- From: Emil Ivov <emcho@jitsi.org>
- Date: Sun, 13 Apr 2014 12:08:27 +0200
- To: Robin Raymond <robin@hookflash.com>
- Cc: "public-ortc@w3.org" <public-ortc@w3.org>
- Message-ID: <CAPvvaaJDZ-rSwQuVzU2VKbADMgsYh1x6fb6oZg39-37-m4K_bw@mail.gmail.com>
Hey Robin, On Sat, Apr 12, 2014 at 7:18 PM, Robin Raymond <robin@hookflash.com> wrote: > > Hi Emil, > > The browser may or may not know when it will deliver the final candidate. > That's why I put in the "caveat". > Right. Seems I missed that. > Sometimes it won't know because the browser discovers the pending network > operation for a relay candidate is for a redundant server reflexive > candidate and thus is not a valid candidate. Other times it may not know > because of a failure to allocate or discover a relay candidate. It is a > matter of operation timing. > > However, under many scenarios the browser fully knows that this it has > gathered the final candidate so delivering that candidate so it doesn't > need to fire a distinct event later. Delivering that candidate with the > "end of gathered candidates" flag as part of remote signaling is important > to avoid an extra needless remote signaling event for the sole purpose of > indicating the final gathered candidate has arrived. In the case where the > browser discovers that it has no more candidates to gather too late, > there's not much that can be done except fire the final "onlocalcandidates" > event with the "end of gathered candidates" flag set to true. > OK. I think the importance of avoiding an extra signalling message that says "it's over" might be a bit exaggerated: it doesn't delay connection establishment, in the worst case it would delay failures only marginally (but that would be a very rare situation) and the incurred extra bandwidth use is basically just an IP+UDP/TCP header. That said, I don't see any problems arising from this so, if you and others find it convenient .... :) Emil -- https://jitsi.org
Received on Sunday, 13 April 2014 10:09:15 UTC