Op-ed: Oracle Attorney Says Google's Court Victory Might Kill the GPL (arstechnica.com)

Annette Hurst, an attorney at Orrick, Herrington & Sutcliffe who represented Oracle in the recent Oracle v. Google trial, has written an opinion piece for Ars Technica in which she urges developers and creators to not celebrate Google's win in the hard-fought copyright case as the decision -- if remains intact -- is poised to make them "suffer" everywhere and also the free software movement itself "now faces substantial jeopardy." As you're aware, in a verdict earlier this week, a federal court announced that Google's Android operating system didn't infringe on Oracle-owned copyrights because its re-implementation of 37 Java APIs is protected by "fair use." Hurst writes: No business trying to commercialize software with any element of open software can afford to ignore this verdict. Dual licensing models are very common and have long depended upon a delicate balance between free use and commercial use. Royalties from licensed commercial exploitation fuel continued development and innovation of an open and free option. The balance depends upon adherence to the license restrictions in the open and free option. This jury's verdict suggests that such restrictions are now meaningless, since disregarding them is simply a matter of claiming "fair use." 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. Software businesses now must accelerate their move to the cloud where everything can be controlled as a service rather than software. Consumers can expect to find decreasing options to own anything for themselves, decreasing options to control their data, decreasing options to protect their privacy.

Comments

Multiple Award Winning

[dcollins]

"Best Sour Grapes of May 2016"

"Best Nail in Coffin for Confidence in Legal Judgement"

Re:Multiple Award Winning

[chrism238]

Well, she would say that, wouldn't she?

Re:Multiple Award Winning

[arglebargle_xiv]

Let me rephrase the lawyer's text into something more comprehensible to the masses:

Ladies and gentlemen, this is Chewbacca. Chewbacca is a Wookiee from the planet Kashyyyk. But Chewbacca lives on the planet Endor. Now think about it; that does not make sense! Why would a Wookiee, an 8-foot-tall Wookiee, want to live on Endor, with a bunch of 2-foot-tall Ewoks? That does not make sense! But more important, you have to ask yourself: What does this have to do with this case? Nothing. Ladies and gentlemen, it has nothing to do with this case! It does not make sense!

Look at me. I'm a lawyer working for a multibillion dollar software company, and I'm talkin' about Chewbacca! Does that make sense? Ladies and gentlemen, I am not making any sense! None of this makes sense! And so you have to remember, when you're in that jury room deliberatin' and conjugatin' the Emancipation Proclamation, does it make sense? No! Ladies and gentlemen of this supposed jury, it does not make sense! If Chewbacca lives on Endor, Google's victory will destroy the GPL!

Re:Multiple Award Winning

[jthill]

So long as you confine yourself to copying the index of the book and taking out the page numbers and every entry that isn't a proper noun, go for it.

Re:Multiple Award Winning

[solidraven]

API description doesn't equal actual code though. So yeah this is a lot of drama for a list of supported functions and methods essentially. The majority of the work (the code behind the API) is still protected. So bugger off.

Re:Multiple Award Winning

[phayes]

No, the point the lawyer for Oracle is trying to make is a distinction that does not exist.

Oracle has been trying to make a case that API=Code so others cannot copy their API's without a licence. For everyone else on the planet, API!=code and APIs are free to use ether because they are not copyrightable (as the judge originally ruled and I agree with personally) or because they are fair use.

Now that the Java APIs used in Android have been determined by the jury to have been fair use, Oracle is now attempting to make the case that because (in their opinion) API=Code, if APIs are fair use then so is ALL code, including the GPL.

Nobody except Oracle & co believe that API=Code so the woman has no point.

Re:Multiple Award Winning

[nyet]

Look genius:

This is an API:

extern int add(int a, int b);

This is code:

int add(int a, int b)
{
        return a+b;
}

Copy the prototype all you want, I don't give a crap.

Re:Multiple Award Winning

[Eunuchswear]

You can make jokes but she has a point...why would I give a shit about the GPL if I'm a company if I can just take the parts I want and claim fair use?

Why sould I worry about laws against murder? I can just shoot you dead and claim self defense.

Re:Multiple Award Winning

[brantondaveperson]

Design of a really good API is a far from easy task - though I'm certainly not going to argue that Java's libraries constitute a "good API". Designing really quality and useful APIs involves organising some fairly abstract ideas into very concrete representations, and designing as few simple operations on those representation as possible, while still managing to get whatever the job is done. Doing so without restricting what can be done, and as the same time without making the effort of getting what you want done, done, is hard. Having written Java for Android in the past, I can claim with a little authority that Java's API fails everywhere, and is dreadful to use - but nevertheless, it does seem to me that API design, and perhaps also the intellectual property therein, might not quite be the trivial thing many people seem to belive it to be.

Whether or not API is code (it isn't, obviously) is beside the point. The question surely is, can API constitute an invention, or a work, or whatever the language might be? And it seems to me that it can.

Re:Multiple Award Winning

[Kiwikwi]

It doesn't matter if designing an API is hard... at least not in the US, which does not follow the sweat of the brow doctrine.

It does matter if API design involves substantial creativity. If the API design is not simple and obvious, and if the work to design the API is a substantially creative endeavor, the API becomes eligible for copyright, and indeed, this is what the courts have now found (at least in the specific case of the Java APIs). As much as I don't like the idea of copyrightable APIs, I can't really fault the ruling.

Fortunately, the court has now also ruled that using APIs is fair use, which is an essential ruling from an interoperability PoV. And despite this op-ed piece, I'm certain that most companies will be relieved that there are limits to Oracle's absurd attempts to squeeze money of out any company that use Java.

Personally, I'm just happy that the company I work for chose C#/.NET as the scripting platform for customers, not Java. It was touch-and-go there for a while, but Microsoft and .NET has turned out to be much more trustworthy platform stewards than Sun/Oracle and Java.

Re:Multiple Award Winning

[turbidostato]

"If Chewbacca is on Endor he is probably a pedophile."

Exactly this. A "Chewbacca defense" by the book (only it makes even less sense now: the court has already ruled so why bother?).

And then, "oh! this will mean everybody will go with an as-a-service model!" as if they wasn't for that anyway.

Oracle is a big behemoth, so it will take time to fall, but they are going the SCO path, so it's not unreasonable to expect they'll follow the same outcome.

Re:Multiple Award Winning

[segedunum]

They've also regurgitated this crap before years ago: https://www.publicknowledge.or...

Re:Multiple Award Winning

[DuckDodgers]

The GPL never prevented reverse engineering or fair use. If I write libfoo and release it under the GPL, and you write libbar with all of the same APIs and release it under another license, that was never illegal or a GPL violation.

As far as I understand it, if you re-implement GPL software using the original code (not API, but implementation code) as a starting point, then you are violating the GPL. But the Java standard library is colossal and the bits that Google copied directly are tiny and obvious. So I genuinely don't think the GPL is weakened by this.

Separately, I agree Google is no friend to GPL. But I don't think this lawsuit is some clever attempt for them to sabotage the GPL so they can make Android 7 fully proprietary from top to bottom. I think this lawsuit is just to avoid paying Oracle licensing fees.

Re:Multiple Award Winning

[Lonewolf666]

As others have written, there is a difference between APIs and implementation. But GPLed software is not more endangered than lots of closed source software is too. Because it does not depend on the sort of license used. With tools like Dependency Walker (http://www.dependencywalker.com/), you can in many cases also discover the interfaces of proprietary code, re-implement them and and try to claim fair use. I guess your chances of succeeding would be about the same as in the Oracle vs. Google lawsuit.

For what it's worth, I agree with the decision of the court, but my point is that the GPL is not really special here. If GPL software is in danger now, lots of closed source software is too.

Considering Google and trying to control Android, I agree that is a dick move but does not make Android worthless for the community. Cyanogen Mod for instance describes itself as
"a customized, aftermarket firmware distribution [...] based on the Android Open Source Project"
So the Android Open Source Project still lives on in ways that serve the end user. I guess if a phone vendor wanted to avoid being pressured by Google, they could try and use Cyanogen Mod instead of Android. Marketing might still be difficult though, depending on how many people know what Cyanogen Mod is all about.

Re:Multiple Award Winning

[Ol Olsoc]

You can make jokes but she has a point...why would I give a shit about the GPL if I'm a company if I can just take the parts I want and claim fair use?

Why sould I worry about laws against murder? I can just shoot you dead and claim self defense.

That's Florida's Stand your ground laws. They work it would appear. http://www.huffingtonpost.com/...

Yeah yeah - I know - Huffpost

Re:Multiple Award Winning

[cwsumner]

I think it is known as "Poisoning the well". 8-(

They are ignorant lawyers and don't know the difference between an info list and operational code. Or, maybe they don't think "the law" knows. Or, maybe they just don't think the public knows. (Or maybe they just don't think...)

Re:Multiple Award Winning

[slashdot_commentator]

why would I give a shit about the GPL if I'm a company if I can just take the parts I want and claim fair use?

Because both you and her are idiots. Both you and her think you can take ideas and make it into exclusive intellectual property. And fatally, both you and her think an API is the same thing as copying code. Source code is the actual human effort and investment used to produce a sequence of machine intelligible instructions that computes something useful. An API is a description of how source code talks to other source code. You can call something tangible property, but you cannot call an idea property. You can't patent or copyright a steering wheel to a car. You can't even do that when the wheel concept is used to steer a boat.

The problem is that patents and copyrights have been distorted for the past twenty five years to try to distort laws to treat ideas a form of property. This Google victory only shows that the system is broken when it overreaches as Oracle tried to do. And not just the GPL has nothing to do with this case, the GPL is not used to protect license owners from being deprived of their projected profits. The GPL is a legalist way to make their source code available to the public for use, without allowing other parties to seizing it, or deny the original owners the modifications to the GPL owner's source code.

Google is not a friend of the GPL, so what? Oracle and Google are not football teams to root for; they both pull annoying legal or market shit on users. But the lawyer has an excuse for her idiocy. She apparently believes that corporations are living beings, rather than fictions created by the legal process. What's your excuse?

Re:Multiple Award Winning

[jedidiah]

Furthermore, the GPL was specifically engineered with jackasses in mind. It creates a framework where entities can cooperate on equal terms despite them being natural enemies and being otherwise un-equal.

Re:Multiple Award Winning

[Kjella]

It does matter if API design involves substantial creativity. If the API design is not simple and obvious, and if the work to design the API is a substantially creative endeavor, the API becomes eligible for copyright, and indeed, this is what the courts have now found (at least in the specific case of the Java APIs). As much as I don't like the idea of copyrightable APIs, I can't really fault the ruling. Fortunately, the court has now also ruled that using APIs is fair use

And that's really what I find strange, because if you think APIs are creative and copyrighted, when if not in the Google case are they infringed? They copied much of it, for profit, reducing the market for licensed use of Java. It feels like one bad twisting of the law to fix another bad twisting of the law. And "fair use" is an affirmative defense that depends on your particular case, just because Google won doesn't mean the APIs are now free. If you use them in some other way in some other context maybe the judge will rule against you. It's not as fucked up as it could have been, but still pretty messed up.

Re:Multiple Award Winning

[unrtst]

why would I give a shit about the GPL if I'm a company if I can just take the parts I want and claim fair use?

Fortunately, that's not the case here.
The case here is that reuse of the API is being deemed ok. There is a little bit of a grey area in defining the borders of said API, but those rarely matter at all because they are almost always cases where there's only one good and obvious way to implement that thing, so it's not defensible anyway. This won't hurt the GPL at all.

In fact, the opposite is true. There is an awful lot of GPL software, as well as software that covers the gamut of closed to public domain, that has re-implemented other API's. I doubt it applies directly to this Oracle v. Google case, but https://en.wikipedia.org/wiki/....

The author of the Op-ed in question seems to be over committing to the false premise that API and code are the same thing, making up a fake world where that was true, and saying that would cause problems for GPL software which, while there would be some minor downsides, is not really true - it would allow us to open up any software after snagging a copy of the source, which would make most GPL advocates quite happy. IE. it'd be at least as bad for proprietary software, and probably far worse. Thankfully, that very premise is bunk, so none of this matters.

Re:Multiple Award Winning

[silentcoder]

Regardless all else, it's long been held by the courts that reverse engineering is fair use, that automatically implies that APIs MUST be fair use as well - because it's impossible to reverse engineer anything without replicating the API and by definition a reimplimented API will not be distinguishable from the original - otherwise it would not be compatible.

If anything it's Oracle's point that could kill the GPL. Almost the entire GNU/Linux userspace is filled with programs that are shell-api replicas of older unix commands, the GLIBC library is a direct replica of the API in the original unix LIBC. That's not even getting to software like WINE.

These projects would all die if APIs can have copyright. Now the earlier court found it can, but the fair use ruling provides a way to keep them alive, that's something we need to cling to because without it - we're all dead in the water and not just the GPL. The entire global software industry would grind to a halt.

Now maybe she has a point that her bizarre interpretation of that concept would scare some companies out of dual-licensing... so be it, the GPL was more than 2 decades old before any dual-licensed software ever existed, the free software movement was close to 3 decades old before there were common dual-license as a business-model companies.
We did just fine without those companies for decades, if they bugger off - we will do just fine without them. Hell we'll take what they built and thanks to the GPL continue on our own forks into the future as we have done many times before when a dual-licence software company failed to play nice with the community - including quite recently with Oracle on two of the most important free software projects they acquired - OpenOffice to LIbreOffice and mysql to MariaDB, but there are many earlier examples - like Hudson/Jenkins.
In most cases where a corporation doesn't play nice with the community and we fork the thing to continue it ourselves - it's the corporate version that ultimately dies off.

Re:Multiple Award Winning

[Kiwikwi]

And that's really what I find strange, because if you think APIs are creative and copyrighted, when if not in the Google case are they infringed?

Well, from a straight-forward reading of the law, it is possible for a method of expression to be subject to copyright, but also essentially always subject to fair use. Which would seem to be what the court has found. I guess a similar case could be made for e.g. a political speech – sure, it's subject to copyright, but by its very nature, any third-party use is almost inevitably fair use.

But I agree, it's still messed up.

Re:Multiple Award Winning

[micahraleigh]

+1 Wish I had mod points

They don't know what they're talking about

[Anonymous Coward]

It's the implementations that they're protecting with the GPL, not the interfaces.

Re:They don't know what they're talking about

[ShanghaiBill]

It's the implementations that they're protecting with the GPL, not the interfaces.

She is a lawyer, not a programmer. What she is saying is nonsense, but it is grammatically correct, and succinctly encapsulates Oracle's outrage at the verdict.

Re:They don't know what they're talking about

[WarJolt]

She is a lawyer, not a programmer. What she is saying is nonsense, but it is grammatically correct, and succinctly encapsulates Oracle's outrage at the verdict.

No where near as outraged as all the OSS developers are at all of Oracle's patent trolling over the years, so what was she thinking trying to co-opt them?

We love copyright law, you're not going to beat us at our own game. Oracle should stick to abusing patent law. They'll still lose. Now they also look like idiots.

Re:They don't know what they're talking about

[MightyMartian]

This, I suspect, is going to form a part of their appeal, a sort of IP version of "what about the children?", except this variant is called "what about the GPL?"

This is about interfaces, not their implementation, and no one is challenging that code can be copyrighted. But I get it, maybe the next judge will be the right kind of fucking moron for Oracle to gain a victory.

Re:They don't know what they're talking about

[Guy Harris]

She is a lawyer, not a programmer. What she is saying is nonsense

Maybe she shouldn't be talking about something of which she is ignorant. Then again, she's an attorney and from what I've been able to observe, attorneys frequently talk out of their collective asses about things which they know nothing about, especially in technical or scientific fields.

A company I worked at a while ago was planning to threaten another company with a claim that "you couldn't possibly have implemented that without stealing our stuff"; I remember the threat letter repeatedly claiming it was "inconceivable" that they could have done that.

"The Princess Bride" came to mind.

Maybe hyperbole is Standard Operating Procedure for lawyers in many situations - perhaps the belief is that if you can get the people to whom the hyperbole is addressed to believe it, you win.

Re: They don't know what they're talking about

[matbury]

Not biscuit. It's a muffin

Not muffin. McMuffin. Try ordering a muffin in McDonald's and wait for the blank stare from the cashier.

Re:They don't know what they're talking about

[dhasenan]

If the interfaces are fair use, I could have a non-GPL project depending on a GPL library, and that dependency won't affect the licensing of my project.

This reduces the difference between the GPL and the LGPL. However, in order to sidestep that difference entirely, you have to distribute your application separately from the GPL'd library.

As a practical matter, I don't think people tend to be that concerned when I, for instance, release code under the MIT license with GPL dependencies. Compile the work and distribute it, and you still have to follow the GPL's restrictions. But I might be wrong about that.

Re: They don't know what they're talking about

[FatdogHaiku]

Not biscuit. It's a muffin

Not muffin. McMuffin. Try ordering a muffin in McDonald's and wait for the blank stare from the cashier.

I just carry a picture that matches the one on the cash register.
Bonus, this will also work with a robot as long as it has a camera as one of its inputs...

Re:They don't know what they're talking about

[Dahamma]

No, that doesn't make sense, and it doesn't change anything. It's not about libraries, it's about interfaces. You could create you own library using the same function names/signatures, but you can't use someone else's code beyond that.

Re:They don't know what they're talking about

[Anonymous Coward]

I think what he is saying is this:

gfoo.so is a GPL binary. It's header (gfoo.h) defines the calls you can make. He could make ngfoo.h that implements the same functions and layout as gfoo.h, but the ngfoo.c file is full of no-ops. Then he could link dynamically against gfoo.so (compiled from his ngfoo project), and successfully compile. His application could then be put together with teh gpl gfoo.so and use it. All without violating copyright.

And yeah, I've basically challenged someone to show why this won't work for years and never gotten a good answer.

Re:They don't know what they're talking about

[currently_awake]

The open source movement wants API's free of copyright, because it means we can make open source versions of closed source software and it will still work. General rule: When a lawyer says something that is clearly the exact opposite of reality, you can assume they are lying not stupid.

Re:They don't know what they're talking about

[Dahamma]

Good point - and if you really think about it, that's the way it SHOULD be. If we prevented the case you describe, all emulators like WINE, MAME, etc would not be possible. It goes all the way back to DR DOS, etc. I think it's pretty clear that the open source community has gained SO MUCH MORE from being able to reverse engineer interfaces than any sort of "workaround" of GPL-like library licenses. Linux itself was based on UNIX/POSIX. If clean-room implementation of interfaces weren't fair game it would never have existed!

Re:They don't know what they're talking about

[currently_awake]

A layers job is like poker, you can win with the cards or you can win by bluffing.

Re: They don't know what they're talking about

[orlanz]

I think that is a grey area historically. It's why LGPL and GPL3 were created. But the community in general is ok with that.

What they are not ok with is if you distribute your interface and the GPL compiled code closed. You would have to either have your installer check and install the GPL version from a 3rd party or distribute the GPL code freely (many consumer products do this). Additionally, any modifications you do to the GPLed code would have to be released too as they are GPLed.

Also, keep in mind this may not count under fair use like Google did. The primary content of Googles deritive work is the implementation. The copied interface is a small piece of it. Your code if defined as a deritive would have the majority of the value come from the GPL implementation, weakening the fair use defense.

Re:They don't know what they're talking about

[whoever57]

She is a lawyer, not a programmer. What she is saying is nonsense, but it is grammatically correct, and succinctly encapsulates Oracle's outrage at the verdict.

Lawyers are today's hired guns. They do and say whatever is in their employer's interests, with little (or no) regard for the truth.

Re:They don't know what they're talking about

[Anonymous Coward]

> It goes all the way back to DR DOS,

It goes back further than that: MS-DOS (1981) used the API of CP/M (1975).

DR-DOS, on the other hand, had permission to use MS/PC-DOS API and FAT which goes back to DRI demonstrating that PC-DOS 1.x could display a hidden DRI copyright notice. Part of the settlement to DRI included to rights to use all PC-DOS features and APIs.

Re: They don't know what they're talking about

[camperdave]

Not biscuit. It's a muffin

Try ordering a muffin in McDonald's and wait for the blank stare from the cashier.

The McDonalds restaurants around here all serve muffins. It's part of their McCafe initiative.

Re: They don't know what they're talking about

[trabby]

Not biscuit. It's a muffin

Not muffin. McMuffin. Try ordering a muffin in McDonald's and wait for the blank stare from the cashier.

I just carry a picture that matches the one on the cash register. Bonus, this will also work with a robot as long as it has a camera as one of its inputs...

Only problem is that even robots can not make a McMuffin that looks remotely like the picture they have in the store of said McMuffin.
It will just cause a system error that the poor mechanical soul will never recover from.

Re:They don't know what they're talking about

[Sun]

First of all, here is RMS's take on this question. I should point out that I do not agree with his analysis.

As a counter example, I like people to consider NDISwrapper. No one in their right mind would claim that just because the user linked a proprietary closed source network driver with a GPL implementation of NDIS, that somehow turns the driver written by Broadcom for Windows into a derivative of the Linux kernel.

When I consulted to clients about the issue, I suggested the following criteria, under the claim that it is not industry standard, but it is defendable in court:

1. How stable the API you are relying on?
2. How documented is it? In other words, how much do you need to see the source in order to understand how to use it?
3. How many different independent implementations are there of the same API? (even if all of them are also GPL)

I claim that these are good criteria to explaining how independent are the APIs from their implementation, and accordingly, how much copyright protection they deserve.

Shachar

Re:They don't know what they're talking about

[Tough Love]

...succinctly encapsulates Oracle's outrage at the verdict

And succinctly encapsulates Oracle's past and ongoing stupidity.

Re: They don't know what they're talking about

[Vlad_the_Inhaler]

A better way of putting this: She is one of the 96% of lawyers who give the others a bad name.

Re: They don't know what they're talking about

[gnupun]

float foo (int bar); // my copyrighted interface .. Then ..

float bunny (int wabbit); // you attempt to do useful work

bunny() may be doing something very different from foo(). bunny() infringes on foo() only if you have a patent on all single-int-argument functions that returns a float. That's right, without a patent, bunny() does not infringe on foo() unless there are a lot of semantic similarities in the spec description of each function.

If they intend to legally protect interfaces, they should come up with laws that are a blend of copyright and patent. I don't think copyright in it's pure form can protect interfaces that can be subtly changed to avoid copyright infringement.

On another topic, does Google's victory mean, anyone can create a J2ME clone like Android without paying any licensing fees to Oracle? I don't understand how you can use a competitor's protected assets (Java APIs here) to create a competing commercial product and claim fair use. It would be like allowing Ford to hack into Toyota's computers, steal their car designs and manufacture based on those designs, legally.

Re: They don't know what they're talking about

[Anonymous Coward]

Then you didn't look very hard. The GPL dies not distinguish between static and dynamic linking http://www.gnu.org/licenses/gpl-faq.en.html#GPLStaticVsDynamic

Re: They don't know what they're talking about

[Hognoxious]

Isn't it more like a third party examining their brake discs and making compatible spare parts?

I thought that was one form of fair use.

Re:They don't know what they're talking about

[Xtifr]

That actually does work. There was even a real-world example: someone created a plugin for some GPL'd app that used the (proprietary at the time) RSA libraries. FSF claimed the code couldn't be distributed, because the intent was clearly to link GPL'd and non-GPL'd code into a single app. So the guy made a stub library with the same interface as the RSA library, and the FSF dropped their objections. Even though the stub library did nothing.

Of course, in such a case, you couldn't distribute the complete system, because that would clearly still be mixing the GPL and non-GPL code--you would no longer be able to call it "mere aggregation"--but he was able to distribute the plugin by itself.

Re:They don't know what they're talking about

[JaredOfEuropa]

A lawyer's allegiance is not to the truth but to their clients. That is their role in the legal system.

Re: They don't know what they're talking about

[ultranova]

Only problem is that even robots can not make a McMuffin that looks remotely like the picture they have in the store of said McMuffin.
It will just cause a system error that the poor mechanical soul will never recover from.

If Skynet decides that human society is beyond saving because Skynet worked a McJob at McDonald's in it's youth, it's... kinda hard to argue, actually.

Re:They don't know what they're talking about

[thegarbz]

She is a lawyer, not a programmer.

This is a problem that will result in open source projects landing in court to defend themselves for doing the right thing.

Re:They don't know what they're talking about

[thegarbz]

This, I suspect, is going to form a part of their appeal, a sort of IP version of "what about the children?", except this variant is called "what about the GPL?"

I wouldn't. Why would they cite the GPL which itself is a legally untested license in the court room? Oh unless their goal is to simply confuse the jury into submission in that case they may as well go ahead with it.

Re:They don't know what they're talking about

[TheRaven64]

The real stupidity from Oracle is that they didn't realise how dangerous it would be to win the lawsuit. If they had come up with a verdict that APIs were covered by copyright and implementations were not fair use, then life would have been very difficult for them. Remember the SCO lawsuit? Now imagine what a similar lawsuit would cover if all of the APIs in POSIX (and the C/C++ specifications) were copyrighted by the first person to propose and implement them. Solaris would be dead in the water, as would most other UNIX clones.

Re: They don't know what they're talking about

[Anonymous Coward]

The lady is making a threat. Oracle may try to violate the GPL to test this decision. Look for an injunction to neutralize the GPL license terms during the duration of the suit. 6+ years?

Re:They don't know what they're talking about

[Kiwikwi]

The GPL has been tested in court numerous times, both in the US and in the EU.

Along with the Artistic License, it is actually the only open source license I know of that has been tested in court. It makes sense that there's not much to go to court over with a BSD-like license, but it'd be nice to have the Apache License or the Mozilla Public License tested.

Re: They don't know what they're talking about

[fche]

"Interfaces can't be protected by copyright"

The first Oracle/Google judgement ruled otherwise.

Re: They don't know what they're talking about

[go-nix.ca]

Kneejerk reaction: No, because this case was about implementation of the interface, not about linking to the library.

Re: They don't know what they're talking about

[gnupun]

The Java API is not the protected asset, the specific implementation of it is.

You're very mistaken here. APIs are copyrightable according to a previous judgement from the appeals court. The trial was whether Google's use of those APIs was fair use or not. After all, you can legally quote a few paragraphs from a copyrighted book you are reviewing, under fair use law. This judge is totally wrong if he thinks a competing product/clone constitutes fair use (in which universe?).

And yet, every car on the market today has a steering wheel, a pedal to slow down and a pedal to speed up.

And there are numerous design decisions made in designing the steering wheel and brake pedal and they vary among manufacturers. You're confusing high level (patentable) function with low-level copyrightable designs.

Both java.io.* classes and C's stdio.h specify the interfaces to access files (high level function == file access), yet they have completely different designs (low-level copyrightable design). When Google created Android, API protection was assumed to not exist and therefore Google copied the entire API assuming there would be no problems. Well, if APIs are now copyrightable, so they must pay a licensing fee.

Re:They don't know what they're talking about

[grumbel5969]

The GPL tries to protect interfaces as well, that's why there is a LGPL. With the GPL the idea is that when you use it, your whole program has to follow it. However when interfaces are no longer copyrightable, then that falls apart and linking to a GPL library from a non-GPL piece of software becomes ok. There are some situations where this might not hold true like code inlining, static linking, etc., but in general the GPL would essentially become a the LGPL when APIs and ABIs are non-copyrightable. The FSF's stance on this always felt more like wishful thinking then hard legal ground and is kind of incompatible with their own idea that APIs shouldn't be copyrightable. That logic they apply however only to other peoples libraries, not their own.

Re:They don't know what they're talking about

[Cederic]

Were she not extrapolating from a position that's already lost in court you might have a point.

As it is, she's clearly talking utter bullshit and just doesn't understand the difference between API and implementation - which is why Oracle lost. Whether that's wilful misinterpretation or genuine stupidity I don't even give a shit.

Re:They don't know what they're talking about

[Anonymous Coward]

Not a lawyer.. But as i understood it it's all about intent..

So from my (limited) understanding you would have the following situations:
* Develop a library that presents the api in gfoo.h. Write application that utilizes gfoo.h
    - User replaces your library with the gpl version of gfoo.so -- Your intention was not to depend on the GPL code.
* Develop a library with noop's for all functions listed in gfoo.h
    - User replaces your library with the gpl version of gfoo.so so it works - Your intention was to allow the user to use the GPL'ed library or else the application would never work.

Re:They don't know what they're talking about

[DuckDodgers]

To be fair, I am a card-carrying FSF member but I work on proprietary software because I have a mortgage and kids to feed. I suspect most lawyers are effectively in the same boat - they got into law to earn a decent living and do the right things, but ended up in a situation where the choice was either "work on something morally questionable or be unemployed" or at best "work on something morally questionable or earn janitor's wages - while still having to make your student loan payments".

Re:They don't know what they're talking about

[edtice1559]

This is a great post. However, the example of a closed-source NIC driver and NDIS doesn't quite work. If the Oracle pretense were true, it would be a copyright violation to create the NDIS wrapper. But if somebody were to do so, it wouldn't create an obligation on the part of a third-party (In this case the owner of the copyright to the proprietary driver). Instead it would be something that could not be legally distributed.

Re:They don't know what they're talking about

[DarkOx]

Somehow I doubt Oracle's legal team is that short sighted. They may for one thing be perfectly content to stop selling Solaris at this point if the trade off was a massive new revenue stream of licensing in the mobile market. Sometimes is chess you sacrifice a man for better position.

However Sun licensed UNIX. Solaris isn't a clone its a branch and without getting deep into the weeds of the particulars of those very old licensing agreements I would image they are worded such that a present day court could be convinced that a license to use the implementation reasonably confers a license to use the API.

Interestingly now that you bring up under oracle logic the IEEE license to use the API in the POSIX standard it self might be on shakier ground than Solaris' use of it.

Re:They don't know what they're talking about

[Immerman]

Actually, no - their first allegiance is supposed to be to the court, and the pursuit of truth. There was even a high profile case recently where the judge sentenced the lawyers to regular ethics course for blatantly lying to the court.

Now in practice... they're supposed to at least keep up appearances. Which seems to be what most ethics courses boil down to anyway.

Re:They don't know what they're talking about

[ceoyoyo]

I think you could do it even more simply than that. Just link against gfoo.so. No actual GPL code is included in your binary except bits of the interface, which is now fair use. The end user has to supply their own gfoo.so.

The reason it wouldn't work before is that linking against gfoo.so (or ngfoo.so) would include some stuff derived from gfoo.h, or a clone of it. If that interface code is copyrightable, the GPL applies.

To me that was always an overstep by the GPL. If GPL code is compiled into your binary, absolutely the whole thing should be open source. If it's dynamically linked, it shouldn't. The user can still swap versions of the library if they so choose, which is the stated intention of that part of the GPL.

Re:They don't know what they're talking about

[John.Banister]

They just have to sue in Texas.

Re:They don't know what they're talking about

[sjames]

Incorrect. The headers would be fair game if you wanted to RE-IMPLEMENT the GPL library (in fact, that is the case now), but you still wouldn't be allowed to actually link against it.

Re:They don't know what they're talking about

[david_thornley]

The GPL has never been all that clear on what constitutes a derivative work. From a copyright point of view, I see nothing wrong with treating the GPL like the LGPL. However, the GPL is what gives you the right to copy gfoo.so, so it's at least arguable that the definitions implied by the GPL are used, and so dynamic linking is still a no-no. Has the Affero GPL been tested in court? It's a more extreme version, requiring the Affero GPLing of server code used by an Affero GPLed web client.

Have to do more than "claim fair use"...

[rbrander]

...I think a judge has to rule that it IS fair use, to render copyright "meaningless". He's just sore the judge ruled that way.

Re:Have to do more than "claim fair use"...

[queazocotal]

Not quite.
The judge could (amongst other things) as I understand it have ruled it was fair use, have ruled it was infringing - or dismissed the case because it was not a copyrightable matter.
The best case would be the latter. The first is still extremely problematic for devs, though for a very different reason the shill above claims.
https://www.eff.org/deeplinks/...
See the later caveats in that celebration.

Re: Have to do more than "claim fair use"...

[bbn]

The judge did rule that interfaces are not copyrightable but that got overturned on appeal. He then asked the jury to decide if it was fair use. The jury voted yes to that.

Re:Have to do more than "claim fair use"...

[Trongy]

The judge (William Alsup) did rule that the java API were not copyrightable in the earlier case. The appeals court overturned this ruling. Hence the current case which was started on the basis that the API was copyrighted and the open question was about fair use.

Re:Have to do more than "claim fair use"...

[Citizen of Earth]

This issue was also decided in the SCO vs. IBM case as well as the IBM-PC BIOS case, not to mention the reverse-engineering exception for interoperability in the DMCA.

Losing Attorney is BSing

[Anonymous Coward]

There never was copyright on the interface code as it is required to function. This was about Oracle bean counters trying to make their balance sheet look better through shenanigans rather than actually doing real work.

This case was not about the core code which is still covered fine by copyright and the GPL.

These comments are just spin.

Re:Losing Attorney is BSing

[Anonymous Coward]

There never was copyright on the interface code as it is required to function. This was about Oracle bean counters trying to make their balance sheet look better through shenanigans rather than actually doing real work.

This case was not about the core code which is still covered fine by copyright and the GPL.

These comments are just spin.

Then what is the LGPL needed for?

Re:Losing Attorney is BSing

[Anonymous Coward]

Of course the losing attorney is lying through their teeth, their firm just lost a multi-billion dollar suit. Everyone's in damage control mode now, which makes the fact that they _still_ have no idea what they're talking about all the more embarrassing.

Wasn't hard to figure that out, I'm guessing the jury themselves were rolling their eyes when Oracle's legal team were pretending they couldn't make any "sense" of the first letter in GNU standing for GNU...apparently no one at Oracle has ever heard of recursion before. The only metaphor I can think of to compare it to is if two absurdly rich businessmen challenged each other to a fight, then one stood there while the other punched himself in the jewels repeatedly.

Re: Losing Attorney is BSing

[BitZtream]

LGPL covers the code that implements the interface, and allows for less restricted use. It covers the same thing as GPL, just like every other license, its just a different one. Its no more or less different from GPL than BSD or MIT, for instance.

Re: Losing Attorney is BSing

[Anonymous Coward]

GPL, you link to the API and distribute, you are required to provide source.

LGPL, you link to the API and distribute, you can keep your source closed.

GPL and LGPL, you re-implement the API, you can keep it closed source. That's actually the POINT of open source. If you're not willing the put the work in, you can share with others. If you put all the work in yourself, you can do whatever you'd like.

Re:Losing Attorney is BSing

[Anonymous Coward]

These comments are just spin.

Precisely. We don't need some ignorant attorney telling us, the ones who write the code, what our business or interests are regarding software. We don't need copyright to help us share or get paid for our work. She doesn't get it. The GPL was a hack of the copyright laws designed to prevent legal interference in the affairs of the open source movement and to neutralize a legal weapon so that it couldn't be used against us to prevent sharing. It was not necessary to enable sharing. If copyright did not exist, software would still be shared freely and the GPL would have been largely unnecessary. There have been a few minor incidents where copyright law was used successfully to force sharing of GPL code from an otherwise unwilling party, but in my opinion none of these cases revealed anything but banal or poorly written additions to the original GPL code, certainly nothing of lasting importance that wouldn't have been shared otherwise. To summarize, copyright is mostly a hindrance and rarely a help to the open source movement. The GPL mostly exists to ensure that a-hole attorneys, like Annette Hurst, leave us alone. We share code because we ourselves benefited from that sharing when we were learning our craft and it would be an act of disrespect and deep ingratitude to refuse to reciprocate when the time came.

Re:Losing Attorney is BSing

[Anonymous Coward]

The LGPL essentially revokes your freedom to be able to modify the library, and this allows that library to be embedded in some other program whose license doesn't allow it to be modified. This mostly relates to static linking, as with dynamic linking, you can recompile the shared objects without modifying the programs using them.

Re:Losing Attorney is BSing

[Guy Harris]

The best part was when that vapid bitch got in front of a microphone to state quite plainly that they "didn't buy Oracle" just to file the suit against Google.

Presumably you're referring to the Oracle co-CEO saying they didn't buy Sun just to file the lawsuit?

Re:Losing Attorney is BSing

[bug1]

Infringement happens when linking occurs, nothing limits distribution prior to that as they are separate parts.
After linking of an incompatible binary with a GPL library you cannot distribute the GPL library, you can still distribute your useless application separately.

Re:Losing Attorney is BSing

[jbolden]

GPL never claimed it could go further than "derived work". Static linking is worse than dynamic linking. Intermixing source code worse than just static linking and everyone agrees this almost invariable creates a derived work. LGPL doesn't allow for linking explicitly. This ruling allows for reimplementing APIs which is even less than dynamic linking.

Re:Losing Attorney is BSing

[K. S. Kyosuke]

For example, for distribution of binaries for which you don't want to release the source code for?

Re:Losing Attorney is BSing

[phantomfive]

Static linking is worse than dynamic linking.

Why? Seems about the same to me, the only difference being the --static flag on the compiler command line (or whatever, depending on your compiler).

Re:Losing Attorney is BSing

[phantomfive]

The LGPL covers more than just the API: it covers all the implementation code in the library as well.

Re: Losing Attorney is BSing

[TheRaven64]

GPL, you link to the API and distribute, you are required to provide source.

But why are you required to provide the source? Remember, the GPL is a copyright license and so can only grant rights that copyright took away, not take away rights. If fair use permits me to use the API, then my code is not a derived work of the GPL'd code and so I can distribute it with whatever license I want. I need to obey the permissions of the GPL if I want to distribute the GPL'd library, but there's nothing stopping me from distributing the code under a different license and requiring end users to install the GPL'd library for themselves. The GPL is explicitly not a usage license, so doing so would not cause the end user to infringe the GPL.

Re: Losing Attorney is BSing

[ceoyoyo]

The GPL (and LGPL) cover distribution. If I write a program that dynamically links to a GPL library, it includes only necessary code only from the interface. If I then distribute that binary, without the actual library, I haven't distributed any GPLed code except the necessary interface. As others have pointed out, if use of the interface is fair use, then I've complied with the GPL.

The biggest difference between the GPL and the LGPL is the extension of restrictions via copyright to the interface.

Re:Losing Attorney is BSing

[ceoyoyo]

"Infringement happens when linking occurs"

I don't think that's true. Copyright restrictions apply to copying and giving copyrighted material to other people. I can do whatever I want, so long as I don't give it to someone else. If I dynamically link my program to a GPL library the only GPLed bits of code in my program are interface code, which is fair use. I can then distribute both my program and the GPL library, with the restriction that I provide all the source necessary to modify and rebuild the GPLed library.

Re: Losing Attorney is BSing

[shawn2772]

Where I disagree with the FSF is when they say that dynamically linking to GPL'd libraries also makes your program a derivative work. Whether your program needs the library depends entirely on the design of your program, so it's a bit of a gray area, but the process of dynamic linking is not far removed from simply loading and running a program.

That would make every program you run (!) a "derivative work" of the operating system, which is patently ridiculous.

Programmers tend to analyze legal codes the way they think about software. Computers are perfectly literal and pedantic, and the notion of reasonableness, or consideration of questions of context or fairness have absolutely no bearing in how they execute their code. But legal codes are executed by humans, and human judges are explicitly tasked with considering reasonableness, context and fairness, at least to some degree. They try to balance this with careful analysis of the text of the law and legislative intent.

In the context of the GPL, you're absolutely right that "linking" (especially dynamic linking) is a fairly arbitrary place to draw the line about what constitutes a derived work. But when judges look at it, it seems like a reasonable and fair line, because in the context of the way computing systems operate today, it works. Judges recognize the fundamentally fair notion of "share and share alike" that the GPL attempts to embody, see that it's a good thing for the world, that GPL authors are acting in good faith to try to provide something of value under reasonable and well-intentioned terms, and that GPL violators are taking advantage and acting in bad faith. As such, they've -- very correctly, in my opinion -- chosen to overlook the technical deficiencies of the line drawn by the FSF, in favor of its fundamental fairness.

Re:Losing Attorney is BSing

[bug1]

Yea, i was should have said 'restrictions on distribution begin when linking occurs'

Re:Losing Attorney is BSing

[Spazmania]

Infringement happens when a non-ephemeral copy is made. That's the only time infringement can happen. If I dynamically link, I consume the API into my binary object, but nothing else. I can then distributed the GPL-covered library and the closed source program in separate files. So long as I also provide the source code for the GPL-covered library, the fact that the consumption of the API was fair use means that no GPL requirements attach to the closed source file.

Whole different story if I statically link it. But why would I?

Re:Losing Attorney is BSing

[bug1]

consume the API into my binary object, but nothing else

Why would you combine a binary API but nothing else, it has no functionality.

In that situation you didnt even need to distribute the GPL'ed parts in the first place because they arent being used.

Re:Losing Attorney is BSing

[radarskiy]

"There never was copyright on the interface code as it is required to function."

A) The appeals court stipulated that it was copyrightable before remanding the case back down to Alsup.
B) There has to be copyright for a fair-use defense to apply.

Bullshit

[somenickname]

Re-implementing an API and wholesale lifting a GPL software package are not even vaguely related. And, the end part of her quote, where she proclaims doom and gloom if we don't all move to the cloud, is ponderous hyperbole. The is just scare tactic garbage from an attorney who wants to bait the waters before she takes on the appeal and charges Oracle millions of dollars for the privilege.

Re:Bullshit

[dwywit]

"Software businesses now must accelerate their move to the cloud where everything can be controlled as a service rather than software. "

I think it's less hyperbole and more a sign of what Oracle is considering. She's let the cat out of the bag - Oracle must be exploring options to charge even more for their products.

Re:Bullshit

[queazocotal]

It's an inherent part of their case that they are intimately related.
There is no bright-line seperating out APIs and code.
And indeed, they won a partial victory on this, in that the decision states that there can be _unfair_ use of APIs.
Their argument is simply that this judgement means that another judge (because there is no bright line) could determine that using (say) half or 90% of the linux kernel against GPL restrictions could be 'fair use'.
Should this be bullshit - yes.
Is it - well - it relies on courts acting rationally.

Re:Bullshit

[rwyoder]

The is just scare tactic garbage from an attorney who wants to bait the waters before she takes on the appeal and charges Oracle millions of dollars for the privilege.

This could cost Oracle millions?
So what is the downside?

Re:Bullshit

[dgatwood]

Re-implementing an API and wholesale lifting a GPL software package are not even vaguely related.

No, they're vaguely related. They both involve taking and using something that you didn't create. Of course, they're related in much the same way that bumming a stick of chewing gum and stealing a car are related....

Re:Bullshit

[Hognoxious]

So what is the downside?

Think of the poor unemployed yacht builders.

Re:Bullshit

[Anonymous Coward]

There's no "taking". An API is an interface (that's what the I stands for, duh). An interface is a requirement on you to do something with your code. An interface is not actually code. It's not even vaguely possible to get this wrong. Every programmer knows the difference intuitively. Lawyers and non-programmers generally, not so much.

Re:Bullshit

[jbolden]

Oracle has a pretty good cloud service. They sell their database to other services especially Verizon. There is no bag they have been openly talking about moving towards administering the database and away from just selling licensing.

Re:Bullshit

[shutdown -p now]

Not only that, but we've already had the debate on interfaces/APIs and GPL.

http://clisp.cvs.sourceforge.n...

And FSF had specifically supported Google in this case.

https://www.fsf.org/blogs/lice...

Oracle is just continuing their long-standing trend of treating both customers and developers as idiots. The only appropriate response is to flip the bird.

Re:Bullshit

[shutdown -p now]

"Fair use" has always been an affirmative defense against copyright infringement - the Oracle ruling doesn't really change that.

Re:Bullshit

[Frosty Piss]

"Software businesses now must accelerate their move to the cloud where everything can be controlled as a service rather than software. "

I think it's less hyperbole and more a sign of what Oracle is considering. She's let the cat out of the bag - Oracle must be exploring options to charge even more for their products.

I'm sorry, this is news? It's been more or less the public strategy for Microsoft and every other big software house for the last few years. There's no cat in any bag anymore, the cat escaped a few years back.

Re:Bullshit

[Fallen Kell]

I don't think you understand what happened in this case. API's have inherently always been allowed to be used essentially since the beginning of coding. A Judge and/or jury didn't rule that the the code in the Java was fair use to use, just the functional names. This retained the same working conditions that have existed for the last 30-40 years of code development, including the white/clean room techniques that have been used by virtually every major development studio, (including Oracle by the way). The underlying code is still protected, but you can't protect the names of the functions. I mean seriously, how many times have people written a function called "length" or "size" (hint, thousands of times).

Are we all suddenly suppose to pay royalties to the first one who called their function by that name? What about if someone wrote a program that then auto-generated creating billions of function names from every language, but each function was simply "return(1)"? Am I to get billions of dollars from every company in existence now for them infringing my copyright on all those function names?

In other words, your argument is ridiculous. The real copyright is and always has been on the specific implementation of the code, not what it is named.

Re:Bullshit

[dgatwood]

I never said that it was taking code. I said it was "taking something". It is downright silly to argue that an API isn't "something", or that it is possible to use an API without first obtaining a copy of it. So how how can you claim that implementing an API does not involve taking that API and using it?

Re:Bullshit

[El_Muerte_TDS]

So, if the Java API was in the cloud then Google couldn't have stolen in?

Re:Bullshit

[drinkypoo]

No, they're vaguely related. They both involve taking and using something that you didn't create. Of course, they're related in much the same way that bumming a stick of chewing gum and stealing a car are related....

No. They're related in much the same way that copying a cookbook and picking up a copy of a take-home menu are related. The cookbook costs money, and you have to pay. The take-home menu is provided for your benefit, and it is free.

Re:Bullshit

[Minupla]

She's let the cat out of the bag - Oracle must be exploring options to charge even more for their products.

Not news for anyone who's looked at their Oracle budget numbers... ever.

Re:Bullshit

[edtice1559]

This is a great post, but length and size are such simple names that they don't contain the modicum of creativity necessary for copyright protection. On the other hand, more complicated APIs may contain a large amount of creative work. Many times, the hardest part of designing a system is figuring out how the pieces and parts are going to fit together. I'm in no way arguing with your point or how the case was decided. Just trying to steer us toward a more apt analogy.

Re:Bullshit

[Cederic]

Thing is, Wine is doing the same thing as Android, and this court case confirms that copyright law in the US doesn't preclude this.

Re:Bullshit

[samwhite_y]

I'll keep this comment short and cryptic. Oracle is really two companies. A proprietary database company (with shallow add ons, such as the cloud) and everything else. Many comments about quality and other issues many times do not apply equally to both parts. I believe that when asserting or refuting statements about Oracle, you should explain which part of Oracle you mean and limit the scope of your comment to that part.

Re:Bullshit

[jbolden]

I not sure I agree with the premise of the question. But assuming it were true both parts openly have and have had a cloud migration strategy for their customers. The corporate managed services and government services parts (which in your premise would be the non database) are the most enthusiastic.

Re:Bullshit

[cwsumner]

When you see the phrase "The Cloud", read it as "Monopoly Control". 8-P

Re:Bullshit

[cwsumner]

... There is no bright-line seperating out APIs and code.

Actually, there is. But if someone can't read the language, they won't see the line.

... Is it - well - it relies on courts acting rationally.

This is true, and has always been a danger to technical work.

Bullocks!

[a_n_d_e_r_s]

Never read such a stupid article. If oracle has won; alot of open source projects had been dead in the USA. In EU they had lived on since there APIs are not copyrightable because of interoperability.

Now they can continue to live even in the US.

As for the point that this makes it possible for people to steal open source code. It wont. Since it's always been possible to take open source code and reimplement it as closed code. Its never been forbidden before and thus this won't change that. APIs has never been seen as protected by the software industry before and its a sad state that USA has changed that.

Re:Bullocks!

[campuscodi]

It's not a stupid article. It's a stupid lawyer's opinion.

Re:Bullocks!

[Anonymous Coward]

Bullocks, the actress? You mean Bollocks and here is a reference https://en.wikipedia.org/wiki/Never_Mind_the_Bollocks,_Here's_the_Sex_Pistols

Re:Bullocks!

[Anonymous Coward]

Now they can continue to live even in the US.

The damage has already been done. The federal circuit court ruling, namely that APIs are copyrightable, still stands. By arguing fair use Google won the consolation prize in this case, but fair use must be argued each time on a case-by-case basis . It's a fig leaf that offers little protection against the aforementioned court ruling. Google or anybody else could be sued again tomorrow for copyright infringement of a different API in a different instance and they would have to defend fair use all over again each and every time they were sued. Google might be able to afford that, but small companies and open source projects cannot.

Re:Bullocks!

[tricorn]

The precedent is extremely narrow, it only applies to Ninth Circuit cases, and then only when appealed with the Federal Circuit because it also involves patent claims. If you do find yourself in that situation, with a similar case, you might be able to get the copyright and patent issues separated. Winning in the Ninth Circuit on appeal would essentially nullify the CAFC decision.

Re:Bullocks!

[Zontar The Mindless]

Not trying to be a Nazi, just offering some grammar help here:

If Oracle HAD won, a lot of open source projects WOULD HAVE BEEN dead in the USA. In [the] EU they WOULD HAVE LIVED on...

English doesn't really have a subjunctive mood (only a few traces), and our conditional tenses can be a bitch sometimes. (Let this stand as a warning to other languages not to adopt some other language family's verb system wholesale and dump it on top of your own.)

Stallman's defense of Oracle

[Anonymous Coward]

Oracle's loss is a huge threat to free software and the GPL. That's why Stallman has been campaigning on their behalf for years now, hanging out with Ellison on his private catamaran, writing op-ed pieces supporting Oracle's treatment of Java post-Sun acquisition, etc.

I'd post the links, but it's easy enough to find them with Google.

Oracle wants us to have crappy computers.

[headkase]

This is plain double-speak. If Oracle had their way they'd kill GPL software. Innovation revolves around an application programming interface. The API is the "shape" of the program. The code inside the shape is the implementation. The GPL revolves around the implementation and has nothing to say about the shape. If shapes were always copyrightable then that would absolutely kill innovation. All of a sudden if you used someone else's shape in a way they didn't like they could totally shut you down with just the threat of a lawsuit - not everyone has deep pockets to fight that. Copyrighted API's would become just another kind of currency much in the way software patents already are. If you can't beat them with money then beat them, forced licensing, with other kinds of currency. In the Oracle world we wouldn't even enjoy the powerful computers we have today. Decades ago Phoenix clean-room reverse-engineered IBM's BIOS and made the same shape with a different implementation. If that shape had never been open we would have never experienced the rapid advancement of a bazaar that component manufacturers can revolve around. We would have been stuck with IBM's will and computing would have stagnated because they would not necessarily have had an interest in advancing it as much as competition does. At the time IBM's BIOS was reverse-engineered they weren't even the best computers. There were others that were much better like the Commodore Amiga, however, when the ecosystem around an open BIOS happened then the feedback effects from that made it win. Without a doubt.

Re:Oracle wants us to have crappy computers.

[NormalVisual]

. At the time IBM's BIOS was reverse-engineered they weren't even the best computers.

IBM's BIOS wasn't reverse-engineered. The source was available directly from IBM via the Technical Reference Manual. Phoenix and Compaq both merely did a clean-room reimplementation based on looking at the function call interfaces in the actual source code and what that code actually did, then writing a detailed set of requirements for the coders. No reverse-engineering was needed.

Re: Oracle wants us to have crappy computers.

[Bing Tsher E]

IBM's BIOS was not reverse engineered. The commented Assembly Language source code for the BIOS is published in the Technical Reference Manual which anybody could purchase.

Phoenix had to hire programmers to read the IBM source, write a human language specification, then hand that specification over to a seperate team to code their version of BIOS. Anybody on the first team, or anybody who had access to and read the published IBM source code was disqualified to work on the reimplementation.

In other words, IBM published their BIOS in commented human readable source code. The PC was in this regard among others, an open architecture.

Re:Oracle wants us to have crappy computers.

[ShanghaiBill]

The source was available directly from IBM via the Technical Reference Manual.

That brings back some sweet memories. I saved up and bought a TRM for $100, which was a lot of money back then. I remember reading every line of the source code, all in 8086 asm. I figured out plenty of tricks and shortcuts by jumping into the BIOS code instead of going through interrupts, and tweaking where the BIOS stored variables. Good times.

Re:Oracle wants us to have crappy computers.

[fluffernutter]

While reading your comment I decided I like triangles.

Re:Oracle wants us to have crappy computers.

[MightyMartian]

It would be like claiming that open use of the QWERTY keyboard risks destroying the writing of books, because, you know, somehow, free use of QWERTY in some way makes it more likely that people will be able to steal novels.

Re:Oracle wants us to have crappy computers.

[Anonymous Coward]

It's still reverse engineering, known as the "clean room" or "Chinese wall" technique:

http://www.computerworld.com/article/2585652/app-development/reverse-engineering.html

Re:Oracle wants us to have crappy computers.

[cwsumner]

It would be like claiming that open use of the QWERTY keyboard risks destroying the writing of books, because, you know, somehow, free use of QWERTY in some way makes it more likely that people will be able to steal novels.

Good Analogy! 8-)

Parade of horribles

[DRJlaw]

Consumers can expect to find decreasing options to own anything for themselves...

Like commercial software, which you do not own but merely use under license...

...decreasing options to control their data...

Like commercial services, which grant themselves increadibly broad licenses to everything that you post, store, and transmit, usually for far longer than merely the duration that you use the service...

... [and] decreasing options to protect their privacy.

Like commercial software and servicves, which report back telemetry data, raid amazing quantities of your PII for the services' own benefit (well I'll just be uploading that contacts list for you...), and generally function as your own personal stalker for the benefit of the service and many other third parties (dear advertising partner, our user is a 18-34 year old female who is 4 months pregnant and just about to walk past your storefront).

Yeah. Having someone reimplement a GPL-licensed software product by mirroring the APIs is certainly going to lead to things worse than that.

Re:Parade of horribles

[edtice1559]

Consumers can expect to find decreasing options to own anything for themselves...

Like commercial software, which you do not own but merely use under license...

Like commercial software that is now sold under "term licenses" so the minute you stop paying it shuts off.

Cue Cartman's extended laugh.....

[GerryGilmore]

...in the "little people" episode. Totally appropriate response to her op-ed.

We lost!

[guygo]

The sky is falling. Ahhhhhhh! Bunch of self-serving bleep.

parse failure

[Runaway1956]

WTF are you trying to say? Is GPL the worst, or not? If not the worst, then what is worse? Never mind, you're just another fuckweed grousing about open source.

Re:parse failure

[cloud.pt]

at least I'm not a yuppie assuming everyone who doesn't like abusive ways of making money is a weed-using commie. I didn't even come here to defend GPL usage. I came here pointing out this lawyer's argument is totally invalid. She's trying to make sense of nonsense with charismatic speeches like this is the republican presidential race

Re:parse failure

[cloud.pt]

apparently, this was a reply to a -1 comment which was hidden on my UI due to filters. As another comment of mine (on the same level as the one hidden) showed right above, the UI showed as your reply was directed at my comment. Sorry for the very misunderstood reply.

Complete utter nonsense!

[Xtifr]

Before Oracle v. Google, everyone assumed (based on extensive legal precedent) that APIs were not subject to copyright at all. Yet the GPL was just fine. Why would the GPL be threatened all of a sudden just because one more API turned out to be copyable?

The only tangible result of this case has been a very slight strengthening of copyrights, since the appeals court rules that APIs might be copyrightable under certain circumstances. How does strengthening copyright weaken a license that relies on copyright?

This is either monumental stupidity, or outright shilling. Hanlon's razor suggests I ought to go with the former, but I'm going to wait and see.

Re:Complete utter nonsense!

[taharvey]

Can you explain this a little?

As programmer it makes no sense to me. Are you saying that the only code protected by copy-write is the code inside the function curly braces, and not the code outside (i.e. the functions, parameters, returns). This seems like an arbitrarily grey line. What about expressive languages that blur the lines between function declarations and executable code? What if significant code is all written in a macros, where there is no difference? This also dismisses code architecture as uncopywriteable, and yet architecture may be one of the hardest to do right, and thus most valuable things in code.

Re:Complete utter nonsense!

[ragahast]

Are you saying that the only code protected by copy-write is the code inside the function curly braces, and not the code outside (i.e. the functions, parameters, returns).

Even in C, this wouldn't be the case, because some code that actually does stuff is outside functions. But your parenthetical is almost right - the function names, and parameter and return types.

What about expressive languages that blur the lines between function declarations and executable code? What if significant code is all written in a macros, where there is no difference?

There is no difficulty here. The specification of what must be given to the program and what will be returned from the program. Everything else can be subject to copyright.

As programmer it makes no sense to me.

Think about what the world would be like if interfaces had been copyrightable. The precedent comes from Borland v. Lotus, where it was ruled that you can't be granted a monopoly on interfaces like "File > Print." One immediate consequence would have been that every program we use would have to have something different from a set of drop down menus containing the names "File" and "Print," where a user interaction and document input lead to a print command. Of course, this would extend to every way in which we interact with standard desktop applications, but also to how you choose to name your functions and choose their signatures. It already sucks that any cool algorithm you come up with might be subjected to litigation attempts based on patents. But now we might not be able to use the names e.g. of other browsers' non-standard CSS either. We'd have to search through every preexisting program to make sure the function names are unique, etc.

Further, the most critical interfaces in modern computing are those defined by programming languages and operating systems. For many languages, one would need to obtain a license in order to write a compiler/interpreter - so no Octave or Scilab, no OpenJDK, many fewer C, C++ and FORTRAN compilers. No UNIX-like operating systems.

As a programmer, what's really terrifying (and makes no sense), is that my own independent developments, let alone attempts at compatibility, might be restricted by government-enforced monopolies on ideas.

Re:Complete utter nonsense!

[Xtifr]

You asked a mouthful there!

Bottom line: copyright is supposed to protect creative and original works. You cannot copyright a simple list of facts, no matter how much research it took. Languages (computer or ConLan) are generally considered tools for expression; something written in Klingon can be copyrighted, but the Klingon language itself cannot be. A dictionary defining the meanings of Klingon words can be copyrighted, but not the language itself.

Note that no computer language ever has been copyrighted. Implementations can be copyrighted, of course, but the languages themselves are not.

APIs are simply the language you use to access the functionality of a library. The library may be copyrighted, like the dictionary of Klingon, but the interface is simply the words of the language of the library, used for creative expression by the application programmer. This is more-or-less the heart of the matter.

Beyond that, purely functional language (not a functional language, but language which is required for functionality) cannot be copyrighted. If there's only one way to describe something, you can't copyright that, because it's defined by the problem-space, so no creativity is required. In most languages, a simple swap function would probably not be copyrightable, because there's really only one, or maybe two, sane ways to implement that. Writing a word processor, on the other hand, allows all sorts of opportunities for creativity.

(The irony of this last point is that a really badly designed API would be more likely to be copyrightable than a sensible, straightforward one, since the sensible, straightforward one is constrained by the problem-space.)

In the case of macros, the names and arguments of the macros would not* be copyrightable, but the bodies would be (assuming they weren't trivial, like swap).

Anyway, I'm really not the best one to explain all this. If you want a really good explanation, I recommend reading the transcripts of the final judgment in the original suit. Its conclusion may have been overturned by the appeals court, but it clearly lays out the reasons why APIs until then had not been considered copyrightable. It also lists a number of applicable precedents, which are still quite relevant.

* At least, till Oracle v. Google, they would not have been. Now they're merely probably not copyrightable.

Re:Complete utter nonsense!

[david_thornley]

A good API typically requires creativity, and when written it isn't purely functional. Later on, use of the API would be purely functional. I don't think copyright law currently captures the nature of APIs well.

Re:Ars Shouldn't Have Given Her a Forum

[Trongy]

What's telling is that she chose to present her viewpoint on a tech site. If she had conviction, she would present it on a forum with an audience of lawyers where she would face the criticism of her profession.

Wrong!

[Greyfox]

I'm pretty sure one of Oracle's IP lawyers would understand the difference between specification and implementation. To claim this could kill the GPL is at best disingenuous and at worst incompetence. Or perhaps the other way around

Re:Good

[MightyMartian]

Which is why it is at the heart of two of the most successful software projects in history; the GNU toolset and the Linux kernel.

In the real world, not the fantasy land you come from, results count, and the result of the GPL is an ecosystem of kernel modules, libraries and applications that is used on hundreds of millions of systems around the world every day.

Bullshit ...

[CaptainDork]

... consumers don't "own" a fucking thing, ever, anywhere.

Read the goddam EULA.

Re:Bullshit ...

[Xtifr]

Read the goddam EULA

I'll be glad to if you can tell me where it is. The only thing on my system that has an EULA is a game I bought about 15 years ago. (Oh, and flash, but I've currently go that disabled.)

Re:Bullshit ...

[RabidReindeer]

Read the goddam EULA

I'll be glad to if you can tell me where it is. The only thing on my system that has an EULA is a game I bought about 15 years ago. (Oh, and flash, but I've currently go that disabled.)

Where THEY are. Even if you're running an open-source OS like Linux there are EULAs. All over the place. The GPL is a EULA. Virtually every application you installed came with a EULA. Usually the EULA is one of the installed files. Sometimes it's embedded in the code. Sometimes it's on the site that you downloaded from or on the box it came in. Assuming you can find a software product in a physical box anymore. EULA frequently come with your hardware and your ebooks.

We've seen so many EULAs and they've become so long and pointless that we've become blind to them, but they're still there.

Re:Bullshit ...

[Xtifr]

As anonymous coward said, the GPL is not an EULA, it's a distribution license. It's also completely optional. Unless you want to redistribute the code, you can completely ignore the GPL and simply use the copy under normal copyright terms. Therefore, until I actually distribute some of this software myself, there is no license!

Likewise, the BSD and MIT license, which simply say "you can copy/modify this as long as you preserve attributions" and not much else. That's not a user license; that's a distribution license.

Re:Bullshit ...

[currently_awake]

I think it would be an interesting experiment to buy and install software in a way that doesn't trigger the EULA. Not open the shrink wrap (read the disk through the plastic?) not run the install program (make your own installer by dis-assembling theirs), and then exercise the first sale doctrine by selling it publicly.

Re:Bullshit ...

[david_thornley]

A license allows you to do things copyright law wouldn't let you do. The GPL itself states that you don't have to accept it, but it's likely to be the only thing permitting you to copy or redistribute the software. A EULA typically adds restrictions. Reverse engineering is fine under copyright law, but it's often banned by EULAs. If you stick to the provisions of copyright law, you can ignore the regular licenses, but in some jurisdictions, including the US, you can't ignore EULAs.

Farm supplies

[Todd Knarr]

There's a farm supply store near me that'd love a steady supply of the kind of high-quality bovine waste product Ms. Hurst is spewing...

HA HA HA

[no-body]

Isn't this corporates attorney cool-aid barf she gets paid for?
As I get this, API - application interface - like the front end, how to talk to an set of functionality of a software package, not the guts behind doing the actual work, such as in Unix/Xenix/Linux commandline/sh/ksh/csh, library definitions, awk,samba and what else there is on GPL goodies around for decades.
And getting this Oracle case decided in Oracle's greedy bosses sick mind's favor, all this would be in jeopardy.
Quite the opposite of this truth-talk barfing attorney lady.

Unknown concept for those dorks:

https://en.wikipedia.org/wiki/....

Re:Nice Work.

[PopeRatzo]

As usual, great Slashdot "editor" work here

It's the first hot holiday weekend of the summer, so give 'em a break. They've probably been drinking since like 11am.

Comment removed

[account_deleted]

Comment removed based on user account deletion

Oracle just trying to save face

[NimbleSquirrel]

Are we supposed to believe that Oracle really had the best interests of the Open Source community at heart? Are we really supposed to believe that this case was all about Oracle's altruistic intentions instead of a $9Billion payday??

Now that they have lost this particular legal battle, Oracle are just trying to save face in front of the Open Source community. This lawsuit has alienated a massive segment of the developer community. Arguably that alienation began the moment Oracle acquired Sun, and this lawsuit simply confirmed many people's worst fears. It is clear, from comments in the trial, that Oracle only acquired Sun to have total control over Java, and anything related to it. Now that this lawsuit has confirmed that Oracle don't have the control they thought they did, their only option is to try and sway public opinion with the developer community. I don't see how that will start to happen unless Oracle abandon any appeal and let this case rest.

Hurst said that the whole Open Source community is in jeopardy because this will allow anyone to ignore copyright on source code and claim 'fair use'. Sure, there may be a possibility that 'fair use' could be pose a risk to enforcing the GPL, but the precedent in this case is limited because it specifically involved APIs. That hardly means that the GPL is now worthless. What is certain is that all developers everywhere, including the Open Source community, would have been in far greater jeopardy had Oracle been victorius.

If Oracle had been victorius, then Annette Hurst would have been busy firing off dozens of other API copyright lawsuits instead of writing Op-ed pieces on ArsTechnica. (The only thing that surprises me about this article is that ArsTechnica were so willing to publish something from such a clearly biased source.) Given that this was published so quickly after the trial, I find it hard to believe that Hurst penned this in her spare time after the trial as her personal opinion instead of the opinion of her client. To me it just seems like a lame 'Plan B' approach to sway public opinion for her client while they work on an appeal.

To be clear, I don't for one second believe that Hurst and Oracle have the best interests of the Open Source community in mind. I also don't believe that this is just about making money out of Google (although that is the starting point). This is about Oracle trying to regain total control over Java and anything related to it. The are billions of devices and programs that use Java or make use Java APIs (and not just Android devices), so the potential licensing revenue stream would be massive for Oracle. This is about Oracle trying to put an Open Source genie back in the bottle, and represents a far greater threat to the GPL than fair use ever will.

...really?

[karnal]

Software businesses now must accelerate their move to the cloud where everything can be controlled as a service rather than software.

Nope, nope nope nope fuck you in the neck nope.

Total BS; I've done the right thing.

[dltaylor]

It is the INTERFACES that are open, not the implementation.

Many years ago, I had to implement a set of printer drivers but, as usual, the printer codes were proprietary. Using the man pages only, which described the interfaces and data format, I wrote a new subset of the pnm functions for a pnmto program (since I could not locate any LGPL implementation of pnm). I did, in fact, type in all of the characters for the new headers myself, and the text did not match the original headers, except for the function names and parameters.

Re:Total BS; I've done the right thing.

[Z00L00K]

However what Oracle wants is to also close the interfaces in order to ensure a completely walled garden solution.

They will in the coming years look into creating a cloud based service that doesn't have any public interfaces to ensure that they own everything in that garden including your data.

revised headline:

[Black Parrot]

"Oracle Makes Shit Up To Inspire FUD"

Re:Nice Work.

[Hognoxious]

Unless it's a transgender "dude" or "lady"...

[SJW mode] And what's wrong with that? [/]

Butt-Hurt lawyer cries over inevitable defeat.

[Anonymous Coward]

Get over yourself. APIs CANNOT be copyrighted, period. It's the one decision from the EU high court system that makes sense.
Software, the "exact" bits of code can be copyrighted, but not patented.

Re:Nice Work.

[K. S. Kyosuke]

Never made a typo? I see...

Re: Nice Work.

[Anonymous Coward]

Dude, stfu. Mistakes happen, get over it. You sound like a child.

Static creates a copy. COPY right

[raymorris]

I think by "worse" GP means "more clearly brings the result under the license used by the library". Static linking creates a strong case that the resulting executable must abide by the license of the library.

The reason why is because copy right is essentially about the right to create copies. Static linking EMBEDS a -copy- of the library into the executable. Therefore you're clearly making and distributing copies, and must follow the GPL is whichever license applies.

With dynamic linking, you're not copying the full compiled code of the library before the program runs. You're "only" copying the external interface of the library, much like Google copied the external interface of Java. Therefore it MAY be fair use and you MAY not have to follow the license of the library.

I capitalize MAY because this case did not rule that all use of any and all APIs, in any way, is always fair use. The jury found that in this particular case, the way that Android used the Java APIs was fair, after considering statutory criteria such as how it affected the market for Oracle's Java products.

Re:Static creates a copy. COPY right

[phantomfive]

I think by "worse" GP means "more clearly brings the result under the license used by the library". Static linking creates a strong case that the resulting executable must abide by the license of the library.

I'm not sure that's true. Because they're both derivative works.

Re:Static creates a copy. COPY right

[raymorris]

Well a statically compiled binary, which contains a copy of the library, is obviously a derived work, no question about it.

Judge Alsup ruled that using an API dynamically (and even re-implementing it) isn't, because the library's API can't be copyright protected anyway. The appeals court disagreed in this case. So it's debatable so using an API via dynamic linking. It may come down to the specific facts of the case, or the mood of the judge or jury.

On to fair use. A key consideration in fair use is how much of the original work is used. If you copy the whole thing verbatim, it's generally unlikely to be fair use. If you use 1% of the original, it's much more likely to be fair. Static linking copies the entire function verbatim. Dynamic copies only the header line. Not much different than copying the full text of a book versus copying only the chapter titles.

With dynamic linking, there's an argument to be made either way. Different courts may decide different cases of dynamic linking differently. Static linking, verbatim copying of the entire implementation - that's not even arguable, not at all.

Re:Static creates a copy. COPY right

[phantomfive]

Judge Alsup ruled that using an API dynamically (and even re-implementing it) isn't, because the library's API can't be copyright protected anyway.

It doesn't matter if the API is copyright or not, because you are still using all the code from the library: the actual implementation (unless you reimplemented it yourself).

Re:Static creates a copy. COPY right

[jbolden]

What is a derivative work is a question of judgement. There are criteria but it is not so clear cut.

If program X can dynamically link to A or B where A and B have different license holders than X is very unlikely to be considered a derived work. The same way that raw HTML code is not considered a derived work of internet explorer.

Re:Static creates a copy. COPY right

[jbolden]

Using something doesn't invoke copyright, I have to copy it. For example if I buy a book and the license holder doesn't get to decide if I use it for kindling, stop a door jam with it or read it. The question with a derived library is whether an act of copying took place. No copying, no copyright violation. What constitutes copying as contrasted with referencing is the legal issue.

Re:Static creates a copy. COPY right

[jbolden]

Good answer across. Yes you understood my intent here and in your other responses.

Re:Static creates a copy. COPY right

[phantomfive]

If it can dynamically link to either A or B, then the case is no different than if you can statically link to either A or B.

I've talked with lawyers about this very topic, and the method of linking isn't particularly important.

Re:Static creates a copy. COPY right

[jbolden]

The binary is almost always considered a derived work. If you statically link and then distribute that's very different, you are making a copy. If you distribute source and the recipients statically link then that's likely the same thing but may not be.

As for the lawyers, the lawyers may not understand this area of law is rather specialized and very technical. The existing case law is the best source.

Re:Static creates a copy. COPY right

[phantomfive]

As for the lawyers, the lawyers may not understand this area of law is rather specialized and very technical.

Yes, that's why I talked to a lawyer who specializes in software copyright in Silicon Valley. But anyway, appeals to authority are lazy arguments, so I should go into more detail.

If you look at the case law, the courts don't consider the linking method at all, it's irrelevant. The courts use a different method to figure out what is a derivative work, the abstraction filtration comparison test.

Re:Static creates a copy. COPY right

[jbolden]

That's an excellent link. But you are forgetting the absolutely key point. What is the scope of the "program" to which the claim is being made? With static linking it is often the linked version, with dynamic the unlinked.

Well, this lawyer has one part right at least...

[Narcocide]

The very idea of copyrighting software is not internally logically consistent. Free software advocates, pirates, and other computer enthusiasts have been saying this for decades now. Probably this is the cause of Hurst's apparent panic and confusion. He/she/other however clearly does not grasp the fundamental conceptual difference between a API and an implementation. Lawyers and Judges still calling shots about stuff they don't understand. Nothing new to see here.

Re:Nice Work.

[Bite The Pillow]

Much as I love poking at the dashslot noneditors, assuming that Annette is a female name is so 1960. If you had posted support for your side, then just maybe.

An *Oracle* win would have killed off FOSS

[DrJimbo]

The Oracle lawyer has it completely backwards. If APIs could be protected by copyright then FOSS could be easily locked out of making compatible implementations. Oracle is not in this battle to get a few billion dollars from Google. They are in this battle to kill off all independent software development. As bad as software patents are, changing the ground rules so APIs can be protected by copyright would be much much worse.

Re:Nice Work.

[simcop2387]

Oh wow, that's much later than usual. Server maintenance last night?

Stallman might agree with Oracle

[ebcdic]

He argued that writing a program which could be linked with readline violated the GPL even if you didn't include readline, because it was a derivative work.

Re:Stallman might agree with Oracle

[Xtifr]

That's not the interface. That's linking with the actual code. Which can be (and is) copyright.

The reason using readline but not including the readline code is still a copyvio is that the only possible use for such code is to create a program linked with the (actually copyrighted) readline library. And the law takes intent into account. If you intend to give people a copyright-violating program, trickery of making them get the copyright-violating part on their own are considered just that: tricks. That's why NeXT gave up and released the code to their Objective-C back end after the FSF objected to their use of GCC. And that was a standalone binary!

Now, in the case of readline, you could get around the problem (and someone did, in a similar case involving a different library) by creating your own stub library which used the same (not-copyrighted) API. Then (and only then), the use of the API would no longer signal your clear intent to distribute a copyright-violating program.

Re:Stallman might agree with Oracle

[ebcdic]

It's not the only possible use for it. You could, as you mention yourself, create your own compatible library and use that.

Re:Stallman might agree with Oracle

[edtice1559]

Whether this is true or not, it's still somewhat of a non sequitor here. The GPL is a license and it can (and does) impose additional terms beyond copyright if you want to distribute. If I write a routine and publish the source on a web page, the RMS argument would have a tough time standing. However, once I start including a build system or detailed instructions on how to link in the third-party code, at some point I'm really just finding a creative way to distribute the binaries. There may not be a nice legal "bright line" here. That's why RMS avoided the whole thing by making libc LGPL! The one thing about the GPL that is hard to argue is that the license is written in a way that attempts to provide a clear bright line of what can and cannot be done. The intent is also very clear. We only get into these situations when people try to find creative ways to follow the letter of the license, not the spirit. Sadly this happens more often than it should.

Re: Nice Work.

[John Allsup]

To put my grammar nazi hat on, it is not incorrect to refer to a female using 'he', but it is incorrect to refer to a male using 'she'. Women are special in the language because they can have babies whilst males cannot, and this little fact, sexist or otherwise, is critical to keeping the human race going. Something like that.

.c is copyright, .h is not

[John Allsup]

The basic rule should be that interfaces are not copyrightable, but implementations are. If the contents of /usr/include can be treated as not subject to the GPL, that is no problem. If the 'commonly available' bit applies as the op-ed mentions, then would this not apply at least to all 'freeware'? And to the space filler that comes on a new laptop's hard drive? Copyright needs fixing anyway.

Re:You know, she's right.

[goose-incarnated]

What Google's winning this case means is that a proprietary company can now copy portions of GPL'd code under the legal protection of "fair use".

No. It means that they can now copy the function and class declarations, but then again - they always could.

you cant kill the GPL

[FudRucker]

somebody will just fork it first

API is not software

[ajyand]

The author seems to be confused between software and API. Google copied API, not the software. GPL is applicable to software, not the API.

Re:API is not software

[ajyand]

I mean to say, interface is not implementation. Implementation is protected, not the interface (like function names).

TL:DR... she's a frigging idiot...

[advocate_one]

everybody else has already posted why..

Re:Nice Work.

[Z80a]

The wrong only starts when all critics gets deflected using the transhield.
And well, the OP did a great job tossing the whole discussion there, probably being a hired gun to do just that.

Re: Nice Work.

[wonkey_monkey]

To put my grammar nazi hat on, it is not incorrect to refer to a female using 'he', but it is incorrect to refer to a male using 'she'.

It's used when gender is unknown, but no-one uses "he" to refer to known females.

Re:Nice Work.

[lucm]

Unless it's a transgender "dude" or "lady"...

[SJW mode] And what's wrong with that? [/]

The real question should be: why would be the same person called "dude" or "lady" depending on his/her gender? Why not dude/chick or gentleman/lady? Does the sex change imply a transformation of the personality or social standing?

But of course it's socially acceptable as long as the woman version is more respectful; nobody would stand for "gentleman or chick". That kind of inconscious bias is an order of magnitude more sexist than the alleged discrepancy in gender income, which is actually explained by pregnancy leaves and the fact that a larger proportion of women choose a career in a field that is more rewarding socially/emotionally than financially (ex: dedicated special ed teacher vs coke snorting stock market broker).

50 years from now they will make TV shows that take in place in the 2010s and just like we're shocked by the blatant systemic discrimination against women in Mad Men, people from the future will be shocked to see the constant belittlement of males in the name of "equality" or "feminism" in our current society.

Rick said it best

[ilsaloving]

"I'm gonna need you to take your opinions, and shove 'em waaaaay up inside your butthole."

Customer USING, you're not COPYING

[raymorris]

> because you are still using all the code from the library

COPYright essentially controls the right to make copies, not the right to use. If I buy a book, I can't legally make and distribute copies of it; I CAN use it to prop up a wobbly chair, swat a bug, etc. Generally I can USE it without worrying about copyright. It's the making of copies that invokes copyright. Static linking makes a copies.

Further, my distribution of a dynamically linked executable doesn't actually even USE the library. The library is used by the recipient when they RUN the program. Presumably they have the right to use the library.

Again, I'm not saying that no court will ever, under any circumstances, hold any kind of dynamic linking to invoke copyrights. Maybe they will. Maybe. I'm saying that making copies by static linking DEFINITELY invokes copyrights. Dynamic, since it doesn't create a copy, is much less likely to involve COPYrights.

Re:Customer USING, you're not COPYING

[phantomfive]

COPYright essentially controls the right to make copies, not the right to use. If I buy a book, I can't legally make and distribute copies of it;

If you create a derivative work, copyright also controls the right to make copies of that derivative work.

Further, my distribution of a dynamically linked executable doesn't actually even USE the library.

The executable won't run at all without the library being present, it pretty clearly uses it.

Re:Customer USING, you're not COPYING

[samwhite_y]

This is both criticism and compliment to the previous post.

The phrase

"The executable won't run at all without the library being present, it pretty clearly uses it." is wrong. Any reasonably written executable that does dynamic linking can choose or not to choose to load a library. The JDBC (Java API to connect to databases) and Crypto Libs all work on this principle. They use discovery to find the implementations of well-defined APIs.

However, if for a particular executable, the phrase "The executable won't run at all without the library being present", is another test you can make to figure out whether the executable is a derivative work of the GPL library. So following a prior post, the levels of "derivatives" measure would now look like this. The following is listed from "most derived" to "least derived".

Source code co-mingled with source code of GPL library.
Statically linked and essentially only does what is in the GPL library.
Statically linked, but GPL library is only a component of entire solution.
Dynamically linked, but the executable won't function without it.
Dynamically linked, but the executable is still quite usable without it.
Dynamically linked against small simplistic rewrite based on APIs, but can adopt the GPL version if it is present.
Inter process calls to another executable that is distributed under GPL that returns the results (example: proprietary program to MYSQL database).
Inter machine calls to another executable that is distributed under GPL that returns the results.

If I were a judge, I would say that GPL covers "derivedness" down to "Dynamically linked, but the executable won't function without it". That is why I believe the prior comment has merit, because I think it points out the dividing line for GPL coverage. As an aside, note that LGPL only prevents the first item in the list (though in very simplistic abuse cases, it may reach down to the second as well).

Re:Customer USING, you're not COPYING

[phantomfive]

Any reasonably written executable that does dynamic linking can choose or not to choose to load a library.

That's true, you can conditionally open a dll in your program using dlopen(), but it's an edge case.

As I mentioned in another post, the method of linkage doesn't matter, but whether it is a derivative work. There is a fairly complex process the court goes through to determine if something is a derivative work or not (more info here, and links to even more info), but it doesn't take into consideration linkage. Your list of linkage types is not relevant. Even inter-machine calls can make your program a derivative work.

It is true that in the 90s and late 80s when the GPL was born, Richard Stallman taught that linkage was an important aspect, and it seemed reasonable at the time; but since then, copyright law developed in a way he didn't expect.

Re:Customer USING, you're not COPYING

[samwhite_y]

Huh. I have read the GPL 3 (which deserves its long term death compared to GPL 2) and its statement on derivative works. It is true that it does not directly address linkage.

Also, dlopen?? Personally, I believe modern virtual machine languages offer far better options.

But to say that if I use a GPL database that is built into my executable that is somehow identical to my using a GPL database that I call on the network is ridiculous, so I still see linkage as being highly relevant. One of the frustrating things I have with the new GPL is that there are no good sets of examples of what would be considered derivative and not derivative. As a particular case which I know is on the edge, in what scenarios can I invoke conversions of HTML documents to PDF using Ghostscript from a proprietary program? I do this as convenience for an end user as part of much larger program. The idea is that for a particular HTML report page, the user can get a PDF rendering of the HTML page they are currently looking at. It is a grace note in the program, but it is not a "must have" feature.

Here are some options.

I use the Ghostscript library directly. I know that violates the GPL.
I write a library module, which has a method to convert HTML to PDF and use a poor version of the conversion if I cannot find a Ghostscript library on the local machine -- the Ghoscript library is not distributed with my program, but do have a tight integration with it when I successfully load it.
I call command line shell commands to do the conversion after instructing the admin of the install program how to install an executable that does the conversion.
I call a 3rd party service across the network that does the conversion.

Which of these violate the GPL for Ghostscript? If you don't take linkage into account, I suspect that there is no coherent answer to that question.

Re:Customer USING, you're not COPYING

[phantomfive]

That's an interesting example, but I can't help getting the feeling that you haven't read the link I gave you.

Can you copyright a language

[samwhite_y]

The general question is whether you can copyright an API specification. Some have argued that you can copyright an API specification because the layout of a coherent solution to a problem in the API might have some real value in of itself. However, there is a more sophisticated version of this question. Can a language, such as Adobe's Postscript, be covered by copyright? The line between languages and APIs is getting increasingly blurred. If you look at API specifications for some Scala libraries, the library is really just creating a "holistic" extension to the Scala language, not necessarily limiting itself to providing simple APIs. Here is a simple example, I can define an API to add two complex numbers or I can extend the language so that the plus symbol will add two complex numbers. Scala lets you go down that second path a long way, and it is one of Scala's selling points that it can do this.

Given this blurring of the line between API and language, I argue that any answer you might make about APIs should apply equally to programming languages and vice versa.

My understanding is that most believe the programming languages cannot be copyrighted, but this understanding have never truly been tested in the courts. I think Adobe's Postscript has come fairly close to being tested, but Adobe never really pulled the trigger on some of its threatened legal action. However, I am having trouble getting an accurate history of Postscript licensing, so if anybody else has more details, they can certainly add to this post.

Cloud Java = reimplement as Cloud Harmony

[iTrawl]

If Java were 100% closed source PaaS and people reimplemented Harmony to mimic its API 100% in another cloud that isn't owned by Oracle, what would these guys say? Basically, this ruling was one between "you can reimplement other people's APIs on your own time and dime, and that's fine" and "My API! Mine! Go make your own!"

GPL isn't dead. Oracle can go implement a closed source version of the Linux kernel API just fine. Make it a drop-in replacement that is better than the GPL one and you might even make some money out of it. GNU/kOracle is perfectly valid without violating any GPL - especially under this ruling.

Most software ever written

[raymorris]

To cite one example from my own experience, I'm not sure how giving away an open source mapping script which works with whatever image-processing library happens to be installed is "cheating the GPL", but okay.

> But the biggest one is: how will your users get that library? If give them both to your customer

Probably the most obvious example is pretty much every C program ever written, which is maybe half of all software ever written. Every C program uses the standard library. Approximately none of them include libc in the installer, or any script to get it.

Another example would be the approximately 60,000 packages available from any major Linux repository. When you create an RPM or a .deb package, you simply declare which other packages it depends on.

API Re-use

[middlebass]

The first time I re-used an API was in 1968, to implement job accounting on an IBM 360/40 running IBM DOS. Their is a huge amount of prior art to re-using an API. To read accounting info added to Job cards, I had to intercept the job card and process it myself before then handing it off to the next step of processing.

Re:API Re-use

[iTrawl]

They didn't patent API re-use, so prior art doesn't apply here. This lawsuit was about if this common practice was legal under copyright law or everybody just broke the law until now because nobody bothered to challenge the practice. It's like piracy in Eastern Europe: it's common, it's been happening since computers appeared around those parts - just ask Microsoft about how they got their software stronghold there, but that doesn't mean it's actually legal to get your latest Windows and films from torrent sites, even if you pretty print all your covers and fill a shelving unit just like people who buy the stuff.

Oracle is ridiculous

[jonjavajones]

I'm not an expert, but I think oracles continued flailing for a way to charge money for Java, after its long been "free" is killing java more than the "gpl".

Your own installer is a DMCA violation

[tepples]

not run the install program (make your own installer by dis-assembling theirs)

That's illegal under 17 USC 1201(a) and foreign counterparts if the installer is encrypted, as doing so would circumvent a "technical protection measure".

Re:Nice Work.

[silentcoder]

>The wrong only starts when all critics gets deflected using the transhield.

Why is that wrong exactly ? There are some things which should not be open to criticism, specifically innate characteristics which a person cannot change about themselves such as their race, parentage, herritage, sexual orientation or gender. Since these things cannot be changed, criticising them cannot possibly serve any useful purpose. I wouldn't suggest the government actually censor such speech - but the world tell you to get your head out of your ass if you express it is definitely fair game.

Stop extrapolating!

[cloud.pt]

businesses now must accelerate their move to the cloud where everything can be controlled (...). Consumers can expect to find decreasing options to own anything for themselves, decreasing options to control their data, decreasing options to protect their privacy.

No! No! And no!

On the cloud, not everything can be controlled. Especially not related with fair use vs commercial use! Are you telling me the fact I am using a cloud service during work hours on a work IP is gonna grant a SaaS provider that I am using the service for commercial reasons? Can't I edit a private document in Word or check my free Outlook email on my work PCs now, even if my work is on a BYOD policy? WTF! This is bollocks right there.

And it continues being bollocks when you tell me I don't own a service I actually purchased for the web! When people buy an online game, and online text/presentation/spreadsheet editor (e.g. Office 365), there is a license that states very well that ownership lasts as long as the service can be provided by the provider, and/or has a fixed, unquestionable time frame such as a 1 year license! Despite this being time-limited ownership, IT IS OWNERSHIP! Stop arguing that people have to be bound to a physical object (a CD/DVD) that has to work to full extent OFFLINE for it to be called ownserhip! Even Oracle knows that and it uses SaaS for most of its core business, and it's not even a recent practice - they been doing this for the last 2 decades! (even when they didn't use cloud services, you pretty much had to have Oracle support for maintaining anything on production every N-months or so!). Want me to blow your mind? Time-shares are ownership. From wikipedia: "A timeshare (sometimes called vacation ownership) is a property with a particular form of ownership or use rights." BAM!

Finally, stop arguing about data control and user privacy like you invented the thing! You're a lawyer lady, you know there's no such thing as privacy in a country with the Patriot Act and the Snowden leaks. You just used widespread, catch-all internet and policy issues to justify a circumvention of your (clients') flawed GPL licencing, and by flawed I mean that Oracle made it to protect a solid royalty revenue, and then it simply didn't! Don't call out civilization's evolutionary problems to fix THAT mess! Tell your client to learn from their mistakes. Or simply to drop the patent-troll scheme like most commercial companies like to do now: buy a FOSS company in order to nullify competition by making it non-FOSS. Because unlike that "loss of privacy induced by cloud/SaaS gold rush induced by dual-licensing issues", that is an ACTUAL PROBLEM!

Re:Good

[cloud.pt]

I'm assuming this was directed at the "parse failure" comment right?

Heh

[Runaway1956]

Cool. I didn't understand you, TBH, and I was just letting is slide as something stupid. But, now I see what happened - - - I browse everything. Sometimes those AC's write something worth reading, but you gotta wade through a lot of crap to get anything "good". And, TBH, I shouldn't have fed the troll to whom I replied.