The 03.05.25 at 15:22, Jon Clausen wrote:
-and I haven't even *received* any spam since!!! ;)
Good! :-)
It was nice while it lasted... but they're back at it now :P
:-( Almost all the spam I get goes to this address... I wish I could fake it.
Speed was the main factor in my wanting to use the daemon/client rather than spamassassin.
The rule I have in .procmailrc is:
I put it in /etc/procmailrc instead. Just in case I ever get any users (other than myself) on that server.
It doesn't work for me: "/etc/procmailrc" is not even read, it is ignored, ever since I changed to postfix. Procmail is executed by user "nobody", so that it can not read that file, nor "/root/.procmailrc"
:0fw | /usr/bin/spamc
Hmmm... I'd better go change that line, so it's got the full path instead of relying on $PATH...
I think it will run faster as well :-?
Some people send it to dev/null directly, but I don't trust it so much. I have some legitimate emails marked as spam now and then...
Indeed. I get the impression that SA 2.5 is smarter than 2.3, but in any case I like it that it now puts the report in the body of the mail, rather than keeping it in the headers.
I'm not sure I like it better... maybe when I get used to it :-) It is smarter, true. But I prefer to play safe.
I'll have to read the docs carefully, I saw something about sending spam that passes the filter somewhere so that it is detected next time. I wonder if that address is local or remote :-?
You mean sa-learn ?
AFAICT one needs a fairly large library of spam/ham before sa-learn has much to add... (?)
No, not that, or maybe yes :-? I don't really know. I mean just someway to tell it that that email it passed as "good" was really "bad". Also, I'd like a way of automatically add the senders addresses used for spam to a local database so that next time they are rejected before download - like one I got the other day claiming to be from M$ support, and containing a 70Kb executable, obviously malign :-} I wonder if somebody falls to that trick? -- Cheers, Carlos Robinson