Bret Miller wrote: > Yes and no. How an e-mail gets forwarded is very much > client-dependent.
You could put it that way. <g> > I find that with Microsoft products, there isn't any easy way to get > it to forward a message and include the full headers. My experience has been that once you get users to understand the operation, Outlook Express does this very nicely- right-click the message in the message list, click "Forward as attachment", enter the destination, and you're happy. Netscape provides an almost identical option on the right-click menu; the major difference is that it doesn't forward "subject.eml" attachments. Eudora doesn't seem to have any way to forward complete original messages; KMail can get *most* of the message headers but it still mangles one or two and sticks in a header or two of its own; at least one version Opera coudn't forward as an attachment of any kind. And then there's Microsoft Outlook. Ugh. Between the 3 most recent releases (I think; I haven't been able to figure out Outlook's versioning) there have been 3 *very* different interfaces, and 3 ways to attach a complete message (as sent by the POP3 server) to a new message. None of them work with all 3 versions; all fail at random in completely unpredictable ways, and there are a host of other problems I've had to deal with doing ISP tech support- prime among them the fact that the interface keeps changing. > And even if you > could, you'd then have to have a way to strip off the outer message > and only learn on the attached message. A relatively trivial task, compared to getting users to forward the untagged spam properly in the first place. > Plus, then you are putting the accuracy of your spam filter in the > hands of your users. Definitely dangerous. ;) > I have a few select users that I allow to drop their spam > in a shared mailbox. I then review the spam and send it to the > learner. We don't limit who can send in untagged spam or tagged nonspam, but I too check each message. I look to see what triggered, and how I might make certain that that particular message and any similar ones get tagged in the future- sometimes it results in another URL component getting added to the growing list of spamsites, sometimes a set of phrases wrapped in a meta rule for .5 points; sometimes a set of overlapping phrases with a combined score over the default threshold in the case of particularly nasty pornspam. > I send probably 80% or more on, but the rest is misclassified by the > user because they don't understand how to get off a mailing list, > whether it's their bank or some online store or whatever. I've seen a few of those, but not many recently. > In my > experience, I can't trust users to know the difference between ham > and spam, so there is still the manual review. After the review, I > drag the message into another mailbox where it gets sa-learn'ed by > cron job and autodeleted eventually. What I'd *like* to do is have per-user Bayes dbs; at the moment I can't afford the disk space to do so. (I would need to take the server down to add a ~10G+ disk for SA misc data alone; I don't want the Bayes and AWL dbs to be part of a user's quotas.) A system-wide Bayes db is not too bad; the spam and ham that people report is remarkably similar across ~350 accounts... -kgd -- <erno> hm. I've lost a machine.. literally _lost_. it responds to ping, it works completely, I just can't figure out where in my apartment it is. ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
