[Spambayes] Trouble with the IMAP filter

[Tony Meyer]

> Here are the results from a run of sb 1.0rc1.  I had 3 
> messages (obvious spams) in INBOX, and nothing in any of
> the training folders.  Both db files existed prior to running.  
> The result was INBOX had 2 copies of each message, one 
> deleted with a spambayes ID only added to the header, the
> other looked to be the original message untouched (no sb
> headers added).
> A copy was also placed in INBOX.SPAM: Spam, with all the 
> normal sb headers added.

This is very strange.  There should be these three versions - the one with
just the SpamBayes header should be there, but deleted, and obviously the
spam folder one should be there.  However, the original should also have
been deleted, and the log says that it was:

  26:07.42 > DEN14 UID STORE 14891 +FLAGS.SILENT (\Deleted \Seen)
  26:07.52 < DEN14 OK STORE completed.

(and twice more for the other two messages).

I wonder if perhaps Evolution was open during this process and maybe
interfering?  It might not like other processes changing flags for messages
that it already knows about.  In any case, since:

[later message]
> When adding -e y to sb_imapfilter (and using version rc1),
> I'm having no errors.
> I'm also using the move_trained_X_to_folder settings now, 
> without issue.

I presume there's nothing to worry about for now.  Let me know if there are
further troubles.  I suspect this fixes it because it forces the deletion of
the original message and the temporary one, and Evolution doesn't ignore
that, while it might ignore a message it knows about having the /Deleted
flag set.  (But that's just a guess).

> I presume this is because evolution adds it's own headers to 
> a message, possible changing the message id, and sb saw it
> as new.

Yes - because IMAP doesn't actually let you add headers to a message
Evolution will be doing what we are: creating a new message with the
additional information and deleting the old one.  Doing this probably made
the message come up in the Recent list, and so SpamBayes would process it.
In proper operation, though, the messages would have the X-SpamBayes-ID
header (even after Evolution, presumably), and still be ignored, because it
knows that it has processed that message already.

=Tony Meyer

---
Please always include the list (spambayes at python.org) in your replies
(reply-all), and please don't send me personal mail about SpamBayes. This
way, you get everyone's help, and avoid a lack of replies when I'm busy.