Re: [rtcweb] Fwd: New Version Notification for draft-thomson-rtcweb-alpn-00.txt

[Paul Kyzivat <[e-mail]>]

On 4/9/14 5:25 PM, Martin Thomson wrote:
> On 9 April 2014 13:56, Paul Kyzivat <[e-mail]> wrote:
>> I expect two different protocols to look different on the wire.
>>
>> You seem to be saying that SMTP used to talk to ietf mailing lists is a
>> different protocol from SMTP used to talk to my lawyer, because I expect my
>> lawyer to keep the communications confidential.
>
> They are different on the wire.  They use different identifiers.
>
> That's hair splitting, but there's a real difference between the two
> usages.  And I think that it's important enough to do this.
>
> Do you perhaps have an alternative, or is it just that this lack of
> solidity is giving you heartburn?  Because I can appreciate that.

Heartburn.

It seems like a hack - tunneling - just looking for *something* that can 
be used to convey one more bit of data.

I don't have another suggestion. I only half understand the problem. But 
to the extent that I do, it seems ill-defined. AFAIK it only really has 
meaning in the context of a browser. Once you start signaling it, it 
gets fuzzier. If you *knew* both ends were controlled by a browser 
following these conventions then it would be a bit clearer. But when 
that ceases to be true I can make no sense of what it all means. And as 
best I understand, the isolation is intended to be per-media-stream, so 
solutions that are at a different level seem problematic.

	Thanks,
	Paul