diff options
author | Anton Khirnov <anton@khirnov.net> | 2023-03-09 16:59:12 +0100 |
---|---|---|
committer | Anton Khirnov <anton@khirnov.net> | 2023-03-20 10:42:09 +0100 |
commit | e6126abc6997058ca49ee596b70611bbe367163e (patch) | |
tree | d291ee8ee6fd8e657264c6c6fdba6c29e73033ff /libavcodec/x86/huffyuvencdsp_init.c | |
parent | 0ad64cdd92a132b57748bd8b84db955e1369f74e (diff) | |
download | ffmpeg-e6126abc6997058ca49ee596b70611bbe367163e.tar.gz |
fftools/ffprobe: stop using AVFrame.pkt_{pos,size}
These fields are ad-hoc and will be deprecated. Use the recently-added
AV_CODEC_FLAG_COPY_OPAQUE to pass arbitrary user data from packets to
frames.
Changes the result of the flcl1905 test, which uses ffprobe to decode
wmav2 with multiple frames per packet. Such packets are handled
internally by calling the decoder's decode callback multiple times,
offsetting the internal packet's data pointer and decreasing its size
after each call. The output pkt_size value before this commit is then
the remaining internal packet size at the time of each internal decode
call.
After this commit, output pkt_size is simply the size of the full packet
submitted by the caller to the decoder. This is more correct, since
internal packets are never seen by the caller and should have no
observable outside effects.
Diffstat (limited to 'libavcodec/x86/huffyuvencdsp_init.c')
0 files changed, 0 insertions, 0 deletions