On Fri, Apr 05, 2013 at 02:19:01PM -0700, Brendan Cully wrote: > On Friday, 05 April 2013 at 15:12, Nicolas Bock wrote: > > On Fri, Apr 05, 2013 at 10:52:51PM +0200, Patrick Ben Koetter wrote: > > > * Nicolas Bock <nicolasb...@gmail.com>: > > > > Hi, > > > > > > > > I have about 190,000 messages in my +[Gmail]/All Mail folder. Changing > > > > to that > > > > folder always takes a minute or two despite the fact that I am using > > > > header_cache. Is there something else I could do to speed up opening > > > > large > > > > folders? > > > > > > Something along these lines? > > > > > > # Caching > > > set header_cache="~/.mutt/.cache/" > > > set tmpdir="~/.mutt/.tmp/" > > > set message_cachedir="~/.mutt/.messages/" > > > set message_cache_clean = no > > > > > Thanks. So far it doesn't seem to make much difference though. I compiled > > mutt > > with gdbm, does anyone know whether other database backends make a > > difference > > in terms of performance? > > tokyocabinet should be noticeably faster. >
I have recompiled mutt a few times now with different database backends. This is really only a quick and dirty benchmark, so I might very well have missed something. First I deleted the header cache and timed switching to "All Mail": Cold connection without headers database: 2'56" Then I switched back and forth between "All Mail" and INBOX a few times and timed opening "All Mail" when it was cached: Tokyocabinate without compression: 59" Tokyocabinate with compression: 54" qdbm with compression: 56" gdbm: 56" As you can see, I didn't measure much performance difference between the different backends, although tokycabinet is slightly faster than the rest. Then I did the same test on a local copy of the folder (using offlineimap): on local copy of All Mail: 6" Again there wasn't much difference between the backends. The local read test really seems to indicate that it's not the database backend that is controlling performance when switching folders here. It is presumable network communication with Google's imap servers. And that presumably means that I can't do much about it, or can I? nick
signature.asc
Description: Digital signature
