On Wed, Jun 29, 2011 at 15:51, aviad rozenhek <[email protected]> wrote: > On Tue, Jun 28, 2011 at 13:43, aviad rozenhek <[email protected]> wrote: >> >> On Mon, Jun 27, 2011 at 00:41, Luca Barbato <[email protected]> wrote: >>> >>> On 6/24/11 4:38 AM, Måns Rullgård wrote: >>>> >>>> Luca Barbato<[email protected]> writes: >>>> >>>>> If the Continuity Check fails drop the whole packet >>>> >>>> That's a bit harsh. There may still be useful data in the parts >>>> present. >>> >>> The current patch makes it return the incomplete frame and resync if I >>> read it correctly. >>> >>>> Consider the situation of a lost TS packet near the end of a frame (not >>>> necessarily end of a PES packet). The decoder will probably skip to the >>>> next frame header and resume decoding normally there. By dropping the >>>> entire PES packet, the damage is propagated until the _next_ frame >>>> starts. >>> >>> I'm not sure about it, let's test with the following adaptation, Aviad >>> could you provide the samples or try yourself? >>> >>> lu >> >> I'm not sure I understand what you mean regarding the adaptation, but the >> sample [that crashes without the patch] can be accessed from here >> http://www.datafilehost.com/download-cb99c0d7.html >> additionally, the original patch you proposed seems to cause problem with >> some samples [which dont feature any actual data loss], to the point they >> dont ffplay anymore, >> i've also uploaded them: >> http://www.datafilehost.com/download-16ab1765.html >> http://www.datafilehost.com/download-af0b2237.html >> ps. datafilehost is relatively painless and requires no login. I would >> have uploaded to your favorite ftp site had it been available. sorry in >> advance. >> > > I think the continuity check might be wrong, as it reports continuity check > failures even for files which playback perfectly. the two samples I posted > above show this problem > http://www.datafilehost.com/download-16ab1765.html > http://www.datafilehost.com/download-af0b2237.html
Actually, already read_packet() returns corrupted data for some reason. When I dumped the raw data, the continuity errors were there. -- Jindrich Makovicka _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
