Again: bugs in contrib/solaris/opensshd.in and buildpkg.sh

[Michael Tokarev]

mouring at etoh.eviladmin.org wrote:
> 
> On Wed, 17 Oct 2001 j.petersen at msh.de wrote:
[]
> > --- contrib/solaris/buildpkg.sh Fri Oct 12 22:30:53 2001
> > +++ contrib/solaris/buildpkg.sh Tue Oct 16 13:53:07 2001
[]
> > -ln -s $FAKE_ROOT/etc/init.d/opensshd $FAKE_ROOT/etc/rcS.d/K30opensshd
> > -ln -s $FAKE_ROOT/etc/init.d/opensshd $FAKE_ROOT/etc/rc1.d/K30opensshd
> > -ln -s $FAKE_ROOT/etc/init.d/opensshd $FAKE_ROOT/etc/rc2.d/S98opensshd
> > +ln -s /etc/init.d/opensshd $FAKE_ROOT/etc/rcS.d/K30opensshd
> > +ln -s /etc/init.d/opensshd $FAKE_ROOT/etc/rc1.d/K30opensshd
> > +ln -s /etc/init.d/opensshd $FAKE_ROOT/etc/rc2.d/S98opensshd

That better should be:
     ln -s ../init.d/opensshd $FAKE_ROOT/etc/rcS.d/K30opensshd
           ^^
this is common practice and Sun people follows it carefully: no
absolute symlinks, only relative ones whenever possible (and the
startup scripts are *good* example where relative symlinks should
be used).  That is to allow mounting root filesystem on another
machine and have all links to point to the right place.  Well, a
minor issue.

> Hmm.. Eep.. Your right.  I forgot to verify the init stuff on my test box.
[]
> Damien, Can you commit the above patch.  It is the correct thing to do.

Regards,
 Michael.