Re: [rtcweb] Friday Agenda: Re: Friday Call details for signaling discussion

[Harald Alvestrand <[e-mail]>]

On 10/20/11 12:45, Iñaki Baz Castillo wrote:
> 2011/10/20 Magnus Westerlund<[e-mail]>:
>> The proposed agenda for Firday is as follows:
>>
>> 10 min introduction from each signaling proposal:
>> - draft-jennings-rtcweb-signaling-00 (ROAP) Cullen
>> - draft-partha-rtcweb-signaling-00 (Standard signaling protocol) Partha
>> - ? (No Protocol) ?
>>
>> In the above only clarifying questions may occur.
>>
>> 20 min discussion of each proposal
>>
>> 30 min concluding discussion
>>
>>  From my perspective both draft-beck-rtcweb-alt-ic-00 and
>> draft-ibc-rtcweb-sip-websocket-00 are relevant documents to the
>> discussion as they provides useful proposals on how interconnect and SIP
>> interop respectively can be done. But as they aren't proposals for how
>> the actual signaling solution should work. Thus these are homework but
>> don't get presentation time.
> Hi Magnus. Honestly I don't consider that discussing about
> draft-ibc-rtcweb-sip-websocket-00 should take place. It's just a
> suggestion about a signaling protocol in RTCweb (in this case pure SIP
> over WebSocket). It's not my aim that the WG considers such spec as a
> standard signaling for RTCweb. Well, this is basically the same you
> have said :)
>
> In the other said, I'd really would like that, before the meeting, all
> the folks could take some time to read:
>
>    http://public.aliax.net/RTCweb_Signaling_Components.html
> and
>    http://www.ietf.org/mail-archive/web/rtcweb/current/msg02158.html
>
> This means not wasting time if somebody proposes ROAP as a "default
> signaling protocol" because ROAP is not that and cannot do that (more
> info in the given links). Also, given the general confusion when the
> term "signaling" appears in this WG, I've tryied to clarify its
> meaning(s) in RTCweb context (first link).
>
> For those who advocate for a "default signaling protocol", I hope
> second link should make them to re-think about what such erroneous
> decision would entail in current and *existing* WWW world.
While I may quibble with Inaki about the terminology he proposes, I 
think I support what he's driving at in the "components" Web page.

In the -overview-02 draft section 7, the term "media negotiations" is 
used for what he calls "media signalling"; it is trying to get at the 
same distinction.

In his category of "info signalling", I would quibble a bit in that I 
see "channel-related signalling" (a term I just invented, describing 
which media is sent on which SSRC on which port) as distinct from 
"identity-related signalling", where the whole question of "who am I 
talking to" is relevant.

I believe "channel-related signalling" needs to be treated together with 
"media signalling".