Hi,
I am interested in implementing a spamassassin solution similar to the one described here:
http://wiki.apache.org/spamassassin/SiteWideBayesFeedback
Specifically, point 5 indicates a clean solution using IMAP: add 2 shared IMAP accounts (one for spam, the other for nospam), so that all users can drag-n-drop into those folders from their own POP3 folders.
My company uses a wide array of email clients, from M$ Outlook to Thunderbird and Mutt. The problem I am experiencing is that the original mail headers change from the time they are dropped off in /var/mail by the local MTA, to the time they are moved into the IMAP folder. Here is an example of the first couple of lines in the original email header, as found in /var/mail/someuser:
From [EMAIL PROTECTED] Sat Aug 28 01:01:57 2004 Return-Path: <[EMAIL PROTECTED]>
Now, if I download this mail via my POP3 account, and then move it into an IMAP account called "[EMAIL PROTECTED]" (using Thunderbird), I notice that the email header has been mangled to look like this:
From [EMAIL PROTECTED] Sat Aug 28 01:10:03 2004 -0400 Status: R X-Status: X-Keywords: X-UIDL: 2OJ!!%,B!!Unf!!K4B"! Return-Path: <[EMAIL PROTECTED]>
Is this normal behaviour, and will the salearn functionality still behave appropriately? The rest of the headers remain untouched, it is just the first few lines which are mangled during the (I assume) delivery process of the mail.
I am not using cyrus' IMAP, I'm using the default one with sendmail in /usr/sbin/imapd. Perhaps imapd is mangling the headers?
Some direction on this would be greatful, as I couldn't seem to find any answers on google or the spamassassin wiki.
Regards,
_keith
