Re: [dispatch] I-D Action: draft-pd-dispatch-msrp-websocket-03.txt

[Peter Dunkley <[e-mail]>]

Hello,

Because WebSockets is an asynchronous protocol two MSRP peers (that are
clients) cannot use MSRP over WebSockets.

It could be possible for a WebSocket server to be an MSRP peer (for
example, in the scenario where the server is B2B'ing MSRP like an SBC will
do in IMS for CEMA).  Christer Holmberg has expressed an interest in this
scenario and may provide language relating to it for the draft - but for
now this is out of the scope of the draft which explicitly deals just with
with the MSRP through relay scenario.  If Christer does provide some
additional language then the draft will cover this without relay scenario,
but if not nothing in the draft should prevent someone from addressing it
in the future with an additional document (should there be interest).

>From my point-of-view I am only interested in the with relay scenario at
this time.

Regards,

Peter


On 13 February 2014 21:18, Sergio Garcia Murillo <
[e-mail]> wrote:

>  Hi Peter,
>
> I think that you misunderstood me. I was not asking you to change the MSRP
> over websocket draft to set the transport to TCP/WSS/MSRP and do it like
> the BFCP over websocket draft, but to find a way that we could both use the
> same solution with the principles that you are mentioning, that is:
>
> -Do not break backward compatibility
> -Allow easy implementation for servers which wants to support both TCP/TLS
> and WS/WSS "flavors"
>
> In my case, I will implement both BFCP and MSRP over TCP and Websockets in
> my MCU, so my interest is genuine.. :)
>
> I have a doubt regarding the draft. If I understood it correctly, you are
> only considering P2P in he examples and using a msrp relay in the middle to
> be able to connect a WS to a non WS, but I am not sure how would you handle
> the case when you don't use a relay. Is that possible at all?
>
> Best regards
> Sergio
> El 13/02/2014 17:26, Peter Dunkley escribió:
>
> Hello,
>
>  The method chosen for advertising WebSockets in the SDP for MSRP was
> picked because it was would enable interoperability with existing MSRP
> implementations through an MSRP relay without requiring changes to those
> implementations.  In this scenario advertising WebSockets in the SDP would
> mean that the existing implementation - which does not need to support
> WebSockets (the MSRP relay deals with this) - would need to be able to cope
> with seeing TCP/WSS/MSRP in the SDP.  In all probability existing
> implementations will not cope with seeing "TCP/WSS/MSRP" in the SDP instead
> of "TCP/TLS/MSRP" - this would be a big problem.
>
>  I have no issue with consistency between the drafts using WebSockets
> where possible, but in this case doing what is done in the BFCP draft
> (explicitly putting TCP/WSS/MSRP into the SDP) may not be practical (and in
> fact may be quite undesirable).
>
>  Regards,
>
>  Peter
>
>
>
> On 13 February 2014 16:07, Sergio Garcia Murillo <
> [e-mail]> wrote:
>
>> What I mean is that I expect quite a lot of "over websocekts" drafts and
>> we should try to use the same solution for advertising it in the SDP, and
>> not have each one have their own way of handling it.
>>
>> Best regards
>> Sergio
>> El 13/02/2014 16:49, "Mary Barnes" <[e-mail]>
>> escribió:
>>
>>  Sergio,
>>>
>>>  The draft you mention is being discussed in the BFCPBIS WG and any
>>> comments that others might have on that document should go to that list.
>>>
>>>  It's not clear to me whether your suggestion is that changes are
>>> needed to this MSRP document or whether you are just proposing to make the
>>> BFCPBIS consistent with this MSRP document?
>>>
>>>  Mary.
>>>
>>>
>>> On Thu, Feb 13, 2014 at 9:38 AM, Sergio Garcia Murillo <
>>> [e-mail]> wrote:
>>>
>>>> Hi,
>>>>
>>>> We also have recently published a different draft for BFCP over
>>>> websockets:
>>>> http://tools.ietf.org/html/draft-pascual-bfcpbis-bfcp-websocket-00
>>>>
>>>> I beleive that we should harmonize who the WS transport is signaled and
>>>> how to signal the WS version if also the "normal" TCP version is supported.
>>>>
>>>> In your case you seem to be using the same transport line TCP/MSRP and
>>>> use the path to signal the ws part, in our case, we choose to signal it in
>>>> the transport TCP/WS/BFCP and include a new attribute ws-uri to signal the
>>>> full url (I could not reuse the path attribute as it is restricted to msrp
>>>> urls only).
>>>>
>>>> Best regards
>>>> Sergio
>>>> El 13/02/2014 16:17, Gonzalo Salgueiro (gsalguei) escribió:
>>>>
>>>>>  Hi Mary -
>>>>>
>>>>> Thanks for taking the time to review the document.  We have published
>>>>> an -05 that (hopefully) addresses all your feedback.
>>>>>
>>>>> Inline, trimming to only the points requiring responses...
>>>>>
>>>>>
>>>>> On Jan 10, 2014, at 5:58 PM, Mary Barnes <[e-mail]>
>>>>> wrote:
>>>>>
>>>>>  I have agreed to shepherd this document.  I've reviewed the document
>>>>>> in anticipation of doing the PROTO write-up and have the following comments
>>>>>> and questions.  Ben Campbell has agreed to do the required expert review
>>>>>> and that should be posted within the next week or so.    This is also a
>>>>>> good time for anyone in the WG that hasn't previously reviewed this
>>>>>> document to review and provide any final comments.  Note, that this
>>>>>> document was agreed to be AD sponsored around the IETF-86 timeframe.
>>>>>>
>>>>>> Regards,
>>>>>> Mary.
>>>>>>
>>>>>> Review Summary: Almost ready. Comments & questions below.
>>>>>>
>>>>> <snip/>
>>>>>
>>>>>  5) Section 10.1. Since securing the connection is just RECOMMENDED,
>>>>>> what are the implications and risks if the MSRP traffic isn't transported
>>>>>> over a secure connection?
>>>>>>
>>>>> Other review comments indicated that it was problematic to downgrade
>>>>> the 4976 MUST requirement for TLS between a client and a server. Thus, the
>>>>> document has been updated so that MSRP traffic transported over WebSockets
>>>>> MUST be protected by using a secure WebSocket connection (i.e., using TLS).
>>>>>  I believe this renders this point moot.
>>>>>
>>>>> <snip/>
>>>>>
>>>>>  8) It's typical for documents that are updating existing RFCs to have
>>>>>> a section that summarizes the updates to the existing RFCs that are made by
>>>>>> this document.
>>>>>>
>>>>> This was the intent of Sections 5.2 and 5.3.  Is this sufficient? Or
>>>>> did you have something else in mind?
>>>>>
>>>>> Regards,
>>>>>
>>>>> Gonzalo
>>>>>
>>>>>   _______________________________________________
>>>>> dispatch mailing list
>>>>> [e-mail]
>>>>> https://www.ietf.org/mailman/listinfo/dispatch
>>>>>
>>>>
>>>> _______________________________________________
>>>> dispatch mailing list
>>>> [e-mail]
>>>> https://www.ietf.org/mailman/listinfo/dispatch
>>>>
>>>
>>>
>> _______________________________________________
>> dispatch mailing list
>> [e-mail]
>> https://www.ietf.org/mailman/listinfo/dispatch
>>
>>
>
>
>  --
>  Peter Dunkley
> Technical Director
>  Crocodile RCS Ltd
>
>
>


-- 
Peter Dunkley
Technical Director
Crocodile RCS Ltd