* build-aux/gendocs.sh (version):
* doc/gendocs_template:
* doc/gendocs_template_min:
* doc/gnulib.texi:
* lib/version-etc.c (COPYRIGHT_YEAR):
Update copyright dates by hand in templates and the like.
* all files: Run 'make update-copyright'.
* build-aux/gendocs.sh (version):
* doc/gendocs_template:
* doc/gendocs_template_min:
* doc/gnulib.texi:
* lib/version-etc.c (COPYRIGHT_YEAR):
Update copyright dates by hand in templates and the like.
* all files: Run 'make update-copyright'.
The data passed to re_search by the test for glibc bug 15078 is a
multi-character collating element followed by a single character.
According to POSIX, "It is unspecified whether a non-matching list
expression matches a multi-character collating element that is not
matched by any of the expressions."
One of differences between glibc and gnulib implementations of
re_search is that glibc re_search matches multi-character
collating elements in that case while gnulib re_search doesn't.
Since both re_search implementations conform to standard, change
the test to allow glibc re_search behavior.
* tests/test-regex.c (main): In test for glibc bug 15078, reformat
re_search input data to make the multi-character collating element
in it clearly visible, and treat re_search return code 0 as valid.
* m4/regex.m4 (gl_REGEX): Likewise.
test-regex failed on mingw; and I traced it in gdb to an instance
of init_dfa() setting dfa->is_utf8 to 0 in spite of setlocale()
claiming success for "en_US.UTF-8". test-wcwidth already has
precedent for skipping utf-8 tests where the system (or gnulib
setlocale replacement, in this case) lies about utf-8 support.
* modules/regex-tests (Depends-on): Add localcharset.
* tests/test-regex.c (main): Use it to skip test on mingw.
Signed-off-by: Eric Blake <eblake@redhat.com>
On mingw (at least, when cross-compiling with Fedora 18's
mingw32-headers-2.0.999-0.15.trunk.20121110.fc18.noarch build),
compilation of test-regex fails:
test-regex.c: In function 'main':
test-regex.c:42:11: error: 'SIGALRM' undeclared (first use in this function)
test-regex.c:42:11: note: each undeclared identifier is reported only
once for each function it appears in
test-regex.c:43:3: warning: implicit declaration of function 'alarm'
It turns out that recent mingw64 added an export of alarm() and
SIGALRM, but guarded their declarations behind __USE_MINGW_ALARM
(default off, and with alarm() only in the non-standard <io.h>);
so the m4 tests were setting HAVE_ALARM to 1 based on link success
but then failing to compile.
* doc/posix-functions/alarm.texi (alarm): Document that alarm
exists but still doesn't work in newer mingw.
* m4/frexp.m4 (gl_FUNC_FREXP_WORKS): Check for alarm declaration,
not existence. Ensure SIGALRM is not trapped.
* m4/mktime.m4 (gl_FUNC_MKTIME): Likewise.
* m4/regex.m4 (gl_REGEX): Likewise.
* m4/remainderf.m4 (gl_FUNC_REMAINDERF_WORKS): Likewise.
* tests/test-regex.c (main): Use correct probe for alarm.
Signed-off-by: Eric Blake <eblake@redhat.com>
* tests/test-regex.c: Don't include regex.h twice. Include
string.h, to declare memset. Christensen's report also mentioned
this issue.
* m4/regex.m4 (gl_REGEX): Keep test program more in sync with
test-regex.c, to avoid future problems like this. Remove
AC_INCLUDES_DEFAULT. Include <string.h>. Don't include <regex.h>
twice.
