Prev: OOo problems
Next: Another Micoshfat centric myth dies: 75% of Linux developers are now paid
From: Tony Houghton on 19 Jan 2010 18:07
In <slrnhl91iq.dhd.catwheezel(a)ID-107770.user.individual.net>,
Whiskers <catwheezel(a)operamail.com> wrote:

> On 2010-01-18, Robin H <robinh(a)anon.anon> wrote:
>> Can anybody recommend a newsreader that will kill crossposts from a given
>> group -- 24hoursupport.helpdesk, say?
>
> Can the version of Pan you seem to be using, not manage that? The
> 'stable' version 0.14.2 can, I'm sure. You might have to edit a score
> file manually.
>
> Slrn can of course, and I expect Gnus and Tin can too. Indeed, it might
> be easier to list the newsreaders that can't ...
>
> The main trick is to use the Xref header, not the Newsgroups header; the
> latter is 'expensive' as that header is not part of the standard
> 'overview' file on the news-server. Note that the Xref header sent from
> the server may only list newsgroups that are carried by that server.

Articles crossposted to two or even three newsgroups are often sensible,
but any more than that and they're usually trolls etc. A rule matching
one of these patterns should help:

Newsgroups: .*,.*,.*,
Xref: .*:.*:.*:.*:

--
TH * http://www.realh.co.uk
From: Nigel Wade on 20 Jan 2010 04:50
On Tue, 19 Jan 2010 17:31:52 +0000, Dave Gibson wrote:

> Nigel Wade <nmw(a)ion.le.ac.uk> wrote:
>> On Mon, 18 Jan 2010 18:41:25 +0000, Paul Martin wrote:
>>
>>> In article <hj1vnd$1g4$1(a)speranza.aioe.org>,
>>> Robin H wrote:
>>>> Can anybody recommend a newsreader that will kill crossposts from a
>>>> given group -- 24hoursupport.helpdesk, say?
>>>
>>> slrn
>>>
>>>
>>> [uk.comp.os.linux]
>>> Score: =-9999
>>> Newsgroups: 24hoursupport\.helpdesk
>>
>> That match doesn't work for me in Pan. It seems it only matches the
>> first newsgroup in the list. If 24hoursupport is not first in the list
>> the match fails.
>
> Insert ".*" to skip over leading newsgroup names.
>
> Newsgroups: .*24hoursupport\.helpdesk

You think I didn't try that?

I do know how to use RE...

--
Nigel Wade
From: Nigel Wade on 20 Jan 2010 05:58
On Tue, 19 Jan 2010 13:35:13 -0600, Moe Trin wrote:

> On Tue, 19 Jan 2010, in the Usenet newsgroup uk.comp.os.linux, in
> article <hj3sc9$48i$1(a)south.jnrs.ja.net>, Nigel Wade wrote:
>
>>Paul Martin wrote:
>
>>> Robin H wrote:
>
>>>> Can anybody recommend a newsreader that will kill crossposts from a
>>>> given group -- 24hoursupport.helpdesk, say?
>
>>> slrn
>
>>> [uk.comp.os.linux]
>>> Score: =-9999
>>> Newsgroups: 24hoursupport\.helpdesk
>
> As noted by others, that's expensive scoring, requiring the news reader
> to ask for that header for every article in the '[uk.comp.os.linux]'
> group.
>
>>That match doesn't work for me in Pan. It seems it only matches the
>>first newsgroup in the list. If 24hoursupport is not first in the list
>>the match fails. And the documentation for the Score file in Pan is
>>rather limited.
>
> news.software.readers Discussing software for reading network news
> (Usenet).
>
> though I'm not sure how much it would help. You may want to consider a
> different news reader.

I've considered, and tried quite a few. Each has it's own idiosyncrasies
and limitations. Pan is about the best overall compromise. I did used to
use Knode, but the list of rules I'd built up made Knode so slow to load
some newgroups that it became intolerable. Pan is much quicker at
filtering, but clearly has limitations on what it can filter - that's
probably why it's so much quicker.

> Hmmm... does adding a wild-card fix it?
>
> Newsgroups: .*24hoursupport\.helpdesk
>
> (I don't think it works in slrn, but the original score does.)

Unfortunately not. I tried just about every variety of RE that I could
think of. None matched.

I've been doing some more extensive tests now, and it seems that my
previous post where I stated that it only matched if 24hoursupport was
first in the list was false. The match which I did get must have been a
false positive, i.e. the posts matched against some other rule in the
file, not the Newsgroups: match.

Doing more tests it seems that Newsgroups: will only match against
newsgroups which the news server carries. Quite why this is I don't know,
maybe it's some optimization which Pan performs in order to avoid the
slow search which matching on Newsgroups: is supposed to entail.

Anyway, the upshot is that I can match against an RE such as:

advocacy|kooks

and it removes all posts with newsgroups containing those words. However,
I am completely unable to match 24hoursupport, or the scorched-earth
group mentioned in one sub-thread.


>>Xref does appear to work on any group in the Xref list, but most of the
>>idiot posts don't have an Xref to that group.
>
> Most probable reason is that the news server you are reading from
> doesn't carry the '24hoursupport.helpdesk' newsgroup, and therefore has
> no need to create an Xref entry for a group that (as far as it is
> concerned) doesn't exist. See RFC1036 section 2.2.13 and RFC5536
> section 3.2.14.

That would almost certainly account for it then. Our news service (Janet
academic newsserver) does not carry that newsgroup.

--
Nigel Wade

From: David Gibson on 20 Jan 2010 11:23
On Wed, 20 Jan 2010 09:50:12 +0000, Nigel Wade wrote:

> On Tue, 19 Jan 2010 17:31:52 +0000, Dave Gibson wrote:
>
>> Nigel Wade <nmw(a)ion.le.ac.uk> wrote:
>>> On Mon, 18 Jan 2010 18:41:25 +0000, Paul Martin wrote:
>>>
>>>> In article <hj1vnd$1g4$1(a)speranza.aioe.org>,
>>>> Robin H wrote:
>>>>> Can anybody recommend a newsreader that will kill crossposts from a
>>>>> given group -- 24hoursupport.helpdesk, say?
>>>>
>>>> slrn
>>>>
>>>>
>>>> [uk.comp.os.linux]
>>>> Score: =-9999
>>>> Newsgroups: 24hoursupport\.helpdesk
>>>
>>> That match doesn't work for me in Pan. It seems it only matches the
>>> first newsgroup in the list. If 24hoursupport is not first in the list
>>> the match fails.
>>
>> Insert ".*" to skip over leading newsgroup names.
>>
>> Newsgroups: .*24hoursupport\.helpdesk
>
> You think I didn't try that?

Did you add the rule to ~/.pan2/Score with a text editor or did you use
that dismal graphical scoring widget?

As a test, I added the following to ~/.pan2/Score (pan 0.132) and the
appropriate posts (including those where comp.os.linux.setup was not the
first group in the Newsgroups: header) were marked accordingly.

[comp.os.linux.misc]
Score: =-100
Newsgroups: .*comp\.os\.linux\.setup

> I do know how to use RE...
From: Nigel Wade on 20 Jan 2010 12:34
On Wed, 20 Jan 2010 16:23:11 +0000, David Gibson wrote:

> On Wed, 20 Jan 2010 09:50:12 +0000, Nigel Wade wrote:
>
>> On Tue, 19 Jan 2010 17:31:52 +0000, Dave Gibson wrote:
>>
>>> Nigel Wade <nmw(a)ion.le.ac.uk> wrote:
>>>> On Mon, 18 Jan 2010 18:41:25 +0000, Paul Martin wrote:
>>>>
>>>>> In article <hj1vnd$1g4$1(a)speranza.aioe.org>,
>>>>> Robin H wrote:
>>>>>> Can anybody recommend a newsreader that will kill crossposts from a
>>>>>> given group -- 24hoursupport.helpdesk, say?
>>>>>
>>>>> slrn
>>>>>
>>>>>
>>>>> [uk.comp.os.linux]
>>>>> Score: =-9999
>>>>> Newsgroups: 24hoursupport\.helpdesk
>>>>
>>>> That match doesn't work for me in Pan. It seems it only matches the
>>>> first newsgroup in the list. If 24hoursupport is not first in the
>>>> list the match fails.
>>>
>>> Insert ".*" to skip over leading newsgroup names.
>>>
>>> Newsgroups: .*24hoursupport\.helpdesk
>>
>> You think I didn't try that?
>
> Did you add the rule to ~/.pan2/Score with a text editor or did you use
> that dismal graphical scoring widget?
>

I did it properly, using a fine text editor (vi - the best configuration
tool available).

> As a test, I added the following to ~/.pan2/Score (pan 0.132) and the
> appropriate posts (including those where comp.os.linux.setup was not the
> first group in the Newsgroups: header) were marked accordingly.
>
> [comp.os.linux.misc]
> Score: =-100
> Newsgroups: .*comp\.os\.linux\.setup
>

See my other post. There are strange things with Newsgroup: matching. My
setup won't match 24hoursupport with any RE, despite what I thought
earlier. I presume it's because my newsserver doesn't include this group,
I can match other groups which are served. I don't know why Pan will only
match groups which are available from the server, my guess is some kind
of optimization by Pan.

--
Nigel Wade

First  |  Prev  |  Next  |  Last
Pages: 1 2 3 4 5
Prev: OOo problems
Next: Another Micoshfat centric myth dies: 75% of Linux developers are now paid