Possibly offtopic : Binary only driver

[Stefan Sonnenberg-Carstens]

What I think about.
>From a users point of view, an average user
could expect the following :
Buy hardware, insert installation cd, start installer,
plug-in hardware. Go working.
On Linux it depends on which hardware is supported by which distribution
out of the box. That is one of the reasons why SUSE is so beloved
here in germany by so called daus.
Normally the linux drivers on the customers homepages don't fit
your distro/kernel pair.
And how many users are able to start a simple make ?
Sad enough, not many. There are bothered buy a system which
"forces" them to dig in mud.
Users don't want to dig in mud, they want to dig the mud.
We support many users with a well written windows software.
"Oh, you inserted the USB device before installing, the driver ...
... oh, you did not read the manual, which states that ..."
Want more ?
The industrie is telling the user that everything with a PC
is as easy as 1,2,3.
It's not, and there pissed if someone tells them : "no, you can't..."
That sounds to them like "you are an asshole not knowing that you
must ...".
You take every experience away and declare it useless.
That is the *real* reason many companies fear the penguin - or the fish.
What they've learned and believed in over decades is shit then.
So I think that is the real challenge of a modern Linux system.
Give the user (the end-user, not the seasoned UNIX admin or techie)
the feeling he's the dude.
If that means offering the same methods other systems offer, ok.
Who cares ?
People expect executables to be ending to *.exe.
I have a co-woker, who started using apache after I created a symlink
called "apache.exe" to apachectl.
"apache.exe start", I said.
Surprise, surprise.
Now he's the dude.
So what about drivers ?
People expect setup.exe oder usb-device.msi under their
pillow.
That *should* be possible.

Cheers,
Stefan

Am Sonntag, 21. November 2004 19:06 schrieb Mike Hearn:
> On Sun, 21 Nov 2004 12:37:48 -0500, Sean wrote:
> > Both of which run under the protection of the kernel.
>
> So what? It still involves debugging programs with binary components.
>
> > It's not capitalism per se, but unfortunate patent laws and other legal
> > entanglements.
>
> I think you missed the point: if nVidia took a (let's be optimistic) 5%
> performance hit to move stuff out of the kernel, ATI would make a big deal
> of the fact that their systems were faster. Performance is one of the
> primary factors people use to choose which vendor to go with, so more
> people would go with ATI.
>
> If you keep making those sorts of decisions you end up bankrupt, because
> ultimately you're not meeting the customers needs. For once, patents and
> other "legal entanglements" don't enter into it.