I have to wonder how much work is going to be required to bring qmail up to a modern standard. There's no doubt that the author is an extremely intelligent programmer with a deep focus on security. In the intervening period Postfix has really gained in popularity, and is shipped pre-configured out of the box on some major Linux distributions. It is the default MTA for Fedora/Redhat etc for some years now.

I wanted to like qmail, it was definitely an improvement over the complicated disaster that is Sendmail, however, it always seemed to require patches to get it to do things that Postfix could already do out of the box. In addition I didn't like the dependency on the inetd replacement thing (daemontools?). It just felt so un-ergonomic.

In addition, the Maildir mailbox format (while very cool in its own right) usually required third-party software to be patched manually, for example WU-IMAP.

A day or two before the September 11 attacks, the UNIX time_t value rolled over to 1 billion, and went from 9 digits long to 10 digits long. The moment this happened, a comparison function in the WU-IMAP maildir patches failed, and our IMAP server started sorting and displaying all our emails out of order.

After scratching our heads and poring over the source code for a couple of hours, a co-worker and I found the problem and submitted a patch to the maintainer.

Funnily enough, the imap-maildir page still references this event even today (it probably hasn't been updated in years):

http://www.davideous.com/imap-maildir/

Under the section "10-digit unix date rollover problem". But it drive home to me as a fairly young sysadmin at the time of the dangers (at least for the inexperienced) of using third-party patches just in order to enable a non-standard mailbox format (eg. Maildir).