On 9/28/2016 8:09 PM, James Almer wrote: > On 9/28/2016 7:35 PM, Michael Niedermayer wrote: >> On Wed, Sep 28, 2016 at 07:23:37PM -0300, James Almer wrote: >>> Numerical values for channel layout can be confusing and unintuitive, >>> especially >>> when no channel count is printed. >> >> doesnt this break the format in libavformat/hashenc.c >> for format_version=2 ? > > Yeah, it would break both versions 1 and 2, assuming anything tries to > parse the hex value for some purpose, as both are expected to have it. > And for that matter it would also break parsing framecrc in a similar > fashion.
Then again, framecrc doesn't report any kind of version, so a standardized output shouldn't be expected. Ok if i make this version 3 for framehash/framemd5 then? I can also keep the hex value, so something like #channel_layout 0: 0x3 (stereo) would be printed. > > This can be solved with a new version 3. But that only covers the > muxers framehash and framemd5, and not framecrc as used by most fate > tests, which also calls ff_framehash_write_header(), the function that > prints this line. > > What would be the best course of action here? > _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
