The Friday 2003-12-26 at 12:26 +0100, Jon Clausen wrote:
Now, since setting up {spam|ham|forget}@domain seems to require too much of the users (in that they would be required to edit the headers, before forwarding to those accounts) I think the {spam|ham|forget}-mailbox+cronjob-sa-learn is probably the better solution. Because this would only require the users to move the message(s) to said mailbox(es).
This, however, begs the next (couple of) question(s);
I have another thought: It may happen that what one user considers spam, for another one it might be useful mail. There would be problems if spam criteria is not unified amongst your users. That must be the reasoning behind the idea of setting spam filtering for each user, instead of globally - or at least, the customization or bayesian filters and such things. On the other hand, some of those files are big, with one set for each user: 659456 auto-whitelist 258068 auto-whitelist.dir 258068 auto-whitelist.pag 907 bayes_msgcount 1294336 bayes_seen 5148672 bayes_toks Just an idea :-} -- Cheers, Carlos Robinson