[news.software.b] more ways to kill articles

eric@snark.UUCP (Eric S. Raymond) (05/01/88)

In article <2182@rayssd.ray.com>, gmp@rayssd.RAY.COM (Paris) writes:
>What I'd like:
>
>1.  A way to kill articles by amount of included text, accounting
>    for various indenting characters like >, <, -, and <tab>.

Interesting idea. I might have time to add this to 3.0 before release.
Seems a bit unselective, though.

>2.  A way to kill articles by long signature, especially articles
>    with signatures not preceded by the "--<space><return>" sequence.

Um, how would you tell what things were signatures?

>3.  A way to kill followups (efficiently) to articles that were
>    auto-killed, for example, followups to webber@ articles.

3.0 has this, sort of, for one major case. There's an 'unsuscribe discussion'
command that kills all followups to a given ID.
-- 
      Eric S. Raymond                     (the mad mastermind of TMN-Netnews)
      UUCP: {{uunet,rutgers,ihnp4}!cbmvax,rutgers!vu-vlsi,att}!snark!eric
      Post: 22 South Warren Avenue, Malvern, PA 19355   Phone: (215)-296-5718

heiby@mcdchg.UUCP (Ron Heiby) (05/03/88)

Although this doesn't actually kill the article, it is possible to tell
rn to "hide" the normal inclusion lines.  The HIDELINE variable can be
set to something like "^>" and any line beginning with a "greater-than"
symbol will be hidden when you view the article.  You can even use "|"
to hide more than one pattern.  The hidden lines can be viewed by restarting
the article with the 'v' command.
-- 
Ron Heiby, heiby@mcdchg.UUCP	Moderator: comp.newprod & comp.unix
"I believe in the Tooth Fairy."  "I believe in Santa Claus."
	"I believe in the future of the Space Program."

gmp@rayssd.ray.com (Greg Paris /exaflop meatbop/) (05/06/88)

In article <7531@mcdchg.UUCP> heiby@mcdchg.UUCP (Ron Heiby) writes:
> Although this doesn't actually kill the article, it is possible to tell
> rn to "hide" the normal inclusion lines.  The HIDELINE variable can be
> set to something like "^>" and any line beginning with a "greater-than"
> symbol will be hidden when you view the article.  You can even use "|"
> to hide more than one pattern.  The hidden lines can be viewed by restarting
> the article with the 'v' command.

This is a nice feature of rn, but it's not good enough.  I don't have a
problem with included text per se, so I don't want to hide all included
text by default.  (I do use HIDELINES to hide multiply-included text,
though.)

What I want to be able to do is automatically (and with as little
overhead as possible) junk articles that contain predominately more
included text than original text.  The 50% rule is probably extreme,
but articles that contain 75% or more included text are nearly always
not worth reading (my opinion).

Another possibility might be to have a HIDELINES-like variable that
will come into play after N contiguous lines of included text.

-- 
Greg Paris /exaflop meatbop/ <gmp@rayssd.ray.com>
{att,decuac,gatech,necntc,sun,uiucdcs}!rayssd!gmp

gmp@rayssd.ray.com (Greg Paris /exaflop meatbop/) (05/06/88)

In article <227b3604:9d29@snark.UUCP> eric@snark.UUCP (Eric S. Raymond) writes:
> >2.  A way to kill articles by long signature, especially articles
> >    with signatures not preceded by the "--<space><return>" sequence.
> 
> Um, how would you tell what things were signatures?

Starting at the end of the article, count as a signature line any line
that has a preponderance of the characters "!@+=-$*_" (and other
signature favorites).  Count blank (and empty) lines too.  Large
amounts of white space in the line are probably a good indicator also.

This method can be easily defeated by putting a line that looks like
text at the end of the article.  Better ideas are welcome!

By the way, I've been informed by Amir Plivatsky that articles
transmitted via BITNET have all trailing spaces removed, so the
"--<space><return>" sequence isn't good enough for recognizing the
beginning of a signature.

-- 
Greg Paris /exaflop meatbop/ <gmp@rayssd.ray.com>
{att,decuac,gatech,necntc,sun,uiucdcs}!rayssd!gmp