GitLab

Fix the following warning:

../../modules/video_output/wayland/shm.c:266:60: warning: excess elements in struct initializer
  266 |     Close, Prepare, Display, Control, ResetPictures, NULL, NULL,
      |                                                            ^~~~
../../modules/video_output/wayland/shm.c:266:60: note: (near initialization for 'ops')

Before calling ThreadChangeFilters().

No need to pass a value for deinterlacing anymore.

They do the same thing and the string is always used in ThreadChangeFilters no
matter what.

We always call this function under filter lock.

We don't use the control lock anymore (from vout_control_PushBool), just the
filters lock.

Now the interlacing mode change is applied in frame by frame mode.

We don't need to wait when enabling filters, it's OK if we miss the current
frame and apply it in the next loop iteration. The filter changes will
be applied just before the picture is rendered, only once between frames. It is
now done after the control pop deadline, so will eat some of the rendering
time.

We don't use the control lock anymore (from vout_control_PushString), just the
filters lock.

Now the filter changes are applied even in frame by frame mode.

We can do this in the UI thread, the processing is light.
Locking the display_lock doesn't have a big impact on lag as it's only taking
time when reloading a display module, which is very rare.

They don't need to know about it.

It should already be done by opengl_interop_init in case of success.

Fix #25077

on seek or activation, the seek reference point is
inside a segment and if that segment is longer than
buffering duration, it could never start.
(segment start + buffering amount < deadline)

Only 4 modules actually use the Reset Pictures feature to create a new filter
pipeline when they can't handle the source anymore. It's still faster than
closing the module and probing for a new display module.

Alon Sirota authored Sep 08, 2020 and Hugo Beauzée-Luyssen committed Sep 14, 2020

Signed-off-by: Hugo Beauzée-Luyssen <hugo@beauzee.fr>

Alon Sirota authored Sep 06, 2020 and Hugo Beauzée-Luyssen committed Sep 14, 2020

All selected subtitles are now downloaded, instead of only one.

Change 'get_first_sel' function to return ALL selected subtitles
instead of only one, and refactored name to 'get_all_sel'.

Change 'download_subtitles' function to iterate over 'get_all_sel'
returned table instead of downloading the only subtitle from old
'get_first_sel'.
Signed-off-by: Hugo Beauzée-Luyssen <hugo@beauzee.fr>

Steve Lhomme authored Aug 10, 2020 and Felix Paul Kühne committed Sep 11, 2020

Signed-off-by: Felix Paul Kühne <felix@feepk.net>

This solves compilation issues with contrib libraries using meson for future macOS architectures

It's supposed to be handled in the window, not in the display module.

It's likely broken at this stage.

It's supposed to be handled in the window, not in the display module.

In the future it may be called directly from the thread generating the event to
lower the latency.

The temporary and output format are the same at this stage.

It also shows potential that might need some attention when vd->source changes in push.

In some cases the display modules may rotate the video format, resulting in
inverted and non matching SAR. But in that case the width/height will also be
rotated and thus not matching either.

Also the comparision between the input and preferred format of the display
module explicitly zero the SAR for comparison (although it may hide some needed
conversion).

Instead of adding explicit opt-out for the known unsupported
architectures (armv7, aarch64), just enable it for the specific
architectures we know are supported.

And some code cleaning.

No functional changes.

If there are some extra data, the chroma should be constant and we can estimate
it so that we create the decoder output with the proper chroma on open.

The VLC chroma contains both the subsampling and the bitdepth.

VLC_CODEC_GREY* is used for monochrome.