[Spambayes] How well does sb_imapfilter.py work?

[Tony Meyer]

[It doesn't appear that this messages to spambayes at python.org has been
answered yet - sorry about that! I've been away and so are catching up with
the messages. Please ignore if I'm wrong and this has been resolved
already.]

> I learnt from my short experiments with imapfilter the following:
> 
> * imapfilter seemed to work quite slowly when it processed a 
> lot of messages. It seemed to hang for a while. If I just 
> waited, then it finished, though.

The 1.1 version of sb_imapfilter should be a fair bit faster, I hope.  Feel
free to grab the source from CVS and give it a go!

> * When I interrupted it with Ctrl+Break the database was 
> broken (AssertionError spamcount <= nspam).

Yes, interrupting any SpamBayes script is asking for database trouble.  I
think most of us would be happy to move away from the bsddb database that we
use at the moment (hopefully solving some of the problems); unfortunately
it's a lot of work, and requires being familiar with some db system.

> * The web interface is out of sync, if you start it first, 
> and then do some filtering. The statistics are not correct.

The web interface won't run at the same time as the filter, will it?  (It's
been a while since I've used it).  If there is a problem like this, please
open a bug tracker on sourceforge and I'll get to it when I can.

> * The web interface had problems with showing the folders 
> selected for filtering and unsure/spam. Now it seems to work.

This should also be improved in 1.1 (probably also 1.0.1).

> * The web interface shows the password in clear text on 
> screen.

Feel free to open a feature request tracker on sourceforge about this.  I'm
not sure exactly what to do, other than special-case this, though, since all
the other options are just shown in plain text.  You can use the -P option,
of course.

> Also -i 4 shows the password.

I've tried to think of something to do about this before, since it's easy to
accidentally submit your username and password when debugging.  Perhaps it
should set the debug level to 0 when logging in, and then reset?  Again, a
bug tracker would encourage me to get something done about it.

> It is impossible to configure settings when some 
> other people are in the same room...

Suggestions are welcome :)

> * If I didn't use the expunge option, then the messages were 
> tripled in Outlook 2003 in Cached Exchange mode.

1.1 will handle this nicer.

> Imapfilter has two shortcomings:
> * It can't process more than one server/login.

The infrastructure is all there for this, but no-one has ever asked for it
before.  A feature request tracker for this would probably get done
reasonably promptly (for 1.1, not for 1.0.1).

> * It can't move the filtered good messages to another folder.

There's already a feature request open about this, I think.  Add your plea
to it, and that'll encourage me to get to it sooner.

> I thought about adding those, but probably I'll wait for 
> Spambayes 1.1 and Python 2.4 first.

If you are considering working on it (that would be great!), then you should
really work from the CVS version.  Do an anonymous checkout and run with
that.  Otherwise you'll always be working against code that's quite out of
date.

=Tony Meyer

---
Please always include the list (spambayes at python.org) in your replies
(reply-all), and please don't send me personal mail about SpamBayes. This
way, you get everyone's help, and avoid a lack of replies when I'm busy.