Jason Haar wrote:
> On Tue, 2003-12-02 at 09:15, Dennis Duval wrote:
>> FWIW opinion:  the developers of SA are gods.   But they
>> need to recognize a few things and fix a few things for
>> ISPs who are trying to configure things on a site-wide
>> basis and keep uneducated uses from botching settings.
>> First, they need multiple thresholds, like low, high and
>> medium and let user assign them scores.   Second,
>> Subject line tagging.  *******SPAM******** and  the
>> levels thing of 47 stars for a score of 47.0 is pretty
>> worthless too for people using Outlook Express.  The
>> other problems deal with spamc.
>
> This is because OE doesn't have sophisticated filtering
> rules - right?

Exactly.   There is no capability for filtering on header information.

>
> I think you're onto something there. Have you talked to
> the SA people about this - you seem to have valid points.

I have not posted concerning the problem with tagging the subject line.  I
have posted on the SA list concerning the problem with -R on spamc.

>
> What about if the "***SPAM***" string added to Subject
> lines was more like "***SPAM-low***", "***SPAM-medium***"
> and "***SPAM-high***"? Could OE filter on that?
>

I believe it would work.  Anything that can be matched as a string is fine.
It cannot do numerical operations to evaluate the SA score.  I find a tag
like this too long and sort of unsightly.  Pushes the orginal text too far
to the right, especially if the subject line is already tagged with a long
list name, ie [Qmail-scanner-general].  But I'm being picky there.  The way
I am doing it now is {Spam* - 5.2} or
{Spam** - 8.3} or {Spam*** - 17.5}, for example.  This looks good, takes
little space and gives a good amount of usable info on the subject line.
Three stars is the max.  The little dash between the stars and SA score was
added as a necessary evil.  Good ole OE will drop the trailing space from a
message rule.  So, for example, if you want to delete everything with 2
stars or more, create a rule that looks for the string "{Spam* ", it will
drop that trailing space after the star and catch all three cases instead of
just one star.  With the dash separator, I can set up a rule "{Spam* -" and
it will catch only one star spams.  There are better ways of doing this,
like {8.3 Spam**}, but I had gone too far down the road when I realized OE
dropped trailing spaces, so I just threw in the dash.

> No - Qmail-Scanner operates correctly with spamc -
> perhaps you've misunderstood what it does? Don't forget
> "fast_spamassassin" runs "spamc -c" and only cares for
> the score - no other aspect of what headers/whatever that
> SA can do to the e-mail applies. If you want the "full
> force" of SA alterations applied, then you have no option
> but to run "verbose_spamassassin" - then you get all the
> extra headers/etc.
>
I agree that spamc -c works as advertised, qmail-scanner-queue.pl works
out-of-the-box with SA.  I wish it returned more info, but it works as
designed I think.  I believe that spamc -R is broken.  If you use it from
the command line with some sample spam and non-spam, it does not return the
information one would expect based on the documentation.  It also seems to
destroy the original message.  My hacked up version of qmail scanner uses
spamc -f , with no other option.  I'll be the first to admit I don't know
much about the way spamd and spamc interact, but I did do quite a bit of
testing of spamc from the command line without using qmail-scanner-queue.pl
at all.  I don't understand "fast_spamassassin" and "verbose_spamassassin"
as you used them above.

Dennis Duval



-------------------------------------------------------
This SF.net email is sponsored by OSDN's Audience Survey.
Help shape OSDN's sites and tell us what you think. Take this
five minute survey and you could win a $250 Gift Certificate.
http://www.wrgsurveys.com/2003/osdntech03.php?site=8

_______________________________________________
Qmail-scanner-general mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/qmail-scanner-general

Reply via email to