Subject: Laptop Wireless

Having spent an obscene amount of time working on this off and on over
the last year, the short answer is: none.

The somewhat less short answer is: atheros-based devices

1. atheros devices using the 'proper' drivers and kernel revs works
well enough most of the time if you don't mind cussing and tweaking
everytime someone sneezes. I use an atheros pcmcia in my laptop, but I
hang on to the drivers that are working, and don't 'upgrade' either the
madwifi, wireless-tools or wpa_supplicant or any of their dependencies
until I'm bludgeoned into it.

2. devices using the zydas (1211b?) chipset works well enough most of
the time if you don't mind cussing and tweaking everytime someone
sneezes. =20
I've got a couple of zydas-based usb devices that I use with fair
success most of the time; they do tend to drop offline at random times
so they have to be attended all the time or have a watchdog running.

The (newer) orinocos use the atheros drivers, but those use a closed
binary-only 'blob' in a wrapper (similar to how ndiswrapper works).
Sometimes those work and play well with iwconfig and wpa_supplicant.
Sometimes they don't. Sometimes both in the same day.

To be fair, I'm prejudiced against closed drivers, but that's me. To be
fair, I think most of the atheros driver issues not working and playing
well with others has do do with the mad_wifi developer and or the
wpa_supplicant developer and his/their 'quirks'.

The zydas and ralink chipsets have drivers that have been gpl'ed from
the ground up with the manufacturers full cooperation (apparently), but
both are still early-beta at best. Wpa_supplicant seems to be
problematic at best with both. I never did succeed in getting a ralink
device (I had three) to work reliably and repeatably unencrypted, and
never did get them to work at all with wpa. Don't remember if I tried
wep.

Someone asked a while back what windows did that linux didn't do as well
or better. =20

Wireless. =20

Not that that's entirely or even mostly due to 'linux'; much of it is
manufacturer's closed mindedness (TI and atheros-sort-of and
Intel-sort-of come to mind), or the quirks of the developers, or the
'random' changes that any given package (including the kernel devs) may
implement. Or simply the time commitments required to do a driver; most
of these devs are volunteers after all.

Oops...rant-mode snuck on again.../rant

---------
To unsubscribe, send email to <aklug-request@aklug.org>
with 'unsubscribe' in the message body.
Received on Wed Oct 4 09:13:48 2006