on Sun, Apr 03, 2005 at 08:23:36AM -0700, Doctorcam ([EMAIL PROTECTED]) wrote: > * Karsten M. Self ([email protected]) wrote: > > So, I've got problems with some really big email archives. And mutt > > bogs heavily opening (or searching) same. > > > So I _thought_ I had mutt's header cache working. Well... Just having > > the hack compiled in isn't enough. If you don't set "header_cache_ to a > > location where the cache is maintained, you don't have it. > > This made me go looking at my own. Same thing - I never actually > created the directory. > > Thanks, Karsten - this was very helpful.
Innit just? So the results. First time opening a given folder, there's hit as the index is built. When reopening the folder when no mail's been added, action's _far_ faster, though sorting contents still takes time. The problem remains, though that when reading a large folder which has been incrementally updated, there's still a long delay just moving between messages. E.g.: debian-user, current 5749 messages, just took 4m50s or thereabouts going from viewing a message to displaying the folder index. Maildir format. System load hit about 9 during this process. So it's an improvement, but still not bliss. There's a related option, "header_cache_verify", which checks to see if the folder's been modified by a non-mutt program. I'm not sure if procmail counts as same, but I'm going to try disabling this. The option (default=yes) requires one stat() per message in the folder for each read of the folder. Which sounds expensive over time. Peace. -- Karsten M. Self <[email protected]> http://kmself.home.netcom.com/ What Part of "Gestalt" don't you understand? She was asking about you earlier, Rick, in a way that made me extremely jealous. - Casablanca
signature.asc
Description: Digital signature
_______________________________________________ vox-tech mailing list [email protected] http://lists.lugod.org/mailman/listinfo/vox-tech
