[PLUG] Fedora or Red Hat (or OSX)

[AthlonRob]

On Fri, 2003-12-12 at 09:08, Carla Schroder wrote:

> 1. their products need software drivers to work at all
> 2. good drivers and docs = less cranky customers pestering tech support, 
> and maybe even increased sales

IME, hardware makers, if they provide any technical support at all, make
it very difficult to locate and make it cost money in terms of long
distance phone calls...

> 3. one nice *nix tarball feeds every single Linux, BSD, and Unix. Nice 
> package maintainers roll them into handy binary packages for those who 
> prefer such

I'm not too up on cross-platform driver development, but I really don't
think you can write one nice little driver and expect it to compile
under Linux, FreeBSD, and Slowaurus.  If that were the case, FreeBSD
would have the same driver support Linux has, or very close to it... and
from what I understand, that just isn't the case.

> 4. But each winduhs - 9x, NT, ME, 2000, and XP- needs its own binary 
> driver. Which often must be modified because the latest Service Pack 
> hoses it. Plus we all know how upgrading to the next winduhs means 
> either waving bye to many of your peripherals, or spending wads of 
> money on new hardware. But that's OK, somehow

Just playing devil's advocate...

A few things... 9x and Me can share binary drivers.  2K and XP can share
binary drivers.  NT and 2K can share some binary drivers.

There are, I believe, more folks buying new hardware devices to run them
on Windows 98 than there are buying new hardware devices to run them
with any Unix workalike.  On a cost/benefit analysis, if you have
limited driver writing resources, it seems to make sense to me to build
one driver for 9x and one for XP (which will probably work with 2K). 
Right there you've covered what... 95% of the market?

That said, if they would just release a little bit more information
about their hardware products, there are a multitude of volunteers out
there who will take that and turn it into a real driver for Linux.

> 5. Linux developers are valiantly reverse-engineering and writing 
> drivers anyway. But supporting this free labor pool with specs and 
> product is, somehow, Wrong.

One interesting perspective on drivers I read a little while ago had to
do with Patents.  It may have been here, so this may all be a rerun. 
Patents are so crazy these days, everything is patented.  You cannot
design a piece of hardware today without inadvertently stepping on a
half dozen patents.  If they released the specs of the hardware so OSS
writers could design drivers for it, you would also be opening up enough
of the hardware to the public so your competitors could see where you
accidentally stepped on their patent and then... you get sued.

> A little hobby of mine is every so often I make the rounds of various 
> hardware companies, and ask them about Linux support. Why don't they 
> have any. Do they have any future plans to offer it. The 
> Linux-unfriendly companies, without exception, respond in one of two 
> ways: they ignore me totally, or they claim "when customers ask for it, 
> we'll look at it." Which everyone here knows is complete bunk. 
> 
> So, according to the responses in this thread so far, the main reasons 
> for the lack of Linux hardware support are incompetence and confusion. 
> This is what we trust our computing infrastructure to? People who can't 
> say duh without stuttering? People who can't tie their own shoelaces? 
> Or have an independent thought, or make an independent decision?
> 
> I'll stick with my collusion-with-m$ theory. I'm not discounting 
> incompetence and confusion, but it's hard to believe that's the primary 
> reason. Why would any businessperson in their right mind purposely 
> ignore significant new markets? It makes no sense.

They don't see Linux *as* a significant new market.  Not only is it
still not very popular (a few years ago weren't we all saying it was
going to be *way* more popular than it is today, today?)... I think
Linux users tend to be cheaper when buying hardware than M$ users.

Rob