Re: [rtcweb] [MMUSIC] Quick comments on draft-roach-rtcweb-glareless-add-00

On 5/22/13 1:56 PM, Harald Alvestrand wrote:
> On 05/14/2013 11:17 AM, Emil Ivov wrote:
>>
>> On 14.05.13, 11:24, Stefan Håkansson LK wrote:
>>> On 2013-05-13 19:58, Bernard Aboba wrote:
>>>>
>>>> On May 12, 2013, at 10:18, "Stefan Håkansson LK"
>>>> <stefan.lk.hakansson@ericsson.com> wrote:
>>>>
>>>>>> I would hope there will normally not be SDP changes for either of
>>>>>> those events.
>>>>> I would like to avoid them too, but I'm not sure how.
>>>> [BA] It is probably not possible to avoid them in all cases, but we
>>>> can define RTCP functionality which along with some set of
>>>> functionality (e.g. Simulcast, layered coding) will minimize them.
>>>> This is worth doing (and soon).
>>> I agree to this!
>> RTCP `certainly sounds a lot better than SDP O/A
>>
>> Emil
>>
> If something's doable doing app-level signalling, it can be deployed at
> the speed of app updates - but each app has to do it on its own.
>
> If something's doable using RTCP, it can be deployed at the speed of
> browser updates - but you have to convince the browser people to do it
> first.
>
> SDP O/A extensions are in an uncomfortable place between these extremes,
> I think. But it's not obvious to me that one of the three always "sounds
> better" than the other ones.

I think that having things like pause/resume signaling (and perhaps 
resolution and the like) in RTCP would be better than SDP, not least 
because you escape the session related state model for this kind of 
signaling (which I can see can be frequent in scenarios when you want to 
save transmission). And I think that standardized signaling it preferred 
since the functionality enabled would then be easily available also to 
more naive developers.

But, I think what you prefer to a large extent depends on your 
viewpoint, and also what scenarios you consider, so I could agree that 
it is not obvious that one way would always be better than the other ones.

There was a discussion on RTCP vs SDP signaling at an earlier IETF 
meeting, and the group at that time preferred SDP signaling. But I don't 
think a concrete proposal on how to do pause request/indication, resume 
request/indication etc. was ever put together (there are hints in the 
draft-juberti-plan-00). Perhaps that should be the first thing to do?

>
>
>
>
> _______________________________________________
> rtcweb mailing list
> rtcweb@ietf.org
> https://www.ietf.org/mailman/listinfo/rtcweb
>