RE: Getting rid of RTP

Inaki said: 

"I hope we can discuss about encoders/decoders without having to also mention QUIC."

[BA] Definitely. In ORTC, the RtpSender/RtpReceiver is distinct from the transport.  I think what is being suggested here is a more granular pipeline so as to be able to distinguish encoding from encapsulation and decoding from decapsulation.  Controlling transport(s) separately would remain. 

"ORTC can already handle forking, trickle ICE and TURN mobility."

[BA] Yes, but something like multi-path ICE would be trickier.  Not sure how we can guarantee that a design would be compatible with a version of ICE that doesn't yet exist though.

Received on Monday, 5 March 2018 22:42:25 UTC