On Saturday 10 January 2004 10:10 am, JoeHill wrote: > As long as KMail is configured to look at each message through the eyes of > Spamassassin, it shouldn't matter where the mail is actually being > retrieved from.
Actually, Joe, your solution provides for more control with SA as well. Postfix receives the mail and then directs it through Procmail which passes it through SpamAssassin before delivering to the local spool. That allows SA global filters to be used to filter out everything over say a score of 10-15 which is definitely going to be spam, with probably extremely few or no false positives. Now, users can configure Kmail to pass all local spool mail through SpamAssassin yet again, this time using their own custom recipes and bayesian filter addons with perhaps a custom score that works for them, say a 4-5 which is more likely to grab false positives but also will get much more spam. They drop those into a folder and quickly look through the subject lines before dumping them. Added benefit is that they can use bayesian filters to "learn" from false positives and negatives so that the custom filters get better over time. Eventually, the admin could, if he wanted, apply some of that learning to the global filters and allow other users to benefit from those lessons. That particular method gives you the benefit of having custom created filters in SA, which means that one user can use the DSBL dns block list, or the bl.spamcop.net dns blocklist while another user who may not trust that source can pick another one or simply not use any at all. Finally, some users can actually choose to get spam and filtering can be completely turned off for those particular users with SA. Not only do you get more control with procmail, but SA is a natural extension of procmail and also gives you granular levels of control. -- Bryan Phinney Software Test Engineer
Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com