To view parent comment, click here.
To read all comments associated with this story, please click here.
This is not a zealot fight. Kernel developers believe that non-GPL modules are a violation of the license of their code. This is a very simple way they found to help enforce their license.
If your lawyers tell you the Linux developers are wrong with their read of the GPL you can take 2 minutes to modify your copy and use that. You can even distribute it as "Wbeebex" if you want to.
It's not so simple on other cases, though. Apple goes through great lengths to assure their EULA is respected. Even if the law tells you their EULA is wrong there's nothing you can do about it, like installing your copy of Mac OS X on a computer not manufactured by Apple.
Microsoft also spends a lot of time to engineer their WGA scheme, which not surprisingly doesn't work too well. But its purpose is to defend the license of their product.
Some people have different standards. It's okay for Apple and Microsoft to defend their copyrighted work, but when Linux developers do a very minor thing with a similar purpose then it's blind zealotry.
And by the way, unlike someone mentioned there's no "time bomb" on the kernel. Non-GPL modules will continue to work fine after Jan 2008, they will only be blocked on versions of the kernel released after that date.
To sum up:
1) Kernel developers are enforcing the license of their work;
2) People that don't agree with their read of the license are still allowed to disagree with the kernel developers;
3) People are being given over one year of notice before this change happens.
This is not a zealot fight. Kernel developers believe that non-GPL modules are a violation of the license of their code. This is a very simple way they found to help enforce their license.
This is a zealot fight. Some kernel developers believe non-GPL modules are a violation of the GPL. Others - particularly Linus - believe that you cannot make a blanket statement about that.
What's not in dispute:
Your company designs a new network card. You decide to code a Linux driver for it. You do this by taking the code for an existing driver in the kernel and modifying it to work with your card. In this case, your driver is very clearly a derived work of the kernel and must fall under the GPL.
What is in dispute:
Same network card scenario as above. This time, your company decides to implement a Linux driver by creating a little bridge code to interface between the Linux kernel and your existing Windows codebase. You release the bridge code under the GPL, but release the rest of the driver only in binary form. Some developers - the ones arguing for this change - insist your entire driver codebase is a derived work of the kernel, and therefore must fall under the GPL. Others - including Linus - feel that your Linux driver is a derived work of the Windows driver, therefore, it falls under the same restrictions as your Windows codebase. Under this viewpoint, only the Linux specific glue code is a derived work of the kernel and must be GPLed.
While I'm all for enforcing the license/copyright - this is bull.
Seriously. Linux and the GPL are supposed to be all about Freedom. Sure, it needs to remain open - and the GPL (though it has a few flaws) does this well (imho).
The problem is that in many cases these are just individuals or small organizations that use these drivers. I doubt many enterprise Linux customers are installing Nvidia drivers (sure, sure, some might).
So...how's this hurting Linux? How is me installing a driver to make my system as functional (3D wise) as my Windows machine somehow violating copyright?
If Joe Linux installing the NV/ATI driver just to try to make his system comparable is somehow illegal/against the developer's wishes, then when this happens Linux will lose a lot of users. Will it die? Not likely. I'm afraid it will be relegated to a RISC type place in the OS world though.
Maybe I should start looking into BSD
"ome people have different standards. It's okay for Apple and Microsoft to defend their copyrighted work, but when Linux developers do a very minor thing with a similar purpose then it's blind zealotry. "
Actually, Linux fans badmouth MS's and Apples liscenses all the time, much louder than this. Just look at the uproar over running Vista Home in a VM
With the exception of the hard-core user, Linux is going to make themselves unwelcome with these antics. Who wants to get pulled into this zealot's fight? All most folks want to do is run the hardware they purchased and get the best experience possible with as little hassle as possible. Blocking binary-only "non-free" drivers is the ultimate in hassle. They just want it to work.
Talk about short-sighted. You're failing to see the other side to this issue. The reason your Mac works so nicely is that Apple has the specs and the sources to all the major hardware and all the drivers. The reason Apple and Microsoft can go mucking around to their hearts pleasure with new driver models and new 3D models is because they have the source to pretty much everything. The Linux developers don't, and that's a major hinderance. So its not just a zealot issue here, but a very pragmatic one: it's hard to maintain code that you don't have the source for, and if you want a first-class experience on your platform, you need to maintain your own drivers.
What I really don't get is the way Linux and GPL advocates constantly bitch about Microsoft and Apple having the advantage of having all the hardware specs at hand and yet keep on insisting that the quality of their own open source drivers top that of the proprietary ones from Microsoft, Apple and the other 3rd party vendors like e.g. NVidia. If the NVidia driver is this bad and insecure, them why even care about it in the first place? Is it simply because it is 'unfair' that some functions are not available in the open source 'nv' driver compared to the 'nvidia' driver? after all, I bet it's the 3D accelleration stuff that's important to both end users and coders here....
Of course the complete specs of a product are necessary for an open source driver to be able to completely replace a closed source alternative at any given time, as long as the functionality set of the open source driver is otherwise inadequate.
At the same time, though, that does not reflect negatively on e.g. Apple and Microsoft. The true problem is that the GPL puts some restrictions on the way proprietary and non-proprietary software come together and because of this, GPL advocates tend to believe that everyone everywhere should obey similar openness policies as they themselves enforce, simply for the same developers to be able to honor the license they themselves have chosen....
This effectively means that e.g. NVidia should open source their drivers under the GPL, in order for kernel developers not finding their copyrights violated, rather than accepting the current 'driver stub' approach as being close enough to satisfy their own goals for a project they themselves chose the license for... A project they really want to 'take over the world', in turn...
I just don't see how this can be NVidia's problem to solve, since it makes it okay for open source developers to loot on corporate intellectual property as this is being open souced by various parties, like e.g. Sun Microsystems, however simply benefitting from the same intellectual property is just not 'good enough'!
To me it seems like some open source advocates tend to enforce an GPL dictatorship on the world, because since they 'know' what is right in this world, everyone should follow. Examples of this way of thinking:
-Broadcom should open up its specs for their wireless chipsets
-Sun should go GPL with Solaris as with Java
-NVidia and ATi should go completely GPL
-Apple should completely open source Mac OS X so open source developers could get their hands on Aqua, Quartz, Quartz Extreme, Quicktime and every other little interesting tidbit under the hood
etc.
But face it: There's always three sides to every story, yours mine and the truth... And because of the truth existing somewhere in between you and me, co-existance and cooperation ought to be the answer...
There's no need for everything and everyone in the world of software development to following the same rules...
And concerning the ability to change driver models, I'd like to remind you that this has happened on Linux as well, breaking driver compatibility from time to time... As a counter example, Solaris has had a stable driver interface since at least 1998 allowing you to load drivers from at least that time on a modern Solaris 10 or Open Solaris.
Let me rephrase this:
"Shut up and stop having opinions. Your semi-public mailing list antics are scaring high school students off your product."
Unfortunately for you, Linux fails to be a product and few people who don't care at all about these issues read these lists (until some "news" site posts links to them). They're of interest to us, but until policy is inacted no one else cares. Even then few people will care...
Anyway though, now that Linus has ended the debate maybe you can go ahead and take your impatient foot out of your impatient mouth and quit getting mad that people who develop for Linux care about the rights of their creative works.







Member since:
2006-12-11
I really don't need this. I'm voting against this with my wallet. I've already purchased an iMac (17") for one of my daughters for Christmas. I didn't want to get a Windows machine because I don't trust Windows anymore. And I refused to get a machine and install Linux, any Linux distribution, because it's so easy to break and there's little in the way of quality applications that match what my daughter is used to under Windows and Mac OSX from high school. So I bought a Mac. And as time goes along, I'll buy more of them for personal use as well as for work.
With the exception of the hard-core user, Linux is going to make themselves unwelcome with these antics. Who wants to get pulled into this zealot's fight? All most folks want to do is run the hardware they purchased and get the best experience possible with as little hassle as possible. Blocking binary-only "non-free" drivers is the ultimate in hassle. They just want it to work.
And as an embedded developer, I can tell you this: if this really goes into the kernel, I'll look long and hard at another alternative to embedded Linux, starting tomorrow.