Re: [AVT] [Fwd: I-D Action:draft-petithuguenin-avt-multiple-clock-rates-00.txt]

[Colin Perkins <csp at csperkins.org>]

Marc,

On 5 Oct 2009, at 17:19, Marc Petit-Huguenin wrote:
> It seems that there was some interest at the time to have this  
> problem fixed, but I didn't receive any comment or suggestion since  
> I published this I-D.  At a minimum I would be interested in  
> documenting what existing RTP implementations are doing in a  
> multiple clock rate scenario, especially on the sender side.


Looks like a reasonable summary of the issues.

What's your recommendation for preferred sender behaviour? I'm leaning  
towards requiring that the SSRC changes if the timestamp rate changes,  
since that seems to have the least corner cases with RTCP behaviour  
and synchronisation, and the backwards compatibility issues don't seem  
that serious.

It might be possible to eliminate the lip-synchronisation delay if the  
sender were to generate a compound RTCP packet containing two SRs, one  
for each SSRC, at the change-over, with each SR containing a mapping  
of its RTP timestamp to the same NTP time.

--
Colin Perkins
http://csperkins.org/