I just fed sa-learn with a maildir of spam emails and it said it learnt from
them but bayes is reporting a significantly less number of learnt spams.
OK, since noone could offer any ideas, just for kicks I renamed the
bayes* files from the .spamassassin directory and fed it my ham and spam
On Sat, Sep 06, 2003 at 11:54:22PM -0500, Bill Polhemus wrote:
What's the problem with just putting
*:catch-all mailbox
In your ALIASES file?
That would catch all mail with an unknown To: field,
not with unsubscribed 'From:' fields.
I suggest to use procmail to filter your mail
On Sat, 2003-09-06 at 21:03, Theo Van Dinter wrote:
On Wed, Aug 27, 2003 at 01:29:52AM +0100, Yorkshire Dave wrote:
0=
0=Q
A quick roundup. Can anyone fill any gaps?
2queers cipher
--
in html comment
plain = abcdefghijklmnopqrstuvwxyz0123456789-_.@
coded = dO,9F_ [EMAIL
i ran the follwoing command several times withing seconds, why did it
learn from different messages every time?
[EMAIL PROTECTED]:~ sa-learn --spam --dir
evolution/local/Inbox/subfolders/spam/
Learned from 2 messages.
[EMAIL PROTECTED]:~ sa-learn --spam --dir
1 How do I write my own rules? is there a tutorial someplace?
2 Spam changes with time, which means the tests should be updated. How to I
get the current tests performed? ie can I simply copy the new .cf files
from a newer version of spamassassin? or do I have to reinstall
completely? or is
SA thinks mail from TheBat 2.0 is spam:
Date: Sat, 06 Sep 2003 13:17:08 -0400
From: Stefan Tanurkov [EMAIL PROTECTED]
Subject: Re: 2.00: bug in Scheduler
To: Peter Fjelsten [EMAIL PROTECTED]
[...]
Message-id: [EMAIL PROTECTED]
Organization: RITLABS Inc.
MIME-version: 1.0
X-Mailer: The
On Sunday, Sep 7th 2003 at 12:39 +0200, quoth Carlo Wood:
=On Sat, Sep 06, 2003 at 11:54:22PM -0500, Bill Polhemus wrote:
= What's the problem with just putting
=
= *: catch-all mailbox
=
= In your ALIASES file?
=
=That would catch all mail with an unknown To: field,
=not with unsubscribed
On Sun, Sep 07, 2003 at 08:40:38AM -0400, landy wrote:
i ran the follwoing command several times withing seconds, why did it
learn from different messages every time?
impossible to say without seeing the messages. I'd bet it's related to
messages w/out message-id headers.
--
Randomly
On Sun, Sep 07, 2003 at 02:11:58PM +0200, Roman Katzer wrote:
Since Stefan Tanurkov is the head programmer of TheBat, I don't think he
has to fake his mail client ;-)
please post bugs on bugzilla.spamassassin.org, otherwise we may not see
them and/or remember about them.
--
Randomly Generated
On Sunday, September 7, 2003, 18:11:29, Theo Van Dinter wrote:
please post bugs on bugzilla.spamassassin.org, otherwise we may not see
them and/or remember about them.
Thanks for the pointer, I just did.
Roman
--
Roman Katzer, Aachen, Germany
You know those people who won't take no for an
On Sun, 7 Sep 2003, Steven W. Orr wrote:
If someone sends *me* a copy of Sobig:F I catch it as spam just fine.
But if it's sent to listaddr, then it comes in with no processing from
SA. What I want is to say that *all* incoming mail needs to go through
SA.
This probably means that your
I've gotten a few spams which were scored appropriately, such as:
the example below, which got 12.7 spampoints.
BUT: the X-Spam-Status header indicates a score of 0.7 points! Why is this
very partial sum being used for this header?
This problem happens for about 1% of the spams I get. Have
On Sun, Sep 07, 2003 at 10:06:51AM -0700, Jerry Glomph Black wrote:
I've gotten a few spams which were scored appropriately, such as:
the example below, which got 12.7 spampoints.
BUT: the X-Spam-Status header indicates a score of 0.7 points! Why is this
very partial sum being used for
sa-learn is in every SA version 2.50 and up. You can get the
latest stable from the SA web page.
On Sat, 6 Sep 2003, Scott wrote:
Theo Van Dinter wrote:
On Sat, Sep 06, 2003 at 08:04:22PM -0700, Scott wrote:
My redhat 8 box with courier and spamassassin doesn't have sa-learn
Hello, everybody
As a part of my diploma thesis I developed an AI to sort Mails into
Ham/Spam.
The AI alone was able to get about the same results as Spamassassin.
Which means it is very good.
I released it as Open Source and it is downloadable at
http://spam-fitz.de
You just overwrite a few
That's fine, but why is the score number not the same? I've been using procmail
to read the X-Spam-Status numerical value, and thus decide where to deliver a
given message. Is this header now irrelevant?
On Sun, 7 Sep 2003, Theo Van Dinter wrote:
On Sun, Sep 07, 2003 at 10:06:51AM -0700,
On Sunday, Sep 7th 2003 at 10:02 -0700, quoth Bart Schaefer:
=On Sun, 7 Sep 2003, Steven W. Orr wrote:
=
= If someone sends *me* a copy of Sobig:F I catch it as spam just fine.
= But if it's sent to listaddr, then it comes in with no processing from
= SA. What I want is to say that *all* incoming
Folks,
Looking for conventional wisdom on handling the bayes db in a centralized
production environment. I don't intend to do autolearning, but instead to
feed a carefully selected set of messages into a central SA instance,
creating the db via sa-learn. I then want to take that database and
I moved my site from one of my servers to another one of my servers over
the weekend. Everything was copied and user permissions were reset just
in case.
I am now getting the following errors in my log files...
Cannot open bayes_path /home/edwardam/.spamassassin/bayes R/O:
Cannot open bayes_path
This message made is past SpamAssassin without setting off ANYTHING? (Score
of 0.0.)
Return-Path: [EMAIL PROTECTED]
Received: by bluehome.net (CommuniGate Pro PIPE 4.1.3)
with PIPE id 360205; Sun, 07 Sep 2003 12:05:04 -0700
Received: from [68.198.199.174] (HELO atlantech.fr)
by bluehome.net
#Higher scores for bayes
score BAYES_80 2.50
score BAYES_90 4.00
I think that I messed up with the BAYES ones.. BAYES_80 isn't adding the
2.5 points like I wanted. =/
2.60 signifigantly increases default bayes scores.
---
This sf.net
No. I use CGPSA to get the messages to CommuniGate Pro.
I think it's interesting that it Identified non-spam (0.0/5.0) for
default in 0.0 seconds
Seems too quick. Anyway, here's the message.
12:12:30.24 4 EXTFILTER(CGPSA) out(25): 54 FILE Queue/360223.msg\n
12:12:30.28 2 PWD-00035([127.0.0.1])
yes thanks, but how does SA know which rbl checks to do... ie how do I
configure what services I want SA to check?
thanks again
3 I don't see any evidence that RBL checks are being made in my spam
headers. How can I make sure the RBL checks are being done?
enable them in local.cf:
On Sun, 07 Sep 2003 15:55:40 -0400
Burt Juda [EMAIL PROTECTED] wrote:
You can pipe it thru 'formail' to nuke the headers that SA added, like
so:
cat FILE | formail -I'X-Spam-Checker-Version:' -I'X-Spam-Level:
\
-I'X-Spam-Status:' -ds sa-learn --[s]ham
The '-ds'
Thanks to Theo Van Dinter for his polite patience following my 'bug' report.
It was a problem, but related to a known bug in 2.60rc3, that the PREFIX of
installation of the local config files is not being set to the usual
/etc/mail/spamassassin, but rather /mail/spamassassin.Thus my local.cf
At 09:05 AM 9/7/03 -0400, [EMAIL PROTECTED] wrote:
1 How do I write my own rules? is there a tutorial someplace?
Yes there is a tutorial someplace, or a least a fairly decent howto on the
subect.
http://mywebpages.comcast.net/mkettler/sa/SA-rules-howto.txt
2 Spam changes with time, which means
for example use these:
tflagsBL_BLKHOLE_INTERSILnet
headerBL_BLKHOLE_INTERSIL
rbleval:check_rbl('intersil', 'blackholes.intersil.net.')
describeBL_BLKHOLE_INTERSILSpammer list in:
blackholes.intersil.net (Basis entry)
score
At 12:08 7/09/2003 -0700, Jason wrote:
This message made is past SpamAssassin without setting off ANYTHING? (Score
of 0.0.)
[snip]
Very odd. You must have something wrong with your setup there, as I see
heaps of spams identical to that one that score quite highly... (somewhere
between 10 and 20
[EMAIL PROTECTED] wrote:
1 How do I write my own rules? is there a tutorial someplace?
2 Spam changes with time, which means the tests should be updated. How to I
get the current tests performed? ie can I simply copy the new .cf files
from a newer version of spamassassin? or do I have to
Bob Proulx writes:
I prefer the benevolent dictator approach, where I get to choose and
to ignore the dictators of my choice.
Yeah, me too -- a web of trust, basically. If I know that J. Random
Listing Guy is publishing a bl, which a public reviews site measures
as listing 80% of spam with
Robert Menschel writes:
* What specific business are your people in? Much of our mail deals with
lumber and building materials. ...
Ssh! This is a public forum! Don't mention the lumber cartel!!
http://lumbercartel.freeyellow.com/
;)
--j.
trey valenta writes:
On Fri, Sep 05, 2003 at 04:56:45PM +0200, Jochen Tuchbreiter wrote:
Did any of you analyze (profile) where most of the CPU in spamassassin
is spent?
I ran spamassassin --lint through Perl's profiler back in July. I know
this isn't the same as checking mail for
Yorkshire Dave writes:
I don't know if I should be adding these incomplete ciphers to
detoken.pl, they're useful to a lot of people before they're complete.
Any opinions?
I'd say so -- you could leave the unknowns as ., and count the dots once
the RE is generated; if there's too many (ie. 3 or
Jason writes:
No. I use CGPSA to get the messages to CommuniGate Pro.
I think it's interesting that it Identified non-spam (0.0/5.0) for
default in 0.0 seconds
Seems too quick. Anyway, here's the message.
I would guess that's not finding the config files -- if there's no
rules, it'll run
Peter Kiem writes:
I just fed sa-learn with a maildir of spam emails and it said it learnt from
them but bayes is reporting a significantly less number of learnt spams.
OK, since noone could offer any ideas, just for kicks I renamed the
bayes* files from the .spamassassin directory and
Diffenderfer, Randy writes:
Looking for conventional wisdom on handling the bayes db in a centralized
production environment. I don't intend to do autolearning, but instead to
feed a carefully selected set of messages into a central SA instance,
creating the db via sa-learn. I then want to
Okay I did check the output of `file` for the bayes* files and it was
GDBM, I removed them and re-created them. They are now DB files and
everything is working.
Thanks to #spamassassin.
On Sun, 2003-09-07 at 13:14, Edward Muller wrote:
I moved my site from one of my servers to another one of my
trey valenta writes:
On Sun, Sep 07, 2003 at 08:37:59PM -0700, Justin Mason wrote:
BTW I would suggest instead profiling spamd -- --lint is not the
case you want to optimize for ;)
Thanks, I'll try that. I realized --lint isn't the ideal test case, but
made the assumption (which might be
On Sun, Sep 07, 2003 at 08:37:59PM -0700, Justin Mason wrote:
BTW I would suggest instead profiling spamd -- --lint is not the
case you want to optimize for ;)
Thanks, I'll try that. I realized --lint isn't the ideal test case, but
made the assumption (which might be wrong) that Conf::_parse
At 20:37 7/09/2003 -0700, Justin Mason wrote:
trey valenta writes:
On Fri, Sep 05, 2003 at 04:56:45PM +0200, Jochen Tuchbreiter wrote:
Did any of you analyze (profile) where most of the CPU in spamassassin
is spent?
I ran spamassassin --lint through Perl's profiler back in July. I know
40 matches
Mail list logo
