and input_preparser_callbacks_t to struct vlc_metadata_cbs

Instead of libvlc_ArtRequest()

SCOPE_* enums are for the preparser (not for the fetcher) and this
option only touches the preparser.

This will allow to use the preparser to only fetch metadata.

Rémi Denis-Courmont authored 2 years ago and Jean-Baptiste Kempf committed 2 years ago

The 3rd formatted value has a type of vlc_tick_t, a.k.a. int64_t,
not necessary long.

Alexandre Janniaux authored 2 years ago and Steve Lhomme committed 2 years ago

Ensure that the default framebuffer is setup during MakeCurrent(), so
that any vlc_gl_t client can expect the state to be correct after
calling vlc_gl_MakeCurrent() and without changing it.

It seems that the default framebuffer was not always kept, leading to
GL_INVALID_FRAMEBUFFER errors on Linux with a MESA egl_pbuffer (GBM
backend) GLES2 implementation:

    $ MESA_DEBUG=1 ./vlc -vv ~/Video/dog_meme.mp4 --video-filter='opengl{filter=mock{mask},gles=any}'
    ...
    Mesa: User error: GL_INVALID_FRAMEBUFFER_OPERATION in glClear(incomplete framebuffer)
    Mesa: User error: GL_INVALID_FRAMEBUFFER_OPERATION in glDrawArrays
    vlc: ../../modules/video_output/opengl/filter_draw.c:112: Draw: Assertion `!"GL_INVALID_FRAMEBUFFER_OPERATION"' failed.

Alexandre Janniaux authored 2 years ago and Steve Lhomme committed 2 years ago

MakeCurrent() (triggered by vlc_gl_MakeCurrent) can have different
behaviour than just eglMakeCurrent() to match the VLC API semantic.

However, we'll need to reset the default framebuffer from this function
for this very VLC semantic, whereas the default framebuffer is not yet
existing when calling this function from Open.

By using eglMakeCurrent(), we allow future behaviour change to handle
this correctly.

Alexandre Janniaux authored 2 years ago and Steve Lhomme committed 2 years ago

Fix a future issue when using the code against a GLES2 provider, where
the detection was done at build time instead of runtime. The build time
check is needed to prevent using non-existant function and defines, but
the runtime check will be needed whenever OpenGL headers are available.

With the work allowing to use the opengl filter with a gles2 provider on
Linux with OpenGL available (not merged), it leads to:

    $ MESA_DEBUG=1 ./vlc -vv ~/Video/dog_meme.mp4 --video-filter='opengl{filter=mock{mask},gles=any}'
    Mesa: User error: GL_INVALID_ENUM in glTexParameter(pname=GL_TEXTURE_PRIORITY)
    Mesa: User error: GL_INVALID_OPERATION in unsupported function called (unsupported extension or deprecated function?)
    -> GL_ASSERT_NOERROR triggers

Oddly, it was not reproduced with `-V gles2`, but only because the
interop code is built into the convenience libvlc_opengles.la with the
USE_OPENGL_ES2 define, which is what is currently disabling this code.

Alexandre Janniaux authored 2 years ago and Steve Lhomme committed 2 years ago

Ensure that draw commands are not leading to any error later in the
pipeline.

Alexandre Janniaux authored 2 years ago and Steve Lhomme committed 2 years ago

Ensure that draw commands are not leading to any error later in the
pipeline.

Alexandre Janniaux authored 2 years ago and Steve Lhomme committed 2 years ago

Ensure that draw commands are not leading to any error later in the
pipeline.

Alexandre Janniaux authored 2 years ago and Steve Lhomme committed 2 years ago

Ensure that draw commands are not leading to any error later in the
pipeline.

Marvin Scholz authored 2 years ago and Jean-Baptiste Kempf committed 2 years ago

While runtime gettext support was already working,
this adds the missing piece of build integrations.

With these changes, three new targets are now available
to manage translation files:

- vlc-pot: Regenerates the .pot files using xgettext
- vlc-update-po: Regenerates the .po files from the current .pot files
- vlc-gmo: Builds (but not installs) the translations

Benjamin Arnaud authored 2 years ago and Jean-Baptiste Kempf committed 2 years ago

No functional changes

Benjamin Arnaud authored 2 years ago and Jean-Baptiste Kempf committed 2 years ago

No functional changes

Benjamin Arnaud authored 2 years ago and Jean-Baptiste Kempf committed 2 years ago

No functionnal changes

Fatih Uzunoğlu authored 2 years ago and Jean-Baptiste Kempf committed 2 years ago

preserve the interface geometry when going into fullscreen,
in order to restore it later (when exiting the fullscreen
mode).

Alexandre Janniaux authored 2 years ago and Jean-Baptiste Kempf committed 2 years ago

input_SetEsCatDelay will apply the delay synchronously if the input is
not running yet, or push a control to apply it from the input thread
otherwise.

Co-authored-by: Thomas Guillem <thomas@gllm.fr>

Refs #27918

Alexandre Janniaux authored 2 years ago and Jean-Baptiste Kempf committed 2 years ago

When setting delay, audio output flushing and picture drops can happen
and lead to unpleasant result, like stuttering, on both outputs.

Since the delay was set by an input command, demux could send a bit of
data to play before the delay was really applied, leading to this effect
even when starting the playback.

This commit alone doesn't fix the issue, but it ensures that delay is
synchronously set before the start of the input when it's not running
yet, mimicking the other input methods allowing to apply a control
synchronously when the input is not yet running.

Co-authored-by: Thomas Guillem <thomas@gllm.fr>

Refs #27918

We only define them when they are missing from the mingw-w64 headers.

This is more readable considering the comment.

It should be identical to MFVideoFormat_I420 according to
https://learn.microsoft.com/en-us/windows/win32/directshow/yuv-video-subtypes

Otherwise it's missing IID_IDXGIDevice used from C.

This is the proper way according to the documentation.

This is the proper way according to the documentation.