Hello Matt, Andy,

Tuesday, June 8, 2004, 3:14:55 PM, you wrote:

MK> At 06:10 PM 6/8/2004, Andy Jezierski wrote:
>>My system has them in the original SA 2.63 distribution in 20_ratware.cf

MK> Ack, you're right.. my bad!

Still, your other suggestion is a good one -- if these rules cause
false positives for Andy, he could lower their scores so they don't
push the Washington Post emails over the edge.

Those rules still appear in http://www.spamassassin.org/tests.html,
which I believe somewhat reflects the 3.0 development test set at this
time.

Andy, you should probably also check bugzilla, search for these rule
names, and see if there's a bug submitted for these types of false
positives.  If so, upload a couple of copies of your false positives
to it. If not, create a new bug and upload. That should get your
emails added to the dev corpus in time for them to be taken into
account during the next scoring session, so 3.0 will avoid this
problem.

If for some reason you can't work through bugzilla, forward a couple
of those false positives to me as attachments in a zip file (my email
hosts won't accept eml attachments), and I'll get them to the devs.

Thanks.

FYI, I've just signed up for a couple of Washington Post newsletters
myself, so I may have my own examples in a few days.  (It's been a
long time since I've had to do any work on false positives.)

Bob Menschel

Reply via email to