Re: [rtcweb] Clarification on simulcast and RID and RepairedRtpStreamId

[Magnus Westerlund <[e-mail]>]

On 2018-12-03 11:11, Sergio Garcia Murillo wrote:
On 03/12/2018 11:02, Magnus Westerlund wrote:
To attempt to restate your question. For RTP sessions, where there are a
single source media stream per Media Description in SDP (m= block) and
where there is explicit MID signalling and SSRC signalling, then one
will not need RepairedStreamId and can instead rely on explicit MID
signalling for the RTX stream?

I think no. I think using RepairedStreamId independent if there is
multiple source RTP streams or not per media description provides a more
consistent experience

How would that work, if we are not using simulcast, no rid will negotiated in the SDP (AFAIK) so we can't use RepairedStreamId and we will have to rely on mid header extension and pt multiplexing, right?


That is my point, that we should define that when using RFC 4588 RTX one should include RID and use the RepairedStreamID for dynamic bindings of the SSRCs even if one doesn't use Simulcast. That would be a stable and commonly operational solution.

The solution that Iñaki suggested still only works in a sub-set of conditions and you will still need to implement the dynamic late binding solution that is described in RFC4588. That will be the ultimate fallback.

Cheers

Magnus Westerlund

----------------------------------------------------------------------
Network Architecture & Protocols, Ericsson Research
----------------------------------------------------------------------
Ericsson AB                 | Phone  [phone]
Torshamnsgatan 23           | Mobile [phone]
SE-164 80 Stockholm, Sweden | mailto: [e-mail]<mailto:[e-mail]>
----------------------------------------------------------------------