Re: [rtcweb] New version of use-case draft uploaded

[Stefan Håkansson LK <[e-mail]>]

On 2014-02-10 12:02, Magnus Westerlund wrote:
> Hi Partha,
>
> See inline
>
> On 2014-02-08 10:12, Parthasarathi R wrote:
>> Hi Stefan,
>>
>> <snip>
>>> Yes, I did that change (TURN -> ICE). My understanding was that ICE is
>>> what is used/mandated (and it in turn makes use of STUN and TURN).
>>>
>> </snip>
>>
>> ICE is mandated in RTCWeb but I disagree to your assumption that ICE in turn
>> makes use of TURN. My concern is that The requirement document is misleading
>> about TURN requirements. ICE-TCP avoids TURN server as the middle-box in the
>> network. ICE-TCP is host candidate whereas TURN is relay candidate and as
>> per ICE candidate selection, host candidate is preferred over relay
>> candidates.
>
> I will agree that the note as currently worded is not completely
> accurate and clear. Based on the text in the preceding section, it makes
> sense to note that ICE using STUN and TURN is not the only solution
> including additional ICE modes or relay services. I request the authors
> propose a new formulation of the note.

Proposal:

"Note that ICE support being mandatory does not preclude a WebRTC 
endpoint from supporting more traversal mechanisms than ICE using STUN 
and TURN."

OK?