[imp] DSPAM integration

[Liam Hoekenga]

We're testing DSPAM at the University of Michigan, and have been using 
it with our IMP installation for several months now.  Aside for one 
feature in IMP 4.0 that we're waiting for, we've been pretty happy with 
how easy it was to integrate.

> The suggested setup for reporting false positives and spam with DSPAM 
> is to
> forward the message to per user email aliases.
> eg.
>   spam-user at example.com
>   falsepos-user at example.com
>
> which is a bit cumbersome and I don't see my users actually using it.

Is it any less cumbersome than "dspam --user user at domain --addspam"?

How will your users be accessing their email?  Solely through the 
webmail client?  Other desktop clients?  Terminal clients (mutt, mh, 
pine, etc?) Are they going to have the option of running the command 
line anywhere else, or only using the webmail client?

If you go w/ the email reporting method, your users don't have to have 
access to a command line, and can train their spam filters using 
whatever email clients you allow them to use with your servers.

> I'd rather have the 'Report as spam' link run
> dspam --user user at domain --addspam
> and have a second link 'Report as False Positive' run
> dspam --user user at domain --falsepositive

The "report as spam" link can either execute a program or send email to 
an address.

> It works, however I did run into permissions errors.
> the dspam binary is installed 2510, root.mail
> and the dspam USERDIR is 770, root.mail

If you use $conf['spam']['email'] instead of $conf['spam']['program'], 
you don't need to grant your webserver broader privileges on the 
filesystem.  It's probably best to not let your webserver have more 
access than it absolutely needs to.

I can generate a diff of what we're running, if you like.  I've already 
submitted a version of it back, and it was accepted into the CVS 
branch.

Liam