From: Russ Allbery (rra@stanford.edu)
Date: Fri Aug 03 2001 - 18:07:00 CDT
Charles Lindsey <chl@clw.cs.man.ac.uk> writes:
> But that is not what is observed to happen in practice. Experience
> suggests that the old groups tend not to get deleted on many systems,
> and linger on for years, carrying articles between small (and poorly
> connected) groups of users. The main benefit of mvgroup is that the two
> processes (creation and removal) get coupled together, so that a
> sysadmin who deliberately wanted to retain both would have to work quite
> hard at it (and probably wouldn't bother, as oppsoed to the present
> situation where he may have to intervene manually to remove the old
> one).
I think you're incorrectly assuming that those systems are acting on any
control messages at all. More likely they're just picking up any
newsgroup that has traffic in it, which means that mvgroup wouldn't change
the shape of this problem.
> There are two separate issues to consider:
> 1. Would it be a useful feature to have, assuming it worked?
I think mvgroup would be useful since it makes the semantics of control
messages match more closely the operations that hierarchy administrators
are trying to actually perform.
> 2. Can it (or is it likely to) be implemented on a timescale that would
> ensure it would be regarded as normal practice in a few years time?
I highly doubt it.
-- Russ Allbery (rra@stanford.edu) <http://www.eyrie.org/~eagle/>