* configure.in:
* glib/gthread.h: Prevent the compiler from warning about breaking
strict-aliasing rules when using gcc 4.3 and G_LOCK on C sources.
svn path=/trunk/; revision=6691
2008-03-12 Tor Lillqvist <tml@novell.com>
Bug 520914 - win_iconv doesn't support UCS-2
* glib/win_iconv.c: Make UCS-2 just an alias for
UTF-16. Technically this is wrong of course, but shouldn't matter
an awful lot in practice.
svn path=/trunk/; revision=6686
2008-03-11 Murray Cumming <murrayc@murrayc.com>
Bug 521591 – g_markup_parse_context_parse() creates GError message that
is invalid UTF8.
* glib/gmarkup.c (set_error): Make sure that the GError::message is
valid UTF-8 even if it is complaining about invalid UTF-8 in the
markup text, using _g_utf8_make_valid().
svn path=/trunk/; revision=6680
2008-03-07 Alexander Larsson <alexl@redhat.com>
* glocalfile.c:
(g_local_file_query_filesystem_info):
Use struct statfs.f_fstypename if availible (e.g. on OpenBSD)
Patch from Jasper Lievisse Adriaanse
svn path=/trunk/; revision=6638
2008-03-04 Alexander Larsson <alexl@redhat.com>
* gfile.c:
Remove deprecated symbols we kept for one release.
2008-03-04 Alexander Larsson <alexl@redhat.com>
* glib/gurifuncs.c:
Remove deprecated symbols we kept for one release.
svn path=/trunk/; revision=6621
2008-03-03 Matthias Clasen <mclasen@redhat.com>
* glib/gutils.h: Add a version of G_INLINE_FUNC for
__GNUC__ && __GNUC_STDC_INLINE__, patch by Jakub Jelinek
svn path=/trunk/; revision=6616
2008-03-03 Tor Lillqvist <tml@novell.com>
* glib/gspawn-win32.c (read_helper_report): Must set the GError
also in the unexpected EOF case.
(do_spawn_with_pipes): Must protect also new_argv[0].
svn path=/trunk/; revision=6612
2008-02-29 Tor Lillqvist <tml@novell.com>
* glib/win_iconv.c (name_to_codepage): Add some GNU libiconv
compatibility: Recognize "" and "char" as aliases for the current
locale's charset. (We use the system ANSI codepage as returned by
GetACP().) Recognize "wchar_t" as an alias for UTF-16LE.
svn path=/trunk/; revision=6604
2008-02-26 Tor Lillqvist <tml@novell.com>
* glib/gmain.c (g_poll): Further patch by Vlad Grecescu: Drop
the code path that called WaitMessage(), as WaitMessage()
doesn't offer any chance for APCs to run. Instead just use the
code path with MsgWaitForMultipleObjectsEx() even for the
wait-only-for-messages case. (#517484)
svn path=/trunk/; revision=6597
2008-02-25 Alexander Larsson <alexl@redhat.com>
* glib/glib.symbols:
* glib/gurifuncs.[ch]:
Rename g_uri_get_scheme to g_uri_parse_scheme.
Keep g_uri_get_scheme() symbol for this
unstable release to avoid breaking to many apps.
svn path=/trunk/; revision=6588
2008-02-25 Alexander Larsson <alexl@redhat.com>
* glib/gurifuncs.[ch]:
Rename g_uri_get_scheme to g_uri_parse_scheme.
Keep g_uri_get_scheme() symbol for this
unstable release to avoid breaking to many apps.
svn path=/trunk/; revision=6587
2008-02-24 Tor Lillqvist <tml@novell.com>
* glib/gutils.c (_glib_get_installation_directory): New internal function.
* glib/gspawn-win32.c: When spawning the helper process, use an
explicit full path. (#518292)
* glib/gspawn-win32.c
* glib/gspawn-win32-helper.c: Fix race condition when using the
helper process: When the helper process writes the handle of the
actual started user process to the parent process, it must be
duplicated in the parent process with DuplicateHandle() so that it
is a valid handle in that process. However, if the helper process
has happened to exit before the DuplicateHandle() call, the
duplication will fail. Thus we must synchronise the helper
process's exit. Use another pipe for this.
Take care not to inherit the writing end of this pipe to the
helper process. Also, in the helper process, take care not to
inherit either of the pipes used for communication with the parent
process to the started user process.
svn path=/trunk/; revision=6575
2008-02-24 Tor Lillqvist <tml@novell.com>
* glib/gmain.c (g_poll) [Win32]: Use alertable wait functions so
that I/O completion routines or user-mode Asynchronous Procedure
Calls can be run. (#517484, Vlad Grecescu)
svn path=/trunk/; revision=6571
2008-02-24 Tor Lillqvist <tml@novell.com>
* glib/gutils.h: Mention G_WIN32_DLLMAIN_FOR_DLL_NAME() will be
deprecated in the future.
* glib/gutils.c: Drop use of G_WIN32_DLLMAIN_FOR_DLL_NAME(). Use a
minimal DllMain() instead that just saves the DLL handle.
(g_win32_get_system_data_dirs_for_module, _glib_get_locale_dir)
(get_module_share_dir): Use
g_win32_get_package_installation_directory_of_module().
svn path=/trunk/; revision=6570
2008-02-24 Tor Lillqvist <tml@novell.com>
* glib/gwin32.c
(g_win32_get_package_installation_directory_of_module): New
function. Supersedes g_win32_get_package_installation_directory()
and g_win32_get_package_installation_directory().
It makes more sense to have the function for this functionality
take a HMODULE as parameter instead of DLL name. The typical use
scenario has been to have a DllMain() function that retrieves the
full pathname for the DLL in question, and saves just the basename
of that. Then later code passes that saved dll basename to
g_win32_get_package_installation_directory(), which retrieves the
corresponding DLL handle, and then retrieves up its full
pathname. (Which DLlMain() already had.) It is less convoluted to
have a DllMain() that just saves the DLL handle, and then when
needed call this function to get the corresponding installation
folder.
(get_package_directory_from_module): Use
g_win32_get_package_installation_directory_of_module().
(g_win32_get_package_installation_directory)
(g_win32_get_package_installation_subdirectory): Mention these
functions will be deprecated and recommend using
g_win32_get_package_installation_directory_of_module() instead.
* glib/gwin32.h: Declare
g_win32_get_package_installation_directory_of_module().
* glib/glib.symbols: Add it.
svn path=/trunk/; revision=6569
2008-02-23 Matthias Clasen <mclasen@redhat.com>
* tests/testglib.c: Don't test user directories for being
non-null. (#517084, Yevgen Muntyan)
svn path=/trunk/; revision=6562
2008-02-20 Tor Lillqvist <tml@novell.com>
* glib/gtestutils.c (g_test_trap_fork) [Win32]: Change the
g_error() to g_message() to avoid stopping on warnings. At least
now testglib runs to completion and the old tests in it get
exercised even if the newfangled ones don't.
(g_test_trap_assertions) [Win32]: Bypass on Windows.
svn path=/trunk/; revision=6542
2008-02-17 Hans Breuer <hans@breuer.org>
* glib/gutils.c : define CSIDL_MYPICTURES if not available
* gthread/gthread-win32.c : use G_STRFUNC instead of compiler
specific __FUNCTION__
svn path=/trunk/; revision=6526
2008-02-13 Tor Lillqvist <tml@novell.com>
* glib/gwin32.c: Doc change: Deprecate passing anything but NULL
for the "package" parameter to
g_win32_get_package_installation_directory() and
g_win32_get_package_installation_subdirectory().
svn path=/trunk/; revision=6508
2008-02-10 Matthias Clasen <mclasen@redhat.com>
* glib/gtestutils.h: Make the g_test_add macro work with
gcc 4.3
* tests/gobject/paramspec-test.c: Adapt to recent changes in
GParamGType initialization.
svn path=/trunk/; revision=6500
2008-02-07 Tor Lillqvist <tml@novell.com>
* configure.in: Unfortunately the mingw implementations of
C99-style snprintf and vsnprintf don't seem to be quite good
enough, at least not in mingw-runtime-3.14. I don't know exactly
what the problem is, but it is related to floating point
formatting and decimal point vs. comma, and the symptoms show up
in some dialogs in GIMP, presumably also elsewhere. The simple
tests in AC_FUNC_VSNPRINTF_C99 and AC_FUNC_SNPRINTF_C99 aren't
rigorous enough to notice, though. So preset
ac_cv_func_vsnprintf_c99 and ac_cv_func_snprintf_c99 to "no".
svn path=/trunk/; revision=6482
2008-02-07 12:58:54 Tim Janik <timj@imendio.com>
* Makefile.am: fixed build order to build gobject after gmodule gthread
so gobject tests can be threaded.
svn path=/trunk/; revision=6480
2008-02-07 Ryan Lortie <desrt@desrt.ca>
* glib/gmessages.h (g_error): add for(;;); after the g_log call so
that GCC stops issuing false warnings about reachability Bug #514920
svn path=/trunk/; revision=6476
2008-02-06 Behdad Esfahbod <behdad@gnome.org>
* glib/pltcheck.sh: Skip g_bit_*(). Inline functions may end up with
a local plt if the compiler doesn't support what we want. Bug #514702
svn path=/trunk/; revision=6468
2008-02-02 Jonathon Jongsma <jjongsma@gnome.org>
* gio/gbufferedoutputstream.c:
* gio/gbufferedoutputstream.h: modify the new_sized() constructor to take a
gsize param instead of guint to match the GBufferedInputStream constructor.
svn path=/trunk/; revision=6448
2008-01-31 Michael Natterer <mitch@imendio.com>
* glib/gmem.c: use %G_GSIZE_FORMAT instead of %lu since sizes have
changed from gulong to gsize in this file.
svn path=/trunk/; revision=6432
2008-01-29 Christian Persch <chpe@gnome.org>
* docs/reference/glib/tmpl/macros_misc.sgml: G_GNUC_[PRETTY_]FUNCTION
are deprecated since 2.16, not 2.14.
svn path=/trunk/; revision=6419
2008-01-29 14:58:31 Tim Janik <timj@imendio.com>
* glib/gmem.[hc]: changed size argument type from gulong to gsize as
discussed on gtk-devel-list:
http://mail.gnome.org/archives/gtk-devel-list/2007-March/msg00062.html
this should be ABI compatible on all platforms except win64 for which
no ABI binding port exists yet.
svn path=/trunk/; revision=6413
2008-01-29 Sebastian Wilhelmi <wilhelmi@google.com>
* tests/threadpool-test.c (test_thread_pools): Grab
thread_counter_pools LOCK when increasing
leftover_task_counter. Fixes race in test. (#512624, Simon Murray)
svn path=/trunk/; revision=6406
2008-01-27 Matthias Clasen <mclasen@redhat.com>
* configure.in: Fix the pcre Unicode test to work with
LDFLAGS=-Wl,--as-needed. (#484261, Mark Lee)
svn path=/trunk/; revision=6387
2008-01-27 Murray Cumming <murrayc@murrayc.com>
* gio/gfile.c: (g_file_replace_contents),
(g_file_replace_contents_finish): Document that the new_etags output
gchar* should be freed.
svn path=/trunk/; revision=6383
2008-01-25 Loïc Minier <lool@dooz.org>
* glib/goption.c: (group_has_visible_entries),
(group_list_has_visible_entires), (g_option_context_get_help): Pass
context down the implementation to check for the main_group.
Bug #510292.
* glib/tests/option-context.c:
Don't set G_OPTION_FLAG_IN_MAIN in main_entries
(group_captions): only create group when actually adding it to the
context; add an exit(0) to make sure the test succeeds.
svn path=/trunk/; revision=6377
2008-01-23 Jens Granseuer <jensgr@gmx.net>
* glib/gtestutils.c: (g_test_trap_fork):
* glib/tests/testing.c: (test_assertions): Only declare variables at
the beginning of a code block. Bug #511654.
svn path=/trunk/; revision=6361
2008-01-20 Murray Cumming <murrayc@murrayc.com>
* glib/gchecksum.c: (g_checksum_update): Accept -1 for the data
length if the data is a null-terminated string. Bug #510855.
svn path=/trunk/; revision=6340
2008-01-21 Alexander Larsson <alexl@redhat.com>
* inotify/Makefile.am:
* inotify/inotify-helper.c:
* inotify/inotify-kernel.c:
* inotify/inotify-path.c:
* inotify/local_inotify.h: Removed.
* inotify/local_inotify_syscalls.h: Removed.
Removed the included copies of the inotify
headers. We now only use the <sys/inotify.h>
header which exists on modern systems.
This fixes problems on ARM and SH5 (#510448)
but is also generally much cleaner and future
safe. For instance, if other OSes add support
for inotify it should "just work".
svn path=/trunk/; revision=6338
2008-01-18 16:51:23 Tim Janik <timj@imendio.com>
* glib/gutils.h: don't define __GNUC_PREREQ which is not in the glib
namespace. for gcc, define G_INLINE_FUNC to "static inline" as with
all other C compilers, because newer GCC versions incompatibly
changed "extern inline" semantics.
svn path=/trunk/; revision=6333
2008-01-18 Murray Cumming <murrayc@murrayc.com>
* glib/gfileutils.c:
* glib/gsequence.c:
* glib/gstring.c: Fixed some minor typos in the documentation.
svn path=/trunk/; revision=6332
* glib/gchecksum.c (md5_sum_update): Fix another bug (which
doesn't affect the results, but may cause it to read bad memory).
svn path=/trunk/; revision=6323
* glib/gchecksum.c (md5_sum_update): fix this; the previous code
gave the wrong md5sum when called in certain ways with buffers
larger than 64 bytes.
(g_checksum_update): remove the unnecessary "length > 1"
restriction
* tests/checksum-test.c: Rewrite this to be much more exhaustive
(and in particular to test the md5_sum_update bugfix).
svn path=/trunk/; revision=6314
