VLC Media Player

Recent Commits

• François Cartegnie(08 Jul 26)

  demux: avi: fix embedded sub segfault refs #29972 Reported-by: HE WEI (ギカク) <[email protected]>

• François Cartegnie(08 Jul 26)

  demux: ogg: fix skeleton read fix #29959 Reported-by: Tristan Madani <[email protected]>

• François Cartegnie(27 Apr 26)

  codec: cvdsub: check block payload size when rendering refs #29863

• François Cartegnie(27 Apr 26)

  codec: cvdsub: refactor RLE unpacking refs #29864

• François Cartegnie(01 Apr 26)

  codec: png: use checked arithmetic for alloc refs #29719

• François Cartegnie(01 Apr 26)

  codec: jpeg: use checked arithmetic for alloc refs #29719

• Alexandre Janniaux(08 Aug 26)

  test: transcode: check the output order on drain The previous commit fixed the regression from commit d8c883e273f333d7e2d8c2c2df1e35b280021df5 which made drained video frames being mixed up ahead of the frames coming right before. This new test stamps the frames and check that everything is coming in order to avoid this issue. To reproduce the issue, the test first generates regular frames and then makes the transcode drain, leading to the held and encoder drain queues. The decoder releases the pictures it was holding, which are encoded and left pending in the transcode output, and the encoder then releases the frames it was holding for itself when it gets drained in turn. seq: 0 1 2 3 | 4 5 6 7 8 | 9 10 11 -------- ----------- -------- regular held frames drained frames frames To ensure this order, the frame sequence number and test stage are saved into each picture. This test heavily relies on the fact that stream output pipeline is processed synchronously by the input and not queued at the input level.

• Alexandre Janniaux(08 Aug 26)

  transcode: fix the encoded end-of-stream picture order The drained pictures and those that were outputs right before drained but not already sent over were inverted when sent to the downstream stream output component. It put the end of the stream ahead of the blocks that were before it. Regression from d8c883e273f333d7e2d8c2c2df1e35b280021df5.

• Claudio Cambra(13 Aug 26)

  macosx: Fix infinite load on empty removable storages Signed-off-by: Claudio Cambra <[email protected]>

• Claudio Cambra(12 Aug 26)

  macosx: Only dump child cache and reload data when preparse change notification does not target currently-viewed folder Signed-off-by: Claudio Cambra <[email protected]>

• Claudio Cambra(12 Aug 26)

  macosx: Acquire input node child counts on separate dispatch queue (and cache them) Signed-off-by: Claudio Cambra <[email protected]>

• Claudio Cambra(12 Aug 26)

  macosx: Add childCountForInputNode to media source types Signed-off-by: Claudio Cambra <[email protected]>

• Claudio Cambra(11 Aug 26)

  macosx: Handle preparse state on a per-item basis Signed-off-by: Claudio Cambra <[email protected]>

• Claudio Cambra(11 Aug 26)

  macosx: Check specifically for suffix in status notifier view notifications Signed-off-by: Claudio Cambra <[email protected]>

• Claudio Cambra(11 Aug 26)

  macosx: Notify on in-tree preparse end Signed-off-by: Claudio Cambra <[email protected]>

• Claudio Cambra(11 Aug 26)

  macosx: Add preparse meta-info to media source notification emission Signed-off-by: Claudio Cambra <[email protected]>

• Serhii Bykov(10 Aug 26)

  macos: dispatch media source reloads to the main queue # Conflicts: # modules/gui/macosx/library/media-source/VLCMediaSourceBaseDataSource.m

• Steve Lhomme(13 Aug 26)

  contrib: vulkan-headers: remove old headers before building Older headers can lead to build issues.

• Steve Lhomme(13 Aug 26)

  contrib: vulkan*: do not require CMake 3.22 We currently require CMake 3.21. And it builds fine with CMake 3.21. It was just bumped to match SPIRV [^1] [^2]. [^1]: https://github.com/KhronosGroup/Vulkan-Headers/commit/234c4b7370a8ea3239a214c9e871e4b17c89f4ab [^2]: https://github.com/KhronosGroup/Vulkan-Loader/commit/b05d038c3348a8c8d4024abe84c7411b3068ebe7

• Serhii Bykov(11 Aug 26)

  macos: use rounded style for empty library Browse button

• François Cartegnie(12 Jan 26)

  demux: mp4: fix iTun reads refs #29536

• François Cartegnie(19 Mar 26)

  demux: mp4: move APV code to proper handler

• François Cartegnie(19 Mar 26)

  demux: mp4: fix APV invalid deref refs #29677

• Claudio Cambra(12 Aug 26)

  macosx: Ensure thumbnail is correctly acquired in controls bars and decorative views for ml items Signed-off-by: Claudio Cambra <[email protected]>

• Claudio Cambra(12 Aug 26)

  macosx: Ensure expected media item title is presented in playqueue and controls bars Signed-off-by: Claudio Cambra <[email protected]>

• Steve Lhomme(15 Aug 26)

  Revert "CI: use a single job with a matrix for Wasm" For this optimization to work, it required that the script to move the tarball knew about the group name in addition to the job name. But Gitlab doesn't provide this. And we can't really tweak the job name when using grouped jobs... This reverts commit 48428f8375227a68902871774df1c4b793c95cde.

• Steve Lhomme(11 Aug 26)

  demux: aiff: do not try to read a file if seeking is not possible The AIFF parser parses the file until the end (skipping the chunks data) and then seeks to the beginning of the SSND chunk. If seeking is not possible we can't do that.

• Steve Lhomme(31 Jul 26)

  demux: aiff: make i_ssnd_pos unsigned It's either -1 (unset) or comes from vlc_stream_Tell() which gives a uint64_t. As a consequence i_ssnd_start and i_ssnd_end and also unsigned.

• Steve Lhomme(11 Aug 26)

  demux: aiff: use more local variables in Open()

• Steve Lhomme(31 Jul 26)

  demux: aiff: make i_ssnd_offset a uint32_t It's read as an unsigned 32-bit value.