[Pkg-wpa-devel] [RFC]: ifupdown pre-up, post-down scripts
Reinhard Tartler
siretart at tauware.de
Sun Feb 26 22:15:21 UTC 2006
On Mon, Feb 27, 2006 at 07:01:43AM +1000, Kel Modderman wrote:
> >Wow! I love this idea! I think this is even better than the pre-up.d/
> >post-down.d/ approach we tried before.
> >
>
> What approach was/is that?
The one which is currently in the wpasupplicant-0.5 branch in our svn.
It is similar to the old approach using init scripts, but now moved to
those scripts directly.
> Yes, we face the same problem with Kanotix. The current wpasupplicant is
> not interfaces friendly :-)
Perhaps we can work on having the most common use cases (WPA-PSK, PEAP
and similar) configrable in /e/n/i.
> >I suggest to introduce settings like 'wpa-psk-key' and/or
> >'wpa-peap-username'/'wpa-peap-password'. If all given parameters are
> >given for such easy setups, the scripts can create a config file for
> >wpasupplicant on the fly.
> >
> >
>
> Yep. easily supplied via wpa_cli. One would need to handle the check for
> $IF_WPA_CONF differently, I'll look into it.
If we use wpa_cli, it might make sense to introduce a wpasupplicant
group, which is able to access the running wpasupplicant (or
supplicants, in case of multiple interfaces).
> For roaming, I'd eventually love to see a wpa_cli action script plugin,
> somehow providing a (user-configurable) means of silently bringing
> up/down each location "alias" specified in interfaces.
interesting. could become tricky, though.
> Cool, glad you liked the idea, I'll continue to work on these ifupdown
> hooks.
>
> If everyone is in agreement, can I commit these to our wpasupplicant-0.5
> branch at some stage? This could provide the backend for a debconf
> driven configuration asistant as well . . . (and end the wpasupplicant
> init script transition).
Yes, all changes are versioned anyway. If someone disagrees with some
changes or approach, he can branch from an older revision anyway, but I
agree with your approach.
Gruesse,
Reinhard
More information about the Pkg-wpa-devel
mailing list