platform/upstream/glib.git
12 years agogdateparser: Delete unused debug print function
Colin Walters [Fri, 2 Nov 2012 00:10:46 +0000 (20:10 -0400)]
gdateparser: Delete unused debug print function

https://bugzilla.gnome.org/show_bug.cgi?id=687385

12 years agogdbusactiongroup: Add prototype for g_dbus_action_group_sync()
Colin Walters [Thu, 1 Nov 2012 23:43:02 +0000 (19:43 -0400)]
gdbusactiongroup: Add prototype for g_dbus_action_group_sync()

Even private functions that are actually called across compilation
units should have prototypes.  For g_dbus_action_group_sync(), create
one in gdbusactiongroup-private.h

https://bugzilla.gnome.org/show_bug.cgi?id=687385

12 years agotest-pipe-unix: Add missing include
Colin Walters [Thu, 1 Nov 2012 23:40:07 +0000 (19:40 -0400)]
test-pipe-unix: Add missing include

Fixes the build with -Werror=missing-prototypes.

https://bugzilla.gnome.org/show_bug.cgi?id=687385

12 years agogettext: Add missing include
Colin Walters [Thu, 1 Nov 2012 23:39:09 +0000 (19:39 -0400)]
gettext: Add missing include

Fixes the build with -Werror=missing-prototypes.

https://bugzilla.gnome.org/show_bug.cgi?id=687385

12 years agogcharset: Add header file for private API
Colin Walters [Thu, 1 Nov 2012 23:38:44 +0000 (19:38 -0400)]
gcharset: Add header file for private API

This fixes the build with -Werror=missing-prototypes.

https://bugzilla.gnome.org/show_bug.cgi?id=687385

12 years agogslice: Prototype G_ENABLE_DEBUG function that's part of ABI
Colin Walters [Thu, 1 Nov 2012 23:38:09 +0000 (19:38 -0400)]
gslice: Prototype G_ENABLE_DEBUG function that's part of ABI

Sadly, g_slice_debug_tree_statistics is conditionally part of the
public ABI.  We might as well make it conditionally part of the API as
well, even though this will require people actually using it to

https://bugzilla.gnome.org/show_bug.cgi?id=687385

12 years agog_check_setuid: Include glib-private.h
Colin Walters [Thu, 1 Nov 2012 23:37:57 +0000 (19:37 -0400)]
g_check_setuid: Include glib-private.h

Otherwise we fail to build with -Werror=missing-prototypes.

https://bugzilla.gnome.org/show_bug.cgi?id=687385

12 years agotests/1bitmutex: Hack to build with -Werror=missing-prototypes
Colin Walters [Thu, 1 Nov 2012 23:37:38 +0000 (19:37 -0400)]
tests/1bitmutex: Hack to build with -Werror=missing-prototypes

Admittedly, this could probably be better, but it builds.

https://bugzilla.gnome.org/show_bug.cgi?id=687385

12 years agoUse (void) for no parameters, not ()
Colin Walters [Thu, 1 Nov 2012 23:36:52 +0000 (19:36 -0400)]
Use (void) for no parameters, not ()

This ensures we build with -Werror=missing-parameter-type.

https://bugzilla.gnome.org/show_bug.cgi?id=687385

12 years agobuild: Prototype GType accessors for private classes
Colin Walters [Thu, 1 Nov 2012 23:36:15 +0000 (19:36 -0400)]
build: Prototype GType accessors for private classes

Otherwise we fail to build with -Werror=missing-prototypes.

https://bugzilla.gnome.org/show_bug.cgi?id=687385

12 years agobuild: Add missing "static" keyword where it should be used
Colin Walters [Thu, 1 Nov 2012 23:35:38 +0000 (19:35 -0400)]
build: Add missing "static" keyword where it should be used

Otherwise we fail to build with -Werror=missing-prototypes.

https://bugzilla.gnome.org/show_bug.cgi?id=687385

12 years agobuild: Ensure we #include header files for glib-genmarshal code
Colin Walters [Thu, 1 Nov 2012 23:36:41 +0000 (19:36 -0400)]
build: Ensure we #include header files for glib-genmarshal code

Otherwise we fail with -Werror=missing-prototypes.

https://bugzilla.gnome.org/show_bug.cgi?id=687385

12 years agobuild: Don't use C99 declarations
Colin Walters [Thu, 1 Nov 2012 23:40:41 +0000 (19:40 -0400)]
build: Don't use C99 declarations

Since GLib needs to compile with MSVC, we can't use them.  This fixes
compilation when using -Werror=declaration-after-statement.

https://bugzilla.gnome.org/show_bug.cgi?id=687385

12 years agoUse "Returns:" instead of the invalid "@returns" for annotating return values.
Robert Ancell [Wed, 31 Oct 2012 01:56:00 +0000 (14:56 +1300)]
Use "Returns:" instead of the invalid "@returns" for annotating return values.

https://bugzilla.gnome.org/show_bug.cgi?id=673229

12 years agotests/signals: Disable large enumeration value test that is failing on PPC64
Colin Walters [Thu, 25 Oct 2012 19:34:29 +0000 (15:34 -0400)]
tests/signals: Disable large enumeration value test that is failing on PPC64

Basically due to a combination of va_args semantics around
signed/unsigned ints, this test case fails on ppc64.  At the moment,
we have as yet to find any real-world consumer with such a large
enumeration value.

Unfortunately, the possible fixes for this are extremely invasive;
we would have to define a new enum API.

Given both of these facts, we believe it makes the most sense at the
current time to simply not test this. If we at a later time determine
there is such a real-world consumer, we can look at doing the
necessary fixes.

https://bugzilla.gnome.org/show_bug.cgi?id=686662

12 years agopo: Fix Makefile.in.in syntax
Rico Tzschichholz [Wed, 31 Oct 2012 12:14:06 +0000 (13:14 +0100)]
po: Fix Makefile.in.in syntax

Introduced by cff536c99ed22f83a09a1030c720f1555240390b

12 years agoGNetworkMonitor: add missing apostrophe to docstring
Will Thompson [Tue, 30 Oct 2012 17:16:33 +0000 (17:16 +0000)]
GNetworkMonitor: add missing apostrophe to docstring

12 years agoAdd annotations for g_filename_from_uri()
Vincent Untz [Tue, 27 Mar 2012 14:52:24 +0000 (16:52 +0200)]
Add annotations for g_filename_from_uri()

https://bugzilla.gnome.org/show_bug.cgi?id=672924

12 years agogmessages: fix minor typo in documentation
Aleksander Morgado [Tue, 30 Oct 2012 12:15:23 +0000 (13:15 +0100)]
gmessages: fix minor typo in documentation

12 years agoUpdated gujarati file
Sweta Kothari [Tue, 30 Oct 2012 10:51:42 +0000 (16:21 +0530)]
Updated gujarati file

12 years agogthread-posix: always use atomic pointer ops
Ryan Lortie [Mon, 29 Oct 2012 09:13:40 +0000 (10:13 +0100)]
gthread-posix: always use atomic pointer ops

On platforms where dependent loads can be reordered (alpha) and we have
exotic implementation of pthread_mutex_lock() it could be possible that
our implementation of g_mutex_lock() is unsafe.

Always use atomic operations to avoid this possibility.

https://bugzilla.gnome.org/show_bug.cgi?id=686191

12 years agog_dbus_connection_export_menu_model(): fix a crash
Ryan Lortie [Mon, 29 Oct 2012 08:09:49 +0000 (09:09 +0100)]
g_dbus_connection_export_menu_model(): fix a crash

Add some extra protection when 'preparing' a group that doesn't yet
contain any menus.  This can happen if you subscribe to a group that
doesn't yet exist.

It was possible to crash any application using
g_dbus_connection_export_menu_model() by requesting a non-existent
subscription group over the bus.

In practice this only happened in races -- where the proxy sees a group
that exists and queries it, but by the time it does, it's already gone.

https://bugzilla.gnome.org/show_bug.cgi?id=687089

12 years agogmain: Document constraints on waitpid
Paul Eggert [Mon, 29 Oct 2012 14:19:20 +0000 (10:19 -0400)]
gmain: Document constraints on waitpid

Applications that use glib should not invoke waitpid with a first
argument that is nonpositive, because when such a waitpid is run in
one thread and glib waits for a subprocess in another, there is a race
condition, and the former waitpid can reap a process that was intended
for the latter.  Mention this in the documentation for
g_child_watch_source_new, and in the diagnostic generated by
g_spawn_sync when its waitpid fails with errno equal to ECHILD.

Signed-off-by: Colin Walters <walters@verbum.org>
http://bugzilla.gnome.org/show_bug.cgi?id=687075

12 years agomkinstalldirs: Delete from version control
Colin Walters [Sun, 28 Oct 2012 13:46:35 +0000 (09:46 -0400)]
mkinstalldirs: Delete from version control

This is installed by automake.  By maintaining it in git, we create
merge conflicts as people build with different versions of automake.

Just use 'mkdir -p' instead in gettext.  Should be portable enough.

https://bugzilla.gnome.org/show_bug.cgi?id=686839

12 years agoREADME.in: Note undefined behavior with new g_type_init() -> ctor change
Colin Walters [Sat, 27 Oct 2012 16:28:14 +0000 (12:28 -0400)]
README.in: Note undefined behavior with new g_type_init() -> ctor change

https://bugzilla.gnome.org/show_bug.cgi?id=686822

12 years agogio/tests: Fix build with -Werror=format-security
Rico Tzschichholz [Sat, 27 Oct 2012 10:12:29 +0000 (12:12 +0200)]
gio/tests: Fix build with -Werror=format-security

12 years agogio/tests: Fix linker failure
Rico Tzschichholz [Sat, 27 Oct 2012 10:10:44 +0000 (12:10 +0200)]
gio/tests: Fix linker failure

Fixes 8c320d0c34ad5b3a31ac4f9024dba358a46ac725

12 years agogio: No need to specify SOURCES when name matches executable
Stef Walter [Fri, 26 Oct 2012 09:12:46 +0000 (11:12 +0200)]
gio: No need to specify SOURCES when name matches executable

Removes some of the duplication from Makefile.am

https://bugzilla.gnome.org/show_bug.cgi?id=686921

12 years agogio: Remove LDADD repetition from gio/tests Makefile
Stef Walter [Fri, 26 Oct 2012 08:42:17 +0000 (10:42 +0200)]
gio: Remove LDADD repetition from gio/tests Makefile

Using a global LDADD for the basic libraries to link.

https://bugzilla.gnome.org/show_bug.cgi?id=686921

12 years agogdbus: Allow GDBusObjectManagerClient to work on peer connections
Stef Walter [Fri, 26 Oct 2012 08:30:29 +0000 (10:30 +0200)]
gdbus: Allow GDBusObjectManagerClient to work on peer connections

Allow GDBusObjectManagerClient to work on peer to peer DBus
connections. Don't require that a unique bus name is available
for the object manager, if the owned bus name is NULL.

https://bugzilla.gnome.org/show_bug.cgi?id=686920

12 years agoFix typo in G_TEST_DBUS.
Matthew Barnes [Fri, 26 Oct 2012 17:20:29 +0000 (13:20 -0400)]
Fix typo in G_TEST_DBUS.

12 years agofile-info: catch thumbnail files in large directory as well
Cosimo Cecchi [Tue, 1 May 2012 02:39:57 +0000 (22:39 -0400)]
file-info: catch thumbnail files in large directory as well

When building the file attribute table info for local files, use
thumbnail paths in $XDG_CACHE_DIR/thumbnails/large in addition to
$XDG_CACHE_DIR/thumbnails/normal.

Failing to do this would cause an application that creates large
thumbnails by default to never find any value for
G_FILE_ATTRIBUTE_THUMBNAIL_PATH, with no
G_FILE_ATTRIBUTE_THUMBNAILING_FAILED set, which might cause the
application to either think thumbnailing is still in progress, or
blindly requeue thumbnail operations in a loop.

Large thumbnails are generally preferred, so we now default to the path
of a large thumbnail (in case both are present).

https://bugzilla.gnome.org/show_bug.cgi?id=686895

12 years agog_unix_signal_source_new: Allow SIGUSR1 and SIGUSR2
Colin Walters [Thu, 25 Oct 2012 22:21:59 +0000 (18:21 -0400)]
g_unix_signal_source_new: Allow SIGUSR1 and SIGUSR2

These are user defined, it makes sense to allow watching them.  This
is needed to port gnome-session and gdm over.

https://bugzilla.gnome.org/show_bug.cgi?id=686898

12 years agoGIOChannel: Add missing annotations
Martin Pitt [Thu, 25 Oct 2012 12:55:30 +0000 (14:55 +0200)]
GIOChannel: Add missing annotations

The various read and write methods have several out arguments which were not
previously marked as such. Also, as GIOChannel supports binary data with a NULL
encoding, the buffers need to be uint8 arrays instead of utf8 strings.

12 years agoRevert "Box GPollFD to make it introspectable"
Martin Pitt [Thu, 25 Oct 2012 09:27:39 +0000 (11:27 +0200)]
Revert "Box GPollFD to make it introspectable"

This reverts commit 932f4250b88a50059330a9df8224feeab6b0ffd7.

This got pushed accidentally and has not been accepted yet. It's also not clear
whether we want this in the first place.

https://bugzilla.gnome.org/show_bug.cgi?id=686797

12 years agoBox GPollFD to make it introspectable
Martin Pitt [Wed, 24 Oct 2012 14:30:37 +0000 (16:30 +0200)]
Box GPollFD to make it introspectable

https://bugzilla.gnome.org/show_bug.cgi?id=686797

12 years agoGIOChannel: Add missing (allow-none) annotations
Martin Pitt [Thu, 25 Oct 2012 05:51:40 +0000 (07:51 +0200)]
GIOChannel: Add missing (allow-none) annotations

g_io_channel_set_line_term() and g_io_channel_set_encoding() can get NULL
arguments, mark them as such.

12 years agogvariant: Make g_variant_new_from_bytes() public
Colin Walters [Tue, 23 Oct 2012 14:11:33 +0000 (16:11 +0200)]
gvariant: Make g_variant_new_from_bytes() public

Now that GBytes has been made public, we should make
g_variant_new_from_bytes() public too.

Add g_variant_get_data_as_bytes() to match.

https://bugzilla.gnome.org/show_bug.cgi?id=677062

12 years agoRelease GLib 2.35.1 2.35.1
Ryan Lortie [Mon, 22 Oct 2012 20:20:47 +0000 (22:20 +0200)]
Release GLib 2.35.1

12 years agoFix this cost -> the cost typos
Murray Cumming [Sun, 21 Oct 2012 11:08:01 +0000 (13:08 +0200)]
Fix this cost -> the cost typos

12 years agoUpdated Slovenian translation
Matej Urbančič [Fri, 19 Oct 2012 22:14:37 +0000 (00:14 +0200)]
Updated Slovenian translation

12 years agoAllow slightly too big poll duration in /socket/timed_wait test
Antoine Jacoutot [Fri, 19 Oct 2012 12:34:18 +0000 (14:34 +0200)]
Allow slightly too big poll duration in /socket/timed_wait test

Sometimes the poll duration in the /socket/timed_wait test is slightly
bigger than the requested 100000, causing failures like:

GLib-GIO:ERROR:socket.c:620:test_timed_wait:
    assertion failed (poll_duration < 110000): (110057 < 110000)

Adjust the test to allow some jitter in the "too high" direction.

https://bugzilla.gnome.org/show_bug.cgi?id=686458

12 years agoGBytes: deal with a corner-case
Matthias Clasen [Wed, 17 Oct 2012 01:07:13 +0000 (21:07 -0400)]
GBytes: deal with a corner-case

When calling g_bytes_unref_to_data on a GBytes with NULL data
we could end up with double-free or use-after-free issues.
https://bugzilla.gnome.org/show_bug.cgi?id=686091

12 years agoAdd a test for the previous fix
Matthias Clasen [Wed, 17 Oct 2012 01:16:50 +0000 (21:16 -0400)]
Add a test for the previous fix

Running tests/bytes with G_SLICE=always-malloc MALLOC_CHECK_=2
was aborting before.

12 years agoGBusNameVanishedCallback: document NULL connection
Ryan Lortie [Tue, 16 Oct 2012 16:29:22 +0000 (12:29 -0400)]
GBusNameVanishedCallback: document NULL connection

@connection can be NULL for this callback.  Document this possibility
and describe the circumstances.

https://bugzilla.gnome.org/show_bug.cgi?id=686231

12 years agoGMenuExporter: allow NULL bus on _name_vanished
Ryan Lortie [Tue, 16 Oct 2012 16:27:12 +0000 (12:27 -0400)]
GMenuExporter: allow NULL bus on _name_vanished

GBusNameVanishedCallback is called with a NULL GDBusConnection in the
case that the connection has vanished.  We were doing an assert to
verify that it was the same as we had exported the menu on and that
assert was failing.

https://bugzilla.gnome.org/show_bug.cgi?id=685995

12 years agoAdd note about g_type_init() to README
Ryan Lortie [Tue, 16 Oct 2012 13:40:45 +0000 (09:40 -0400)]
Add note about g_type_init() to README

12 years agoProperly deprecate g_type_init()
Ryan Lortie [Mon, 15 Oct 2012 16:01:01 +0000 (12:01 -0400)]
Properly deprecate g_type_init()

Now that all internal users of it are gone.

https://bugzilla.gnome.org/show_bug.cgi?id=686161

12 years agoRemove g_type_init() calls
Ryan Lortie [Mon, 15 Oct 2012 16:00:51 +0000 (12:00 -0400)]
Remove g_type_init() calls

Very many testcases, some GLib tools (resource compiler, etc) and
GApplication were calling g_type_init().

Remove those uses, as they are no longer required.

https://bugzilla.gnome.org/show_bug.cgi?id=686161

12 years agoConvert g_type_init() to a ctor
Ryan Lortie [Mon, 15 Oct 2012 15:53:25 +0000 (11:53 -0400)]
Convert g_type_init() to a ctor

Move the guts of g_type_init() into a ctor and turn g_type_init() itself
into a do-nothing function.

g_type_init_with_debug_flags() now ignores its arguments, but it has
always been possible to achieve the same effect via environment
variables.

https://bugzilla.gnome.org/show_bug.cgi?id=686161

12 years agoAdd missing semicolons to gobject_probes.d
Matthias Clasen [Mon, 15 Oct 2012 23:48:44 +0000 (19:48 -0400)]
Add missing semicolons to gobject_probes.d

Pointed out in
https://bugzilla.gnome.org/show_bug.cgi?id=686119

12 years agogdbus-testserver: port to pygi
Marc-Antoine Perennou [Sat, 6 Oct 2012 12:56:36 +0000 (14:56 +0200)]
gdbus-testserver: port to pygi

Signed-off-by: Marc-Antoine Perennou <Marc-Antoine@Perennou.com>
https://bugzilla.gnome.org/show_bug.cgi?id=685608

12 years agoAdd return-if-fail checks to g_application_hold/release
Matthias Clasen [Mon, 15 Oct 2012 23:34:36 +0000 (19:34 -0400)]
Add return-if-fail checks to g_application_hold/release

This is our common practice for public API.
https://bugzilla.gnome.org/show_bug.cgi?id=685208

12 years agog_settings_bind: use canonical property name
Ryan Lortie [Mon, 15 Oct 2012 23:28:28 +0000 (19:28 -0400)]
g_settings_bind: use canonical property name

We were using the user-passed value of the @property argument for
several purposes in g_settings_bind(): error messages, binding
uniqueness (ie: one-binding-per-property-per-object) and most
importantly, connecting to the detailed notify:: signal.

The user may pass a string like "property_name" when the property's
canonical name is "property-name".  g_object_class_find_property() will
find the property under these circumstances, but a connection to
"notify::property_name" will not notice notifies emitted for
"property-name".

We can solve this by using the user's string to perform the lookup and
then using pspec->name for everything after that.

https://bugzilla.gnome.org/show_bug.cgi?id=684882

12 years agoAdd Win32 versions of some new content type APIs
Matthias Clasen [Mon, 15 Oct 2012 23:22:50 +0000 (19:22 -0400)]
Add Win32 versions of some new content type APIs

This commit adds minimal versions of
g_content_type_get_symbolic_icon and
g_content_type_get_generic_icon_name.
https://bugzilla.gnome.org/show_bug.cgi?id=683642

12 years agoUpdated Danish translation
Ask H. Larsen [Mon, 15 Oct 2012 04:33:25 +0000 (06:33 +0200)]
Updated Danish translation

12 years agogsignal: really fix closure invalidation
Ryan Lortie [Sat, 13 Oct 2012 16:16:32 +0000 (12:16 -0400)]
gsignal: really fix closure invalidation

Commit 66b0d95f0ba1939882368b47b01f93289c42ae07 missed this part of the
patch.

12 years agogtimezone: Indentation adn comment fixes
Arnel A. Borja [Fri, 12 Oct 2012 16:58:05 +0000 (09:58 -0700)]
gtimezone: Indentation adn comment fixes

12 years agoBug 631382 - GTimeZone support for zoneinfo version 1
John Ralls [Mon, 24 Sep 2012 17:54:38 +0000 (10:54 -0700)]
Bug 631382 - GTimeZone support for zoneinfo version 1

12 years agoUpdated Norwegian bokmål translation.
Kjartan Maraas [Thu, 11 Oct 2012 18:27:30 +0000 (20:27 +0200)]
Updated Norwegian bokmål translation.

12 years agogtestdbus: correct documentation typos
Will Thompson [Tue, 9 Oct 2012 08:30:06 +0000 (09:30 +0100)]
gtestdbus: correct documentation typos

https://bugzilla.gnome.org/show_bug.cgi?id=685787

12 years agogio/tests: port from GSimpleAsyncResult to GTask
Dan Winship [Thu, 2 Aug 2012 19:45:24 +0000 (15:45 -0400)]
gio/tests: port from GSimpleAsyncResult to GTask

https://bugzilla.gnome.org/show_bug.cgi?id=661767

12 years agogio: port networking classes from GSimpleAsyncResult to GTask
Dan Winship [Thu, 2 Aug 2012 19:48:22 +0000 (15:48 -0400)]
gio: port networking classes from GSimpleAsyncResult to GTask

https://bugzilla.gnome.org/show_bug.cgi?id=661767

12 years agogio: port GAsyncInitable from GSimpleAsyncResult to GTask
Dan Winship [Thu, 2 Aug 2012 19:51:37 +0000 (15:51 -0400)]
gio: port GAsyncInitable from GSimpleAsyncResult to GTask

https://bugzilla.gnome.org/show_bug.cgi?id=661767

12 years agogio: deprecate gioscheduler, soft deprecate GSimpleAsyncResult
Dan Winship [Mon, 21 Nov 2011 14:19:56 +0000 (09:19 -0500)]
gio: deprecate gioscheduler, soft deprecate GSimpleAsyncResult

Reimplement gioscheduler in terms of GTask, and deprecate the original
gioscheduler methods. Update docs to point people to GTask rather than
gioscheduler and GSimpleAsyncResult, but don't actually formally
deprecate GSimpleAsyncResult yet.

https://bugzilla.gnome.org/show_bug.cgi?id=661767

12 years agogio: port GThreadedResolver from GSimpleAsyncResult to GTask
Dan Winship [Tue, 11 Oct 2011 18:57:35 +0000 (14:57 -0400)]
gio: port GThreadedResolver from GSimpleAsyncResult to GTask

https://bugzilla.gnome.org/show_bug.cgi?id=661767

12 years agoGTask: new GAsyncResult implementation / threaded task manager
Dan Winship [Wed, 5 Oct 2011 14:05:50 +0000 (10:05 -0400)]
GTask: new GAsyncResult implementation / threaded task manager

GTask is a replacement for GSimpleAsyncResult and GIOScheduler, that
also allows for making cancellable wrappers around non-cancellable
functions (as in GThreadedResolver).

https://bugzilla.gnome.org/show_bug.cgi?id=661767

12 years agoghash.c: fix docs
Dan Winship [Wed, 10 Oct 2012 14:02:39 +0000 (10:02 -0400)]
ghash.c: fix docs

A previous commit left an unclosed XML tag. Fix that.

12 years ago[gsignal] Fix closure invalidation
Ryan Lortie [Mon, 8 Oct 2012 22:20:24 +0000 (18:20 -0400)]
[gsignal] Fix closure invalidation

If the closure is invalidated we drop the ref on the signal handler
node, but if the signal is currently being dispatched, the ref could be
held elsewhere.

Flag that we no longer have an outstanding invalidation handler so that
we don't try to unregister ourselves when the other ref drops.

Add a testcase that catches this situation.

12 years agogdbus: minor documentation fix in g_dbus_interface_skeleton_has_connection()
Aleksander Morgado [Mon, 8 Oct 2012 08:03:43 +0000 (10:03 +0200)]
gdbus: minor documentation fix in g_dbus_interface_skeleton_has_connection()

Fixes https://bugzilla.gnome.org/show_bug.cgi?id=685697

12 years ago[gobject] set all properties before constructed()
Ryan Lortie [Mon, 8 Oct 2012 15:40:00 +0000 (11:40 -0400)]
[gobject] set all properties before constructed()

Move the constructed() call to happen after all of the properties are
set (not just the construct properties).

This is an incompatible change but we are making it under the belief
that it should be safe.  If this change impacts you in a negative way
please comment on the bug.

https://bugzilla.gnome.org/show_bug.cgi?id=685733

12 years agofix g_signal_connect_object() documentation
Ryan Lortie [Mon, 8 Oct 2012 15:20:07 +0000 (11:20 -0400)]
fix g_signal_connect_object() documentation

g_signal_connect_object() now works properly, so we can remove the note
in the docs about it being broken.

https://bugzilla.gnome.org/show_bug.cgi?id=118536

12 years ago[gsignal] fix up a crasher in previous commit
Ryan Lortie [Mon, 8 Oct 2012 15:18:00 +0000 (11:18 -0400)]
[gsignal] fix up a crasher in previous commit

The previous commit introduced a new variable in the Handler struct but
didn't initialise it.  This was causing some tests to crash.

https://bugzilla.gnome.org/show_bug.cgi?id=118536

12 years ago[gsignal] disconnect invalidated closures
Matthias Clasen [Wed, 23 Jun 2004 05:49:24 +0000 (01:49 -0400)]
[gsignal] disconnect invalidated closures

Modify gsignal to automatically disconnect a GClosure that becomes
invalid (in the g_closure_invalidate() sense).

Previously, when g_signal_connect_object() was used with a GObject as
the user_data and that object was destroyed, the handler would no longer
be called but the signal handler was itself was not disconnected (ie:
the bookkeeping data was kept around).

The main effect of this patch is that these signal handlers will now
be automatically disconnected (and fully freed).

The documentation for g_signal_connect_object() has anticipated this
change for over 10 years and has advised the following workaround when
disconnecting signal handlers connected with g_signal_connect_object():

 if (g_signal_handler_is_connected (instance, id))
   g_signal_handler_disconnect (instance, id);

If your code follows this practice then it will continue to work.

If your code never disconnects the signal handler then it was wasting
memory before (and this commit fixes that).

If your code unconditionally disconnects the signal handler then you
will start to see (harmless) g_critical() warnings about this and you
should fix them.

https://bugzilla.gnome.org/show_bug.cgi?id=118536

12 years agoghash: Add documentation
Jasper St. Pierre [Sun, 7 Oct 2012 23:10:36 +0000 (20:10 -0300)]
ghash: Add documentation

12 years agoglib.py: Remove old debugging code
Jasper St. Pierre [Wed, 26 Sep 2012 14:36:30 +0000 (11:36 -0300)]
glib.py: Remove old debugging code

This was accidentally copy/pasted from gobject.py

12 years agoGFileMonitor: thread-safety fix for non-default-main-context monitors
Dan Winship [Fri, 5 Oct 2012 13:32:24 +0000 (09:32 -0400)]
GFileMonitor: thread-safety fix for non-default-main-context monitors

When queuing events to another thread, we need a mutex around
priv->pending_file_changes and priv->pending_file_change_source.

https://bugzilla.gnome.org/show_bug.cgi?id=682950

12 years agocodegen: Install Python files in $(datadir)/glib-2.0, rather than $(libdir)
Colin Walters [Fri, 28 Sep 2012 00:11:27 +0000 (20:11 -0400)]
codegen: Install Python files in $(datadir)/glib-2.0, rather than $(libdir)

These files are actually architecture-indepdendent; using $(libdir)
for them means that /usr/bin/gdbus-codegen varies between
architectures, which is problematic for (mis)uses of multilib.

See https://bugzilla.redhat.com/show_bug.cgi?id=718404

https://bugzilla.gnome.org/show_bug.cgi?id=685012

12 years agocodegen: Explicitly close output
Colin Walters [Wed, 26 Sep 2012 19:44:33 +0000 (15:44 -0400)]
codegen: Explicitly close output

This is just cleaner rather than relying on the GC, and maybe
if we're lucky it will actually solve a problem.

https://bugzilla.gnome.org/show_bug.cgi?id=684909

12 years agoAdd GLIB_VERSION_2_36 and related
Dan Winship [Wed, 5 Sep 2012 13:44:24 +0000 (09:44 -0400)]
Add GLIB_VERSION_2_36 and related

12 years agogdbus-codegen: Update tests for new inheritance-semantics for Since
David Zeuthen [Wed, 3 Oct 2012 15:20:51 +0000 (11:20 -0400)]
gdbus-codegen: Update tests for new inheritance-semantics for Since

Signed-off-by: David Zeuthen <zeuthen@gmail.com>
12 years agogdbus-codegen: make members of an interface inherit the "Since" annotation
David Zeuthen [Wed, 3 Oct 2012 15:07:57 +0000 (11:07 -0400)]
gdbus-codegen: make members of an interface inherit the "Since" annotation

This is the expected (and sane) behavior - without this bug-fix you'd
have to add "Since" to every member of a newly added D-Bus interface.

Also show-case this in the codegen example.

Signed-off-by: David Zeuthen <zeuthen@gmail.com>
12 years agounicode: Add new tests for unicode 6.2
Christian Persch [Sat, 7 Jul 2012 22:24:58 +0000 (00:24 +0200)]
unicode: Add new tests for unicode 6.2

12 years agounicode: Update to unicode 6.2.0 beta
Christian Persch [Sat, 7 Jul 2012 21:56:18 +0000 (23:56 +0200)]
unicode: Update to unicode 6.2.0 beta

12 years agounicode: Add new line breaking class from unicode 6.2
Christian Persch [Sat, 7 Jul 2012 21:53:23 +0000 (23:53 +0200)]
unicode: Add new line breaking class from unicode 6.2

Regional Indicator (RI) is new in unicode 6.2.

https://bugzilla.gnome.org/show_bug.cgi?id=684912

12 years agounicode: Use g_assert_cmp{int,uint,hex} in tests
Christian Persch [Sat, 7 Jul 2012 22:21:59 +0000 (00:21 +0200)]
unicode: Use g_assert_cmp{int,uint,hex} in tests

This gives more info when the test goes wrong.

12 years agobuild: Post branch version bump
Christian Persch [Wed, 3 Oct 2012 11:58:04 +0000 (13:58 +0200)]
build: Post branch version bump

12 years agoReally fix the leak
Matthias Clasen [Sun, 30 Sep 2012 23:24:13 +0000 (19:24 -0400)]
Really fix the leak

Pointed out by Josselin Mouette.

12 years agoFix memory leaks in glib-compile resources
Matthias Clasen [Sun, 30 Sep 2012 18:07:37 +0000 (14:07 -0400)]
Fix memory leaks in glib-compile resources

Based on a patch by Josselin Mouette,
https://bugzilla.gnome.org/show_bug.cgi?id=685069

12 years agoCorrect to docs for g_strcmp0
Matthias Clasen [Sun, 30 Sep 2012 17:37:39 +0000 (13:37 -0400)]
Correct to docs for g_strcmp0

Don't promise to return -1 or 1 - we fall back to strcmp which
does not guarantee this.
https://bugzilla.gnome.org/show_bug.cgi?id=685037

12 years agoUpdated Lithuanian translation
Žygimantas Beručka [Sun, 30 Sep 2012 10:01:14 +0000 (13:01 +0300)]
Updated Lithuanian translation

12 years agoFix gdb script for new quark variable names
Benjamin Otte [Thu, 27 Sep 2012 13:31:46 +0000 (15:31 +0200)]
Fix gdb script for new quark variable names

... and don't spam stderr with exceptions if someone renames things
again.

Last but not least, keep the old names as a fallback, so that LD_PRELOAD
with an older libglib still works.

12 years agoUpdate config.h.win32(.in) and glibconfig.h.win32(.in)
Chun-wei Fan [Wed, 26 Sep 2012 09:47:52 +0000 (17:47 +0800)]
Update config.h.win32(.in) and glibconfig.h.win32(.in)

-Make config.h.win32(.in) have entries that more resembles the generated
 config.h.in
-Move the ALIGNOF_* #define's from glibconfig.h.win32(.in) to
 config.h.win32(.in), where they were supposed to be.

12 years agoVisual C++ property sheet updates
Chun-wei Fan [Wed, 26 Sep 2012 07:42:53 +0000 (15:42 +0800)]
Visual C++ property sheet updates

Make up for the missed GIO headers that need to be installed.

12 years ago[l10n] Updated Catalan (Valencian) translation
Carles Ferrando [Tue, 25 Sep 2012 23:50:49 +0000 (01:50 +0200)]
[l10n] Updated Catalan (Valencian) translation

12 years ago[l10n] Updated Italian translation.
Luca Ferretti [Tue, 25 Sep 2012 06:34:44 +0000 (08:34 +0200)]
[l10n] Updated Italian translation.

12 years agoUpdate Czech translation
Petr Kovar [Mon, 24 Sep 2012 16:23:18 +0000 (18:23 +0200)]
Update Czech translation

12 years agoBump version
Matthias Clasen [Mon, 24 Sep 2012 02:39:20 +0000 (22:39 -0400)]
Bump version

12 years ago2.34.0 2.34.0
Matthias Clasen [Mon, 24 Sep 2012 02:33:21 +0000 (22:33 -0400)]
2.34.0

12 years agoMore updates
Matthias Clasen [Mon, 24 Sep 2012 01:20:06 +0000 (21:20 -0400)]
More updates