platform/upstream/glib.git
14 years agoGApplication: add remote commandline support
Ryan Lortie [Sat, 9 Oct 2010 21:24:09 +0000 (17:24 -0400)]
GApplication: add remote commandline support

14 years agoGApplication test case
Ryan Lortie [Wed, 6 Oct 2010 18:06:26 +0000 (14:06 -0400)]
GApplication test case

14 years agoGApplication: fix inactivity-timeout
Ryan Lortie [Wed, 6 Oct 2010 18:05:40 +0000 (14:05 -0400)]
GApplication: fix inactivity-timeout

Create the gobject property for it.

Tweak the logic of having a pending timeout at the time that the
application starts -- run the mainloop with a use count of zero if there
is a timeout active.

14 years agonew GApplication implementation
Ryan Lortie [Wed, 6 Oct 2010 17:08:26 +0000 (13:08 -0400)]
new GApplication implementation

14 years agoUse CSIDL_LOCAL_APPDATA on Windows
Tor Lillqvist [Thu, 14 Oct 2010 19:47:25 +0000 (22:47 +0300)]
Use CSIDL_LOCAL_APPDATA on Windows

Make g_get_user_data_dir() return the CSIDL_LOCAL_APPDATA folder on
Windows, and not CSIDL_PERSONAL. On Windows 7, that corresponds to the
subfolders AppData\Local vs. Documents under the profile ("home")
folder. This matches what Qt does, for instance, and has been widely
requested.

Also make g_get_user_config_dir() return this and not the (roaming)
CSIDL_APPDATA folder. The reason for this change is that it would be
surprising and hard to justify if g_get_user_data_dir() returned the
local application data folder while g_get_user_config_dir() would
return the roaming one. Nothing in the function names or the XDG specs
suggests that any roaming vs. local dichotomy would be involved.

Document the new semantics and the fact that these two functions now
return the same directory on Windows.

Note that in reality, code that really truly wants to support Windows
as well as possible probably will not use these GLib functions anyway,
but Win32 APIs directly to be sure what it is doing...

Should hopefully satisfy complaints in bug #620710 and related bugs.

14 years agoPrevent error pileup
Matthias Clasen [Sun, 17 Oct 2010 03:30:30 +0000 (23:30 -0400)]
Prevent error pileup

14 years agoGSignal: fix trivial docs typo
Ryan Lortie [Wed, 13 Oct 2010 01:34:26 +0000 (21:34 -0400)]
GSignal: fix trivial docs typo

14 years agoGObject: add g_signal_accumulator_first_wins
Ryan Lortie [Wed, 13 Oct 2010 00:58:43 +0000 (20:58 -0400)]
GObject: add g_signal_accumulator_first_wins

Deals with the case where we want exactly one signal handler to run.

14 years agogthemedicon: Fix annotation for g_themed_icon_get_names
Colin Walters [Tue, 12 Oct 2010 16:54:36 +0000 (12:54 -0400)]
gthemedicon: Fix annotation for g_themed_icon_get_names

14 years agoUpdated Spanish translation
Jorge González [Sat, 9 Oct 2010 12:38:08 +0000 (14:38 +0200)]
Updated Spanish translation

14 years agogtypemodule: Mention g_object_run_dispose correctly in a warning
Javier Jardón [Fri, 8 Oct 2010 15:21:53 +0000 (17:21 +0200)]
gtypemodule: Mention g_object_run_dispose correctly in a warning

https://bugzilla.gnome.org/show_bug.cgi?id=630797

14 years agogobject: Improve install_properties()
Emmanuele Bassi [Fri, 8 Oct 2010 15:09:10 +0000 (16:09 +0100)]
gobject: Improve install_properties()

Cache the parent type outside of the loop, and fix a typo in the
documentation.

14 years agoInitialise lengths in GvdbReader to silence warnings
Christian Dywan [Fri, 8 Oct 2010 14:33:33 +0000 (16:33 +0200)]
Initialise lengths in GvdbReader to silence warnings

14 years agoInitialise lengths in GDbusAuth to silence warnings
Christian Dywan [Fri, 8 Oct 2010 14:33:04 +0000 (16:33 +0200)]
Initialise lengths in GDbusAuth to silence warnings

14 years agoMention g_object_run_dispose correctly in tutorial
Christian Dywan [Fri, 8 Oct 2010 13:17:31 +0000 (15:17 +0200)]
Mention g_object_run_dispose correctly in tutorial

Fixes: https://bugzilla.gnome.org/show_bug.cgi?id=630797
14 years agoUpdated British English translation
Bruce Cowan [Thu, 7 Oct 2010 20:27:17 +0000 (21:27 +0100)]
Updated British English translation

14 years agoRemove g_desktop_app_info_lookup_get_type symbol from gio.types
William Jon McCann [Thu, 7 Oct 2010 17:51:21 +0000 (13:51 -0400)]
Remove g_desktop_app_info_lookup_get_type symbol from gio.types

It was removed in 9b262f1c5fe5a6fd879f17cd7b80d8c54e33d80c

14 years agoDon't call close() on -1
Tor Lillqvist [Thu, 7 Oct 2010 09:04:52 +0000 (12:04 +0300)]
Don't call close() on -1

Of course, a proper implementation of close() will just ignore an
invalid parameter silently, and set errno. But apparently the "debug"
version of the Microsoft C library generates some noise in this
case. So avoid that. Thanks to John Emmas for reporting.

14 years agoUpdated Hebrew translation.
Yaron Shahrabani [Wed, 6 Oct 2010 00:18:30 +0000 (02:18 +0200)]
Updated Hebrew translation.

14 years agoReplace "gio-desktop-app-info-lookup" extension point
Bastien Nocera [Tue, 5 Oct 2010 14:02:37 +0000 (15:02 +0100)]
Replace "gio-desktop-app-info-lookup" extension point

With a native version, that looks for desktop items supporting
x-scheme-handler/foo, when looking for a handler for the "foo"
URI scheme handler.

https://bugzilla.gnome.org/show_bug.cgi?id=631410

14 years agovolume monitor: don't unref NULL
Ryan Lortie [Tue, 5 Oct 2010 06:29:47 +0000 (02:29 -0400)]
volume monitor: don't unref NULL

Fix a couple more cases of blindly calling g_object_unref() on the
result of a function that is documented as sometimes returning NULL.

14 years agoGVariant tests: fix bad allignment assumption
Ryan Lortie [Tue, 5 Oct 2010 06:11:22 +0000 (02:11 -0400)]
GVariant tests: fix bad allignment assumption

One of the GVariant test cases allocates a pair of character arrays on
the stack and then passes them to functions that assume that they will
be aligned to a multiple of two.

This is not the case for some versions of GCC (4.0.3 on PowerPC).

14 years agoGSettings test: fix error match strings
Ryan Lortie [Tue, 5 Oct 2010 01:07:50 +0000 (21:07 -0400)]
GSettings test: fix error match strings

The name of the internal function that appears in an assertion message
has changed.  Update the tests.

14 years agogsettings-tool: Add 'range' subcommand
Ryan Lortie [Mon, 4 Oct 2010 07:41:03 +0000 (03:41 -0400)]
gsettings-tool: Add 'range' subcommand

Provides access to the g_settings_get_range() functionality, converting
its return value to something that's reasonable for printing at the
console and potentially parseable.  The format may change.

Bug #631264.

14 years agogsettings-tool: implement range-checking
Ryan Lortie [Mon, 4 Oct 2010 07:40:22 +0000 (03:40 -0400)]
gsettings-tool: implement range-checking

Prevent assertion messages from spewing forth and also ensure that we
exit with an error status in the event that the value was out of range.

Bug #631264.

14 years agoUpdate symbols and docs sections
Ryan Lortie [Mon, 4 Oct 2010 07:36:09 +0000 (03:36 -0400)]
Update symbols and docs sections

14 years agoGSettings: add g_settings_range_check() API
Ryan Lortie [Mon, 4 Oct 2010 07:33:06 +0000 (03:33 -0400)]
GSettings: add g_settings_range_check() API

Checks if a given value is within the correct range for a key.

14 years agoBug 631263 - GSettings needs range/choice APIs
Ryan Lortie [Mon, 4 Oct 2010 06:58:46 +0000 (02:58 -0400)]
Bug 631263 - GSettings needs range/choice APIs

Add g_settings_get_range() to describe the possible values that may be
provided to g_settings_set_value() without causing an error.

Add a test case.

14 years agoschema compiler: Don't store zero-valued flags
Ryan Lortie [Mon, 4 Oct 2010 06:57:06 +0000 (02:57 -0400)]
schema compiler: Don't store zero-valued flags

Don't store the "none" value for flags into the compiled schema file.
"none" should never appear as a value -- no flags set is indicated by an
empty array.

14 years agoBug 627126 - gsettings schemas on FreeBSD
Ryan Lortie [Mon, 4 Oct 2010 05:51:11 +0000 (01:51 -0400)]
Bug 627126 - gsettings schemas on FreeBSD

Rewrite the install rule for GSettings schemas to not depend on an
obscure chunk of non-portable sed that nobody understands the purpose
of.

Instead, use make's VPATH feature to resolve the paths of the files that
need to be installed.  No need to depend on the .valid targets here
since automake already ensures that 'make all' is complete before 'make
install' is permitted to run.

14 years agoglib-compile-schemas: write strinfo little endian
Ryan Lortie [Mon, 4 Oct 2010 03:26:18 +0000 (23:26 -0400)]
glib-compile-schemas: write strinfo little endian

Ensure that the strinfo is output in little-endian byte order on big
endian machines.

GSettings is now passing all of its tests on PowerPC.

Bug #630968 is closed.

14 years agoGSettings strinfo: byteswap integers
Ryan Lortie [Mon, 4 Oct 2010 03:25:29 +0000 (23:25 -0400)]
GSettings strinfo: byteswap integers

strinfo is always strictly little endian, so ensure that we byteswap to
native when comparing and returning.

14 years agoGSettings endian: missed a spot
Ryan Lortie [Mon, 4 Oct 2010 03:15:27 +0000 (23:15 -0400)]
GSettings endian: missed a spot

Missed an instance of get_value -> get_raw_value search/replace.

14 years agoGSettings big endian tweaks
Ryan Lortie [Mon, 4 Oct 2010 03:04:00 +0000 (23:04 -0400)]
GSettings big endian tweaks

GSettings relies on parts of the schema infromation remaining
unbyteswapped (the strinfo database, for example) while requiring other
parts to be in native order (the default value, constraints, etc.).

Lift the byteswapping into a place where we can do it selectively.

14 years agoMerge remote branch 'gvdb/master'
Ryan Lortie [Mon, 4 Oct 2010 03:03:12 +0000 (23:03 -0400)]
Merge remote branch 'gvdb/master'

14 years agoAdd gvdb_table_get_raw_value() API for GSettings
Ryan Lortie [Mon, 4 Oct 2010 02:54:03 +0000 (22:54 -0400)]
Add gvdb_table_get_raw_value() API for GSettings

14 years agoGSettings test: fix unsafe GObject properties use
Ryan Lortie [Mon, 4 Oct 2010 02:53:49 +0000 (22:53 -0400)]
GSettings test: fix unsafe GObject properties use

The test case was passing a guint16 to g_object_get() for a guint
property.  That's invalid on all systems, although it works (more or
less) on little endian ones.  On big endian it's a total no-go.

14 years agoGVariant: avoid byteswapping in some cases
Ryan Lortie [Mon, 4 Oct 2010 02:39:47 +0000 (22:39 -0400)]
GVariant: avoid byteswapping in some cases

Make g_variant_byteswap() merely return a new reference on the given
value in the event that we know that byteswapping will have no effect
(ie: types which have no alignment requirement).

This fixes a somewhat complicated interaction between GVariant,
GSettings and GVDB on big endian machines:  GSettings assumes that it
can unref values returned from GVDB without losing access to the
underlying data.  This only works if the underlying data is in the
mapped file -- not a freshly-allocated buffer that GVariant byteswapped
into.

14 years agoMerge remote branch 'gvdb/master'
Ryan Lortie [Mon, 4 Oct 2010 01:11:17 +0000 (21:11 -0400)]
Merge remote branch 'gvdb/master'

14 years agoFix reader on big endian
Ryan Lortie [Mon, 4 Oct 2010 01:10:35 +0000 (21:10 -0400)]
Fix reader on big endian

Some of the hashtable code was failing to byteswap properly.

14 years agoallow libtool 2.4
Ryan Lortie [Mon, 4 Oct 2010 00:47:26 +0000 (20:47 -0400)]
allow libtool 2.4

It's released now and it seems to work OK.

14 years agoBug 623400 - acquire context before dispatching
Ryan Lortie [Sun, 3 Oct 2010 21:30:10 +0000 (17:30 -0400)]
Bug 623400 - acquire context before dispatching

For GSettings.

Use the functionality introduced in the last commit to simplify our
notify dispatching and increase the safety of doing so (by ensuring that
the context is acquired in the current thread for the duration of the
dispatch).

This closes bugs #623400 and #629849.

14 years agoBug 618737 - "dispatch to context" functionality
Ryan Lortie [Sun, 3 Oct 2010 21:26:37 +0000 (17:26 -0400)]
Bug 618737 - "dispatch to context" functionality

Adds a new function g_main_context_invoke() (and _full() variant).

This function takes a main context, a function and a user_data.  If the
main context is already acquired in the current thread, the function is
invoked directly.  If the main context is the default main context of
the current thread and it can be acquired then the function is invoked
directly while the context is owned.  Otherwise, the function is
scheduled as an idle on the context.

14 years agoUpdated Lithuanian translation by Aurimas Černius <aurisc4@gmail.com>
Žygimantas Beručka [Sun, 3 Oct 2010 13:45:33 +0000 (15:45 +0200)]
Updated Lithuanian translation by Aurimas Černius <aurisc4@gmail.com>

14 years agogsettings-tool: Rewrite
Ryan Lortie [Sun, 3 Oct 2010 06:40:48 +0000 (02:40 -0400)]
gsettings-tool: Rewrite

Rewrite the GSettings tool.

Improvements/changes:

 - simplify the code by performing common actions (like creating a
   schema) in only one place instead of one per-command

 - new features (list schemas, list keys, monitor multiple, etc)

 - factor-out bash completion and implement in shellscript

 - input validation: should never abort due to invalid inputs

Still to do:

 - proper error checking for ranges/choices

 - support for querying range/choice information

 - bash completion support for enums

Closes bug #629289, possibly among others.

14 years agoGSettings: implement .property_get('path')
Ryan Lortie [Sun, 3 Oct 2010 05:53:09 +0000 (01:53 -0400)]
GSettings: implement .property_get('path')

This was unimplemented in g_settings_get_property(), leading to a failed
'g_assert_not_reached()'.

14 years agoAdd 'Since:' tags for schema listing APIs
Ryan Lortie [Sun, 3 Oct 2010 02:46:53 +0000 (22:46 -0400)]
Add 'Since:' tags for schema listing APIs

14 years agoClean up g_settings_list_schemas()
Ryan Lortie [Sun, 3 Oct 2010 02:42:02 +0000 (22:42 -0400)]
Clean up g_settings_list_schemas()

In its previous form, g_settings_list_schemas() was not useful as a tool
to prevent aborts due to using g_settings_new() with an invalid schema
name.  This is because g_settings_list_scheams() also listed relocatable
schemas, and calling g_settings_new() for those would abort just the
same as if you called it for a non-existent schema.

Modify g_settings_list_schemas() so that it only returns schemas for
which it is safe to call g_settings_new().  Add another call for sake of
completeness: g_settings_list_relocatable_schemas().

14 years agoUpdated Bulgarian translation
Damyan Ivanov [Sat, 2 Oct 2010 05:57:09 +0000 (08:57 +0300)]
Updated Bulgarian translation

14 years agoImprove .gitignore
Ryan Lortie [Mon, 27 Sep 2010 15:14:08 +0000 (11:14 -0400)]
Improve .gitignore

14 years agoBug 628937 - gracefully handle broken schemas
Ryan Lortie [Mon, 27 Sep 2010 15:10:11 +0000 (11:10 -0400)]
Bug 628937 - gracefully handle broken schemas

Implement the second feature requested in the bug: silently ignore
override files that attempt to override schemas that are not currently
installed.

Also, support 'strictness' being optional for other errors when parsing
override files (ie: inability to open the file, unknown key name, parse
errors, out of range).  We don't completely back out the file in this
case — as that is difficult with the current implementation — but just
ignore the override for the single key.

14 years agoglib-compile-schemas: improve error accuracy
Ryan Lortie [Mon, 27 Sep 2010 14:36:11 +0000 (10:36 -0400)]
glib-compile-schemas: improve error accuracy

We wrote "<enum> must contain at least one <value>" for empty <flags>.
Fix that.

14 years agoBug 628937 - gracefully handle broken schemas
Ryan Lortie [Mon, 27 Sep 2010 14:27:11 +0000 (10:27 -0400)]
Bug 628937 - gracefully handle broken schemas

Implement the first of two features requested in the bug: when
encountering a broken .xml schema file, back out the changes in that
file and continue to parse other files.

This prevents a single broken .xml file from messing up GSettings for
everyone else.

Add a --strict option to get the old behaviour.  Use this from the test
cases.

14 years agoBug 630077 - GDateTime week number support
Ryan Lortie [Mon, 27 Sep 2010 13:06:24 +0000 (09:06 -0400)]
Bug 630077 - GDateTime week number support

Fully implement support for ISO 8601 week dates in GDateTime and
document that this is the case.

Add an exhaustive test case to ensure correctness.

14 years agomessage_to_write_data_free: Don't unref data->message if it is NULL
Matthias Clasen [Thu, 30 Sep 2010 18:40:50 +0000 (14:40 -0400)]
message_to_write_data_free: Don't unref data->message if it is NULL

After the recent changes to message filtering, it can happen that
data->message is NULL when we get here.

14 years agol10n: Updated Greek translation for glib
Giannis Katsampirhs [Thu, 30 Sep 2010 11:47:58 +0000 (14:47 +0300)]
l10n: Updated Greek translation for glib

14 years agoUpdated Polish translation
Piotr Drąg [Wed, 29 Sep 2010 16:46:10 +0000 (18:46 +0200)]
Updated Polish translation

14 years agointrospection: Fix one annotation syntax
Colin Walters [Wed, 29 Sep 2010 14:38:25 +0000 (10:38 -0400)]
introspection: Fix one annotation syntax

14 years agoUpdated Hungarian translation
Gabor Kelemen [Wed, 29 Sep 2010 12:07:05 +0000 (14:07 +0200)]
Updated Hungarian translation

14 years agoUpdated French translation
Bruno Brouard [Wed, 29 Sep 2010 07:02:51 +0000 (09:02 +0200)]
Updated French translation

14 years agoUpdate Czech translation
Petr Kovar [Wed, 29 Sep 2010 02:41:25 +0000 (04:41 +0200)]
Update Czech translation

14 years agoUpdated Slovenian translation
Peter Kragelj [Tue, 28 Sep 2010 19:21:38 +0000 (21:21 +0200)]
Updated Slovenian translation

14 years agoUpdated Galician translations
Fran Diéguez [Tue, 28 Sep 2010 15:35:23 +0000 (17:35 +0200)]
Updated Galician translations

14 years agoUpdated Slovenian translation
Peter Kragelj [Tue, 28 Sep 2010 13:12:00 +0000 (15:12 +0200)]
Updated Slovenian translation

14 years agoUpdated Slovenian translation
Peter Kragelj [Tue, 28 Sep 2010 13:10:33 +0000 (15:10 +0200)]
Updated Slovenian translation

14 years agoUpdated Lithuanian translation (thanks Aurimas Cernius).
Gintautas Miliauskas [Mon, 27 Sep 2010 21:12:42 +0000 (00:12 +0300)]
Updated Lithuanian translation (thanks Aurimas Cernius).

14 years ago[i18n] Updated German translation
Mario Blättermann [Mon, 27 Sep 2010 18:19:46 +0000 (20:19 +0200)]
[i18n] Updated German translation

14 years agoUpdated Brazilian Portuguese translation
Fabrício Godoy [Mon, 27 Sep 2010 11:49:51 +0000 (08:49 -0300)]
Updated Brazilian Portuguese translation

14 years agoUpdated Bulgarian translation
Damyan Ivanov [Mon, 27 Sep 2010 10:31:09 +0000 (13:31 +0300)]
Updated Bulgarian translation

14 years agoclean up g_date_time_get_week_day()
Ryan Lortie [Sun, 26 Sep 2010 22:51:23 +0000 (18:51 -0400)]
clean up g_date_time_get_week_day()

No functionality changes here.

Vastly simplify the algorithm for calculating the day of the week.

Fix the documentation (which is incorrectly stating that 1 means
Sunday) and clarify that the number we return is in line with ISO 8601
week day numbering.

14 years agoUpdated Dutch translation by Wouter Bolsterlee
Wouter Bolsterlee [Sun, 26 Sep 2010 20:59:38 +0000 (22:59 +0200)]
Updated Dutch translation by Wouter Bolsterlee

14 years agoUpdated Romanian translation
Lucian Adrian Grijincu [Sun, 26 Sep 2010 19:39:14 +0000 (22:39 +0300)]
Updated Romanian translation

14 years agoAdd a lot of missing annotations
Johan Dahlin [Fri, 24 Sep 2010 21:24:41 +0000 (18:24 -0300)]
Add a lot of missing annotations

14 years ago[introspection] Add GObject annotations
Johan Dahlin [Fri, 24 Sep 2010 19:23:31 +0000 (16:23 -0300)]
[introspection] Add GObject annotations

These were moved in from gobject-introspection

14 years ago[introspection] Move over annotations
Johan Dahlin [Fri, 24 Sep 2010 18:51:26 +0000 (15:51 -0300)]
[introspection] Move over annotations

Move all the annotations over from gobject-introspection.

They will not be used directly by the introspection scanner for now,
instead they will be extracted by a script and updated manually
until introspection is properly integrated into the glib build

14 years agoCorrect error message when GUnixOutputStream fails to write
Christian Dywan [Fri, 24 Sep 2010 11:56:35 +0000 (13:56 +0200)]
Correct error message when GUnixOutputStream fails to write

Fixes: https://bugzilla.gnome.org/show_bug.cgi?id=628876
14 years agoAdd missing file to POTFILES.in
Andre Klapper [Fri, 24 Sep 2010 11:15:23 +0000 (13:15 +0200)]
Add missing file to POTFILES.in

14 years agoGDBus: Don't use abstract sockets in test code
David Zeuthen [Thu, 23 Sep 2010 23:16:25 +0000 (19:16 -0400)]
GDBus: Don't use abstract sockets in test code

It doesn't really work right now because of a dbus-daemon(1) bug - see
the comment added in the TODO section of gdbusconnection.c. So revert
to old behavior. The downside is a lot of files in /tmp but right now
that's better than not being able to run tests in a loop.

Signed-off-by: David Zeuthen <davidz@redhat.com>
14 years agoGDBus: Use abstract namespace in test cases to avoid littering all over /tmp
David Zeuthen [Thu, 23 Sep 2010 22:57:35 +0000 (18:57 -0400)]
GDBus: Use abstract namespace in test cases to avoid littering all over /tmp

Signed-off-by: David Zeuthen <davidz@redhat.com>
14 years agoGDBusConnection: Use correct GMainContext when invoking free functions
David Zeuthen [Thu, 23 Sep 2010 21:23:30 +0000 (17:23 -0400)]
GDBusConnection: Use correct GMainContext when invoking free functions

Without this fix, the ./gdbus-connection test case occasionally fails, see

 https://bugzilla.gnome.org/show_bug.cgi?id=629945#c5

like this

 /gdbus/connection/basic: OK
 /gdbus/connection/life-cycle: **
ERROR:gdbus-connection.c:223:test_connection_life_cycle: assertion failed:
(!quit_mainloop_fired)
 cleaning up bus with pid 21794
 Aborted (core dumped)

because the callback didn't happen on the same thread as where we are
running the loop.

Signed-off-by: David Zeuthen <davidz@redhat.com>
14 years agobuild: fix out of srcdir build
Stefan Kost [Thu, 23 Sep 2010 15:41:57 +0000 (18:41 +0300)]
build: fix out of srcdir build

The gtk-doc makefile rules do cd srcdir && cp $(HTML_IMAGES) target/ and this
breaks for "srcdir" != ".".

14 years agoGDBus: fix name test cases
David Zeuthen [Thu, 23 Sep 2010 20:27:52 +0000 (16:27 -0400)]
GDBus: fix name test cases

Since we make message buses come and go, we need to ensure that the
singleton connection instance goes away before attempting to call
g_bus_get_sync() or similar.

Signed-off-by: David Zeuthen <davidz@redhat.com>
14 years agoGDBus: bump timeout for some tests
David Zeuthen [Thu, 23 Sep 2010 20:09:25 +0000 (16:09 -0400)]
GDBus: bump timeout for some tests

When under load, a one second timeout is just not enough. This can be
observed by e.g. restarting a CPU- and IO-intensive application like a
web browser with many tabs while running the test cases. Therefore,
bump the timeouts to 30 seconds.

Signed-off-by: David Zeuthen <davidz@redhat.com>
14 years agoGDBus: Move "slow" connection test cases into separate test program
David Zeuthen [Thu, 23 Sep 2010 19:47:46 +0000 (15:47 -0400)]
GDBus: Move "slow" connection test cases into separate test program

Signed-off-by: David Zeuthen <davidz@redhat.com>
14 years agoGDBusConnection: Avoid callbacks on finalized connection
David Zeuthen [Thu, 23 Sep 2010 19:10:50 +0000 (15:10 -0400)]
GDBusConnection: Avoid callbacks on finalized connection

Turns out that GDBusWorker will issue callbacks (in its own thread)
even after g_dbus_worker_stop() has been called. This would rarely
happen (and unreffing a connection is even rarer) so only saw this bug
occasionally when running the gdbus-connection test case in a loop.

Fix up this issue by maintaining a set of GDBusConnection objects that
are currently "alive" and do nothing in the callbacks if the passed
user_data pointer is not in this set.

Also attempted to fix up a race condition with
_g_object_wait_for_single_ref_do() and its use of GObject toggle
references - for now, just resort to busy waiting, thereby
sidestepping the toggle reference mess altogether.

Signed-off-by: David Zeuthen <davidz@redhat.com>
14 years agoUpdated Japanese translation (symbolic link related)
Takayuki KUSANO [Thu, 23 Sep 2010 17:09:10 +0000 (02:09 +0900)]
Updated Japanese translation (symbolic link related)

14 years agoFix memory leak in SOCKSv5 implementation
Nicolas Dufresne [Wed, 22 Sep 2010 17:11:59 +0000 (13:11 -0400)]
Fix memory leak in SOCKSv5 implementation

14 years agoglib-2.0.m4: Use unsigned variables for version numbers
Benjamin Otte [Thu, 23 Sep 2010 11:57:22 +0000 (13:57 +0200)]
glib-2.0.m4: Use unsigned variables for version numbers

When using signed, we get complaints from gcc about comparing signed to
unsigned with -Wsign-compare. And combined with -Werror in users' CFLAGS
it breaks configure runs.

14 years agoUpdated French translation
Bruno Brouard [Thu, 23 Sep 2010 06:47:54 +0000 (08:47 +0200)]
Updated French translation

14 years agoGSocketControlMessage: clean up confusing code
Ryan Lortie [Wed, 22 Sep 2010 10:45:02 +0000 (06:45 -0400)]
GSocketControlMessage: clean up confusing code

It looks like the deserialisation function in GSocketControlMessage can
potentially leak a reference to the class structure of a
GSocketControlMessage subclass (although the particular code path is
probably never hit).

Clean up the code a bit.

Also, make sure that the GUnixCredentialsMessage type is registered
before attempting deserialisation.

Closes bug #629687.

14 years agofix argument order for g_date_time_difference()
Ryan Lortie [Wed, 22 Sep 2010 10:28:49 +0000 (06:28 -0400)]
fix argument order for g_date_time_difference()

The names of the arguments in the header file was reversed.

Closes bug #630000

14 years agoAdd translator comments for command parameter translation
Claude Paroz [Wed, 22 Sep 2010 08:42:55 +0000 (10:42 +0200)]
Add translator comments for command parameter translation

14 years ago[l10n] Updated Estonian translation
Mattias Põldaru [Tue, 21 Sep 2010 16:18:38 +0000 (19:18 +0300)]
[l10n] Updated Estonian translation

14 years agoUpdated Basque language
Inaki Larranaga Murgoitio [Tue, 21 Sep 2010 12:42:23 +0000 (14:42 +0200)]
Updated Basque language

14 years agoBug 630185 — Allow NULL strings in g_quark_try_string()
Philip Withnall [Mon, 20 Sep 2010 16:56:30 +0000 (17:56 +0100)]
Bug 630185 — Allow NULL strings in g_quark_try_string()

Allow NULL strings to be passed to g_quark_try_string(), returning a 0 quark
in that case, as per the existing documentation. Closes: bgo#630185

14 years agoFix typo
Behdad Esfahbod [Tue, 21 Sep 2010 02:51:26 +0000 (22:51 -0400)]
Fix typo

14 years agogdir: Clarify in documentation that ordering is not defined
Colin Walters [Mon, 20 Sep 2010 17:20:04 +0000 (13:20 -0400)]
gdir: Clarify in documentation that ordering is not defined

14 years agol10n: Updated Greek translation for glib
Michael Kotsarinis [Mon, 20 Sep 2010 16:40:53 +0000 (19:40 +0300)]
l10n: Updated Greek translation for glib

14 years agoUpdated Hebrew translation.
Yaron Shahrabani [Mon, 20 Sep 2010 13:15:46 +0000 (15:15 +0200)]
Updated Hebrew translation.

14 years agodocs: shorten var name in example
Stefan Kost [Sun, 19 Sep 2010 19:06:10 +0000 (22:06 +0300)]
docs: shorten var name in example

We use 'pspec' instead of 'maman_param_spec' elsewhere too.