Re: [MMUSIC] Offer/Answer PT Questions - text proposal

[Colin Perkins <[e-mail]>]

> On 2 Mar 2016, at 22:32, Paul Kyzivat <[e-mail]> wrote:
> 
> On 3/2/16 5:03 PM, Colin Perkins wrote:
>>> On 1 Mar 2016, at 19:06, Paul Kyzivat <[e-mail]> wrote:
>>> On 3/1/16 1:23 PM, Roman Shpount wrote:
>>> 
>>>> What really bugs me about this discussion is that it is extremely
>>>> difficult to determine when RTP session is started or stopped based on
>>>> O/A exchanges. Because of this it is difficult to determine when it is
>>>> safe to reuse PT for something else just by looking at O/A.
>>> 
>>> Well, can we assume that if the addr/port changes at one of the ends then this is to be a new RTP session?
>> 
>> The RTP session is tied to the shared SSRC space. If the same SSRCs are valid after the change, it’s the same RTP session.
> 
> Fine. But how do the participants in the O/A session discover that the SSRC space has changed?

I don’t know - I’d always thought the lifetimes were aligned. That is, the o/a exchange set of a set of parallel RTP sessions (one per media, unless bundle is used) that had lifetime matching the o/a session. 

Perhaps that works, perhaps not. For the simple cases I expect it does work. For cases with middleboxes and multiple participants, I expect things get rapidly complex. But, anything to do with middleboxes and RTP gets complex, because we have multiple topologies many of which don’t really conform to how RTP handles sessions (this is what RFC 7667 tries to sort out).

> In general I assume that the O/A is what tells them this sort of stuff. I guess it can also be learned via RTCP. But if the one end decides to switch to a new RTP session, without signaling this change via O/A, I assume that things will be messed up until a bunch of RTCP stuff has been exchanged. That doesn't seem graceful.
> 
> So, while mechanically it is possible to make such a change without signaling it, in practice it seems like a bad idea.
> 
> But I also don't see how to signal a change in RTP session via O/A. Superficially, it seems like changing the addr/port is *likely* to mean you have a new RTP session. But that isn't *necessarily* true. It could be you are simply changing to a different path to the same target.
> 
> 	Thanks,
> 	Paul
> 



-- 
Colin Perkins
https://csperkins.org/