Re: [exim] header_syntax

On Fri, 4 Mar 2005, Michael F. Sprague wrote:

> Marc Sherman <msherman@???> wrote:
> >
> > The sad fact is, on today's internet, header_syntax-based rejection is
> > not viable. You really shouldn't use it for anything more than scoring.
>
> I've had a similar experience.

Interesting. That prompted me to look at our logs.

In 10 weeks of logs, it seems we've rejected a grand total of 2 mails
on header syntax errors. One is an obvious spam; the other appears to
be otherwise bona-fide but has attempted to set an unusable Reply-to:
address. One might have hoped that a serious particle-physics
research project would wish to get informed that they are asking to
get replies at an invalid address, rather than us silently accepting
their mail. But I could be wrong...

This feels very different from a year or two back, when we were
rejecting spams at a rate of knots on their defective header syntax.

Of course, we have plenty of other tests which are rejecting a
reasonable proportion of spams before we even reach the DATA ACL: we
would never know whether their headers had syntax errors.