I just wondered about redirecting mail to learn as Spam, as mentioned in FAQ 5.1.
This will add extra headers (especially Received headers). Won't these headers poison the Bayesian database?
This varies a lot from mail client to mail client. With some mail clients, the redirect is almost more of a re-queue of the original message with a new SMTP envelope... In 5.1 Justin is speaking specifically about outlooks resend ability.
Others mail clients (such as my own) do it more like a forward.
That's why FAQ 5.3 exists, to emphasize that is kind of thing is really not as easy as it may first seem, but with some MUA's it may be possible.
And now that I look at it, FAQ 5.3 and FAQ 5.1 really should be merged into one.. maybe I'll ping justin on that at some point.
Thus far I can only account for Eudora, where it's absolutely impossible as the client will immediately discard any text sections of a multipart/alternative message as soon as you download it, making good learning impossible even if you could get forwarding with no header modifications.
btw: I upgraded to 2.60 yesterday and even spoiled SA with Razor and DCC (well, Pyzor cowardly refuses to install) -- and the results are great! Very good job, folks, thanks! :-)
-------------------------------------------------------
This SF.net email sponsored by: Enterprise Linux Forum Conference & Expo
The Event For Linux Datacenter Solutions & Strategies in The Enterprise Linux in the Boardroom; in the Front Office; & in the Server Room http://www.enterpriselinuxforum.com
_______________________________________________
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
