Re: [MMUSIC] Proposal for what bundle should say about demux

[Colin Perkins <[e-mail]>]

On 27 May 2013, at 20:37, Paul Kyzivat wrote:
...
> I agree there is no *need* to relax the PT uniqueness requirement in order to map the packets (or sources if you prefer) to m-lines.
> 
> But (at least with plan A) there is a concern that there will be insufficient PTs to keep them unique per-m-line.
> 
> What I am saying is that if you do the mapping to m-line *before* the mapping from PT to payload format, then you *can* relax the PT uniqueness requirement.

That would mean using signalled SSRC values to map from an RTP source to an m= line. Once you've done that, though, it's not clear that you need that many unique payload types (since if you're using the same payload format in two m= lines you can use the same payload type in both, when using a signalled SSRC to distinguish the m= lines). 

> This isn't violating precedent, because there is no precedent. Neither the problem nor the solution are present without bundling.


There's precedent in that the RTP specifications assume the mapping is fixed in a single RTP session. 



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