Re: Circumventing the GPL via open standards

On 11 Jun 2002, Martin Wilck wrote:

> This thread is slowly getting OT (this is wine-license, remember).

Yeah, I know.  But I didn't elaborate about UDI at first, but only after 
someone else responded about it...

> Deven, why don't you open a website,
> e.g. www.devens-daily-commentary-on-GPL-evilness.com,
> to publish your ever-new insights ?

I'm not interested in becoming an anti-GPL crusader.  I have concerns about 
the GPL, and I'm no longer the unqualified GPL supported I once was, but 
that doesn't mean I consider the GPL evil.  I consider Microsoft evil, and 
I've concluded that the GPL is proprietary.  However, unlike RMS, I don't 
automatically equate "proprietary" with "evil".

> I shouldn't reply to this, but I can't help myself, thus just a few
> remarks:
> 
> - If Linux needs proprietary UDI modules to run on tomorrow's
>   hardware, it will cease to be a free operating system. The kernel may 
>   still be free, but usable on modern hardware only with the non-free
>   drivers. The free OS we are all enjoying now will be history.

Not everyone is as rabidly opposed to proprietary software as Stallman is.

Another point to consider is that UDI compatibility is only guaranteed at 
the SOURCE level, which gives vendors an incentive to release the source.  
It would be better to lobby hardware vendors to release their drivers under 
a free license like the BSD license and try to make THAT standard practice.

Linux has always had to play "catch up" with hardware driver support, and 
other free operating systems now also have to play "catch up" with Linux.
UDI could improve the reliability of device drivers (they can run protected 
in user mode, to keep the kernel safe, if the environment permits that) and 
allow all free operating systems to freely share device drivers and stop 
duplicating efforts so much.  This is a good thing to anyone who wants all 
free operating systems to prosper.  Maybe it's a bad thing to GPL zealots 
who see anything non-GPL as a threat to their proprietary GPL model.

Yes, it's possible that fewer device drivers would be written.  Why would 
that be?  Because many of them are written out of necessity, not because of 
an innate, blind hatred for all things proprietary.

> - Availability of proprietary UDI drivers will certainly diminish the 
>   willingness of people to write free ones. This is about the same
>   reasoning Alexandre used wrt Direct3d.

It would diminish the NECESSITY.  Anyone who believes in Stallman's vision 
of "freedom" is free to implement "free" driver alternatives for any such 
"important" hardware as Stallman refers to.  (Apparently, users stuck with 
"unimportant" hardware don't matter to him.)

Moreover, the UDI driver environment is VERY constrained, which means it 
would probably be EASIER to reverse-engineer a UDI driver than a Windows 
driver, which is what's commonly done now.  It could be made even easier 
yet by designing a custom UDI environment designed to provide maximum 
transparency into the workings of the driver being used, which might make 
it even easier to reverse-engineer UDI drivers.

In short, it's possible that UDI could INCREASE the supply of free drivers.  
Certainly, if it threatens the supply at all, it's because the supply of 
free drivers has been artificially enhanced by necessity, the mother of 
invention.  If Stallman really wants people to develop free software for 
its own sake, he can be more certain that's what's happening when there's 
proprietary options to be turned down.  It's not a real choice if there 
aren't multiple options.  (But like any monopolist, Stallman doesn't really 
want the users to have real choices, only choices among those he offers...)

> These are exactly the main concerns expressed by RMS. IMO he isn't so
> wrong, after all.

Even if he's right about these concerns (which I doubt), is that reason 
enough to demand that device driver development remain Balkanized as it 
currently is?  Is it worth retarding the advancement of computer science 
for political gain?

> - It is close to ridiculous to claim that Linux developers blindly
>   follow RMS's views. Linus himself is most often viewed as RMS's 
>   opponent. Look at this list! The accusations against Stallman as
>   a totalitarian world domination seeker go so far that I, being 
>   certainly no follower of his, feel forced to take his side.

A large fraction of the Linux community is composed of GPL zealots who 
treat RMS as some sort of prophet.  Linus himself is NOT one of them.  
Neither is Alan Cox.  They are pragmatists, and would not refuse to use
UDI on the basis of Stallman's arguments.  However, they have their own 
concerns about UDI, mainly that they believe there would be a performance 
cost inherent to the design, and they're not willing to take that risk.  
(Even if the penalty is only 5-10%, they'd be opposed.)

Interestingly, some of the work with the existing UDI prototypes suggests 
that the UDI model's better design might actually allow UDI drivers to be 
FASTER than traditional drivers, especially if the UDI drivers are not 
layered on top of traditional drivers.  SCO has found that prototype 
drivers have had comparable performance to native drivers despite being 
layered on top of traditional drivers!  Considering that those drivers 
(being prototypes) weren't optimized for performance, that bodes well for 
the possible performance of native, optimized UDI drivers/environments...

But, being pragmatists, Linus and Alan will probably need to see proof 
before they'll be willing to sign on board.  Of course, if they do sign
on board with UDI and start actively supporting it, it will go a long way 
to counter Stallman's shortsightedness regarding UDI; they're almost as
revered as Stallman is, and much more respected by many.  (While the Linux 
community has many Stallman followers, there are plenty of his critics in 
the ranks of the Linux community as well.)

I might point out that the Linux community wasn't asked by Project UDI to 
provide a Linux UDI environment implementation -- one of the proprietary 
companies (STG) involved in Project UDI wrote a reference implementation 
for Linux and released it under minimal license -- it only says that the 
software is "as is" with no warranties and no support, and disclaims 
liability.  That's it.  They probably should have explicitly permitted use, 
modification and redistribution.  I think that's implied, but it would be 
better stated outright...

I still think UDI is the wave of the future, and a great opportunity to 
unify all of Mirosoft's opponents (free and proprietary) to help break 
Microsoft's stranglehold over the operating system market.  We should jump 
at the chance to fully take advantage of this hard work being freely given 
to everyone to use...

Deven