- Apr 11, 2016
-
-
Hugo Beauzée-Luyssen authored
-
- Mar 11, 2016
-
-
Hugo Beauzée-Luyssen authored
-
- Feb 12, 2016
-
-
Signed-off-by:
Hugo Beauzée-Luyssen <hugo@beauzee.fr>
-
- Jan 28, 2016
-
-
Hugo Beauzée-Luyssen authored
-
Hugo Beauzée-Luyssen authored
-
- Jan 27, 2016
-
-
Hugo Beauzée-Luyssen authored
-
- Dec 21, 2015
-
-
Hugo Beauzée-Luyssen authored
-
Hugo Beauzée-Luyssen authored
-
- Dec 15, 2015
-
-
Signed-off-by:
Hugo Beauzée-Luyssen <hugo@beauzee.fr>
-
- Nov 25, 2015
-
-
Signed-off-by:
Hugo Beauzée-Luyssen <hugo@beauzee.fr>
-
- Nov 24, 2015
-
-
Hugo Beauzée-Luyssen authored
-
Hugo Beauzée-Luyssen authored
-
Hugo Beauzée-Luyssen authored
-
- Nov 20, 2015
-
-
Hugo Beauzée-Luyssen authored
Since callbacks are now shared among instances, we don't need to store them as a shared_ptr. Though to be fair, I'm not sure we ever needed it since we always use the raw pointer.
-
Hugo Beauzée-Luyssen authored
-
Hugo Beauzée-Luyssen authored
-
Hugo Beauzée-Luyssen authored
This fixes an issue where having a callback set on a specific object would make an application crash. For instance, having 2 VLC::Instance objects and calling logSet on one would cause the wrapper libvlc_instance_t to invoke this callback. If the instance on which logSet was called gets destroyed, the object wrapping the callback will be destroyed as well, thus causing the next log callback invocation to crash. All callbacks are now shared among objects that wrap the same underlying libvlc_xxx_t object. In addition to that, we now ensure the wrapped vlc object gets destroyed before our callbacks wrappers.
-
- Sep 28, 2015
-
-
Hugo Beauzée-Luyssen authored
-
Hugo Beauzée-Luyssen authored
-
Hugo Beauzée-Luyssen authored
-
- Sep 17, 2015
-
-
Hugo Beauzée-Luyssen authored
-
- Sep 15, 2015
-
-
Signed-off-by:
Hugo Beauzée-Luyssen <hugo@beauzee.fr>
-
- Aug 21, 2015
-
-
Signed-off-by:
Hugo Beauzée-Luyssen <hugo@beauzee.fr>
-
Hugo Beauzée-Luyssen authored
-
- Jul 30, 2015
-
-
Hugo Beauzée-Luyssen authored
We don't want to hold a copy of the event manager, but rather a reference. The C++/CX type is intented to be solely a wrapper to the actual type. Furthermore, if due to C++/CX being garbage collected, holding a copy means the EventManagerCX destructor might be called after MediaPlayerCX's destructor, since it only decreases the EventManagerCX's refcount. Calling EventManagerCX's destructor without a valid libvlc_media_player_t (ie. VLC::MediaPlayer being deleted) will cause undefined behavior. By storing a reference, we ensure we just forward the events registrations to the actual VLC::EventManager, and that the GC won't bite us when releasing resources.
-
- Jul 29, 2015
-
-
Signed-off-by:
Hugo Beauzée-Luyssen <hugo@beauzee.fr>
-
- Jul 21, 2015
-
-
Signed-off-by:
Hugo Beauzée-Luyssen <hugo@beauzee.fr>
-
- Jun 23, 2015
-
-
the SwapChain layout dimensions are passed via private data in the DXGI swap chain
-
- Jun 16, 2015
-
-
Jean-Baptiste Kempf authored
-
Signed-off-by:
Hugo Beauzée-Luyssen <hugo@beauzee.fr>
-
Signed-off-by:
Hugo Beauzée-Luyssen <hugo@beauzee.fr>
-
Useful, if not using the system headers Signed-off-by:
Hugo Beauzée-Luyssen <hugo@beauzee.fr>
-
- Jun 15, 2015
-
-
Signed-off-by:
Hugo Beauzée-Luyssen <hugo@beauzee.fr>
-
- May 30, 2015
-
-
Signed-off-by:
Hugo Beauzée-Luyssen <hugo@beauzee.fr>
-
- May 20, 2015
-
-
Hugo Beauzée-Luyssen authored
-
Hugo Beauzée-Luyssen authored
An exception might be thrown by emplace_back
-
Hugo Beauzée-Luyssen authored
vector::emplace_back might throw and prevent libvlc_audio_output_device_list_release from being called
-
Hugo Beauzée-Luyssen authored
-
Hugo Beauzée-Luyssen authored
This reverts commit 9d2e6edb. This makes more sense, since most of libvlc_*_release functions will either assert or assume that the given pointer is non null. Since shared_ptr will call the deleter for any pointer (including nullptr), we're better of refusing to wrap a pointer and avoid undefined behavior
-
Hugo Beauzée-Luyssen authored
This reverts commit 8e8c1672. This feels too much like a bad idea.
-