Apple Originally Tried To Give GPL'ed LLVM To GCC

[dee.]

Michael, please stop deliberately stirring up shit, while fansplurging constant "PRO LLVM ANTI FSF" flamebait.

Yeah Michael... why all the FSF hate? Without FSF, your site wouldn't be here bringing us news of FOSS developments, because there wouldn't BE any FOSS.

[Del_]

There seems to be come confusion here.

This is not the same thing at all, for at least a couple of different reasons.

I don't think so. The + is enough for FSF to relicense a project under a new license, namely any future version of GPL. As explicitly mentioned it removes the ability of FSF to enforce the license, so yes not requiring copyright transfer does come with a cost. Another aspect that is well worth putting on the table is that not requiring copyright transfer also reduces FSF's influence on GNU projects.

So when GPL3 addressed Tivoization, the FSF can now re-license all files as GPL3+ only, preventing people using such shady practices from using new versions of FSF software.

Anybody can do that to a code with the +. Of course, the original code that was GPL2+ will remain available under GPL2 forever too. If Linus was so inclined he could also make a GPL2 only branch of GCC based on the latest version of GCC distributed under GPL2+, he can also create a GPL3 only version of GCC based on todays GCC.

Understand that without copyright ownership, the FSF would not be able to stop people from releasing code under older versions of the GPL

As long as they distribute the project, and act as project managers, they certainly can. The + gives FSF the exclusive right of making a new license for any project having the +. Yes, I agree though, FSF's influence and practical ability to remove support for older version of GPL would probably diminish. Hence I am not at all sure what "the right thing" would be. All things considered I lean towards making copyright transfer to FSF voluntary, not mandatory. One of the key observations leading me to his conclusion is the success of projects with distributed copyright. Linux, Samba and Debian comes to mind among many others.

[mrugiero]

All things considered I lean towards making copyright transfer to FSF voluntary, not mandatory. One of the key observations leading me to his conclusion is the success of projects with distributed copyright. Linux, Samba and Debian comes to mind among many others.

I'd agree with you if it weren't because it sounds like it makes the assignment completely pointless. If you assign only voluntarily, there is still the same chance you can't enforce the license (because you don't own the whole copyright; BTW, I never understood why is it that you can't enforce the license if you don't own the whole copyright, or if it's actually that way, could someone explain this to me?) than if you don't assign at all. Also, this doesn't take into consideration the possibility that it is actually one of the copyright holders the one infringing. Although I think they are in the same position as not being copyright holders (as part of the work they are using is not theirs), I don't know if a court will treat it the same. In case you need all of the copyright holders to agree on enforcing the license, any contributor can, with optional assignment, make a loophole, just by not suing themselves.

EDIT: Note I'm not specifically in favor of copyright assignments (I mean, I'd probably assign mine, but I'm not completely in favor of making it mandatory), but I think it's almost pointless if it's optional. I'm OK with FSF projects either not asking for it, asking for it but making it optional or making them mandatory, but the case where it is optional just seems pointless to me, and I wouldn't bother if I were them.

[devius]

Of course it didn't. Apple wanted to pull the old Microsoft EEE trick with GCC but FSF anticipated that.

No, that's not possible! Haven't you been reading Phoronix articles? If you had you would know by now that Apple is the best thing in the world! They are so in love with Free and open-source software that it almost makes you puke! Not like that evil FSF that hates freedom and is trying to kill your cat

[Marc Driftmeyer]

Not at all.
Let's be honest, GCC has only the speed on his part. And compatibilty with the various GCC extensions to C/C++.

Beside that, LLVM is way, way better. It's modular architetture allow anyone to quickly create a new language on top of it, or adapt the various part (like the JIT) for different projects. It also allow for great tools like LLDB, and stuff like SafeCode (compile-time check for buffer overflows and similar errors).
It's faster, way faster at compiling.
It's generally friendlier at developers. The output is nicer.

I like GCC, but let's not be delusional.
In my regard, the rule of thumb is: compile/program with LLVM, release with GCC.

In 12 months time you won't even touch GCC, with the state LLVM/Clang will have advanced.

[prodigy_]

Eheh. Now you're just clutching at straws. Apple wished to provide a legitimate contribution. FSF rejected it -- they had their reasons -- but they still rejected it. Drop the blind Apple-hatred and look at the facts.

Oh, so Apple wanted to contribute to a GPL open source project without having an ulterior motive? Yeah, right. And Steve Jobs real name was Santa Claus.

[xeekei]

One of the key observations leading me to his conclusion is the success of projects with distributed copyright. Linux, Samba and Debian comes to mind among many others.

I don't think Linux or Debian have any copyright reassignments, since both Linus Torvalds and the Debian devs seem to be very much against that.

[Pajn]

I don't think Linux or Debian have any copyright reassignments, since both Linus Torvalds and the Debian devs seem to be very much against that.

Eee, yeah. That was what he said.

[Del_]

Eheh. Now you're just clutching at straws. Apple wished to provide a legitimate contribution. FSF rejected it

Actually, that is not clear to me at all. From what I have seen, Apple never succeeded in neither copyright transfer nor re-licensing of LLVM, and hence never got to the point where FSF could accept or reject (if anyone have other information, then please come forward). There was one Apple developer that tried to make it happen, but it seemed he was not able to pull it through. To the contrary all correspondence I have seen indicates that GCC developers were positive towards integrating LLVM (again, please step forward if you have information indicating otherwise).

I don't think Linux or Debian have any copyright reassignments, since both Linus Torvalds and the Debian devs seem to be very much against that.

That is what I mean by distributed copyright, all contributors retain the copyright to their contribution and provides access to their contribution through the license. As opposed to one entity having copyright to everything. Samba is also like that, and OpenWrt and KDE and many other successful open projects.

(because you don't own the whole copyright; BTW, I never understood why is it that you can't enforce the license if you don't own the whole copyright, or if it's actually that way, could someone explain this to me?)

Sure. I am no lawyer, but I do have colleagues that are which help me out on these issues. My understanding is that you need to have copyright in order to enforce the license, and this is invested in copyright law. In short, if you don't have copyright to the code, then no harm has been done to your property. Moreover, I believe it suffices to have copyright only to parts of the code, since those parts are also covered by the law, and is infringed. Hence, enforcing the license should be possible for anybody holding copyright to any of the code. It is important to note that GPL heavily relies on, and is ingeniously taking advantage of, copyright law.

[xeekei]

Eee, yeah. That was what he said.

Okay, I misunderstood the term "distributed copyright". My apologies.