On Thu, 27 Dec 2001, Andrew Doran wrote:

> kpneal@pobox.com wrote:
>
> > On Thu, Dec 27, 2001 at 12:08:33PM -0500, Andrew Brown wrote:
> > > >It seems to me that some of the problems noted here could be dealt with in a
> > > >more simple way by having /usr/{local,pkg,X11R6}/etc/rc.d included in
> > > >rcorder's pass during boot. I know it has been said that there were problems
> > > >with this. What are they?
> > >
> > > well...first of all, there's the problem of whether or not /usr is
> > > *present* at the time that rc starts running...
> >
> > So have an rc script run after /usr is mounted and execute all the
> > /usr/* rc.d scripts then?

Who says that pkgsrc scripts should only start after the base scripts
are done. What if you're running bind9 on your fileserver?

> That would satisfy me (having a clean seperation between system and local
> scripts, raising visiblity), since I can accept making changes to the system
> itself for anything more complicated (like a new version of named, where
> named where named is truly a dependancy, for instance). Having some
> subdirectories under /etc/rc.d is another scheme that comes to mind.

If you're going to execute every single thing in /usr/pkg/etc/rc.d,
/usr/X11R6/etc/rc.d, and /usr/local/etc/rc.d anyway, where's the seperation?
Why not just put everything into /etc/rc.d?

We've been through all this before, as a search over the tech-pkg
archives would show. The only new thing here, is my idea to permit
changing the path in "/etc/rc.conf". What's wrong with that?

Frederick