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

[planet10]

> The main idea behind the restriction on Savannah is pretty simple:
> actually, it is now possible to use a completely free system, as
> GNU/Linux exists.
>
> As now this completely free system exists, new bricks to add to this
> free system mustn't depend on non-free bricks. It would be adding
> proprietary software in the chain.
>
> You said that people having brand new 3d cards do not want to use 2d
> graphics. I have an ATI Radeon at home. Indeed, I use them in 3d when
> playing at game, that actually are proprietary, running on a
> proprietary system.
> While I use a free system (98% of the time, when I'm not playing), I
> use VESA drivers. Sure I do not take full advantages of my hardware,
> sure I do not play tremendous 3d games. But I won't spend time on this
> to make my free system dependant. While using proprietary driver on a
> system which is 98 % non free does not really disturb me, I would
> never add them on a free system... and neither I would promote 3d
> games that would incitate me to install proprietary stuff on my free
> system to use them.

So, because you don't have free 3D drivers, and don't want to install the
proprietary ones...  you use proprietary drivers anyway, to play
proprietary games, on a proprietary operating system.  (I'm guessing you
use Windows.)

Having a "completely free system" is essentially meaningless in that case.
Your physical box has a ton of non-free stuff on it.  The "free"
distinction does not exist for practical purposes.

So you encourage non-free games and drivers by purchasing those products,
which does not help the cause of either free games or free drivers on
GNU/Linux.

> If new free 3d games appears to be dependant on proprietary drivers,
> 3d cards manufacturers will never have interest on contributing on
> free drivers.

I think the opposite may actually be true; I think that if cutting-edge
free games appear, that will show 3D vendors that people are serious about
3D graphics on free systems. We don't have enough people asking for free
drivers since they are not exposed to the whole FSF philosophy when they
play their games on Windows.

Who do ATI, NVidia, et al. listen to?  Id software, Epic, Hollywood F/X
shops, and so forth.  Id makes games that help sell the hardware.  If we
stick with feature sets from two generations back for our free games, we
aren't going to be on their radar, and we won't get enough users to reach
the critical mass needed to successfully lobby for free drivers.

Reverse-engineering is too slow for something as complex as 3D hardware;
we either need help from the current vendors, or we need a new
graphics card manufacturer that already digs free software.

> ILM recently accepted to free some software. I think that the good
> way.

They released OpenEXR.  It contains a 16-bit floating-point data type
("Half"), which is identical to the 16-bit float used by, guess who,
NVidia in their latest hardware/drivers.  You can read/write directly
from/to your floating-point framebuffer without needing an extra
conversion step.

That's yet another example of new features that aren't supported in the
free drivers, so 3D on free systems gets even further behind.

> To promote free drivers, it would be surely helpfull to have
> high-visibility projects relying on them.

I'm not sure how that would help, because the end-user with the vendor's
proprietary drivers will be able to run the same software that any partial
free implementation would be able to run.

Additionally, in order to test free drivers, you need software which uses
those features.  Right now that means proprietary games.  I don't see how
to get away from proprietary software completely when making a replacement
for either a driver or the software which uses it.

If I create a free game now, I can develop against the vendor's reference
implmentation (proprietary drivers) for correctness.  That enables people
working on free drivers to test against my previously-developed,
known-behavior game.