[plug] New PLUG news server

[Bernd Felsche]

James Devenish <devenish at guild.uwa.edu.au> writes:

>In message <s7rp22x24b.ln2 at innovative.iinet.net.au>
>on Wed, Sep 29, 2004 at 12:45:16PM +0800, Bernd Felsche wrote:
>> Not at all. The novel idea you appear to have is that somehow they
>> can ascertain an explicit interest without seeing the header. :-)

>We are obviously talking about different things, then, as the above
>claim has never been my thought or intention. If you are interested
>in a million articles, then there's no way around it. But if you
>aren't interested, then you don't touch them and there is no delay.

I'm not interested in a million articles, but you might be
interested in a hundred out of the million. So the "only" way to
determine which ones they are out of the million, is to get the
overview containing key headers for the million.

AFAIK, there's no thread-based overview specification available in
the protocol... It's been a long time since I looked at it and the
protocol may have grown such an option with the trend towards
thread-based readers. leafnode2 may well then provide support.

It should be obvious that such exercises are best undertaken on an
archive server (micro-google) of sorts.

>> It's a real barrier until you alter the source code and
>> rebuild leafnode.

>Okay, you and I disagree on your use of the word "configurable".

You can configure leafnode2 to shift the barrier to 68 years by
editing the /etc/leafnode/config text file and a companion filter
file if defined. The barrier is then still no further than 24800
days.  It's still there; i.e. not removed.

I don't understand how you can disagree with my use of the word
"configurable".

>> Yep. It's arbitrary in some ways. But valid in that the archived
>> articles posted before the news server existed cannot logically be
>> found in the newsgroup.

>By that logic, we had better remove a lot of articles from the
>Mailman archives!!! And Cameron had better dump his backdated
>off-top Lurker archives!

>> Each header takes a finite amount of time to transmit.

>Then don't transmit them!

>> Simply knowing that there are 52,678 (for example) messages
>> available, how can you tell which ones you want to read?

>You apply your own judgement. If you are interested in a sample of
>50, then take a sample of 50 and see how you go. 

The "sample" of 50 is the most-recent 50. Hardly a "sample" at all
because it might be ranting about nothing more than the NNTP server.

I know that what you an others probably want is the past 50 active
threads, but I've not seen a server that delivers such, nor a client
that does it. fetchnews that's part of leafnode2 can probably be
modified to do that; feel free to do so and contribute diffs back to
the leafnode2 project.

Article expiry in leafnode2 is already thread-based. Active threads
never expire (overlooking the bug in limiting the length of the
References header). Threads that aren't active for a pre-determined
time, are expired.

>When you've read and reconstructed the ones you're interested in,
>you can work your way further backwards. If you don't care for the
>past, then go recall ZERO messages of the messages and suffer no
>delays.

Sort of defeats the purpose in having everything in the main
newsgroup.

I urge you to read the overview on leafnode2 so that we can be
talking about the same thing.
-- 
/"\ Bernd Felsche - Innovative Reckoning, Perth, Western Australia
\ /  ASCII ribbon campaign | I'm a .signature virus!
 X   against HTML mail     | Copy me into your ~/.signature
/ \  and postings          | to help me spread!