1. is my approach wide of the mark or are there better, smarter solutions to achieve this? (I've read about people doing this using a deliver wrapper which calls spamassassin and then deliver. but I don't like this kind of "fork and exec orgies")
I'm in your situation - some of my users _need_ spam filtering, others absolutely don't want it. I've found the best solution to be simply tagging messages using SpamAssassin (X-Spam-Flag: true|false) then letting filtering be done in sieve using a simple header check. It's even possible for users to pick different thresholds - one of my users likes 8 instead of the default 5, so they just use sieve to check if X-Spam-Score contains 8 or more asterisks. You can also use Junk mailboxes or instant spam deletion configured on a per-user basis this way.
It wouldn't be hard to whip up a CGI script to do the sieve configuration for them but my site is not large I'm currently handling it manually.
It's also possible for users to configure their clients to do client-side filtering, but server-side with sieve is much nicer.
Craig Ringer
