[MMUSIC] RTSP timeout issue

[RÃmi Denis-Courmont <rem at videolan.org>]

	Hello,

I suppose this has been discussed many times over the past decade, but 
anyway.

One thing I keep wondering, and never came to a satisfactorily safe 
conclusion is, "who decides" how to refresh an RTSP session, so that it 
does not timeout.

The specs (both 1.0 and current draft of 2.0) say the client can send 
either a dummy RTSP request with the Session-ID to the server, or send 
RTCP-RR. Hmm, so the client decides then... But but, what if the server 
specifies a nul bandwidth for RTCP-RR in the SDP? Is it so that the 
client MUST use RTSP-level refreshes, or that no timing out is used at 
all?

I have no strong opinions on using either mechanism, but shouldn't 
RTSP/2.0 mandate one of them specifically (and if RTCP is selected, 
that the RR bandwidth specification MUST be "compatible"...), as well 
as a minimum timeout that the server shall apply? Or then mandate 
ICE-style Binding Indication.


Then of course there is the side issue, of proving that the 
client "owns" the IP address it receives data on - especially when a 
proxy is used. I guess nobody wants to require congestion control (-> 
DCCP) support usage at this point.

Regards,

-- 
RÃmi Denis-Courmont
http://www.remlab.net/

_______________________________________________
mmusic mailing list
mmusic at ietf.org
https://www1.ietf.org/mailman/listinfo/mmusic