Open Source Stack Exchange is a question and answer site for people organizing, marketing or licensing open source development projects. It only takes a minute to sign up.
Sign up to join this community
I'm interested in the Copyfree Open Innovation License as a simpler alternative to Apache 2.0 (it is a permissive license with an explicit patent grant). But because it's not so popular, not much information is available about it.
In particular, where does it fit into this graph?
Image credit: David A. Wheeler
Is it GPL-compatible? With which versions? And where does it fit in with the permissive licenses?
Not a lawyer, but I disagree with Free Radical's interpretation.
COIL (it's not CFOIL) includes "Permission to reproduce, modify, distribute, publish, sell, sublicense, use, and/or otherwise deal in the licensed material without restriction." It fits precisely with the MIT part of the graph in the question here.
That means one can completely sublicense, i.e. use the work in a derivative released on GPL or any other license, period.
The misinterpretation comes from thinking that the patent clause must be strong. In fact, the patent clause is weak for the reasons Free Radical points out. If I want to bypass the patent clause, all I need to do is sublicense under MIT-style license, and the patent clause is gone.
Effectively, COIL is a copyfree (as in the name) license with a patent clause. Being that permissive, it is not strong. It does explicitly state in your use of the license that you are including a patent license, so if you then turned around and sued someone over a patent, they would have some defense. But it fails to ensure that the patent grant remains permanently for further distribution. The patent clause stays only as long as people continue to choose COIL.
Indeed, it would give you no legal recourse if someone else stripped out the patent clause for their derivative. Think of COIL as the lightest-weight version of a patent-sensitive license stating that you will never sue anyone over a patent you hold related to this work. It's unilateral patent disarmament.
The copyfree concept basically rejects the whole nature of copyright law. It's not useful for those who intend to sue others over copyright violation. The only use of a copyfree license is to assure others that they can use the work and that they won't be sued. In that regard, COIL assures them that you won't even sue over a patent either.
In short: Apache v2 and GPLv3 both provide stronger and more persistent patent protection. COIL allows you to have a weak license similar to MIT-style while still addressing the patent issue as long as people choose to keep the reference.
The Copyfree Open Innovation License License (CFOIL) is significantly shorter than the Apache 2.0 license. CFOIL grants:
provided that:
Unlike Apache 2.0, CFOIL lacks the additional supporting elements for license enforcement that are present in Apache License 2.0. Specifically, it lacks language throughout the license that indicates that the license grants has the terms and conditions of the license is a precondition for the license grant.
Apache 2.0, by constrast, is quite explicit about this:
Subject to the terms and conditions of this License, each Contributor hereby grants ...
Unlike Apache 2.0, CFOIL fails to make explicit that the specific intent of the drafters is to provide the license grant to only to those individuals or entities that abide by the terms and conditions contained in the license. Such an additional provision in the Apache 2.0 License strengthen the conclusion that the conditions in the license is, legally speaking a "precondition" rather than a "covenant". It also lacks the Apache 2.0 patent termination clause:
If You institute patent litigation against any entity (including a cross-claim or counterclaim in a lawsuit) alleging that the Work or a Contribution incorporated within the Work constitutes direct or contributory patent infringement, then any patent licenses granted to You under this License for that Work shall terminate as of the date such litigation is filed.
which makes its embedded patent-license weaker than the one in Apache 2.0.
In fact, its patent terms seems incompatible with any copyleft license.
In practice, these flaws in this license will limit your options (compared to Apache 2.0) if you want to seek legal remedy for license violations. If all you want to have a really permissive license with a (sort of) patent license embedded, and do you will not even think about pursuing license violations, and do not fear patent suits filed from contributors will pose a problem, and you don't care about license compatibility, then it may be OK.
But if your alternatives are this license and Apache 2.0, you should be aware of the differences. Apache 2.0 is IMHO superior.
As far as license compatibility goes, I don't think it is compatible with any version of the GPL or any other copyleft license. It looks like a permissive license but its messed up patent license clause rules out GPL comptability and compatibility with Apache 2.0
So I can understand why this one is not much used, as it:
So to conclude: I don't think it fits anywhere on the license map reproduced in the question.
Disclaimer: I am not a lawyer. This should not be regarded as legal advice. If you require legal advice, I recommend you seek the counsel of a bar certified attorney, or equivalent licensed professional for whatever jurisdiction you call home.
TLDR: The short version of the below is that I believe at least fifty percent of Free Radical's answer is mistaken or imprecise, and the COIL fits more closely with the PD/MIT/BSD group than any of the other licenses in that list, apart from the fact the COIL comes with a patent clause protecting recipients from patent litigation.
As noted by both MSalters and Wolftune, it appears that Free Radical has made some errors of interpretation. The COIL does indeed contain the language "subject to these conditions", and does indeed provide for sublicensing. Free Radical's explanation seems self-contradictory on the point of conditional licensing, in fact, where it says abiding by the license terms is not a precondition for the grant of license, just after saying that the license applies "provided that the user retains applicable copyright or other legal privilege notices, these conditions, and the following license terms and disclaimer." In my estimation, the "covenant" versus "condition" complaint raised by Free Radical should concern you exactly as much as it does for the MIT/X11 License or the Simplified BSD License (which is to say: probably not at all).
The concern Free Radical shows for license "enforcement" is largely inapplicable, because the license is (apart from requirement for copyright and license notice requirements) composed entirely of permissions, which requires no enforcement, as Wolftune pointed out. I further agree with Wolftune's assessment of how to classify "strong" versus "weak" patent terms in the context of granting permissions/license to recipients, and with MSalters' assessment in the context of the kinds of guarantees for the recipients of the licensed work. While I have no particular objection to a patent termination clause strictly as an anti-aggression provision, I also do not think that the lack of such a thing particularly devalues the patent clause itself.
Regarding GPL compatibility, it is worth mentioning that the Apache License 2.0 (AL2) is incompatible with every version of the GPL except GPLv3, and Apache License 1.0 is incompatible with all versions of GPL. In addition, AL2 and AL1 are mutually incompatible. Meanwhile, as Wolftune points out, COIL should not in fact be incompatible with any version of the GPL for a number of reasons, including the fact that it does not contain copyleft provisions (that is, copyright license terms affecting any parts of the work other than the parts explicitly covered by the COIL itself). Free Radical's primary claim for license incompatibility seems to center around patent conditions, but GPL incompatibility on that, I think, would have to involve the GPL's requirement that other licenses combined with it cannot be more restrictive than the GPL, and the terms of the COIL are in every way more permissive instead.
The AL2 has additional problems, outside of this discussion's subject matter, that are worth investigating before you adopt it for any project. While it is far from an authoritative resource, WikiVS offers some comparison in its Apache License vs COIL article.
In addition to the above, the COIL is also more generally applicable than the other licenses in the original question's attendant diagram, as the rest of them are software-specific, while the COIL refrains from tying itself specifically to software while still covering the categories of conditions applied to software by the other licenses, making combination of works covered under the terms of the COIL with other, non-software works (including documentation, artwork, et cetera -- issues of some concern to game development I would imagine) more predictably addressed by those terms.
Depending on how you interpret that diagram in relation to patent clauses in your software copyright licenses, I think COIL should fit in there either between public domain and the MIT/X11 License, or between "BSD-new" (also known as Revised BSD License or 3-Clause BSD License) and the rest of the world below and to the right of it (including AL2, MPL, and various *GPL options).
EDIT: I disagree quite a bit with Free Radical's assertions that COIL "creates legal uncertainty" (a statement made without meaningful support) and "creates license proliferation for no good reason" (there are very good reasons for the existence of the COIL, such as offering a "permissive" copyfree license with a patent clause; AL2 is not copyfree, is not nearly as permissive as some people think, and comes with some problematic clauses unrelated to the subject matter directly addressed in this discussion).
2021 EDIT: Version 1.0 of the COIL (abbreviated COIL-1.0 in the SPDX license database) has been assessed by a lawyer and fits the picture painted in these comments, except that it contradicts what Free Radical said.
