Thanks for the hints - the main problem ist an architectural one, though. My machine is a UUCP leaf node, and I have currently no intention of changing this. Although this may sound 90's, it has a number of advantages - I don't need to keep my incoming SMTP port clean (with the downside that I cannot filter directly there), and I have a complete mail subdomain for myself.
It's actually a single "egrep -i -o -f " per mail. For each mail, egrep is forked exactly once, which means writing my own tool will not reduce the OS overhead.
I might give perl a try though, but I doubt that forking perl will be much faster than forking grep.
Yes sure, SA is also in the chain. But SA alone was never able to do the job even if I fed each mail that got through to sa-learn --spam (which I still do). This is why I started the local spam pattern file.
It's a personal account btw, maybe that was unclear.
Slashdot Mirror
You may want to try Boxcryptor https://www.boxcryptor.com/
Thanks for the hints - the main problem ist an architectural one, though. My machine is a UUCP leaf node, and I have currently no intention of changing this. Although this may sound 90's, it has a number of advantages - I don't need to keep my incoming SMTP port clean (with the downside that I cannot filter directly there), and I have a complete mail subdomain for myself.
It's actually a single "egrep -i -o -f " per mail. For each mail, egrep is forked exactly once, which means writing my own tool will not reduce the OS overhead. I might give perl a try though, but I doubt that forking perl will be much faster than forking grep.
Yes sure, SA is also in the chain. But SA alone was never able to do the job even if I fed each mail that got through to sa-learn --spam (which I still do). This is why I started the local spam pattern file. It's a personal account btw, maybe that was unclear.