Re: Alternate checks for reinjected mail

Magnus B=E4ck wrote:
> On Tuesday, May 24, 2005 at 20:11 CEST,
> Jorey Bump <list@joreybump.com> wrote:
>=20
> [...]
>=20
>>My content filter, spamfilter.sh:
>>
>>#!/usr/bin/bash
>>/usr/bin/spamc | /usr/sbin/sendmail -i "$@"
>>exit $?
>=20
> This is not robust. What happens if either of the forks fail?

Apparently, the mail gets delivered without filtering. At least this is=20
what happens when spamd is not running. I think if the sendmail binary=20
failed, it would indicate a more serious problem that I could not easily=20
ignore.

But this is partly what spurred my original inquiry: I'd like to avoid=20
redundant checks not only for efficiency, but to avoid any new condition=20
that might arise from SpamAssassin's modifications to the original messag=
e.

I haven't encountered any problems so far, and I am thrilled with both=20
the performance and quality of SpamAssassin, but I am definitely open to=20
suggestions for improvement.