MailScanner 5 on Gentoo

[Michael Weiser]

Hey guys and especially Jerry,

I know I'm ridiculously late to the party here but I still need to get
this off my chest: Jerry, you're my hero! :)

I just updated my Gentoo-based MailScanner installation from something
ancient to MailScanner 5.0.3-7 and just couldn't believe how insanely
sane this new MailScanner is. So I went right ahead and reworked the
Gentoo package to reflect that.

I couldn't resist sticking with Gentoo's /etc/conf.d for what
/etc/MailScanner/defaults does now. Also I couldn't bring myself to even
try to install ms-init and stuck with an updated openrc
start-stop-daemon-based init script. And finally I patched around in
ms-check because I couldn't really wrap my head around the benefit of
having a stopped_lockfile *and* a run_mailscanner setting. I've tried to
formulate my thoughts on all this in a README.Gentoo
(https://584524.bugs.gentoo.org/attachment.cgi?id=453228).

Anyway, I found an already open bug on updating the Gentoo package and
stuck my stuff into it. There's lots of explanations and rationale in
the ebuild, the ms-check patch and the README.Gentoo if anyone's
interested: https://bugs.gentoo.org/show_bug.cgi?id=584524.
Now we'll have to see if it gets picked up.

Here are some additional points for your consideration hidden away as
comments in the ebuild:

On /etc/conf.d/MailScanner (aka /etc/MailScanner/defaults):
- nothing is using the ms_re2c setting (ms-update-sa finds it on its own)
- ms_etc isn't used anywhere

On calls to ms-init in /usr/sbin/ms-update-bad-emails and
/usr/sbin/ms-update-sa:

change restarts via ms-init after e.g. rules updates into
reloads via /etc/init.d/MailScanner
a.) reloads should be enough because they restart all children and the
    parent MailScanner doesn't do anything so doesn't need the update
b.) this way we'll never accidentally start MailScanner when it's
    supposed to be switched off without mucking about with
    $run_mailscanner and $stopped_lockfile

What I mean there is: A simple SIGHUP instead of ms-init restart
should do the trick and be safer at the same time, shouldn't it?

Feel free to bash me (a bit :) for stuff I got wrong.

And thanks again! This is awesome! :)
-- 
Michael