Re: [MMUSIC] New Version Notification for draft-ietf-mmusic-rfc4566bis-29.txt

[Paul Kyzivat <[e-mail]>]

Colin,

On 6/15/18 8:06 AM, Colin Perkins wrote:

>>>> 1) Section 5.14 - Media Descriptions
>>>> The 6th paragraph talks about use of the rtcp attribute with 
>>>> hieararchically encoded streams, however AFAIK, the rtcp attribute 
>>>> does not support. If it does, then I think the explanation needs to 
>>>> be elaborated on (basically, you would need to have more than one of 
>>>> them I assume).
>>>
>>> This is text from 4566. I don't recall the discussions leading to 
>>> that text, and I don't really understand what is intended here. Is 
>>> RTP/RTCP intended to be an instance of a hierarchically encoded 
>>> stream, or is that considered a distinct usage? What is another 
>>> example of hierarchically encoded streams that should apply here.
>>>
>>> It is interesting that the examples here all use port/2 for RTP 
>>> streams, but that is clearly not current practice.
>> Maybe one of Colin, Magnus or Jonthan can chime in ? 
> 
> Hierarchically encoded streams, in this context, is layered coding. The 
> draft gives an example:
> 
>            m=video 49170/2 RTP/AVP 31
> 
>        would specify that ports 49170 and 49171 form one RTP/RTCP pair
>        and 49172 and 49173 form the second RTP/RTCP pair.  RTP/AVP is the
>        transport protocol and 31 is the format (see below).  If non-
>        contiguous ports are required, they must be signalled using a
>        separate attribute (for example, "a=rtcp:" as defined in
>        [RFC3605]).
> 
> The /2 /does not/ refer to the port used for RTCP, it refers to the 
> number of RTP streams in the layered stream.

Is this form used in practice? I had the impression that hierarchically 
encoded streams were using separate m= lines for each layer.

And Flemming pointed out to me that a=rtcp isn't defined so it it has 
meaning with such an m= line. That would limit its usage further.

Should this /n usage be deprecated?

	Thanks,
	Paul