From: Bill Davidsen (davidsen@prodigy.com)
Date: Tue Aug 21 2001 - 07:28:05 CDT
Russ Allbery <rra@stanford.edu> noted:
> Bill Davidsen <davidsen@prodigy.com> writes:
> > Andrew Gierth <andrew@erlenstar.demon.co.uk> declared:
>
> >> Personally I'd need an awful lot of convincing before I even considered
> >> supporting the =group functionality in a server.
>
> > How about a decade of experience with INN? It's been in production use
> > for years, and I would think the continued use and track record would
> > stand as proof of concept.
>
> It's not in particularly wide use at INN sites and it has a few odd corner
> cases because no one ever really completely wrote down the details of the
> extension and how it interacts with the rest of the NNTP protocol. (A
> common problem with these sorts of things, unfortunately.)
On the other hand it has lots of experience, is useful for doing group
renames, particularly when a previous group becomes a hierarchy:
rec.travel 0000000000 0000000001 =rec.travel.misc
comp.mail 0000000000 0000000001 =comp.mail.misc
rec.toys 0000000000 0000000001 =rec.toys.misc
rec.auto 0000000001 0000000001 =rec.auto.misc
And I recently got a post from a group which was renamed several years
ago when I took over another ISP news, their usenet discussion was
renamed to mine. Still solving problems two years later, for users who
haven't updated their newsgroups in that long :-(
I think it has been proved both useful and possible to implement, and
therefore would certainly support standardization in the proper group.
As you note, the standard needs a state table to cover all the corner
cases, but that's what standards are for.
I still feel that lots of use over time without any major problems
surfacing constitutes "an awful lot of convincing" for a feature which
is at least as useful as some of the other things I've seen proposed!
-- -bill davidsen (davidsen@prodigy.com) "The secret to procrastination is to put things off until the last possible moment - but no longer" -me