On Tue, 18 Sep 2001, Will Newton wrote:
> On Monday 17 Sep 2001 9:16 pm, you wrote:
>
> > There's no good reason why you couldn't put an NVIDIA card in a sever
> > and use the open source 2D driver. You wouldn't want to enable DRI on
> > you server, either, because of lockup issues.
>
> Look at it this way: a couple of security holes have been found in the DRI
> drivers, and fixed. What's in the nVidia drivers?
What's in any binary application or driver you install?
> > You may feel that the NVIDIA driver sets a bad precedent, but it also
> > sets a high standard. Their drivers are high performance, complete, and
> > stable. They have several people with histories in open source 3D
>
> I've never used them, but a lot of people have told me they can't get them to
> run stable.
I'd guesstimate that it's roughly the same percentage of people who can't
get the DRI drivers for their card to run stable, either. Hell, my
Voodoo3 still locks up my machine :-)
> > efforts, people who care about 3D on Linux. You trusted these people
> > with your drivers when they worked on Utah and DRI, why shouldn't you
> > trust them when they work for NVIDIA?
>
> Because I can't see the source. No, I don't read the source of every program
> I install, but I do read a lot of source code. Why do I trust myself more
> than the people who wrote the driver? I don't, but that doesn't mean I, or
> some other person wround the world looking at the code will not see something
> the authors didn't.
>
> > If binary-only is the only way to generate enough revenue to support 3D
> > driver development for Linux, it should at least be considered.
>
> People are willing to use them, so fair enough. But whilst I have a
> reasonable choice I will use an open source driver.
Unfortunately, your reasonable choice doesn't look like it's going to last
much longer.
> > The important thing is to have drivers available from a source you have
> > some confidence in. If Linus wrote a binary-only kernel module that
> > provided some functionality you require, would you have faith in it?
>
> Not particularly, and that's the point. I don't think Linus would have that
> much confidence in the driver either (hubris aside).
>
> > No, it's not that simple. How long have the G400 specs been available?
> > I still have lockups and general flakeyness with DRI/G400. Stable
> > drivers don't just magically appear when specs arrive. You need talent,
> > and you need time. Even with the amount of talent working full time on
> > the DRI project, things have been moving slowly. It's a big, difficult
> > job that needs a lot of attention to do correctly.
> >
> > All the open source cheerleading the in world isn't going to make
> > drivers appear from thin air.
>
> Seriously, if I had a G400, I would help fixing bugs in it.
Not everyone has the programming skills to do that. And even many of
those that do, don't want to have to fix the bugs in a driver just to get
their hardware working :-)
> Unfortunately I'm a poor student so I have a crappy old 3dfx
> Banshee[1] card. So unless I win the lottery there is not much chance
> of me personally writing/fixing any DRI code in the near future.
And it's because so many people feel that way that there are so few
volunteers with the DRI development.
> [1] Only has one bug that I have run into, which is so esoteric it's hardly
> worth fixing.
I wish I were that lucky :-)
Adam
_______________________________________________
Dri-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/dri-devel
