Les Mikesell wrote:
On Fri, 2006-02-17 at 15:45, Ask Bjørn Hansen wrote:
On Feb 17, 2006, at 9:43 AM, Les Mikesell wrote:
Wouldn't it at some point be simpler to run sendmail as the
front end since it already knows how to do this stuff?
It depends.
I run qmail-smtpd (with TLS and AUTH patches) on the client relays,
but if I had a user database integrated with qpsmtpd already it'd
probably be easier to set it up in qpsmtpd instead. Or if you want
other qpsmtpd plugins running, for example virus filtering on
outgoing mail... (the usual reasons to run qpsmtpd ;-) )
Yes, but MimeDefang does all the same stuff that you can do
in qpsmtpd - or if it doesn't it would be easy to duplicate
there, and it works with sendmail. And the things you can
control in sendmail's access file are handled more effeciently
than you can do it in perl. Qpsmtpd is a neat idea but there
is just a lot of functionality to duplicate to match sendmail
plus a milter.
Yes, you could use sendmail and a milter instead of qpsmtpd, or you
could use a milter with qpsmtpd. What is your point? What
functionality is missing in qpsmtpd to which you're referring? I would
argue that it would be easy to add anything to qpsmtpd that is missing.
I added LDAP authentication and recipient verification to qpsmtpd in
about 45 minutes, for example. The framework is there to do whatever
you want. It's also a bit easier to read than sendmail (in my opinion,
of course), from a configuration standpoint as well as a code standpoint.
I noticed you posted a while ago about using qpsmtpd's milters.. That
seems to be the best of both worlds for you: mimedefang with qpsmtpd
(which I would personally be interested in, too.) Did you have success
with that?
