Re: [gentoo-amd64] Re: URGENT: No more fonts after xorg changes

[Boyd Stephen Smith Jr.]

On Sunday 03 December 2006 09:24, Neil Bothwick <neil@××××××××××.uk> wrote 
about 'Re: [gentoo-amd64]  Re: URGENT: No more fonts after xorg changes':
> On Sun, 3 Dec 2006 07:32:23 -0600, Boyd Stephen Smith Jr. wrote:
> > > That leaves Nvidia two choices, distribute closed source drivers for
> > > Linux, or remove that code from the Linux drivers
> >
> > ...or both (release a "free" version and a "full" version), or release
>
> They'd probably argue that would be extra work.

I agree; it's still a choice.  Just exposing the lack of truth is 
the "false dichotomy" that was originally proposed.  Maintaining two 
separate drivers would probably be hell; even if both derived from a 
common source-tree it would require a code-audit before every release of 
the "free" version.

> > specs so we can write our own drivers (we don't need to know anything
> > patentable or even trade-secret-able, just how to talk to the
> > hardware) or a half-dozen other options.
>
> Haven't they done that already?

No.  The nv driver is reverse engineered, IIRC.

> It's just that this doesn't work with
> 3D, which is tied into the proprietary code. Without knowing the
> details, which would require an NDA and then not being able to discuss
> this, there's no way of knowing how closely the 3D stuff is tied into
> the proprietary code.

What we it is something similar to e.g. the x86 instruction set.  That 
doesn't tell you anything about the specifics inside the chip, but allows 
to write software that takes advantage of all features.

To write a 3D accelerated driver, we need to know what primitive operations 
the cards support and how to invoke those primitive operations.  How to 
translate application-level function calls (opengl, xlib, whatever) into 
those primitives is something the F(L)/OSS community would have to figure 
out.

I'm sure it would take a year or more to get to the performance and 
stability of the closed driver, but at least it would be feasible to write 
and maintain a 3d-accelerated driver in the "free world".

It's possible that we could reverse engineer this information in the 
future, but that could also mean some broken cards, and it would be nearly 
impossible to maintain.

> > NB: I use the nvidia driver; I don't like it, but I do use it since I
> > do occationally play games that require accelerated 3d.
>
> So use a different make of card that provides equivalent performance
> with a free driver.

When next I change video cards, I will be doing just that.

-- 
"If there's one thing we've established over the years,
it's that the vast majority of our users don't have the slightest
clue what's best for them in terms of package stability."
-- Gentoo Developer Ciaran McCreesh