RE: ICE in Javascript (Re: Microsoft API Proposal)

These concerns (largely unwarranted, but that's no longer relevant) lead us in the new proposal to split the timing-critical parts (implemented by the browser) and the choice of what to do with that information (in Javascript or server-side).

What is proposed is substantially different from what was in my I-D, and I'd encourage basing analysis on the specification itself rather than Martin's or my commentary about it.

Matthew Kaufman
________________________________________
From: Harald Alvestrand [harald@alvestrand.no]
Sent: Tuesday, August 07, 2012 12:11 PM
To: public-webrtc@w3.org
Subject: ICE in Javascript (Re: Microsoft API Proposal)


Martin, the last time the RTCWEB/WEBRTC people discussed whether ICE
should be in Javascript or in the browser machine, the consensus
recorded was, I believe, that we'd do ICE in the browser, and that we
wouldn't consider the question again until someone could demonstrate
(not just theorize) that an ICE machine written in Javascript was able
to satisfy the ICE timing constraints.

Has that demonstration been successful?

                          Harald




Received on Tuesday, 7 August 2012 19:30:42 UTC