Re: WPA status 2006-01-03

[Dan Williams <dcbw redhat com>]

On Wed, 2006-01-04 at 01:56 +0100, Kay Sievers wrote:
> On Tue, Jan 03, 2006 at 05:19:29PM -0500, Dan Williams wrote:
> > Another WPA-related update.
> > 
> > Major Changes since Dec 22
> > ------------------------------
> > 
> > 1) NMDevice has been refactored and GObject-ified.  Wired and wireless
> > are now subclasses.  This should make a Bluetooth DUN device quite a bit
> > easier to write, for example.  Anyone?
> > 
> > 2) The applet respects capabilities for both access points and devices
> > 
> > 3) Normalization of how generic wireless ciphers in libnm-util return
> > hashed key information
> > 
> > 
> > What's left to be done?
> > -----------------------
> > 
> > o Hook up WPA options in the applet
> > o Create a supplicant_manager object that controls invocations of
> >     wpa_supplicant
> >     o Write out correct wpa_supplicant config file and ask the
> >         NMAPSecurity objects to write out their security information
> >     o Connect to wpa_supplicant's control socket to monitor association
> >         status
> 
> Dan, can you comment on:
>   http://mail.gnome.org/archives/networkmanager-list/2005-December/msg00193.html
> 
> and let me know, what you think we need more than this. The code
> should work for setting up WPA PSK from NM and listen to wpa_supplicant
> events from NM.

Hi,

That looks like a great starting point, thanks.  I'm in the process of
sucking in bits of it to NM as we speak.  Of note is that for the
moment, I'm going to run wpa_supplicant in non-daemon mode, and also
that we'll run _all_ 802.11 stuff through wpa supplicant, including WEP
and unencrypted.  Makes our code a lot simpler.  Though I'm quite
curious what's going to happen when we trigger scan, whether
wpa_supplicant is able to hold onto authentication and whatnot after the
scan is complete.

Eventually we need a dbus interface for wpa_supplicant, something Jouni
said was cool as long as it was #ifdef-ed.  Which sucks, but hey,
whatever.

Dan