Re: [rtcweb] Agenda time request for draft-marjou-rtcweb-audio-codecs-for-interop-01

[Adam Roach <[e-mail]>]

On 3/14/13 17:09, Andrew Allen wrote:
> Koen is right that there are many more obstacles to RTCweb and legacy network interop than just a common codec. First there will need to be RTCweb signaling gateways to interface between the RTCweb signaling and the legacy networks (SIP, H.323 etc) and there will need to be in place mechanisms for peering, federation and address resolution between networks as well as service agreements in place between the players.


We had this mostly working at the SIPit in Boston, using the SIP over 
Websockets spec. The only reason we weren't actually setting calls up 
was that we couldn't find any clients there that supported both ICE and 
DTLS-SRTP. What's neat is that this isn't even really "gatewaying" in 
the way that you mean it -- all you need is a SIP proxy that supports 
one more (easy-to-implement) transport protocol.

And once you hit that proxy, all of the peering, federation, and address 
resolution solutions developed for SIP apply.

I don't think this points to a need to push AMR into the web browsers, 
mind you. To take Justin's earlier point a few steps further: if we're 
going on number of shipping units, the mere existence of Chrome and 
Firefox's WebRTC implementation means that the number of deployed Opus 
codecs far overwhelms the number of deployed AMR codecs. Appeals to the 
size of codec deployment would more reasonably reach the conclusion that 
Opus should be MTI for the next 3GPP release. ;-)

/a