RFC: configuration broker design

[Steve Langasek]

Hi Martin,

I've read over the configuration broker design document, and have a few
comments/questions:

What is the difference between "methods" and "configuration methods", the
latter not previously defined in the design doc?

Methods are implemented as shell scripts -- is this a hard requirement, or
will the interface support methods implemented in any language that are 
invoked via exec()?

"[T]he interface needs not be unconfigured and reconfigured" -- how will
this interact with services that need to know that a connection has been
lost, so that they can deconfigure network-specific setups (e.g., mount
points)?

Does netlink give us information about how an interface's configuration has
changed? (current weakness of ifupdown)

Interfacing with wpa-supplicant: has it been tested yet?  (currently
wpa-supplicant sucks for roaming, I have to manually bounce interfaces and
fiddle with wpa_cli to reliably move between essids even though
wpa_supplicant is supposed to take care of this for me)

Thanks,
-- 
Steve Langasek                   Give me a lever long enough and a Free OS
Debian Developer                   to set it on, and I can move the world.
vorlon at debian.org                                   http://www.debian.org/