| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
| |
The GLES2 shaders were considered private API until 1.6.0, see
discussion:
https://bugzilla.gnome.org/show_bug.cgi?id=661071
Let's update the variable names and depend on 1.6.0
(cherry picked from commit 11cce755880127565e88bd50c63c6f0b7ee6051f)
|
|
|
|
|
|
| |
And allows to ensure the VideoTexture and the VideoSink are correctly
finalized.
(cherry picked from commit a1a5a7221ed21f5ea43833cdd7839e2130fc6919)
|
|
|
|
|
|
|
|
|
|
|
| |
The sink does not need to hold a reference to the Texture, it better to
leave the life cycle of the ClutterTexture entirely to Clutter.
If the texture disapears when the sink is still live and pushing buffers
to it, you will get warnings not nothing dramatic. It's your
responsibility to ensure the GStreamer pipeline has been stopped when
removing/destroying the texture.
(cherry picked from commit ddbb091b4a4d4cce0f3b324819af975a246faf4d)
|
|
|
|
|
|
|
|
|
|
| |
VideoTexture -> GstPipeline -> VideoSink -> VideoTexture
Break the circular dependency by putting clutter_gst_player_deinit() in
the dispose.
This means that we need to make clutter_gst_player_deinit() idempotent
as well.
(cherry picked from commit 9f9ad6266c9e7cd870c86c8d5a346f9912cdad42)
|
| |
|
| |
|
| |
|
|
|
|
|
|
| |
These days, you get warnings from your compiler when using deprecated
function. That's enough for me.
(cherry picked from commit a290bf5747ce8307aa636a2b4861d02f933b9a6a)
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=661096
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
| |
Partially revert commit f92a80288a57cc0fbc58a74e3a143b7318649557
This fixes video playback on systems that uses the nouveau classic mesa driver.
You can't query an int field of a GstStructure as uint.
https://bugzilla.gnome.org/show_bug.cgi?id=660732
|
| |
|
|
|
|
|
| |
When the info is not ready yet, tag lists are NULL. Don't try to free
NULL tag lists...
|
| |
|
| |
|
|
|
|
| |
OCD.
|
| |
|
|
|
|
|
|
| |
Instead of having a list of strings, we give the full GstTagList to the
application so it can do whatever it wants to display information about
the tracks of the media.
|
|
|
|
|
|
|
|
|
|
|
|
| |
Since Cogl 1.8.0 we can use cogl_renderer_get_n_fragment_texture_units()
to know the number of texture image units available for fragment
shaders and we can thus avoid using the direct GL call glGetIntegerv and
linking against libGL.
This does not work on OS X though as we can't get the CoglContext out of
the ClutterBackend is that case. Make sure to special case this.
Fixes: https://bugzilla.gnome.org/show_bug.cgi?id=657225
|
|
|
|
| |
This allows to override implementation of the player if need be.
|
|
|
|
| |
And do that that after having set the new URI, not before.
|
| |
|
| |
|
|
|
|
|
|
| |
So we don't show a black frame when looping the video.
Fixes https://bugzilla.gnome.org/show_bug.cgi?id=657773
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
| |
Time to fix a bit the player. I was getting annoyed that everytime I
need to test something the player just fullscreens itself on me. First
step to have a more useful example that can test more features than just
the basic playback.
|
|
|
|
| |
As GNOME is moving to this format.
|
| |
|
| |
|
| |
|
|
|
|
| |
Let's fix those and enabled --warn-all to catch the next ones.
|
|
|
|
|
|
|
|
|
|
|
|
| |
We are using a gl call (glGetIntergerv). However, detecting wich GL
headers and library to use is getting impossible (as nowedays, cogl even
has a runtime selection of which library should be dlopened, so it could
be either one of flavours of the GLES1 DSO or the GLES2 DSO, or the Big
GL one).
So, really, we should not have any direct GL call in clutter-gst. The
solution is coming, but in the mean time, just link against -lGL when
it's safe to assume so.
|
|
|
|
|
|
|
| |
This is a API change, but I can because we are still in a unstable
branch. Really noone should have overriden that anyway.
This allows the signal to be handled in the Player and that's it.
|
|
|
|
| |
Nothing needs it, so just remove it.
|
|
|
|
|
|
| |
When ClutterGstPlayer has been split out, the deinit() code was
forgotten. It's back in with test-video-texture-new-unref-loop showing 0
leak. Yeah!
|
|
|
|
|
| |
And while I fixed it a few commits ago, we are leaking again with the
ClutterGstPlayer split.
|
| |
|
| |
|
| |
|
| |
|
|
|
|
| |
Add new API to get the list of subtitles in a stream.
|
|
|
|
| |
But rely on the idle state kept by the player.
|
|
|
|
|
|
| |
The eos_cb was split between VideoTexture and Player at first, but
then the implementation of Player changed witout and it was possiblet to
have the eos_cb handle everything. Except that it was not ported.
|