- Mar 20, 2025
-
-
-
Alexandre Janniaux authored
Add a unit test checking the behaviour of the input clock. For now, the test mainly setup the infrastructure for testing the input clock and checks some basic fundamentals. Note that the test doesn't pass without the previous commit removing the CR_PCR_GAP hack for DVD, so the test actually still check something that was failing before.
-
Alexandre Janniaux authored
The check was added back then for a special case with DVDs, but it should be moved to the DVD demuxer instead if it is really needed. However, it actually request the system time to have a min value which is almost always the case now, so it's probably not needed anymore.
-
Alexandre Janniaux authored
We can use the internal logger reference now.
-
Alexandre Janniaux authored
The logger is used in the input_clock_Update() function, but it's more suitable to define the logger at the object creation rather than being able to "switch" logger instance at call site.
-
Alexandre Janniaux authored
The structure is already used in the Rust bindings and has as low dependencies as possible to anything. Exposing it allows creating logger that will work in a minimal test environment.
-
Alexandre Janniaux authored
-
Alexandre Janniaux authored
Those functions are currently specific interpretation of the values that are already returned by input_clock_GetState(), ie. computed from system start, stream start, system buffering duration and stream buffering duration as well as the clock pause date or the current date. The rationale for separating the computation in those functions instead of input_clock_GetState is that the latter exposes an internal state of the input_clock buffering code, which prevents modifying it in a different way -- like exposing multiple buffering segments for instance. Exposing the computations by "intent" instead of by "current state" enables abstracting away the buffering from the es_out.
-
- Mar 19, 2025
-
-
Signed-off-by:
Claudio Cambra <developer@claudiocambra.com>
-
Signed-off-by:
Claudio Cambra <developer@claudiocambra.com>
-
- Mar 18, 2025
-
-
Our CI tend to run tests extremely slowly when heavy loaded. Qt tests will timeout by themselves after a global timeout and exit gracefully. As these checks are not supposed to timeout we can just remove the explicit timeout with no penalty for the case where the test runs under normal conditions. fix: #29077
-
- Mar 17, 2025
-
-
-
- Now that size change can be requested from the UI at most once per UI frame rendered, WindowResizer has become less relevant. - !1978 (vout: Use a fair display lock and interrupt wait) already reduced the waiting time considerably. Since the UI uses VSync (except CompositorX11), it should be fine to have blocking stuff that only take couple of milliseconds to execute.
-
Currently the new approach is only enabled for `CompositorWayland` and `CompositorDirectComposition`.
-
So that we don't need to require adjusting the size before the scale.
-
-
As per Qt Quick Effects' `CMakeLists.txt`, the excluded resources are the ones that use masking. We do not use mask, so we don't need to initialize these shaders. This should save some disk space for the static build. We could eliminate more of these, for example depending on the blur level, only a certain subset of these shaders are used. However, I decided to not be that aggressive for the moment.
-
Signed-off-by:
Claudio Cambra <developer@claudiocambra.com>
-
Signed-off-by:
Claudio Cambra <developer@claudiocambra.com>
-
Signed-off-by:
Claudio Cambra <developer@claudiocambra.com>
-
- Mar 16, 2025
-
-
- thread id - thread naming
-
If the internal vlc input item t is null Signed-off-by:
Claudio Cambra <developer@claudiocambra.com>
-
Signed-off-by:
Claudio Cambra <developer@claudiocambra.com>
-
Prevent crashes from accessing null pointer Signed-off-by:
Claudio Cambra <developer@claudiocambra.com>
-
The background color is the same as header color, there is no need to have extra coloring in the header.
-
Having a header color by default does not make sense here: - There is header background color but not background color for the rest. - Views are not controls, and are often placed within controls where background is handled properly. - A purely transparent rectangle does not create a scene graph node thus not rendered. This is considered an optimization for rendering. In this case, since the background is likely to be the same as the background color of the parent control, the area is painted twice with the same color. If the color needs to be different, the property can be overridden to deviate from transparent color.
-
-
The UPnP option was not being properly built with the contribs on Apple platforms.
-
This is because `updateIntfScaleFactor()` no longer emits the signal `intfScaleFactorChanged()` unconditionally.
-
-
- Mar 15, 2025
-
-
As in configure.ac, look for "freerdp2" package.
-
- Mar 14, 2025
-
-
-
- Implicit size is not calculated correctly. - Control's padding is not respected. - Control's spacing is not respected. - Text is not centered when there is more than enough width. - Text is not elided when the control has less than enough width.
-
Sometimes the array elements may be variant (`QVariant`), even though the actual type is a `QUrl` or a `QString`. This is an odd case because `QUrl` is automatically converted to `url` value type when passed to QML. However, the source url may already be provided as a `QVariant`, where in that case this conversion would not occur.
-
Copy elf_aux_info bits from OpenBSD for CPU feature detection on Arm, AArch64 and RISC-V.
-
Signed-off-by:
Claudio Cambra <developer@claudiocambra.com>
-
-
- Mar 13, 2025
-
-
Wrong alignment: GEN util/qmldir GEN widgets/qmldir MOC dialogs/bookmarks/bookmarks.moc.cpp MOC dialogs/dialogs/dialogmodel.moc.cpp UIC dialogs/extended/ui_video_effects.h UIC dialogs/firstrun/ui_firstrunwizard.h
-
-
The output value is bounded by the size parameter, and is used for indexing elements in arrays.
-