Updating <Rule> in shibboleth2.xml (was: Shibboleth 2.x packages updated in unstable)

[Kristof BAJNOK]

On Wednesday 16 September 2009 16.06.40 Peter Schober wrote:
> > > I suppose if people never changed the defaults, we could do
> > > something simpler.
> >
> > No, they don't, but guaranteeing that isn't so easy.
>
> Wouldn't conditionally applying a patch[1] in the postinst phase be
> sufficient? Something along the lines of:
> $ patch --dry-run -p0 -s < r2962.diff 1>/dev/null 2>&1 && patch -p0 -s
> 1>/dev/null 2>&1

Yes, this is what I was wondering about.
However, as long as the only practical consequence of _not_ doing this is 
that users get warnings in the logfile on startup (which they can trivially 
diagnose and correct if they care), well, I'm not in favour of hacking with 
people's config files anymore.

Kristof