[patch] wireless network support

[Mathieu Lacage]

On Wed, 2004-05-26 at 20:29, David Zeuthen wrote:
> On Wed, 2004-05-26 at 15:10 +0200, Mathieu Lacage wrote:
> > I am not sure if this address is subscribed to the list so I apologize
> > if it gets moderated.
> >
> > On Wed, 2004-05-26 at 11:15, David Zeuthen wrote:
> > > My only other comment is that the properties should be called
> > > net.ethernet.wireless instead of just net.wireless, since other
> > > wireless networking technologies may or may not exist.
> > 
> > I would suggest net.80211 instead. It is clearly more precise: saying
> > that 802.11 is based on ethernet is, at best, a misconception.
> > Furthermore, other similar wireless standards are being developped. It
> > will make the naming convention for these newer standards more
> > reasonable if they can be made .net.standard.
> > 
> 
> I'm not sure myself, but looking at Joe's lshal output you will have to
> use both net.ethernet.mac_address and stuff from net.wireless.* to get
> the full story, since net.wireless.* alone doesn't tell you the MAC
> address of your wireless device.

well, copy the mac address from net.ethernet to net.ieee80211.
Generally, there is no reason other wireless network technologies will
use the ieee MAC addresses used in ethernet. So, it would make sense to
add the address of each device in the namespace of its underlying LAN
technology.

> At least the Linux kernel seem to treat 802.11 as an extension to
> Ethernet, so I think it makes sense if we do the same. I'm perfectly

No, it does not really. The linux kernel treats all LAN devices the
same. Most drivers for 802.11 devices present themselves as fake
ethernet devices but there is no reason this will continue forever. For
example, hopefully, someone will integrate someday a device-independent
802.11 MAC layer in Linux (such as the one in the MADWIFI drivers) which
will make it much more interesting for most wifi drivers to stop faking
ethernet devices.

> fine with calling it net.ethernet.80211 or something, I just think it's
> important that the properties appear in a namespace underneath
> net.ethernet. And, oh, the device should have three capabilities namely

I think this is fundamentaly wrong. Generally, this kind of relationship
implies the "is-a" relationship. Here, I believe it is not the case. I
find it very confusing.

regards,
Mathieu
-- 
Mathieu Lacage <mathieu.lacage at sophia.inria.fr>


_______________________________________________
hal mailing list
hal at freedesktop.org
http://freedesktop.org/mailman/listinfo/hal