On 07/11/2012 02:47 PM, Måns Rullgård wrote: > Justin Ruggles <[email protected]> writes: > >> On 07/11/2012 01:14 PM, Justin Ruggles wrote: >>> The information provided by compute_pkt_fields2() is all related to >>> timestamps. >>> --- >>> libavformat/utils.c | 20 ++++++++++++-------- >>> 1 files changed, 12 insertions(+), 8 deletions(-) >> >> The mxf FATE tests fail with this patch because the muxer sets >> AVFMT_NOTIMESTAMPS but actually uses timestamps if available. It just >> has a fallback for when they're not. I don't really like this >> behavior... Thoughts? > > Fix it? What happens if that flag is removed from mxf?
It will reject timestamps that are non-monotonic even though they're not actually coded so it doesn't really matter. We could downgrade it to AVFMT_TS_NONSTRICT, which would at least allow non-increasing timestamps. The packet timestamps and duration are used in ff_audio_rechunk_interleave(). -Justin _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
