Re: [rtcweb] ICE-Mismatch and WebRTC

[Roman Shpount <[e-mail]>]

On Fri, Jul 11, 2014 at 12:07 PM, Justin Uberti <[e-mail]> wrote:

> The fact that WebRTC implementations MUST ignore the address and port in
> the c=/m= lines will be written into JSEP, S 5.6/5.7.
>
>
> Is it going to completely ignore these parameters or would it still honor
port 0 to indicate that m line disabled?

What about 0.0.0.0 address to indicate hold? These should not be allowed
per RFC 5245 section 9.1.1.1 but I still do see it used in SIP, especially
with various B2BUA scenarios where this is the only way for a pure
signaling agent to put a call on hold or to resolve collision. There are
other ways to deal with this in WebRTC but this will still come up during
interop with SIP.

Would it also document on what happens when WebRTC implementation receives
an answer SDP with ice-mismatch attribute? An error in such case is
probably the most appropriate action.
_____________
Roman Shpount