Re: [MMUSIC] RID simplification re: PT

[Magnus Westerlund <[e-mail]>]

Hi Peter,

If the offerer includes PT=*, then I think it is reasonable to require 
that to continue to be PT=* in the answer. The reason is that this RID 
is stating that these limitations apply to all Payload Types present in 
the m= section. Changing that to a sub-set of the PTs in the answer 
appears to introduce the complexity I thought was the whole goal here to 
avoid.

If one introduces a RID that has a sub-set of all PTs, then I think it 
is straightforward to remove rejected PTs, that is basic logic in 
contructing messages, and the implication on the offer side when 
receiving the answer is clear. I can't send codec Foo, thus the 
limitations doesn't matter.

If we go with your less restrictive rule, then the offerer has to 
understand what it means on a session level to not get the limitations 
it intended on both A and B codec to only apply to one of them, and the 
other becomes unrestricted in this RID. I personally will not have to 
implement this, but people that has to do, should think about what they 
think becomes most straightforward. I do note that the implementations 
always will have to deal with a RID being refused by the answerer.

So to be clear I don't have a strong preference between these two 
alternatives I can go with either.

Cheers

Magnus


Den 2015-12-17 kl. 20:31, skrev Peter Thatcher:
> I am also fine with the proposal, including Magnus's tweak of allowing
> removal of PTs via the m= line and removing RIDs that have an empty set
> of PTs.
>
>
> However, having thought about it some more, I think the following rule
> would be more simple and suffice those implementers seeking sensibility:
>   "If the offer a=rid line contains pt=*, the corresponding answer a=rid
> line must contain pt=*".
>
> In other words, if the offer has "pt=A,B", the answer *can* remove A or
> B on a per-rid level with "pt=A" or "pt=B", but if an offerer does not
> want to support reduction of payload types, it can simply put in "pt=*".
> ​  This allows more control of a=rid lines in the answer than the
> originally proposed rule (in the case where an implementation decides to
> support pt= other than pt=*), but it still allows implementations the
> sensible ease of putting "pt=*" in the offer and not allowing reduction,
> but with a more simple rule.
>
> Would that also work?  Or am I missing something?  If I am and that rule
> doesn't work, I still support the original proposal (with Magnus's tweak).
>
>
> And by the way, does the order of the PTs in a RID line even matter?
> What effect would the answer reordering the PTs even have?


-- 

Magnus Westerlund

----------------------------------------------------------------------
Services, Media and Network features, Ericsson Research EAB/TXM
----------------------------------------------------------------------
Ericsson AB                 | Phone  [phone]
Färögatan 6                 | Mobile [phone]
SE-164 80 Stockholm, Sweden | mailto: [e-mail]
----------------------------------------------------------------------