Eric Shubert wrote:
I've read alot about both, I am mainly looking
at it, cause I've been running spamassassin for a while and it's good,
but I wanted to try something different and dspam sounds really neat.
DSPAM would be a good one to try. I think that it's bayes processing is
better suited to a global (shared) setup. That's just from what I've read
though, I haven't actually tried it (yet).
I've used DSPAM before - it's okay. It does not perform well until it's
been trained, but it does have some nice features that spamassassin
didn't have - outside users are automatically whitlisted after a couple
emails or if you reply to a couple; they do not get scanned for spam
anymore after that point unless you send one of their messages to the
+spam address. A few other nice things like that.
It would be suited for environments where your users are semi-technical
(or at least have more than a room-temperature IQ) and the users must be
willing to send spam in to train it. I've personally found that less
than 1% of users will actually do this, YMMV.
