Hi, On Wed, Mar 25, 2009 at 1:33 AM, Michael Broughton <Michael_Broughton at advanis.ca> wrote: > > > Interesting. So this was a bug introduced by the fix for ticket #744? > I think so, it may need to handle such special cases (e.g: your case as described below). In your "special" case, it didn't detect the frame seq restart because the difference is not really large enough, in other words, the sender seemed to restart the RTP seq and timestamp oftenly and jitter buffer treated the frame as a late/duplicated frame, while normally seq & timestamp should be restarted only when the values reach its maximum value (around max of 16bit unsigned for seq & max of 32bit unsigned for timestamp). The log shows that within a minute, the sender restarted the seq (and timestamp) twice, restarting once is fine, but twice is rather strange. However, possibility of infinite loop is really a problem, so fix is needed. > I don't mean to pester you, but can you provide an rough ETA for a fix? I > mean, will it take days or weeks? I need to decide if I am going to attempt > a fix myself. > I am afraid that all I can say now is by default, any fixes should be available in the next release (normally new release is launched in every 6 weeks). Regards, nanang
