rpms/rpcbind/devel rpcbind.init,1.7,1.8 rpcbind.spec,1.21,1.22

[Michael Schwendt]

On Wed, 17 Oct 2007 18:36:25 +0300, Ville Skyttä wrote:

> On Wednesday 17 October 2007, Michael Schwendt wrote:
> > On Tue, 16 Oct 2007 12:12:26 -0400, Steve Dickson (steved) wrote:
> > > Author: steved
> > >
> > > Update of /cvs/pkgs/rpms/rpcbind/devel
> > > In directory cvs-int.fedora.redhat.com:/tmp/cvs-serv31038
> > >
> > > Modified Files:
> > > 	rpcbind.init rpcbind.spec
> > > Log Message:
> > > Corrected a typo in the initscript from previous commit.
> > >
> > >  	# Check that networking is up.
> > > -	[ "$NETWORKING" = "no" ] || exit 6
> > > +	[ "$NETWORKING" = "no" ] && exit 6
> > >  	[ -f /sbin/$prog ] || exit 5
> >
> > Isn't
> >
> > 	[ "$NETWORKING" = "yes" ] || exit 6
> >
> > preferred, so the service is off by default unless networking
> > is turnt on explicitly?
> 
> There have also been opinons that sourcing /etc/sysconfig/network and then 
> testing $NETWORKING to see if networking is available is primarily legacy 
> cruft that should be just removed.
 
Fine.

I don't disagree.

But the initscripts doc files still cover this variable (albeit
without any details), and many initscripts do the ||'n'&& dance based
on its value (albeit without a policy). /etc/init.d/network even
exists with 0 (zero!) when NETWORKING=no, so not evaluating it in
network service scripts would be inconsistent.

Low hanging fruit for Fedora 9?