"Dave Cridland" <dave at cridland.net> writes: >> > So what does "idle" mean >> >> So that no packets are sent or received on that interface. But IIRC, >> we ignore DHCP packets but any other IP packet is tracked. >> >> > Okay, so it's tracking interface level activity? Yep. It's checking if WLAN or BT DUN PPP are idle. >> > , and what does the 770 do when it sees an "idle" connection? >> > >> > Please tell me it doesn't kill the connection. >> >> It closes the connection. But that's use configurable, it can be set >> from the Connectivity Control Panel applet. > > Shuts down the association or the physical interface? (And I know, > I've suddenly switched to OSI-speak to avoid the word "connection" > here) > > And is this done merely if associations are idle, or only if there are > no current associations? Sorry, I don't quite understand your terms here. But basically when idle timer triggers, the WLAN or BT DUN connection is disconnected similarly as user would have clicked disconnect button in Connection Manager or in Connection Status Bar Applet. Anyways, the default is to have idle timer enabled only for BT DUN Circuit Switched Data connections (ie. data calls), which makes sense because they are usually charged per minute. WLAN and BT DUN Packet Switched connections do _not_ have idle timeout set by default. > It makes a huge difference to the viability of the 770 or N800 as a > platform for mobile email, and I'd have thought it'd degrade the user > experience for XMPP as well. Take a look at the highly readable > http://www.ietf.org/internet-drafts/draft-ietf-lemonade-deployments-04.txt > for some general guidance for the mobile email case. We aim for excellent mobility experience. Don't worry, we wouldn't want to harm that in any way. Hopefully this makes things a bit more clear. Do you still see a problem with our idle timer implementation? -- Kalle Valo
