[pkg-wpa-devel] What's wrong with the new configuration scheme, IMO

[Reinhard Tartler]

On Wed, Apr 26, 2006 at 12:15:45AM +0200, Felix Homann wrote:
> >  * Make the daemon use an action script.
> >  * the action script checks what location we connected to (look at the
> >    essid), and does this:
> >    * determine which network alias in /e/n/i this location is associated
> 
> I guess that this isn't necessary in a lot of cases, since DHCP is very common 
> in the wireless world.

If this was true, then ifupdown wouldn't be that needed here. 'dhclient
$IFACE' is enough to configure an interface with dhclient.

> >    * call ifupdown like this: 'ifup $IFACE=$LOCATION'
> >  * On disconnection, the action script ifdowns $IFACE
> 
> Why not integrate nicely with existing packages: let ifplugd do this task.

How do you tell ifplug which mapping an interface should get? Yes, with
an ifupdown mapping script. Can you please write such an mapping script?
How can a package (not ifplugd) make ifplugd 'monitor' the interface.
Please consider the cornercase that the interface may not exist (yet) at
start time of ifplugd. 

> > This package makes the following assumptions:
> >  * Only one interface is to be managed
> 
> Why this restriction? Would it make the action script too complex? (Then you'd 
> better use ifplugd+guessnet; you can easily run 1 wpa_supplicant for 2 
> interfaces or 2 wpa_supplicants for 2 interfaces.)

not the action script, but the init script + its configuration.

> >  * by default, it assumes that the alias in /e/n/i matches the essid
> >  * otherwise check a mapping file [optional]
> 
> Then I would have to edit wpa_supplicant.conf and /e/n/i for every single AP I 
> want to connect to even if they are all using DHCP? 

I was thinking about a mapping 'default', which is plain dhcp.
 
> > Problems which I still have to think about:
> >
> >  * What happens if we connect to a yet unconfigured location?
> 
> It should not happen without user interaction.

Are you kidding? This is an important cornercase, which DOES happens.

> >  * What if there have been no locations configured yet? (shall
> >    wpasupplicant run anyway, allowing managing of wpa_supplicant.conf
> >    via wpa_cli/wpa_gui, or should it doesn't start the daemon at all?)
> 
> Ask the user on install. 

what about installations in chroot? No, I'm not happy at all with user
interaction at install time if avoidable.

> >  * ifdown will bring the link down. Does this confuse wpasupplicant too
> >    much?
> 
> No problem on my machine.

On my machine, wpasupplicant seems to stop scanning after bringing down
the interface. I have to ifconfig up it to make it start scanning again.

> > I think this follows the guessnet/ifplugd configuration scheme, but I
> > haven't looked too much into guessnet yet.
> 
> It's actually so very close to a wpa_supplicant+ifplugd+guessnet (I'll call it 
> WIG) setup that I'd rather suggest to put the init script back into 
> wpasupplicant (don't activate it by default) and add some documentation on 
> how to configure a WIG system.
> 
> I don't really stick to WIG - I know a lot of things a WIG system cannot 
> (easily) do - but the system you propose seems inferior, sorry. 

Ok, you just convinced me that working on such a package is not a
priority for me anymore. I'm tired and angry about your continous
vituperating on this list, and don't want to read that anymore.

Please stop trolling by proving a package which implements your 'WIG'
behavior.

Reinhard