On Tue, Aug 06, 2002 at 11:59:30PM -0400, Dan Merillat wrote:
<> 
> Then why is the NAT case impossible?  This whole thing started specifically
> because when people asked about NAT they were told "The connection will
> be busy, performance will suck, fuckoff."
> 
> If this DOES work properly, holding open 'waiting' connections for a
> longer period will let transients behind NAT work most of the time.

If you go back to my original message you'll see that I wrote
specifically that making transient nodes work behind a NAT would not
require any major changes to the protocol. Adding ways to force the
connections to stay open and to keep opening more connections for every
query (or queuing replies) are exactly the protocol cludges, layer
breaches, and resource wastes that I was referring to.

> If it's due to threadcount, then we're back to my original summary: wait
> on NBIO, then it will just work.

Even with NBIO it would not be desirable to keep an endless number of
connections open. So it offsets any major resource waste, but the other
issues remain, and having the users in question use FCP is still
preferable.

And anyways, I seriously doubt you will ever see non-blocking IO in Fred.
I'm not going to do it, and around here that usually means it won't get
done.

-- 

Oskar Sandberg
oskar at freenetproject.org

_______________________________________________
devl mailing list
devl at freenetproject.org
http://hawk.freenetproject.org/cgi-bin/mailman/listinfo/devl

Reply via email to