1) I don't see how it is the GPL's fault that the Open CASCADA license is GPL incompatible. The Open CASCADA license was created way after the GPL, and so if they wanted to use their code with GPL code, they shouldn't have made a license that was GPL incompatible. If they thought it was compatible, but it turned out it wasn't, then maybe their legal team doesn't have the acumen to fully design their own license, and they should have used one of the existing known GPL-compatible licenses for their work.
2) The FSF has long anticipated potential incompatibilities between GPL versions, which is why its recommended to license your software under the GPL or any later version, and indeed such language is provided in the license by default. LibreCAD's original developers changed the license to make it GPLv2 only, intentionally breaking license compatibility with software licensed under future versions of the GPL.
If these people did these actions by accident without realizing the ramifications then the GPL could be faulted for being confusing (despite the very comprehensive FAQ page provided by the FSF), but I don't see how it's GPL's fault when people go out of their way to be GPL incompatible, or incompatible with future versions of GPL.
It's up to the authors to decide which license they will choose. Complaining that you can't use a GPL3 DWG library in your code is no different than complaining that you can't use Autocad's DWG library in your code: it's not yours; you do not get to decide the rules for using it.
The GPL prevents you from shipping compiled binaries or modified source.
A LibreDWG plugin to LibreCAD is likely to be easily distributable purely as source. FSF doesn't like to talk about it (it gets into quite dangerous territory for them when talking about interpreted languages), but the reality is that GPL relies on copyright. Specifically, it relies on their ability to make a copyright claim to things you are trying to share. If they have no actual copyright claim to something you're distributing, they are shit out of luck.
FSF and licensing@ really likes to pretend that they can prevent this. But they never give specifics and always talk in generalities. They are bluffing and have no actual legs to stand on. If they have no copyright claim to something, they have no actual power over it. At all. It's out of their GPL's grasp.
Of course, the FSF also wants it both ways when convenient. For example, they want to co-opt Matlab's MEX API for GNU/Octave, and simultaneously prevent you from doing the same to their code. This is allowed because of their magical "compatibility" wand. Apparently, they may copy interfaces as they wish, but for nebulous reasons you are expected to trust them that you may not do the same.
See also the existence of ZFS on Linux distributed as patches/source-code shims (not a FSF issue as FSF does not own the Linux kernel, but ultimately it's the same licensing issue).
tl;dr: GNU LibreDWG is licensed GPL3+, incompatible with projects that use other open source licenses, including "GPLv2 only".
I'm a little surprised that a library such as LibreDWG is not licensed LGPL, but come to think of it, this isn't unexpected from the hardline FSF (Stallman). They want to create a coherent free ecosystem, and are unwilling license their work with licenses that have known loopholes.
That's a damn shame for these almost-but-not-quite-as-libre software projects such as LibreCAD, FreeCAD, Open Asset Import Library and Bender.
What's wrong with the OpenDWG [1] libraries, licensed GPLv2?
I don't think it's fair to blame GPLv3 alone for this problem. Really the problem is with all so-called free software licenses that impose conditions (or restrictions) that might be inconvenient, difficult, or even (legally) impossible to meet.
The problem here is that the GPLv3 has more requirements of the licensee than the GPLv2 does, but the GPLv2 says "you may not impose any further restrictions on the recipients' exercise of the rights granted herein".
Some extra requirements of licensees of GPLv3 programs:
- Preventing tivoization: they must provide information on how to install and execute modified versions of anything they build with the code, when they provide the corresponding source;
- Preventing DRM: anything they build with the code won't be legally considered a DRM system, and they "waive any legal power to forbid circumvention" of that system;
- Protecting against patents: they must provide every recipient of the program (in source or binary) with any patent licenses necessary to exercise the rights that the GPL gives them, and if they try to use a patent suit to stop another user from exercising those rights, their license will be terminated.
I would place the blame of this situation at the maintainers of the GPLv2-only libraries: why don't they want to use a license that prevents tivoization, DRM, and patent abuse? If they've changed their minds on copyleft, why not switch to the BSD license?
I'm not sure exactly what the LibreDWG developers asked of Stallman, but if it was to change the GPLv2 or GPLv3, that's stupid. Even if he did, it wouldn't help their situation.
On the trollish title: the GPLv3 stifling "open-source" development? Really? (Edit: the title has since been changed) Let's talk about how all the non-copyleft licenses let companies run off with the code to build their own systems behind closed doors, and only contribute what they want, when they want -- if at all? How does that support free software development?
I'm licensing my projects under the AGPLv3, because I believe that nonfree software is harmful, and I don't want to contribute to its development at all. I believe that a free society must necessarily operate on free software. I want to encourage the development of free software, and discourage the development of nonfree software.
The GPL was designed to ensure that the software stays free software; "to ensure that every user has freedom". The GPL's protections may have sufficed in 1990, but they don't in 2014. The GPL doesn't consider users of a web service to be users of the software implementing that server. Thankfully, the AGPL does.
The AGPL ensures that if the code is used to implement a web service, then the entire source code of that web service must be free software. This way, I'm not contributing to nonfree software, whether it's executed locally or provided over a network.
How does the 'all or nothing' approach work in this situation? LibreDWG stays GPLv3, but no projects can use it. Eventually the development team gives up and leaves because their work isn't being used. But it's a win for Free Software because the license remained unchanged! A victory! Round of drinks for everyone! We made sure that our work remained pure, despite everyone else and their 'practical' concerns!
> Let's talk about how all the non-copyleft licenses let companies run off with the code to build their own systems behind closed doors, and only contribute what they want, when they want -- if at all? How does that support free software development?
It supports free software development because contributing to free software works in practice, so less complex requirements means lower legal barriers means more people participate.
> The GPL's protections may have sufficed in 1990, but they don't in 2014.
The GPL's restrictions may have been necessary to motivate people to contribute back in 1990, but they aren't in 2014. The people making the practical case for free software succeeded, and active communities around permissively licensed Free software abound. Sure, proprietary forks and proprietary value-added enhanced versions also abound, but the maintainers of those (at least the latter) often also contribute (both code and money) back to the Free upstream project -- or, actually, are simultaneously the maintainer of the Free upstream project.
This is an ideological argument rather than a practical one, but the usual counterpoint is that even pure leeches of OS development contribute to its success by displacing competitors, training users and developers, and there is always a chance they'll change their position on code contributions when it's in their interests. I personally prefer 3 clause BSD because it gets used without some developer, entrepreneur or corporation engaging lawyers to sort out these issues. Freedom from legal BS is the best freedom I can gift to users.
IMO the AGPL cuts off your nose to spite your face. I understand the rationale, but I find it excessively onerous and complex for the users I really care about (other devs).
The most free license, as in completely unlimited freedom as opposed to this kind of "freedom forced down your throat" approach, is the one used by SQLite.
[+] [-] tbirdz|11 years ago|reply
1) I don't see how it is the GPL's fault that the Open CASCADA license is GPL incompatible. The Open CASCADA license was created way after the GPL, and so if they wanted to use their code with GPL code, they shouldn't have made a license that was GPL incompatible. If they thought it was compatible, but it turned out it wasn't, then maybe their legal team doesn't have the acumen to fully design their own license, and they should have used one of the existing known GPL-compatible licenses for their work.
2) The FSF has long anticipated potential incompatibilities between GPL versions, which is why its recommended to license your software under the GPL or any later version, and indeed such language is provided in the license by default. LibreCAD's original developers changed the license to make it GPLv2 only, intentionally breaking license compatibility with software licensed under future versions of the GPL.
If these people did these actions by accident without realizing the ramifications then the GPL could be faulted for being confusing (despite the very comprehensive FAQ page provided by the FSF), but I don't see how it's GPL's fault when people go out of their way to be GPL incompatible, or incompatible with future versions of GPL.
[+] [-] cjensen|11 years ago|reply
[+] [-] fluidcruft|11 years ago|reply
A LibreDWG plugin to LibreCAD is likely to be easily distributable purely as source. FSF doesn't like to talk about it (it gets into quite dangerous territory for them when talking about interpreted languages), but the reality is that GPL relies on copyright. Specifically, it relies on their ability to make a copyright claim to things you are trying to share. If they have no actual copyright claim to something you're distributing, they are shit out of luck.
FSF and licensing@ really likes to pretend that they can prevent this. But they never give specifics and always talk in generalities. They are bluffing and have no actual legs to stand on. If they have no copyright claim to something, they have no actual power over it. At all. It's out of their GPL's grasp.
Of course, the FSF also wants it both ways when convenient. For example, they want to co-opt Matlab's MEX API for GNU/Octave, and simultaneously prevent you from doing the same to their code. This is allowed because of their magical "compatibility" wand. Apparently, they may copy interfaces as they wish, but for nebulous reasons you are expected to trust them that you may not do the same.
See also the existence of ZFS on Linux distributed as patches/source-code shims (not a FSF issue as FSF does not own the Linux kernel, but ultimately it's the same licensing issue).
[+] [-] AceJohnny2|11 years ago|reply
I'm a little surprised that a library such as LibreDWG is not licensed LGPL, but come to think of it, this isn't unexpected from the hardline FSF (Stallman). They want to create a coherent free ecosystem, and are unwilling license their work with licenses that have known loopholes.
That's a damn shame for these almost-but-not-quite-as-libre software projects such as LibreCAD, FreeCAD, Open Asset Import Library and Bender.
What's wrong with the OpenDWG [1] libraries, licensed GPLv2?
[1]: http://sourceforge.net/projects/opendwg/
[+] [-] r0muald|11 years ago|reply
The Open Design Alliance [0] despite its name does not make available any open source implementation of DWG.
[0] <http://www.opendwg.org/>
[+] [-] retroafroman|11 years ago|reply
Edit: Not much:http://libregraphicsworld.org/blog/entry/librecad-gets-nativ... but now QCAD can import some features of older versions of DWG.
[+] [-] clarry|11 years ago|reply
[+] [-] Alphasite_|11 years ago|reply
[+] [-] imanaccount|11 years ago|reply
[+] [-] majika|11 years ago|reply
Some extra requirements of licensees of GPLv3 programs:
- Preventing tivoization: they must provide information on how to install and execute modified versions of anything they build with the code, when they provide the corresponding source;
- Preventing DRM: anything they build with the code won't be legally considered a DRM system, and they "waive any legal power to forbid circumvention" of that system;
- Protecting against patents: they must provide every recipient of the program (in source or binary) with any patent licenses necessary to exercise the rights that the GPL gives them, and if they try to use a patent suit to stop another user from exercising those rights, their license will be terminated.
I would place the blame of this situation at the maintainers of the GPLv2-only libraries: why don't they want to use a license that prevents tivoization, DRM, and patent abuse? If they've changed their minds on copyleft, why not switch to the BSD license?
I'm not sure exactly what the LibreDWG developers asked of Stallman, but if it was to change the GPLv2 or GPLv3, that's stupid. Even if he did, it wouldn't help their situation.
On the trollish title: the GPLv3 stifling "open-source" development? Really? (Edit: the title has since been changed) Let's talk about how all the non-copyleft licenses let companies run off with the code to build their own systems behind closed doors, and only contribute what they want, when they want -- if at all? How does that support free software development?
I'm licensing my projects under the AGPLv3, because I believe that nonfree software is harmful, and I don't want to contribute to its development at all. I believe that a free society must necessarily operate on free software. I want to encourage the development of free software, and discourage the development of nonfree software.
The GPL was designed to ensure that the software stays free software; "to ensure that every user has freedom". The GPL's protections may have sufficed in 1990, but they don't in 2014. The GPL doesn't consider users of a web service to be users of the software implementing that server. Thankfully, the AGPL does.
The AGPL ensures that if the code is used to implement a web service, then the entire source code of that web service must be free software. This way, I'm not contributing to nonfree software, whether it's executed locally or provided over a network.
[+] [-] pyre|11 years ago|reply
[+] [-] dragonwriter|11 years ago|reply
It supports free software development because contributing to free software works in practice, so less complex requirements means lower legal barriers means more people participate.
> The GPL's protections may have sufficed in 1990, but they don't in 2014.
The GPL's restrictions may have been necessary to motivate people to contribute back in 1990, but they aren't in 2014. The people making the practical case for free software succeeded, and active communities around permissively licensed Free software abound. Sure, proprietary forks and proprietary value-added enhanced versions also abound, but the maintainers of those (at least the latter) often also contribute (both code and money) back to the Free upstream project -- or, actually, are simultaneously the maintainer of the Free upstream project.
[+] [-] polemic|11 years ago|reply
IMO the AGPL cuts off your nose to spite your face. I understand the rationale, but I find it excessively onerous and complex for the users I really care about (other devs).
[+] [-] bane|11 years ago|reply