Bug#412179: [pkg-wpa-devel] Bug#412179: D-Bus interface

[Kel Modderman]

On Monday 05 March 2007 01:05, Michael Biebl wrote:
<snip>
> >
> > Ok. How would the sequnce number of the wpa_supplicant event.d scriplet
> > be determined?
>
> NetworkManager 0.7.x is currently the only application I know of, which
> uses the D-Bus interface of wpa_supplicant. It runs at 25, so, yeah, 23
> should be fine.

Good. I just pulled 23 out of my behind ;-)

>
> > Attached is the patch I recently commited to SVN, can you please check
> > it?
> >
> > Also, an experimental package has been prepared at:
> >
> > http://users.tpg.com.au/sigm/debian/pkg-wpa/wpasupplicant_0.6.0~cvs200702
> >24-1.dsc
>
> Seems to work fine here, although I can't really test the D-Bus
> interface. NetworkManager 0.7 from SVN is currently horribly broken and
> immediately crashes after startup (not because of wpa_supplicant I might
> add).
>
> Maybe it would make sense to provide a /etc/default/wpasupplicant with a
> var DAEMON_START or the like which defaults to 0. The D-Bus init script
> would check this var and so it would be easy to enable/disable the D-Bus
> service.

Ok. In the future, how would we make sure that NetworkManager "JustWorks" (as 
is their philosphy on homepage) when the backend it depends on, 
wpa_supplicant, is disabled by default?

Will the user be presented with an obtuse error message by NetworkManager 
indicating that the wpa_supplicant interface is not available and be forced 
into editing one or more config files to get it working?

> IIRC avahi-daemon does it similar.

Ok. 

# 0 = don't start, 1 = start
AVAHI_DAEMON_START=1

To suggest an answer to my pesky questions above, would could ship a conffile 
similar to this now, but disabled, and when networkmanager 0.7.X is in the 
archive "flip the switch" and upload new wpasupplicant with conffile set to 
activate wpa_supplicant dbus daemon by default.

> With no application using the D-Bus interface atm. (NM 0.7 will still
> take some time to be released) I think people would complain if we start
> wpa_supplicant by default.

Ack. I would complain ;-)

> On the other hand, if wpa_action could be made to reuse the running
> D-Bus enabled wpa_supplicant instance instead of spawning its own copy
> (dunno if that would be possible/sensible) it would be different.

I don't think that is feasable a present.

Thanks, Kel.