Re: [rmcat] Do we need a generic feedback message, should we work on this now?

[Karen Elisabeth Egede Nielsen <[e-mail]>]

Hi Colin, All

The idea was indeed to do exactly what you propose Colin.
I.e., to collect requirements and then design a feedback message
within the RTP framework.

This was exactly what the chair slides said (or aimed to say ..).
Please see the last slide of
http://www.ietf.org/proceedings/94/slides/slides-94-rmcat-0.pdf.

There was no reference to any current proposal in the discussion in RMCAT wg
session on Friday.

BR, Karen

> -----Original Message-----
> From: rmcat [mailto:[e-mail]] On Behalf Of Colin Perkins
> Sent: 6. november 2015 13:55
> To: Michael Welzl <[e-mail]>
> Cc: rmcat WG <[e-mail]>
> Subject: Re: [rmcat] Do we need a generic feedback message, should we
> work on this now?
>
> Michael,
>
> If we’re to define a common feedback packet format, then it should a)
> convey the information needed by the various congestion control algorithm
> candidates; and b) match the RTP architecture, since it is providing RTP
> congestion control. I don’t believe the current proposal fits with the RTP
> framework, and it’s not clear that it conveys the right information, since
> there hasn’t been a requirements analysis of the feedback needed by the
> different algorithms.
>
> I don’t object to a common feedback packet format, although I’m perhaps
> less convinced of its benefits, but I do object to pushing forward with a
> proposal, such as that discussed in the meeting, that’s inconsistent with
> the
> way RTP feedback works.
>
> Colin
>
>
> > On 6 Nov 2015, at 18:54, Michael Welzl <[e-mail]> wrote:
> >
> > Hi,
> >
> > I just watched the RMCAT recording at
> >
> http://recs.conf.meetecho.com/Playout/watch.jsp?recording=IETF94_RMCA
> T
> > _II&chapter=chapter_1 and I can't help but be amazed at the discussion
> > of the generic feedback message.
> >
> > I strongly support doing this work NOW because delaying has the risk of
> never getting it done: code will be shipped, and then has to be
> retro-fitted
> with a generic feedback message... and then we may end up with an
> incentive problem too (why support this standard, when MY system works?).
> >
> > Given all the work that the RTCWEB WG has put into agreeing on common
> codecs I honestly think it's ridiculous to even imagine that we in RMCAT
> may
> end up having senders and receivers that are not be able to talk to each
> other because we don't have a common feedback message and no
> agreement on what should be on the sender or receiver side.
> >
> > At the risk of being a smart-ass, I can't help but repeat that I did say
> > these
> things at the very first RMCAT meeting, in November 2012:
> > http://www.ietf.org/proceedings/85/slides/slides-85-rmcat-0.pdf
> > (slides
> 2-7)
> >
> > I do believe we should have agreed on sender vs. receiver - based
> > control
> and on a common feedback message before even doing anything else.
> > Okay, this is all voluntary work and there's no point complaining or
> > just
> demanding certain work to be done, I understand that .... but: now that we
> have a proposal for a common feedback message, I absolutely think we
> should move ahead with this, try to come to a decision about sender vs.
> receiver-based control, and align the existing congestion controls with it
> ASAP.
> >
> > Cheers,
> > Michael
> >
>
>
>
> --
> Colin Perkins
> https://csperkins.org/
>
>
>