Re: [Savannah-hackers] submission of Waves, Clouds, and Sand - savannah.nongnu.org

[planet10]

> > What about features that are part of the OpenGL spec, e.g.
> > ARB_vertex_program instead of *_NV or ATI_* ?  Mesa now supports
> > OpenGL 1.4.
>
> If those features have not been implemented on any free driver yet, then they
> should be avoided in a Savannah project.

Mesa is free (MIT license).

> This would not be the first time we fall way behind in some technological
> aspect for the sake of ethical concerns which we put above technical
> concerns.

I might be reassured if you could tell me what positive steps are being
taken to get modern, free 3D working on GNU systems.

Of course I understand the ethical concern, but usually it's more
effective to have a 2-part plan: avoid proprietary component while
lobbying for/creating free replacement.

> Some of us do. I have a 3D NVidia card and I cannot even play tuxracer. I've
> seen it working in someone's computer with an nvidia driver and I must confess
> that it looks nice, but I'd rather stick to 2D games for the time being rather
> than installing a driver that is not even in the official Debian CD.

Spending the money to get a 3D card and then not using it doesn't help
convince NVidia to release a free driver, either.

> If those who took the free-software ride for its technical advantages decide
> to jump into another train when this one seems to go too slow, it will be a
> pity for them but we will be left with those who share our ethical views.

Rather than planning to be left behind, it makes sense to plan for the
future where everyone will own a 3D card and use it; when this happens,
I'd prefer to be able to do so on a GNU system than to sit there being sad
about all the sell-outs.

I don't know what's going on behind the scenes, but at the moment it looks
to me like 3D isn't a big concern for the FSF.  Maybe I can help motivate
others to make it a priority.

> Not really. The GNU project attempts to create a complete free operating
> system. The GNU manifesto does not say "thy shall do everything proprietary
> software developers do".

An "operating system" includes all the components that people generally
like to use, right?  I mean, we're not just talking "login" and "halt".
Well, a great many people like to play games that use facilities currently
unavailable on a GNU system.  It's not just games -- 3D geographic
studies, medical imaging, weather simulations, etc. will all be relegated
to proprietary land if we don't get crackin'.

Drivers are certainly part of the OS, and thus of concern to GNU/FSF.

> Or we can use our programming skills to make free drivers for the new cards,
> rather than developing games that depend on the proprietary drivers. If that
> task is impossible because the vendors refuse to release some important
> information, we will then have to live with 2D or come up with a new 3D scheme
> that relies on the CPU and not on the graphic card's chipset.

Mesa and a 10 GHz CPU ought to work.

In my case, I want to work on the "it actually does something" side of
programming as opposed to the "it enables you to write a program that does
something" side.

> > Having a large number of free 3D programs available which use modern
> > hardware capabilities seems to be an improvement over the current
> > situtation of only the proprietary programs being able to do so;
>
> An improvement from the technical point of view, not from the ethical point of
> view.

No, an improvement in BOTH areas.  Right now we have proprietary games
with proprietary drivers; in the future we will have free games with
proprietary drivers, then free games with free drivers.  It's the middle
step.

Having people learning from and sharing advanced 3D games is surely better
than binary-only games with 10-page EULAs.

> the end does not justify the means; and I think it is very doubtful what will
> make proprietary vendors change their narrow minds. We should rather make
> plans that depend on our own (the free software community) actions.

Software doesn't exist in a vacuum -- okay, well maybe embedded software
does ;) -- you always depend on hardware.  So, you can't make plans
completely independently of what hardware manufacturers do.  Even if you
think, "I'll always have my Pentium 100", it will eventually stop working.

The best way to avoid a proprietary future is to convince lots of people
that your philosophy is correct.  If you can't even run a 3-year-old
game on your system, it's going to be a hard sell.

Technical considerations can't be ignored.  Look how many people the Hurd
is converting to the FSF philosophy.

> can install it in my 100% free system because it is part of Debian; it runs,
> but it works very slow and not very smooth. It could be a Savannah package
> anyway, but if it was, we would expect its developers to try to resolve our
> bug reports complaining that it does not work well in our systems, without
> telling us "go and install the nvidia package from Debian non-free".

Of course it's easier preaching to the choir than doing missionary work.

However, it doesn't make sense to expect game developers to make their
game run better on an incomplete, buggy driver.

Direct the pressure gradient outward and talk to the hardware
manufacturers.


> And how do you intend to pass the word to NVidia that people are using their
> cards to run tuxracer rather than some proprietary game? All they will see is
> an increase in their sells, which is what they want, and will think that their
> policy of keeping technical details hidden is working very well.

Tux Racer is not exactly a high-visibility project.  You need a popular
free software project that attracts enough users that "get it", who are
clamoring for NVidia to release a free driver.

It worked with TrollTech and QT, and MySQL.