On Mon, 21 Mar 2005, Hubert Feyrer wrote:

> > It's rather critical to have a front-and-center indicator of whether
> > installed files are different.  Otherwise, users having trouble with a
> > package will have a much more difficult time trying to communicate which
> > "version" of the package is in use.
>
> Yes. Actually, I'd make this the central point to judge if a PKGREVISION bump
> is appropriate: If you can/have to tell a user "use THAT (new) revision" in
> order to make it differentiate from an old, possibly
> buggy/insufficient version, then the bump is justified. For whatever changes
> led to the change in behaviour that may be of interest to the user.
>
> Personally I think trying to list all those possible changes would be too
> much, and should be left to developers' judgement. :)

Why isn't "if any installed file changes" sufficient as the criterion?
Judgement calls about whether or not a user might "see" a change are prone
to... well, misjudgement.

-- 
-- Todd Vierling <tv@duh.org> <tv@pobox.com>