Matt Kettler a écrit :
That really makes me wonder if DJB is intentionally trying to make plain
qmail an unusable POS so that nobody will be able to use it without
patching. Or is it a purely accidental side effect?
Any MTA that can ever fail to report the local hostname in the "by"
clause scares me. Ditto for the source IP delivering mail (in the case
of SMTP).
I would then say:
Any filter that can ever fail to filter mail based on the client, the
sender and/or the recipient scares me.
but then SA works this way....
don't be too harsh with qmail (I don't use it and don't care). most
software requires that people understand how it works. the fact that you
can run it without understanding how it works shouldn't be used against
that software.
In the present case, the idea of letting another server manage the
network connection isn't bad. It's even more sound from an architecture
stand point. needing to copy the same code all over again (daemonize,
listen, accept, ...) is just silly. and we see bugs all over again.
I have a lot of things to say against qmail, but not here. just because
a silly guy found a way to generate silly Received headers, we're not
going to blame an MTA.
It should at least be really hard to misconfigure an MTA to the point it
does that. Like hacking the source code hard, or at least mucking around
with several options the manpage or conf file warn you not to touch.
This means the MTA should have things hardcoded, which is really bad
design. an MTA gets its IPs from the system. This may be the kernel or a
master process. In both cases, bad things can happen.
now it seems most people would blame qmail for anything, sometimes
because of real issues, but too often because of emotional "reasons".
