- Feb 08, 2025
-
-
Signed-off-by:
Claudio Cambra <developer@claudiocambra.com>
-
Signed-off-by:
Claudio Cambra <developer@claudiocambra.com>
-
macosx: Implement standard pasteboard filename handling as playqueue items in file drag recognising view Signed-off-by:
Claudio Cambra <developer@claudiocambra.com>
-
Signed-off-by:
Claudio Cambra <developer@claudiocambra.com>
-
Signed-off-by:
Claudio Cambra <developer@claudiocambra.com>
-
Signed-off-by:
Claudio Cambra <developer@claudiocambra.com>
-
Signed-off-by:
Claudio Cambra <developer@claudiocambra.com>
-
Signed-off-by:
Claudio Cambra <developer@claudiocambra.com>
-
Signed-off-by:
Claudio Cambra <developer@claudiocambra.com>
-
Signed-off-by:
Claudio Cambra <developer@claudiocambra.com>
-
Signed-off-by:
Claudio Cambra <developer@claudiocambra.com>
-
Signed-off-by:
Claudio Cambra <developer@claudiocambra.com>
-
Signed-off-by:
Claudio Cambra <developer@claudiocambra.com>
-
Given playback was stopped, many issues happened when the playback was resumed from the main controls bar. The media time/progress slider wasn't visible anymore and app crashed unexpectdly when picture in picture was activated. This was caused by the `_currentMedia` var that was set to `nil` when playback is stopped but never reassigned when playback was resumed. This change will reassign the var with the current played back input item from the vlc player when the `currentMedia` getter is called.
-
The wrong method parameter type was used for the PIP activation method. These changes are fixing that. The notification associated `VLCPlayerController` object is also used directly in the method implementation instead of accessing it through a global object.
-
- Feb 07, 2025
-
-
Not needed since decoders are now drained.
-
This was a temporary commit, to not break bisect while fixing dvdnav.
-
-
-
SPUs packets might be beyond the final reference frame sectors (vobu_*ref_ea), causing them to be skipped if we drain the es_out immediately after the last I-frame. By preferring vobu_ea when it is nonzero, we ensure SPUs to be properly displayed. cf. fdb3c736
-
Wait for the es_out to be empty/drained in order to avoid flushing past video/subtitles. This also fixes an undefined behavior (that could lead to no audio) when DemuxForceStill() was called but the PCR was not reset (cf. the other call to DemuxForceStill() not showing in this diff).
-
Temporary commit that restore the behavior from 0dd8096f. This allows demuxers to check if empty without draining the decoders. This behavior i undefined as this could lead to audio underrun. This is needed to no break bisect while fixing the dvdnav module. This will be reverted just after.
-
but return if the call was successful.
-
This control is called after EOF, no need to stop buffering.
-
Call EsOutDrain(), that is only stopping buffering for now. Therefore, ES_OUT_IS_EMPTY can be called from input.c when EOF.
-
Check if the es_out is empty without touching the buffering state.
-
Into EsOutDrain() and EsOutIsEmpty() EsOutIsEmpty() must be called while buffering.
-
To make it more explicit.
-
Because it is actually stopping buffering and often used to for that purpose. This will be modified to drain after stopping the buffering in a upcoming commit, since all users also except this behavior.
-
-
When `::destroyMainInterface()` already gets called before, `::unloadGUI()` gets called again as well and with the recent introduction of `m_quickView->setSource(QUrl());` there without first checking for null pointer it causes null pointer dereference as `::destroyMainInterface()` already destroys the `::m_quickView` instance. At the same time, calling `::destroyMainInterface()` during destruction causes the same null pointer dereference, since if initialization fails there is no `::m_quickView` instance. This is a case with Windows 7 where direct composition is not available, which causes crash at start.
-
Steve Lhomme authored
No need to translate to UTF-8 first (which is also not used as such later).
-
Steve Lhomme authored
So we don't have to release it manually. This should be backported as Qt5 with QT_USE_QSTRINGBUILDER doesn't like the + in onUpdate(). For some reason Qt6 is OK with that.
-
Signed-off-by:
Claudio Cambra <developer@claudiocambra.com>
-
Signed-off-by:
Claudio Cambra <developer@claudiocambra.com>
-
macosx: Move all the connect/disconnect related properties and method to the abstract media library segment view controller These methods are not useful for the media source view controller Signed-off-by:
Claudio Cambra <developer@claudiocambra.com>
-
macosx: Use VLCMediaSbtractMediaLibrarySegmentViewController for all media library segment view controllers Signed-off-by:
Claudio Cambra <developer@claudiocambra.com>
-
Signed-off-by:
Claudio Cambra <developer@claudiocambra.com>
-
- Feb 06, 2025
-
-
Thomas Guillem authored
Fixes the following assert when seeking more than one time past the end of the file. src/audio_output/dec.c:1107: vlc_aout_stream_Drain: Assertion `!atomic_load_explicit(&stream->drained, memory_order_relaxed)' failed. Calling `pw_stream_flush(s->stream, true)`, when empty, leads to undefined behavior and might cause the `stream_drained()` callback to be called in loop (leading to the mentioned assert after a flush).
-
Thomas Guillem authored
Remove the not up to date pi_allowed_chromas. Use directly the vlc_chroma_probe.h API to get the list of chroma fallbacks. Remove recursivity handling: Before, the chain module could load an other chain module for conversions requiring an extra step (OPAQUE -> I420 -> RV24 for example). Now that the conversion list is acquired from start, we can add each converter filters in an iterative loop.
-