- From: Iñaki Baz Castillo <ibc@aliax.net>
- Date: Sat, 16 Apr 2016 13:25:40 +0200
- To: Randell Jesup <randell-ietf@jesup.org>
- Cc: "public-webrtc@w3.org" <public-webrtc@w3.org>
2016-04-15 16:26 GMT+02:00 Randell Jesup <randell-ietf@jesup.org>: > On 4/15/2016 7:33 AM, Iñaki Baz Castillo wrote: >> >> 2016-04-14 20:45 GMT+02:00 Cullen Jennings <fluffy@iii.ca>: >>> >>> The sender can change the SSRC whenever it becomes aware of a SSRC >>> collisions and the receiver has to deal with that. It's one of the many >>> reasons for use of identifiers other than SSRC (like rid) >> >> SSRC is a 32 bit long integer. There are no SSRC collisions. And if >> there were, there could also be RID or MID collisions. > > > SSRC collisions happen in real life. Some devices like to use non-random > SSRCs (0 is a good value, right? Or 1? ;-) ), and they can end up > colliding. Or they seed poorly. Or just bad luck. If a device uses a fixed SSRC with value 0 or 1 I would not expect that it will implement RID or MID. And if it did that, it would set "" for both values XD BTW: If an endpoint decides to change a SSRC, is a SDP re-offer needed by indicating the new SSRC in use? (regardless it is not needed if RID/MID is in use and so on...). -- Iñaki Baz Castillo <ibc@aliax.net>
Received on Saturday, 16 April 2016 11:26:28 UTC