From mboxrd@z Thu Jan  1 00:00:00 1970
X-Msuck: nntp://news.gmane.io/gmane.emacs.gnus.general/9950
Path: main.gmane.org!not-for-mail
From: visigoth@naiad.fac.cs.cmu.edu
Newsgroups: gmane.emacs.gnus.general
Subject: Re: Why so many k data when just fetching headers.
Date: 15 Feb 1997 03:58:21 -0500
Sender: visigoth@naiad.fac.cs.cmu.edu
Message-ID: <vpdk9oatphu.fsf@naiad.fac.cs.cmu.edu>
References: <ug1z17gei.fsf@pm1.contactor.se> <m2sp2z68o0.fsf@proletcult.slip.ifi.uio.no>
NNTP-Posting-Host: coloc-standby.netfonds.no
X-Trace: main.gmane.org 1035149898 21538 80.91.224.250 (20 Oct 2002 21:38:18 GMT)
X-Complaints-To: usenet@main.gmane.org
NNTP-Posting-Date: Sun, 20 Oct 2002 21:38:18 +0000 (UTC)
Return-Path: <ding-request@ifi.uio.no>
Original-Received: from ifi.uio.no (0@ifi.uio.no [129.240.64.2])
	by deanna.miranova.com (8.8.5/8.8.5) with SMTP id BAA06463
	for <steve@miranova.com>; Sat, 15 Feb 1997 01:19:16 -0800
Original-Received: from naiad.fac.cs.cmu.edu (NAIAD.FAC.CS.CMU.EDU [128.2.191.173]) by ifi.uio.no with SMTP (8.6.11/ifi2.4) 
	id <JAA13315@ifi.uio.no> for <ding@ifi.uio.no> ; Sat, 15 Feb 1997 09:58:46 +0100
Original-To: ding@ifi.uio.no
In-Reply-To: Lars Magne Ingebrigtsen's message of 14 Feb 1997 10:25:51 +0100
Original-Lines: 34
X-Mailer: Red Gnus v0.82/Emacs 19.34
Source-Info: Sender is really visigoth@naiad.fac.cs.cmu.edu
Xref: main.gmane.org gmane.emacs.gnus.general:9950
X-Report-Spam: http://spam.gmane.org/gmane.emacs.gnus.general:9950

Lars Magne Ingebrigtsen <larsi@ifi.uio.no> writes:

> It depends on your variable settings (gnus-fetch-old-headers in
> particular) and whether the 30 unread messages were spread out
> numerically. 

Hmm.  Does Gnus do anything fancy to try to detect if the news server
provides overview lines?  We use a little proxy thing locally which
will provide overviews, but produces errors on "list overview.fmt".
Will Gnus switch to using headers because of this?

On a related note--the tool that we use is called dnntpd.  It merges
the output of two or more news servers into one.  Using this, we have
separate machines to spool "frivolous" groups and "important" groups,
and could split further.  This allows the machines to be less heavily
loaded and allows us to better manage disk space.  dnntpd, though, is
not the best program in the world.  How hard would a dnntp-like
backend be to write?  Specifically, a backend that would take a list
of virtual servers and provide each newsgroup from the first server on
the list that carries it, and merge ACTIVE files from those servers?

Gnus can sort of already do this by having multiple servers, but it
doesn't really support transparently moving groups from one server to
another and back, which we can do with dnntpd.  The main problem I see
with this is if two new servers don't use the same numbering
information--but this shouldn't be a problem if, as in our case, one
server carries some groups, and the other carries some other groups.
Or, in another case, the preferred server never "loses" the group.

Can I get this on the list of things for Quassia Gnus, or is it too
special-purpose?

John.