Re: Mailing lists and news archiving

[Lars Magne Ingebrigtsen <larsi@gnus.org>]

(Chris Beggy ) news@kippona.com writes:

> http://www.mail-archive.com does a good job of this, and we can
> even read these as gnus groups with nnwarchive!

Yeah, mail-archive.com is pretty good, but I still prefer reading from
an nntp server.  Even though nnwarchive makes it possible to read the
archives with Gnus, it's so much smoother when reading from an nntp
server... 

> For now, mail-archive.com is a gift from Jeff Briedenbach and
> VALinux.  In the future, who knows?  We've already lost
> deja-news.com, which was archiving usenet in a similar way.  Some
> people lost faith in VALinux's approach to http://sourceforge.net
> and are cloning it at http://savannah.gnu.org, which is another
> indication.

Yeah.  Difficult to say what's going to happen with all that.  People
seem quite jittery about sourceforge these days...

> I *already* think this is a great idea for distribution.  Do you
> have the disk space for archiving these, though?

Sure.  Disks are cheap.  :-)

If I get serious about doing a real big news server, then I'll
probably extend the machine with a nice RAID-ey disk setup, but for
now, it just has one biiig IDE disk.

> In addition, are newsreaders really good at perusing archived
> groups?  Even with gnus, I know only how to select the N oldest
> articles or M most recent articles, but I can't get range N to M.
> (I'm just kidding. I'm certain gnus can do this, but what about
> other newsreaders :-) ) Downloading the *whole* archive isn't a good
> use of resources.  Might it be better to clone mail-archive.com's
> approach for archiving?

I was actually pondering these very things.  How to do a convenient
search from Gnus from such a repository?

Ok, say you go enter the gnus.xemacs.beta group.  (I should come up
with a better top level hierarchy, though.)  You press the "search"
button, and type in a text.  Gnus queries Quimby, and gets a list of
Message-IDs back.  Gnus then requests these headers from the news
server, and displays them in the buffer.

The query can be via HTTP to some search engine or other running
separately on Quimby.  For instance a Glimpse thing, or whatever is
the done thing these days.

I think such a thing could be done without too much work...

-- 
(domestic pets only, the antidote for overdose, milk.)
   larsi@gnus.org * Lars Magne Ingebrigtsen