That is not exactly the same thing, since with "milter", you are adding one more layer.
exim loads in spamassassin?
spamd_address = 127.0.0.1 783
Then you just add the spam rules on the mime ACL.
:D. spamassassin-milter = spamd. The milter is a running process that loads in spamc and is multi-threaded. That should make it equal :)
Yes, but the virtual mail backend of postfix is much less flexible, and so using vpopmail makes sense.
Interesting comment...for the umpteenthed time I must take a look at exim :P
The main point is that Exim ACLs really rock. You can do everything there, ranging from SPF, Greylisting, spamassassin and av, mixing them to your delight. I can easily say that I don't want to check for spam or greylist authenticated users, unless they come from a give network and are sending e-mail to domain X. You can use some very interesting variables on the acls for control, something like:
warn set acl_m2 = ${lookup mysql{GREYLIST_TEST}{$value}{0}}
and then I can test for it:
condition = ${if eq{$acl_m2}{0}{1}}
I like the if parts :). With postfix, I would have to chain stuff around to get this kind of behaviour.