But what if your code writes files or communicates over the network? And what if, say, Microsoft takes your permissively-licensed code that is just beginning to gain popularity, changes one piece of it to make it incompatible with yours, and doesn't release their change? They use their massive marketing department or pre-bundling to get lots of Windows users to use it. Everyone is still free to use "your" version of the code, but nobody wants to, since it is incompatible with the more popular Microsoft version, now on its way to becoming "the standard."
If you are okay with that, no rabid copyleft zealot like me has any place to criticize you. Release your code with a permissive license. That's fine. (Permissive licenses are considered "Free Software" by FSF and rms as well, after all.)
But to me, that seems grossly unfair for code I spent my valuable time designing and developing, so I don't release my code under a permissive license. If Microsoft wants to build a product with the work I have done, it needs to convey the same freedoms to its users that I did. That's the cost.
And sure, everyone is free to use the v2 version of the code (i.e. gcc 4.2), but nobody wants to, since all new development is happening with the v3 version, which is now the standard.
Linux is a famous exception: it omits the "or any later version" clause, and so avoids getting steamrolled in the way you describe.
How can they relicense the code unless every contributor signed a Contributor License Agreement granting them that right?
A quick search seems to indicate that GCC has no such CLA.
So if you write something under the GPLv2, and I change it and publish it, then the GPL says I need to release the source under the GPLv2 or later. This means that I can publish it under the GPLv3, and you can't take back my changes unless you upgrade your software to the GPLv3.
Linux removed the "or later" text from their version of the GPLv2, meaning that I can't change the Linux kernel and publish the result under the GPLv3.
However, you can "upgrade" LGPL code to GPL, see LGPLv3 "2. Conveying Modified Versions".
There is a FLOSS license that has an implicit upgrade to new version clause: the Mozilla Public License, see "6.2. Effect of New Versions" in MPLv1.1.
It looks like the "or later clause" is a built in option, but not a built in default, of the GPLv2. With that I mean that the meaning of that phrase is explained in the 9th clause.
At least to my knowledge the opposite seems to be more common. Experts release a library under a very permissive licence, and everybody wins. One such example is libPNG.
KHTML (LGPL, not GPL) taken by Apple to make WebKit (LGPL), later "rewritten" and licensed BSD for 2.0.
Apple did also the dirty trick of giving a single huge patch against the previous release of KHTML making it almost impossible to integrate back to the main codebase. Even LGPL can be abused. Lesson learned the hard way by the community.
And you now have everyone parroting how great WebKit is, nobody remembers the original team at KDE. It was brutal.
> While Microsoft uses the Kerberos protocol, it does not use the MIT software.
The license of the original implementation is thus irrelevant. Pretty much the only thing that would have prevented MS would have been if there were patents on Kerberos.
How would the Oracle lawsuit change anything? Protocols are almost explicitly the things that are not protected by copyright. The problem is, people here are confusing interfaces in human-readable code (copyright-eligible) with interfaces for binary interoperability between systems (not eligible).
1. http://www.heise.de/newsticker/meldung/Microsoft-und-Open-So... 2. http://slashdot.org/story/00/05/02/158204/kerberos-pacs-and-... 3. http://features.slashdot.org/story/00/05/18/2015251/our-atto...
Or the zillion of extensions being made by companies on top of LLVM.
> "Embrace, extend, and extinguish", is a phrase that was used internally by Microsoft to describe its strategy for entering product categories involving widely used standards, extending those standards with proprietary capabilities, and then using those differences to disadvantage its competitors.
https://en.wikipedia.org/wiki/Embrace,_extend_and_extinguish
MS did embrace and extend standards, but that is a very different thing from code.
I feel this argument is weak: even if I license my works under BSD, no one can make my BSD licensed code that has already been distributed into to a proprietary product, even not myself.
To make that software that I licensed under BSD a proprietary product they would need to somehow convince all users of my software to get rid of all copies as well as convincing me to stop distributing it under a free licence.
Building a proprietary product that includes my code does NOT remove anyones freedom to use my code, no?