5 ms·
The minute it is required to hand over anything, even a single line of code, you will find someone raising the objection you mentioned (“it’s too burdensome to
by toyg 6y ago
The minute it is required to hand over anything, even a single line of code, you will find someone raising the objection you mentioned (“it’s too burdensome to figure out if I have to release 6 characters or 5 characters!”). Stock corporate lawyers don’t like uncertainty, and the smallest of grey areas will be seized upon to block anything “in order to protect the company”.
The reality is that in 99.99999% of the cases, nobody will ever come after you for trying to do the right thing. If you engage the GPL community in good faith, things will work out fine with minimum fuss.
So no, we don’t need another license, we just need people willing to do the right thing.
- _ph_ 6y agoThe problem with GPL is, that it doesn't only require you to hand over any modifications to the GPL code - this would be easily to do, but to any code in a product built containing GPL code. And that is where the problems usually start. The reality is, not everyone agrees what the reality is and especially a company which takes licenses serious will be very careful to check, whether using GPL code is possible in a project. And no, we don't need another license, there are plenty of licenses available which resolve the issues. Be it the permissive licenses (MIT, BSD, ...) or GPL with linking exceptions.
- arminiusreturns 6y ago> but to any code in a product built containing GPL code. Citation please. I don't think this is accurate, and is another common misunderstanding of the gpl license family. I frequently see proprietary systems that have a notice list of the licenses they are using, including gplv2/3. Are they all in violation? I doubt it. GPL is my license of choice, (I have an almost completely gpl daily stack), so I'm not just being rhetorical, if there is something I'v missed I want to learn!
- bartread 6y ago> Citation please. I don't think this is accurate, and is another common misunderstanding of the gpl license family. I believe you, but I also think (as stated in another comment) that if this is the case the GPL doesn't do a very good job of explaining itself. At the very least this creates friction when considering its adoption.
- arminiusreturns 6y agoI think there is some truth there, but I think a lot more of the FUD on the copyleft licenses originates not from unclear licenses, but from a network effect where so many people repeat falsehoods, and to be honest, I wouldn't be surprised if some entities actively encouraged them.
- toyg 6y ago> the GPL doesn't do a very good job of explaining itself The copious amount of side-notes [1][2][3] is likely to clarify most doubts, if one is willing to go through it. Obviously "contract law applied to software" is not a fun read though. [1] https://www.gnu.org/licenses/gpl-faq.html https://www.gnu.org/licenses/gpl-faq.html [2] https://www.gnu.org/licenses/gpl-howto.html https://www.gnu.org/licenses/gpl-howto.html [3] https://www.gnu.org/licenses/quick-guide-gplv3.html https://www.gnu.org/licenses/quick-guide-gplv3.html
- _ph_ 6y agoThat is the problem, it entirely depends on how you use the GPL code and there is no definite legal precedence. If you use GPL code as part of your executable, the whole code used for building this executable falls under GPL. See for example the question, whether the non-GPL ZFS code can be linked against the GPL Linux kernel and the result distributed. See also all the GPL-only symbols the kernel exports. Another example is libreadline, which is put under GPL to force all applications using it to be licensed under GPL too. So there are clearly cases where using GPL code in your product can require the whole product be distributed under GPL. Of course, there are also cases, where this can be avoided. If I ship a product and I also ship emacs along with it, that doesn't influence my product. But if the product executable itself is derived from the emacs sources, it would.
- topkai22 6y agoSurprisingly hard to find a simple explanation from FSF, but here is a "proof by contradiction" explaining why should use GPL instead of the LGPL- https://www.gnu.org/licenses/why-not-lgpl.html https://www.gnu.org/licenses/why-not-lgpl.html Basically, the LGPL exists because the GPL requires any binaries linking to GPLed code to inherit the GPL. My understanding is that properitary programs can utilize / incorporate GPL binaries via interprocess communication (pipes, TCP, shared files, etc), but if they cannot do intra process communication without potentially violating the GPL. So it depends on how your define "build" here :) Regardless, because my role involves creating and distributing potentially proprietary software, I generally handle GPL projects with caution. I'll happily use Linux, but I don't look at the sources. If I see a GPL license on github, I'll generally move on. If I see something is AGPL licensed, I'll move on very quickly.
- bartread 6y ago> Basically, the LGPL exists because the GPL requires any binaries linking to GPLed code to inherit the GPL. This was my understanding too, but other comments in this discussion seem to imply that this might not be the case. If you could statically or dynamically link to GPL'd binaries without having to license your own code in the same way, then in the real-life examples I've seen we'd have had no concerns about either using GPL'd projects, or contributing modifications to the GPL'd projects back.
- _ph_ 6y agoRight. Contributing changes to the GPL'd software itself back isn't the issue. It is the intended "viral" nature of it, that also enforces the GPL onto all code linked to it. See for reference the issues with GPL-symbols in the Linux kernel, ZFS, or libreadline,
- dragonwriter 6y ago> but to any code in a product built containing GPL code. No, to any code that is a part of a derivative work of GPL code; a product may often be an aggregate of code that is not a derivative work of GPL code and derivative-of-GPL code. What is legally a derivative work, OTOH, is not always simple and straightforward to determine, so to avoid conflict one might go the whole way out to the product boundary.
- _ph_ 6y agoWith "containing" I didn't mean just delivered together as in shipping emacs together with your product, but making the GPL code part of it, especially linking with GPL code or even compiling together with the GPL code.
- dragonwriter 6y agoLinking is an interesting case, because side notes from the FSF indicate that they think linking makes things a derivative works, but its far less clear if that is true under the law, especially for dynamic linking (API copyrightability, which is currently before the court, might have a lot to say about this).
- bartread 6y ago> The reality is that in 99.99999% of the cases, nobody will ever come after you for trying to do the right thing. If you engage the GPL community in good faith, things will work out fine with minimum fuss. > So no, we don’t need another license, we just need people willing to do the right thing. What I'd actually like is for the license to be clear enough in its intent that I don't need to engage anyone in a consultation with either a lawyer or the community to figure out whether I can use a project or not.
- lawtalkinghuman 6y ago> If you engage the GPL community in good faith, things will work out fine with minimum fuss. Who is the "GPL community"? Stallman and FSF/GNU/SFLC? Linus Torvalds? A non-profit foundation like Apache/Mozilla/GNOME? A community of enthusiasts with a web forum and a GitHub org? A random solo author library of a random JS package on GitHub? Google? Microsoft? Oracle? Because the sort of intuitive risk feelings I'd have for each differ quite considerably.
- toyg 6y agoFirst port of call should be authors, since they are the copyright holders and can clarify intent beyond the letter of the licence. Failing that, you can ask FSF/GNU. (Not sure why you'd bring Apache or Mozilla into this, considering they went to the trouble of writing their own licenses rather than using the GPL...)
- lawtalkinghuman 6y agoThe intent needs to be clear in the licence text itself. In a dispute, courts will look to intent somewhat, but generally the law assumes that when parties to a contract or licence or deed state their intentions in the form of written text, that text will be primary in understanding the content of the agreement that was made. The authors of legal texts don't have magic powers. Let's say you buy a product from a company, C. They wrote a contract which you both agreed to. They are a big company, and you are just an individual person, so your bargaining power is pretty minimal compared to them, and basically amounts to "take it or leave it". So you buy a thing on their terms. Some term of the contract later turns out to be unclear, so you ask them to clarify it. Is the clarification they provide going to be beneficial to their interests or yours? Theirs, obviously. Stallman's interpretation of the GPL in the case of uncertainty or lack of clarity is going to be to read it in such a way as serves his goals: making sure there is more free software in the world, and protecting the Four Freedoms. A corporate lawyer working for Oracle is going to interpret unclear provisions in a perhaps rather more conservative way, especially if it financially benefits Oracle. The fact Stallman wrote the license doesn't give him any particular special hermeneutical status when working out how a court will interpret it. "Hey, just trust us" isn't useful either, because there's a difference between being asked to trust randomhacker849 and being asked to trust a multinational megacorporation like Google or Oracle. While randomhacker849 may follow the broad spirit of any particular open source licence and expect licensees to follow the broad spirit too, I'm pretty sure Oracle are going to follow the exact letter if doing so is in their financial interest. The whole point of legal instruments is the need for trust is reduced because you have rights and obligations that a court will enforce. And the whole reason we turn agreements into written legal texts is precisely because "hey, just trust us" doesn't scale—when millions or billions are on the line, you want those agreements formalised. (I included Apache and Mozilla because they represent examples of open source non-profit foundations, not because they use the GPL, incidentally. Other than GNOME and FSF, most of the projects that have their own foundation—LibreOffice's Document Foundation, Perl, Python, Mozilla—have permissive licenses rather than copyleft. The point isn't about the wording of the text, but more the question of "if I were going to rely on a disputed bit of license text in my use of this software, what risk would I face of litigation?"... and that risk differs considerably between randomhacker849, a non-profit foundation, Stallman/FSF, and a huge scary litigious corporate beast like Oracle.)