GCC RISC-V Support Allegedly Held Up Due To University Lawyers

[phoronix]

Phoronix: GCC RISC-V Support Allegedly Held Up Due To University Lawyers

While there has been talk about RISC-V architecture support in the GCC compiler and for LLVM too going back months, a developer is reporting that the GCC RISC-V support is being delayed due to UC Berkeley lawyers...

http://www.phoronix.com/scan.php?page=news_item&px=GCC-RISCV-Lawyer-Hold

[bitman]

Or maybe it is being held up by requirements of FSF. I mean they are asking to give away copyright. It is not a small thing to ask and i do not think they should. Is it not licensing code as GPLvX or any later good enough for FSF?

[Inopia]

Or maybe it is being held up by requirements of FSF. I mean they are asking to give away copyright. It is not a small thing to ask and i do not think they should. Is it not licensing code as GPLvX or any later good enough for FSF?

Why did they even started working on GCC if copyright assingmet to FSF is a problem? It's not like it's a secret requirement only revealed when you try to upstream patches.

[microcode]

Or maybe it is being held up by requirements of FSF. I mean they are asking to give away copyright. It is not a small thing to ask and i do not think they should. Is it not licensing code as GPLvX or any later good enough for FSF?

They need copyright assignment to make licensing changes in the future. Though I would say that the world would probably be a better place if GCC were still GPLv2. Same deal with LLVM/Clang; which is also the reason why UCB hasn't yet upstreamed with them either.

[phoron]

Or maybe it is being held up by requirements of FSF. I mean they are asking to give away copyright. It is not a small thing to ask and i do not think they should. Is it not licensing code as GPLvX or any later good enough for FSF?

Yes, It's being held by both the FSF requirements and UC Berkeley lawyers. The FSF requirements were clearly known before the work started, so there's also some regret on starting a job that you could have foreseen to need legal clearance before upstreaming. And thanks to authors to having done it.

But the FSF requirements are not so much to ask. Nobody has any obligation to comply, of course, but if the FSF is going to take the maintaing of the code, both technically and legally, they need to have something to hold on to. In other projects copyrights is distributed and that makes somewhat more difficult to make GPL an effective license, that is, to sue. People license under GPL but it is not so clear that the GPL is as strong as possible because nobody knows what everyone will do if someone violates the GPL. And there's no way to change the license when laws or markets changes (for good or for bad).
Of course you'll get more contributions when you don't require assigning copyright than when you require it, but you may want to make sure you can relicense under GPL4 when it comes to that, or that you can sue infringers, or so.

I think (but haven't checked, I might be confused with the FSFE fiduciary agreement) that the FSF gives back rights to the author when the author gives the copyright to the FSF, so the author is not effectively conceding all rights. And I find it easier to trust on what the FSF will do in the future than trusting a company. But it is still an act of faith that UCB does not have an obligation to take. I just hope it does.

The good news is that at least the code is out there and anyone can patch GCC to incorporate it. Let's hope we get even better news and UCB considers in their interest to help the public by giving the copyright to FSF. I very much doubt it'll be the first University contributing to GCC.

[ssokolow]

Or maybe it is being held up by requirements of FSF. I mean they are asking to give away copyright. It is not a small thing to ask and i do not think they should. Is it not licensing code as GPLvX or any later good enough for FSF?

First, it's for enforcement. The copyright assignment ensures that they have legal standing to go after any GPL violations, since they'll own the code being infringed no matter which piece it is.

Second, they've been requiring it for decades. Palmer Dabbelt should have made sure the lawyers were OK with that before starting work on the code for GCC. Hell, the FSF even has a "What if my university won't let me GPL my degree work?" FAQ which also applies well to this.

[nils_]

Second, they've been requiring it for decades. Palmer Dabbelt should have made sure the lawyers were OK with that before starting work on the code for GCC. Hell, the FSF even has a "What if my university won't let me GPL my degree work?" FAQ which also applies well to this.

Why would the university own the copyright to his work?

[bitman]

Second, they've been requiring it for decades.

It is a really bad argument to justify something. As for enforcing - i am pretty sure with minimal effort FSF can act as representative to copyright owner. Not like people represent themselves in courts nowdays.

[pal666]

It is a really bad argument to justify something.

it is good argument to argue "you had to solve this before writing code"

As for enforcing - i am pretty sure with minimal effort FSF can

i am sure fsf will accept your legal advice if you can give one. if you can't, then you have to accept what their lawyers say. btw, it is not only for enforcement, it is also for relicensing like from v2 to v3. and it is not your usual canonical cla, they don't take away your rights to code and they have binding contract with you that the code will stay free.

[mmstick]

This is why MIT-licensed software like LLVM is so successful.