Re: suggestion for static ip

[Dan Williams <dcbw redhat com>]

On Mon, 2006-03-27 at 22:50 +0200, Kasper Peeters wrote:
> > Really?  It's true that for wireless networks you can make an
> > educated guess based on the ESSID, but we all know how much we can
> > count on the fact that an ESSID of "linksys" means that we're
> > talking to the AP we think we're talking to...
> 
> I'd be happy to have it for wireless too, but right now I don't need
> it, and it's probably easier to add it to the wired part only. In any
> case: is someone familiar with the NM code willing to give this a
> shot, or shall I try to cook up a proof-of-concept?

A proof-of-concept, even if it doesn't involve any NM code at all, would
be greatly neat.  Bonus points for doing it using glib, since that's
what NM uses and we'd have to convert the proof over to it anyway.

The configuration information, ie what MAC address to arping and what
corresponding static IP to assign to the card would be derived from the
distribution's specific configuration files though NM's configuration
abstraction stuff.

I'm assuming that when we notice a link switch in NM, that we can fire
off the arping requests and then, with a short delay, start a DHCP
transaction.  If we receive a reply from the arping, we cancel DHCP and
set the static IP on the device.  If no reply from arping, we wait until
DHCP has finished with success, or timed out, and continue with the
existing behavior.

Dan