I had no idea there was a base library exception in the GPL. I guess I’ve never read that far into it (finished the BSD, ISC, and MIT licenses a bunch of times, though).
It’s surprisingly just one line, which says “System Libraries” are excluded from the requirement to convey source code. It was needed to allow GPL software to run on proprietary operating systems, since you typically have to link with a not-open-source libc or win32.dll or such.
Should this really be what free software development is about? I hate the GPL on many levels and enjoy the liberty the BSD licenses give.
Reading the GPL licenses, especially version 3, puts me into an orwellian state where not only all corporations are under state control, but also every free thought of every individual.
How can it be that we really need to discuss if it is permitted to statically link a GPL-library into a BSD-licensed project. Why does the GPL need to be such a cancerous disease forcing itself upon everything that touches it? It’s not like it’s stopping big companies from abusing free software, and don’t tell me about those few lawsuits won by free software activists.
Most people endulging in free software activism aren’t even developers. Most of them are just people who want to impose power on other people, stopping people from doing their job, like those fine folks over there working on this great piece of software called LibreSSL.
Freedom is about choice. And imposing a limitation of choice through a license is not freedom, it’s abuse of power through a legal construct that definitely requires a reformation.
Why does the GPL need to be such a cancerous disease forcing itself upon everything that touches it?
Please don’t. I’m aware that I’m in the minority for being a copyleft advocate on lobste.rs, but language like this is incredibly unproductive, inflammatory, and generally not appropriate for the level of discourse we’re aiming for here.
This particular situation has nothing whatsoever to do with the GPL. It’s one developer making controversial statements about how they wish their license to be interpreted.
Do you also consider the likes of Apple or VMware to be engaged in totalitarian thought control, since their chosen licenses (“all rights reserved”) are a “cancerous disease” that restrict any derivative works you link with them? Honestly I’m not even sure how to respond to a comment like this, with such a high ratio of over-the-top bluster to substantive comments. You can disagree with different companies' and individuals' licensing choices without becoming totally unhinged.
Of course not! Thing is, if they publish their code under an “ARR”-license, it’s their free choice.
The conditions imposed by the GPL are worse though. Or have you ever heard Apple complain about somebody publishing an .app including a certain set of Apple libraries? Every OS X .app is practically statically linked.
Not so with the GPL. If you would do the same with a GPL library, the authors could potentially hire lawyers up your asses, even though both parties have a mutual goal of bringing free software into the world.
I definitely understand your notion, but you got to realize that these totalitarian licenses are not helping our cause. I don’t know if you’re even a developer, but I’ve seen countless times when perfectly valid GPL-projects were rewritten to fit a BSD-licensed environment. Do we really need that?
The GPL strictly gives you more freedom than “all rights reserved”. It cannot take away any rights that you don’t have by default, since all it is is a conditional permission grant. If you don’t like the conditions, you don’t have to accept the license, and you’re back at the default rights that copyright law gives you.
It sounds like you have a problem with copyright law in general, i.e. you think there should be a universal right to make derivative works without having to get a license from the copyright holder?
you think there should be a universal right to make derivative
works without having to get a license from the copyright holder?
No definitely not! If you knew the GPLv3 you would know the situation regarding static linking.
I wrote a BSD-licensed tox-client, but toxcore is licensed under the GPLv3. Any packager who would distribute statically linked binaries (client + static toxcore) would incriminate himself.
This is the thing I criticize, besides the fact that the GPL is too bloated and complex to comprehend for us humans. ^^
I like the GPL, but there is some merit in “no holds barred” rights, get rid of trademarks, copyright, patents, licenses, no reverse engineering clauses. Allowing anyone to copy/use whatever they like however they like. All music/movies/pictures/software/games would effectively become “public domain”. I would expect use of DRM to spike as companies try to prevent losses, the public may naturally select them out of business by favouring easy to share non-DRM versions, we might even see a rise in people buying non-DRM content (If the prices are right). I’m not saying that is better (Prices for non-essential goods may well plummet, people will rip of or even sell copies of your work, some artists would not bother creating some works for example) it’s just an interesting side tangent.
I’m thinking of the license for OSX. It has some kind of funky custom terms that try to restrict what you do with it. For example the license is conditional on hardware you run it on: running it on Apple hardware is licensed, but running it on non-Apple hardware is a license violation. This even “looks through” VMs, so virtualizing OSX is legal iff the VM itself runs on Apple hardware. There are further terms restricting what you can do with a number of the components, e.g. you can’t expose iTunes via an SaaS (not even if you’re running it on top of a genuine OSX on top of genuine Apple hardware). Much of the software also comes with an EULA where you agree (if EULAs are valid in your jurisdiction) not to try to reverse engineer it or its protocols.
Reading the GPL licenses, especially version 3, puts me into an orwellian state where not only all corporations are under state control, but also every free thought of every individual.
How many software licenses have you read? The GPL is much longer and more complex than the BSD, sure - but it’s short and readable in comparison to most commercial licenses. And any control over thoughts was inherent to copyright law, since all the GPL does is give you extra permissions that you wouldn’t normally have.
How can it be that we really need to discuss if it is permitted to statically link a GPL-library into a BSD-licensed project.
We don’t. This is only an issue because OpenSSL has its own special (non-BSD) license. Honestly I find that behaviour far more irritating. The GPL is widely known, widely used and widely understood. If you want a GPL-like license, use the GPL! If you want a BSD-like license, use the BSD license! Don’t write your own different version of it just because.
Why does the GPL need to be such a cancerous disease forcing itself upon everything that touches it? It’s not like it’s stopping big companies from abusing free software, and don’t tell me about those few lawsuits won by free software activists.
If I’m not supposed to tell you about the times the GPL was effective then what can I tell you?
I have a game that I bought that I can only play on my preferred computer thanks to the GPL - the developers released the source because of the GPL, and so people were able to port it to other operating systems. That’s real freedom that I value, and the only way to ensure that the “end users” get that freedom is something like the GPL.
f you want a BSD-like license, use the BSD license! Don’t write your own different version of it just because.
That’s oversimplifying a bit. The OpenSSL license is essentially the Apache 1 license, which is pretty well understood. It’s also about the same as the original 4 clause BSD license, also GPL incompatible.
Fair enough. I’d argue that the advertising clause achieves very little in comparison to the trouble it causes when integrating code with something else, but that’s a separate discussion.
Although there are exceptions, I generally feel confident knowing what I can and can’t do with proprietary software. I find the GPL more confusing. It may grant more freedom, but it contains a lot of conditionals.
I usually find the opposite. Since this thread seems to be mainly about linking with libraries, it’s really a minefield to link your software with proprietary libraries and distribute the result. By default it’s not allowed, but sometimes it is. The terms are almost always custom, buried in a hugely complex license drawn up by the company’s lawyers. When can your code, which depends on Matlab libraries, be distributed? Depends on which Matlab license you have, what kind of entity you are, what your field of business is, and who you want to distribute it to. Similarly with, say, proprietary GIS libraries, databases, or most scientific/finance/business/simulation packages.
A very partial list of common problems:
The type of use of your software impacts the license terms, for example it may be legal to link your code with theirs and distribute the result, including for commercial gain, but not to use it as the basis of an SaaS. The GPL doesn’t care.
The type of user impacts the license terms, for example if you’re an academic entity, terms are different than if you’re a business entity. What if you’re an academic entity, previously happily complying with licenses, who now has a joint DARPA grant with Lockheed which will use some of the software? Oops, better call the lawyers, your licenses may or may not be valid anymore. The GPL doesn’t care here, either.
Value-add and non-competition requirements. To keep you from just reselling the base product as your own rebadged product (or offering it as an SaaS), your software that links with it has to be sufficiently different in functionality, market, etc. The line here is usually pretty fuzzy. The GPL doesn’t judge whether your product has sufficient value-add over the base library.
Architectural requirements. Want to move your beefy server sitting in the closet to a distributed set of EC2 instances, or vice-versa? Better ask a lawyer first if our license allows this. The GPL has no opinion on distributed vs. consolidated architectures.
I would gladly take a GPL’d version of just about any proprietary library I’ve had to use in the past, to avoid most of this mess. And people do switch for that reason; the clearer terms around distribution are a big driver of Matlab->Octave switches.
All fair points. I’ll just say I had the opposite experience. We paid $500 a seat for some library and that gave us binary distribution rights. It can go either way. Some of the derived work questions apply to GPL too. (To say nothing of affero…)
The question of whether merely optionally linking with readline forces all of clisp to be GPL may have been settled favorably for stallman, but otherwise threw up a giant “here be dragons” sign to others.
You mention os x restrictions to apple hardware. That may be onerous or immoral or whatnot, but I’ve rarely questioned whether my hardware counted as apple hardware or not. It seemed clear enough.
For the record, I have my ideological differences with the GPL but I understand what people are trying to do. And it’s the authors right to do that. But there are practical implications I find… Bitter tasting? I like the LGPL a lot more, because compliance is much easier even in the edge cases.
Speaking only for myself: http://www.superloopy.io/articles/2007/gpl-vs-bsd-license.html – mind you, my motivation wasn’t all about freedom. It was just as much about convenience and “marketing” for me and my projects. I used BSD license for my most popular project ever, started that year. (It might indeed have been the project that prompted that post–I cannot remember.)
Stunnel’s author might do well to consider the lessons of history in order to see how this kind of thing has played out in the past. The case of XFree86 comes to mind, and I’m sure there are plenty of more recent examples.
[Comment removed by author]
He appears to particularly object to the removal of EGD and FIPS support.
Based on the linked articles/tweets, the stunnel author is under the impression that LibreSSL exists “only to benefit OpenBSD,” whatever that means.
I had no idea there was a base library exception in the GPL. I guess I’ve never read that far into it (finished the BSD, ISC, and MIT licenses a bunch of times, though).
It’s surprisingly just one line, which says “System Libraries” are excluded from the requirement to convey source code. It was needed to allow GPL software to run on proprietary operating systems, since you typically have to link with a not-open-source libc or win32.dll or such.
Should this really be what free software development is about? I hate the GPL on many levels and enjoy the liberty the BSD licenses give. Reading the GPL licenses, especially version 3, puts me into an orwellian state where not only all corporations are under state control, but also every free thought of every individual.
How can it be that we really need to discuss if it is permitted to statically link a GPL-library into a BSD-licensed project. Why does the GPL need to be such a cancerous disease forcing itself upon everything that touches it? It’s not like it’s stopping big companies from abusing free software, and don’t tell me about those few lawsuits won by free software activists.
Most people endulging in free software activism aren’t even developers. Most of them are just people who want to impose power on other people, stopping people from doing their job, like those fine folks over there working on this great piece of software called LibreSSL.
Freedom is about choice. And imposing a limitation of choice through a license is not freedom, it’s abuse of power through a legal construct that definitely requires a reformation.
Please don’t. I’m aware that I’m in the minority for being a copyleft advocate on lobste.rs, but language like this is incredibly unproductive, inflammatory, and generally not appropriate for the level of discourse we’re aiming for here.
This particular situation has nothing whatsoever to do with the GPL. It’s one developer making controversial statements about how they wish their license to be interpreted.
Do you also consider the likes of Apple or VMware to be engaged in totalitarian thought control, since their chosen licenses (“all rights reserved”) are a “cancerous disease” that restrict any derivative works you link with them? Honestly I’m not even sure how to respond to a comment like this, with such a high ratio of over-the-top bluster to substantive comments. You can disagree with different companies' and individuals' licensing choices without becoming totally unhinged.
Of course not! Thing is, if they publish their code under an “ARR”-license, it’s their free choice. The conditions imposed by the GPL are worse though. Or have you ever heard Apple complain about somebody publishing an .app including a certain set of Apple libraries? Every OS X .app is practically statically linked. Not so with the GPL. If you would do the same with a GPL library, the authors could potentially hire lawyers up your asses, even though both parties have a mutual goal of bringing free software into the world. I definitely understand your notion, but you got to realize that these totalitarian licenses are not helping our cause. I don’t know if you’re even a developer, but I’ve seen countless times when perfectly valid GPL-projects were rewritten to fit a BSD-licensed environment. Do we really need that?
The GPL strictly gives you more freedom than “all rights reserved”. It cannot take away any rights that you don’t have by default, since all it is is a conditional permission grant. If you don’t like the conditions, you don’t have to accept the license, and you’re back at the default rights that copyright law gives you.
It sounds like you have a problem with copyright law in general, i.e. you think there should be a universal right to make derivative works without having to get a license from the copyright holder?
No definitely not! If you knew the GPLv3 you would know the situation regarding static linking. I wrote a BSD-licensed tox-client, but toxcore is licensed under the GPLv3. Any packager who would distribute statically linked binaries (client + static toxcore) would incriminate himself. This is the thing I criticize, besides the fact that the GPL is too bloated and complex to comprehend for us humans. ^^
I like the GPL, but there is some merit in “no holds barred” rights, get rid of trademarks, copyright, patents, licenses, no reverse engineering clauses. Allowing anyone to copy/use whatever they like however they like. All music/movies/pictures/software/games would effectively become “public domain”. I would expect use of DRM to spike as companies try to prevent losses, the public may naturally select them out of business by favouring easy to share non-DRM versions, we might even see a rise in people buying non-DRM content (If the prices are right). I’m not saying that is better (Prices for non-essential goods may well plummet, people will rip of or even sell copies of your work, some artists would not bother creating some works for example) it’s just an interesting side tangent.
Which licenses?
LLVM’s BSD license? CUPS’s GPL license?
I’m thinking of the license for OSX. It has some kind of funky custom terms that try to restrict what you do with it. For example the license is conditional on hardware you run it on: running it on Apple hardware is licensed, but running it on non-Apple hardware is a license violation. This even “looks through” VMs, so virtualizing OSX is legal iff the VM itself runs on Apple hardware. There are further terms restricting what you can do with a number of the components, e.g. you can’t expose iTunes via an SaaS (not even if you’re running it on top of a genuine OSX on top of genuine Apple hardware). Much of the software also comes with an EULA where you agree (if EULAs are valid in your jurisdiction) not to try to reverse engineer it or its protocols.
How many software licenses have you read? The GPL is much longer and more complex than the BSD, sure - but it’s short and readable in comparison to most commercial licenses. And any control over thoughts was inherent to copyright law, since all the GPL does is give you extra permissions that you wouldn’t normally have.
We don’t. This is only an issue because OpenSSL has its own special (non-BSD) license. Honestly I find that behaviour far more irritating. The GPL is widely known, widely used and widely understood. If you want a GPL-like license, use the GPL! If you want a BSD-like license, use the BSD license! Don’t write your own different version of it just because.
If I’m not supposed to tell you about the times the GPL was effective then what can I tell you?
I have a game that I bought that I can only play on my preferred computer thanks to the GPL - the developers released the source because of the GPL, and so people were able to port it to other operating systems. That’s real freedom that I value, and the only way to ensure that the “end users” get that freedom is something like the GPL.
That’s oversimplifying a bit. The OpenSSL license is essentially the Apache 1 license, which is pretty well understood. It’s also about the same as the original 4 clause BSD license, also GPL incompatible.
Fair enough. I’d argue that the advertising clause achieves very little in comparison to the trouble it causes when integrating code with something else, but that’s a separate discussion.
Because authors are free to license their work however they like, even if doing so would be a personal inconvenience to you,
Well put. A license describes how the author would like you to use their work, it definitely doesn’t have to describe how you wanted to use it.
Well, won’t prohibit me from criticizing those ugly licenses. ;)
Maybe you should preferentially criticize the proprietary-software licenses where you don’t even get access to the source code — they’re even uglier!
Although there are exceptions, I generally feel confident knowing what I can and can’t do with proprietary software. I find the GPL more confusing. It may grant more freedom, but it contains a lot of conditionals.
I usually find the opposite. Since this thread seems to be mainly about linking with libraries, it’s really a minefield to link your software with proprietary libraries and distribute the result. By default it’s not allowed, but sometimes it is. The terms are almost always custom, buried in a hugely complex license drawn up by the company’s lawyers. When can your code, which depends on Matlab libraries, be distributed? Depends on which Matlab license you have, what kind of entity you are, what your field of business is, and who you want to distribute it to. Similarly with, say, proprietary GIS libraries, databases, or most scientific/finance/business/simulation packages.
A very partial list of common problems:
The type of use of your software impacts the license terms, for example it may be legal to link your code with theirs and distribute the result, including for commercial gain, but not to use it as the basis of an SaaS. The GPL doesn’t care.
The type of user impacts the license terms, for example if you’re an academic entity, terms are different than if you’re a business entity. What if you’re an academic entity, previously happily complying with licenses, who now has a joint DARPA grant with Lockheed which will use some of the software? Oops, better call the lawyers, your licenses may or may not be valid anymore. The GPL doesn’t care here, either.
Value-add and non-competition requirements. To keep you from just reselling the base product as your own rebadged product (or offering it as an SaaS), your software that links with it has to be sufficiently different in functionality, market, etc. The line here is usually pretty fuzzy. The GPL doesn’t judge whether your product has sufficient value-add over the base library.
Architectural requirements. Want to move your beefy server sitting in the closet to a distributed set of EC2 instances, or vice-versa? Better ask a lawyer first if our license allows this. The GPL has no opinion on distributed vs. consolidated architectures.
I would gladly take a GPL’d version of just about any proprietary library I’ve had to use in the past, to avoid most of this mess. And people do switch for that reason; the clearer terms around distribution are a big driver of Matlab->Octave switches.
All fair points. I’ll just say I had the opposite experience. We paid $500 a seat for some library and that gave us binary distribution rights. It can go either way. Some of the derived work questions apply to GPL too. (To say nothing of affero…)
The question of whether merely optionally linking with readline forces all of clisp to be GPL may have been settled favorably for stallman, but otherwise threw up a giant “here be dragons” sign to others.
You mention os x restrictions to apple hardware. That may be onerous or immoral or whatnot, but I’ve rarely questioned whether my hardware counted as apple hardware or not. It seemed clear enough.
For the record, I have my ideological differences with the GPL but I understand what people are trying to do. And it’s the authors right to do that. But there are practical implications I find… Bitter tasting? I like the LGPL a lot more, because compliance is much easier even in the edge cases.
[Comment removed by author]
… and that less and less libre software is community-driven.
(C.f. http://ebb.org/bkuhn/blog/2014/12/03/conservancy-supporter.html for a discussion.)
Citation needed.
Legitimately asking for a citation, since I see no signs of the GPL loosing ground…
Speaking only for myself: http://www.superloopy.io/articles/2007/gpl-vs-bsd-license.html – mind you, my motivation wasn’t all about freedom. It was just as much about convenience and “marketing” for me and my projects. I used BSD license for my most popular project ever, started that year. (It might indeed have been the project that prompted that post–I cannot remember.)
Stunnel’s author might do well to consider the lessons of history in order to see how this kind of thing has played out in the past. The case of XFree86 comes to mind, and I’m sure there are plenty of more recent examples.