Re: [Gaim-devel] STUNT integration

[Nathan P. <gai...@gl...>]

Ethan Blanton wrote:
> Haig Didizian spake unto us the following wisdom:
>> I think it's all about the "T" at the end of STUNT. 
>>
>> While STUN was designed for UDP, STUNT is for TCP communication, too.
> 
> Ahh, I missed that on the first read -- if it works, go to it.  We are
> slowly growing a number of mechanisms for making NAT traversal more
> plausible, and as long as they don't interfere with normal operation
> and are not overly invasive, I see no reason not to include others.

Please take heed of this warning in the FAQ:

"...the STUNT library has not been optimized for 
connection-establishment-latency. Therefore, it make take little longer 
to connect to end-point using STUNT that you could already connect to 
with plain sockets. As a result, the recommended way of using the STUNT 
library is to try the plain-socket connection first and if that fails, 
then fall back to the STUNT library."

Also, perhaps I am wrong, but it is my understanding that STUNT requires 
special cooperation from the other client for TCP hole punching.  If 
that is the case, then maybe STUNT is not so useful for gaim since it 
would only work with other gaim clients.

-Nathan