Re: nm-system-settings problem

[Dan Williams <dcbw redhat com>]

On Tue, 2008-05-13 at 16:12 -0400, Gene Czarcinski wrote:
> On Tuesday 13 May 2008 15:59:39 Dan Williams wrote:
> > You actually do want to match the watch descriptor to the one that we're
> > saving; the patch (while it works) is essentially a big club when we
> > need to be a bit more elegant.  The problem is that the watch
> > descriptors increase as you add new watches, and therefore if I added a
> > watch to /var/foo/bar that would now trigger the check there, which is
> > wrong.  But you're on the right path :)
> 
> I never claimed my patch was anything but a big club! ... but it did work.
> 
> Part of my problem was that I could understand what was happening in 
> stuff_changed() and in sc_plugin_inotify_init() but the other functions and 
> how they worked/interacted was not all that clear.
> 
> >
> > I've fixed this "correctly" upstream.  Turns out that GIO in glib
> > doesn't handle hardlinks either.  But I've worked around that by doing
> > parallel watches on ifcfg files themselves (instead of the
> > network-scripts directory) to catch the hardlink case.  I'll build a new
> > testing version of NM presently.
> 
> When you have something ready to test, give a shout and I will give it a try.

http://koji.fedoraproject.org/koji/taskinfo?taskID=607805

Dan