platform/upstream/gstreamer.git
13 years agodocs: Add GstDateTime section and update it for latest symbols
Edward Hervey [Thu, 2 Dec 2010 12:03:57 +0000 (13:03 +0100)]
docs: Add GstDateTime section and update it for latest symbols

13 years agoqueue2: Only call update_buffering if needed.
Edward Hervey [Tue, 19 Oct 2010 16:09:53 +0000 (18:09 +0200)]
queue2: Only call update_buffering if needed.

update_buffering is so big it will never be inlined (and shouldn't),
we therefore move the check outside of it.

13 years agoqueue2: Avoid re-checking many times whether an item is a buffer
Edward Hervey [Tue, 19 Oct 2010 15:45:16 +0000 (17:45 +0200)]
queue2: Avoid re-checking many times whether an item is a buffer

Avoids calling 6 times gst_buffer_get_type() for every item coming
through queue2

13 years agoqueue2: Reduce amount of time locks are taken
Edward Hervey [Tue, 19 Oct 2010 15:43:56 +0000 (17:43 +0200)]
queue2: Reduce amount of time locks are taken

13 years agoqueue2: Fixup documentation of some properties
Edward Hervey [Tue, 19 Oct 2010 15:42:39 +0000 (17:42 +0200)]
queue2: Fixup documentation of some properties

13 years agoqueue2: Avoid useless segment_to_running_time() calculations.
Edward Hervey [Tue, 19 Oct 2010 15:40:13 +0000 (17:40 +0200)]
queue2: Avoid useless segment_to_running_time() calculations.

* Cache src and sink time
* Use a boolean to known whether src/sink time needs to be recalculated

Avoids 50% of calls to gst_segment_to_running_time()

13 years agobasesink: Re-using GstClockID instead of constantly recreating one
Edward Hervey [Wed, 20 Oct 2010 15:41:28 +0000 (17:41 +0200)]
basesink: Re-using GstClockID instead of constantly recreating one

Makes _sink_wait_clock at least 2 times faster.

https://bugzilla.gnome.org/show_bug.cgi?id=632778

13 years agogstclock: New API to re-use a single shot GstClockID
Edward Hervey [Wed, 20 Oct 2010 15:40:43 +0000 (17:40 +0200)]
gstclock: New API to re-use a single shot GstClockID

API: gst_clock_single_shot_id_reinit

https://bugzilla.gnome.org/show_bug.cgi?id=632778

13 years agobasesink: Pass along miniobject type through various functions
Edward Hervey [Wed, 20 Oct 2010 11:52:02 +0000 (13:52 +0200)]
basesink: Pass along miniobject type through various functions

Avoids doing useless GST_IS_*

https://bugzilla.gnome.org/show_bug.cgi?id=632778

13 years agobasesink: Switch enable_last_buffer to an atomic int
Edward Hervey [Wed, 20 Oct 2010 11:08:08 +0000 (13:08 +0200)]
basesink: Switch enable_last_buffer to an atomic int

Avoids having to take a lock to read/write it.

https://bugzilla.gnome.org/show_bug.cgi?id=632778

13 years agoqueue: Remove useless checks from e406f7
Edward Hervey [Tue, 19 Oct 2010 13:53:26 +0000 (15:53 +0200)]
queue: Remove useless checks from e406f7

srcresult was being rechecked in places it couldn't have changed.
queue level was being rechecked in places it couldn't have changed.

https://bugzilla.gnome.org/show_bug.cgi?id=632780

13 years agogstdataqueue: Only emit g_cond_signal when needed
Edward Hervey [Wed, 13 Oct 2010 11:50:22 +0000 (13:50 +0200)]
gstdataqueue: Only emit g_cond_signal when needed

Keep track of which cond we're waiting for and only emit when needed.

https://bugzilla.gnome.org/show_bug.cgi?id=632779

13 years agosegment: move g_return_if_fail to where it is needed
Wim Taymans [Wed, 20 Oct 2010 15:12:29 +0000 (17:12 +0200)]
segment: move g_return_if_fail to where it is needed

13 years agoversion: Take nano version into account in GST_CHECK_VERSION()
Sebastian Dröge [Wed, 3 Nov 2010 10:14:03 +0000 (11:14 +0100)]
version: Take nano version into account in GST_CHECK_VERSION()

If the nano is > 0 the current version should be handled the same as
micro + 1.

13 years agopad: Set the event source object if none is set yet in gst_pad_push_event()
Sebastian Dröge [Mon, 1 Nov 2010 15:34:46 +0000 (16:34 +0100)]
pad: Set the event source object if none is set yet in gst_pad_push_event()

Otherwise the source will stay at NULL, the event is passed to the
peerpad via gst_pad_send_event() and then the peerpad is set as
source of the event instead of the originating pad.

13 years agoqueue2: Remove dead assignment and unused variable
Sebastian Dröge [Sun, 31 Oct 2010 17:48:19 +0000 (18:48 +0100)]
queue2: Remove dead assignment and unused variable

13 years agoqueue2: Remove dead assignment and move variable declaration into inner block
Sebastian Dröge [Sun, 31 Oct 2010 17:46:43 +0000 (18:46 +0100)]
queue2: Remove dead assignment and move variable declaration into inner block

13 years agoqueue2: Remove redundant variable
Sebastian Dröge [Sun, 31 Oct 2010 17:23:00 +0000 (18:23 +0100)]
queue2: Remove redundant variable

Other than saving an immense amount of 4 bytes of memory this
prevents clang from complaining and keeps the ring buffer state
in a single variable instead of two.

13 years agogsttask: Set thread names on Windows with MSVC if a debugger is attached
David Hoyt [Wed, 20 Oct 2010 08:18:18 +0000 (10:18 +0200)]
gsttask: Set thread names on Windows with MSVC if a debugger is attached

Fixes bug #632168.

13 years agomacros: Define restrict keyword if not available
Sebastian Dröge [Tue, 19 Oct 2010 13:52:30 +0000 (15:52 +0200)]
macros: Define restrict keyword if not available

This change always defines the restrict keyword if a
non-C99 C compiler is used. In the case of GCC >= 4
it will be defined to __restrict__, in all other
cases to nothing. This allows to use the restrict
keyword unconditionally.

13 years agoBump GLib requirement to >= 2.22
Tim-Philipp Müller [Wed, 1 Dec 2010 23:57:36 +0000 (23:57 +0000)]
Bump GLib requirement to >= 2.22

See http://gstreamer.freedesktop.org/wiki/ReleasePlanning/GLibRequirement

13 years agoBack to development
Tim-Philipp Müller [Wed, 1 Dec 2010 23:56:45 +0000 (23:56 +0000)]
Back to development

13 years agoRelease 0.10.31 RELEASE-0.10.31
Tim-Philipp Müller [Tue, 30 Nov 2010 17:40:47 +0000 (17:40 +0000)]
Release 0.10.31

13 years agoutils: avoid 'unused argument' warnings caused by GST_BOILERPLATE_FULL
Tim-Philipp Müller [Sat, 27 Nov 2010 19:13:35 +0000 (19:13 +0000)]
utils: avoid 'unused argument' warnings caused by GST_BOILERPLATE_FULL

The unused data parameter in the class_init trampoline function
seems to cause warnings with some c++ compilers.

https://bugzilla.gnome.org/show_bug.cgi?id=635869

13 years agodocs: some small fixes to the plugin writer's guide
reynaldo [Wed, 10 Nov 2010 02:27:17 +0000 (23:27 -0300)]
docs: some small fixes to the plugin writer's guide

Fix wrongly placed example and weirdly phrased 'note' lacking proper
formatting. Fix missing hint for autogen.sh location and rephrase
'built and installed' sentence. Fix wrongly phrased and redundant
paragraph in PWG

https://bugzilla.gnome.org/show_bug.cgi?id=634921

13 years agomanual: Improve states documentation
Thiago Santos [Sat, 27 Nov 2010 14:02:48 +0000 (11:02 -0300)]
manual: Improve states documentation

Be more explicit about being on NULL before unrefs

13 years agointrospection: Include exported packages information in GIRs
Evan Nemerson [Sat, 20 Nov 2010 22:54:23 +0000 (14:54 -0800)]
introspection: Include exported packages information in GIRs

https://bugzilla.gnome.org/show_bug.cgi?id=635389

13 years ago0.10.30.5 pre-release
Tim-Philipp Müller [Thu, 18 Nov 2010 00:29:41 +0000 (00:29 +0000)]
0.10.30.5 pre-release

13 years agoelementfactory: fix caps leak in element factory list utility functions
Tim-Philipp Müller [Thu, 18 Nov 2010 00:29:19 +0000 (00:29 +0000)]
elementfactory: fix caps leak in element factory list utility functions

13 years agopo: update translations
Tim-Philipp Müller [Wed, 17 Nov 2010 23:55:45 +0000 (23:55 +0000)]
po: update translations

13 years agodatetime: Add _from_unix_epoch variants
Thiago Santos [Wed, 17 Nov 2010 12:39:34 +0000 (09:39 -0300)]
datetime: Add _from_unix_epoch variants

Adds 2 variants for the gst_date_time_from_unix_epoch function,
one for UTC and another for local time.

API: gst_date_time_new_from_unix_epoch_utc
API: gst_date_time_new_from_unix_epoch_local_time

Fixes #653031

https://bugzilla.gnome.org/show_bug.cgi?id=635031

13 years agomath-compat: don't re-define _USE_MATH_DEFINES if already defined
Vladimir Eremeev [Wed, 3 Nov 2010 14:21:02 +0000 (14:21 +0000)]
math-compat: don't re-define _USE_MATH_DEFINES if already defined

This avoids compiler warnings.

https://bugzilla.gnome.org/show_bug.cgi?id=633886

13 years ago0.10.30.4 pre-release
Tim-Philipp Müller [Mon, 1 Nov 2010 16:06:43 +0000 (16:06 +0000)]
0.10.30.4 pre-release

13 years agopo: update translations
Tim-Philipp Müller [Mon, 1 Nov 2010 15:36:54 +0000 (15:36 +0000)]
po: update translations

13 years agoqueue2: don't leak pad ref in pull mode when flushing
Tim-Philipp Müller [Sun, 31 Oct 2010 20:17:05 +0000 (20:17 +0000)]
queue2: don't leak pad ref in pull mode when flushing

Fix pad leak when queue2 is flushing or being shut down.

13 years agoqueue2: don't send seeks beyond the end of the file upstream in pull mode
Tim-Philipp Müller [Sun, 31 Oct 2010 19:47:25 +0000 (19:47 +0000)]
queue2: don't send seeks beyond the end of the file upstream in pull mode

If downstream is operating in pull mode, short-circuit any pulls beyond
the end of the file and return FLOW_UNEXPECTED immediately instead of
sending a seek beyond the end of the file upstream, since this might
confuse upstream elements (and/or http servers, for example). Fixes
playback of apple trailers in totem and youtube/html5 clips in
WebkitGTK+.

https://bugzilla.gnome.org/show_bug.cgi?id=632977

13 years agobasetransform: use input position for queries if we have no output position
Jonathan Matthew [Thu, 28 Oct 2010 13:28:15 +0000 (23:28 +1000)]
basetransform: use input position for queries if we have no output position

13 years agobasetransform: fix reverse negotiation
Wim Taymans [Thu, 28 Oct 2010 12:29:31 +0000 (13:29 +0100)]
basetransform: fix reverse negotiation

When the downstream element suggests a new format, pass the suggestion
upstream if we can't convert to it.

Fixes #633147

13 years agomultiqueue: Fix tracking of unlinked streams.
Jan Schmidt [Wed, 27 Oct 2010 16:12:36 +0000 (18:12 +0200)]
multiqueue: Fix tracking of unlinked streams.

33082eb9e42c52e4df848195946f1b7bbce768c5 introduced a bug
preventing sparse unlinked streams from advancing properly,
leading to the queue blocking.

Fixes: #633176

13 years agotests: Add a multiqueue sparse streams test
Jan Schmidt [Wed, 27 Oct 2010 16:11:35 +0000 (18:11 +0200)]
tests: Add a multiqueue sparse streams test

13 years agoAutomatic update of common submodule
Jan Schmidt [Wed, 27 Oct 2010 12:16:11 +0000 (13:16 +0100)]
Automatic update of common submodule

From 7bbd708 to 011bcc8

13 years ago0.10.30.3 pre-release
Tim-Philipp Müller [Fri, 22 Oct 2010 16:35:23 +0000 (17:35 +0100)]
0.10.30.3 pre-release

13 years agobytereader: const-ify byte reader argument in peek/scan API
Tim-Philipp Müller [Mon, 18 Oct 2010 16:42:27 +0000 (17:42 +0100)]
bytereader: const-ify byte reader argument in peek/scan API

Because we can.

13 years agoelementfactory: Fix 64bit constant
Edward Hervey [Fri, 22 Oct 2010 09:52:47 +0000 (11:52 +0200)]
elementfactory: Fix 64bit constant

Basically we're not meant to put anything more complex than simple numbers,
due to the definition of G_GUINT64_CONSTANT:
G_GUINT64_CONSTANT(val) (val##UL)

Which previously resulted in .... 1 << 49UL

13 years agobasesink: recompute correct running time for buffer ending flushing step
Mark Nauwelaerts [Mon, 18 Oct 2010 08:46:59 +0000 (10:46 +0200)]
basesink: recompute correct running time for buffer ending flushing step

Prevents delaying/hanging when resuming PLAYING.

Fixes #632433.

13 years agodocs: fix misc. gtk-doc warnings in libs
Tim-Philipp Müller [Sat, 16 Oct 2010 18:19:47 +0000 (19:19 +0100)]
docs: fix misc. gtk-doc warnings in libs

(for gtk-doc 1.15)

13 years agoRevert "libs/gst/dataqueue: Document gst_data_queue_new_full"
Tim-Philipp Müller [Sat, 16 Oct 2010 17:26:20 +0000 (18:26 +0100)]
Revert "libs/gst/dataqueue: Document gst_data_queue_new_full"

This reverts commit 80727c117703507f790a86b0962ab3d915e5a491.

This doesn't make sense. gst_data_queue_new_full() is already
documented above. And we need the doc blurb for _new() here.

13 years agodocs: flesh out release instructions a bit more
Tim-Philipp Müller [Sat, 16 Oct 2010 16:00:17 +0000 (17:00 +0100)]
docs: flesh out release instructions a bit more

13 years agodocs: add some gtk-doc Since: markers
Tim-Philipp Müller [Sat, 16 Oct 2010 15:53:49 +0000 (16:53 +0100)]
docs: add some gtk-doc Since: markers

Add some gtk-doc Since: markers, fix one Since: marker,
fix typo.

13 years ago0.10.30.2 pre-release
Tim-Philipp Müller [Fri, 15 Oct 2010 23:25:52 +0000 (00:25 +0100)]
0.10.30.2 pre-release

13 years agopo: update translations
Tim-Philipp Müller [Fri, 15 Oct 2010 23:14:13 +0000 (00:14 +0100)]
po: update translations

13 years agogst-inspect: useful factory names for uri-handlers output
Mark Nauwelaerts [Fri, 15 Oct 2010 17:45:14 +0000 (19:45 +0200)]
gst-inspect: useful factory names for uri-handlers output

Fixes #632236.

13 years agoAutomatic update of common submodule
David Schleef [Thu, 14 Oct 2010 19:31:32 +0000 (12:31 -0700)]
Automatic update of common submodule

From 5a668bf to 7bbd708

13 years agodatetime: Use seconds as double
Thiago Santos [Tue, 12 Oct 2010 18:13:48 +0000 (15:13 -0300)]
datetime: Use seconds as double

Use seconds as double to make API similar to glib's
gdatetime. Also move timezone parameter to the
first position, just like glib's.

https://bugzilla.gnome.org/show_bug.cgi?id=628408

13 years agogstdatetime: Move doc outside the ifdefs
Thiago Santos [Mon, 11 Oct 2010 19:15:29 +0000 (16:15 -0300)]
gstdatetime: Move doc outside the ifdefs

Move the datetime documentation of the functions outside the
ifdefs

https://bugzilla.gnome.org/show_bug.cgi?id=628408

13 years agodatetime: Use GDateTime if available
Thiago Santos [Mon, 27 Sep 2010 22:35:08 +0000 (19:35 -0300)]
datetime: Use GDateTime if available

Use GDateTime internally on GstDateTime if glib already
provides it.

https://bugzilla.gnome.org/show_bug.cgi?id=628408

13 years agoglib-private: Add include protection macro
Thiago Santos [Tue, 28 Sep 2010 20:46:29 +0000 (17:46 -0300)]
glib-private: Add include protection macro

13 years agobuffer: add guard to buffer_set_caps() that checks if caps are simple
Tim-Philipp Müller [Wed, 13 Oct 2010 11:51:00 +0000 (12:51 +0100)]
buffer: add guard to buffer_set_caps() that checks if caps are simple

13 years agosystemclock: add a missing G_PARAM_STATIC_STRINGS
Stefan Kost [Wed, 13 Oct 2010 12:56:26 +0000 (15:56 +0300)]
systemclock: add a missing G_PARAM_STATIC_STRINGS

13 years agolibs/gst/dataqueue: Document gst_data_queue_new_full
Edward Hervey [Thu, 1 Oct 2009 09:14:06 +0000 (11:14 +0200)]
libs/gst/dataqueue: Document gst_data_queue_new_full

13 years agobase/gstdataqueue: inline some functions, get levels with memcpy.
Edward Hervey [Mon, 28 Sep 2009 11:35:35 +0000 (13:35 +0200)]
base/gstdataqueue: inline some functions, get levels with memcpy.

13 years agoqueue2: Remove unused argument from find_range()
Edward Hervey [Wed, 13 Oct 2010 09:54:04 +0000 (11:54 +0200)]
queue2: Remove unused argument from find_range()

13 years agofdsink: cleanup get_property/set_property
Edward Hervey [Wed, 13 Oct 2010 09:52:25 +0000 (11:52 +0200)]
fdsink: cleanup get_property/set_property

13 years agoqueue: avoid g_cond_signal when we can
Wim Taymans [Tue, 12 Oct 2010 16:48:10 +0000 (18:48 +0200)]
queue: avoid g_cond_signal when we can

Keep track of what cond we are waiting on and only signal when some other thread
is waiting.

13 years agominiobject: avoid duplicate type check when freeing miniobject
Tim-Philipp Müller [Mon, 11 Oct 2010 18:27:54 +0000 (19:27 +0100)]
miniobject: avoid duplicate type check when freeing miniobject

gst_mini_object_unref() has guards that check the type already, so
we don't really need to re-check it here again while getting the
class (there's not really much point to that anyway, since we don't
check the return value of the get_class, so we'd crash anyway if
we're not dealing with a mini object, the only question would
be if there'd be a warning before the crash or not).

13 years agominiobject: Directly increate mini_object in mini_object_free()
Edward Hervey [Mon, 11 Oct 2010 16:55:14 +0000 (18:55 +0200)]
miniobject: Directly increate mini_object in mini_object_free()

Speeds up mini_object_unref by 25% by avoiding the typecheck which
is avoidable here since it is only called on existing miniobjects.

13 years agominiobject: Remove confusing DEBUG_REFCOUNT define
Edward Hervey [Mon, 11 Oct 2010 16:30:54 +0000 (18:30 +0200)]
miniobject: Remove confusing DEBUG_REFCOUNT define

the debugging statements will be silenced automatically if debugging
is disabled, and the type check is actually required.

13 years agoqueue2: release queue2 lock before notify
Wim Taymans [Mon, 11 Oct 2010 16:10:07 +0000 (18:10 +0200)]
queue2: release queue2 lock before notify

Make sure that we don't hold the lock when we notify the temp-location
property,

Fixes #631853

13 years agoqueue: add debug statement
Wim Taymans [Mon, 11 Oct 2010 14:45:16 +0000 (16:45 +0200)]
queue: add debug statement

13 years agoqueue: apply sink segment on the source if queue is empty
Ognyan Tonchev [Mon, 11 Oct 2010 08:27:52 +0000 (10:27 +0200)]
queue: apply sink segment on the source if queue is empty

Apply the sink segment on the source immediatly when it is received
and there is nothing in the queue.

Solves #482147

13 years agobin: fix documentation for iterate_sources
Wim Taymans [Mon, 11 Oct 2010 13:51:21 +0000 (15:51 +0200)]
bin: fix documentation for iterate_sources

13 years agodocs: use the gtk-doc shortcuts to get coloured and xrefed example
Stefan Kost [Mon, 11 Oct 2010 13:41:26 +0000 (16:41 +0300)]
docs: use the gtk-doc shortcuts to get coloured and xrefed example

13 years agobin: Initialize variable
Sebastian Dröge [Mon, 11 Oct 2010 12:20:15 +0000 (14:20 +0200)]
bin: Initialize variable

13 years agobin: Improve tracking of source elements
Wim Taymans [Mon, 11 Oct 2010 08:56:56 +0000 (10:56 +0200)]
bin: Improve tracking of source elements

Track elements tagged with the IS_SOURCE flag in a similar way we track the sink
elements. This allows us to efficiently dispatch downstream events to the right
elements.

13 years agobasesrc: tag as a SOURCE element
Wim Taymans [Mon, 11 Oct 2010 08:55:04 +0000 (10:55 +0200)]
basesrc: tag as a SOURCE element

Tag all elements deriving from the basesrc with the IS_SOURCE flag.

13 years agoelement: add IS_SOURCE flag
Wim Taymans [Mon, 11 Oct 2010 08:53:41 +0000 (10:53 +0200)]
element: add IS_SOURCE flag

Add the GST_ELEMENT_IS_SOURCE flag so that we can tag source elements like we
can with sink elements.

13 years agoregistry: g_mapped_file_unref exists already since GLib 2.21.3
Vincent Penquerc'h [Sat, 9 Oct 2010 13:18:44 +0000 (14:18 +0100)]
registry: g_mapped_file_unref exists already since GLib 2.21.3

13 years agobasetransform: Report the output position on POSITION queries on the srcpad
Sebastian Dröge [Sun, 10 Oct 2010 16:14:40 +0000 (18:14 +0200)]
basetransform: Report the output position on POSITION queries on the srcpad

There can be a difference between input and output last_stop.

Fixes bug #629410.

13 years agoAutomatic update of common submodule
David Schleef [Fri, 8 Oct 2010 19:43:20 +0000 (12:43 -0700)]
Automatic update of common submodule

From c4a8adc to 5a668bf

13 years agomultiqueue: Remove unused variable and simplify code
Edward Hervey [Fri, 8 Oct 2010 10:54:52 +0000 (12:54 +0200)]
multiqueue: Remove unused variable and simplify code

oldid was only used when we were doing multiple pops per loop.

13 years agoAutomatic update of common submodule
Sebastian Dröge [Fri, 8 Oct 2010 10:50:10 +0000 (12:50 +0200)]
Automatic update of common submodule

From 5e3c9bf to c4a8adc

13 years agobytewriter: Add missing file
Sebastian Dröge [Fri, 8 Oct 2010 10:48:42 +0000 (12:48 +0200)]
bytewriter: Add missing file

13 years agobytewriter: Add unchecked variants of the writing functions
Sebastian Dröge [Fri, 8 Oct 2010 10:18:23 +0000 (12:18 +0200)]
bytewriter: Add unchecked variants of the writing functions

These don't check if there's enough free space available and are
available as inline functions only.

API: gst_byte_writer_put_int8_unchecked
API: gst_byte_writer_put_int16_be_unchecked
API: gst_byte_writer_put_int16_le_unchecked
API: gst_byte_writer_put_int24_be_unchecked
API: gst_byte_writer_put_int24_le_unchecked
API: gst_byte_writer_put_int32_be_unchecked
API: gst_byte_writer_put_int32_le_unchecked
API: gst_byte_writer_put_int64_be_unchecked
API: gst_byte_writer_put_int64_le_unchecked
API: gst_byte_writer_put_uint8_unchecked
API: gst_byte_writer_put_uint16_be_unchecked
API: gst_byte_writer_put_uint16_le_unchecked
API: gst_byte_writer_put_uint24_be_unchecked
API: gst_byte_writer_put_uint24_le_unchecked
API: gst_byte_writer_put_uint32_be_unchecked
API: gst_byte_writer_put_uint32_le_unchecked
API: gst_byte_writer_put_uint64_be_unchecked
API: gst_byte_writer_put_uint64_le_unchecked
API: gst_byte_writer_put_float32_be_unchecked
API: gst_byte_writer_put_float32_le_unchecked
API: gst_byte_writer_put_float64_be_unchecked
API: gst_byte_writer_put_float64_le_unchecked
API: gst_byte_writer_put_data_unchecked
API: gst_byte_writer_fill_unchecked

13 years agocontroller, dataprotocol: make public enum _get_type() functions thread-safe
Tim-Philipp Müller [Fri, 8 Oct 2010 08:34:47 +0000 (09:34 +0100)]
controller, dataprotocol: make public enum _get_type() functions thread-safe

Not that it is likely to matter in practice, but since these are public
API they should probably be thread-safe.

13 years agodataprotocol, lfocontrolsource: fix enum value name in enums that are public API
Tim-Philipp Müller [Thu, 7 Oct 2010 23:38:39 +0000 (00:38 +0100)]
dataprotocol, lfocontrolsource: fix enum value name in enums that are public API

So run-time bindings can introspect the names correctly (we abuse this
field as description field only in elements, not for public API
(where the description belongs into the gtk-doc chunk).

https://bugzilla.gnome.org/show_bug.cgi?id=629946

13 years agobytewriter: Fix possible infinite loop caused by an overflow
Sebastian Dröge [Fri, 8 Oct 2010 07:47:12 +0000 (09:47 +0200)]
bytewriter: Fix possible infinite loop caused by an overflow

13 years agoelements: minor performance improvement when doing g_object_notify() for the "last...
Tim-Philipp Müller [Thu, 7 Oct 2010 17:46:26 +0000 (18:46 +0100)]
elements: minor performance improvement when doing g_object_notify() for the "last-message" property

Make sure property names passed to g_object_notify() are in the canonical form
(ie. "last-message" not "last_message"), so that g_param_spec_pool_lookup()
doesn't have to do strdup/canonicalize/free for every single notify call.
This only applies when building against older GLib versions (< 2.26).

13 years agofakesink: use g_object_notify_by_pspec() and remove work-around for old GLib versions...
Tim-Philipp Müller [Thu, 7 Oct 2010 17:27:06 +0000 (18:27 +0100)]
fakesink: use g_object_notify_by_pspec() and remove work-around for old GLib versions if possible

Use more efficient g_object_notify_by_pspec() if we're compiling against
GLib >= 2.26, and also remove work-around for g_object_notify() thread-
safety issues with older GLib versions if it's not needed any more.

13 years agotee: use g_object_notify_by_pspec() if possible
Tim-Philipp Müller [Thu, 7 Oct 2010 17:21:13 +0000 (18:21 +0100)]
tee: use g_object_notify_by_pspec() if possible

Use more efficient g_object_notify_by_pspec() if we're compiling against
GLib >= 2.26.

13 years agofakesrc: use g_object_notify_by_pspec() if possible
Tim-Philipp Müller [Thu, 7 Oct 2010 17:19:31 +0000 (18:19 +0100)]
fakesrc: use g_object_notify_by_pspec() if possible

Use more efficient g_object_notify_by_pspec() if we're compiling against
GLib >= 2.26.

13 years agoidentity: use g_object_notify_by_pspec() and remove work-around for old GLib versions...
Tim-Philipp Müller [Thu, 7 Oct 2010 16:53:18 +0000 (17:53 +0100)]
identity: use g_object_notify_by_pspec() and remove work-around for old GLib versions if possible

Use more efficient g_object_notify_by_pspec() if we're compiling against
GLib >= 2.26, and also remove work-around for g_object_notify() thread-
safety issues with older GLib versions if it's not needed any more.

13 years agopads: use new g_object_notify_by_pspec() for caps notifies if available
Tim-Philipp Müller [Thu, 7 Oct 2010 16:23:10 +0000 (17:23 +0100)]
pads: use new g_object_notify_by_pspec() for caps notifies if available

If we're building against GLib >= 2.26.0, we can use the more efficient
g_object_notify_by_caps(), which avoids the param spec lookup.

13 years agoclock: remove unnecessary g_object_notify() call
Tim-Philipp Müller [Thu, 7 Oct 2010 15:27:20 +0000 (16:27 +0100)]
clock: remove unnecessary g_object_notify() call

GObject will do that for us when g_object_set*() is called.

13 years agodocs: update qos design doc
Wim Taymans [Thu, 7 Oct 2010 17:18:03 +0000 (19:18 +0200)]
docs: update qos design doc

Fix some typos.
change the definition of the quality field for video decoders to something that
makes more sense.

13 years agoAdd gobject-introspection temp directories to CRUFT_DIRS
Tim-Philipp Müller [Tue, 5 Oct 2010 16:02:58 +0000 (17:02 +0100)]
Add gobject-introspection temp directories to CRUFT_DIRS

13 years agolfocontrolsource: use math-compat.h for M_PI
Tim-Philipp Müller [Tue, 5 Oct 2010 14:05:43 +0000 (15:05 +0100)]
lfocontrolsource: use math-compat.h for M_PI

13 years agogstinfo: remove random MSVC compatibility define for M_PI that doesn't belong here
Tim-Philipp Müller [Tue, 5 Oct 2010 13:45:02 +0000 (14:45 +0100)]
gstinfo: remove random MSVC compatibility define for M_PI that doesn't belong here

Code that needs this should include gst/math-compat.h or use G_PI.

13 years agogst: add math-compat.h header
Tim-Philipp Müller [Tue, 5 Oct 2010 13:16:19 +0000 (14:16 +0100)]
gst: add math-compat.h header

Add minimal math-compath.h header where we can define fallback
versions for miscellaneous math functions that aren't always
available, so we don't have to duplicate this in plugins.
The header is not included by default, so needs to be
included explicitly for now.

https://bugzilla.gnome.org/show_bug.cgi?id=630802

13 years agotools: fix parsing of timestamp in gst-plot-timeline
Thijs Vermeir [Tue, 5 Oct 2010 09:47:59 +0000 (11:47 +0200)]
tools: fix parsing of timestamp in gst-plot-timeline

13 years agobasesink: don't take preroll-lock in get_property
Stefan Kost [Sat, 25 Sep 2010 11:24:46 +0000 (14:24 +0300)]
basesink: don't take preroll-lock in get_property

Use atomic ops to read and write more properties. Taking the preroll lock in get_property
can lock up applications reading the property during preroll.