Re: [rtcweb] Comment on draft-ietf-rtcweb-overview-02 (Browser RTC trapezoid)

["Ravindran Parthasarathi" <[e-mail]>]

Saul,

I have already mentioned in the mailing alias that *Please don't delete the existing mail* because we often miss the original intent of the mail and goes in the circular discussion.

My original reply was intended only to answer Inaki query on how I come to the conclusion that "I'm not favor Inaki solution of SIP over websocket because it is overkill for signaling protocol". 

Thanks
Partha

>-----Original Message-----
>From: Saúl Ibarra Corretgé [mailto:[e-mail]]
>Sent: Thursday, October 20, 2011 3:24 AM
>To: Ravindran Parthasarathi
>Cc: Iñaki Baz Castillo; [e-mail]
>Subject: Re: [rtcweb] Comment on draft-ietf-rtcweb-overview-02 (Browser
>RTC trapezoid)
>
>Hi,
>
>On Oct 19, 2011, at 11:39 PM, Ravindran Parthasarathi wrote:
>
>> Inaki,
>>
>> There are fundamentally two aspects in RTCWeb:
>>
>> 1) (on-wire) protocol mechanism - How protocol is designed between
>RTCWeb client (browser) and RTCWeb server, RTCWeb client and RTCWeb
>client. The protocol design does not focus on RTCWeb client only but
>also in RTCWeb server because the protocol has to be implemented in both
>entities. IOW, it is not 20 lines of JavaScript but impact of RTCWeb
>server design has to be noted as well. Here IETF has significant role
>and it has to come up with best protocol design between RTCWeb entities.
>>
>> 2) API design (JavaScript in browser) - Browser provides RTCWeb client
>framework and exposes its API as JavaScript and JavaScript is the
>language/script of choice for browsers. The interaction is between
>browser and JavaScript only. IMO, W3C plays the vital role in coming up
>with best API set and also, there is no on-wire stuff here. Here, the
>discussion focus is whether low level API or ROAP API or any other high
>level API. Of course, I have less experience in JavaScript based API
>design. Please note that I have designed and worked in protocol
>framework in other languages like C, TCL.
>>
>
>If point 2, the JS API is flexible enough there is no need for a new
>protocol design, we can reuse an existing one and use the JS API for the
>media layer. I've lost count on the number of times this has already
>been said, lets move on.
>
>> My comment on "SIP over websocket is an overkill" is purely based on
>(on-wire) protocol mechanism and it is nothing do with API design.
>Irrespective of whether "SIP over websocket" is implemented in browser
>or in JavaScript or in other language, it is not the best protocol
>design. Hope this clarifies my stance.
>>
>
>Nobody is advocating for SIP over WebSocket nor XMPP over PBTL (pigeon
>based transport layer), what some of us want is to let it up to the
>developers.
>
>--
>Saúl Ibarra Corretgé
>AG Projects
>
>