Re: [MMUSIC] TCP simultaenous open

[Bruce Lowekamp <lowekamp at sipeerior.com>]

I think we need to give up on the idea that we're going to find a single 
solution for TCP NAT traversal that works on all platforms.  TCP-SO is 
great when it works.  If you're on a platform that doesn't support it, 
don't offer a TCP-SO candidate.  The same goes for teredo or any other 
type of candidate.  The fact that Windows XP doesn't support TCP-SO 
isn't that important to the many systems that do support it (and many, 
many deployments will consist entirely of platforms other than windows xp).

Bruce



Rémi Denis-Courmont wrote:
> 	Hello,
>
> Unfortunately, I could not make it to the MMUSIC session. Reading the ICE-TCP 
> slides, there was an open issue that Windows XP might not support TCP 
> simultaenous open.
>
> Now, I am _not_ a Windows expert (has anyone asked, say Dave Thaler or another 
> Microsoft guy at IETF?). But, I can confirm that TCP-SO does not seem to work 
> quite right. Trying to get a (statefully firewalled) Linux box to talk to a 
> Windows XP box... I could open both active and passive sockets on the same 
> port, but it seemed like the incoming connections could not be dequeued from 
> the passive socket. Even though everything was fine on the wire, and the 
> Linux box reported successful connection establishment.
>
> Now, I could have gotten things wrong, but if not, I would say it's time to 
> burry the TCP-SO idea for good :/
>
>
> Regards,

_______________________________________________
mmusic mailing list
mmusic at ietf.org
https://www.ietf.org/mailman/listinfo/mmusic