On 1/12/07, Hugo Monteiro <[EMAIL PROTECTED]> wrote:
Hello all,
I have a doubt about DSPAMs behaviour when submitting the same message
multiple times for retrain with the same classification.
I've wondered about this, but never tested it.
Scenario:
I'm using dovecot as pop/imap server and there is a nice dspam plugin
for it that allows instant retrain with the simple move of messages
to/from a defined spam folder, while using imap.
All messages (spam+nonspam) are being delivered to the users INBOX,
since the user has the possibility to retrieve the messages using pop.
Messages are being tagged in the headers only, and for users to be able
to separate messages they have configured a filter in their clients.
A possible problem, depending on DSPAMs behavior, is that if a message
gets delivered tagged as spam, the client will filter and throw it in a
spam folder. For pop users that's not a problem. But for imap users, the
message will be moved to the spam folder, where the dovecot plugin acts
as trigger. So, each message delivered as spam, will be dropped in the
users spam folder, making dspam retrain that same message as spam.
For imap users, are you able to do server side filtering? While this
isn't a direct "fix" for dspam, the way the plugin is written, is to
take advantage of server side filtering, so spam messages start in the
spam folder, and then the only way to retrain it as spam without
retraining it as ham first, is to do a complicated move to trash, move
to inbox, then move to spam again (seeing as moves between trash and
spam aren't normally processed as a retrain).
I have no idea though about how dspam should, or does handle
retraining the same message over and over. (I use dovecot LDA with
sieve filtering in my setup).
Tim
--
Linux Counter user #273956
Don't email [EMAIL PROTECTED]
