Re: [rtcweb] Feedback messages (and problem with RTP usage document)

[Magnus Westerlund <[e-mail]>]

Hi Bernard and Tim,
(As document author)

I have reviewed this issue and have the following comments regarding it.
So, both VP8 and H.265 RTP Payload formats specifies how, but does not
mandate support of RPSI or SLI/SPLI. To my understanding such
requirement would be expressed by a user of the codec, like WebRTC. The
RTP usage document has so far avoided specifying any specifics for a
particular video codec and its payload format. And I would note that the
solution so far allows negotiation of these features, at least as long
as one accept to arrive at symmetric capabilities in send vs receive.

Thus the question is in which direction you think this should be addressed:

A) A generally changing the requirement on RPSI and SLI to enable its
usage with all video codecs (capable of using them) in WebRTC. Raising
the requirements level either to RECOMMENDED or SHALL.

B) Write a draft-ietf-rtcweb-video that has a section saying:
Codec Specific Requirements
---------------------------
- If you implement VP8 in WebRTC also RPSI and SLI SHALL be supported.
- If you implement H.265 in WebRTC also RPSI and SPLI SHALL be supported.

Note that SHALL could be replaced by RECOMMENDED depending on the WGs
consensus for what is appropriate here.

Note that the WG can choose to do both of the above alternative (A and
B) but for example use A) to raise these to RECOMMENDED and use B to
mandate them for specific codecs.

For the H.265 RTP Payload format and the SPLI message we might have
issues with its publication not finishing, less so with VP8, but not
impossible. We might have to deal with more unfinished pointers as we
get closer to the publication request state.

I believe that we need more input from the WG about this question. Both
what an appropriate way forward for this issue is, and what requirements
level people are interested in.

Cheers

Magnus

On 2013-12-30 16:28, Bernard Aboba wrote:
> Tim Panton said: 
> 
>> No - just that if you disable rtcp-fb in a VP8 session, you'll have a
> poor experience - try it in a few wireless environments.
> 
> [BA] Totally agree -- and yet the RTP usage document does not properly
> account for this.  
> 
> draft-ietf-payload-vp8 Section 5 refers to the RPSI and SLI feedback
> messages. 
> draft-ietf-payload-rtp-h265 Section 8 refers to the RPSI and (newly
> defined) SPLI messages. 
> 
> And yet draft-ietf-rtcweb-rtp-usage Section 5.1.3 indicates that support
> for SLI is OPTIONAL while Section 5.1.4 indicates that support for RPSI
> is OPTIONAL.  
> 
> 
> 
> 
> 


-- 

Magnus Westerlund

----------------------------------------------------------------------
Multimedia Technologies, Ericsson Research EAB/TVM
----------------------------------------------------------------------
Ericsson AB                | Phone  [phone]
Färögatan 6                | Mobile [phone]
SE-164 80 Stockholm, Sweden| mailto: [e-mail]
----------------------------------------------------------------------