ICE restart behavior still needs to be resolved

[juberti]

#59 attempted to address this, but issues were discovered:

"I don't think this is sufficient given the need for TransportController and freezing policy. The TransportController really needs to be able to redo its thing, suggesting that ICE restart needs to be internal to an IceTransport rather than configured externally. (Unless we want to go down the route of having multiple TCs, which starts getting into questionable territory)."

[robin-raymond]

Just to clarify, are you saying we need an explicit "restart()" method on the IceTransport? I'm okay with that, just want to clarify.

[aboba]

I'd suggest we assume that a restart() method on the IceTransport is what is needed.

[aboba]

restart() method added in the latest editor's draft:
http://ortc.org/wp-content/uploads/2014/06/ortc.html

[aboba]

Can we reopen this? In the June 24 ORTC meeting, I believe we concluded that we had not fully addressed this.

[aboba]

RFC 5245 9.1.1.1 indicates that an ICE restart can involve changing endpoints, so that the DTLS Transport would also need to be changed:
http://tools.ietf.org/html/rfc5245#section-9.1.1.1

[aboba]

Yet another restart-related issue from Shijun Sun:
Section 3.3. IceTransport. restart(). If this is to reset to gathering, the spec should be clear on the impact to “state”, local candidate list, etc. And, should we allow any change to the getherPolicy when restart()? And, from another perspective, can we recommend just constructing a new IceTransport as in Section 3.1, i.e. removing “restart()” from the interface?

[robin-raymond]

$related #101

[aboba]

I believe the resolution is to add restart() to the RTCIceTransportController, no?