Inexplicable all.SCORE analysis

Inexplicable all.SCORE analysis

[Karl Kleinpaste]

[-- Attachment #1: Type: text/plain, Size: 440 bytes --]

I'm very confused.  q0.10 is insistent that the article below matches
this all.SCORE entry, intended to squash excessive cross-posting:

 ("xref"
  (".* .* .* .* .* .* .* .*" -10 nil r))

as indicated by the score trace:

(".* .* .* .* .* .* .* .*" -10 nil r)  ->  all.SCORE

There's nothing in here to justify such a perspective, but q0.10's
belief is persistent, over Gnus quit/restart and fresh XEmacs (19.14)
start up.

Clues, anyone?


[-- Attachment #2: Type: text/plain, Size: 448 bytes --]

914	AOL and TCP/IP	c.m.manson@eqcity.ktb.net (C M Manson)	Sat, 20 Sep 1997 06:58:00 GMT	<970923100046611@eqcity.ktb.net>		1059	14        	Xref: news.jprc.com comp.protocols.tcp-ip.domains:914 	Keywords: dos aol win internationally earthlink windholes computer internet arachne drivers program quicken running access enough friend insert packet switch hello james liked mouth point back blue data echo foot need open slip wave work ppp qwk tcp thx


[-- Attachment #3: Type: message/rfc822, Size: 1059 bytes --]

From: c.m.manson@eqcity.ktb.net (C M Manson)
Subject: AOL and TCP/IP
Date: Sat, 20 Sep 1997 06:58:00 GMT
Message-ID: <970923100046611@eqcity.ktb.net>

Hello 

I have a friend (not me, I have Earthlink and DOS) who has AOL, and 
Win95. He's going to switch back to Win3.11 and DOS because he saw a DOS 
Internet program on my computer that he really liked (Arachne). My point is,
is there any way to get DOS PPP or SLIP packet drivers to work with AOL? We'd
like to get this up and running enough so that he would only need to go into
Windholes to access his Quicken data. 

Thx James 
DOS TCP/IP

... Open mouth, insert foot, echo internationally.
___ Blue Wave/QWK v2.12

Re: Inexplicable all.SCORE analysis

[SL Baur]

Karl Kleinpaste <karl@jprc.com> writes:

> [1  <text/plain; US-ASCII (7bit)>]
> I'm very confused.  q0.10 is insistent that the article below matches
> this all.SCORE entry, intended to squash excessive cross-posting:

>  ("xref"
>   (".* .* .* .* .* .* .* .*" -10 nil r))

> as indicated by the score trace:

> (".* .* .* .* .* .* .* .*" -10 nil r)  ->  all.SCORE

> There's nothing in here to justify such a perspective, but q0.10's
> belief is persistent, over Gnus quit/restart and fresh XEmacs (19.14)
> start up.

> Clues, anyone?

The one unusual thing I see is the additional Keywords: entry.  Does
Gnus handle fields beyond Xref: (which is the last field on the local
news server) properly?  If it's getting tacked on to the Xref, I can
see how the regexp above would match.

> 914	AOL and TCP/IP	c.m.manson@eqcity.ktb.net (C M Manson)	Sat, 20 Sep 1997 06:58:00 GMT	<970923100046611@eqcity.ktb.net>		1059	14        	Xref: news.jprc.com comp.protocols.tcp-ip.domains:914 	Keywords: dos aol win internationally earthlink windholes computer internet arachne drivers program quicken running access enough friend insert packet switch hello james liked mouth point back blue data echo foot need open slip wave work ppp qwk tcp thx

$ telnet newshost nntp
Trying 199.74.141.7...
Connected to newshost.
Escape character is '^]'.
200 zinger.callamer.com InterNetNews NNRP server INN 1.5.1 17-Dec-1996 ready (posting ok).
group comp.protocols.tcp-ip.domains
211 43 11898 11942 comp.protocols.tcp-ip.domains
xover 11939
224 data follows
11939   AOL and TCP/IP  c.m.manson@eqcity.ktb.net (C M Manson)  Sat, 20 Sep 1997 06:58:00 GMT   <970923100046611@eqcity.ktb.net>                1144    14              Xref: zinger.callamer.com comp.protocols.tcp-ip.domains:11939 
.

Re: Inexplicable all.SCORE analysis

[Lars Magne Ingebrigtsen]

Karl Kleinpaste <karl@jprc.com> writes:

> I'm very confused.  q0.10 is insistent that the article below matches
> this all.SCORE entry, intended to squash excessive cross-posting:
> 
>  ("xref"
>   (".* .* .* .* .* .* .* .*" -10 nil r))
> 
> as indicated by the score trace:
> 
> (".* .* .* .* .* .* .* .*" -10 nil r)  ->  all.SCORE

[...]

> 914	AOL and TCP/IP	c.m.manson@eqcity.ktb.net (C M Manson)	Sat, 20 Sep 1997 06:58:00 GMT	<970923100046611@eqcity.ktb.net>		1059	14        	Xref: news.jprc.com comp.protocols.tcp-ip.domains:914 	Keywords: dos aol win internationally earthlink windholes computer internet arachne drivers program quicken running access enough friend insert packet switch hello james liked mouth point back blue data echo foot need open slip wave work ppp qwk tcp thx

Currently Gnus uses everything in the last slot as the Xref header.
Which means that you have a very long one.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: Inexplicable all.SCORE analysis

[Karl Kleinpaste]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:
> Currently Gnus uses everything in the last slot as the Xref header.
> Which means that you have a very long one.

(*sigh*  I suppose I should have expected that.)

No, it doesn't, in my case, as I've got a patch to handle Keywords: in
overviews, auto-generated by INN (a patch I wrote for $inn/innd/art.c,
which is now part of 1.6bN).  For a contrary example, the article
showing this overview does not score-down on a misguided Xref: analysis.

435	US-PA-PGH-Records Management Systems Consultant	nescopgh@nauticom.net (NESCO Technical Services)	22 Sep 1997 21:58:19 GMT	<606plr$5ac@news.psc.edu>		3341	66	Xref: news.jprc.com pgh.jobs.offered:435 	Keywords: management nesco document company systems page consultant experience pittsburgh service please resume home http mail plus www manufacturing assignment technical nauticom nescopgh services records windows timing visit wish faq fax net implementation compensation construction nescoservice organization experienced information knowledable monroeville opportunity confidence responding specifying technology candidate computing corporate deploying designing expertise immediate important 101

This item scores up via the local scorefile, and doesn't trip the xref
entry in all.SCORE.  The total *Score Trace* buffer for this article:

("pitt\\|pgh" nil nil r)  ->  SCORE

Anyhow, something else funny is going on, and I'll chase it down somehow.

Re: Inexplicable all.SCORE analysis

[Lars Magne Ingebrigtsen]

Karl Kleinpaste <karl@jprc.com> writes:

> Anyhow, something else funny is going on, and I'll chase it down somehow.

To see what Gnus thinks the References header is, say
`(mail-header-references (gnus-summary-article-header))' on the
article you're interested in.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

SPCs can hurt overviews (was Inexplicable all.SCORE analysis)

[Karl Kleinpaste]

The erroneously-scored articles previously mentioned contain a Lines:
header which include trailing SPCs, e.g., "^Lines: 15        $".  This
confuses overview parsing, because the SPCs manage to survive into
INN's overview files, and Gnus does not hunt down the next TAB properly.

Looks like a user agent flaw, to start with; the articles in question
claim an agent tag of QWK v2.12.  Thereafter, it's INN's fault for
letting trailing SPCs into the overviews, though one could argue that
INN was doing the correct thing (copy blindly).  And lastly, it's
Gnus' mistake not to have hunted for the separating TAB properly.

Please experiment by implementing an "xref" score rule such as
	("xref" (".* .* .* .* .* .*" -10 nil r))
in all.SCORE and siccing it on <970923100046612@eqcity.ktb.net> in
comp.protocols.tcp-ip.domains.

(q 0.10, xemacs 19.14, inn 1.5.1.)

regards,
--karl

Re: SPCs can hurt overviews (was Inexplicable all.SCORE analysis)

[Lars Magne Ingebrigtsen]

Karl Kleinpaste <karl@jprc.com> writes:

> The erroneously-scored articles previously mentioned contain a Lines:
> header which include trailing SPCs, e.g., "^Lines: 15        $".  This
> confuses overview parsing, because the SPCs manage to survive into
> INN's overview files, and Gnus does not hunt down the next TAB properly.

Uhm... aha!  A bug in `gnus-nov-read-integer'.  Fix in Quassia Gnus
v0.12.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen