…d return D3DFMT_A8R8G8B8 instead. libass returns the glyphs in a 8bit monochrome format, which must be expanded to the alpha channel of the 32bits D3DFMT_A8R8G8B8 format. The rest of the overlay code assumes the data is in the alpha channel, the others don't matter.
…with python. Maybe we should deinstall XBMC completely (keep userdata) before installing? (fixes #11860)
…took the precompiled dll from the rtmpdump homepage and made a package out of it.
…epeat*) set of built in commands. If present, the user will be notified of the new repeat state. e.g. XBMC.PlayerControl(Repeat,Notify) Partially fixes #11902
…m*) built in commands. If present, the user will be notified of the new shuffle state. e.g. XBMC.PlayerControl(Random,Notify) Partially fixes #11902
… Airplay is using 2 persistent tcp connections. The second one is a reversed HTTP connection for sending events from XBMC to the airplay device. Without this the play button of the device doesn't get noticed about play/pause state. Implemented the reverse HTTP connection for making it more airplay compliant as of reverse engineering. Hey i think we are the first project doing this ;) [clean] - make the xml consistent (line endings \r\n) and make them better visible in wireshark by using \t
…correct fps report Original patch by Voyager-xbmc <email@example.com>
…g the virtual nodes (genre,studio and the likes) - only titles node will show the sets item
…sic setting if the video one is in the gui, the music one should be as well.
…a set with only one movie - thanks Danone-KiD (closes #10965 and #11907)
…of one single movie (closes #10561)
…re decoding is used. The dxva CreateSurface() function is meant for output surfaces (render targets), not inputs.
…te the processor with that format instead of YUV. Former code worked so it seems that the drivers work around an incorrect value but it's better to fix. Maybe that's what confused the Poulsbo? Former code was c/p from Boxee. No idea why they tried NV12 first and fell back on RGB. Their processor render target was always RGB and the dxva blit was followed by a quad copy to the final render target. For a format conversion StretchRect is supposed to be used, not a quad copy AFAIK
… for c37c29e) (thanks to Garbear for the heads up) changed: Cleanup CFileItem::FindTrailer() a bit