworthy in NEWS, then either you forgot
or you shouldn't be releasing.
* build-aux/announce-gen: Die if the NEWS delta is effectively empty.

13 years agomaint.mk: exclude compiler symbols from sc_tight_scope
Pádraig Brady [Tue, 17 May 2011 18:08:14 +0000 (19:08 +0100)]
maint.mk: exclude compiler symbols from sc_tight_scope

13 years agointprops: add doc
Paul Eggert [Tue, 17 May 2011 20:09:08 +0000 (13:09 -0700)]
intprops: add doc

* doc/intprops.texi: New file, documenting intprops.
* doc/gnulib.texi (Particular Modules): Include it.

13 years agoverify: add doc to gnulib manual and fix example
Paul Eggert [Tue, 17 May 2011 20:07:28 +0000 (13:07 -0700)]
verify: add doc to gnulib manual and fix example

* doc/gnulib.texi (Compile-time Assertions): New node, for 'verify'.
* doc/verify.texi (Compile-time Assertions): Update 'assert' doc.
(Compile-time Assertions): Fix example so it can't overflow.

13 years agowarnings.m4: don't usurp save_CPPFLAGS variable name
Jim Meyering [Tue, 17 May 2011 12:10:16 +0000 (14:10 +0200)]
warnings.m4: don't usurp save_CPPFLAGS variable name

* m4/warnings.m4: Prefix local temporary variable name with gl_.

13 years agoChangeLog: s/^ /\t/
Jim Meyering [Tue, 17 May 2011 06:19:00 +0000 (08:19 +0200)]
ChangeLog: s/^        /\t/

13 years agodoc: fix typo
Jim Meyering [Tue, 17 May 2011 06:17:42 +0000 (08:17 +0200)]
doc: fix typo

* doc/gnulib-intro.texi (Target Platforms): s/is/are/

13 years agodoc: Tweak recent change.
Paul Eggert [Tue, 17 May 2011 00:23:22 +0000 (02:23 +0200)]
doc: Tweak recent change.

* README (Portability guidelines): Tweak new text.
* doc/gnulib-intro.texi (Target Platforms): Likewise. Mention
Interix 6.1.

13 years agointtypes: avoid autoconf warning
Eric Blake [Mon, 16 May 2011 23:20:47 +0000 (17:20 -0600)]
inttypes: avoid autoconf warning

I noticed this while updating libvirt to the latest gnulib.

configure.ac:80: warning: AC_REQUIRE: `gl_STDINT_H' was expanded before it was required
gnulib/m4/inttypes.m4:18: gl_INTTYPES_INCOMPLETE is expanded from...
gnulib/m4/inttypes.m4:12: gl_INTTYPES_H is expanded from...
gnulib/m4/gnulib-comp.m4:379: gl_INIT is expanded from...
configure.ac:80: the top level

* m4/inttypes.m4 (gl_INTTYPES_INCOMPLETE): Only expand once.
* m4/stdint.m4 (gl_STDINT_H): Likewise.

Signed-off-by: Eric Blake <eblake@redhat.com>
13 years agoTypo.
Bruno Haible [Mon, 16 May 2011 22:45:12 +0000 (00:45 +0200)]
Typo.

13 years agovc-list-files: accept multiple directory operands
Eric Blake [Mon, 16 May 2011 22:02:07 +0000 (16:02 -0600)]
vc-list-files: accept multiple directory operands

* build-aux/vc-list-files: Iterate over all remaining operands.

Signed-off-by: Eric Blake <eblake@redhat.com>
13 years agoFix confusion regarding deprecated modules.
Bruno Haible [Mon, 16 May 2011 22:27:54 +0000 (00:27 +0200)]
Fix confusion regarding deprecated modules.

* modules/calloc (Status, Notice): Mark module as deprecated, not
obsolete.
* modules/fnmatch-posix (Status, Notice): Likewise.
* modules/getdate (Status, Notice): Likewise.
* modules/getopt (Status, Notice): Likewise.
* modules/malloc (Status, Notice): Likewise.
* modules/pipe (Status, Notice): Likewise.
* modules/realloc (Status, Notice): Likewise.
* modules/rename-dest-slash (Status, Notice): Likewise.
* modules/unictype/bidicategory-all (Status, Notice): Likewise.
* modules/unictype/bidicategory-byname (Status, Notice): Likewise.
* modules/unictype/bidicategory-name (Status, Notice): Likewise.
* modules/unictype/bidicategory-of (Status, Notice): Likewise.
* modules/unictype/bidicategory-test (Status, Notice): Likewise.

13 years agodoc: List the target platforms.
Bruno Haible [Mon, 16 May 2011 22:25:35 +0000 (00:25 +0200)]
doc: List the target platforms.

* doc/gnulib-intro.texi (Target Platforms): New section.
* doc/gnulib.texi (Introduction): Update menu.
* README (Portability guidelines): Refer to the new section. Update
statement about oldest supported environment. Remove rationale why
<errno.h>, <string.h>, <stdlib.h> are assumed. Update example of an
unportable C89 function.
Reported by Bastien Roucariès <roucaries.bastien@gmail.com> and
Charles Wilson <cygwin@cwilson.fastmail.fm>. Feedback from Paul Eggert.

13 years ago* build-aux/bootstrap (gnulib_tool): Handle symlink timestamps better.
Paul Eggert [Mon, 16 May 2011 16:17:48 +0000 (09:17 -0700)]
* build-aux/bootstrap (gnulib_tool): Handle symlink timestamps better.

13 years agointprops-tests: new module
Paul Eggert [Sun, 15 May 2011 16:37:21 +0000 (09:37 -0700)]
intprops-tests: new module

* modules/intprops-tests, tests/test-intprops.c: New files.

13 years agointprops: add safe, portable integer overflow checking
Paul Eggert [Sun, 15 May 2011 16:35:12 +0000 (09:35 -0700)]
intprops: add safe, portable integer overflow checking

* lib/intprops.h (_GL_INT_CONVERT, _GL_INT_TWOS_COMPLEMENT):
(_GL_INT_SIGNED, _GL_INT_MINIMUM, _GL_INT_MAXIMUM):
(_GL_SIGNED_INT_MINIMUM, INT_ADD_RANGE_OVERFLOW):
(INT__SUBTRACT__RANGE_OVERFLOW, INT_NEGATE_RANGE_OVERFLOW):
(INT_MULTIPLY_RANGE_OVERFLOW, INT_REMAINDER_RANGE_OVERFLOW):
(INT_LEFT_SHIFT_RANGE_OVERFLOW, _GL_ADD_OVERFLOW):
(_GL__SUBTRACT__OVERFLOW, _GL_MULTIPLY_OVERFLOW, _GL_DIVIDE_OVERFLOW):
(_GL_REMAINDER_OVERFLOW, _GL_UNSIGNED_NEG_MULTIPLE, INT_ADD_OVERFLOW):
(INT__SUBTRACT__OVERFLOW, INT_NEGATE_OVERFLOW, INT_MULTIPLY_OVERFLOW):
(INT_DIVIDE_OVERFLOW, INT_REMAINDER_OVERFLOW):
(INT_LEFT_SHIFT_OVERFLOW, _GL_BINARY_OP_OVERFLOW): New macros.

13 years agofnmatch: add a test for glibc's Bugzilla bug #12378
James Youngman [Thu, 12 May 2011 21:14:12 +0000 (22:14 +0100)]
fnmatch: add a test for glibc's Bugzilla bug #12378

* m4/fnmatch.m4: Use gnulib's fnmatch if the system fnmatch
doesn't allow the literal matching of a lone "[" (which is
required by POSIX).
* tests/test-fnmatch.c (main): Check that "[/b" matches itself.

13 years agofnmatch: sync glibc change fixing Bugzilla bug #12378
James Youngman [Thu, 12 May 2011 21:14:11 +0000 (22:14 +0100)]
fnmatch: sync glibc change fixing Bugzilla bug #12378

* lib/fnmatch_loop.c (FCT): When matching '[' keep track of
beginning and fall back to matching as normal character if the
string ends before the matching ']' is found.  This is what POSIX
requires.

13 years agoupdate from texinfo
Karl Berry [Fri, 13 May 2011 23:09:30 +0000 (16:09 -0700)]
update from texinfo

13 years agogetcwd-lgpl: relax test for FreeBSD
Eric Blake [Fri, 13 May 2011 16:26:08 +0000 (10:26 -0600)]
getcwd-lgpl: relax test for FreeBSD

getcwd(NULL, 1) mallocs a larger buffer on BSD, rather than failing
with ERANGE as on glibc.  This behavior difference is not worth
coding around, as it is an uncommon use of getcwd in the first place.

* doc/posix-functions/getcwd.texi (getcwd): Document portability
issue.
* tests/test-getcwd-lgpl.c (main): Relax test.
Reported by Matthias Bolte.

Signed-off-by: Eric Blake <eblake@redhat.com>
13 years agotest-fflush: silence compiler warning
Eric Blake [Thu, 12 May 2011 01:33:46 +0000 (19:33 -0600)]
test-fflush: silence compiler warning

Detected by clang.

* tests/test-fflush.c (main): Don't fclose a NULL pointer.

Signed-off-by: Eric Blake <eblake@redhat.com>
13 years agocanonicalize, canonicalize-lgpl: Avoid crash dialog on MacOS X.
Bruno Haible [Thu, 12 May 2011 00:10:44 +0000 (02:10 +0200)]
canonicalize, canonicalize-lgpl: Avoid crash dialog on MacOS X.

* m4/canonicalize.m4 (gl_FUNC_REALPATH_WORKS): Add GL_NOCRASH code.
* modules/canonicalize (Depends-on): Add 'nocrash'.
* modules/canonicalize-lgpl (Depends-on): Likewise.
* doc/posix-functions/realpath.texi: Update platforms list.
Reported by Ryan Schmidt <ryandesign@macports.org>.

13 years agogroup-member: Declare function in <unistd.h>.
Bruno Haible [Thu, 12 May 2011 00:03:07 +0000 (02:03 +0200)]
group-member: Declare function in <unistd.h>.

* lib/unistd.in.h (group_member): New declaration.
* lib/group-member.h: Remove file.
* lib/group-member.c: Include <unistd.h> instead of group-member.h.
* tests/test-unistd-c++.cc: Check signature of group_member.
* m4/group-member.m4 (gl_FUNC_GROUP_MEMBER): Require
gl_UNISTD_H_DEFAULTS. Set HAVE_GROUP_MEMBER.
* m4/unistd_h.m4 (gl_UNISTD_H): Check whether group_member is declared.
(gl_UNISTD_H_DEFAULTS): Initialize GNULIB_GROUP_MEMBER,
HAVE_GROUP_MEMBER.
* modules/group-member (Files): Remove lib/group-member.h.
(Depends-on): Add unistd. Specify conditions.
(configure.ac): Invoke gl_UNISTD_MODULE_INDICATOR.
(Include): Change to <unistd.h>.
* modules/unistd (Makefile.am): Substitute GNULIB_GROUP_MEMBER,
HAVE_GROUP_MEMBER.
* NEWS: Mention the change.
* lib/euidaccess.c: Don't include group-member.h.

13 years agogroup-member: Document module.
Bruno Haible [Thu, 12 May 2011 00:01:42 +0000 (02:01 +0200)]
group-member: Document module.

* doc/glibc-functions/group_member.texi: Mention the 'group-member'
module.

13 years agofclose: Fix mistake earlier today.
Bruno Haible [Thu, 12 May 2011 00:01:00 +0000 (02:01 +0200)]
fclose: Fix mistake earlier today.

* lib/fclose.c (rpl_fclose): Don't assume that EOF < 0.

13 years agofclose: preserve fflush errors
Eric Blake [Wed, 11 May 2011 17:35:06 +0000 (11:35 -0600)]
fclose: preserve fflush errors

Caught by gcc -Werror=unused-but-set-variable.

* lib/fclose.c (rpl_fclose): Don't lose fflush errors.
Reported by Jim Meyering.

Signed-off-by: Eric Blake <eblake@redhat.com>
13 years agobootstrap: support a prereq of 'rpcgen -' on RHEL5
Eric Blake [Wed, 11 May 2011 17:21:01 +0000 (11:21 -0600)]
bootstrap: support a prereq of 'rpcgen -' on RHEL5

On RHEL 5, 'rpcgen --version' spews usage to stderr with status 1,
rather than a version string to stdout with status 0.  But libvirt
merely requires a prereq of 'rpcgen -' - that is, it must exist,
but need not have a well-behaved --version.

* build-aux/bootstrap (check_versions): When no specific version
is required, merely check that the app produces an exit status
that indicates its existence.

Signed-off-by: Eric Blake <eblake@redhat.com>
13 years agomaint.mk: drop redundant check
Eric Blake [Wed, 11 May 2011 14:22:18 +0000 (08:22 -0600)]
maint.mk: drop redundant check

* top/maint.mk (sc_the_the): Delete; sc_prohibit_doubled_word does
the same but better.

Signed-off-by: Eric Blake <eblake@redhat.com>
13 years agoautoupdate
Karl Berry [Wed, 11 May 2011 13:58:35 +0000 (06:58 -0700)]
autoupdate

13 years agofclose: Fix possible link error.
Bruno Haible [Wed, 11 May 2011 11:14:46 +0000 (13:14 +0200)]
fclose: Fix possible link error.

* lib/fclose.c (rpl_fclose): Invoke _gl_unregister_fd, not
unregister_shadow_fd. Improve comments.
* lib/sockets.c (close_fd_maybe_socket): Add comments. Reported by
Eric Blake.

13 years agomaint.mk: adjust previous commit
Jim Meyering [Wed, 11 May 2011 08:41:23 +0000 (10:41 +0200)]
maint.mk: adjust previous commit

Change "misc" in names to "undesirable_word_seq".
"misc" is rarely desirable, in a name ;-)
Define and use ignore_undesirable_word_sequence_RE_ properly.

13 years agomaint.mk: improve "can not" detection and generalize rule name
Jim Meyering [Wed, 11 May 2011 08:07:32 +0000 (10:07 +0200)]
maint.mk: improve "can not" detection and generalize rule name

* top/maint.mk (sc_prohibit_misc): Renamed from sc_prohibit_can_not,
since we'll probably add a few more word pairs here.
Use the same technique as in sc_prohibit_doubled_word, so that
we recognize "can not" also when the words are separated by a newline.
Suggested by Eric Blake.
(perl_filename_lineno_text_): Define.  Factored out of...
(prohibit_doubled_word_): ...here.  Use the new definition.
(prohibit_misc_): New var.  Use it here, too.
(prohibit_misc_RE_): New overridable variable.

13 years agofclose: avoid double close race when possible
Eric Blake [Tue, 10 May 2011 21:28:48 +0000 (15:28 -0600)]
fclose: avoid double close race when possible

Calling close(fileno(fp)) prior to fclose(fp) is racy in a
multi-threaded application - some other thread could open a new file,
which is then inadvertently closed by the fclose that we thought
should fail with EBADF.  For mingw, this is no worse than the race
already present in close_fd_maybe_socket for calling closesocket()
prior to _close(), but for all other platforms, we might as well be
nice and avoid the race.

* lib/fclose.c (rpl_fclose): Rewrite to avoid double-close race on
all but WINDOWS_SOCKETS.

Signed-off-by: Eric Blake <eblake@redhat.com>
13 years agoopenat: correct new comment
Bastien Roucariès [Tue, 10 May 2011 09:30:19 +0000 (11:30 +0200)]
openat: correct new comment

* lib/openat-proc.c (openat_proc_name): Correct the comment.

13 years agoopenat: add comments
Jim Meyering [Tue, 10 May 2011 07:46:59 +0000 (09:46 +0200)]
openat: add comments

* lib/openat-proc.c (openat_proc_name): Add comments,
mostly from Eric Blake.

13 years agoopenat: reduce syscalls in first probe of /proc
Eric Blake [Mon, 9 May 2011 22:27:35 +0000 (16:27 -0600)]
openat: reduce syscalls in first probe of /proc

open/access/close is cheaper than open/stat/stat/close.

Use O_DIRECTORY for safety.

* lib/openat-proc.c (openat_proc_name): Require that /proc/self/fd
be a directory.  Simplify the probe for .. bugs.
* modules/openat (Depends-on): Drop same-inode.
Reported by Bastien ROUCARIES.

Signed-off-by: Eric Blake <eblake@redhat.com>
13 years agoautoupdate
Karl Berry [Mon, 9 May 2011 18:06:26 +0000 (11:06 -0700)]
autoupdate

13 years agomaint.mk: change semantics/name of tight_scope variables
Jim Meyering [Mon, 9 May 2011 17:28:11 +0000 (19:28 +0200)]
maint.mk: change semantics/name of tight_scope variables

* top/maint.mk (_gl_TS_var_match, _gl_TS_function_match):
Rename variables to align with semantics that make them more useful.

13 years agomaint.mk: fix oops
Jim Meyering [Mon, 9 May 2011 11:29:38 +0000 (13:29 +0200)]
maint.mk: fix oops

* top/maint.mk (sc_tight_scope): Fix last change: missing backslash.

13 years agomaint.mk: tweak new rule's name not to impinge
Jim Meyering [Mon, 9 May 2011 11:27:55 +0000 (13:27 +0200)]
maint.mk: tweak new rule's name not to impinge

* top/maint.mk (_gl_tight_scope): Rename from sc_tight_scope-0.
(sc_tight_scope): Use new rule name rather than $@-0.

13 years agomaint.mk: add a syntax-check rule to ensure tightly-scoped symbols
Jim Meyering [Mon, 9 May 2011 10:30:55 +0000 (12:30 +0200)]
maint.mk: add a syntax-check rule to ensure tightly-scoped symbols

* top/maint.mk (sc_tight_scope): New rule.
(sc_tight_scope-0): New rule, ifdef'd out.
(_gl_TS_dir): Default.
(_gl_TS_unmarked_extern_functions, _gl_TS_function_regex): Define.
(_gl_TS_unmarked_extern_vars, _gl_TS_var_regex): Define.

13 years agogc: Remove gl_PREREQ_GC (not used).
Simon Josefsson [Mon, 9 May 2011 07:07:13 +0000 (09:07 +0200)]
gc: Remove gl_PREREQ_GC (not used).

* m4/gc.m4: Remove gl_PREREQ_GC (not used).  Reported by Bruno
Haible <bruno@clisp.org>.

13 years agoComments.
Bruno Haible [Sun, 8 May 2011 23:11:59 +0000 (01:11 +0200)]
Comments.

* m4/isnanf.m4: Add comment.
* m4/isnanl.m4: Likewise.

13 years agoglob: Remove obsolete macro.
Bruno Haible [Sun, 8 May 2011 18:39:07 +0000 (20:39 +0200)]
glob: Remove obsolete macro.

* m4/glob.m4 (gl_GLOB_SUBSTITUTE): Remove macro.

13 years agointprops: Sun C 5.11 supports __typeof__
Paul Eggert [Sun, 8 May 2011 19:28:01 +0000 (12:28 -0700)]
intprops: Sun C 5.11 supports __typeof__

* lib/intprops.h (_GL_HAVE___TYPEOF__): New macro, which is set
for either GCC 2 or later, as before, or for Sun C 5.11 or later,
which is new.
(_GL_SIGNED_TYPE_OR_EXPR): Use it.

13 years agointprops: switch to usual gnulib indenting and naming
Paul Eggert [Sun, 8 May 2011 18:53:53 +0000 (11:53 -0700)]
intprops: switch to usual gnulib indenting and naming

* lib/intprops.h (_GL_INTPROPS_H): Rename from GL_INTPROPS_H.
(_GL_SIGNED_TYPE_OR_EXPR): Rename from signed_type_or_expr__.
* tests/test-inttostr.c (IS_TIGHT): Adjust to above renaming.

13 years agomaint.mk: suppress "Entering/Leaving directory" diag in announcement
Jim Meyering [Sun, 8 May 2011 16:13:59 +0000 (18:13 +0200)]
maint.mk: suppress "Entering/Leaving directory" diag in announcement

* top/maint.mk (release-prep): Use make's --no-print-directory
option when generating the announcement.  This eliminates the
pesky "make[2]: Entering/Leaving directory" diagnostics in the
generated announcement template.

13 years agotzset: Fix gettimeofday wrapper on Solaris 2.6.
Bruno Haible [Sun, 8 May 2011 16:09:39 +0000 (18:09 +0200)]
tzset: Fix gettimeofday wrapper on Solaris 2.6.

* m4/tzset.m4 (gl_FUNC_TZSET_CLOBBER): When invoking
gl_GETTIMEOFDAY_REPLACE_LOCALTIME, also set REPLACE_GETTIMEOFDAY.

13 years agoignore-value, verify: Omit include files from lib_SOURCES.
Paul Eggert [Sat, 7 May 2011 21:29:30 +0000 (14:29 -0700)]
ignore-value, verify: Omit include files from lib_SOURCES.

* modules/ignore-value, modules/verify (Makefile.am):
Don't put ignore-value.h, or verify.h, into lib_SOURCES, as
that leads Automake to duplicate use of am__objects_... variables
in Makefile.in.  See
<http://lists.gnu.org/archive/html/emacs-devel/2011-05/msg00257.html>.

13 years agofclose: Simplify autoconf macro.
Bruno Haible [Sat, 7 May 2011 11:05:51 +0000 (13:05 +0200)]
fclose: Simplify autoconf macro.

* m4/fclose.m4 (gl_FUNC_FCLOSE): Assume gl_FUNC_FFLUSH_STDIN is
defined.

13 years agocanonicalize-lgpl: Fix autoconf macro ordering bug.
Bruno Haible [Sat, 7 May 2011 10:50:30 +0000 (12:50 +0200)]
canonicalize-lgpl: Fix autoconf macro ordering bug.

* m4/canonicalize.m4 (gl_CANONICALIZE_LGPL): Require
gl_STDLIB_H_DEFAULTS.

13 years agomaintainer-makefile: make sc_po_check easier to tune
Eric Blake [Fri, 6 May 2011 21:38:46 +0000 (15:38 -0600)]
maintainer-makefile: make sc_po_check easier to tune

* top/maint.mk (sc_po_check): Allow overriding which non-VC files
to probe for strings, such as an alternate location for gnulib.

Signed-off-by: Eric Blake <eblake@redhat.com>
13 years agofclose: guarantee behavior on seekable stdin
Eric Blake [Fri, 6 May 2011 19:01:17 +0000 (13:01 -0600)]
fclose: guarantee behavior on seekable stdin

* modules/fclose (Depends-on): Add fflush.
* doc/posix-functions/fclose.texi (fclose): Document this.
* tests/test-fclose.c (main): Make test for this unconditional.

Signed-off-by: Eric Blake <eblake@redhat.com>
13 years agofflush, fpurge: Relicense under LGPLv2+.
Bruno Haible [Fri, 6 May 2011 18:14:41 +0000 (20:14 +0200)]
fflush, fpurge: Relicense under LGPLv2+.

* modules/fflush (License): Change from LGPLv3+ to LGPLv2+.
* modules/fpurge (License): Likewise.
With permission from Eric Blake and Jim Meyering.
Suggested by Eric Blake.

13 years agorm exit module
Karl Berry [Fri, 6 May 2011 16:31:16 +0000 (09:31 -0700)]
rm exit module

13 years agomaint.mk: use info-gnu@ as the default only for a stable release
Jim Meyering [Fri, 6 May 2011 06:09:21 +0000 (08:09 +0200)]
maint.mk: use info-gnu@ as the default only for a stable release

* top/maint.mk: Don't default to info-gnu for alpha or beta releases.
For those, just use $(PACKAGE_BUGREPORT), in which case we don't have
to set the Mail-Followup-To header.  Prompted by Reuben Thomas in
http://thread.gmane.org/gmane.comp.lib.gnulib.bugs/26488