On Tue, 2009-06-23 at 13:49 +0200, Samuel Ortiz wrote: > > One question I just had -- if bssid is NULL, and the card selects the > > BSSID, I suppose it will also roam by itself if it can [1]. Don't we > > need a "roamed" event for that then, even if we're not yet adding a > > ->roam() call? > I guess userspace will get 2 events: a disconnect one and then a connect one. > Not exactly a "roamed" event though, I agree. Ok, I guess then it needs to distinguish between a disconnect that it caused locally (or some other userspace tool), and a disconnect that the AP caused. OTOH, I don't think this is appropriate. I think after a disconnect event, the kernel/device should stop doing _anything_. Otherwise we'll, on a smaller scale, repeat the wext problem where you don't know whether the device is currently doing something or not. johannes
Attachment:
signature.asc
Description: This is a digitally signed message part
