platform/upstream/gstreamer.git
9 years agocaps: gst_caps_is_any() should return TRUE or FALSE
Tim-Philipp Müller [Sun, 22 Jun 2014 11:50:42 +0000 (12:50 +0100)]
caps: gst_caps_is_any() should return TRUE or FALSE

Not some flag value instead of TRUE. Fixes code like
gst_caps_is_any() == TRUE.

https://bugzilla.gnome.org//show_bug.cgi?id=731704

9 years agodevice: rename "klass" and get_klass() to "device-class" and _get_device_class()
Tim-Philipp Müller [Sun, 1 Jun 2014 15:56:41 +0000 (16:56 +0100)]
device: rename "klass" and get_klass() to "device-class" and _get_device_class()

There's some precedent in GstElementFactory, but a
"klass" property just seems weird.

9 years agobenchmarks: capsnego: add --loops command line option
Tim-Philipp Müller [Fri, 20 Jun 2014 17:34:44 +0000 (18:34 +0100)]
benchmarks: capsnego: add --loops command line option

And default to 50 loops.

9 years agobenchmark: capsnego: use GOptionContext for option parsing
Tim-Philipp Müller [Fri, 20 Jun 2014 16:14:52 +0000 (17:14 +0100)]
benchmark: capsnego: use GOptionContext for option parsing

9 years agotests: fix compiler warnings in gstvalue tests
Tim-Philipp Müller [Thu, 19 Jun 2014 11:10:23 +0000 (12:10 +0100)]
tests: fix compiler warnings in gstvalue tests

Calling GST_VALUE_HOLDS_*(&v) now results in a compiler
warning about value!=NULL always being false, so check
type directly in those cases.

9 years agovalue: simplify GST_VALUE_HOLDS for our boxed and fundamental types
Tim-Philipp Müller [Tue, 17 Jun 2014 21:45:57 +0000 (22:45 +0100)]
value: simplify GST_VALUE_HOLDS for our boxed and fundamental types

Boxed types can't be derived from, and we don't support
deriving from our special fundamental types (the code
checks for GType equality in most places.

9 years agoGstDevice: Document GstDevice and related classes
Olivier Crête [Fri, 20 Jun 2014 20:55:06 +0000 (16:55 -0400)]
GstDevice: Document GstDevice and related classes

9 years agoFix funnel EOS handling and wrong unittest
Srimanta Panda [Mon, 16 Jun 2014 11:47:55 +0000 (13:47 +0200)]
Fix funnel EOS handling and wrong unittest

When no data is coming from sinkpads and eos events
arrived at one of the sinkpad, funnel forwards the EOS
event to downstream. It forwards the EOS because lastsink pad
is NULL. Also the unit testcase of the funnel is not checking
the correct behavior as it should. The unit test case should
fail if one of the sink pad has already EOS present on it and
we are trying to push one more EOS.

https://bugzilla.gnome.org/show_bug.cgi?id=731716

9 years agogstvalue: optimise checks for lists
Tim-Philipp Müller [Thu, 19 Jun 2014 07:09:55 +0000 (08:09 +0100)]
gstvalue: optimise checks for lists

Our fundamental types are non-derivable, so we can
just check for equality. Also avoid doing the same
check multiple times in a couple of places.

9 years agogstvalue: use g_assert() in internal function for already-checked things
Tim-Philipp Müller [Thu, 19 Jun 2014 07:06:31 +0000 (08:06 +0100)]
gstvalue: use g_assert() in internal function for already-checked things

So these get compiled out for releases.

9 years agogstvalue: add internal _can_compare_unchecked()
Tim-Philipp Müller [Thu, 19 Jun 2014 07:05:40 +0000 (08:05 +0100)]
gstvalue: add internal _can_compare_unchecked()

9 years agogstvalue: add internal _list_concat() that takes ownership of input values
Tim-Philipp Müller [Thu, 19 Jun 2014 07:03:37 +0000 (08:03 +0100)]
gstvalue: add internal _list_concat() that takes ownership of input values

Avoids unnecessary copies.

9 years agostructure: simplify value type checks in getters
Tim-Philipp Müller [Wed, 18 Jun 2014 18:06:58 +0000 (19:06 +0100)]
structure: simplify value type checks in getters

Just check for GType equality in common cases.

9 years agovalue: Add a FIXME 2.0 for a fraction ranges optimization
Sebastian Dröge [Thu, 19 Jun 2014 07:29:18 +0000 (09:29 +0200)]
value: Add a FIXME 2.0 for a fraction ranges optimization

Currently we leak the internal representation of them as two GValues that
contain a fraction. Without this we could store fraction ranges as
  data[0] = (min_n << 32) | (min_d)
  data[1] = (max_n << 32) | (max_d)
and wouldn't require an additional allocation per range.

9 years agovalue: Make sure to cast int range values to guints before storing them
Sebastian Dröge [Thu, 19 Jun 2014 07:23:56 +0000 (09:23 +0200)]
value: Make sure to cast int range values to guints before storing them

Otherwise negative values will sets all of the 64 bits due to two's
complement's definition of negative values.

Also add a test for negative int ranges.

9 years agowin32: update exports
Tim-Philipp Müller [Thu, 19 Jun 2014 06:57:11 +0000 (07:57 +0100)]
win32: update exports

9 years agovalue: Store integer ranges directly in a GValue without additional allocation
Sebastian Dröge [Thu, 19 Jun 2014 07:05:18 +0000 (09:05 +0200)]
value: Store integer ranges directly in a GValue without additional allocation

Micro optimization to save some allocations. Next step to do this
with fraction ranges too.

9 years agogst_private: Fix duplicate definition
Edward Hervey [Thu, 19 Jun 2014 06:43:02 +0000 (08:43 +0200)]
gst_private: Fix duplicate definition

9 years agogst: Store more basic type GTypes in variables
Sebastian Dröge [Thu, 19 Jun 2014 06:05:03 +0000 (08:05 +0200)]
gst: Store more basic type GTypes in variables

Micro optimization to change a function call to a variable access
for all our basic types.

9 years agovalue: Store our fundamental type GTypes in variables
Sebastian Dröge [Thu, 19 Jun 2014 06:04:01 +0000 (08:04 +0200)]
value: Store our fundamental type GTypes in variables

Micro optimization to change a function call to a variable access
for all our basic types.

9 years agogstvalue: Speed up gst_value_intersect/_subtract
Edward Hervey [Tue, 17 Jun 2014 05:31:48 +0000 (07:31 +0200)]
gstvalue: Speed up gst_value_intersect/_subtract

Both gst_value_intersect and gst_value_subtract will call
gst_value_compare if one of their arguments isn't a list.

gst_value_compare will then re-do a check to see if one of
the arguments is a list (for the special case of comparing a unitary
value with a list of length 1).

The problem is that the various G_VALUE_HOLDS represent an expensive
amount of calling gst_value_compare (almost half of it) to see if
the provided arguments are list. These checks can be done without
when we know that the arguments aren't lists.

* Create a new "nolist" gst_value_compare which avoids that special
  case comparision

Benchmarks:
 valgrind/callgrind: average speedup in instruction calls for
 gst_value_intersect and gst_value_subtract is around 56% (Makes 63%
 of the calls it used to take previously)

 tests/benchmarks/capsnego: With default settings (depth 4, children 3
 607 elements), time taken for transition from READY to PAUSED:
   Before : 00.391519153
   After  : 00.220397492
    56% of the time previously used, +77% speedup

https://bugzilla.gnome.org/show_bug.cgi?id=731756

9 years agotests: remove some cruft from the bufferlist test
Tim-Philipp Müller [Tue, 17 Jun 2014 13:39:00 +0000 (14:39 +0100)]
tests: remove some cruft from the bufferlist test

Buffers no longer carry caps, and bufferlists don't have
groups where buffers may need to be merged into one any more.

9 years agotests: add test for gst_buffer_list_remove()
Tim-Philipp Müller [Mon, 16 Jun 2014 19:30:13 +0000 (20:30 +0100)]
tests: add test for gst_buffer_list_remove()

9 years agobufferlist: fix buffer leak in _remove()
Tim-Philipp Müller [Mon, 16 Jun 2014 19:29:56 +0000 (20:29 +0100)]
bufferlist: fix buffer leak in _remove()

9 years agoflowcombiner: fix g-i transfer annotations
Tim-Philipp Müller [Mon, 16 Jun 2014 08:18:45 +0000 (09:18 +0100)]
flowcombiner: fix g-i transfer annotations

9 years agoflowcombiner: Fixed GBoxedCopyFunc
Edward Hervey [Mon, 16 Jun 2014 06:41:48 +0000 (08:41 +0200)]
flowcombiner: Fixed GBoxedCopyFunc

I'll just quote the most interesting man in the world:

"I don't usually push commits, but when I do I don't compile it
first"

9 years agodevicemonitor: some docs additions and fixes
Tim-Philipp Müller [Sat, 14 Jun 2014 15:30:49 +0000 (16:30 +0100)]
devicemonitor: some docs additions and fixes

9 years agowin32: add exports for new get_type() function
Tim-Philipp Müller [Sat, 14 Jun 2014 15:28:48 +0000 (16:28 +0100)]
win32: add exports for new get_type() function

9 years agoflowcombiner: keep a ref to the pads we're using
Tim-Philipp Müller [Sat, 14 Jun 2014 10:31:44 +0000 (11:31 +0100)]
flowcombiner: keep a ref to the pads we're using

Needed for use via the boxed type.

https://bugzilla.gnome.org/show_bug.cgi?id=731355

9 years agoflowcombiner: add boxed type for bindings
Tim-Philipp Müller [Sat, 14 Jun 2014 09:54:41 +0000 (10:54 +0100)]
flowcombiner: add boxed type for bindings

https://bugzilla.gnome.org/show_bug.cgi?id=731355

9 years agointrospection: minor annotation additions
Evan Nemerson [Wed, 11 Jun 2014 23:28:51 +0000 (16:28 -0700)]
introspection: minor annotation additions

https://bugzilla.gnome.org/show_bug.cgi?id=731541

9 years agointrospection: include gstversion.h in GIR generation
Evan Nemerson [Thu, 12 Jun 2014 02:08:04 +0000 (19:08 -0700)]
introspection: include gstversion.h in GIR generation

https://bugzilla.gnome.org/show_bug.cgi?id=703021

9 years agobytereader: Use concistant derefence method
Nicolas Dufresne [Tue, 10 Jun 2014 14:23:13 +0000 (10:23 -0400)]
bytereader: Use concistant derefence method

This is minor style fix to not mix *var and var[N].

9 years agobytereader: Use pointer instead of index access
Sungho Bae [Tue, 10 Jun 2014 13:35:38 +0000 (09:35 -0400)]
bytereader: Use pointer instead of index access

Currently the scan uses Boyer-moore method and its performance is good.
but, it can be optimized from an implementation of view.

The original scan code is implemented by byte array and index-based access.
In _scan_for_start_code(), the index is increasing from start to end and the
base address of the byte array is referred to as return value.

In the case, index-based access can be replaced by pointer access, which
improve the performance by removing index-related operations.

Its performace is enhanced by approximately 8% on arm-based embedded devices.
Although it seems trivial, it can affect the overall performance because the
_scan_for_start_code() function is very often called when H.264/H.265 video is
played.

In addition, the technique can apply for all architectures and it is good in
view of readability and maintainability.

https://bugzilla.gnome.org/show_bug.cgi?id=731442

9 years agoglobaldevicemonitor: prettify header
Tim-Philipp Müller [Sat, 7 Jun 2014 09:13:56 +0000 (10:13 +0100)]
globaldevicemonitor: prettify header

9 years agotests: add unit test for queuearray expansion from 1
Tim-Philipp Müller [Sat, 7 Jun 2014 08:46:42 +0000 (09:46 +0100)]
tests: add unit test for queuearray expansion from 1

https://bugzilla.gnome.org/show_bug.cgi?id=731349

9 years agoqueuearray: fix expanding size of queue from 1
Evan Nemerson [Fri, 6 Jun 2014 23:36:00 +0000 (16:36 -0700)]
queuearray: fix expanding size of queue from 1

Without we would not actually expand and access
memory beyond the allocated region for the array.

https://bugzilla.gnome.org/show_bug.cgi?id=731349

9 years agodataqueue: clear up documentation of gst_data_queue_new
Evan Nemerson [Thu, 5 Jun 2014 23:55:15 +0000 (16:55 -0700)]
dataqueue: clear up documentation of gst_data_queue_new

The gpointer argument is passed to all three callbacks, not just one.

https://bugzilla.gnome.org/show_bug.cgi?id=731302

9 years agointrospection: fix some minor annotation bugs
Evan Nemerson [Fri, 30 May 2014 07:17:06 +0000 (00:17 -0700)]
introspection: fix some minor annotation bugs

https://bugzilla.gnome.org/show_bug.cgi?id=730982

9 years agobase: use correct syntax in documentation more consistently
Evan Nemerson [Thu, 5 Jun 2014 19:38:20 +0000 (12:38 -0700)]
base: use correct syntax in documentation more consistently

Previously, many constants were prefixed with # or unprefixed,
some functions and macros were prefixed with # instead of suffixed
with (), etc.

https://bugzilla.gnome.org/show_bug.cgi?id=731293

9 years agobaseparse: Pass rate of input segment to output segment
zhouming [Wed, 7 May 2014 10:26:38 +0000 (18:26 +0800)]
baseparse: Pass rate of input segment to output segment

https://bugzilla.gnome.org/show_bug.cgi?id=729701

9 years agogstbuffer: factor three flags-to-string loops
Vincent Penquerc'h [Mon, 7 Apr 2014 13:49:59 +0000 (14:49 +0100)]
gstbuffer: factor three flags-to-string loops

9 years agoinfo: make printing datetimes work with GST_PTR_FORMAT
Tim-Philipp Müller [Tue, 3 Jun 2014 22:42:45 +0000 (23:42 +0100)]
info: make printing datetimes work with GST_PTR_FORMAT

9 years agodatetime: change internal implementation to mini object
Tim-Philipp Müller [Tue, 3 Jun 2014 22:38:28 +0000 (23:38 +0100)]
datetime: change internal implementation to mini object

And move type stuff from GstValue to GstDateTime.

9 years agodownloadbuffer: fix uninitialized variable
Wim Taymans [Tue, 3 Jun 2014 20:19:33 +0000 (22:19 +0200)]
downloadbuffer: fix uninitialized variable

9 years agodownloadbuffer: improve start/stop in buffering query
Wim Taymans [Tue, 3 Jun 2014 20:12:13 +0000 (22:12 +0200)]
downloadbuffer: improve start/stop in buffering query

The start and stop should represent the currently downloading region.
The estimated-total should represent the remaining time to download
the currently downloading region. This makes it a lot more useful
for applications because they can then use those values to update
the fill region and use the estimated time to delay playback.
Update the docs with this clarification.

9 years agoidentity: add static and const where appropriate
Vincent Penquerc'h [Mon, 7 Apr 2014 13:35:04 +0000 (14:35 +0100)]
identity: add static and const where appropriate

9 years agoidentity: fix potential buffer overflow
Vincent Penquerc'h [Mon, 7 Apr 2014 13:31:17 +0000 (14:31 +0100)]
identity: fix potential buffer overflow

Coverity 1037155

9 years agodownloadbuffer: reset read and write positions
Wim Taymans [Tue, 3 Jun 2014 12:49:44 +0000 (14:49 +0200)]
downloadbuffer: reset read and write positions

Reset the read and write positions right after we open the file or flush
it. We are also in the buffering state with 0 percent buffered when we
start.

9 years agoinfo: first handle all miniobjects, then GObjects
Wim Taymans [Tue, 3 Jun 2014 12:47:17 +0000 (14:47 +0200)]
info: first handle all miniobjects, then GObjects

First handle all miniobjects before we attempt to dereference the first
field pointer and look at the GType. With the recent glib change to
speed up G_IS_OBJECT, this causes crashes on miniobjects otherwise.

9 years agoinfo: GstDateTime does not have a GType as first field
Wim Taymans [Tue, 3 Jun 2014 12:46:11 +0000 (14:46 +0200)]
info: GstDateTime does not have a GType as first field

GstDateTime does not have the GType as the first field so we can't use
it to detect its type.

9 years agoinfo: use macros to check types
Wim Taymans [Tue, 3 Jun 2014 12:45:22 +0000 (14:45 +0200)]
info: use macros to check types

Use the macros to check the type of objects instead of directly poking
at the first field.

9 years agoglobaldevicemonitor: connect sync-message signal on the right object
Tim-Philipp Müller [Sun, 1 Jun 2014 22:51:20 +0000 (23:51 +0100)]
globaldevicemonitor: connect sync-message signal on the right object

Fixes criticals at runtime and makes stuff actually work.

9 years agotypefind: Keep still meaningfull pending events on FLUSH_STOP
Sebastian Dröge [Sat, 31 May 2014 15:35:52 +0000 (17:35 +0200)]
typefind: Keep still meaningfull pending events on FLUSH_STOP

Only EOS and segment should be deleted in that case.

https://bugzilla.gnome.org/show_bug.cgi?id=709868

9 years agoRevert "miniobject: Add missing (nullable) annotations"
Sebastian Dröge [Fri, 30 May 2014 07:13:12 +0000 (09:13 +0200)]
Revert "miniobject: Add missing (nullable) annotations"

This reverts commit 96361e9b5c5d00dc7712ff3a9acfbe10df7cd9fe.

This was not supposed to be pushed yet!

9 years agobufferpool: It's pool, not poo... even when talking about flushing
Sebastian Dröge [Fri, 30 May 2014 07:12:14 +0000 (09:12 +0200)]
bufferpool: It's pool, not poo... even when talking about flushing

9 years agominiobject: Add missing (nullable) annotations
Philip Withnall [Wed, 28 May 2014 09:14:45 +0000 (10:14 +0100)]
miniobject: Add missing (nullable) annotations

gst_mini_object_replace() can take NULL mini-objects.

https://bugzilla.gnome.org/show_bug.cgi?id=730873

9 years agotests: multiqueue: fix leaks
Thiago Santos [Fri, 30 May 2014 04:42:17 +0000 (01:42 -0300)]
tests: multiqueue: fix leaks

9 years agodocs: convert NULL, TRUE, and FALSE to %NULL, %TRUE, and %FALSE
Evan Nemerson [Thu, 29 May 2014 21:54:34 +0000 (14:54 -0700)]
docs: convert NULL, TRUE, and FALSE to %NULL, %TRUE, and %FALSE

This should help improve documentation generated for
languages other than C.

https://bugzilla.gnome.org/show_bug.cgi?id=730961

9 years agodocs: fix type in GstObject docs
Tim-Philipp Müller [Thu, 29 May 2014 23:13:30 +0000 (00:13 +0100)]
docs: fix type in GstObject docs

9 years agobufferpool: fix gst_buffer_pool_has_option() documentation
Evan Nemerson [Thu, 29 May 2014 22:04:45 +0000 (15:04 -0700)]
bufferpool: fix gst_buffer_pool_has_option() documentation

https://bugzilla.gnome.org/show_bug.cgi?id=730962

9 years agoelement: set pads need-parent flag to false when removing
Thiago Santos [Thu, 29 May 2014 17:07:15 +0000 (14:07 -0300)]
element: set pads need-parent flag to false when removing

When a pad is added the need-parent flag is set to true, so when
they are removed the flag should be set back to false

This was preventing GstPads to be reused in elements (removed and
later re-added). A unit tests was added to verify that this is
working now.

The use case is tsdemux that has a program-number property and
allows the user to switch programs. In order to do that tsdemux
will remove the pads of the current program and add from the new
ones. The removed pads are kept in the demuxer for later if the
user selects the old program again.

9 years agomultiqueue: post buffering message when queues flush
Thiago Santos [Tue, 27 May 2014 11:09:36 +0000 (08:09 -0300)]
multiqueue: post buffering message when queues flush

The buffering status goes back to 0, so inform the application about it

https://bugzilla.gnome.org/show_bug.cgi?id=726423

9 years agogitignore: Ignore VIM swap files
Nicolas Dufresne [Thu, 29 May 2014 18:39:36 +0000 (14:39 -0400)]
gitignore: Ignore VIM swap files

9 years agopad: two minor docs fixes
Tim-Philipp Müller [Tue, 27 May 2014 12:36:29 +0000 (13:36 +0100)]
pad: two minor docs fixes

10 years agoflowcombiner: beautify headers a little
Tim-Philipp Müller [Tue, 27 May 2014 09:09:02 +0000 (10:09 +0100)]
flowcombiner: beautify headers a little

10 years agodocs: add GstFlowCombiner
Tim-Philipp Müller [Tue, 27 May 2014 09:05:51 +0000 (10:05 +0100)]
docs: add GstFlowCombiner

10 years agobase: include flowcombiner header from base.h
Tim-Philipp Müller [Tue, 27 May 2014 08:55:27 +0000 (09:55 +0100)]
base: include flowcombiner header from base.h

10 years agoflowcombiner: add GstFlowCombiner
Thiago Santos [Mon, 26 May 2014 15:31:33 +0000 (12:31 -0300)]
flowcombiner: add GstFlowCombiner

Adds a utility struct that is capable of storing and aggregating flow returns
associated with pads.

This way all demuxers will have a standard function to use and have the
same expected results.

Includes tests.

https://bugzilla.gnome.org/show_bug.cgi?id=709224

10 years agopad: store last flow return and provide acessor function
Thiago Santos [Fri, 23 May 2014 16:25:35 +0000 (13:25 -0300)]
pad: store last flow return and provide acessor function

Stores the last result of a gst_pad_push or a pull on the GstPad and provides
a getter and a macro to access this field.

Whenever the pad is inactive it is set to FLUSHING

API: gst_pad_get_last_flow_return

https://bugzilla.gnome.org/show_bug.cgi?id=709224

10 years agobufferpool: Add method and virtuals to set flushing state
Nicolas Dufresne [Fri, 23 May 2014 19:26:59 +0000 (15:26 -0400)]
bufferpool: Add method and virtuals to set flushing state

Currently there is no other way to unlock a buffer pool other then
stopping it. This may have the effect of freeing all the buffers,
which is too heavy for a seek. This patch add a method to enter and
leave flushing state. As a convenience, flush_start/flush_stop
virtual are added so pool implementation can also unblock their own
internal poll atomically with the rest of the pool.  This is fully
backward compatible with doing stop/start to actually flush the pool
(as being done in GstBaseSrc).

https://bugzilla.gnome.org/show_bug.cgi?id=727611

10 years agobasetransform: Passthrough ALLOCATION queries in passthrough mode even if we had...
Sebastian Dröge [Mon, 26 May 2014 12:23:13 +0000 (14:23 +0200)]
basetransform: Passthrough ALLOCATION queries in passthrough mode even if we had no caps yet

Or if the element does not care about caps at all.

Also remove an assigned but unused local variable.

https://bugzilla.gnome.org/show_bug.cgi?id=710268

10 years agopo: update
Tim-Philipp Müller [Sun, 25 May 2014 15:10:30 +0000 (16:10 +0100)]
po: update

10 years agopo: update POTFILES
Piotr Drąg [Sun, 25 May 2014 14:57:59 +0000 (16:57 +0200)]
po: update POTFILES

https://bugzilla.gnome.org/show_bug.cgi?id=730718

10 years agoBack to development
Sebastian Dröge [Wed, 21 May 2014 11:23:21 +0000 (13:23 +0200)]
Back to development

10 years agoRelease 1.3.2 1.3.2
Sebastian Dröge [Wed, 21 May 2014 11:06:34 +0000 (13:06 +0200)]
Release 1.3.2

10 years agoUpdate .po files
Sebastian Dröge [Wed, 21 May 2014 09:39:53 +0000 (11:39 +0200)]
Update .po files

10 years agoAutomatic update of common submodule
Sebastian Dröge [Wed, 21 May 2014 08:50:43 +0000 (10:50 +0200)]
Automatic update of common submodule

From 211fa5f to 1f5d3c3

10 years agovalue: Add some positive testcase for string deserialization
Sebastian Dröge [Mon, 19 May 2014 09:05:12 +0000 (11:05 +0200)]
value: Add some positive testcase for string deserialization

10 years agodocs: remove reference to Mandrake and packages we no longer provide
Tim-Philipp Müller [Sun, 18 May 2014 09:49:50 +0000 (10:49 +0100)]
docs: remove reference to Mandrake and packages we no longer provide

https://bugzilla.gnome.org/show_bug.cgi?id=730312

10 years agodocs: fix typo
Wim Taymans [Thu, 15 May 2014 14:41:58 +0000 (16:41 +0200)]
docs: fix typo

10 years agopluginloader: fix compiler warning on windows
Tim-Philipp Müller [Wed, 14 May 2014 12:40:03 +0000 (13:40 +0100)]
pluginloader: fix compiler warning on windows

gstpluginloader.c:584:1: error: label 'beach' defined but not used

https://bugzilla.gnome.org/show_bug.cgi?id=730125

10 years agoelements: don't depend on libgio just for g_io_error_from_errno()
Tim-Philipp Müller [Tue, 13 May 2014 18:51:34 +0000 (19:51 +0100)]
elements: don't depend on libgio just for g_io_error_from_errno()

https://bugzilla.gnome.org/show_bug.cgi?id=729949

10 years agosparsefile: keep it private as helper API for downloadbuffer
Tim-Philipp Müller [Tue, 13 May 2014 18:30:38 +0000 (19:30 +0100)]
sparsefile: keep it private as helper API for downloadbuffer

There's no expectation that any other element or applications
might want to use this helper API any time soon, so keep it
private for the time being. There were open questions regarding
portability and binding-friendliness too.

This also removes the gio dependency of -base again.

https://bugzilla.gnome.org/show_bug.cgi?id=729951
https://bugzilla.gnome.org/show_bug.cgi?id=729949

10 years agodocs: pick up GstBaseParse hierarchy and properties
Tim-Philipp Müller [Tue, 13 May 2014 18:14:08 +0000 (19:14 +0100)]
docs: pick up GstBaseParse hierarchy and properties

10 years agodocs: expose GstPushSrcClass in documentation
Tim-Philipp Müller [Tue, 13 May 2014 18:10:43 +0000 (19:10 +0100)]
docs: expose GstPushSrcClass in documentation

Might come in handy in case someone wants to derive from it.

10 years agopluginloader: Don't leak pluginloader in error cases
Edward Hervey [Mon, 12 May 2014 15:03:46 +0000 (17:03 +0200)]
pluginloader: Don't leak pluginloader in error cases

CID #1212154

10 years agocaps: Don't leak features on error cases
Edward Hervey [Mon, 12 May 2014 14:59:29 +0000 (16:59 +0200)]
caps: Don't leak features on error cases

If we fail to parse fields, we would end up leaking the features we
parsed just before

CID #1212152

10 years agobasetransform: Correctly reset configuration
Nicolas Dufresne [Fri, 9 May 2014 18:28:59 +0000 (14:28 -0400)]
basetransform: Correctly reset configuration

When pool can't we use, and we fall back to default pool, we need to
correctly reset that pool configuration.

10 years agonettimeprovider: Use non-freed variable
Edward Hervey [Fri, 9 May 2014 12:46:59 +0000 (14:46 +0200)]
nettimeprovider: Use non-freed variable

address is only used temporarily. Use the proper variable instead.

CID #1212189

10 years agotests: multiqueue: test to check queue overrun with pts=none
Thiago Santos [Thu, 8 May 2014 20:33:37 +0000 (17:33 -0300)]
tests: multiqueue: test to check queue overrun with pts=none

Checks if buffers with pts=none can break the queue time size limit
and allow more buffers than expected

10 years agodownloadbuffer: Fix 32bit build
Nicolas Dufresne [Thu, 8 May 2014 18:48:00 +0000 (14:48 -0400)]
downloadbuffer: Fix 32bit build

format '%lli' expects argument of type 'long long int', but argument 8 has type 'gsize'

10 years agopool-nego: Correctly reset the configuration
Nicolas Dufresne [Thu, 8 May 2014 18:12:16 +0000 (14:12 -0400)]
pool-nego: Correctly reset the configuration

When pool cannot be used, correctly reset the configuration before
configuration a default pool.

10 years agopool-nego: Retry setting configuration with modified config
Nicolas Dufresne [Tue, 15 Apr 2014 18:17:00 +0000 (14:17 -0400)]
pool-nego: Retry setting configuration with modified config

Buffer pool set_config() may return FALSE if requested configuration needed
small changes. Reget the config and try setting it again (validating the
changes first). This ensure we have a configured pool if possible.

https://bugzilla.gnome.org/show_bug.cgi?id=727916

10 years agobufferpool: Add an helper to validate config
Nicolas Dufresne [Thu, 8 May 2014 16:47:43 +0000 (12:47 -0400)]
bufferpool: Add an helper to validate config

When we call gst_buffer_pool_set_config() the pool may return FALSE and
slightly change the parameters. This helper is useful to do the minial required
validation before accepting the modified configuration.

https://bugzilla.gnome.org/show_bug.cgi?id=727916

10 years agobufferpool: Update the configure even if set_config() returned false
Nicolas Dufresne [Tue, 8 Apr 2014 23:27:55 +0000 (19:27 -0400)]
bufferpool: Update the configure even if set_config() returned false

According to the documentation, when set_config() return false, it should be
possible to read the modified version of the config. This patch fixes the
implementation so it is now according to the documentation.

https://bugzilla.gnome.org/show_bug.cgi?id=727916

10 years agobufferpool: Add support for reconfiguring a pool
Nicolas Dufresne [Tue, 6 May 2014 19:35:14 +0000 (15:35 -0400)]
bufferpool: Add support for reconfiguring a pool

If a pool config is being configured again, check if the configuration have changed.
If not, skip that step. Finally, if the pool is active, try deactivating it.

https://bugzilla.gnome.org/show_bug.cgi?id=728268

10 years agovalue: Add support for GstAllocationParams comparision
Nicolas Dufresne [Tue, 6 May 2014 20:59:34 +0000 (16:59 -0400)]
value: Add support for GstAllocationParams comparision

This is useful to compare buffer pool configuaration.

https://bugzilla.gnome.org/show_bug.cgi?id=728268

10 years agovalue: Add support for GObject comparising in structures
Nicolas Dufresne [Tue, 6 May 2014 20:46:55 +0000 (16:46 -0400)]
value: Add support for GObject comparising in structures

This is useful to allow comparing pool configuration where a GstAllocator
is set.

https://bugzilla.gnome.org/show_bug.cgi?id=728268

10 years agoplugin: fix case where gst_plugin_load_file() didn't set the error on failure
Tim-Philipp Müller [Thu, 8 May 2014 16:50:50 +0000 (17:50 +0100)]
plugin: fix case where gst_plugin_load_file() didn't set the error on failure