On Fri, 2006-04-28 at 20:20 +0100, Ian bell wrote:
So to summarise, ndiswrapper is buggy and doesn't work for some people and the rt2x00 driver is buggy and doesn't work for some people
Well yes and no.
I think what Adam is saying is that one is a buggy driver that is in development and will improve with bug reports and further development.
The other is a (admittedly quite clever) hack that approximates the Windows network driver API just well enough to load Wireless network device drivers.
There are several reasons why the second one can only ever be a hack.
The driver API is a moving target, so with this method we have two moving targets (the API and the driver) Both of which are proprietary and some of which may be patent encumbered, therefore this solution can never be 'perfect'
The driver itself is closed source..This is even worse than using things like the closed Nvidia drivers because at least there is a degree of support from Nvidia as they are running on the intended target platform.
It is restricted to only architectures that both Windows and the hardware manufacturer support.
Also there is the classic anti-wine argument.
If wine was perfect then it would probably result in less native linux software being developed or ported because there would be less itches to scratch, therefore Wine solves short term goals at the expense of long term ones.*
So really (and perhaps in this case seeing as the native driver didn't work for you this was the only viable option) you should only use ndiswapper when there is no other alternative, it's a last resort. In the same way that running stuff under wine is really a last resort because you can't find the native software to achieve the same functionality.
So basically you should put as much effort as you can afford into getting the first option working first, before resorting to the second one.
* regarding the anti-wine argument, I don't completely subscribe to this view but it's a view that many hold so it's worth mentioning.