But keep int timeout in milliseconds for functions requiring a timeout.

Do not review directly, all commits from
videolan/vlc!3503 squshed
together.

"Interactive" filters can only do one output per input.
Only "static" filters are allowed to generate more frames, like the deinterlacers.

There is no way at this point to probe the module that will be used
if it will output many frames or not. And even if it could the order of
filters would need to be respected. We cannot dynamically add filters
on one chain or the other.

With future drivers and old binaries it may trigger different code
pathes than the latest version.

The decoded video format may not use the final frame rate, which is used to
detect lateness.

We only allow overriding the source values if they make sense
and if the frame rate wasn't already settled.

Pierre Lamot authored 7 months ago and Steve Lhomme committed 7 months ago

We need to track the position of the VideoSurface globally, if the position of
one of its parent change, the VideoSurface isn't notified, so we need to track
the position of all of its parent

Pierre Lamot authored 7 months ago and Steve Lhomme committed 7 months ago

Qt renderer may mess with scissor/viewport

Prince Gupta authored 7 months ago and Steve Lhomme committed 7 months ago

fixes new indicator not visible anymore

fix regression from cbc63a0a
since the commit, content item is raised you can't put
element there anymore

Thomas Guillem authored 7 months ago and Steve Lhomme committed 7 months ago

It was only used by synchro.c.

Thomas Guillem authored 7 months ago and Steve Lhomme committed 7 months ago

It was a libmpeg2 specific option, not used anymore.

Thomas Guillem authored 7 months ago and Steve Lhomme committed 7 months ago

It was only used by libmpeg2.

Thomas Guillem authored 7 months ago and Steve Lhomme committed 7 months ago

cf. previous commit.

Thomas Guillem authored 7 months ago and Steve Lhomme committed 7 months ago

Lower priority than avcodec since VLC 3.0, so never used by default
since then.

Thomas Guillem authored 7 months ago and Steve Lhomme committed 7 months ago

cf. previous commit.

Thomas Guillem authored 7 months ago and Steve Lhomme committed 7 months ago

Lower priority than avcodec since VLC 3.0, so never used by default
since then.

Thomas Guillem authored 7 months ago and Steve Lhomme committed 7 months ago

cf. previous commit.

Thomas Guillem authored 7 months ago and Steve Lhomme committed 7 months ago

Lower priority than avcodec since VLC 3.0, so never used by default
since then.

Thomas Guillem authored 7 months ago and Steve Lhomme committed 7 months ago

Use newest version without a52 forced build.

Pierre Lamot authored 7 months ago and Steve Lhomme committed 7 months ago

Binding on `state` property was needed to avoid activating the visual transition
when component loads

fix: #28804

Pierre Lamot authored 7 months ago and Steve Lhomme committed 7 months ago

when a video becomes embed the playlist is hidden, the playlist visibility
state is kept but the playlist was hidden, clicking the playlist button was
toggling the playlist visiblity. As a result, if the playlist was visible, it
required the user to click twice to show the playlist.

Now when the button is clicked with video embed, it will visually toggle the
playlist and update the playlist "visible" value to the visual state.

Pierre Lamot authored 7 months ago and Steve Lhomme committed 7 months ago

fix: #28803

Martin Finkel authored 7 months ago and Steve Lhomme committed 7 months ago

Undefined symbols for architecture arm64:
  "_CACurrentMediaTime", referenced from:
      -[VLCPictureInPictureController pictureInPictureControllerTimeRangeForPlayback:] in libpictureinpicturecontroller_plugin_la-VLCPictureInPictureController.o
  "_CMTimeGetSeconds", referenced from:
      -[VLCPictureInPictureController pictureInPictureController:skipByInterval:completionHandler:] in libpictureinpicturecontroller_plugin_la-VLCPictureInPictureController.o
  "_CMTimeMakeWithSeconds", referenced from:
      -[VLCPictureInPictureController pictureInPictureControllerTimeRangeForPlayback:] in libpictureinpicturecontroller_plugin_la-VLCPictureInPictureController.o
  "_CMTimeRangeMake", referenced from:
      -[VLCPictureInPictureController pictureInPictureControllerTimeRangeForPlayback:] in libpictureinpicturecontroller_plugin_la-VLCPictureInPictureController.o
  "_kCMTimeNegativeInfinity", referenced from:
      -[VLCPictureInPictureController pictureInPictureControllerTimeRangeForPlayback:] in libpictureinpicturecontroller_plugin_la-VLCPictureInPictureController.o
  "_kCMTimePositiveInfinity", referenced from:
      -[VLCPictureInPictureController pictureInPictureControllerTimeRangeForPlayback:] in libpictureinpicturecontroller_plugin_la-VLCPictureInPictureController.o

Martin Finkel authored 7 months ago and Steve Lhomme committed 7 months ago

Undefined symbols for architecture arm64:
  "_CGRectZero", referenced from:
      -[VLCVideoUIView viewContainerBounds] in libuiview_window_plugin_la-VLCVideoUIView.o

This reverts commit 94db2f10.

These patches seems to break DVD playback with gcc on Linux.
It has not been merged upstream anyway.

For Windows LLVM builds !2400 is still necessary.

Signed-off-by: Steve Lhomme <robux4@ycbcr.xyz>

François Cartegnie authored 7 months ago and Steve Lhomme committed 7 months ago

Current pts/dts are bogus (pts < dts).
It should not depend on frame rate, but only source pictures.

Fatih Uzunoğlu authored 7 months ago and François Cartegnie committed 7 months ago

Qt 5.15 used to cause `QDragEnterEvent` to be
posted before `QDragLeaveEvent` is posted.

The ordering seems to be fixed with Qt 6. Qt 6.2
and 6.7 are checked.

Alexandre Janniaux authored 7 months ago and François Cartegnie committed 7 months ago

Now that the interfaces are not run from the main thread directly, it's
possible to dispatch the QApplication startup and UI creation to the
main thread directly on Darwin. It matches the behaviour of the
interface on other platform that Darwin, where a thread is created to
handle the interface, except that the thread is already existing as the
main thread for Darwin.

It means that there won't need any specific code (except for Thread()
startup) to handle the specificities of the Qt interface startup on
Darwin anymore.

Ticket #27103 showed issues when leaving the interface with the
medialibrary, because the path releasing the medialibrary was not taken
correctly, but other issues could arise, including deadlocks.

Fixes #27103

Alexandre Janniaux authored 7 months ago and François Cartegnie committed 7 months ago

Interfaces from libvlc were started from the main thread directly. But
UI inserting their own event loop inside the main runloop like Qt were
taking over the interface insertion. It was preventing other interfaces
to load correctly, required dedicated code to handle the startup in the
Qt interface and it was not possible to exit the interface correctly.

By moving the creation of the interfaces in a dedicated thread, we can
ensure all UI are created and destroyed correctly, but the macosx
interface must now dispatch its own creation into the main thread to
match Cocoa requirement on main thread usage for UI components.

Note that libvlc_new itself must be moved to the intf_queue given that
it starts additional extraintf.

The macosx interface must now start its own NSApp inside the main loop,
via a dispatch call, which simplify the main loop into running the
default main loop.

Alexandre Janniaux authored 7 months ago and François Cartegnie committed 7 months ago

VLA are not compatible with blocks, and error handling is pretty simple
here, so there's no reason not to use dynamic allocation instead.

Includes our UWP patch and some CVE fixes from 3.7.4 and 3.7.5.

François Cartegnie authored 7 months ago and Steve Lhomme committed 7 months ago

refs #28799