rtpdec_vp8: Don't trim too much data from broken frames
authorMartin Storsjö <martin@martin.st>
Wed, 9 Jan 2013 12:25:59 +0000 (14:25 +0200)
committerMartin Storsjö <martin@martin.st>
Thu, 10 Jan 2013 07:43:01 +0000 (09:43 +0200)
commit76c40fbef06aaf5013450a4b51dd3c49955777a1
treef9fbb8b8f7c0b63f620e1f3eb84504dc9a2b0ff8
parent3b366c3aa0800d2df9f0678f6f4d7791b8adf2a9
rtpdec_vp8: Don't trim too much data from broken frames

Previously, for broken frames, we only returned the first partition
of the frame (we would append all the received packets to the packet
buffer, then set pkt->size to the size of the first partition, since
the rest of the frame could have lost data inbetween) - now instead
return the full buffered data we have, but don't append anything more
to the buffer after the lost packet discontinuity. Decoding the
truncated packet should hopefully get better quality than trimming out
everything after the first partition.

Signed-off-by: Martin Storsjö <martin@martin.st>
libavformat/rtpdec_vp8.c