On Mon, 24 Oct 2016 12:18:07 -0700 Aman Gupta <ffm...@tmm1.net> wrote:
> On Sun, Oct 23, 2016 at 11:20 PM, Aman Gupta <a...@tmm1.net> wrote: > > > > > > > On Sunday, October 23, 2016, Richard Kern <ker...@gmail.com> wrote: > > > >> > >> > On Oct 19, 2016, at 8:45 PM, Aman Gupta <ffm...@tmm1.net> wrote: > >> > > >> > From: Aman Gupta <a...@tmm1.net> > >> > > >> > ff_videotoolbox_avcc_extradata_create() was never being called if > >> > avctx->extradata_size==0, even though the function does not need or use > >> > the avctx->extradata. > >> > >> Could this be a bug in another part of the code? It seems like extradata > >> should be set. > > > > > > Yes it definitely could be. I can verify if extradata is present on macOS > > where the same sample and ffmpeg version worked. > > > > Looks like missing extradata has nothing to do with the platform, but > rather only happens when you skip calling avformat_find_stream_info(). > > I am skipping the find_stream_info() call to reduce time spent probing and > begin playback more quickly, since the initial avformat_open_input() is > enough to discover available streams on mpegts containers. > > That means this issue is not as widespread as I initially thought, but I > still think it's an optimization worth making. That makes sense. I think a decoder shouldn't really rely on this pseudo extradata that libavformat constructs. That this code is run only with extradata_size set makes no sense at least in the h264 code and was an oversight. _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
