> Simon Waters wrote:
>>>> I think you are making the set up over elaborate.
>>> I know that I can do the spam detection in my email client (thunderbird
>>> here). That's how I do it now and it works relatively well with only one
>>> PC. But I want to be able to access my mailbox from other PCs as well.
>>> That's why I want to setup an internal mailserver, where I can access
>>> the mailbox over imap. And then I could as well adding the spam/virus
>>> detection at the server level.
>> Pipe into spamassassin via .procmailrc ? Worked reasonably well here
>> for a long time before with braved greylisting and RBLs, and took
>> advantage of the fact we are terminating the SMTP transaction.
> The advantage of using postfix instead is that your mail is queued
> quickly, so fetchmail doesn't need to keep the connection with the
> pop/imap server open until mail is filtered. Now, this doesn't seem to
> be what the OP is asking for as he apparently wants to keep a copy of
> non-spam mail on the pop/imap server.
> If the OP can make sure the right error is given back to fetchmail, then
> procmail/maildrop is probably the right way. Otherwise, using an MTA (+
> forward mail to another address to be able to read it from other places)
> may be another solution.
Leaving the non-spam on the pop server is not absolutely required, but
rather a nice thing to have. Forwarding to another address has the
downside that every (non-spam) message has to be re-send, and my upload
speed/volume is rather limited by my isp.