Re: Do we really need TCP/DTLS/SCTP proto field?

On 2/16/17 12:47 PM, Ben Campbell wrote:
> On 16 Feb 2017, at 9:51, Christer Holmberg wrote:
>
>> Hi,
>>
>> Note that the mechanism is also used at least by CLUE, which does not
>> mandate ICE (or JSEP).
>
> ISTM that this argument trumps the ongoing ICE usage discussion. Am I
> missing something?

First, lets note that TCP/DTLS/SCTP is never a desirable thing. It is 
simply a fallback for when for some reason UDP/DTLS/SCTP is impossible.

So something attempting */DTLS/SCTP without ICE would need to have some 
other knowledge that UDP was impossible. This is of course *possible*. I 
don't know how probable a use case it is for something new now that ICE 
exists.

IMO this is really another piece of evidence that we ought to have ICE 
as a transport: ICE/DTLS/SCTP.

	Thanks,
	Paul