platform/upstream/glib.git
12 years agoMore updates
Matthias Clasen [Tue, 30 Aug 2011 14:25:42 +0000 (10:25 -0400)]
More updates

12 years agogio/tests/gdbus-peer: make this work on non-Linux unixes
Dan Winship [Sat, 27 Aug 2011 15:33:43 +0000 (11:33 -0400)]
gio/tests/gdbus-peer: make this work on non-Linux unixes

The test was using a socket in a temporary directory, but not actually
creating that temporary directory. This worked fine on Linux since it
actually ended up using an abstract socket instead, but failed on
unixes without abstract sockets.

https://bugzilla.gnome.org/show_bug.cgi?id=657517

12 years agoAdd a forgotten section
Matthias Clasen [Tue, 30 Aug 2011 04:00:37 +0000 (00:00 -0400)]
Add a forgotten section

12 years agoMore doc additions
Matthias Clasen [Tue, 30 Aug 2011 04:00:21 +0000 (00:00 -0400)]
More doc additions

12 years agoFix a doc comment
Matthias Clasen [Tue, 30 Aug 2011 03:55:48 +0000 (23:55 -0400)]
Fix a doc comment

12 years agoClean up gobject docs
Matthias Clasen [Tue, 30 Aug 2011 03:42:24 +0000 (23:42 -0400)]
Clean up gobject docs

12 years agoCorrect a doc typo
Matthias Clasen [Tue, 30 Aug 2011 03:36:27 +0000 (23:36 -0400)]
Correct a doc typo

12 years agoRemove no-longer existing api
Matthias Clasen [Tue, 30 Aug 2011 03:33:48 +0000 (23:33 -0400)]
Remove no-longer existing api

12 years agoGSocketAddress: create IPv4 addresses for IPv6 mapped-IPv4 addresses
Dan Winship [Sat, 27 Aug 2011 23:46:51 +0000 (19:46 -0400)]
GSocketAddress: create IPv4 addresses for IPv6 mapped-IPv4 addresses

IPv6-mapped-IPv4 addresses are annoying. Just translate them to real
IPv4 addresses.

based on a patch from Jonny Lamb
https://bugzilla.gnome.org/show_bug.cgi?id=646082

12 years agoAdd a 'since 2.30' index to gobject docs
Matthias Clasen [Tue, 30 Aug 2011 01:43:50 +0000 (21:43 -0400)]
Add a 'since 2.30' index to gobject docs

12 years agoDon't distribute MAINTAINERS
Matthias Clasen [Mon, 29 Aug 2011 23:02:33 +0000 (19:02 -0400)]
Don't distribute MAINTAINERS

...it does not exist anymore.

12 years agoFirst round of NEWS updates for 2.29.18
Matthias Clasen [Mon, 29 Aug 2011 22:29:16 +0000 (18:29 -0400)]
First round of NEWS updates for 2.29.18

12 years agoGDBusObjectManagerServer: Add locking
David Zeuthen [Mon, 29 Aug 2011 20:46:45 +0000 (16:46 -0400)]
GDBusObjectManagerServer: Add locking

Signed-off-by: David Zeuthen <davidz@redhat.com>
12 years agoGDBusObjectManagerClient: Add locking
David Zeuthen [Mon, 29 Aug 2011 20:20:50 +0000 (16:20 -0400)]
GDBusObjectManagerClient: Add locking

Signed-off-by: David Zeuthen <davidz@redhat.com>
12 years agoGDBusObjectProxy: Add locking
David Zeuthen [Mon, 29 Aug 2011 19:29:27 +0000 (15:29 -0400)]
GDBusObjectProxy: Add locking

Signed-off-by: David Zeuthen <davidz@redhat.com>
12 years agoDon't call close on invalid fds
Matthias Clasen [Mon, 29 Aug 2011 19:18:29 +0000 (15:18 -0400)]
Don't call close on invalid fds

If an fd is -1, don't call close() on it, since that leads to
EBDAF and complaints from valgrind.

https://bugzilla.gnome.org/show_bug.cgi?id=657593

12 years agoGDBusObjectSkeleton: Add locking
David Zeuthen [Mon, 29 Aug 2011 19:19:42 +0000 (15:19 -0400)]
GDBusObjectSkeleton: Add locking

Signed-off-by: David Zeuthen <davidz@redhat.com>
12 years agoGDBusInterfaceSkeleton: Add locking
David Zeuthen [Mon, 29 Aug 2011 18:23:02 +0000 (14:23 -0400)]
GDBusInterfaceSkeleton: Add locking

Signed-off-by: David Zeuthen <davidz@redhat.com>
12 years agoPrint out URL to HTML coverage report after 'make lcov'
Stef Walter [Sun, 28 Aug 2011 04:58:29 +0000 (06:58 +0200)]
Print out URL to HTML coverage report after 'make lcov'

* For easy clicking

https://bugzilla.gnome.org/show_bug.cgi?id=657540

12 years agogdatetime: Tweak g_date_time_source_new() documentation
Colin Walters [Mon, 29 Aug 2011 19:05:34 +0000 (15:05 -0400)]
gdatetime: Tweak g_date_time_source_new() documentation

12 years agoSpelling fixes
Matthias Clasen [Mon, 29 Aug 2011 18:49:32 +0000 (14:49 -0400)]
Spelling fixes

Spelling fixes in comments and docs, provided by
Kjartan Maraas in bug 657336.

12 years agoAdd Linux timerfd_create() backend for g_date_time_source_new()
Colin Walters [Wed, 17 Aug 2011 17:09:27 +0000 (13:09 -0400)]
Add Linux timerfd_create() backend for g_date_time_source_new()

This makes the source efficient on Linux.

Tested on Fedora 15 x86_64 + updates, kernel-2.6.40-4.fc15.x86_64
Also tested fallback code for unsupported flag TFD_TIMER_CANCEL_ON_SET
on kernel 2.6.38.

https://bugzilla.gnome.org/show_bug.cgi?id=655129

12 years agogdatetime: Add g_date_time_source_new()
Colin Walters [Sat, 13 Aug 2011 12:55:20 +0000 (08:55 -0400)]
gdatetime: Add g_date_time_source_new()

Several different codebases in GNOME want to implement wall clocks.
While we could pretty easily share a private library, it's not a
substantial amount of code, and GLib already has a lot of the
necessary system-specific detection and handling infrastructure.

Note this initial implementation just wakes up once a second in the
cancel_on_set case; we'll add the Linux-specific handling in a
subsequent commit.

https://bugzilla.gnome.org/show_bug.cgi?id=655129

12 years agoUpdated Traditional Chinese translation(Hong Kong and Taiwan)
Chao-Hsiung Liao [Mon, 29 Aug 2011 11:18:28 +0000 (19:18 +0800)]
Updated Traditional Chinese translation(Hong Kong and Taiwan)

12 years agoUpdated Russian translation
Yuri Myasoedov [Sun, 28 Aug 2011 18:24:49 +0000 (22:24 +0400)]
Updated Russian translation

12 years agoUpdated Norwegian bokmål translation
Kjartan Maraas [Sun, 28 Aug 2011 17:07:09 +0000 (19:07 +0200)]
Updated Norwegian bokmål translation

12 years agoUpdated galician translations
Fran Dieguez [Sun, 28 Aug 2011 14:01:03 +0000 (16:01 +0200)]
Updated galician translations

12 years agogio/tests/gtlsconsoleinteraction: Work around missing getpass() on win32
Kalev Lember [Mon, 15 Aug 2011 13:09:18 +0000 (16:09 +0300)]
gio/tests/gtlsconsoleinteraction: Work around missing getpass() on win32

Loosely based on a patch by Sam Thursfield <ssssam@gmail.com>.
https://bugzilla.gnome.org/show_bug.cgi?id=656341

12 years agoUpdated Spanish translation
Jorge González [Sun, 28 Aug 2011 08:14:54 +0000 (10:14 +0200)]
Updated Spanish translation

12 years agoUpdated Spanish translation
Jorge González [Sun, 28 Aug 2011 08:08:05 +0000 (10:08 +0200)]
Updated Spanish translation

12 years agoFix symbol lists
Matthias Clasen [Sun, 28 Aug 2011 03:30:43 +0000 (23:30 -0400)]
Fix symbol lists

12 years agoSpell out file descriptors in a message
Matthias Clasen [Sat, 27 Aug 2011 17:18:45 +0000 (13:18 -0400)]
Spell out file descriptors in a message

Translators don't necessarily know what 'fds' are.
Pointed out in bug 657454.

12 years agoGDBusMessage: Use ngettext() where appropriate
Matthias Clasen [Sat, 27 Aug 2011 17:15:42 +0000 (13:15 -0400)]
GDBusMessage: Use ngettext() where appropriate

Pointed out in bug 657452.

12 years agogsocketaddress: don't return ABSTRACT if the OS doesn't support it
Dan Winship [Sat, 27 Aug 2011 16:08:32 +0000 (12:08 -0400)]
gsocketaddress: don't return ABSTRACT if the OS doesn't support it

FreeBSD apparently has non-0-length (but nameless) ANONYMOUS sockets.
Fix the heuristics here.

12 years agogunixcredentialsmessage: fix this on FreeBSD
Dan Winship [Sat, 27 Aug 2011 15:36:05 +0000 (11:36 -0400)]
gunixcredentialsmessage: fix this on FreeBSD

pointed out by Henry Hu
https://bugzilla.gnome.org/show_bug.cgi?id=628904

12 years agoGSocket: fix GIOCondition on timed-out socket operation
Dan Winship [Sat, 27 Aug 2011 13:59:02 +0000 (09:59 -0400)]
GSocket: fix GIOCondition on timed-out socket operation

The docs for g_socket_set_timeout() claimed that if an async operation
timed out, the GIOCondition passed to the source callback would be
G_IO_IN or G_IO_OUT (thus prompting the caller to call
g_socket_receive/send and get a G_IO_ERROR_TIMED_OUT), but in fact it
ended up being 0, and gio/tests/socket.c was erroneously testing for
that instead of the correct value. Fix this.

12 years agoUpdated Brazilian Portuguese translation.
Og B. Maciel [Fri, 26 Aug 2011 15:07:57 +0000 (11:07 -0400)]
Updated Brazilian Portuguese translation.

12 years agoFixed typo: s/ether/enter.
Og B. Maciel [Fri, 26 Aug 2011 15:07:32 +0000 (11:07 -0400)]
Fixed typo: s/ether/enter.

12 years agoUpdated Brazilian Portuguese translation.
Og B. Maciel [Fri, 26 Aug 2011 15:06:13 +0000 (11:06 -0400)]
Updated Brazilian Portuguese translation.

12 years agogiomodule: When loading GIO_EXTRA_MODULES skip duplicates
Stef Walter [Fri, 26 Aug 2011 13:27:19 +0000 (15:27 +0200)]
giomodule: When loading GIO_EXTRA_MODULES skip duplicates

 * Load modules from paths listed in GIO_EXTRA_MODULES environment
   variable first.
 * Ignore duplicate modules based on module basename.
 * Add the concept of GIOModuleScope which allows other callers to
   skip duplicate loaded modules, or block specific modules based on
   basename.
 * Document behavior.

https://bugzilla.gnome.org/show_bug.cgi?id=656914

12 years agoUpdated Brazilian Portuguese translation.
Og B. Maciel [Fri, 26 Aug 2011 13:04:41 +0000 (09:04 -0400)]
Updated Brazilian Portuguese translation.

12 years agoMake GTlsInteraction virtual methods cancellable
Stef Walter [Sat, 13 Aug 2011 12:03:53 +0000 (14:03 +0200)]
Make GTlsInteraction virtual methods cancellable

 * Add cancellable argument to g_tls_interaction_ask_password
   and g_tls_interaction_ask_password_async.
 * This is API breakage, but this API has not yet been released
   in a stable release (and very unlikely used yet).
 * Since we're breaking unreleased API, expand amount of padding
   on GTlsInteractionClass because we're going to need it.

https://bugzilla.gnome.org/show_bug.cgi?id=656443

12 years agoAdd gdbus-codegen to glib-zip.in
Dieter Verfaillie [Tue, 23 Aug 2011 05:37:05 +0000 (07:37 +0200)]
Add gdbus-codegen to glib-zip.in

https://bugzilla.gnome.org/show_bug.cgi?id=650763

12 years agoMake gdbus-codegen 'relocatable' at runtime on Windows.
Dieter Verfaillie [Tue, 23 Aug 2011 05:36:47 +0000 (06:36 +0100)]
Make gdbus-codegen 'relocatable' at runtime on Windows.

https://bugzilla.gnome.org/show_bug.cgi?id=650763

12 years agogdbus-codegen: Revert to using /usr/bin/env for Python
Dieter Verfaillie [Thu, 25 Aug 2011 09:45:40 +0000 (10:45 +0100)]
gdbus-codegen: Revert to using /usr/bin/env for Python

After a lot of discussion, we decided /usr/bin/env is the most
flexible for now.

https://bugzilla.gnome.org/show_bug.cgi?id=650763

12 years agoIntroduce the UNINSTALLED_GLIB_SRCDIR environment variable
Dieter Verfaillie [Tue, 23 Aug 2011 15:38:36 +0000 (16:38 +0100)]
Introduce the UNINSTALLED_GLIB_SRCDIR environment variable

This makes it possible to avoid setting PYTHONPATH in
Makefile.am files.

https://bugzilla.gnome.org/show_bug.cgi?id=650763

12 years agoIntroduce the UNINSTALLED_GLIB_BUILDDIR environment variable
Dieter Verfaillie [Tue, 23 Aug 2011 15:37:35 +0000 (17:37 +0200)]
Introduce the UNINSTALLED_GLIB_BUILDDIR environment variable

This makes it possible to also use relative imports for
gdbus-codegen's config module.

https://bugzilla.gnome.org/show_bug.cgi?id=650763

12 years agoUse relative imports for the gdbus/codegen package
Dieter Verfaillie [Tue, 23 Aug 2011 05:35:01 +0000 (07:35 +0200)]
Use relative imports for the gdbus/codegen package

But not yet for the config module (but is taken care
of in a followup patch).

https://bugzilla.gnome.org/show_bug.cgi?id=650763

12 years agoAvoid using - (hyphen) in gdbus-codegen directory name
Dieter Verfaillie [Tue, 23 Aug 2011 05:32:18 +0000 (07:32 +0200)]
Avoid using - (hyphen) in gdbus-codegen directory name

It's an invalid character in Python module names and prevents us from
being able to import it.

https://bugzilla.gnome.org/show_bug.cgi?id=650763

12 years agoUse detected PYTHON shebang in gdbus-codegen
Ionut Biru [Wed, 24 Aug 2011 19:09:53 +0000 (19:09 +0000)]
Use detected PYTHON shebang in gdbus-codegen

Change the hardcoded /usr/bin/python shebag from gdbus-codegen.in into
@PYTHON@. Is used in Makefile.am to use detected python binary.

$(AM_V_GEN) sed -e 's,@libdir\@,$(libdir),' -e 's,@PYTHON\@,$(PYTHON),'
$< > $@.tmp && mv $@.tmp $@

Signed-off-by: Ionut Biru <ibiru@archlinux.org>
https://bugzilla.gnome.org/show_bug.cgi?id=657274

12 years agoUpdated Russian translation
Yuri Myasoedov [Thu, 25 Aug 2011 14:59:56 +0000 (18:59 +0400)]
Updated Russian translation

12 years agoglib.doap: Fix download location
Javier Jardón [Thu, 25 Aug 2011 11:26:27 +0000 (12:26 +0100)]
glib.doap: Fix download location

12 years agoGCancellable: Small annotation additions
Matthias Clasen [Thu, 25 Aug 2011 01:23:36 +0000 (21:23 -0400)]
GCancellable: Small annotation additions

https://bugzilla.gnome.org/show_bug.cgi?id=657243

12 years agoUpdated Galician translations
Fran Dieguez [Wed, 24 Aug 2011 18:59:50 +0000 (20:59 +0200)]
Updated Galician translations

12 years agoUpdated Spanish translation
Daniel Mustieles [Wed, 24 Aug 2011 17:57:14 +0000 (19:57 +0200)]
Updated Spanish translation

12 years agogfileutils: fix docs/annotations for temp file methods
Dan Winship [Mon, 22 Aug 2011 15:37:48 +0000 (11:37 -0400)]
gfileutils: fix docs/annotations for temp file methods

https://bugzilla.gnome.org/show_bug.cgi?id=657084

12 years agoAdd DOAP file and remove MAINTAINERS
Emmanuele Bassi [Wed, 24 Aug 2011 17:20:32 +0000 (18:20 +0100)]
Add DOAP file and remove MAINTAINERS

12 years agoUpdated Norwegian bokmål translation
Kjartan Maraas [Wed, 24 Aug 2011 13:38:46 +0000 (15:38 +0200)]
Updated Norwegian bokmål translation

12 years agoUpdated Swedish translation
Daniel Nylander [Wed, 24 Aug 2011 07:19:20 +0000 (09:19 +0200)]
Updated Swedish translation

12 years agoBug 657206 — GInputStream leaked in g_file_icon_load_async()
Philip Withnall [Tue, 23 Aug 2011 22:47:54 +0000 (23:47 +0100)]
Bug 657206 — GInputStream leaked in g_file_icon_load_async()

Add a missing unref to g_file_icon_load_async().

Closes: bgo#657206

12 years agoGIO: Add more missing GI annotations
Eduardo Lima Mitev [Thu, 29 Jul 2010 11:59:42 +0000 (13:59 +0200)]
GIO: Add more missing GI annotations

 - g_io_extension_point_register()
 - g_io_extension_point_lookup()
 - g_io_extension_point_implement()

https://bugzilla.gnome.org/show_bug.cgi?id=625383

12 years agoUpdated Serbian translation
Мирослав Николић [Tue, 23 Aug 2011 17:02:55 +0000 (19:02 +0200)]
Updated Serbian translation

12 years agogdbus-codegen: update comment in code to reflect reality
David Zeuthen [Tue, 23 Aug 2011 16:57:14 +0000 (12:57 -0400)]
gdbus-codegen: update comment in code to reflect reality

Signed-off-by: David Zeuthen <davidz@redhat.com>
12 years agogdbus-codegen: Rework C property getters
David Zeuthen [Tue, 23 Aug 2011 16:46:32 +0000 (12:46 -0400)]
gdbus-codegen: Rework C property getters

Rework property getters to use a vfunc so we can take the fast path
and avoid allocating memory for both the skeleton and the proxy
cases. This requires some special case because of how GVariant expects
you to free memory in some cases, see #657100. Add test cases for
this.

Document the _get_ functions as not being thread-safe and also
generate _dup_ C getters (which are thread-safe).

Mark all the generated _get_, _dup_ and _set_ as (skip) as non-C
languages should just use GObject properties and not the (socalled)
"C binding".

Signed-off-by: David Zeuthen <davidz@redhat.com>
12 years agopo: add missing gtls files to POTFILES.in
Dan Winship [Tue, 23 Aug 2011 12:28:15 +0000 (08:28 -0400)]
po: add missing gtls files to POTFILES.in

Pointed out by Daniel Mustieles
https://bugzilla.gnome.org/show_bug.cgi?id=657138

12 years agogdatetime: Don't include langinfo.h unconditionally
Matthias Clasen [Tue, 23 Aug 2011 02:01:39 +0000 (22:01 -0400)]
gdatetime: Don't include langinfo.h unconditionally

It doesn't exist on Windows.

https://bugzilla.gnome.org/show_bug.cgi?id=657083

12 years agoUpdate VS property sheets
Chun-wei Fan [Mon, 22 Aug 2011 16:22:06 +0000 (00:22 +0800)]
Update VS property sheets

Stop the "installation" of gio/gtimezonemonitor.h as it has been removed
from GIO (commit 5b68b49b2072)

12 years agoUpdate GLib Visual C++ Projects
Chun-wei Fan [Mon, 22 Aug 2011 16:08:24 +0000 (00:08 +0800)]
Update GLib Visual C++ Projects

Define USE_SYSTEM_PCRE for all configurations which uses the PCRE that
was already built and "installed" beforehand (i.e. the *_ExtPCRE
configurations) so that the compilation will not pick up the
GLib-bundled pcre.h when one wants to use the PCRE "installation" on
his/her system.

12 years agoUpdate config.h.win32.in
Chun-wei Fan [Mon, 22 Aug 2011 16:03:51 +0000 (00:03 +0800)]
Update config.h.win32.in

Make the pre-configured config.h(.win32.in) for Windows more like the
config.h that would be produced during ./configure on Windows systems.

12 years agoGTimeZoneMonitor: Revert addition of this class
Colin Walters [Fri, 19 Aug 2011 07:27:16 +0000 (03:27 -0400)]
GTimeZoneMonitor: Revert addition of this class

The main rationale for adding it was to avoid having gnome-shell
mmap'ing /etc/localtime once a second.  However, we can just as easily
run inotify there, and given no one else was clamoring for a way to
detect when the time zone changes, I don't see a need for public API
here - at least not yet.

In the bigger picture, I just don't believe that the vast majority of
applications are going to go out of their way to instantiate and keep
around a random GTimeZoneMonitor class.  And if they do, it's has the
side effect that for other bits of code in the process, local GDateTime
instances may start varying again!

So, if code can't rely on local GDateTime instances being in a
consistent state anyways, let's just do that always.  The
documentation now says that this is the case.  Applications have
always been able to work in a consistent local time zone by
instantiating a zone and then using it for GDateTime constructors.

We fix the "gnome-shell stats /etc/localtime once a second" issue by
using timerfd (in glib) and inotify (in gnome-shell).

https://bugzilla.gnome.org/show_bug.cgi?id=655129

12 years agoUpdate VS property sheets
Chun-wei Fan [Mon, 22 Aug 2011 14:35:31 +0000 (22:35 +0800)]
Update VS property sheets

-Added glib/ghmac.h to the list of files to copy during the "install" stage
-Cleaned up a bit (glib-2.0->glib-$(ApiVersion), where $(ApiVersion) is
 2.0)

12 years agogmain: Clarify that timeouts are in terms of monotonic time
Colin Walters [Sat, 13 Aug 2011 10:03:18 +0000 (06:03 -0400)]
gmain: Clarify that timeouts are in terms of monotonic time

Also note that monotonic time does not include time spent while
suspended (at least on Linux).

https://bugzilla.gnome.org/show_bug.cgi?id=655129

12 years agoGDateTime: use nl_langinfo() when available
Matthias Clasen [Mon, 22 Aug 2011 03:48:07 +0000 (23:48 -0400)]
GDateTime: use nl_langinfo() when available

This makes g_date_time_format() react to LC_TIME, which is
what people expect.

Translators: this change means that the GDateTime strings
are only used when the C library does not already provide
suitable translated strings for these (month names, etc).

12 years agoGDateTime: cosmetics
Matthias Clasen [Mon, 22 Aug 2011 03:36:31 +0000 (23:36 -0400)]
GDateTime: cosmetics

Shuffle things around a bit, to move locale-dependent
things together.

12 years agoGDateTime: cosmetics
Matthias Clasen [Mon, 22 Aug 2011 03:32:28 +0000 (23:32 -0400)]
GDateTime: cosmetics

Don't hide the recursion in g_date_time_format() behind
a macro, make it explicit.

12 years agoAdd a quick utility to test datetime formatting
Matthias Clasen [Mon, 22 Aug 2011 01:33:30 +0000 (21:33 -0400)]
Add a quick utility to test datetime formatting

12 years agoGDateTime: don't use separate strings for upper/lowercase am/pm
Matthias Clasen [Mon, 22 Aug 2011 00:45:57 +0000 (20:45 -0400)]
GDateTime: don't use separate strings for upper/lowercase am/pm

We can just as well change the case ourselves.

12 years agog_date_time_format: support %T
Matthias Clasen [Sun, 21 Aug 2011 23:51:45 +0000 (19:51 -0400)]
g_date_time_format: support %T

strftime supports it, and there is no reason not to do the same.

12 years ago[gi] Add signal annotations
Pavel Holejsovsky [Sun, 21 Aug 2011 15:15:52 +0000 (17:15 +0200)]
[gi] Add signal annotations

12 years ago[gi] Add annotations to g_quark functions
Pavel Holejsovsky [Sun, 21 Aug 2011 15:12:18 +0000 (17:12 +0200)]
[gi] Add annotations to g_quark functions

12 years agoAdd missing GClosure annotations
Pavel Holejsovsky [Fri, 19 Aug 2011 08:34:02 +0000 (10:34 +0200)]
Add missing GClosure annotations

12 years agoAdd missing (allow-none) annotations to g_value_set_string functions
Pavel Holejsovsky [Fri, 19 Aug 2011 08:34:37 +0000 (10:34 +0200)]
Add missing (allow-none) annotations to g_value_set_string functions

12 years agoMerge branch 'concurrent-cancellable'
Stef Walter [Fri, 19 Aug 2011 09:21:11 +0000 (11:21 +0200)]
Merge branch 'concurrent-cancellable'

12 years agogio: GCancellable can be used concurrently
Stef Walter [Fri, 12 Aug 2011 09:49:31 +0000 (11:49 +0200)]
gio: GCancellable can be used concurrently

 * Update documentation to note that GCancellable can be used
   concurrently by multiple operations.
 * Add documentation to g_cancellable_reset that behavior is
   undefined if called from within cancelled handler.
 * Add test for multiple concurrent operations using the same
   cancellable.

https://bugzilla.gnome.org/show_bug.cgi?id=656387

12 years agoupdate translation for Punjabi
A S Alam [Fri, 19 Aug 2011 02:56:25 +0000 (08:26 +0530)]
update translation for Punjabi

12 years agoAdded UG translation
Abduxukur Abdurixit [Fri, 19 Aug 2011 17:42:26 +0000 (19:42 +0200)]
Added UG translation

12 years agoUse actual uint64 values to compare uint64 variants
Christian Dywan [Wed, 17 Aug 2011 23:38:34 +0000 (01:38 +0200)]
Use actual uint64 values to compare uint64 variants

Fixes: https://bugzilla.gnome.org/show_bug.cgi?id=656772

12 years agoAdd (allow-none) annotation for GValue setters.
Pavel Holejsovsky [Wed, 17 Aug 2011 05:37:36 +0000 (07:37 +0200)]
Add (allow-none) annotation for GValue setters.

Assorted g_value_set_xxx() and g_value_take_xxx() functions actually
allow NULL in source parameter.

12 years agoFixup max decomposition len guarantee
Behdad Esfahbod [Wed, 17 Aug 2011 10:13:17 +0000 (12:13 +0200)]
Fixup max decomposition len guarantee

Unicode Technical Committee agreed to limit decomposition length to
18 in both cases.  Reflect that.

12 years agovoid functions should not return. Fixes bug #656675.
Brian Cameron [Wed, 17 Aug 2011 08:01:14 +0000 (03:01 -0500)]
void functions should not return.  Fixes bug #656675.

12 years agopost-release bump
Ryan Lortie [Tue, 16 Aug 2011 01:01:52 +0000 (21:01 -0400)]
post-release bump

12 years agoglib 2.29.16 2.29.16
Ryan Lortie [Tue, 16 Aug 2011 01:01:28 +0000 (21:01 -0400)]
glib 2.29.16

12 years agodistcheck fixes
Ryan Lortie [Tue, 16 Aug 2011 01:01:23 +0000 (21:01 -0400)]
distcheck fixes

12 years agoMore NEWS updates
Matthias Clasen [Mon, 15 Aug 2011 22:51:42 +0000 (18:51 -0400)]
More NEWS updates

12 years agoUpdated Esperanto translation
Ryan Lortie [Mon, 15 Aug 2011 17:33:27 +0000 (13:33 -0400)]
Updated Esperanto translation

Esperanto weekdays and month days are written in lowercase (as in
French).

12 years agoAdd a stress-test for GDBusProxy in threads with no default main context
Simon McVittie [Mon, 15 Aug 2011 15:00:51 +0000 (16:00 +0100)]
Add a stress-test for GDBusProxy in threads with no default main context

Destroying a GDBusProxy in a thread used to race with NameOwnerChanged
being delivered to the main context's thread (GNOME #651133).

Also, g_dbus_proxy_call_sync in a thread would race with NameOwnerChanged
being delivered to the main context's thread and rewriting the name_owner
(GNOME #656039).

Bug: https://bugzilla.gnome.org/show_bug.cgi?id=656039
Bug-NB: NB#259760
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Signed-off-by: David Zeuthen <davidz@redhat.com>
12 years agoGDBusProxy: hold properties_lock while using any mutable property
Simon McVittie [Mon, 15 Aug 2011 14:59:45 +0000 (15:59 +0100)]
GDBusProxy: hold properties_lock while using any mutable property

This changes the meaning of "properties_lock" from "lock for D-Bus
properties" to "lock for GObject properties".

The most common problem, and the only one I've reproduced in a regression
test, is name_owner, which can be updated by the thread that owns
the GDBusProxy's main context (i.e. the thread-default main context of
the thread that constructed it) at the same time that a blocking call
is made. When a GDBusProxy is constructed in a thread-pool thread for
short-term use, the main context will typically be the global default
main context (which is actively running in the main thread!), making
this extremely problematic.

The interface info is perhaps a theoretical concern - one thread could
conceivably set it at the same time that another thread uses it, but only
in relatively pathological situations. The current API for this does have
the problem that it returns a borrowed ref, but interface info is
hopefully permanent anyway.

The default timeout is probably only a theoretical concern - it's just an
int, so writes are indivisible, and there's no worry about whether
something has been freed - but to be safe, let's hold the lock for that
too.

Bug: https://bugzilla.gnome.org/show_bug.cgi?id=656039
Bug-NB: NB#259760

Signed-off-by: David Zeuthen <davidz@redhat.com>
12 years agoGDBusProxy: factor out async_init_data_set_name_owner
Simon McVittie [Mon, 8 Aug 2011 17:29:20 +0000 (18:29 +0100)]
GDBusProxy: factor out async_init_data_set_name_owner

This removes the need for async_init_get_name_owner_cb to cope with being
called without a real GAsyncResult, and will simplify the addition of
correct thread-locking.

In async_init_data_set_name_owner, use the name_owner parameter instead
of the corresponding member of GDBusProxyPrivate, partly to reduce
pointer-chasing but mainly to avoid needing to hold the lock.

Signed-off-by: David Zeuthen <davidz@redhat.com>
12 years agog_dbus_proxy_get_property: use accessors for all mutable state
Simon McVittie [Mon, 8 Aug 2011 17:22:42 +0000 (18:22 +0100)]
g_dbus_proxy_get_property: use accessors for all mutable state

These ought to have thread-locking, and having it in the accessor seems
better than duplicating it here.

Signed-off-by: David Zeuthen <davidz@redhat.com>
12 years agoGDBusProxy: if a well-known name is not owned, don't GetAll from the dbus-daemon
Simon McVittie [Mon, 15 Aug 2011 14:57:59 +0000 (15:57 +0100)]
GDBusProxy: if a well-known name is not owned, don't GetAll from the dbus-daemon

If you run:

    ( cd gio/tests && G_DBUS_DEBUG=all ./gdbus-proxy-well-known-name )

you can see that in the case where the name com.example.TestService isn't
owned yet, the GDBusProxy calls GetAll() with no destination, resulting
in an error reply from the peer (the dbus-daemon itself). That's clearly
not right!

However, if priv->name is NULL, that indicates the special case where we
really do want to talk directly to a peer, instead of via the bus daemon
(most likely to be used on peer-to-peer connections); in that special
case, do call GetAll().

Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Signed-off-by: David Zeuthen <davidz@redhat.com>