cheap ubuntu laptop with wireless that works right "out of the box".

[Derek Broughton]

Christopher Copeland wrote:

> I never said everything needs to be open-source, but given the choice
> I'll prefer a driver that is. 

Sooner or later when you're dealing with NICs you get down to a level of
closed source - it might be firmware and it might be an NDIS driver, but it
really makes no difference in the long run.

> Same when it comes to choice of OS. Watch 
> out now because here comes my argument.. 

:-)

> I've used ndiswrapper and 
> native drivers on a dozen or more wireless chipsets (always testing both
> on supported cards) and in my experience have found the native drivers
> more stable and offer better performance. 

I confess to have only used 3 different wireless chipsets (I keep meaning to
try out the D-Link that my wife used to use on her now-deceased Windows
machine).  One (intel) worked fine with a native driver.  Two (a TI
acx1100 - I think - and this one, a BCM94311MCG) worked only intermittently
with native drivers, but flawlessly with ndiswrapper.

> This includes long distance point to point connections, 

1000ft to the router connected to my satellite uplink...

> running high throughput always on 
> connections etc. Then consider the chipset features unavailable to
> ndiswrapper users.. try putting a card into master or monitor mode. 

I accept there are times it may _not_ be sufficient to use ndiswrapper, but
not for most people.

> What about link quality? 

Not noticeably degraded - certainly better than with the native drivers.

> *my* choice would be to use hardware 
> from manufacturers that release code or specs and have native drivers
> available. Then I can always use ndiswrapper as a backup.

That was my choice with my last laptop - I specifically ordered a Dell with
an intel card.  Then Dell refused to honour the warranty I paid for, and I
bought a cheap NIC that worked with ndiswrapper.  Next laptop, I just
decided it wasn't worth the aggravation, and bought cheap, planning to use
the internal wireless if it worked, or the cheap NIC if it didn't.

>> And, in what way, is selling hardware that works well in Linux with
>> ndiswrapper _not_ playing well with the community?
>>   
> Surely you can see how releasing a product with only windows drivers is
> *less* supportive than releasing hardware specs or a native linux
> driver?

They're not.  They're releasing hardware that works with a known software
interface.  I will accept only that if they haven't actually tested it with
ndiswrapper and provided installation instructions for Linux systems using
ndiswrapper that they really couldn't care less about Linux, but imo if
they've done that much they're playing nice.
-- 
derek