Mike Gorse [Tue, 18 Sep 2012 00:06:40 +0000 (19:06 -0500)]
Fix an errant g_object_unref
If an object is deregistered but not in the cache, then only unref it if
it was found in the queue of objects to be added (ie, it is possible
that it is leased, rather than queued to be cached, in which case it
should not be unrefed).
William Jon McCann [Fri, 14 Sep 2012 21:29:10 +0000 (17:29 -0400)]
Don't leak pending calls
https://bugzilla.gnome.org/show_bug.cgi?id=684077
William Jon McCann [Fri, 14 Sep 2012 21:11:37 +0000 (17:11 -0400)]
Don't leak message
https://bugzilla.gnome.org/show_bug.cgi?id=684077
William Jon McCann [Fri, 14 Sep 2012 20:39:58 +0000 (16:39 -0400)]
Don't leak strings
https://bugzilla.gnome.org/show_bug.cgi?id=684077
William Jon McCann [Fri, 14 Sep 2012 20:28:48 +0000 (16:28 -0400)]
Close the connection when we are done with it
Since it is no longer a shared connection
https://bugzilla.gnome.org/show_bug.cgi?id=684077
William Jon McCann [Fri, 14 Sep 2012 18:20:51 +0000 (14:20 -0400)]
Make cleanup more robust
https://bugzilla.gnome.org/show_bug.cgi?id=684042
Cosimo Cecchi [Thu, 13 Sep 2012 17:37:50 +0000 (13:37 -0400)]
atk-adaptor: plug some refcount leaks
cache->add_traversal holds GObject references, so when we remove/pop
items from it, we should also drop the reference, unless it's transfered
to another queue.
https://bugzilla.gnome.org/show_bug.cgi?id=683979
Mike Gorse [Thu, 13 Sep 2012 17:59:20 +0000 (12:59 -0500)]
Remove erroneous debug print introduced in last commit
Mike Gorse [Thu, 13 Sep 2012 17:55:36 +0000 (12:55 -0500)]
Only initialize the cache when an AT is running
If no AT is running, we shouldn't need to be maintaining the cache, so
only initialize when an AT is active. Note that this change does not
deinitialize the cache when an AT is no longer running. Deinitializing
and setting spi_global_cache to NULL might be a good idea, but trying to
keep risk to a minimum this close to 3.6.
Colin Walters [Thu, 6 Sep 2012 17:46:47 +0000 (13:46 -0400)]
Fix build for removal of po/ directory
Various things missed.
Claude Paroz [Thu, 6 Sep 2012 16:13:08 +0000 (18:13 +0200)]
Removed po directory
i18n support has been removed from at-spi2-atk
(See commits ed260a6 and bc07db9, bug #678035)
Mike Gorse [Tue, 4 Sep 2012 21:01:06 +0000 (16:01 -0500)]
Correctly cast hyperlinks in GetURI
https://bugzilla.gnome.org/show_bug.cgi?id=683182
Mike Gorse [Mon, 3 Sep 2012 23:12:03 +0000 (18:12 -0500)]
Release v2.5.91
Piotr Drąg [Sat, 1 Sep 2012 18:50:35 +0000 (20:50 +0200)]
Updated Polish translation
Rūdolfs Mazurs [Fri, 31 Aug 2012 21:09:41 +0000 (00:09 +0300)]
Updated Latvian translation
Alejandro Piñeiro [Tue, 21 Aug 2012 16:05:17 +0000 (18:05 +0200)]
Removing gtk+3 directory from AC_CONFIG_FILES on configure.ac
That directory was removed on a previous commit
https://bugzilla.gnome.org/show_bug.cgi?id=678315
Joanmarie Diggs [Mon, 18 Jun 2012 13:01:09 +0000 (09:01 -0400)]
Fix for Bug 678315 - Remove the bridge module for Gtk+ 3
Mike Gorse [Mon, 20 Aug 2012 20:19:19 +0000 (15:19 -0500)]
Release v2.5.90
Mike Gorse [Mon, 20 Aug 2012 19:40:43 +0000 (14:40 -0500)]
When a client is removed, don't continue looking through the list of clients
https://bugzilla.gnome.org/show_bug.cgi?id=679297
Mike Gorse [Mon, 20 Aug 2012 19:36:52 +0000 (14:36 -0500)]
Don't leak GError when option parsing fails
https://bugzilla.gnome.org/show_bug.cgi?id=679296
Mike Gorse [Mon, 20 Aug 2012 19:29:22 +0000 (14:29 -0500)]
Don't iterate over the list twice when removing events
https://bugzilla.gnome.org/show_bug.cgi?id=679295
Mike Gorse [Mon, 6 Aug 2012 20:59:01 +0000 (15:59 -0500)]
Release v2.5.5
Daniel Narvaez [Thu, 19 Jul 2012 20:12:04 +0000 (22:12 +0200)]
Allow to build out of source directory
https://bugzilla.gnome.org/show_bug.cgi?id=680280
Mike Gorse [Mon, 30 Jul 2012 17:03:25 +0000 (12:03 -0500)]
Implement GetLocalizedName for actions
https://bugzilla.gnome.org/show_bug.cgi?id=680598
Mike Gorse [Mon, 16 Jul 2012 21:41:07 +0000 (16:41 -0500)]
Release v2.5.4
Mike Gorse [Mon, 9 Jul 2012 22:41:38 +0000 (17:41 -0500)]
Only send property-change signals relevant for caching, absent listeners
Previously, all property-change signals were passed on, regardless of any
registered listeners, because some of them are used by libatspi to keep the
cache up-to-date. However, gtk sends many notifications for property changes
that would have no impact on libatspi's caching code, so we should not waste
time building and sending D-Bus messages for them unless something is
interested in the event.
https://bugzilla.gnome.org/show_bug.cgi?id=674719
Mike Gorse [Mon, 9 Jul 2012 22:36:26 +0000 (17:36 -0500)]
Only add items to the cache on children-changed if they are passed in
Gtk tree views, for instance, send children-changed signals without passing
an object. If a toolkit does this, then assume that it is doing so because
fetching the object would be expensive, and/or the object is transient, so
just have the cache bypass the signal rather than ref the child in order to
add it. Helps with performance when creating large gtk+ tree views.
https://bugzilla.gnome.org/show_bug.cgi?id=674719
Mike Gorse [Fri, 6 Jul 2012 19:37:29 +0000 (14:37 -0500)]
Don't try to set up the server with a main loop if there's no server
Mike Gorse [Fri, 6 Jul 2012 00:58:43 +0000 (19:58 -0500)]
Only create a directory and a socket when requested
Currently, atk-bridge clutters XDG_RUNTIME_DIR with directories for sockets,
which is especially annoying since gtk currently does not call
atk_bridge_adaptor_cleanup, so the directories never go away. This change does
not really solve the problem--it only hides it from users who do not really
need AT-SPI--but, nevertheless, there is no reason to create a directory or a
socket if nothing has requested it, so doing this lazily makes sense.
Cosimo Cecchi [Mon, 2 Jul 2012 20:14:48 +0000 (16:14 -0400)]
atk-adaptor: plug a refcount leak
https://bugzilla.gnome.org/show_bug.cgi?id=679285
Cosimo Cecchi [Mon, 2 Jul 2012 20:14:12 +0000 (16:14 -0400)]
atk-adaptor: plug a refcount leak
https://bugzilla.gnome.org/show_bug.cgi?id=679285
Mike Gorse [Wed, 4 Jul 2012 01:52:25 +0000 (20:52 -0500)]
Don't create a sub-directory for the socket when running as root
If an application is running as root but within a user's runtime directory,
then creating a temporary directory will result in the directory being owned
by root, and the user will not have permission to search the directory to
access the socket, so an AT running as the normal user will not be able to
connect to the application running as root.
Fixes regression introduced by the fix for BGO#678348.
Joanmarie Diggs [Wed, 27 Jun 2012 21:42:25 +0000 (17:42 -0400)]
Fix for bug #679013 - AtspiCollection should be implemented for all containers
Mike Gorse [Mon, 25 Jun 2012 23:21:36 +0000 (18:21 -0500)]
Release v2.5.3
Mike Gorse [Mon, 25 Jun 2012 23:14:57 +0000 (18:14 -0500)]
Add atkbridge.symbols to EXTRA_DIST
Mike Gorse [Mon, 25 Jun 2012 22:49:17 +0000 (17:49 -0500)]
Fix some more warnings
Mike Gorse [Mon, 25 Jun 2012 22:36:28 +0000 (17:36 -0500)]
Reinstate gnome_accessibility_module_init/shutdown for the gtk 3 module
Removing gnome_accessibility_module_init/shutdown will break Unity, for one,
and there is no real reason to remove it, other than it being an old API that
is to be deprecated and should not be needed anymore, but the new API is not
necessarily finalized, so might as well reinstate the old method for now until
we are certain what the new API will be and can then request that, ie, Unity be
patched accordingly.
Mike Gorse [Thu, 21 Jun 2012 21:56:40 +0000 (16:56 -0500)]
Use XDG_RUNTIME_DIR to hold sockets, and do not make a world-writable dir
If we use XDG_RUNTIME_DIR, then the directory should be owned by the
appropriate user, so it should not need to be world-writable. Hopefully this
won't break accessibility for administrative apps on some distro.
https://bugzilla.gnome.org/show_bug.cgi?id=678348
Alejandro Piñeiro [Mon, 18 Jun 2012 13:06:42 +0000 (15:06 +0200)]
Only register events at beginning if AT is listening
https://bugzilla.gnome.org/show_bug.cgi?id=678475
John Ralls [Sun, 17 Jun 2012 20:24:00 +0000 (21:24 +0100)]
Make extern the globals in introspection.h
To fix duplicate symbol _spi_org_a11y_atspi_Accessible in
../atk-adaptor/adaptors/.libs/libatk-bridge-adaptors.a(libatk_bridge_adaptors_la-accessible-adaptor.o)
and .libs/libatk_bridge_2_0_la-introspection.o
Ryan Lortie [Sun, 17 Jun 2012 00:01:04 +0000 (20:01 -0400)]
more uint -> guint fixes
Missed one with the last commit.
Ryan Lortie [Sat, 16 Jun 2012 23:59:12 +0000 (19:59 -0400)]
bridge.c: change 'uint' for 'guint'
'uint' seems to work on Linux (GNU extension?) but it breaks on Mac OS,
so use 'guint' instead.
Bastien Nocera [Fri, 15 Jun 2012 16:01:43 +0000 (17:01 +0100)]
Remove duplicate code from GTK+ modules
Now that the AT_NO_BRIDGE check lives in atk-bridge-2.0
Bastien Nocera [Fri, 15 Jun 2012 16:00:20 +0000 (17:00 +0100)]
Put declarations at the beginning of the block
Bastien Nocera [Fri, 15 Jun 2012 15:58:54 +0000 (16:58 +0100)]
Reset inited variable on cleanup
Alejandro Piñeiro [Thu, 14 Jun 2012 19:41:57 +0000 (21:41 +0200)]
Move NO_AT_BRIDGE test to the library
https://bugzilla.gnome.org/show_bug.cgi?id=678125
Bastien Nocera [Thu, 14 Jun 2012 18:43:54 +0000 (19:43 +0100)]
Fix undeclared G_BEGIN_DECL usage
Bastien Nocera [Wed, 13 Jun 2012 17:10:02 +0000 (18:10 +0100)]
Fix compile-time warnings
Unused variables, assignment used as truth values and unused functions.
https://bugzilla.gnome.org/show_bug.cgi?id=678045
Bastien Nocera [Wed, 13 Jun 2012 18:22:48 +0000 (19:22 +0100)]
Namespace the exported API
https://bugzilla.gnome.org/show_bug.cgi?id=678051
Bastien Nocera [Wed, 13 Jun 2012 18:18:34 +0000 (19:18 +0100)]
Export the atk-bridge-2.0 library
https://bugzilla.gnome.org/show_bug.cgi?id=678051
Bastien Nocera [Wed, 13 Jun 2012 16:01:02 +0000 (17:01 +0100)]
Link the GTK+ modules against the private library
https://bugzilla.gnome.org/show_bug.cgi?id=678037
Bastien Nocera [Wed, 13 Jun 2012 15:29:22 +0000 (16:29 +0100)]
Remove unneeded X11 dependency
https://bugzilla.gnome.org/show_bug.cgi?id=678036
Bastien Nocera [Wed, 13 Jun 2012 16:07:52 +0000 (17:07 +0100)]
Move adaptor_*() functions into their own header
https://bugzilla.gnome.org/show_bug.cgi?id=678037
Bastien Nocera [Wed, 13 Jun 2012 15:17:17 +0000 (16:17 +0100)]
Move GTK+ modules to their own source files
https://bugzilla.gnome.org/show_bug.cgi?id=678037
Bastien Nocera [Wed, 13 Jun 2012 16:31:00 +0000 (17:31 +0100)]
Remove unused variables
https://bugzilla.gnome.org/show_bug.cgi?id=678039
Bastien Nocera [Wed, 13 Jun 2012 16:31:26 +0000 (17:31 +0100)]
Remove unused functions
https://bugzilla.gnome.org/show_bug.cgi?id=678040
Bastien Nocera [Wed, 13 Jun 2012 15:55:54 +0000 (16:55 +0100)]
Create an uninstalled library
https://bugzilla.gnome.org/show_bug.cgi?id=678037
Bastien Nocera [Wed, 13 Jun 2012 15:15:45 +0000 (16:15 +0100)]
Export adaptor_init() and adaptor_cleanup()
https://bugzilla.gnome.org/show_bug.cgi?id=678037
Bastien Nocera [Wed, 13 Jun 2012 16:33:39 +0000 (17:33 +0100)]
build: Quiet by default
https://bugzilla.gnome.org/show_bug.cgi?id=678041
Bastien Nocera [Wed, 13 Jun 2012 15:52:20 +0000 (16:52 +0100)]
Remove duplicate use of DBUS_LIBS
https://bugzilla.gnome.org/show_bug.cgi?id=678038
Bastien Nocera [Wed, 13 Jun 2012 15:12:58 +0000 (16:12 +0100)]
Remove unused X11 display code
https://bugzilla.gnome.org/show_bug.cgi?id=678036
Bastien Nocera [Wed, 13 Jun 2012 15:49:00 +0000 (16:49 +0100)]
po: No files left to translate
https://bugzilla.gnome.org/show_bug.cgi?id=678035
Bastien Nocera [Wed, 13 Jun 2012 15:07:07 +0000 (16:07 +0100)]
No need to include i18n support
There's nothing to translate here.
https://bugzilla.gnome.org/show_bug.cgi?id=678035
Mike Gorse [Wed, 6 Jun 2012 01:32:39 +0000 (20:32 -0500)]
Release v2.5.2
Joanmarie Diggs [Thu, 31 May 2012 17:30:20 +0000 (13:30 -0400)]
Fix for bug 677211 - The collection interface's MATCH_ANY fails for states
Mike Gorse [Tue, 1 May 2012 13:52:45 +0000 (08:52 -0500)]
Update autogen.sh, in order to support automake 1.12
Mike Gorse [Mon, 30 Apr 2012 21:29:04 +0000 (16:29 -0500)]
Release v2.5.1
Mike Gorse [Mon, 30 Apr 2012 20:52:11 +0000 (15:52 -0500)]
Remove a translation from a g_warning call
The consensus seems to be that messages written out as a result of calls to
g_warning/g_log should ot be translated; ie:
http://mail.gnome.org/archives/desktop-devel-list/2011-March/msg00070.html
Åsmund Skjæveland [Sun, 29 Apr 2012 10:24:00 +0000 (12:24 +0200)]
Updated Norwegian Nynorsk translation
Mike Gorse [Mon, 26 Mar 2012 20:53:28 +0000 (15:53 -0500)]
Release v2.4.0
Rajesh Ranjan [Mon, 26 Mar 2012 11:05:17 +0000 (16:35 +0530)]
hindi translation by Chandan Kumar
Mike Gorse [Mon, 19 Mar 2012 21:40:41 +0000 (16:40 -0500)]
Release v2.3.92
Praveen Arimbrathodiyil [Mon, 12 Mar 2012 16:09:53 +0000 (21:39 +0530)]
Malayalam translation by Anish A
Timo Jyrinki [Mon, 12 Mar 2012 08:43:34 +0000 (10:43 +0200)]
Finnish translation update from l10n.laxstrom.name/wiki/Gnome_3.4 translation sprint
Khoem Sokhem [Sat, 10 Mar 2012 12:48:01 +0000 (13:48 +0100)]
Add initial Khmer translation
Mike Gorse [Tue, 6 Mar 2012 00:17:30 +0000 (18:17 -0600)]
Release v2.3.91
Mike Gorse [Mon, 5 Mar 2012 23:20:19 +0000 (17:20 -0600)]
Have DoAction send the reply message *before* invoking atk
In the past, a gtk button's do_action handler added an idle to invoke
the button and then returned, but now the idle has been removed, and the
do_action call activates the button directly, meaning that, if the
button invokes a dialogue, then atk_action_do_action will not return
until the dialog closes. So, to be safe, we need to send a reply before
invoking atk. This means that atk's return value gets ignored, although
it was somewhat meaningless in gtk's case anyhow. This required that
droute's behavior be changed so that, if a handler does not return a
message, droute will now assume that the handler already sent a reply,
rather than synthesizing a default empty reply. Thus, handlers are now
required to return a value DBusMessage.
Perhaps the API should really be asynchronous, with a callback to be
invoked when the action finishes.
Mattias Põldaru [Thu, 1 Mar 2012 17:38:24 +0000 (19:38 +0200)]
[l10n] Updated Estonian translation
Bahodir Mansurov [Sun, 26 Feb 2012 05:45:12 +0000 (00:45 -0500)]
Adding Uzbek@cyrillic translation.
Mike Gorse [Thu, 23 Feb 2012 22:52:33 +0000 (16:52 -0600)]
Have GetIndexInParent return a signed int, per the spec
Mike Gorse [Mon, 20 Feb 2012 23:08:06 +0000 (17:08 -0600)]
Release v2.3.90
Mike Gorse [Fri, 10 Feb 2012 12:11:37 +0000 (06:11 -0600)]
Check that we have a valid path when emitting a signal
If we don't have a valid object path in emit_event (ie, if NULL is
passed in, for some strange reason), then return rather than crashing in
libdbus.
https://bugzilla.gnome.org/show_bug.cgi?id=669794
Mike Gorse [Tue, 7 Feb 2012 23:17:43 +0000 (17:17 -0600)]
Have GrabFocus return a bool, per the spec, rather than a uint32
Mike Gorse [Mon, 6 Feb 2012 23:18:27 +0000 (17:18 -0600)]
Release v2.3.5
Mike Gorse [Mon, 6 Feb 2012 22:35:43 +0000 (16:35 -0600)]
Don't cache objects that return NULL when their state set is queried
If atk_object_ref_state_set returns NULL, then perhaps the object is
going away or not yet initialized, so caching it may not be a good idea.
Also stops some warnings.
Mike Gorse [Mon, 30 Jan 2012 01:03:13 +0000 (19:03 -0600)]
Don't try to unref an object if atk returned NULL
Kjartan Maraas [Tue, 24 Jan 2012 13:55:18 +0000 (14:55 +0100)]
Updated Norwegian bokmål translation
Mike Gorse [Sat, 21 Jan 2012 18:13:02 +0000 (12:13 -0600)]
When an object goes away, remove from queue of pending cache additions
Items are added to the cache in an idle (see commit 75ff19), meaning
that it is possible for an object to go away between the time it is
queued for addition and the time the additions are made.
https://bugzilla.gnome.org/show_bug.cgi?id=666371
Mike Gorse [Sat, 21 Jan 2012 11:14:55 +0000 (05:14 -0600)]
Don't use /a11y/ as a dconf path
Change schema path from /a11y/ to /org/a11y/.
https://bugzilla.gnome.org/show_bug.cgi?id=663967
Ray Strode [Sat, 21 Jan 2012 11:06:27 +0000 (05:06 -0600)]
bridge: drive-by list usage fixes
There are a few places where the glib list apis
aren't being used right:
1) There is code that use GList functions on GSLists
2) There is code that uses the list node itself instead
of the node data
3) There is code that removes nodes from a list as its iterating
over that list.
https://bugzilla.gnome.org/show_bug.cgi?id=659967
Mike Gorse [Thu, 19 Jan 2012 16:41:11 +0000 (10:41 -0600)]
Remove methods to set an atk object's name and description
It doesn't make sense to expose atk_object_set_name and
atk_object_set_description over D-Bus. Removing the ability to set those
properties.
Mike Gorse [Mon, 16 Jan 2012 03:53:37 +0000 (21:53 -0600)]
Release v2.3.4
Mike Gorse [Thu, 5 Jan 2012 20:38:26 +0000 (14:38 -0600)]
Make keystroke listeners work even if no other event listeners present
Fetch a list of active keystroke and mouse event listeners, and monitor
for changes. Enable event watching if keyboard listeners are present.
This requires the newest revision of at-spi2-core.
Fixes http://bugzilla.gnome.org/show_bug.cgi?id=666870
Mike Gorse [Tue, 3 Jan 2012 02:17:18 +0000 (20:17 -0600)]
Fix a warning caused by a missing const
Mike Gorse [Mon, 19 Dec 2011 19:17:47 +0000 (14:17 -0500)]
Release v2.3.3
Benjamin Otte [Mon, 19 Dec 2011 07:51:55 +0000 (08:51 +0100)]
events: Only deregister objects when they are defunct
We don't want to deregister them when they are changing from defunct to
functioning. And GTK 3.4 emits that change after setting up the
accessible.
Mike Gorse [Thu, 15 Dec 2011 02:14:27 +0000 (21:14 -0500)]
Change a : to a / in the suffix to an event
at-spi2-core now uses arg0path to create match rules to catch events
with suffixes (ie, :system), but this is kind of stretching the intended
purpose of arg*path, which expects paths delimited with /'s, so replace
a : with a /. The / is changed back to a : in at-spi2-core.
Mike Gorse [Mon, 12 Dec 2011 17:25:44 +0000 (11:25 -0600)]
Don't leak ref when processing a child-added signal with no object value
Mike Gorse [Sat, 26 Nov 2011 21:50:55 +0000 (15:50 -0600)]
In GetItems, unref objects in two passes
It is dangerous to unref cache objects inside a g_hash_table_foreach,
since, if the object is finalized, it will be removed, which can cause
problems with the iterator. Instead, make a GSList of the objects to
unref, then use this GSList to unref.
Mike Gorse [Fri, 25 Nov 2011 18:56:37 +0000 (12:56 -0600)]
Ref all cached accessibles before querying them for GetItems
It is possible, at least in theory, that querying one accessible could
have the side effect of another accessible being dereferenced. If this
results in the accessible being finalized, then this would likely cause
a crash when potentially iterating over the newly-finalized object.
Hopefully fixes http://bugzilla.gnome.org/show_bug.cgi?id=664822