- From: Harald Alvestrand <harald@alvestrand.no>
- Date: Thu, 07 Feb 2013 20:30:49 +0100
- To: public-webrtc@w3.org
On 02/07/2013 04:24 AM, Martin Thomson wrote: > On 6 February 2013 21:54, Stefan Hakansson LK > <stefan.lk.hakansson@ericsson.com> wrote: >>> The legacy case is somewhat sticky: I don't know what >>> (MediaStreamTrack) identifiers are attached to those streams. What is >>> most interesting is the case where msid attributes arrive after the >>> first packets, which is possible...maybe. >> I think the browser could not do anything sensible until it gets the answer. >> If it has msid data in it, then it can assemble the MediaStream(s) >> accordingly. If there is no msid info it would have to do the default thing >> (which I think currently is to create one MediaStream object with tracks for >> all RTP streams in it). > This is, of course, the exact path we followed to conclude that we > needed dummy streams. I'll try to find some time to pick derf's brain > on what his ideas are here. The only things I could come up with are > a little stinky, but no doubt he has something far better than > anything I could come up with. > Wasn't the track I heard followed in Lyon; the track I heard followed in Lyon involved making sure there was a video m= line with adequate candidates in the answer from Bob's JS that is sent before Bob's JS knows that Bob is giving permission to use his camera. I may have missed some tracks.
Received on Thursday, 7 February 2013 19:31:39 UTC