__/ [Roy Culley] on Wednesday 08 February 2006 01:19 \__
> begin risky.vbs
> Roy Schestowitz <newsgroups@xxxxxxxxxxxxxxx> writes:
>> __/ [Roy Culley] on Sunday 05 February 2006 00:26 \__
>>> ngstats.pl version 1.18
>>> This report covers 2093 articles received by this system to
>>> newsgroup comp.os.linux.advocacy 
>>> <snip />
>>> <snip rel="again" />
>> Could you please include/implement a times or dates section to keep
>> abreast of your new competition from Turquoise user?
> I used to use turqstat for the stats I post. In fact I just took over
> from Terry Porter who started posting them. I've never found the times
> feature of turqstat very interesting except on occasion when I used it
> to show some posters who post round the clock.
Times are the least intersting stats once you look at them the first time and
spot the statistical pattens. They are very stable unlike components that
involve participants who come and go, rise and fall. Moreover, people are
known to care about affairs of other people (names, faces and the like), so
it's natural to look at names rather than numbers. It's what gives numerical
analysis some character.
>> I sometimes wonder if your script is a public one and how one
>> intervenes or complements the in- put,
> When I got fed up with turqstat (some article headers would crash it,
> it uses very 'fuzzy' logic when working with article dates and the
> code is horrible. When I wanted to reduce the width of the reports it
> uses hard coded numbers for each section of the report) I decided one
> Saturday afternoon if I could write a perl script to do the same.
Why is it not public [polite tone]? Is it difficult for the user to maintain
unless changes are made in-line? Is the Perl notoriously intractable? I was
hoping to make use of your script (exposing it to some more light), but
ended up compiling turqstat, which now runs as a cron job to generate weekly
and fortnightly stats for the search engines and Palm newsgroups
> I was surprised how easy it was. By evening ngstats.pl, which is just
> a crude hack, produced identical output using the same articles as
> turqstat. turqstat does a lot more of course but as I didn't use the
> extra features it didn't matter. Surprisingly my script at that time
> ran faster than turqstat given the same articles as input.
> Since then of course I've added new features which slow it down a bit.
> It takes 2 or 3 seconds to process a weeks COLA articles. I can live
> with that. :-)
What?!?! That's quick indeed, compared with turqstat that fetches the feed
almost immediately on this 100Mbit connection and then labours for about 1
minute per 1,000 posts (it doesn't scale linearly, I know).
>> which is a raw newsgroups feed.
> I run leafnode and have 2 newsfeeds. Very few articles are missed
> other than those filtered by leafnode such as mail to news gateways.
That's probably a positive thing.
>> Toby Inkster (of alt.www.webmaster, alt.html and some *nix groups)
>> has a script which is definitely better than Turquoise.
> Do you have a link to his script? ngstats.pl is a bit too messy for me
> to make public. If there's an interest in it then I would be happy to
> clean it up and offer it to all. Under the GPL of course. :-)
Toby keeps a mirror of all stats on his Web site. He doesn't refer to the
tool(s) as far as I can recall (I might re-check), but only appends
"Compiled by Toby Inkster" along with contact details.
My question regarding publicising the script has been asked before I read
your message carefully enough. Let's share the goods, not doing the
equivalent of monopolisation or patent. Self-glorification is an act of
attempting to dominate by keeping others in the dark while your desire to
GPL it shows that you want to _contribute_, which is anything _but_ selfish.
Roy S. Schestowitz | while (sig==sig) sig=!sig;
http://Schestowitz.com | SuSE Linux | PGP-Key: 0x74572E8E
7:20am up 22 days 2:36, 12 users, load average: 1.00, 0.87, 0.73
http://iuron.com - Open Source knowledge engine project