Alexandre Janniaux authored 1 year ago and Steve Lhomme committed 11 months ago

Automatically enable the headers, but detect the currently available
header version with cppcheck since the pkgconfig file will actually
detect the version of the loader.

cppcheck cannot understand the full header version, which was also not
available before 1.3 anyway. Moreover, the major and minor versions are
not defined directly, but instead either through cast or not at all
(before 1.3), so we use the header guards defines and the
VK_HEADER_VERSION for the last patch revision number, which directly
matches the actual package revision.

v1.3.239 is the version provided by the vulkan-loader package on debian
bookworm, and the main constraint we have is the presence of the
vulkan/vulkan_beta.h header which should be present and the presence or
absence of the old/new video_decode extension that was promoted. 238 is
the first revision removing the previous video_decode extension and
adding the new one.

Tested on linux with headers present and android with system header
being obsolete.

Alexandre Janniaux authored 1 year ago and Steve Lhomme committed 11 months ago

libplacebo depends directly on vulkan-headers, not transitionally.
The constraint was written like this to not always install the headers
if some were provided by the system.

This is the latest in the 6.x branch.
We already build with 6.1 in the Debian unstable Docker.

This is the latest in the 5.x branch.

Fatih Uzunoğlu authored 1 year ago and Steve Lhomme committed 11 months ago

spirv-opt provided by spirv-tools is used by
Qt Shader Builder (qsb) for shader optimization.

Alexandre Janniaux authored 11 months ago and François Cartegnie committed 11 months ago

Since the video output cannot be the master clock, there's no reason to
artificially change the date matching with the displayed frame. Indeed,
the video output was using two points that felt weird:

 - When the picture was late, it was "forcing" the timestamp by using
   VLC_TICK_MAX as system time.

 - When the picture was rendered before the deadline, it was using
   system_pts, ie. the date at which the clock computed when it should
   be rendered, as the system time for the clock.

In both cases, it was basically only disabling drift computation in
traces, since the drift is not used in the video output so changing its
value doesn't change the behaviour and the clock points will never be
used in a master clock. Drift would always be zero on the traces.

This commit re-establish the drift computation, using the time at which
the frame is finished being displayed as system time. It won't match the
"exact" time since it doesn't account display latency nor supports using
dates from VSYNC (ref #26184) but the error will almost always be a
constant part (for the latency not being used) plus a bounded part (for
the VSYNC sampling delay), meaning that every non-stable behaviour like
drifting can be measured.

Reference also commit 207fcd92 which
tried to apply that to the clock wait.

Reference commit abb6e135 which asserted
that drift should be 0 given that the whole sampling process is
controlled by the clock, which is only true if the vout_display doesn't
stall the video_output.

Reference commit 013aa3aa which started
to use VLC_TICK_MAX (INT64_MAX back then) so that the clock update would
notify the player.

Important: The intent of the commit above is disabled by this commit.

Note that this commit virtually introduce a huge initial drift. This
drift is actually expected given that we draw the first frame regardless
of the actual date, and playback constraints might have the video track
earlier at the beginning than audio track.

This also introduce increasing drift when a frame is re-drawn, which is
also by design since we didn't update the frame. A motion-interpolating
filter would generate intermediate frames and avoid this, which makes it
more representative of the true video output drift.

Steve Lhomme authored 11 months ago and François Cartegnie committed 11 months ago

Otherwise we carry over some scaling rounding in the original region, even if nothing was changed.

Steve Lhomme authored 1 year ago and François Cartegnie committed 11 months ago

This reverts commit 372e62c0.

This commit broke SPU positioning for this DVD:
https://streams.videolan.org/samples/MPEG-VOB/menus%20DVD/Avcd%2023032b

Prefixes and namespaces are required to be able
to correctly parse multiple namespaces documents
or simply prefixed documents

UDP is unreliable

Alexandre Janniaux authored 1 year ago and François Cartegnie committed 11 months ago

Forward the tracing key/value entries as an array instead of using
va_list. Using an array is easier to interop with in other languages,
but also and more importantly much easier to store and forward.

This will allow to implement hierarchical tracing context.

Alexandre Janniaux authored 11 months ago and François Cartegnie committed 11 months ago

The module was added in 4d691734.

Alexandre Janniaux authored 11 months ago and François Cartegnie committed 11 months ago

The file was added in 7c9f7725 to
provide common code between h264 and h265 handling.

See:
https://opus-codec.org/demo/opus-1.5/

Was not inserted if there's no pre-frame NAL units

Claudio Cambra authored 1 year ago and François Cartegnie committed 11 months ago

macosx: Set VLCLibraryWindow XIB constraint for path control view to top of superview as a placeholder

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

Claudio Cambra authored 1 year ago and François Cartegnie committed 11 months ago

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

Claudio Cambra authored 1 year ago and François Cartegnie committed 11 months ago

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

Claudio Cambra authored 1 year ago and François Cartegnie committed 11 months ago

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

Marvin Scholz authored 1 year ago and François Cartegnie committed 11 months ago

The linkers platform_version argument does not support
the value "macosx" but expects "macos" here instead.

Ash V authored 1 year ago and François Cartegnie committed 11 months ago

Added isADir() method in NetWorkmediaModel's anonymous namespace to list folders-before-files through the NetWorkmediaModel's getSortFunction().

fix: #28371

Alexandre Janniaux authored 1 year ago and Jean-Baptiste Kempf committed 1 year ago

Video using the ffmpeg qtrle decoder are failing to play with:

    [qtrle @ 0x76117804fc00] Unsupported colorspace: 0 bits/sample?
    [0000761178024760] avcodec decoder error: cannot start codec (qtrle)
    [0000761178024760] main decoder error: Codec `rle ' (Apple QuickTime RLE Video) is not supported.
    [0000761178024760] main decoder error: Codec not supported
    [0000761178024760] main decoder error: VLC could not decode the format "rle " (Apple QuickTime RLE Video)

Indeed, QTRLE format can compress source with either 1, 2, 4, 8, 16, 24
or 32 bpp, and the actual bpp is stored in the stst atom, not in the
decoder payload. The demux has to forward this value to the decoder one
way or another.

Regression from 12030f5ccac852aaa3f9f66b20606d2fc344f8f4, which starts
setting the bpp to zero in avcodec because there is no chroma yet.
Defining a chroma instead of forwarding the bpp would be possible for 8,
16, 24 RLE data cases, but would not work for lower RLE data cases.

The correct i_bits_per_pixel field being removed in commit
1d46d8b7, a hack is implemented instead.
The hack is from 3e5f2e5d and other
places that needed it.

Fixes #28550