On Sat, 18 Jan 2003 the voices made Kai Schaetzl write:
KS> Tony L. Svanstrom wrote on Fri, 17 Jan 2003 23:42:57 +0100 (CET):
KS> > Just picture you being an american
KS> > user of SA, suddenly all business e-mails you're getting during the=
KS> > get a small positive score simply because you get these e-mails dur=
KS> > local business hours... wouldn't work.
KS> Which I think I already said :-) If there are rules it doesn't mean t=
KS> are set to score by default at all. There could be 24 rules, one for =
KS> and an admin who feels that this could apply to his users/setup/timez=
KS> then simply add appropriate scores.
If they feel that they need it they can always add such a rule themselfs=
need to give them a gun and say "please don't shoot yourself and your use=
the foot", not when they with some thinking of their own could do it anyw=
less things that can go wrong equals less things going wrong equals less =
KS> The chance that a mail arriving at three in the night is spam is sign=
KS> higher than for one arriving at three in the afternoon. I'm quite sur=
KS> that general assumption.
[removed list of examples of non-spam arriving at night]
Good for you that you're sure about it, but it simply isn't true for me;=
if I assumed that no Date:-headers are wrong and compared them to the tim=
the e-mail was sent from I wouldn't get a pattern good enough to be used.
Although, in some cases you could find these patters if you looked at the=
I usually get e-mails from a limited few individuals; but those that I se=
e-mails to are already whitelisted, so those e-mails are never checked to=
if they're spam.
KS> announcements and what else - and there's whitelisting ... That's why=
The whitelist in SA is a hack... oh, sorry, meant feature *G*... used wh=
SA isn't doing what it should do; any and all features added with a "and
there's whitelisting"-note are broken by design.
KS> I still don't know if this would make a difference but we'd need to h=
ave a rule
KS> built in the SA source to check it out. It would simply need to use t=
KS> when the message is getting processed. I think that's better than usi=
ng the last
KS> Received header and parsing out the date from it (which could be done=
KS> change in the source).
Nah, can't use the time the e-mail is processed, not without messing thi=
for people using fetchmail and/or running SA on their local not-always-co=
computer and/or people reprocessing e-mails (maybe because SA was down, o=
doing scoring for a new version etc); you have to use the date in From_, =
oldest trusted Received-header.
\_@ @_/ \_@ @_/
| Per scientiam ad libertatem! // Through knowledge towards freedom! |
\O/ \O/ (c)1998-2003 tony@... \O/ \O/