Fatih Uzunoğlu authored 6 months ago and Felix Paul Kühne committed 5 months ago

In small screen mode, playlist pane acts as an overlay. However, the
items that are stationed behind can not be seen since the background
directly reflects the window background in order to provide backdrop
blur effect.

In order to prevent confusion, the background is made opaque.

Alexandre Janniaux authored 11 months ago and Felix Paul Kühne committed 5 months ago

Now that we can express module dependencies directly, there's no need to
remove the test if some of the modules are not present. In the future,
we'll be able to inject this module list to mark the test as skipped
instead of disabling it.

Alexandre Janniaux authored 11 months ago and Felix Paul Kühne committed 5 months ago

This declares the modules unconditionnally to be able to use them in
the module_depends: key for tests, so as to disable the tests when the
matching module is disabled automatically and ensure the tests is built
and run after the module has been built.

Alexandre Janniaux authored 5 months ago and Felix Paul Kühne committed 5 months ago

We want to actually enable those frameworks when the condition is true.

Alexandre Janniaux authored 5 months ago and Felix Paul Kühne committed 5 months ago

Copy-paste typo between QuartzCore and CoreText.

Fatih Uzunoğlu authored 6 months ago and Steve Lhomme committed 5 months ago

For internal drag operations, it is possible to start the
drag before the input items are resolved because it is
feasible to wait for the input items to be resolved on
drop.

However, for external drag operations, the mime data
needs to be set before drop as we don't have the luxury
to wait for mime data on drop. In fact, even if there
was such a chance we would still need to make sure that
the mime data is set as soon as the drag starts because
the receiver application might do actions based on drag
enter event and the accompanied mime data.

Since it is not possible to know whether a drag operation
is an internal or external operation, we should pre-
emptively set the mime data before the drag operation
starts.

It's always the same as the source format.
Sometimes it had the rotation of vd->source applied and sometimes not.
In both cases it's used to get a rotated version of vd->source.

When opening a display module, vd->source is the same as the output video format.

They will likely not change but the value in vd->source is always the
most up to date.

The AWindowHandler_setVideoLayout() is the same with or without crop.

The value is valid when we get a SAR event.

onNewVideoLayout() in libvlcjni [1] will apply the values regardless
if the SAR is passed or not.

[1] https://code.videolan.org/videolan/libvlcjni/-/blob/5076ef767eb4126c74a496033ace54649ac200e7/libvlc/src/main/java/org/videolan/libvlc/VideoHelper.java#L323

If it's 0/0 then the code is equivalent.
If it has an actual crop value it's valid even during crop.

onNewVideoLayout() in libvlcjni [1] will apply the values regardless
if the crop is passed or not. And it will discard it if one SAR value is 0.

[1] https://code.videolan.org/videolan/libvlcjni/-/blob/5076ef767eb4126c74a496033ace54649ac200e7/libvlc/src/main/java/org/videolan/libvlc/VideoHelper.java#L323

VOUT_DISPLAY_CHANGE_SOURCE_ASPECT only changes the SAR values
VOUT_DISPLAY_CHANGE_SOURCE_CROP only changes offset+visible values
But the values always valid during each call.

Alexandre Janniaux authored 5 months ago and Felix Paul Kühne committed 5 months ago

Fixes a memory leak with extension manager not being unloaded when the
application exits. Created from the following stacktrace:

    generic_start
    vlc_module_load
    module_need
    -[VLCExtensionsManager loadExtensions]
    -[VLCExtensionsManager buildMenu:]
    -[VLCMainMenu awakeFromNib]

Alexandre Janniaux authored 5 months ago and Felix Paul Kühne committed 5 months ago

Stacktrace from the memory graph:

    _malloc_zone_calloc_instrumented_or_legacy
    vlc_ml_media_t* CreateAndConvert<vlc_ml_media_t, medialibrary::IMedia>(medialibrary::IMedia const*)
    MediaLibrary::Get(int, char*)
    vlc_ml_get
    lc_ml_get_media_by_mrl
    +[VLCMediaLibraryMediaItem mediaItemForURL:]
    -[VLCControlsBarCommon updateCurrentItemDisplayControls:]
    -[VLCMainVideoViewControlsBar update]
    -[VLCMainVideoViewController viewDidLoad]
    -[NSViewController _sendViewDidLoad]
    -[NSViewController _loadViewIfRequired]
    __24-[NSViewController view]_block_invoke
    NSPerformVisuallyAtomicChange
    -[NSViewController view]
    -[VLCLibraryWindow awakeFromNib]
    ...

  This release includes new codec interfaces, compression efficiency and
  perceptual improvements, speedup and memory optimizations and many bug
  fixes. This release is ABI compatible with the last release.
  The definitions of the internal macros AOM_INLINE and AOM_FORCE_INLINE
  have been removed from the public header aom/aom_integer.h.
  - New Features
    * New codec controls:
      * AV1E_SET_AUTO_TILES
      * AV1E_GET_HIGH_MOTION_CONTENT_SCREEN_RTC
      * AV1E_SET_POSTENCODE_DROP_RTC: Post encode frame drop feature.
      * AV1E_SET_MAX_CONSEC_FRAME_DROP_MS_CBR
    * New key-value pair for aom_codec_set_option():
      * "auto-tiles": equivalent to the new codec control
        AV1E_SET_AUTO_TILES.
  - Deprecated Features
    * Deprecated codec control:
      * AV1E_SET_MAX_CONSEC_FRAME_DROP_CBR: Use the new codec control
        AV1E_SET_MAX_CONSEC_FRAME_DROP_MS_CBR instead.
    * The sframe_mode field in the aom_codec_enc_cfg_t struct is not
      implemented.
  - Compression Efficiency Improvements
    * BD-rate gain of 0.7 - 1.3% (by enabling global motion tool) for
      speed 5 and speed 6 with ~5% encode time increase.
    * RTC speed 11 video: ~3-5% BD-rate gain for VGA and QVGA.
  - Perceptual Quality Improvements
    * RTC quality improvements for slide changes and scrolling content.
  - Speedup and Memory Optimizations
    * RTC screen content speedups:
      * ~2x speedup for high motion content for speed 11.
      * ~2x speedup on key frame coding for speed >= 10.
    * Arm: Significant uplifts in speed in this release (vs v3.9.1) have
      come from tuning the various convolutions according to filter size
      (doing 8-tap when only 2-tap is required is inefficient) and also
      deploying Armv8.6 USMMLA instructions in 6-tap and 12-tap standard
      bitdepth convolutions.
      * Standard bitdepth RTC:
        * speed 5: +5%
        * speed 6: +4%
        * speed 7: +5%
        * speed 8: +4%
        * speed 9: +6%
        * speed 10: +6%
      * Standard bitdepth VoD:
        * speed 0: +9%
        * speed 1: +12%
        * speed 2: +9%
        * speed 3: +3%
        * speed 4: +3%
        * speed 5: -9% (expected due to global motion changes)
        * speed 6: -3% (expected due to global motion changes)
      * High bitdepth VoD:
        * speed 0: +4%
        * speed 1: +19%
        * speed 2: +23%
        * speed 3: +1%
        * speed 4: +1%
        * speed 5: -8% (expected due to global motion changes)
        * speed 6: -3% (expected due to global motion changes)
      * Standard bitdepth 2x1 horizontal super-resolution/scaling
        encoding: +101%
  - Other Improvements
    * Reduce bit rate overshoot on slide content.
  - Bug Fixes
    * rtc: Bug fix for active_maps with sb_size=128.
    * b:343429036: rtc: Fix source_sad setting near boundary.
    * Fix to QP for temporal enhancement after key frame.
    * b:343429192: rtc: Condition QP adjustment on rc->q_1/2_frame > 0.

  See:
  https://aomedia.googlesource.com/aom/+/refs/tags/v3.10.0/CHANGELOG

Fatih Uzunoğlu authored 5 months ago and Felix Paul Kühne committed 5 months ago

If player is visible, it should not push player again.

Steve Lhomme authored 5 months ago and Felix Paul Kühne committed 5 months ago

We don't have to patch it for Android anymore.

Pierre Lamot authored 5 months ago and Felix Paul Kühne committed 5 months ago

errors are handled in the MessageDialog and in the Qml

Pierre Lamot authored 5 months ago and Felix Paul Kühne committed 5 months ago

b_select is redundant with the ActionFlag

Pierre Lamot authored 5 months ago and Felix Paul Kühne committed 5 months ago

no need to specify it explicitly

Pierre Lamot authored 5 months ago and Felix Paul Kühne committed 5 months ago

this allows triggering the model without creating the dialog

Fatih Uzunoğlu authored 6 months ago and Felix Paul Kühne committed 5 months ago

- Currently style handling is handled in a really fragile
  manner. The order of keys provided by `QStyleFactory`
  is expected to directly match the combobox. At the
  same time, if the style is not overridden, then there
  should be no need to save this in the settings.
- Although the first entry is the default style, it does
  not provide what the style is. Now, the style itself is
  printed as well.

Fatih Uzunoğlu authored 6 months ago and Felix Paul Kühne committed 5 months ago

When the parent top-level window receive a drag or drop
event, Qt does not propagate the event through the child
windows.

All handling is done in `QQuickWindow`, so if I manually
forward the events to the quick window, drag and drop
works as expected.

Fatih Uzunoğlu authored 6 months ago and Felix Paul Kühne committed 5 months ago

It is documented that the static function `QQuickWindow::graphicsApi()` may
not yield the ultimate graphics API used by the scene graph and that the way
to get the "true, real result" is to wait until the scene graph is initialized.

However, at the same time, it is said that the result would be the "graphics
API that would be used by the scene graph if it was initialized at this point
in time". This is what is really necessary to know here, as we don't (and
should not) override the graphics API in compositors, rather the compositor
should tell if it is compatible with the graphics API. If the compositor is
not compatible (e.g. using QSGRendererInterface::OpenGL with compositor
dcomp), then it should fail at the earliest convenience so that the application
can try other compositors.

As of Qt 6.7, Qt Quick does not implement a fallback procedure in case the
initialization goes wrong due to the chosen graphics API that it tries a
different graphics API. In other words, Qt Quick does not have a cross-graphics
API fallback procedure unlike us. And the cross-graphics API fallback that
we use gets engaged earlier, so it can be assumed with certainty that the
graphics API would not change by the time scene graph is initialized.

Waiting for scene graph initialization has also become less relevant since
0dbfb4b9, where RHI probing and `QQuickWindow::setGraphicsApi()` became
being used in almost all cases. If we set the graphics API explicitly,
it means that the same API can be considered by the compositor directly
without getting concerned by scene graph initialization and it getting end
up with a different graphics API.

Pierre Lamot authored 6 months ago and Felix Paul Kühne committed 5 months ago

extension_dialog_t is not a registered type in qt so the signal was failing.

It's usually dangerous to pass VLC C object in Qt signals regarding the lifetime
of these objects, so rather than registering the type, the object is passed
though lambda capture assumming that the current comment and code is correct.

Fatih Uzunoğlu authored 6 months ago and Felix Paul Kühne committed 5 months ago

When last vout stops, a dummy vout is created.
For the interface, this has no relevance and
"has video" property should be set to false.

Tristan Matthews authored 5 months ago and Felix Paul Kühne committed 5 months ago

When using var_GetBool, "--sout-aom-row-mt" was a noop.

Tristan Matthews authored 5 months ago and Felix Paul Kühne committed 5 months ago

This will output all spatial layers if requested, defaults to false (the built-in dav1d
behavior would defaults this to true).

Fix suggested-by Maryla Ustarroz

Fixes #28776