2002-04-14 Jakub Jelinek <jakub@redhat.com>
* elf/dl-lookup.c (_dl_lookup_symbol): Move add_dependency call to
the end of the function. Pass original flags to recursive call if
add_dependency failed.
(_dl_lookup_versioned_symbol): Likewise.
2002-04-13 Jakub Jelinek <jakub@redhat.com>
* time/mktime.c (__mktime_internal): If year is 69, don't bail out
early, but check whether it overflowed afterwards.
* time/tst-mktime.c (main): Add new tests.
* debug/xtrace.sh: Fix program name in help message.
Patch by Roger Luethi <rl@hellgate.ch>.
2001-05-21 Bruno Haible <haible@clisp.cons.org>
* wcsmbs/mbrtowc.c (mbrtowc): Remove local variable 'flush', always
use 0 instead, and rely on the converter to do the flush.
* wcsmbs/tst-mbrtowc.c (utf8_test_1): New function, taken from
utf8_test.
(utf8_test_2, utf8_test_3): New function.
(utf8_test): Call utf8_test_1, utf8_test_2, utf8_test_3.
2000-09-30 Bruno Haible <haible@clisp.cons.org>
* tests-mbwc/tst_wcscmp.c (tst_wcscmp): Take the sign of ret.
* tests-mbwc/dat_wcscmp.c (tst_wcscmp_loc): Don't expect precise
return values. Only the sign matters.
2000-09-23 Bruno Haible <haible@clisp.cons.org>
* iconvdata/gbk.c (USE_PRIVATE_AREA): Define to 0.
(__gbk_to_ucs): Conditionalize private area mappings.
(__gbk_from_ucs4_tab9): Likewise.
(BODY for TO_LOOP): Likewise.
* iconvdata/testdata/GBK: Don't use characters not yet in Unicode.
* iconvdata/testdata/GBK..UTF-8: Likewise.
* iconvdata/tst-tables.sh: Enable GBK test.
2000-09-23 Bruno Haible <haible@clisp.cons.org>
* iconvdata/gbk.c (__gbk_to_ucs): Swap U+2014 and U+2015.
(__gbk_from_ucs4_tab4): Swap entries for U+2014 and U+2015.
(BODY for FROM_LOOP): Reject input > 0xFEA0, avoids out-of-bounds
array access.
* iconvdata/gbgbk.c (BODY for FROM_LOOP): Map 0xA844 to 0xA1AA.
* iconvdata/testdata/GBK..UTF8: Swap U+2014 and U+2015.
2000-09-23 Bruno Haible <haible@clisp.cons.org>
* iconvdata/johab.c (final_to_ucs): Fix typos.
(jamo_from_ucs_table): Likewise.
(BODY for FROM_LOOP): Map 0x5c to U+20A9. Reject ranges
0xD9E6..0xD9FE and 0xDEF2..0xDEFE.
(BODY for TO_LOOP): Map U+20A9 to 0x5c. Don't produce values in
the range 0xD9E6..0xD9FE.
* iconvdata/tst-tables.sh: Enable JOHAB testing.
2000-07-27 Bruno Haible <haible@clisp.cons.org>
* locale/C-ctype.c (_nl_C_LC_CTYPE): Swap the two names in
_NL_CTYPE_MAP_NAMES.
* locale/localeinfo.h (__TOW_toupper, __TOW_tolower): New enum values.
* wctype/wcfuncs.c (towlower, towupper): Use them.
* wctype/wcfuncs_l.c (__towlower_l, __towupper_l): Likewise.
* wctype/wctrans.c (wctrans): Likewise.
* wctype/wctrans_l.c (__wctrans_l): Likewise.
2000-07-27 Bruno Haible <haible@clisp.cons.org>
* posix/fnmatch.c (is_char_class): Fix logic. Avoid buffer overrun.
Use the optimized version of this function outside glibc as well.
2000-07-07 Ulrich Drepper <drepper@redhat.com>
* locale/programs/locale.c (write_locales): Don't simply add all
directories found in the subdir, test whether at least the
LC_CTYPE file is in there.
* stdlib/strfmon.c: Move somewhat closer to the standard wrt to
handling the internation currency symbol. POSIX says that the
fourth character is used to separate the currency symbol from the
value. Therefore it does not have to be printed. But we cannot
remove the space if the currency symbol is printed before the
number since this is what many locales expect.
* localedata/tests-mbwc/dat_strfmon.c: Remove #ifdefs introduced to
work around DEM problem.
* localedata/tst-fmon.data: Change back entries with DEM.
2000-06-29 Ulrich Drepper <drepper@redhat.com>
* tst-fmon.data: Until we know what is correct, but the spaces
after DEM back in.
* tests-mbwc/dat_mbstowcs.c: Fix various bugs. Comment out
unresolved issues.
* tests-mbwc/dat_mbtowc.c: Likewise.
* tests-mbwc/dat_strcoll.c: Likewise.
* tests-mbwc/dat_strfmon.c: Likewise.
* tests-mbwc/dat_strxfrm.c: Likewise.
* tests-mbwc/tst_mbtowc.c: Likewise.
($(addsuffix .out,$(addprefix $(objpfx),$(locale_test_suite)))):
* wcsmbs/mbrtowc.c: Set flush to 1 for conversion function calls
depending on whether the input string is empty or not.
* wcsmbs/mbrtowc.c: Calling function without input means clearing the
given state and not avoiding this by modifying a local object.
* wcsmbs/wcrtomb.c: Likewise.
* iconv/skeleton.c: If no EMIT_SHIFT_TO_INIT is defined clear state
object since some incomplete characters might be in there.
* locale/programs/ld-ctype.c (ctype_finish): Take all characters from
the input charset into account when generating the hash table.
(allocate_arrays): Correct setting default width. Not all empty slots
in the table are filled, only those not covert explicitly by the
locale description and in the charset.
* stdio-common/vfscanf.c: Make sure to always return WEOF and EOF for
wide character version.
For %C handling, test correct pointer variable for NULL.
* wcsmbs/wctob.c: Handle WEOF special.
* wcsmbs/wcwidth.h: 0xff in width array means invalid character.
* wctype/wctype.h: Protect gcc-isms with __extension__. Avoid
always-true test to avoid warning.
comment them out for now.
(subdir-dirs): New for tests-mbwc, add also vpaths.
* Makefile (tests): Add tests from tests-mbwc subdirectory,
comment them out for now.
(subdir-dirs): New for tests-mbwc, add also vpaths.