[Standards-JIG] Re: The Ack Hack.

Dave Cridland dave at cridland.net
Wed May 10 16:54:02 UTC 2006

On Wed May 10 17:45:38 2006, Pavel Šimerda wrote:
> I think the server should ping silent connections too... just use 
> longer timeouts... like 2 minutes... or 5 minutes... or whatever.
No, because there's no need. A client might want to do this in order 
to prevent a NAT (or connection tracking firewall) timeout, or to 
support any other kind of limited connectivity, but there's no reason 
for a server to do that, it has no reason to think a silent 
connection is anything but silent.

TCP is designed to survive a temporary loss of the underlying network 
layer, it'd be a terrible shame to throw that away, and a silent TCP 
connection doesn't cost much in terms of resources. (Or shouldn't, 

When stanzas get sent to that connection and there's still no acks, 
it's worth the server doing something. It might even be worth 
soliciting an ack when the first stanza "for a while" gets sent, too, 
to shorten the timeout.

           You see things; and you say "Why?"
   But I dream things that never were; and I say "Why not?"
    - George Bernard Shaw

More information about the Standards mailing list