Jonathan Rosenberg wrote:
As folks noticed, I posted a revision of ice-tcp. However, the
document itself has not changed as I wasn't able to invest any time in
it since last IETF.
Since Dublin we did get confirmation that SO does not work on XP, thus
making ICE-tcp really ineffective. I think this is the impetus for
Bruce's framework draft.
The main questions then for the ICE-tcp draft, beyond the framework
stuff proposed by Bruce and Adam, is:
* should we REMOVE S-O from ICE-tcp, given the problems supporting it
on XP
does it make sense to remove S-O because it does not work on 1 given
platform? I think this approach is very Softphone oriented..
There are plenty of embedded SIP devices with their own OS/IP-stack
that might be able to support S-O.
if you want to make ICE-TCP more modular, we can also consider moving
the TCP S-O function to a separare ICE-TCP extension document.
/alfred
* should we instead try to use one of the other mechanisms from
Saikat's paper
Comments welcome.
Thanks,
Jonathan R.