Re: [PATH] Support resolvconf

[Dan Williams <dcbw redhat com>]

On Sun, 2008-08-24 at 12:13 +0100, Roy Marples wrote:
> On Sun, 2008-08-24 at 12:58 +0200, Alexander Sack wrote:
> > On Sun, Aug 24, 2008 at 08:24:59AM +0100, Roy Marples wrote:
> > > 
> > > There's no real need to push it back.
> > > NM currently folds domain into searches, so there's no need to actually
> > > use domain at all.
> > > 
> > > Attached is a new patch, taking all the above into account.
> > > Also makes this less intrusive than the first pass :)
> > 
> > 
> > OK, this patch looks fine. Only glitch is that it doesnt properly pop
> > the config in all cases. The following does the trick for me (by
> > checking whether searches/nameservers actually have a non-null-length
> > string at all).
> > 
> > diff -u b/src/named-manager/nm-named-manager.c network-manager-0.7~~svn20080818t061112+eni1/src/named-manager/nm-named-manager.c
> > --- b/src/named-manager/nm-named-manager.c	(working copy)
> > +++ network-manager-0.7~~svn20080818t061112+eni1/src/named-manager/nm-named-manager.c
> > @@ -289,7 +289,8 @@
> >  			return FALSE;
> >  	}
> >  
> > -	if (searches || nameservers) {
> > +	if ( (searches && *searches && strlen(*searches)) ||
> > +	     (nameservers && *nameservers && strlen(*nameservers))) {
> >  		cmd = g_strconcat (resolvconf, " -a ", "NetworkManager", NULL);
> >  		nm_info ("(%s): writing resolv.conf to %s", iface, resolvconf);
> >  		if ((f = popen (cmd, "w")) == NULL)
> > 
> 
> That should not be needed, unless a config entry itself contains a null
> string. In this instance, probably better to fix it at source if input
> rather than in the processing here.

100% correct.

Dan