GCC To No Longer Require Copyright Assignment To The Free Software Foundation

[stikonas]

IMHO, GPL only makes sense for developers who want to make profit from their code: Lots of GPL code in the scientific field falls in this category, they offer a GPL version for "free", and a non-GPL version when you pay for it. On the other hand, when you are interested in just giving your code for free, for the entire World, then MIT or similar is the way to go.

That has nothing to do with whether you want to make a profit from your code or not. I can release my code under GPLv3 without ever intending to make any profit from it. But I also want to protect other users from somebody rogue taking my code, making a small modification and keeping it proprietary (it's not even about selling, since GPL allows selling).

[torsionbar28]

It's worse than that. If a company develops some thing that would be useful in GCC then assigns the rights to FSF. But then the company wants to use that tech in another project they have the potential risk of FSF coming after them. Even if we assume FSF would never do that the lawyers are going to have to get involved and that drives up the price of contributing.

Precisely. Anytime the legal department has to get involved, the project cost goes through the roof. Requiring legal review and approval is an instant death to small internal projects at most companies.

[Alexmitter]

The copyright assignment to FSF was likely something that did protect the project from some large contributor stepping back asking for his code to be removed. A famous project that this did happen to is Bukkit, never recovering from that.
GPLv3 fixed that, so it probably can make sense to remove that.
What worries me is that it probably happens in a step to separate from the FSF in a act of hate against Stallman.

[strtj]

NetBSD what a nightmare to contribute to. I hate mailing lists, and I don't know anyone under the age of 30 who use mailing lists.
NetBSD use CVS, some kind of old, version control system. I have no idea how to use it, and I don't know anyone who knows how to use it.

Saying "I don't like mailing lists" is a ridiculous reason not to contribute to something. Most of the time you don't even have to sign up for the list itself, you just send an email to an list email address and say, "Hi, here's my patch, I'm not subscribed to the list so please copy me on replies." There are plenty of projects I've sent patches to where I wouldn't have commit or pull request access anyway, so someone else has to review/commit the patch, and a mailing list a perfectly decent publicly visible option.

NetBSD is a good example of a place where I've sent several patches but I've never had to learn one bit of CVS, because I don't have commit access and don't need/want it. I can actually only think of one project, GMP, where I sent a combination patch/patch outline and was told to jump through a bunch of ridiculous hoops before they would even consider it. Saying "I don't want to jump through the sort of hoops GCC used to have" I think is a perfectly reasonable argument, but saying "I don't want to send a quick email" is just beyond lazy as far as I'm concerned.

[RahulSundaram]

Sadly a step in the wrong direction, free software != open source. If free software fails in the future, big corps will close the code and commodify all projects, then open source will loose its importance and cease to exist.

The license doesn't change. So I am not sure what your point is here.

[RahulSundaram]

The copyright assignment to FSF was likely something that did protect the project from some large contributor stepping back asking for his code to be removed. A famous project that this did happen to is Bukkit, never recovering from that.
GPLv3 fixed that, so it probably can make sense to remove that.
What worries me is that it probably happens in a step to separate from the FSF in a act of hate against Stallman.

If a contributor leaves, they have zero rights to demand that their code be removed. Copyright assignment has no relevance to this.

[Cape]

It's worse than that. If a company develops some thing that would be useful in GCC then assigns the rights to FSF. But then the company wants to use that tech in another project they have the potential risk of FSF coming after them. Even if we assume FSF would never do that the lawyers are going to have to get involved and that drives up the price of contributing.

That's fine. We don't need better performing software.
We need better freedom respecting software.

I'm fine with loosing 0.1 second each time I run my application, if i know that all the software involved is GPL3 and respect the user (coincidentally, me).

[kpedersen]

NetBSD what a nightmare to contribute to. I hate mailing lists, and I don't know anyone under the age of 30 who use mailing lists.
NetBSD use CVS, some kind of old, version control system. I have no idea how to use it, and I don't know anyone who knows how to use it.

To be fair, if someone is unable to learn how to use CVS, or any other kind of version control, it means they are unlikely able to learn the internals of an operating system project and again, make a meaningful commit.

Many of the old guys can adapt (and have done to Git for many projects). At the very least they probably expect the younger generation to be able to adapt too. And once people hit the industry and are paid to learn to use crusty version control systems... they will

Mailing lists and CVS are probably a little before my time too but... it ain't exactly rocket science! Perhaps you need to change your workflow a little. Try some different mail clients like Mutt, Alpine and see if you can better handle filters than web mail.

[RahulSundaram]

That's fine. We don't need better performing software.
We need better freedom respecting software.

I'm fine with loosing 0.1 second each time I run my application, if i know that all the software involved is GPL3 and respect the user (coincidentally, me).

Copyright assignment is not a requirement to have free software. Since the software stays under the same license as before, there is no loss of freedom. Also OP is not talking about runtime performance but the effort involved in contributions which isn't related at all. Another thing to keep in mind is there there is no direct tradeoff between runtime performance and freedoms. You can and should expect both things at the same time.

[wizard69]

Apple doesn't want to contribute code back to the wider world, their LLVM is not mainline LLVM :^)

Obviously you don’t know!