Apache License Updated to 2.0

Roy_Fielding writes "The Apache Software Foundation has approved an update to the open source Apache License (Version 2.0) that will be mandated across all Apache projects starting on March 1st. I have been working on variations of this license for the past three years, trying to balance the many different goals of the revision. That includes making the license easier for non-ASF projects to use, improving compatibility with GPL-based software, allowing the license to be included by reference instead of listed in every file, clarifying the license on contributions, and requiring a patent license on contributions that necessarily infringe a contributor's own patents. The result is a license that is compatible with other open source licenses, such as the GPL, and yet still remains true to the original goals of the Apache Group and supportive of collaborative development across both nonprofit and commercial organizations." While it has yet to become OSI-certified, it will probably will be so Real Soon Now, and in the meantime it's fun to compare licenses.

Interesting...

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.

I'm no lawyer, but this seems to be saying that if you sue anyone for breach of patent for something in apache, then you lose your patanet license?

Can someone explain this onein plain english, please?

Re:Interesting...

I read it as, "If you are native american and soux you may not wear patent leather shoes"

I could be totally wrong.

Re:Interesting...

Looks like a standard tit-for-tat clause. Sue me for breaking patent rights and I will rescind your patent rights. Point being that companies using Apache intellectual property should be willing to loan back to the community any IP they develop, and if they aren't willing to make that contribution they shouldn't be a member of the Apache IP community.

Re:Interesting...

I think it means that if you sue somebody alleging that, say, Apache, infringes upon a patent, you lose your license for Apache.

If this is true, doesn't this mean that it's not Free Software and non-compliant with the Debian requirements? It seems to be placing a restriction on use rather than redistribution, and is therefore EULA-like.

Re:Interesting...

I think it means that if you sue somebody alleging that, say, Apache, infringes upon a patent, you lose your license for Apache.

Not exactly. You don't lose your copyright license to Apache (which is what the Apache license mainly is), you lose the patent licenses. Picture this: suppose five companies contributed to Apache and thus granted you licenses to use their appropriate patents. You then sue one of these companies for patent infringement. The new Apache license means that in this case you may still redistribute Apache (since you still have a copyright license), but you've opened yourself up to being sued by any of the five companies for patent infringement, as you no longer have a license to those patents.

If this is true, doesn't this mean that it's not Free Software and non-compliant with the Debian requirements? It seems to be placing a restriction on use

So now you see why this is not the case. You may still use it, but you now run the risk of being sued over patents.

IANAL, of course.

Re:Interesting...

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.

I'm no lawyer, but this seems to be saying that if you sue anyone for breach of patent for something in apache, then you lose your patanet license?

I'm no lawyer either, but the way I read is that there is software in Apache that is covered by patents, but the owners have given people who use Apache a licence to use them. However, if someone decides to sue over one of their patents in Apache, then they lose the licence from the other patent holders, posibly leaving them open to an infringment suit themselves.

In short, is seems to say: you play nice, we'll play nice, and we'll all play with everyone's toys, but if you won't let someone play with your toys, then everyone else will gang up on you and not let you play with any of theirs either.

Maybe I am just stupid but...

Can someone explain to me some specific problems there were with the old license? Not having it in every file I am sure makes things a bit easier, but what else does this do exactly? The original poster didn't make this too clear, and I don't exactly go around studing licenses unless I know it conflicts with something for me personally.

Different definition of fun

in the meantime it's fun to compare licenses.

I don't know about CowboyNeal, but I sure have a different definition of fun than he does!

And yet...

After something of the same journey with licenses for my own open source work, I finally came to the conclusion that Richard Stallman had seen the inevitable truth clearly when he designed the GPL, namely that free software thrives best when there is a definite barrier between it and commercial software.

Our software now uses a dual license model in which it's either licensed for free under the GPL, or licensed for a fee under a standard commercial license.

Without exageration, nor wanting to start a religious war, I believe the GPL is an astonishingly robust answer to the question of how to share creative works without subsidising commercial interests that inevitably seek to quash the independent creative spirit.

Dual licensing

I thought it was clear but I'll explain again.

My company writes software that we like to distribute as free software. I started doing this in 1995, and the tools and packages I make are quite widely ported and used. When I started my company our policy was that anything we could not rapidly turn into a product we would release as open source. Software that is not used dies, we figured.

Our license was a BSD-style license that basically allowed people to do what they liked with the software.

In 2001 I noticed that some commercial products incorporating our code were being sold. Very good, I thought, it's nice to see our work being used. But when I asked them to provide us with some of the extensions and patches they'd made, the answer was "no, this is not possible". Now, seeing people use the results of years of work then refusing to contribute anything back rather annoyed me. My company was selling support licenses for our products, and these were in fact our competitors.

The solution came in the form of an article by Richard Stallman [linuxtoday.com] which explained why using the LGPL was in fact giving help to closed-source developers who directly or indirectly compete with open source developers.

We decided to switch to the GPL, and in 2002 we moved all our OSS products to this license. At the same time, we had a number of commercial licensees. To give these groups a non-viral package, we developed a dual licensing model.

Since the code is ours, it's our right to license it to specific users under specific terms. The GPL is not incompatible with commercial licenses, so long as it's the copyright holder who decides what license to apply in each case.

To summarize: for OSS developers we use the GPL, for commercial developers we use a commercial license.

It works well. We've had no GPL violations, and enough commercial licensees to make it worth developing our core packages further.

Software patents

The Apache License 2 is just a workaround for a real problem. Software patents are bad for development and bad for the economy. The US Federal State Commission called for change in a recent report, the benefits of software patents are falsified by emirical ressearch.

However, as software patents serve for the benefit of patent attorneys the institutions are intrested in an extension of the system, by widening the scope of patentability regardless of an economic foundation. Politicians like this quantitative patent approach, the result are many trivial patents of low quality and disfunction of the patent system atlarge. Many low quality patents endanger our information society. So it is nice to see that organisation like Foundation for a Free Information Infrastructure [ffii.org] build a counter-force to patent lawyer interest groups in Europe. So far the lobby work against software patents and the Eurolinux petition [noepatents.org] were very succesful. In my opinion we need a world wide movement in order to avoid Eolas vs. MS oder Amazonvs.Barnes&Noble ecc. will happen again. The GNU Public License is incompatible with Patent law and most projects and SME cannot afford to get patents. They, the innovators, don't want or need software patents.

100% correct and in agreement

Software (et al.) patents are essentially:

- legalized monopolies, and monopolies are bad.
- legal minefields for software initiatives, and mines are extremely counterproductive in (real and analogous) terms.
- an involuntary subsidy by the software industry for the self-serving legal industry.
- an impediment to open (and commercial) software development.
- an idea based solution to an implementation driven market (ideas are cheap; implementation is hard).
- a backwards implementation benefitting manipulators of the system, not the brilliance in innovation.
- an upside down system benefitting those with big pockets (and fat lawyers) instead of the underdogs (for whom patents were very originally intended).
- a joke in any industry (e.g., software) where production is a mere disk copy away (i.e., there is no manufacturing impediment in software, so even the original concept of patent protection is absurd)
- completely absurd to implement in theory (due to impossibility of any patent examiner's job) and practice considering the infinitude of like ideas in different forms (with patents as with novels, plots, etc.)
- absurd since software is a written form of thought --- and you shouldn't be able to patent (anything but especially) thoughts.

That's just off the top of my head.

IOW, patents are a Lehman (prounounced "lemon") law perpetuated so the legal industry can forcefeed carcinogenic lemonade to the software industry.

Re:Software patents

Because you don't patent ideas. You patent implementiations.

Or at least you should. This is where the current system has gone out of whack. But the point of patents is to encourage people to develop ideas into actual implementations, and then share those with the world.

Software is something different; it exists in the grey area between an idea and an implementation. It is an expression of an idea. Luckily, we already have something designed to deal with expressions of ideas: copyright. Which is all that should apply to software.

Is it actually GPL compatible?

It would be really great if this license was GPL compatible since that would allow much more cooperation and use of Apache licensed works in other free software projects. But is it actually GPL-compatible?

I couldn't find the Apache 2.0 license on the FSF license list [fsf.org]. The only "official" remark seems to be the following email thread [apache.org] which says:

Whether or not they are considered compatible by the FSF is an opinion only they can make, but given that a derivative work consisting of both Apache Licensed code and GPL code can be distributed under the GPL (according to *our* opinion), there really isn't anything to be discussed.

Maybe it was a bit premature to announce this license without waiting for OSI approval and requesting feedback from the FSF. Of course the Apache group can do whatever they want without asking for approval and blessing from other Open Source and Free Software groups. But it would have been nice to try to cooperate a bit more.

Re:Is it actually GPL compatible?

But it would have been nice to try to cooperate a bit more.

Well, not to flame or anything, but cooperation works both ways. The FSF appears (to me) to be completely unbending in its interpretation of the spirit of the GPL, and what is and isn't compatible with it. They'll help you to change your licence to make it compatible, but they won't shift a millimetre on their own position.

Okay, so their philosphical and moral stance more or less requires that behaviour, but it does seem a little unfair to criticise other groups for not cooperating.

Re:Is it actually GPL compatible?

When you make a work consisting in the combination of works covered by two different licenses, the resulting work must be licensed under the terms of both licenses at the same time. If they have conflicting clauses, then the licenses are incompatible and the work is not distributable at all.

The GPL says a couple of things about this: in clause 2 it says: "[...] when you distribute the same sections as part of a whole which is a work based on the Program, the distribution of the whole must be on the terms of this License, whose permissions for other licensees extend to the entire whole, and thus to each and every part regardless of who wrote it." In clause 4 it says "You may not copy, modify, sublicense, or distribute the Program except as expressly provided under this License." And in clause 6 it says "Each time you redistribute the Program (or any work based on the Program), the recipient automatically receives a license from the original licensor to copy, distribute or modify the Program subject to these terms and conditions. You may not impose any further restrictions on the recipients' exercise of the rights granted herein."

In plain English this means: when you make the combined work I talked about before, the whole work must be distributed under the terms of both licenses combined; but as the GPL requests that it be distributed under the terms of the GPL (no less, no more), if the "other" license includes restrictions the GPL does not have, then both licenses are incompatible and you cannot distribute the resulting work.

So, whether your license is GPL-compatible is not just a matter of opinion. You only have to ask yourself: "does this license have any restrictions the GPL does not have?" If it does, then the license is GPL-incompatible. If it does not, then the license is GPL-compatible.

looks trivialy GPL-incompatible

They tried "improving compatibility with GPL-based software", but is it compatible or not?

From a quick read it seems to be a valid Free Software license, but clause 4.d may make it incompatible with the GPL. This would be unfortunate for such a trivial condition. GPL doesn't allow placing additional restrictions on distribution, so is requiring a NOTICE file, and additional restriction?

The patents section might also be GPL-incompatible, but it might be GPLv3-compatible when GPLv3 comes out.

Has anyone seen of any comments from FSF about this?

Altogether, it's a good license, and vastly superior to the last proposal which was ~100 pages long. (slight exageration)

Lawschool?

Is going to lawschool a requirement for beeing a geek nowadays?

Ciryon

question about apache license

Does the Apache license allow commercial distribution of code based on the apache code, in the case where the modified source code is included as part of the commercial package, but the end-user is prohibited from redistributing the source code?

Strange

What is unusual about this license is that it grants a license to the copyright. Most licenses only grant a right contained and protected by copyright law. For example, a company may grant a license to distribute a good, or to reproduce elements of an original work of authorship. I have never heard of being able to license a copyright. We assign and transfer copyrights and we license trademarks and rights granted under tm, (c), and or patent. I wonder if this agreement is really valid. Lets say that someone modifies some code and then registers the copyright and does not inform the Library of Congress about the Apache license. Moreover, what if the "thief" then files suit against derivative authors. The problem here is that fed law trumps the law of contracts and under federal law you can share copyright in a work- it's called joint ownership. One suggestion is that Apache could grant a nonexclusive right to create derivative works so long as they meet the highest standards in the industry. It would be interesting to see if Apache filed the copyright through the LOC? It might be interesting to look into how foreign civil law treats software because it is more natural law/moral rights based and does not allow corporate entities to own the full ip rights of a creator.

IANAL but

it's fun to compare licenses.

IANAL but you sure sound like one :)

Yeah, real fun... (rolls eyes)

"...and in the meantime it's fun to compare licenses."

You're a real fan of root canals and Britney Spears too, aren't you?

How does a big project do a license change?

This might be a tad offtopic but anyway, a while ago I was wondering how a big OSS project (like Apache) does a license change? If there are dozens of people involved who have contributed their work to the project under license X, wouldn't you need to ask each and every one of those people if you want to change that license in any way?

GPL "compatibility" is a bad move.

One of the greatest virtues of the old Apache license was that it was not "viral." If a viral license can be stamped on the code at will, it may eventually lead to a situation in which one must be subjected to the restrictions of the GPL to use the work. This would not be a good thing.

PHP & MySQL

I bet the driving motivation behind this license update was to resolve legal issues with PHP (which is under the Apache license) linking to the MySQL client library (GPL!). Due to the grey area licensing issues between the Apache and GPL licenses, the PHP guys were actually planning to drop MySQL support from PHP! Good to know they are working on fixing that.

Re:Patents

why should I donate all my software patents to the FSF for contributing to Apache ?

You shouldn't have software patents in the first place. But if you do, and you contribute to an Apache Licensed project, you need to license under specific terms (not donate) the relevant (not all) patents to the users of that project (not the FSF, which has nothing to do with this story, or even the ASF). If you aren't willing to let the users use the code you wrote, then the project doesn't want it.

Re:Offtopic : Why bother with the Apache section?

I am going to guess that it is because apache is one of the MAJOR reasons linux ever took off in any commercial setting in the first place. Preconfigured apache on linux boxes that were extremely cheap made very attractive web servers to many people and businesses. This was a major boost to linux awareness and usage. I think we all know how popular apache is as a webserver, and I think it is fair to say that the majority of apache servers out there are running on a *nix platform. Without apache, these platforms may not have ever got much exposure.

Re:GPL

The GPL requires to share the source code for modifications and also forces any software linked to GPL code to be licensed under the GPL as well. The Apache license does neither. From my POV the Apache License is more free then the GPL while the GPL tries to keep free software free. IMHO a good compromise is the LGPL which protected the freedom of your work but does not require others to open their source when linking to your stuff.

Re:wtf is with firebird's page rendering on slashd

Works fine for me on Firebird 0.7 on Fedora and Win XP Pro.

Re:wtf is with firebird's page rendering on slashd

Mozilla Firebird 0.7 (Gecko/20031103) here; and I definitely have this problem. I'd say approximately every fifth view of a Slashdot page only the menu on the left and the page header appear. Refreshing once or twice normally solves the problem.

I don't blame the Gecko engine for this but rather the Slashdot HTML which fails to validate [w3.org] as any version of HTML.