A couple of months ago, Ian Jackson started working with qmail. He
observed what he thought was a bug---namely, qmail rejects
To: @one.host:iwj@???
On the basis of this, he asserted on exim-users that qmail's RFC 822
header parsing is ``very ropey,'' that qmail has ``outright bugs,'' and
that qmail's ``implementation quality is poor.''
Meanwhile, he let me know about the ``bug.'' I pointed out that his
header field violated RFC 822.
He apologized to me for his error, but apparently he never apologized on
exim-users for misleading all of you.
I'm pleased to report that the qmail beta test has gone very smoothly.
The quality of the code is demonstrated by the tiny number of bugs that
have shown up in five months of heavy use since the initial release: 21,
of which 8 affected users. qmail has never helped an intruder and has
never lost a message; it is the fastest available Internet MTA.
Ian also stated that qmail ``has a number of unexpected behaviors.''
Other than the above ``bug,'' the only thing he's complained to me about
is how qmail handles address quoting. There are two approaches here: the
MMDF approach, which qmail follows, and the sendmail approach. Since the
sendmail approach violates all relevant standards---for some examples,
see
ftp://koobera.math.uic.edu/pub/docs/1996-mail-errors---I think I'll
stick with the MMDF approach.
If there are other ``unexpected behaviors,'' I'd be happy to hear what
they are, so that I can either improve the interface or improve the
documentation.
Now, I suppose you'd all like to hear about exim, not qmail, so I have a
quick question: What protection does exim have against system crashes?
It seems to me, from a quick glance through the code, that messages can
be lost, misdirected, or corrupted. If so, exim is completely unsuitable
for production use; anyone who deploys it on the Internet is violating
RFC 1123, section 5.3.3, and violating the trust that users have in the
mail system. Will this be fixed before exim is released?
---Dan
