p_sample_offset_pts must be NULL if MP4_TRUN_SAMPLE_TIME_OFFSET is not
present. Then MP4_TrackGetPTSDelta would return -1 (instead of 0
previously), and PTS of video frames would be set to VLC_TS_INVALID,
which is the right thing to do in that case.
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Different return values have been removed, as they were not used.
Fix #7079

bitrate

Also fix a profile height error and missing baseline options for
android.

This reverts commit 15498896.

This reverts commit 32596ea8.

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

According to http://developer.apple.com/library/mac/#documentation/cocoa/conceptual/Strings/Articles/formatSpecifiers.html
%C is the right one for unicode characters.

I have been testing without this code for quite a while now, and I never
got the crash responsible for this hack anymore. So, lets remove the
hack, and if the crash occurs again, we should find a proper way to
fix it.

Now, we have two classes (instantiated from the xib file for each window)
with controls bar stuff:
- VLCControlsBarCommon holds all code common for main and detached window
- VLCMainWindowControlsBar adds code specific for the main window bar

With that, we can avoid all these redundant code for o_detached_*, furthermore
this decouples all detached window control bar stuff from MainWindow.m.

The objects can be accessed through the controlsBar method.

It's called O_NOINHERIT, but it really is the same as O_CLOEXEC.
As with POSIX, there is no way to set it for dup()/dup2() though.

This is more consistent with vlc_open() and vlc_pipe() behaviour.

On the Galaxy S3 at least, the component OMX.SEC.vc1.dec only
exposes this role, not video_decoder.vc1.

The role name video_decoder.vc1 is pretty recent (from OMX IL 1.2),
and in that spec version the coding field should be OMX_VIDEO_CodingVC1
instead of OMX_VIDEO_CodingWMV as it currently is. The omxil module
currently only uses OMX IL 1.1 headers, and so does Android.
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

This isn't necessary on Galaxy S3 for VC1 (since WMV3/VC1 is the
default type), but might be necessary for WMV1/2 (untested yet).
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

This is necessary for decoding of VC1 on Galaxy S3. (WMV3 still
doesn't work on that device, I haven't been able to figure out
what is missing yet.)
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

This tries to fix wrong behavior when optical cable is unplugged while spdif-output is running.

Currently, you will probably hear pops and you only get sound from internal speakers,
after you changed system volume.

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Threading does not currently work with some codec profiles and is thus
not suited as default. Threading is also known to deadlock in certain
circumstances, such as if no data is ever received (a rather common
scenario with network or DVB reception).

Fixed avcodec version should be checked before this is re-enabled.
Acked-by: Ilkka Ollakka <ileoo@videolan.org>

Less technical profiles, directly targeting devices.
More to be added.

macosx: work-around a playlist core limitation, which doesn't allow the UI to set the currently displayed root properly, so triggering a play can lead to unwanted results (close #7484)

- adjust detached time field margin to be as in main window
- set time slider to hidden by default in order to avoid small flashes on startup

Interestingly, the order inside the xib matters here...

Also shows possible warning for non native muxers.