Well, the FSF has a lot of arguments against copyright and IP, and in a lot of cases they make a good point.
That does not change the fact that the GPL in itself is based upon copyright law, and the author, for as long as he or she retains copyright, can license it anyway they want to anyone.
Cygnus does this with their Cygwin32 environment, for example. You write GPL code, fine, you can user it. You want proprietary code, you buy the the rights from Cygnus. Another example is Be, who bought out some GPL drivers from the copyright holders.
Actually, I think most developers *do* realize what they do when they place their code under GPL. I most certainly do.
I think most free software developers are not anti-commercial. However, I do not want my GPL work to be a free lunch for whoever wants to proprietarize it. I write GPL code to be free at release and remain free in the future. Not to be 'embraced and extended'. The GPL is a guarantee for the codes perpetuated freedom.
The GPL also prevents something else we've witnessed several times in the UNIX world. The old not-so-enjoyable proprietary competetive forking, closely related to embrace-and-extend. With X and BSD we ended up with multiple proprietary versions supporting their own features with the end result that only the bare minimum was possible to use with any crossplatform certainty.
I have nothing against the BSD people. I just believe that perpetuated 'code freedom' is more important than 'developer freedom' for code I donate. If they changed the BSD license to prevent proprietarization of free code I'd be pleased to have them use my code... but of course, that wont happen. So the one-wayness remains.
Of course, with your example about Mesa, it's easy to solve. Simply integrate XFree86 into Mesa rather than the other way around:). Seriously tho, I think you could probably co-distribute them under an LGPL license with XFree86 retaining it's licensing, but the Mesa extensions, loaded through an object interface, would remain LGPL. Of course, that would mean that any commercial implementation would have to be LGPL/XFree licensed or they'd have to do their own implementation for GL.
Heh. The Magic Eight Ball does a better job of understanding the buiness and financial world than Wall Street. But for real market knowledge you should seek out the professional pyramid game creators. Because that is what the stock market is about today.
But hey, you can do it yourself. Take a piece of paper. Sell it to a friend for a buck. You've made a buck. He can sell it to another friend for two bucks. Now you both have made money off it. Go on repeating until you reach the last poor sucker who suddenly realizes that nobody will pay $2000 for a worthless piece of paper...
Even QPL Qt will not qualify as core component in the GPL sense. It doesnt matter if you distribute it with the OS, it is still not an essential system component of the kind referred to. You cant take out the kernel, libc, etc (which are the kind of components that qualify as essential) and expect the OS to work. You can remove Qt without any serious effect. Unauthorized linking of GPL code even to QPL Qt will remain a violation of the GPL.
Except, the good news isnt. Shipping Qt in a distribution does not make it a 'core os exception' component. Core OS components are essential system components like kernel and libc, without which you cannot within reason make a useful program. It's pretty obvious that a Linux system and most programs will work very well without Qt, and it's equally obvious that getting around the GPL by claiming core os component is just license evasion.
I *would* take appropriate measures in case any of my GPL code was misused by linking to even QPL Qt without prior permission.
However, if *asked* permission in advance, I would probably grant it. The issue here isn't Qt or Troll Tech, because QPL Qt qualifies as free. The issue is that we cannot allow the GPL to be legally undermined, because other, not so nice, companies will do the same thing in the future, and if they have a large set of precedents showing it is ok to circumvent the GPL in this way, they will be free to proprietarize GPL code at will by building proprietary dependencies on 'core components'.
Note that the core component issue has nothing to do with wether the software is free or not. Only the practice of the current largest distributions make it seem an easier problem, but if that changes in the future, allowing arbitrary components to qualify as core os would open the way to any form of proprietary library dependence.
If you need an example, just look at the unfortunate Motif (which could arguably be more core component than Qt, on merit of UNIX standards and system vendor control) problems. There is GPL software using Motif. There are serious limitations to the freeness of that GPL code under linux, since you have to either buy Motif, hope it works with lesstif, or obtain a statically linked copy. Neither alternative is very good, and the software isnt as free as most GPL applications in reality. That's where we'll be going if we allow arbitrary libraries to fall under the core component exception.
Slashdot Mirror
Well, the FSF has a lot of arguments against copyright and IP, and in a lot of cases they make a good point.
That does not change the fact that the GPL in itself is based upon copyright law, and the author, for as long as he or she retains copyright, can license it anyway they want to anyone.
Cygnus does this with their Cygwin32 environment, for example. You write GPL code, fine, you can user it. You want proprietary code, you buy the the rights from Cygnus. Another example is Be, who bought out some GPL drivers from the copyright holders.
Actually, I think most developers *do* realize what they do when they place their code under GPL. I most certainly do.
:). Seriously tho, I think you could probably co-distribute them under an LGPL license with XFree86 retaining it's licensing, but the Mesa extensions, loaded through an object interface, would remain LGPL. Of course, that would mean that any commercial implementation would have to be LGPL/XFree licensed or they'd have to do their own implementation for GL.
I think most free software developers are not anti-commercial. However, I do not want my GPL work to be a free lunch for whoever wants to proprietarize it. I write GPL code to be free at release and remain free in the future. Not to be 'embraced and extended'. The GPL is a guarantee for the codes perpetuated freedom.
The GPL also prevents something else we've witnessed several times in the UNIX world. The old not-so-enjoyable proprietary competetive forking, closely related to embrace-and-extend. With X and BSD we ended up with multiple proprietary versions supporting their own features with the end result that only the bare minimum was possible to use with any crossplatform certainty.
I have nothing against the BSD people. I just believe that perpetuated 'code freedom' is more important than 'developer freedom' for code I donate. If they changed the BSD license to prevent proprietarization of free code I'd be pleased to have them use my code... but of course, that wont happen. So the one-wayness remains.
Of course, with your example about Mesa, it's easy to solve. Simply integrate XFree86 into Mesa rather than the other way around
Heh. The Magic Eight Ball does a better job of understanding the buiness and financial world than Wall Street. But for real market knowledge you should seek out the professional pyramid game creators. Because that is what the stock market is about today.
But hey, you can do it yourself. Take a piece of paper. Sell it to a friend for a buck. You've made a buck. He can sell it to another friend for two bucks. Now you both have made money off it. Go on repeating until you reach the last poor sucker who suddenly realizes that nobody will pay $2000 for a worthless piece of paper...
Even QPL Qt will not qualify as core component in the GPL sense. It doesnt matter if you distribute it with the OS, it is still not an essential system component of the kind referred to. You cant take out the kernel, libc, etc (which are the kind of components that qualify as essential) and expect the OS to work. You can remove Qt without any serious effect. Unauthorized linking of GPL code even to QPL Qt will remain a violation of the GPL.
Except, the good news isnt. Shipping Qt in a distribution does not make it a 'core os exception' component. Core OS components are essential system components like kernel and libc, without which you cannot within reason make a useful program. It's pretty obvious that a Linux system and most programs will work very well without Qt, and it's equally obvious that getting around the GPL by claiming core os component is just license evasion.
I *would* take appropriate measures in case any of my GPL code was misused by linking to even QPL Qt without prior permission.
However, if *asked* permission in advance, I would probably grant it. The issue here isn't Qt or Troll Tech, because QPL Qt qualifies as free. The issue is that we cannot allow the GPL to be legally undermined, because other, not so nice, companies will do the same thing in the future, and if they have a large set of precedents showing it is ok to circumvent the GPL in this way, they will be free to proprietarize GPL code at will by building proprietary dependencies on 'core components'.
Note that the core component issue has nothing to do with wether the software is free or not. Only the practice of the current largest distributions make it seem an easier problem, but if that changes in the future, allowing arbitrary components to qualify as core os would open the way to any form of proprietary library dependence.
If you need an example, just look at the unfortunate Motif (which could arguably be more core component than Qt, on merit of UNIX standards and system vendor control) problems. There is GPL software using Motif. There are serious limitations to the freeness of that GPL code under linux, since you have to either buy Motif, hope it works with lesstif, or obtain a statically linked copy. Neither alternative is very good, and the software isnt as free as most GPL applications in reality. That's where we'll be going if we allow arbitrary libraries to fall under the core component exception.