Paul Johnson wrote:
Steve Lamb <grey@dmiyu.org> writes:Currently I am running uw-imapd.
Good. Hold the course. You don't need to change your IMAP server to achieve what you want to do here.
Uh, yes, I do. UW-IMAP does not offer shared folders AFAIK. In fact it barely has any configuration options to speak of.
Newer versions of uw-imapd seem to break Squirrelmail. Every time I upgrade it breaks. Every time I set it back to a good version it does not.
I don't have this problem...
Happens every time here. I am held at the last good version. I update, Squirrelmail breaks and no matter what I do to reconfigure it it will not work with newer UW-IMAPDs. I downgrade and it works again. That is the one and only variable. I'd say it is something in UW-IMAPD's configuration except I think it is limited to, what, 3 items? Yeah, slight exaggeration there but you get the idea.
I'm loking for shared folders so I can offer global spam/ham folders for my users. I know this is generally a nono but in this instance I am willing to run with it given two facts.
Bayesian filter pollution between users would make it more or less useless, though.
3 people, maybe 6 accounts total. Our reading habits are that so disparate that what they consider ham/spam is going to instantly reject my Debian lists and let the spam come flooding through the gates. There is a time for purity and a time for practicality. I have judged that this is the time for practicality because the purity will not suffer *enough to make a difference*. If it were 12 people I'd be disinclined to take this route. But it isn't and chances are it won't. If it were 60 or 120 I most certainly would look for other options. In fact in another context where I am investigating spam prevention for a large number of people I have rejected global bayesian filters.
The first is that spam scanning happens at SMTP. Per user bayes filters do not apply at that time, it has to be global.
Sure they do.
No, they don't. They are technically impossible.
rcpt to: joe@domain.com
rcpt to: jane@domain.com
Joe's bayes DB marks it as spam. Jane's does not. Do you reject or
accept *AT SMTP TIME*? Per user bayes cannot be used at that time because you
either have to accept for all recipients or deny for all recipients. You
cannot selectively accept for one and reject for another. As a result per
user bayes filtering must happen *after* SMTP and part of delivery.
> dman downplays the effectiveness of his original
solution to this problem, but I tend to prefer it myself. It allows for each user to have their own spamassassin settings apply. http://dman13.dyndns.org/~dman/config_docs/exim-spamassassin/
Nice, a cite to an index of other pages. Would you care to cite the specific example. If you're referring to "Integration Method 1" then I have to point out that it is not an SMTP time rejection and therefore not under consideration.
If these users have normal shell logins, then all you have to solve, optionally, would be folders to assign spam and ham to that gets run through spamassassin -r and spamassassin -k at given intervals.
What you meant to say here was, "If these users have normal shell logins.... and know how to access them.... and know how to use them.... then all you have to solve..." Sorry, while all the users do have shell accounts outside of myself none of the users are technically competent enough to use a shell account nor are they inclined to learn. The whole reason for a shared IMAP folder is so they can use a familiar interface, their client (be it TBird or Squirrelmail) to dump the mail into a single folder which is swept on a regular basis.
I've been kind of vaguely wondering in the background if spamassassin has support form mbox input since that would greatly simplify things.
I never understood why people insist on using spamassassin when it comes with a perfectly good tool; sa-learn. From sa-learn's manpage:
--mbox Input sources are in mbox format
--
Steve C. Lamb | I'm your priest, I'm your shrink, I'm your
PGP Key: 8B6E99C5 | main connection to the switchboard of souls.
-------------------------------+---------------------------------------------
Attachment:
signature.asc
Description: OpenPGP digital signature