[Users] Filtering spam difference - auto v. manual

[Pierre Fortin]

Hi guys,

Hope I'm not missing something obvious; but I recently created a top
level domain (TLD) filter to catch some spam messages which are not yet
learned by spamd -- it colors these messages to distinguish them from msgs
processed automatically by spamd.

The incoming messages are filtered as expected, and moved to my SPAM
folder; but if I get a message with a new TLD sourcing mostly spam, I add
the new TLD to the rule, then manually run 'Tools->Filter selected
messages' on the new msg. Everything occurs as expected, except for the
move to the SPAM folder -- both auto and manual filter log entries are
identical, ending with:
[10:18:35] processing rule 'SPAM-tld' [ from regexpcase ".bid>" | from
regexpcase ".date>" | from regexpcase ".download>" | from regexpcase
".stream>" | from regexpcase ".top>" | from regexpcase ".trade>" | from
regexpcase ".tv>" color 7 mark_as_spam move "#mh/Mailbox/inbox/SPAM" ]
[10:18:35] checking if message matches [ from regexpcase ".bid>" ]
[10:18:35] checking if message matches [ from regexpcase ".date>" ]
[10:18:35] checking if message matches [ from regexpcase ".download>" ]
[10:18:35] checking if message matches [ from regexpcase ".stream>" ]
[10:18:35] checking if message matches [ from regexpcase ".top>" ]
> message matches
[10:18:35] applying action [ color 7 ]
[10:18:35] applying action [ mark_as_spam ]


   Aside: I was careful to make move the last action; but I see in the
   --debug log that sa-learn is called, so the move is redundant. I know I
   can just hit the Spam button which will move the message to SPAM; but
   manually filtering was to ensure my filter was correct.

Any reason for the move not to occur when manually filtering?  Bug?

Thanks,
Pierre