| Moe Trin 2007-10-03, 7:12 pm |
| On Tue, 02 Oct 2007, in the Usenet newsgroup alt.os.linux.debian, in article
< mqydnSiXEbufAZ_anZ2dnUVZ_rSinZ2d@bresnan
.com>, sk8r-365 wrote:
>Horton heard a Who named Mark South saying:
[vbcol=seagreen]
That's a bit on the low side. For _most_ of the 84 groups I try to scan
daily, I allow up to 4 groups. OK, inexpensive killing of cross-posts:
% When a newsreader contacts the newsserver, it requests a so-called Xover,
%which gives a subset of the headers for the requested article. Headers
%present in this list include From:, Subject:, References:, Date:, Bytes:,
%Lines:, Message-ID: and Xref:. Newsreaders use this to compose their
%header list, and to perform threading. OTHER headers can be queried
%from the server, but this is expensive and should be avoided over the
%Internet.
%
%Killfile rules based on these headers can be implemented quite efficiently,
%since this information has to be requested for normal operation anyway.
so given the choice of 'From:', 'Subject:', 'References:', 'Date:',
'Bytes:', 'Lines:', 'Message-ID': and 'Xref:':
-----
From: sk8r-365 <sk8r-365@sk8r.debian.etch.invalid.org>
Subject: Re: (snip) slrn scoring help needed
References: <0aadnfMN8ZJErZ_aRVn_vwA@giganews.com>
< QNadnUok1aOp35_anZ2dnUVZ_hKdnZ2d@bresnan
.com>
<47025d2e$1_4@news.bluewin.ch>
Date: Tue, 02 Oct 2007 14:29:06 -0500
Bytes: 2115
Lines: 25
Message-ID: < mqydnSiXEbufAZ_anZ2dnUVZ_rSinZ2d@bresnan
.com>
Xref: number1.nntp.dca.giganews.com alt.os.linux.debian:7519
-----
the latter is the only one with any indication of newsgroup. From
another posting today in another newsgroup:
-----
Xref: number1.nntp.dca.giganews.com news.announce.newgroups:6172
news.groups:430617 news.groups.proposals:3031
-----
you said two groups are OK? Try
Xref: :.*:.*:
which should kill posts where the Xref: header contains three colons.
[vbcol=seagreen]
but no hints?
[vbcol=seagreen]
>I have it figured out - filtering by keyword ... of course, when I
>apply it I won't see further messages for this post due to word
>matching; which would have included this reply, except that I changed
>the Subject so as to *not* have a match:
which is why filtering by keyword isn't the solution _here_ (for the
problem of cross-posters).
>[*] % filter for all NG
>Score: -9999 % absolute death of article
Score: =-9999 %kill it, and don't bother searching for more score terms
According to the 'changes' file, that mode was included way back in
slrn-0.8.0. You may also want to think about 'Score:' verses 'Score::'.
The single colon is a logical AND indicator so all conditions/rules for
that score have to be true. The double colon is a logical OR indicator,
and _any_ condition/rule that matches sets the score.
>Subject: market % any match of this word and _poof_ !
[compton ~]$ grep -vE '^([%\[ ]|Score|$)' /var/spool/slrnpull/score | cut
-d' ' -f1 | sort | uniq -c | column
899 From: 7 Message-ID: 78 Xref:
2 Lines: 282 Subject: 2 ~Subject:
[compton ~]$
Imagine what that would look like if I were grumpy instead of my nice
cheery disposition. (That also is an indication why I use the '=' in
score values.)
Old guy
|