3

On the article:

  1. Op-ed: Oracle attorney says Google’s court victory might kill the GPL

They say:

It is hard to see how GPL can survive such a result. In fact, it is hard to see how ownership of a copy of any software protected by copyright can survive this result.

But I could not figure out why would it kill GPL?

  • 1
    It looks to me like the looser of the case, is trying to brain wash us. There is little new in this case. And the GPL is not open in the way that Oracle describes. GPL does not give freedom to do as you want, however what about permissive licences, does it affect them. Well yes because they would fall into this category, and no because they are already permissive. They are correct that if it affects the GPL, then it also affects all proprietary licences. However I think they are just trying to scare us into the cloud, to get developers to use their cloud products. – ctrl-alt-delor Jul 12 '17 at 18:07
9

"Fair use" is the legal doctrine in the United States that says sometimes use of copyrighted material without permission is "fair" and allowed. The author of this op-ed is concerned that Google had a very weak fair use case, but won the case anyway:

Google won a verdict that an unauthorized, commercial, competitive, harmful use of software in billions of products is fair use. No copyright expert would have ever predicted such a use would be considered fair.

The thesis here is:

...Google's narrative boiled down to this: because the Java APIs have been open, any use of them was justified and all licensing restrictions should be disregarded. In other words, if you offer your software on an open and free basis, any use is fair use.

This author is concerned that anyone will be able to use this case a precedent to defend any use whatsoever of publicly-available source code, even if that use ignores requirements like those imposed by the GPL. People can just reuse GPL-licensed code however they want, says this author, and if they get sued they can just defend their use under fair use.


I'm not a lawyer, but I don't find this argument very convincing. I do agree that the fair use finding in Oracle v. Google is somewhat surprising, since fair use looks unfavorably at commercial use in general (with notable exceptions). However, in this case, the use under consideration was Google's reimplementation of Oracle's Java APIs. That is, Google didn't reuse any part of Java code -- they instead used the API specification (function names, arguments, return values) to write their own brand new code base.

Oracle v. Google was the landmark case that decided that the structure of an API was protected under copyright. Google argued that API structure was purely functional and not part of the creative work protected by copyright, but lost that argument. Immediately following this loss, there was another case to decide if Google's unauthorized reuse of the Java API was protected by fair use. The jury in that case decided that it was.

Considering that this was the first-ever case to consider the application of fair use to API structure, I don't think we can draw any sweeping generalizations from it. As one counterargument, perhaps future case will acknowledge that API structure is covered by copyright, but it constitutes so small a part (and so functional a part) of the work that it will be relatively easy to defend unauthorized reuse of an API. If such a legal future came to pass, that won't have much bearing on any other kinds of reuse outside of API reuse.

5

GPL is a very strong copyleft license. Part of this strength stems from the claim from the FSF that mere linking (even dynamic linking) with a library creates a derivative work of the library and thus the GPL restricts the licensing options for this derivative. This allows the FSF to propose a distinct lighter license (the LGPL) that allows this kind of linking more broadly.

This claim has been contested a lot over time and this is yet another blow against it as use of an API that was shared publicly (as with a GPL library) could be considered fair use and wouldn't allow the author of the API to impose restrictions on the resulting product, making thus the GPL actually not very different than the LGPL. The same issue could arise with the AGPL.

I want to emphasize though that I have no legal training and wouldn't take a bet in the final outcome of this (and we might not know for years because a court ruling on this particular issue would be needed to give a definite answer).

Interestingly the counter-opinion from Pamela Samuelson linked at the bottom of the article you shared says:

Annette Hurst, one of Oracle's lawyers, has recently argued that Google's victory will be harmful to open source developers. I have a great deal of respect for her as a lawyer, but this claim is not just greatly exaggerated—it's wrong.

But then goes on making an argument on why it is good that an API can be copied (which most free software developers would approve) but doesn't defend against the aforementioned issue.

It is also interesting to note the difference in vocabulary: one says free software and talks only about the GPL, the other says open source and makes no specific claim about the GPL. Oracle's lawyer wants to appeal to FSF / Richard Stallman's supporters while Google's lawyer wants to appeal to the larger crowd of open source enthusiasts who care about free reuse but less about copyleft.

  • 1
    I don't see much of a link between reimplementing an API and dynamic linking to a library. – curiousdannii Jul 12 '17 at 14:25
  • 1
    Well. I am not the original author of the argument but this is how I understood it. When your code uses a library it is only a derivative of the API of the library. If there is a broad interpretation of fair use in the case of an API, creating such derivatives could also be considered fair use. I agree though that this is far from established. – Zimm i48 Jul 12 '17 at 14:31
  • Take it to a logical extreme: imagine library A is GPL'd, you create library B with the same API but a very poor quality implementation. Then you link your proprietary app against B and provide instructions for linking it to A instead at runtime. That would be an end run around the GPL's linking restriction. Personally, I'm not really a fan of that restriction in the first place, but I can see how it would be an issue for projects like GNU Readline, where the API has a trivial implementation. – Kevin Jul 13 '17 at 0:52
  • @Kevin My personal opinion is if you don't distribute GPL licensed stuff then its license doesn't apply to you. It may be against the spirit of the license to tell your users how to combine things you can't combine for them, but the freedom to run and use should still allow that. – curiousdannii Jul 13 '17 at 4:24
  • @curiousdannii: Now I'm tempted to release a libcrappyreadline under the MIT license and watch the fireworks. – Kevin Jul 13 '17 at 4:31

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.