Re: [rtcweb] Please require user consent for data channels

[Iñaki Baz Castillo <[e-mail]>]

2015-07-18 1:09 GMT+02:00 Justin Uberti <[e-mail]>:
> It is pretty clear from the diagram on page 10 that for each host candidate
> X:x, the agent will gather a srflx candidate X1':x1' and relay candidate
> Y:y. This is also spelled out in the text underneath the diagram:
>
>    When the agent sends the TURN Allocate request from IP address and
>    port X:x, the NAT (assuming there is one) will create a binding
>    X1':x1', mapping this server reflexive candidate to the host
>    candidate X:x.
>
> Ergo, one tries to gather srflx and relay candidates for each host candidate
> X:x.

That's not what I understand.

What I understand is that the browser should bind on 0.0.0.0 (or ::0)
and send the STUN/TURN request, and the underlying OS would decide the
proper interface and source IP. So probably just one or two (IPv4 and
IPv6) STUN/TURN attempts would be performed during the gathering, and
when *that* happens the quoted text above applies (being X the IP of
the candidate matching the source IP chosen by the OS to route the
STUN/TURN request).

Said that, and omitting the above confusing text: what is the purpose
of sending STUN/TURN requests over all the interfaces? How is that
good at all? AFAIK in most cases that will just mean timeouts waiting
for STUN responses (which is bad if trickle ICE is not being used).

To be even more clear: STUN/TURN servers are typically in the public
Internet, although they may also be placed in local networks. Anyhow,
computers within that network have a single and already configured
route to reach the STUN/TURN server (usually the default route).
Sending STUN/TURN requests from other interfaces will just fail.


-- 
Iñaki Baz Castillo
<[e-mail]>