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

[Martin Thomson <[e-mail]>]

On 14 January 2014 23:26, Magnus Westerlund
<[e-mail]> wrote:
> I definitely wants more feedback on this. But, I think I am not
> disagreeing with your view. Most of the video codecs that there exist
> RTP payload formats today can use SLI and RPSI. RPSI don't make sense
> for Motion JPEG and the video/raw or video/SMPTE292M that don't use
> reference structures and SLI should be possible to use with also the RAW
> formats so that has even wider applicability.

Those formats that don't need SLI/RPSI can just ignore this guidance,
or you can add a specific exception clause if you feel magnanimous,
i.e., "The SLI and RPSI feedback messages MUST be supported and acted
upon for video streams unless the codec does not support the concept
of macroblocks or reference frames (respectively)."  Or a statement to
that effect.

Basically, SLI and RPSI make sense for the codecs we actually care
about, so let's require them.  If people want to do raw video, then
they are either smart enough to know that SLI/RPSI can't help them, or
nothing an RFC says will help fix their issues.