News for hackers. Stuff that matters.
Re:GCC drops its copyright-assignment requirement by notRobot on Tuesday June 01, 2021 @01:46PM (#27358446)
Can someone please tl;dr why this was a requirement in the first place?

I trust the FSF and it seems like a good idea, but I'm not sure why.

Re:GCC drops its copyright-assignment requirement by gumby on Tuesday June 01, 2021 @02:08PM (#27358787)
The concern is that in a legal dispute, the FSF might not have legal standing (I.e. in a lawsuit) if the code had been contributed by and remained the property of someone else. And what if that person could not be found.

I wrote the blanket assignment back in 1989 or 1990. I still consider this a legitimate risk.

Re:GCC drops its copyright-assignment requirement by tobias3 on Tuesday June 01, 2021 @05:39PM (#27361705)
One could see this in action in the Hellwig vs VMWare court case (one of the few GPL court cases).

As a first step they had to show that Hellwig had made changes to code that VMWare uses that can be copyrighted. Especially difficult since every line in Linux is modified by a few different people. They failed at that and that ended the lawsuit if I remember correctly.

Re:GCC drops its copyright-assignment requirement by ajross on Tuesday June 01, 2021 @01:59PM (#27358653)
The copyright assignment policy dates from very early in GPL history, when it was broadly thought that in order to make copyleft work the FSF was going to have to sue entities that had misappropriated GPL software. Doing that is harder when you start from a position of having to prove that you even own the relevant copyright in the first place.

It turns out this didn't happen. People who want to use free software and to stay legal just... honored the license. People who didn't want to do so used other software. And the only people who actually were breaking the license turned out to be tiny fly-by-night operations in parts of the world where copyright litigation wasn't a very useful tool anyway.

Re:GCC drops its copyright-assignment requirement by nullc on Tuesday June 01, 2021 @02:17PM (#27358949)
The GPL has been regularly violated by huge multi-billion dollar corporations.

But the FSF has had a policy of favoring compliance and forgiveness over punishment, and they've aggressively avoided litigation-- instead preferring to negotiate for compliance with goodwill loss as the threat-- to the point of contributing to some splits in free software. E.g. this is a point of disagreement between SFC and FSF.

Re:GCC drops its copyright-assignment requirement by warp on Tuesday June 01, 2021 @02:33PM (#27359199)
There doesn't seem to be any disagreement on GPL enforcement between SFC and FSF:

- https://sfconservancy.org/copyleft-compliance/principles.htm...

- https://www.fsf.org/licensing/enforcement-principles

Re:GCC drops its copyright-assignment requirement by nullc on Tuesday June 01, 2021 @02:50PM (#27359466)
They don't disagree on the general ideology, at least on paper, but when it comes to enforcement actions SFC is much more prone to litigate than the FSF.

IIRC FSF has only sued once, conservancy has like 100x the rate of lawsuits per years-existed. :P

Re:GCC drops its copyright-assignment requirement by bluGill on Tuesday June 01, 2021 @02:49PM (#27359442)
Compliance and forgiveness is probably the best policy anyway: courts will consider intentions. Companies that are sued for a violation and can tell the judge "look at the evidence: we have a history of doing the right thing and messed up in this one case" get a tiny slap and continue one. By having a reputation of make right and we won't sue companies have nothing to lose by making right. By having a history of suing they will win some cases, but many of them the total gained will be less than laywer fees because of all the companies that are able to say "we tried to make right but they rejected this reasonable offer".

Also, what if a company decides to ignore the GPL and pay up? The courts and law understand monetary damages. The FSF runs the risk of the judge deciding that the code is already out there, and the value the code is $X: therefore the infringer should reasonably buy a license for $X and doesn't have to give source code back. This would violate the spirit of the GPL, but would fit the letter of the law.

Re:GCC drops its copyright-assignment requirement by nullc on Tuesday June 01, 2021 @02:53PM (#27359513)
I'm not disagreeing with the FSF's approach.

But the fact that the FSF hasn't in fact litigated isn't by itself a reason that preserving the clear ability to litigate isn't useful. We don't know how many cases an infringer was made more likely to come into compliance by the fact that work had clear ownership, particularly because the FSF has avoided litigation when they had other avenues open.

Re:GCC drops its copyright-assignment requirement by bluGill on Tuesday June 01, 2021 @03:23PM (#27359933)
The FSF only needs some of the code to litigate though. It is better own everything, but that only is a factor after the case is done, if the FSF owns 20% of the code, then they get 20% of the total possible value of the infringement. (likely after lawyer fees, and class-action can apply)

It is easier if they own all the code, but it isn't required.

Re:GCC drops its copyright-assignment requirement by wahern on Tuesday June 01, 2021 @06:00PM (#27361989)
> It is easier if they own all the code, but it isn't required.

It's not that simple. For joint works any single author can grant rights to the entire work. Do you know what the caselaw is regarding joint works as applied to typical FOSS projects? I don't. But I suspect it's less than crystal clear (see, e.g., https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2999185) and therefore creates significant risk when attempting enforcement. Litigation is costly, and a wrinkle like this could potentially be exploited by an infringer to drag out an enforcement case for many years.

Re:GCC drops its copyright-assignment requirement by lmm on Tuesday June 01, 2021 @11:06PM (#27364338)
> For joint works any single author can grant rights to the entire work.

That doesn't sound right - why would anyone ever contribute to a joint work if the other author can sell it to a publisher on their own? Wouldn't they then just take all the royalties and leave you with nothing?

Re:GCC drops its copyright-assignment requirement by wahern on Tuesday June 01, 2021 @11:34PM (#27364535)
1) A joint author can individually only grant non-exclusive a license. See https://copyright.universityofcalifornia.edu/ownership/joint... Traditionally, publishers usually require exclusive licensing.

2) A joint author still must share profits. See previous link and also https://repository.law.umich.edu/cgi/viewcontent.cgi?article... (But can a joint author claim waste--loss of commercial profit--if another licenses as FOSS?)

3) Joint authors in the business of commercializing their works (as opposed to accidental joint authors) would usually have a contract between themselves controlling licensing.

4) If it doesn't sound right, that's exactly my point. Copyright law is more complex than discussed on internet forums. Just because something doesn't make sense to a non-practitioner doesn't mean it's completely senseless to a practitioner. Corporate lawyers are often much more risk averse than law professors counseling FOSS advocacy organizations. But as mentioned by the previous SSRN paper I linked[1], joint authorship is typically considered to be detrimental and even fatal to the objectives of many FOSS licenses, which usually assume a model of distinct, serial contributions creating derivative (not joint) works.

[1] Chestek, Pamela S, A Theory of Joint Authorship for Free and Open Source Software Projects (July 2, 2017). Available at SSRN: https://ssrn.com/abstract=2999185 or http://dx.doi.org/10.2139/ssrn.2999185

Re:GCC drops its copyright-assignment requirement by lmm on Wednesday June 02, 2021 @04:01AM (#27366137)
Hmm, fair enough, I'm more used to England and Wales. Sounds like US copyright law is pretty crazy.
Re:GCC drops its copyright-assignment requirement by wahern on Wednesday June 02, 2021 @12:16PM (#27370478)
The UK also has joint authorship. The test is slightly different, though I'm not sure how much of a difference it would make in the context of FOSS. See Elena Cooper, Joint Authorship in Comparative Perspective: Levy v. Rutley and Divergence Between the UK and USA, http://eprints.gla.ac.uk/100708/7/100708.pdf

To be clear, I'm not trying to say that all or even any FOSS works are or should be considered joint works. Just that the possibility could be a concern for lawyers contemplating enforcement. Certainly in other areas where you often see informal cooperation in the creation of collaborative works, such as in music, the issue absolutely does come back to bite people.

Re:GCC drops its copyright-assignment requirement by ajross on Tuesday June 01, 2021 @06:01PM (#27361997)
What about arguments like "the code the FSF owns isn't what I infringed?", or "the FSF doesn't own the code because this other developer wrote this bit and isn't a party to the suit; let me call them as a witness"?

I mean, it's true that you could still win a case having to win those arguments, but it's harder.

Defense in depth isn't just a computer security technique. Any lawyer will tell you that the easiest argument to win is one that can't be brought in the first place.

Re:GCC drops its copyright-assignment requirement by steerablesafe on Wednesday June 02, 2021 @05:40AM (#27366732)
> Also, what if a company decides to ignore the GPL and pay up? The courts and law understand monetary damages. The FSF runs the risk of the judge deciding that the code is already out there, and the value the code is $X: therefore the infringer should reasonably buy a license for $X and doesn't have to give source code back. This would violate the spirit of the GPL, but would fit the letter of the law.

IANAL but I don't think it works that way. They could pay up $X for damages and even get their license revoked. They still don't buy a license to perpetually violate the GPL.

Re:GCC drops its copyright-assignment requirement by bluGill on Wednesday June 02, 2021 @08:30AM (#27367900)
Perhaps. I wouldn't be surprised if their license was revoked. I wouldn't be surprised if the courts assigned a value to the license and told the FSF to accept that.

There isn't much case law here. It will probably end up in appeals for years. In the end it doesn't really benefit anyone to take their chances in court.

Re:GCC drops its copyright-assignment requirement by steerablesafe on Wednesday June 02, 2021 @10:13AM (#27368913)
> I wouldn't be surprised if the courts assigned a value to the license and told the FSF to accept that.

You basically say that the court can order the copyright holders to offer an exclusive proprietary license to the violators for a "reasonable" price.

Re:GCC drops its copyright-assignment requirement by bluGill on Wednesday June 02, 2021 @11:38AM (#27369942)
right. courts have a lot of power which they will use if they decide someone is being unreasonable. I'm not making a prediction of what the courts will do. There isn't enough case law to be sure (and nobody really wants to chance it so there probably never will be)
Re:GCC drops its copyright-assignment requirement by steerablesafe on Thursday June 03, 2021 @03:45AM (#27378301)
> There isn't enough case law to be sure

Isn't there any case law for license violations in general? For the purpose of this discussion this is just yet an other license violation, not necessarily tied to the GPL.

Re:GCC drops its copyright-assignment requirement by michaelmrose on Wednesday June 02, 2021 @03:29AM (#27365939)
Paying damages for copyright infringement doesn't entitle one to infringe in the same way paying restitution or damages for auto theft doesn't entitle one to keep the car.

Also statutory damages for willful infringement are very high As much as 150k per work which again still doesn't grant you a license.

Re:GCC drops its copyright-assignment requirement by bombcar on Tuesday June 01, 2021 @02:47PM (#27359422)
Haven't a significant number of the multi-billion dollar violations turned out to be rather "oh we didn't know" as opposed to intentional malice?
Re:GCC drops its copyright-assignment requirement by sigjuice on Tuesday June 01, 2021 @01:47PM (#27358458)
Re:GCC drops its copyright-assignment requirement by jhonsrid on Tuesday June 01, 2021 @01:50PM (#27358509)
This seems to imply copyright assignment to the FSF is _required_ for GNU projects? I wonder what RMS's take is on a unilateral change like this in one of the highest profile GNU projects?
Re:GCC drops its copyright-assignment requirement by dec0dedab0de on Tuesday June 01, 2021 @01:58PM (#27358647)
This may not be the FSF stance, but I can think of 3 reasons.

The main one is it is easier to change the license if you want to.

IANAL so I'll probably mess up these next two, which I believe are more theoretical than anything in regard to software. At least in the US.

It is possible for a copyright holder to revoke a license. so there could be an important contribution living in the code base and then one day the author could revoke the license and they would have to remove the code, and somehow prove their replacement was not infringing.

There is a concept of Joint Authorship. Where it could be possible for a contributor to claim joint copyright over the entire project, and then release it on their own under any license they want. I could be wrong, but I believe it has happened where book editors that didn't have an air tight contract, were able to claim joint copyright over a book they edited, and release their own versions with a different publisher.

Re:GCC drops its copyright-assignment requirement by mcguire on Tuesday June 01, 2021 @02:24PM (#27359067)
"It is possible for a copyright holder to revoke a license. so there could be an important contribution living in the code base and then one day the author could revoke the license and they would have to remove the code, and somehow prove their replacement was not infringing."

I don't believe that is true. A license is a contract; there are only two ways to revoke a contract: if the contract is illegal or if the contract provides terms to let you revoke it. If you publish software under the GPL, there is no way to revoke the license.

You could stop supporting the contribution in the code base and release alternate versions under different licenses, but the versions in the code base under the GPL remain available under the original terms.

Re:GCC drops its copyright-assignment requirement by Macha on Tuesday June 01, 2021 @02:29PM (#27359133)
The third is specific laws having specific terms that supercede the contract, e.g. discharging contractually obligated repayments in personal bankruptcy
Re:GCC drops its copyright-assignment requirement by tzs on Tuesday June 01, 2021 @05:09PM (#27361330)
Nonexclusive licenses granted without consideration are by default revocable. Fortunately, GPLv3 section 2 explicitly states that it is irrevocable.

GPLv2 omits such language, so to make it irrevocable you are going to want to find consideration. You can probably whack it a few times with your promissory estoppel stick hard enough to get some consideration to bleed out of it.

That's probably good enough to get irrevocability.