p_reg is never NULL with a vlc_list and we don't have access to the
iterator outside of the loop.

Alexandre Janniaux authored 1 year ago and Steve Lhomme committed 1 year ago

Now that frameworks are probed into the frameworks dictionary, use them
everywhere a framework was set as dependency. See previous commit for
rationale.

Alexandre Janniaux authored 1 year ago and Steve Lhomme committed 1 year ago

Since the dependencies are declared conditionnally, it was not possible
to unconditionnally create plugins that would depend on them. By always
declaring them, they are now defined as a disabler() when the condition
are not met, without even the need to probe them.

In addition, the setup factors together the conditions that are needed
to probe or not the dependencies, and put everything in a single
location that can be shared between the different Apple platforms. It
also makes it clearer what kind of dependency is linked, though meson
was already abstracting all of the framework handling, and what is the
real name that should be found in the documentation.

Alexandre Janniaux authored 1 year ago and Steve Lhomme committed 1 year ago

Some tests will require some modules to be present to run correctly, but
it cannot be probed at compile time. However, as soon as every modules
will be declared unconditionnally, we'll be able to depend on a disabled
module and disable the matching tests.

Ideally, we'd skip it so that the test is still reported, but there's no
code to do that yet.

Alexandre Janniaux authored 1 year ago and Steve Lhomme committed 1 year ago

The video_chroma meson definition file defines chroma_copy_lib, which is
used by codecs like omxil. It must be sourced before then.

Alexandre Janniaux authored 1 year ago and Steve Lhomme committed 1 year ago

The 'enabled' field will be used to define a condition for enabling the
plugin. Having this as as separate field will allow to define every
modules unconditionnally, regardless of whether they will be compiled or
not, and will allow to express runtime dependencies for executables.

Claudio Cambra authored 1 year ago and Steve Lhomme committed 1 year ago

Signed-off-by: Claudio Cambra <developer@claudiocambra.com>

Claudio Cambra authored 1 year ago and Steve Lhomme committed 1 year ago

Signed-off-by: Claudio Cambra <developer@claudiocambra.com>

The proper value is already set in the toolchain.cmake since a6dcec33.
If the value is not set for a cross compiled target, building fails since f2e67d22.

They are supposed to be "window pixel coordinates" inside the display area.
The core does the coordinate translation.

It wasn't set in the test decoder callback.

Rather than create a local variant of it.

Rather than create a local variant of it.

It's cleaned in the destructor.

This is disabling the subtitle area "stacking" as it forces the source
subtitle subpicture to be absolute. But absolute subpictures have absolute
coordinates inside the (unscaled) video. They are not meant to be shown
outside of the video area.

We already pass the position of the video in the display. So we have both information all the time.
Plus the format SAR was already using the display SAR.

The subpicture doesn't need to know more.

To do that we need the video to fill the entire window.

It messes with the source folder location.

With FFmpeg 6.1 an extra item in the buffer is needed.
The API was introduced in https://github.com/FFmpeg/FFmpeg/commit/af94ab7c7c004786084903bcf82b7617e88e3aa9
But it was only made the default in 6.1 since https://github.com/FFmpeg/FFmpeg/commit/83ede01bb08239428fd65de62adc260f4233d229

This results in an out of bounds writing error on Windows.

Khalid Masum authored 1 year ago and Steve Lhomme committed 1 year ago

75b62f44 aims to generalize
HAVE_GLES2 with the introduction of (!HAVE_GL and HAVE_GLES2)
and removes android specific library adding.

However, emscripen, like android falls into this criteria but not
removed. Hence libraries like libvlc_opengles.la is added twice and
causing error: "duplicate symbol" during shared module building.

Fix this by removing emscripten's block of code that causes this.

Fatih Uzunoğlu authored 1 year ago and Steve Lhomme committed 1 year ago

I18n module was intending to provide translation through vlc_gettext().
I'm not sure why QTranslator was not the way chosen before, because it
directly offers this and using qsTr() has the advantage of better integration.

Translation is handled in a lower abstraction level through QJSEngine
(see QJSEngine::TranslationExtension) as opposed to using custom a QML type
through QQmlEngine.

Thanks to that, the `org.videolan.vlc` module import can be eliminated
in some files.

This was forcing the subtitle region to become absolute. This will
later break subtitles in black bar which can be outside of the video
area. Absolute subtitle areas can only describe a position in the video
area.

This was also broken if the subpicture updater decides to set the region
position at each subpicture update. It would update the forced location
we set earlier.

We only need the entry channel order and the subpicture absolute, subtitle and original dimensions.

When they are displayed outside of the black bars they are not cleaned properly.