Re: [Evolution] IMAP Issues ...

[Jeffrey Stedfast <fejj ximian com>]

On Wed, 2002-09-18 at 08:07, Not Zed wrote:
> On Thu, 2002-09-19 at 06:03, Ryan P Skadberg wrote:
> > Subscribing to Mail fixed it.  Two issues though:
> >
> > a) I couldn't do this through evolution, had to telnet in and do it.  If
> > you try and subscribe a directory in evolution, it ignores you.
> > b) After doing this and restarting evolution, it took, oh about 2-3
> > hours to get all of my folders to come up in evolution (approx 600) and
>
> How many total messages is this?  What sort of speed is your box, what
> kind of hard drive?

How fast do other clients load the folders (total time to load the
entire message-lists, not just the time it takes to start populating the
list)?

> > evo was unusable for that whole time. Is this a first time thing or is
> > this going to happen every time?  (Haven't had a chance to run with
> > camel debugging on, as I couldn't lose my mailer for another 3 hours
> > today if it happened again)
>
> its much much slower the first time than subsequent times.  we're well
> aware of the shittiness of the scalability of code, but its hard to fix
> just yet :-(

I've just patched the summary fetching code to not cache to disk and to
only fetch a minimum subset of headers from the server needed to
generate the message list and it is on the order of 2 to 3 times faster
(calculated by timing how long it took to load my bugzilla folder and my
folder for this list and comparing those times with the times it took to
load those 2 folders with my optimisation patch).

Anyways, with my optimisation patch - it is the absolute best speed you
will ever see Evolution do for loading a folder for the first time. It
really can't get any faster unless you want to give up threading
capabilities and/or mailing-list vfolders/filtering (thus allowing use
to just use FETCH ENVELOPE). And even then it wouldn't get a whole lot
faster.

Jeff

-- 
Jeffrey Stedfast
Evolution Hacker - Ximian, Inc.
fejj ximian com  - www.ximian.com