Re: [MMUSIC] Offer/Answer PT Questions

[Roman Shpount <[e-mail]>]

On Thu, Feb 25, 2016 at 5:06 AM, Christer Holmberg <
[e-mail]> wrote:

> Not necessarily. Even if PT is per direction is allowed as such, we still
> need to clarify whether the same PT/codec binding is per direction, or
> whether it applies to the whole session.
>

Yes, this needs to be clarified, but general consensus is that each
direction is independent.


> > The complication here is that Alice will not be allowed to reuse PT 111
> for AMR(mode-set=0,2,5,7) in the answer. Alice will need to pick a
> > different PT for AMR(mode-set=0,2,5,7), if it is supported and end
> points will use asymmetric payload types.
>
> So, you are saying that Alice is not allowed to reuse a PT value that
> Alice has previously used for another PT/codec binding - no matter if Bob
> supported the codec or not?
>
> For example: assume Alice offered PT 111: VP8, but Bob did not include VP8
> in the answer. Then Bob offers PT 111:AMR. Now, Alice is not allowed to use
> PT 111 for AMR, since Alice previously used it for VP8.
>

This is exactly what I am saying. The fact that Bob did not include VP8 in
the answer does not mean it is not sending VP8 to Alice. Because of this
Alice is not allowed to use PT 111 for AMR since there can be VP8 media in
flight with PT 111.

What I am trying to highlight here that if an end point is reusing a PT
which was used in one direction only, it is forcing asymmetric PT
assignment. This, of cause, breaks the SHOULD from section 6.1 in RFC 3264:

In the case of RTP, if a particular codec was referenced with a
specific payload type number in the offer, that same payload type
number SHOULD be used for that codec in the answer.


Regards,
_____________
Roman Shpount