Hi guys,

I read the draft and I think that it does not discuss the difference in RTCP handling  between RTCP terminating mixer and source projecting mixers or translators. It looks to me that the text addresses only the second topologies.

If the mixer uses its own SSRCs it does not show all the behaviors described in section 5.1. My understanding of RFC3550 is that for this case the mixer does not forward the sender and receiver reports, only the SDES in order to identify SSRC collision and the BYE messages so there will be less messages than in the case of translators or source projecting mixers.

Another comment is that for source projecting mixers in my view, the document provides motivation to use multi sessions over single transport to reduce the number of RTC messages.

Roni Even