platform/upstream/glib.git
11 years agoImprove test coverage in gobject/
Matthias Clasen [Wed, 29 May 2013 04:15:55 +0000 (00:15 -0400)]
Improve test coverage in gobject/

Lines:          6631    8862    74.8 %
Functions:      747     893     83.7 %

11 years agoGIcon: NULLify the `type' out param in the sync methods too
Emanuele Aina [Mon, 20 May 2013 14:53:40 +0000 (15:53 +0100)]
GIcon: NULLify the `type' out param in the sync methods too

Both g_[file|bytes]_icon_load() leave the `type' out parameter
untouched, while the async methods g_[file|bytes]_icon_load_finish()
always set it to NULL.

For consistency's sake NULLify it in the sync methods too.

https://bugzilla.gnome.org/show_bug.cgi?id=700725

11 years agoUpdated Slovenian translation
Matej Urbančič [Tue, 28 May 2013 20:26:50 +0000 (22:26 +0200)]
Updated Slovenian translation

11 years agoupdated kn translations
Shankar Prasad [Tue, 28 May 2013 12:58:09 +0000 (18:28 +0530)]
updated kn translations

11 years agoUpdate Visual Studio Project Sheets
Chun-wei Fan [Tue, 28 May 2013 04:55:56 +0000 (12:55 +0800)]
Update Visual Studio Project Sheets

"Install" the newly-introduced gio/gbytesicon.h...

11 years agoGLib 2.37.1 2.37.1
Ryan Lortie [Mon, 27 May 2013 21:34:21 +0000 (17:34 -0400)]
GLib 2.37.1

11 years agoMore test fixes for builddir != srcdir
Matthias Clasen [Tue, 28 May 2013 02:05:19 +0000 (22:05 -0400)]
More test fixes for builddir != srcdir

11 years agoFix make check with builddir != srcdir
Matthias Clasen [Tue, 28 May 2013 01:20:10 +0000 (21:20 -0400)]
Fix make check with builddir != srcdir

This broke when the tests were converted to be installable.
My apologies.

11 years agogio/tests: Make gdbus-proxy-well-known-name handle srcdir != builddir
Colin Walters [Mon, 27 May 2013 23:30:31 +0000 (19:30 -0400)]
gio/tests: Make gdbus-proxy-well-known-name handle srcdir != builddir

11 years agoFix yet more test regressions...
Ryan Lortie [Mon, 27 May 2013 22:34:33 +0000 (18:34 -0400)]
Fix yet more test regressions...

11 years agoOne more broken test....
Ryan Lortie [Mon, 27 May 2013 22:30:53 +0000 (18:30 -0400)]
One more broken test....

11 years agoMore srcdir != destdir tests fallout
Ryan Lortie [Mon, 27 May 2013 22:27:14 +0000 (18:27 -0400)]
More srcdir != destdir tests fallout

11 years agotests: fix a srcdir vs. destdir issue
Ryan Lortie [Mon, 27 May 2013 22:05:09 +0000 (18:05 -0400)]
tests: fix a srcdir vs. destdir issue

The target file is a script, so it'll always be in SRCDIR for the uninstalled
case.  Just look there and avoid the libtool trickery.

11 years agoUpdated Odia Translation.
ManojKumar Giri [Mon, 27 May 2013 12:39:17 +0000 (18:09 +0530)]
Updated Odia Translation.

11 years agoTamil Translation Updated
Shantha kumar [Mon, 27 May 2013 10:01:22 +0000 (15:31 +0530)]
Tamil Translation Updated

11 years agoFix the GObject Visual Studio Projects
Chun-wei Fan [Mon, 27 May 2013 07:57:54 +0000 (15:57 +0800)]
Fix the GObject Visual Studio Projects

Update G_LOG_DOMAIN to be "GLib-GObject" so that we are consistent with
the autotools builds, and that tests expecting the log domain to be
"GLib-GObject" would not fail.

11 years agoFix the GLib Visual Studio Projects
Chun-wei Fan [Mon, 27 May 2013 05:23:58 +0000 (13:23 +0800)]
Fix the GLib Visual Studio Projects

Define the G_LOG_DOMAIN of the GLib DLL as "GLib", because:
-This makes it consistent with the autotools builds
-Some tests expect the log domain to be "GLib"

11 years agoUpdate config.h.win32(.in)
Chun-wei Fan [Mon, 27 May 2013 04:49:47 +0000 (12:49 +0800)]
Update config.h.win32(.in)

Make the entries of config.h.win32(.in) consistent with the entries
that are generated from the autotools build (config.h.in).

11 years agoAvoid a segfault in gdbus tool
Matthias Clasen [Sun, 26 May 2013 03:07:03 +0000 (23:07 -0400)]
Avoid a segfault in gdbus tool

When the interface name is invalid, we don't get an error
back from g_dbus_connection_call_sync.

11 years agobuild: Fix usage of %.test again
Colin Walters [Fri, 24 May 2013 21:16:44 +0000 (22:16 +0100)]
build: Fix usage of %.test again

We actually need the first dependency because it includes the
final executable name.  Rather, fix the original bug by using
the variable $(EXEEXT).

11 years agotests: Drop unnecessary % from .test pattern match rule
Colin Walters [Fri, 24 May 2013 19:42:12 +0000 (15:42 -0400)]
tests: Drop unnecessary % from .test pattern match rule

On Windows, the executables will have .exe, so this won't
match.  Furthermore, they aren't actually dependent on the
executable to build.

11 years agoUpdated Spanish translation
Daniel Mustieles [Fri, 24 May 2013 11:57:47 +0000 (13:57 +0200)]
Updated Spanish translation

11 years agoAssamese translation updated
Nilamdyuti Goswami [Fri, 24 May 2013 11:58:12 +0000 (17:28 +0530)]
Assamese translation updated

11 years agoUpdated Gujarati Translations
Sweta Kothari [Fri, 24 May 2013 07:51:27 +0000 (13:21 +0530)]
Updated Gujarati Translations

11 years agoG_GNUC_FORMAT: documentation error
Simon Kågedal Reimer [Mon, 6 May 2013 20:04:32 +0000 (22:04 +0200)]
G_GNUC_FORMAT: documentation error

Attribute should be placed just before the semicolon, not after.
As can be seen in the example.

https://bugzilla.gnome.org/show_bug.cgi?id=699779

11 years agospelling fixes of 'runtine' and 'adresses' in cross.xml and running.xml, respectively
Jason L. Quinn [Fri, 12 Apr 2013 03:43:17 +0000 (23:43 -0400)]
spelling fixes of 'runtine' and 'adresses' in cross.xml and running.xml, respectively

https://bugzilla.gnome.org/show_bug.cgi?id=697849

11 years agogsignal: remove some pointless locking
Ryan Lortie [Thu, 21 Feb 2013 17:48:04 +0000 (17:48 +0000)]
gsignal: remove some pointless locking

We previously hold a lock in the loop that collects the arguments for
g_signal_emit(), which we drop before calling into the argument
collection functions and reacquire again at the bottom of the loop (ie:
one release/acquire pair for each argument collected).  To make matters
worse, the lock is just released again after the loop.

Presumably that was done to protect the access to the parameter array,
but it's pretty unlikely that this is needed because the only way it
changes is if the signal is unloaded.  That only happens when unloading
types which is quite unlikely to happen while we are emitting on an
instance of that type (and, as an aside, never happens anymore anyway).

If we move the unlock below the loop up above it and remove the
acquire/release pair from the loop, we improve performance in the new
arg-collecting performance tests by ~15% (more like ~18% in the case
where we only emit to one handler -- where argument collection dominates
more).

https://bugzilla.gnome.org/show_bug.cgi?id=694380

11 years agoperformance test: add signal test with args
Ryan Lortie [Thu, 21 Feb 2013 17:47:08 +0000 (17:47 +0000)]
performance test: add signal test with args

Add a signal that has some typical arguments (a uint and a pointer)
since all of the other signal performance tests are for signals with no
args.

https://bugzilla.gnome.org/show_bug.cgi?id=694380

11 years agoperformance test: share some code
Ryan Lortie [Thu, 21 Feb 2013 17:44:56 +0000 (17:44 +0000)]
performance test: share some code

The handled and unhandled cases share the same data and _run()
functions.  Refactor into a common section.

https://bugzilla.gnome.org/show_bug.cgi?id=694380

11 years agogunixconnection: added ngettext to support plural forms
Elisabeth Henry [Mon, 1 Apr 2013 02:11:01 +0000 (04:11 +0200)]
gunixconnection: added ngettext to support plural forms

While those strings ("Expecting 1 control message, got %d" and
"Expecting one fd, but got %d\n") have same singular/plural form
in english, it is not necessarily the case in other languages.

https://bugzilla.gnome.org/show_bug.cgi?id=695233

11 years agoGLocalFileOutputStream: Further deduplicate error code path
Colin Walters [Thu, 9 May 2013 17:47:50 +0000 (18:47 +0100)]
GLocalFileOutputStream: Further deduplicate error code path

Just code cleanup.

https://bugzilla.gnome.org/699959

11 years agoGLocalFileOutputStream: Deduplicate stream creation code
Colin Walters [Thu, 9 May 2013 17:39:50 +0000 (18:39 +0100)]
GLocalFileOutputStream: Deduplicate stream creation code

Lots of copy/paste of the error handling path, let's deduplicate
so I can sanely patch this code later.

https://bugzilla.gnome.org/699959

11 years agogdbus-peer: Drop some usage of g_thread_yield()
Colin Walters [Wed, 22 May 2013 16:41:32 +0000 (17:41 +0100)]
gdbus-peer: Drop some usage of g_thread_yield()

It's a recipe for race conditions and error; on some hardware
architectures one thread isn't guaranteed to see the results
of writes from another thread without a cache flush.

https://bugzilla.gnome.org/show_bug.cgi?id=700855

11 years agoglib/tests/cond: Fix race condition
Colin Walters [Thu, 23 May 2013 00:07:13 +0000 (01:07 +0100)]
glib/tests/cond: Fix race condition

The producer could push two values with the first one being
lost.  Fix this by blocking the producer until a consumer
reads.

11 years agogtestutils: Use stdio rather than write() to be Windows-friendly
Dan Winship [Wed, 22 May 2013 17:20:08 +0000 (14:20 -0300)]
gtestutils: Use stdio rather than write() to be Windows-friendly

Windows doesn't define STDOUT_FILENO and STDERR_FILENO, and they're
not even guaranteed to be 1 and 2. So just use stdio instead. Also fix
a counting error. Pointed out on gtk-devel-list.

11 years agoTimeout the test if dbus sevice has not appeared in due time.
Josep Puigdemont [Tue, 21 May 2013 14:46:25 +0000 (16:46 +0200)]
Timeout the test if dbus sevice has not appeared in due time.

The test /gdbus/connection/large_message waits for a dbus name to appear.
The dbus name is created by a another process executed in the background.
If for some reason this fails, the test will likely wait forever.
This will avoid this situation by making the test fail if the dbus service
has not appeared after 10 seconds.

https://bugzilla.gnome.org/show_bug.cgi?id=698981

11 years agopo: update for quote changes
Dan Winship [Tue, 21 May 2013 12:52:32 +0000 (09:52 -0300)]
po: update for quote changes

https://bugzilla.gnome.org/show_bug.cgi?id=700746

11 years agoUse 'dumb quotes' rather than `really dumb quotes'
Dan Winship [Mon, 20 May 2013 20:54:48 +0000 (17:54 -0300)]
Use 'dumb quotes' rather than `really dumb quotes'

Back in the far-off twentieth century, it was normal on unix
workstations for U+0060 GRAVE ACCENT to be drawn as "‛" and for U+0027
APOSTROPHE to be drawn as "’". This led to the convention of using
them as poor-man's ‛smart quotes’ in ASCII-only text.

However, "'" is now universally drawn as a vertical line, and "`" at a
45-degree angle, making them an `odd couple' when used together.

Unfortunately, there are lots of very old strings in glib, and also
lots of new strings in which people have kept up the old tradition,
perhaps entirely unaware that it used to not look stupid.

Fix this by just using 'dumb quotes' everywhere.

https://bugzilla.gnome.org/show_bug.cgi?id=700746

11 years agotests: fix a few recently-broken tests
Dan Winship [Tue, 21 May 2013 14:21:36 +0000 (11:21 -0300)]
tests: fix a few recently-broken tests

11 years agogio/tests: Fix gdbus-connection when run from "make check"
Colin Walters [Tue, 21 May 2013 13:21:21 +0000 (09:21 -0400)]
gio/tests: Fix gdbus-connection when run from "make check"

11 years agoConvert more tests to installed tests
Matthias Clasen [Tue, 21 May 2013 01:05:56 +0000 (21:05 -0400)]
Convert more tests to installed tests

This makes the tests in tests installed, except for
the performance tests.

11 years agoConvert more tests to installed tests
Matthias Clasen [Tue, 21 May 2013 01:05:56 +0000 (21:05 -0400)]
Convert more tests to installed tests

11 years agogio/tests: gsettings is a developer-only test
Colin Walters [Mon, 20 May 2013 23:12:31 +0000 (00:12 +0100)]
gio/tests: gsettings is a developer-only test

It tries to run glib-compile-schemas and glib-mkenums, which
we won't have in the runtime tree.

Anyways it's kind of a dumb test since the best test for
compilation tools is...compiling things, which we already
do frequently.

11 years agogmessages: Fix up testing commit to handle fatal masquerading
Colin Walters [Mon, 20 May 2013 23:07:18 +0000 (00:07 +0100)]
gmessages: Fix up testing commit to handle fatal masquerading

This makes glib/tests/testing pass again.

11 years agogio/tests: Disable desktop-app-info test in installed mode for now
Colin Walters [Mon, 20 May 2013 23:02:39 +0000 (00:02 +0100)]
gio/tests: Disable desktop-app-info test in installed mode for now

It wants a writable copy of the applications, and appears to
depend on a prior execution of "mimeapps".  This will take
a bit of work to untangle.

11 years agogio/tests: Make gdbus-peer work in installed mode
Colin Walters [Mon, 20 May 2013 20:50:28 +0000 (21:50 +0100)]
gio/tests: Make gdbus-peer work in installed mode

There's no /etc/hosts in gnome-ostree...

11 years agogio/tests: Fix gapplication test in installed mode
Colin Walters [Mon, 20 May 2013 20:43:52 +0000 (21:43 +0100)]
gio/tests: Fix gapplication test in installed mode

11 years agogio/tests: Some more fixes for installed tests
Colin Walters [Mon, 20 May 2013 20:33:00 +0000 (21:33 +0100)]
gio/tests: Some more fixes for installed tests

file passes now, appinfo needs a bit more work, but is getting
there.

11 years agogio/tests: Deduplicate Makefile.am a bit
Colin Walters [Mon, 20 May 2013 19:15:43 +0000 (20:15 +0100)]
gio/tests: Deduplicate Makefile.am a bit

We can just add all the static test data to EXTRA_DIST.

11 years agogtestutils: Ensure test subprocesses don't dump core
Colin Walters [Mon, 20 May 2013 13:31:48 +0000 (14:31 +0100)]
gtestutils: Ensure test subprocesses don't dump core

Since we expect them to crash, let's not spam the system
core dump collection (systemd, abrt).  At the moment
systemd is not very robust against programs crashing
in loops.

Instead of aborting, we exit(1).

https://bugzilla.gnome.org/show_bug.cgi?id=700714

11 years agoUpdated Norwegian bokmål translation
Kjartan Maraas [Mon, 20 May 2013 12:51:04 +0000 (14:51 +0200)]
Updated Norwegian bokmål translation

11 years agoConvert remaining gio tests to installed
Matthias Clasen [Mon, 20 May 2013 12:46:21 +0000 (08:46 -0400)]
Convert remaining gio tests to installed

11 years agoConvert some gio tests to installed tests
Matthias Clasen [Mon, 20 May 2013 10:38:41 +0000 (06:38 -0400)]
Convert some gio tests to installed tests

11 years agoMake gobject tests installable
Matthias Clasen [Mon, 20 May 2013 01:49:51 +0000 (21:49 -0400)]
Make gobject tests installable

This makes the gobject tests run as part of the ostree integration
tests.

11 years ago[l10n] Updated Italian translation.
Milo Casagrande [Sun, 19 May 2013 11:30:46 +0000 (13:30 +0200)]
[l10n] Updated Italian translation.

11 years agoAdd --enable-installed-tests configure option
Colin Walters [Mon, 6 May 2013 13:47:00 +0000 (09:47 -0400)]
Add --enable-installed-tests configure option

See https://live.gnome.org/GnomeGoals/InstalledTests for more
information.

The tests now support being run both uninstalled and installed, so
'make check' works for those who want it.  For tests which need data
files, the way this works is they look in the compiled in value of
SRCDIR by default, and the generated tests use "env G_TEST_DATA=" to
override that.

This patch only converts glib/tests for now; if this patch looks good,
I'll do the rest of the tests.

https://bugzilla.gnome.org/show_bug.cgi?id=699079

11 years agoglib/tests: Drop unnecessary SRCDIR definitions
Colin Walters [Mon, 6 May 2013 13:24:25 +0000 (09:24 -0400)]
glib/tests: Drop unnecessary SRCDIR definitions

It's always going to be defined, we don't support compiling these
tests "by hand".  This is a cleanup commit for the installed tests
work.

https://bugzilla.gnome.org/show_bug.cgi?id=699079

11 years agoglib/tests: Use explicit file listings
Colin Walters [Fri, 3 May 2013 22:36:43 +0000 (18:36 -0400)]
glib/tests: Use explicit file listings

This is a preparatory commit for installed tests work; having things
cleanly categorized helps.

https://bugzilla.gnome.org/show_bug.cgi?id=699079

11 years agoglib/tests/mappedfile: Copy test file before writing to it
Colin Walters [Fri, 3 May 2013 22:13:51 +0000 (18:13 -0400)]
glib/tests/mappedfile: Copy test file before writing to it

This way we're *always* testing writability rather than only sometimes
doing it in the source directory.

https://bugzilla.gnome.org/show_bug.cgi?id=699079

11 years agoAdded Aragonese translation
Jorge Pérez Pérez [Wed, 15 May 2013 21:05:10 +0000 (23:05 +0200)]
Added Aragonese translation

11 years agom4macros/glib-gettext.m4: Don't use AC_HEADER_STDC
Adrian Bunk [Mon, 13 May 2013 22:48:37 +0000 (01:48 +0300)]
m4macros/glib-gettext.m4: Don't use AC_HEADER_STDC

AC_HEADER_STDC is obsolete (pre-C89 headers are no longer an issue),
and the resulting STDC_HEADERS define is anyway not used.

https://bugzilla.gnome.org/show_bug.cgi?id=700263

11 years agoupdated kn translations
Shankar Prasad [Tue, 14 May 2013 09:56:07 +0000 (15:26 +0530)]
updated kn translations

11 years agotests/slice: use G_GNUC_BEGIN_IGNORE_DEPRECATIONS
Dan Winship [Mon, 13 May 2013 15:47:12 +0000 (11:47 -0400)]
tests/slice: use G_GNUC_BEGIN_IGNORE_DEPRECATIONS

rather than implementing it by hand

11 years agogtestutils: use g_clear_pointer
Dan Winship [Mon, 13 May 2013 15:38:53 +0000 (11:38 -0400)]
gtestutils: use g_clear_pointer

11 years agogtestutils: deprecate g_test_trap_fork()
Dan Winship [Wed, 5 Dec 2012 16:28:53 +0000 (11:28 -0500)]
gtestutils: deprecate g_test_trap_fork()

https://bugzilla.gnome.org/show_bug.cgi?id=679683

11 years agotests: port from g_test_trap_subprocess() to g_test_trap_fork()
Dan Winship [Sat, 24 Nov 2012 20:58:36 +0000 (15:58 -0500)]
tests: port from g_test_trap_subprocess() to g_test_trap_fork()

https://bugzilla.gnome.org/show_bug.cgi?id=679683

11 years agotests/protocol: redo a bit
Dan Winship [Sun, 25 Nov 2012 16:50:09 +0000 (11:50 -0500)]
tests/protocol: redo a bit

Rather than overloading --verbose, just skip the tests that aren't
supposed to be run in the parent process (so that if you do run the
toplevel test with --verbose, it doesn't immediately error out).

https://bugzilla.gnome.org/show_bug.cgi?id=679683

11 years agogtestutils: add g_test_trap_subprocess()
Dan Winship [Sat, 24 Nov 2012 20:58:27 +0000 (15:58 -0500)]
gtestutils: add g_test_trap_subprocess()

g_test_trap_fork() doesn't work on Windows and is potentially flaky on
unix anyway given the fork-but-don't-exec. Replace it with
g_test_trap_subprocess(), which re-spawns the same program with
arguments telling it to run a specific (otherwise-ignored) test case.

Make the existing g_test_trap_fork() unit tests be unix-only (they
never passed on Windows anyway), and add a parallel set of
g_test_trap_subprocess() tests.

Also fix the logic of gtestutils's "-p" argument (which is used by the
subprocess tests); previously if you had tests "/foo/bar" and
"/foo/bar/baz", and ran the test program with "-p /foo/bar/baz", it
would run "/foo/bar" too. Fix that and add tests.

https://bugzilla.gnome.org/show_bug.cgi?id=679683

11 years agog_test_trap_fork: don't blow away the SIGCHLD handler
Dan Winship [Wed, 19 Dec 2012 20:59:55 +0000 (15:59 -0500)]
g_test_trap_fork: don't blow away the SIGCHLD handler

Not sure why it was doing this, but it's not necessary (all of glib's
tests pass fine without it), and it breaks tests that try to use
g_spawn_sync() or GChildWatchSource after doing a g_test_trap_fork().

https://bugzilla.gnome.org/show_bug.cgi?id=679683

11 years agotests/spawn-*.c: fix on Windows
Dan Winship [Sun, 25 Nov 2012 18:52:20 +0000 (13:52 -0500)]
tests/spawn-*.c: fix on Windows

Need to append ".exe" to the spawned binary name on Windows

https://bugzilla.gnome.org/show_bug.cgi?id=679683

11 years agoapp: Document when floating variants are consumed
Christian Persch [Sun, 12 May 2013 19:52:54 +0000 (21:52 +0200)]
app: Document when floating variants are consumed

Bug #700203.

11 years agoUpdated Odia Translation.
ManojKumar Giri [Fri, 10 May 2013 06:38:27 +0000 (12:08 +0530)]
Updated Odia Translation.

11 years agoClarify GValueArray docs
Matthias Clasen [Thu, 9 May 2013 20:02:52 +0000 (16:02 -0400)]
Clarify GValueArray docs

Don't refer to Quicksort in the documentation of
g_value_array_sort, but just to qsort().

11 years agoUpdated Telugu Translations
Krishnababu Krothapalli [Thu, 9 May 2013 12:33:09 +0000 (18:03 +0530)]
Updated Telugu Translations

11 years agoFix minor typos in GSocketConnectable example code
Zeeshan Ali (Khattak) [Tue, 20 Mar 2012 20:40:32 +0000 (22:40 +0200)]
Fix minor typos in GSocketConnectable example code

11 years agoDeclare res_init() for builds against Android's Bionic
Sebastian Dröge [Mon, 6 May 2013 14:06:48 +0000 (16:06 +0200)]
Declare res_init() for builds against Android's Bionic

res_init() exists in the C library here and as such is detected
by configure, but it is not declared in any header.

11 years agoGProxyAddressEnumerator: add default-port property
Dan Winship [Sun, 5 May 2013 20:50:43 +0000 (16:50 -0400)]
GProxyAddressEnumerator: add default-port property

Although none of the in-tree GSocketConnectable types need it, other
types (like SoupAddress) may find it useful to be able to pass a URI
and a default-port to GProxyAddressEnumerator separately (the same way
you can with GNetworkAddress). So add a default-port property.

https://bugzilla.gnome.org/show_bug.cgi?id=698877

11 years agoDocument all gtester-report options
Matthias Clasen [Sun, 5 May 2013 01:10:43 +0000 (21:10 -0400)]
Document all gtester-report options

The --subunit option was missing from the man page.

11 years agoDocument all glib-mkenums placeholders
Matthias Clasen [Sun, 5 May 2013 01:08:18 +0000 (21:08 -0400)]
Document all glib-mkenums placeholders

The @ENUMPREFIX@ placeholder was missing in the man page.

11 years agoDocument all glib-genmarshal options
Matthias Clasen [Sun, 5 May 2013 01:05:24 +0000 (21:05 -0400)]
Document all glib-genmarshal options

The --stdinc option was missing from the man page.

11 years agoFix a typo
Matthias Clasen [Sun, 5 May 2013 01:03:00 +0000 (21:03 -0400)]
Fix a typo

11 years agoDocument all gdbus-codegen options
Matthias Clasen [Sun, 5 May 2013 01:01:35 +0000 (21:01 -0400)]
Document all gdbus-codegen options

The --help and --xml-files options were missing from the man page.

11 years agogbitlock: fix this to not unconditionally use futex emulation
Dan Winship [Thu, 2 May 2013 14:14:57 +0000 (10:14 -0400)]
gbitlock: fix this to not unconditionally use futex emulation

Ryan accidentally committed some debugging code a long time ago,
causing this file to always use futex emulation even when real futex
support was available. I noticed this a while later and pointed it out
to him, and assumed he was going to fix it, but I guess he assumed I
was going to fix it, and then neither of us did...

https://bugzilla.gnome.org/show_bug.cgi?id=699500

11 years agoSilently handle icon being NULL
Matthias Clasen [Fri, 3 May 2013 01:00:57 +0000 (21:00 -0400)]
Silently handle icon being NULL

While an emblemed icon without a base icon is not very
useful, no need to crash here.
This was crashing the object finalization test in gtk.

11 years agobinding: Add an explicit unbind()
Emmanuele Bassi [Sun, 14 Apr 2013 20:04:41 +0000 (21:04 +0100)]
binding: Add an explicit unbind()

Higher order languages with garbage collection can have issues releasing
a binding, as they do not control the last reference being dropped on
the binding, source, or target instances.

https://bugzilla.gnome.org/show_bug.cgi?id=698018

11 years agotests/mappedfile: Also handle ENOMEM
Colin Walters [Thu, 2 May 2013 15:39:11 +0000 (11:39 -0400)]
tests/mappedfile: Also handle ENOMEM

The RHEL6.4 kernel gives me that instead of EINVAL.

https://bugzilla.gnome.org/show_bug.cgi?id=699485

11 years agogobject: rename an unused parameter to make AIX happy
Dan Winship [Thu, 2 May 2013 17:58:25 +0000 (13:58 -0400)]
gobject: rename an unused parameter to make AIX happy

11 years agoGSocks5Proxy: don't crash if parsing negotiation reply fails
Simon McVittie [Thu, 2 May 2013 15:50:01 +0000 (16:50 +0100)]
GSocks5Proxy: don't crash if parsing negotiation reply fails

The GError should be initialized to NULL, otherwise we'll
"pile up" errors, then try to free an uninitialized pointer.

Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Bug: https://bugzilla.gnome.org/show_bug.cgi?id=699493

11 years agohindi update
Rajesh Ranjan [Thu, 2 May 2013 08:15:46 +0000 (13:45 +0530)]
hindi update

11 years agoUpdated Slovenian translation
Matej Urbančič [Wed, 1 May 2013 18:32:44 +0000 (20:32 +0200)]
Updated Slovenian translation

11 years agoBump version to 2.37.1
Matthias Clasen [Wed, 1 May 2013 09:20:02 +0000 (05:20 -0400)]
Bump version to 2.37.1

11 years agoAdd 2.38 indexes to the docs 2.37.0
Matthias Clasen [Wed, 1 May 2013 04:16:25 +0000 (00:16 -0400)]
Add 2.38 indexes to the docs

11 years agoUpdates
Matthias Clasen [Wed, 1 May 2013 04:15:06 +0000 (00:15 -0400)]
Updates

11 years agogio: fix small leak
Tim Lunn [Wed, 1 May 2013 00:39:26 +0000 (10:39 +1000)]
gio: fix small leak

https://bugzilla.gnome.org/show_bug.cgi?id=699361

11 years agoGApplication: don't leak 'hint' on remote Open
Ryan Lortie [Sun, 21 Apr 2013 14:55:40 +0000 (10:55 -0400)]
GApplication: don't leak 'hint' on remote Open

We were using format string "s" to deconstruct the open hint into a
'const gchar *' which, of course, we never freed.  Fix that.

11 years agobytesicon: don't use g_object_unref() on GBytes
Cosimo Cecchi [Fri, 26 Apr 2013 21:11:02 +0000 (17:11 -0400)]
bytesicon: don't use g_object_unref() on GBytes

We need to use g_bytes_unref()

https://bugzilla.gnome.org/show_bug.cgi?id=699001

11 years agobytesicon: fix a memory leak
Cosimo Cecchi [Fri, 26 Apr 2013 21:07:41 +0000 (17:07 -0400)]
bytesicon: fix a memory leak

https://bugzilla.gnome.org/show_bug.cgi?id=698999

11 years agoconfigure: Assume C90 compatible malloc() prototype
Colin Walters [Fri, 26 Apr 2013 12:12:01 +0000 (08:12 -0400)]
configure: Assume C90 compatible malloc() prototype

This ancient code was attempting to cope with (unknown) systems whose
malloc() prototype was incompatible with the standard.  This test was
fragile; it would break if the build environment provided -Wall in
CFLAGS.

Now that it's 2013, let's assume that target systems have a sane
malloc().  If someone complains, we can revisit this.

https://bugzilla.gnome.org/698716

11 years agog_object_new: check for NULL from _constructor()
Ryan Lortie [Fri, 26 Apr 2013 15:27:51 +0000 (11:27 -0400)]
g_object_new: check for NULL from _constructor()

There is some code in the wild (like in gnome-session) that does this
from its custom _constructor() implementation:

{
  GObject *obj;

  obj = ((chain up));

  if (!object_is_viable (obj))
    {
      g_object_unref (obj);
      return NULL;
    }
  else
    return obj;
}

This has never been a valid use of GObject and this code has always
caused memory to be leaked[1] by growing the construction_objects list.
The ability to legitimately return NULL from a constructor was exactly
the reason that we created GInitable, in fact.

That doesn't change the fact that the g_object_new() rewrite will crash
in this case, so instead of doing that, let's emit a critical and avoid
the crash.  This will allow people to upgrade their GLib without also
upgrading their gnome-session.  Meanwhile, people can fix their broken
code.

[1] not in the strictest sense of the word, because it's still reachable

11 years agoTamil Translation Updated
Shantha kumar [Fri, 26 Apr 2013 06:10:25 +0000 (11:40 +0530)]
Tamil Translation Updated