[12:41] <Keybuk> I thought of a much more sick way of dealing with pre-requisite mountpoints :p
[12:42] <_ion> Cool!
[12:42] <_ion> Oh, i misread it as "slick" :-D
[12:42] <Keybuk> lay a tmpfs over the root filesystem using unionfs 
[12:42] <Keybuk> make the mount point directories on it
[12:42] <Keybuk> then you can always mount them :p
[12:43] <Keybuk> (actually, you can get away with just requiring that mount point directories always exist on the root filesystem)
[12:44] <_ion> script
[12:44] <_ion>     mkdir -p "$mountpoint"; mount "$mountpoint"
[12:44] <_ion> end script
[12:44] <_ion> ? :-)
[12:45] <Keybuk> only works if you wait until the root filesystem is writable before mounting under it
[12:46] <Keybuk> and you'd need to use the same bind-mount trick as Ubuntu does for /var/run and /var/lock
[12:46] <Keybuk> (which has the advantage that when you unmount /usr, /usr/local is still mounted <g>)
[12:48] <Keybuk> needs much locking to work too <g>
[12:48] <Keybuk> very sick
[12:49] <_ion> :-)
[12:50] <_ion> Yeah, dbus would be really nice instead of whatd, but it would need to 'start on startup'.
[12:52] <Keybuk> don't Fedora already do that/
[12:52] <_ion> I have no idea.
[12:53] <_ion> If it's plausible that distributions using libwhat would actually do that, dbus would be the obvious solution.
[01:06] <_ion> I added a note about dbus to the page.
[10:09] <_ion> I added a usplash mockup to http://johan.kiviniemi.name/blag/2007/03/21/upstart-and-interaction-with-user/
[10:15] <AlexExtreme> looks interesting
[10:15] <AlexExtreme> I'm still not keen on the idea of using DBUS thought
[10:16] <_ion> It has some pros and a bunch of cons.
[10:17] <AlexExtreme> yes, it does have some pros because it provides an easily available IPC protocol, but IMO the cons outweigh the pros