On 2005-06-07, Ricardo Pinto <rncp@mega.ist.utl.pt> wrote:
>
>
> On Tue, 7 Jun 2005, Rui Paulo wrote:
>
>> On 2005-06-07, Ricardo Pinto <rncp@mega.ist.utl.pt> wrote:
>> > 	The idea is to be as universal as possible... but I think this
>> > implies a slight change in device drivers.
>>
>> What change exactly do you have in mind ?
>
> Just as the ath(4) with dev/ic/athioctl.h , the inclusion of ioctl(2)
> interfaces (and if they were 'standardized', the better)

I think you should rely on net80211 ioctl(2) interfaces, not driver specific
ioctl(2)'s or you'll end up by coding a WiFi browser for every chipset.

>
>> >From wiconfig(8):
>>
>>      -D          This forces the driver to initiate one round of access point
>> 	         scanning.  All of the access points found are displayed.
>>
>> Isn't this functionality what you require or am I missing something ?
>
> Exactly! but wiconfig(8) only works for Prism based cards... and yes, it
> resorts to the same ioctl(2) 'trick'  (see dev/ic/wi_ieee.h)

I see. The problem here is that this kind of functionality seems to be
driver-dependant, so, unless we come up with an abstraction layer, it will
only work for Prism chipsets (for instance, ipwctl(8) doesn't support the
-D switch as wiconfig(8)).

-- 
Rui Paulo <rpaulo@netbsd-pt.org>