Arvinn Løkkebakken wrote:
Theo Van Dinter wrote:
SpamAssassin 3.0.4 is released! SpamAssassin 3.0.4 contains several
important bug fixes and is highly recommended for use over previous
versions.
STILL http://bugzilla.spamassassin.org/show_bug.cgi?id=3949 isn't
included. Why? Am I the only one who thinks this is a
_very_serious bug?
Arvinn
Sorry I didn't read the comments in the bottom there. I have now.
However, I think it's quite odd that this isn't considered a crucial bug:
Some stats from my system:
Number of messages tagged as spam while having the ALL_TRUTED score:
# zcat mail-20050606.gz | awk '{if ($7~/Y/ && $10~/ALL_TRUSTED/) print
$0}' | wc -l
77
# zcat mail-20050605.gz | awk '{if ($7~/Y/ && $10~/ALL_TRUSTED/) print
$0}' | wc -l
80
# zcat mail-20050604.gz | awk '{if ($7~/Y/ && $10~/ALL_TRUSTED/) print
$0}' | wc -l
58
# zcat mail-20050603.gz | awk '{if ($7~/Y/ && $10~/ALL_TRUSTED/) print
$0}' | wc -l
69
Threshold is mostly set to 5.
In addition: the messages that was not clasified as spam because of
ALL_TRUSTED pulled the message below the threshold, therein the worst
cases where false positive spam messages scored < 0.1 (score of < 3.4
before ALL_TRUSTED) and SA autolearned them as ham.
4th and 5th of May I ran spamd the whole day with -D:
Number of unparseable headers:
# zgrep 'received-header: unknown format' mail-20050505.gz | wc -l
1176
# zgrep 'received-header: unknown format' mail-20050504.gz | wc -l
885
#
Most of these messages are likely to get misfired at by ALL_TRUSTED
since the unparseable header often is the only one that is not.
The system scans about 40k-45k messages per day.
I have found out what's "wrong" with most of the headers that doesn't
get recognized by SA and will file a bug-report about it though.
PS: This hasn't anything to do with the trustpath FAQ article.
Arvinn
