[vlc-devel] [vlc-commits] commit: rtsp: implement session timeouts (Pierre Ynard )

Pierre Ynard linkfanel at yahoo.fr
Sat Nov 27 18:10:00 CET 2010


> Le samedi 27 novembre 2010 02:34:27 git at videolan.org, vous avez écrit :
> > vlc | branch: master | Pierre Ynard <linkfanel at yahoo.fr> | Sat Nov 27
> > 01:32:43 2010 +0100| [e59ca64643cb0769467f923667279cb83257e85c] |
> > committer: Pierre Ynard
> > 
> > rtsp: implement session timeouts
> > 
> > Based on keep-alive RTSP requests
> 
> And now if the client uses RTCP-RR instead of RTSP (which the spec
> allows), it will be timed out. Evidently, you don't read the mailing
> list archive because this has been discussed several times already.

What I remember from the previous discussions is vague whining about
the brokenness of VLC/the spec, and you saying that there was little
point in considering RTCP-RR because they were too easy to spoof, so
RTSP keep-alive were necessary. Moreover, I remember too that the
RTSP implementation in VLC uses the SDP line "b=RR:0" which disallows
RTCP-RR, making your point moot.

Anyway, it's easy to change the default to disable timeouts, but then
again do you prefer a secure default, or one that accomodates broken
clients? If a user notices unexpected timeouts, it is easy to find the
problem and disable them with the option.

Also, what is your view on RTCP-RR support, before I possibly start
tackling the task?

-- 
Pierre Ynard
"Une âme dans un corps, c'est comme un dessin sur une feuille de papier."



More information about the vlc-devel mailing list