Re: [rtcweb] Regarding Federation Arguments

[Harald Alvestrand <[e-mail]>]

On 11/09/2011 05:55 PM, Wolfgang Beck wrote:
> On Wed, Nov 9, 2011 at 1:37 PM, Harald Alvestrand<[e-mail]>  wrote:
>> [on Bernard's comments]
>> can we simply say "Existing protocols (for example SIP or XMPP) could be
>> used between the servers"?
>>
>> between servers, while either a standards-based or proprietary protocol
>> could be used between the browser and the web server.
> My concern about this trapezoid style connection of servers is this:
> in order to limit the number protocol translators,
> RTCWEB providers will use only a small number of server-to-server
> protocols, most likely SIP and Jingle.
>
> This will inevitably have influence on the JS client  / server
> protocol design: why bother with innovative features that
> don't translate well to SIP and Jingle anyway? Why not just do
> Jingle/SIP over Websocket? And the opportunity of RTCWEB will be gone.
> We will just see browser-based SIP, Jingle clients with fancy CSS
> themes.
Please!!!!!!!!!!!

The trapezoid is ONE possible deployment scenario.
As the overview document says:

    A commonly imagined model of deployment is the one depicted below.

It is neither the only possible one nor even a recommended one. It is 
just commonly imagined.

If you want innovative applications ..... go innovate, and come back 
when you know *exactly why* something in the specs gets in your way!

> This has happened with SIP. As most of the interconnection is done
> through the PSTN, we don't see many exciting new SIP applications in
> wide use.
>
> It's a pity that the VWRAP people are no longer active in the IETF.
> They really had a fresh view on those issues. Their
> interconnection requirements where just to complex for the trapezoid
> and nobody of them had a limiting telephony background.
Without pointers to docs, I can't evaluate this statement. Be specific!