No, I'm not trying to build a proper mail-server.
I collect email on my server from various (remote) mail-servers.
This is then processed by postfix/amavis/clamav/spamassassin .
Spam is marked by addition of [SPAM] or ***Spam*** to the Subject header,
as well as addition of several other headers.
As I understand it, the email is then passed through dovecot(?),
to ~/Maildir/cur/ .
Dovecot just does pop/imap _after_ the mail is delivered to mbox or Maildir/...
A simple way to do things is postfix->procmail->spamassassin . It works on a per
user basis the . And you can tweak spamassassin to sort by score :
spam
almost-certainly-spam
probably-spam
mbox
With a bit of training ( sa-learn --ham mbox / sa-learn --spam spam say
once a month ) you quickly get down to a few false positive/negative per year.
All the configuration is done in .forward and .procmailrc
Not sure how to do it with Maildir though ...
This must be a standard setup.
So how normally is spam dealt with, at this stage?
You seemed to be suggesting that it could be dealt with earlier, by amavis?
Or it could be left to the client MUA, KMail in my case?
What is the norm?
Ah - but there are so many "standards" .......
peter