Re: [rtcweb] draft-sipdoc-rtcweb-open-wire-protocol-00 (Open In-The-Wire Protocol for RTC-Web)

[Iñaki Baz Castillo <[e-mail]>]

2011/10/27 Hadriel Kaplan <[e-mail]>:
> On Oct 27, 2011, at 4:55 AM, Iñaki Baz Castillo wrote:
>>
>> Hi, just wondering if somebody has read this informational draft. I'd
>> would really appreciate if folks agree with the conclusiones and
>> requirements exposed in the document or want to discuss about them.
>> The draft tries to clarify the scope of RTC-Web, and IMHO that's an
>> important milestone before we can move on.
>
> I read it as well, and agreed with its statements and conclusions.  What I don't know is whether you wrote it for the purpose of becoming a WG document, or simply as an FYI-type draft.  It has some requirements in it, which I think can be moved into draft-ietf-rtcweb-use-cases-and-requirements (assuming we get consensus to do so).  I thought the requirements were like self-obvious, but some people may not agree, so it's great that you documented them and we can debate them.

Hi Hadriel. Indeed I do think that this draft should NOT exist, but
given the fact that some folks in this WG (AFAIK just one) insist on
standarizing and mandating the RTCweb in-the-wire protocol and message
format, we wrote this informational draft just in order to justify and
explain why the in-the-wire protocol and message format MUST be up to
the developer.

IMHO mandating the in-the-wire protocol and message format is just
useful for those who plan to build and sell "RTCWeb-to-SIP" gateway
boxes, and that should never be the motivation of a IETF spec.

So if we all (or all but one) agree on this, I think we can move on.
The draft is just a helper indeed.


> One process-based concern about making requirement 4 a WG requirement: you can't actually do SIP over Websocket with a "pure SIP network" until we get Websocket into a SIP-extending RFC as a new transport type.  I wouldn't want to hold up WebRTC docs becoming RFCs,

Neither me want that :)
Note however that this is just an informational draft (we don't want
it to become an RFC) and we just expose the example of "SIP over
WebSocket" because we strongly believe in it ;)


> waiting for the DISPATCH and probable SIPCORE process to make a websocket SIP transport into a RFC.

That's indeed our next step. We plan to propose the draft
(sip-over-websocket) in DISPATCH or SIPCORE. Let's some time ;)


> I *want* to add Websocket as a SIP transport type, but it's not actually as trivial as one would think.

Regardless it becomes a RFC or not, take into account that the
in-the-wire protocol and message format in RTCweb is free, so people
can implement SIP over WebSocket if they want :)



> One other minor nit: the draft naming scheme doesn't follow IETF conventions.  The convention is to have the main author/editor's last name as the word after "draft-".

Good to know, I will take it into account for a next draft.


Thanks a lot.


-- 
Iñaki Baz Castillo
<[e-mail]>