On Wednesday 25 June 2008 20:04:02 Roland Harnau wrote: > 2008/6/23, Andreas Hartmetz : > > Did you consider making the name cache even more central like kded or > > kdeinit? That should give us more cache hits in some cases. > > My first try to integrate the resolver into KIO was via a standalone DBus > service, i.e. I wrote a DBus adapter with "delayed reply". The result was > completely disappointing because of high latency. Besides, I doubt the > session bus is apt to handle the slave storm usually triggered by > Konqueror. So I think a kded module is out of the question. The location in > the application's address space is of course not forced, but allows fast > name resolution without much additional complexity. OK. Putting the cache into klauncher might still make sense, if it keeps a connection to the slaves somehow that is. David? Anyhow, feel free to ignore that possibility for now. One cache per process is a nice improvement already :) -- Flint Paper is insane. I really respect that.