[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: nvidia



Somebody in the thread at some point said:
> Bruno Wolff III wrote:
> 
>>> In my experience, open source drivers have as many or more problems
>>> than the closed source versions.  _If_ you are Alan Cox or some
>>> number of hackers with equivalently specialized skills that you could
>>> probably count on one hand, having the source code available might be
>>> of some value when the supplied binary doesn't work.  The rest of us
>>> report the bug and wait, and again in my experience over the last
>>> couple of decades, the closed source providers are at least equally
>>> responsive in this scenario.
>>
>> The other category of people helped are those with enough money to hire
>> a driver expert to fix the driver for them. That doesn't cover (almost
>> all) one off users. But large enough organizations could potentially
>> afford this.
> 
> The issue is theoretical at best.  In the unlikely event that access to
> a video card breaks due to undiscovered bugs in the original _and_
> vendor refusal to fix it, I'd expect it to be cheaper to either replace
> Linux or the card than to hire an expert to temporarily revive the
> now-dead combination.

Well whatever your other complaints, I really don't think you take into
account the developer suffering that happens from the unsupported
reverse engineering aspect that is often part of the drivers.  Sometimes
folk don't even get thrown a datasheet from the manufacturer.  You just
stare at hundred and even thousands of 32 bit registers with twinkling
mystical unknown purposes and try to figure out how to make it do
something useful.

It's like the proverbial dog playing the piano, it's not that it didn't
play it perfectly (although it often does in my experience) but that it
can play it at all that is remarkable.  I looked through lshal the other
day and it was listing dozens and dozens of devices in this laptop that
had FOSS drivers.

More than that though I myself have taken advantage of a kernel driver
blowing a panic to look through the source and fix the problem, and send
a patch describing and fixing to problem, which was accepted.  I was
really excited by being able to do that and still think that system must
most times knock crap out of doing it in-house with proprietary code.

As more manufacturers support external efforts with data quality and
completeness will definitely generally rise.

-Andy


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]