| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
| |
GStreamer sinks can implement the buffer_alloc() vfunc to provide
buffers for the upstream elements.
|
| |
|
|
|
|
|
| |
clutter_shader_ref/unref() were forgotten in the last batch of
deprecated symbols and cogl_check_extensions() had to die.
|
|
|
|
| |
Let's depend on a stable version of Clutter, not a development one.
|
|
|
|
|
|
|
|
| |
Now that we use automake's silent rules, I still want GNU make to behave
as if it were called with the -s switch when we are in silent mode.
AS_AM_REALLY_SILENT does just that, respecting the default verbosity and
the V environment variable.
|
|
|
|
|
| |
It's a fact, shave is dead, automake's silent rules are the way forward
to have more silent builds.
|
|
|
|
|
| |
We use gio-2.0 internally (not exposing it in the headers) so it's good
to add a Requires.private field to make that fact clear.
|
|
|
|
|
|
|
|
|
| |
Since clutter 1.1.10, new textures might end up in a big atlas. To be
sure this does not happen for frames, we add the NO_ATLAS flag when
creating the texture.
This is just to be futureproof, the atlas does not currently fiddle with
luminance textures, but you never know.
|
|
|
|
| |
We don't really want automake to pull "random" files into the project
|
| |
|
|
|
|
|
| |
The GMutex of the GSource was not destroyed when the GSource is. Plug
it!
|
|
|
|
|
|
| |
No need to copy the caps when some upstream element is asking what we
can do. Reffing them is enough as the caps are built at object
construction time and and kept until destruction time.
|
|
|
|
|
|
| |
ClutterMedia has a new API (since 1.2) to let the user specify an URI of
a subtitle file and display them along with a video. You can also
specify the font that should be used when rendering.
|
|
|
|
|
| |
cogl_texture_ref/unref() is deprecated in favour of the more generic
cogl_handle_ref/unref(), so let's use that.
|
| |
|
| |
|
| |
|
|
|
|
|
|
| |
Cogl had some CGL_ macros in the public headers, wrapping the GL enums.
Nobody was supposed to use them. Now that they are really gone, we need
to use the real GL define instead.
|
|
|
|
|
|
|
|
|
|
| |
The purpose of the library is to provide ClutterActors. You can play
audio files using a ClutterGstVideoTexture as all you do is giving
playbin2 an URI anyway.
Not having to support an audio only ClutterMedia object will allow us to
add more video related functions inside ClutterMedia without worrying
about an audio-only implementation of the interface.
|
|
|
|
|
|
| |
By adding G_DISABLE_DEPRECATED, COGL_DISABLE_DEPRECATED and
CLUTTER_DISABLE_DEPRECATED to AM_CPPFLAGS. Also disable single includes
for GLib as the trend is going that way.
|
|
|
|
|
|
|
| |
Users of the library should not include individual headers but the top
level one only (<clutter-gst/clutter-gst.h>). This is to allow us to
shuffle things around, remove or rename headers without breaking
anything.
|
|
|
|
|
|
|
| |
As the 1.0.0 release is closer than ever, it's time to fix make
distcheck. The issue was introduced with the introspection commit,
g-ir-scanner needs to access the source files and thus those need to be
prefixed by $(srcdir)/
|
|
|
|
|
|
|
|
|
|
|
| |
Under some circumstances, the default priority of the ClutterGstSource
is not enough to have every frame being uploaded. Setting a higher
priority to ClutterGstVideoSink allows new video frames to be painted
even in a busy mainloop.
Thanks to Bastian Winkler <buz@netbuz.org> for the initial patch
Fixes: OHB#1412
|
|
|
|
|
| |
It's been there since the beginning without have a small sentence to
describe it, let's fix that.
|
|
|
|
|
| |
clutter-gst-debug.h does not need the cluttre_gst_init*() functions, no
idea why that include was there.
|
|
|
|
|
| |
So glib does not have to copy them, less things to do, less memory used,
happy bunnies.
|
|
|
|
|
| |
GStreamer's default subtitle font is a small serif font. For a better
readability, let's default to Sans 16.
|
|
|
|
|
| |
ClutterGstVideoTexture is already using it so try to be consistent and
use it everywhere...
|
|
|
|
|
|
|
|
|
|
|
|
| |
The "duration" message is sent on the GstBus when an element has a new
duration. Then, this duration has to be queried from the pipeline.
Instead of that we tried to parse the duration message that returned
GST_CLOCK_TIME_NONE (which is the way to say "we have a new duration,
please query it.") and we emitted a new signal with that value.
This is basically the commit bd3a5266206ff3c730674655cea93a8d689efd8d
but for ClutterGstAudio.
|
|
|
|
|
| |
ClutterGstAudio has been left behing a bit. Start with adding debugging
traces.
|
|
|
|
|
| |
AM_MAINTAINER_MODE is a relic we do not use here. configure warns that
giving --enable-maintainer-mode is stupid, and it is indeed.
|
|
|
|
| |
We do depend on GNU make, that's a fact. Please don't warn about it.
|
|
|
|
|
| |
The API version is clearly not a good way of saying which version we
document. Switch to the actual library version.
|
|
|
|
| |
It's that time of the year again.
|
|
|
|
|
|
|
|
|
| |
I really don't want to make the user of clutter-gst switch to having
multiple threads issuing GL calls, so we stick with having clutter-gst
upload the texture (or do whatever needed GL call) from the Clutter
thread.
Let's make sure that's documented in the constructors that needs it.
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Instead of creating, every single frame, a new idle source and attach it
to the clutter GMainContext, we now attach, while the video is playing,
a GSource that will signal when there is a new buffer pending.
As a result, we do less work by frame -- which is always good -- and the
code is way more readable -- wich makes me very happy.
Based on an initial patch from Ole André Vadla Ravnås.
Fixes: OHB#1657
|
|
|
|
|
| |
This feature depends on the --add-init-section argument of g-ir-scanner
which is available since mid-v0.6.8.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The "duration" message is sent on the GstBus when an element has a new
duration. Then, this duration has to be queried from the pipeline.
Instead of that we tried to parse the duration message that returned
GST_CLOCK_TIME_NONE (which is the way to say "we have a new duration,
please query it.") and we emitted a new signal with that value.
Let's try to limit the duration signal emission even further. The signal
will only be emitted if it's useful to update a UI (1s precision). if
one need a more accurate duration, he can always get the duration
property when needed.
|
|
|
|
|
|
| |
When setting a new media URI, we look for subtitle files with the same
name than the media file but with a "sub", "srt", "smi", "ssa", "ass" or
"asc" extension. If found we play this file as subtitles.
|
|
|
|
|
|
| |
It's time to catch up with GStreamer and use playbin2 instead of
playbin. It can play more URLs already and will allow us to specify a
subtitle URL.
|
|
|
|
| |
A tab is always 8 spaces wide.
|
|
|
|
|
| |
The API version is not correlated to those numbers, so let's define it
by its own.
|
|
|
|
|
| |
As a rule of thumb, we should only add to EXTRA_DIST the m4 files we
have in the git repository.
|
|
|
|
|
|
| |
Several foreign m4 macros are put in the AC_CONFIG_MACRO_DIR by
autoconf, we want to ignore them. Let's make *.m4 the default ignore
rule and add exceptions for the m4 files we ship.
|
|
|
|
| |
An: ARBfp1.0 program handle is a GLuint, not a GLint.
|