Oracle asks Supremes to snub Google's Java API copyright protest – and have a nice cuppa tea, instead Oracle has filed its formal response to Google's plea in January for the US Supreme Court to review the two tech goliaths' long-running war over Java, Android, APIs, and copyright. On Wednesday, Big Red made its pitch for why America's highest court should put an end to the legal conflict once and for all, by refusing to hear …
Pucks on both their houses. So, when google copied the 11,000 lines of code, did they really think that Oracle would not sue? Or was it more of a gamble based on the possibility of the software being developed becoming abandonware, one of google's most cherished modi. We could clean-room this code but it would take months and by then the project might be abandoned, kind-of-reasoning? The existence of copyright didn't stop Google Books, or whatever it was/is called.
Or is google's argument one that has carried the day in other court cases, involving other parties?
reminds me of a previous case...
SCO vs IBM
and the kerfuffle over the file "errno.h"
Not much has changed then.
If Oracle wins this I fully expect the SCO case to be re-opened.
To Google and ORacle, [see Icon]
And the same goes to the EU for Article 13. Just wait for the lawsuits to start flying just because you provided a hyperlink to something that was copyrighted.
Google didn't actually copy any code at all. In fact in all the code compared, they found only nine lines of code which were the same. Google implemented the interfaces for the Java API to remain compatible with the java infrastructure. Oracle never claimed Google copied the code. Oracle's issue was that the organization and interfaces were subject to copyright and that is what the court finally decided.
Oracles's claim is not without merit but it breaks commonly accepted rules about software development which have always allowed people to "stand on the shoulders" of others and develop new technologies. Even Java itself uses a lot of C-derived interfaces. And that's the problem I have with Oracle's side: you can't have it both ways. Java was built on the ideas from others (byte code, programming interfaces, etc) so to then prevent others from doing the same is hypocritical and very dangerous. If I write a routine to print a PDF with the name "printPDF" with a PDF file as a parameter and the interface matches something Oracle claims it owns then they can sue me. So yeah, this will completely stifle software development.
"You haven't paid attention much to this saga, have you?"
Exactly. I was relying on the story to get the facts, but the story did not make sense to me. I am happy that the Comments now give a more realistic view of the matter, regardless of the 33 (and counting!) downvotes. Thanks to those more knowledgeable for their patient expositions.
I don't think they directly copied Oracle code. The original Android SDK was based on the Apache Harmony project: https://en.wikipedia.org/wiki/Apache_Harmony#Use_in_Android_SDK
That project was created with the blessing of Sun Microsystems: https://www.theinquirer.net/inquirer/news/1019575/apache-to-cleanroom-java-with-suns-blessing
From what I understand, Harmony was developed using "clean room" principles. They had one team of developers use the existing code to write the specs, this is basically the function names and parameters and documenting what the result should be. Then another team was handed this spec and implemented it all using totally new code. This spec is the only part that was copied, everything else was implemented from scratch.
It sucks that this happened with the blessing of Sun back in the day and they were since bought by Oracle. Oracle just wants money and saw this as a way to squeeze a lot out of Google. They don't care about anything else.
This part of the story
Google, which copied about 11,000 lines of Oracle-owned code as a starting point to provide the base Java API for its mobile operating system,
sounds nothing like the clean-room, reverse engineering with consent, and different parties story that you describe.
When code is counted in "lines" that to me means commented or documented code. There is an obvious explanation for the documentation to be visible: it is that its owners, Sun, might have opened their books during the Apache Harmony project. Although I don't know why google wouldn't instead simply use the documentation generated by the Apache Harmony project.
But maybe there's some purpose to be garnered from my post. I am not a software developer. As a lay person I was aware that clean room techniques had been used to create functionally equivalent software. Even though clean rooms were not mentioned in the story, I thought that clean-rooming might create software that does not infringe copyright. I don't think I'm the only one in that boat.
Those "11,000 lines" could well be just the header file's function / class definitions, you know the stuff that has to be exactly the same for an implementation to be compatible?
A crude 'wc -l /usr/include/*.h' on my box reveals 52,771 lines of text for the standard C header files, easily 11k lines of "code" by a certain definition.
There is an obvious explanation for the documentation to be visible: it is that its owners, Sun, might have opened their books during the Apache Harmony project.
Well, the thing with copyright is that there are rules and if you make those files available for one project, you have essentially made them available for everyone. Not that I think that the issue will be decided on this, but worth noting. Google certainly hasn't covered itself with glory either. It could probably have got official blessing from Sun for its approach at the time, but seemed determined to want to avoid this.
The potential fallout for making a public API entirely subject to copyright (you get to decide who can implement it and how) is huge and something that Oracle should itself be worried about: there are bound to be parts of its database code built in a similar way.
Oracle didn't invent SQL for starters, has C code for which they include the standard headers, and numerous other things that if they win, they could be sued for doing - much more of the same thing - that IBM and others have done.
Windows.h is kinda large all by itself.
Oh, they support POSIX too. So by some of the liar interpretations of open source such as the one SCO used, they need to make 100% of their other code - including this - open source!
See how disruptive this can become with lawyers involved? Everyone needs to use headers from other code to interoperate with it and be useful on any opsys whatever. If an opsys wants to be usable without needing completely custom application code for every little thing, it itself needs to comply with some standards.
Copyrighting a standard is horribly bad ju ju. Haven't we all railed about similar issues with hardware standards that turned out to have submarine patents on them and companies coming out of the woodwork demanding excessive royalties on things like dram or radio parts? Similar issues here.
It'll be interesting to see whether the new standard that a copyright needs to be registered before you can lob a sueball and whether Oracle had done so, ignoring whether Sun had done so. Lots of conditionals there. You couldn't get me anywhere in the same country doing interoperable systems development right now and if Oracle wins, never again.
...with a fairly liberal license IIRC.
After buying Sun, Oracle are trying to close Java down again, but it's a little late for that. I'm thinking that Google should just pay off Oracle, so that Oracle can continue to do nothing of value, and slowly die.
Why hasn't Java been forked into a non-Oracle consortium that will actually do something with it?
Actually they hadn't. Confusingly there were _three_ variants of Java at the time that Android was being developed; JavaME (for mobile), JavaSE (for desktop), JavaEE (for 'enterprise' or web servers). Sun had only open sourced JavaSE at that time and was still selling JavaME licenses to various device makers. There was an explicit restriction of use on the test suite to _certify_ a fork of JavaSE that prohibited it from being used in the mobile space where it would have competed with Sun's mobile offering. This is why Apache Harmony was never officially blessed by Sun. The license that Harmony was created under didn't allow for restrictions on use. It was JavaSE in everything but name, hence the 'Harmony' moniker. Without passing the still closed test suite forks couldn't legally call themselves Java.
Google wanted to use Java as the basis for the user land applications so that it could leverage the large number of existing Java developers. Unfortunately, JavaME was/is rubbish. Sun wanted Google to license JavaME for Android, but even though Google could have afforded the license, it wasn't fit for purpose. Google wanted to start with JavaSE and create an 'Android' version of JavaSE for Android. So they started with Apache Harmony and abiding by Harmony's license (perfectly legal) made the changes they felt were needed to get it to run better on a smartphone. They didn't NEED to do a clean room implementation, the Apache Harmony developers had already done that. Then Google wrote their own virtual machine Dalvik (as opposed to Sun's jvm) that runs it's own compiled version of the android 'Java' program. .dex (as opposed to Sun's .jar). APT has recently replaced Dalvik and Kotlin is rapidly replacing Java as the preferred language for development of android applications.
So as you can see, it's a rather confused mess and Sun hadn't 'actually' _open_sourced_ Java at this point. Sun had opened the JavaSE specs, but not the JVM or the test suite needed to certify an alternate implementation. JavaME and JavaEE were still closed and Sun was still getting licensing revenue (although rapidly dwindling by this point) from handheld manufacturers licensing JavaME. Oracle, bought Sun and sued Google for basically using Apache Harmony in a manner consistent with it's license and against what everyone else in the industry thought was propper. The judge took the time to learn to program and _correctly_ declared that you couldn't copyright API's etc. The Fed. Appeals Court, that has never seen a patent or copyright that should be granted/expanded regardless of the underlying law, overturned it. It went back and Google's use was ruled 'Fair Use'. Appealed back to the same court and again overturned. The only result the appeals court would allow was Copyright good, Oracle gets the money.
Hopefully the Supreme's correct this.
My understanding is that Google routinely tries to patent such stuff largely to prevent patent trolls from having the same idea, because the US patent system is such a mess. The On8 and On9 codecs were done with this in mind and AV1 has been done in exactly the same way. Don't know the details on this but I'd be surprised if it wasn't another defensive patent.
OK I make a widget that does the same thing as the widget you invented, i name that widget the same generic thing as yours, but my widget works completley different to your widget but achieves the same goal, i invented how my widget works, and you try to sue me because i copied your generic name and fuctions.
This is effectivley the case.
This is ineffect Hoover suing Dyson for making a vacuum cleaner.
All this is really out of my field of expertise and I'd appreciate an analogy to understand what the case is about. The way that @Enviable one describes it could still breach copyright in the following manner:
I write a story who's purpose is to entertain with tales of, say, young SW developers who get the better of their evil PM overlords. I deliver the story using a paper book - the widget.
Someone invents a different widget which can also deliver tales but it does so via a screen on a battery-driven plastic widget. They take my entertaining story of SW developers who get the better of their evil PM overlords, put it in their widget and sell it but seem to claim that they've not infringed my copyright because their widget is clearly different from my widget.
I'm not trying to be contrary here, I'm just trying to understand the case 'cos I don't do any tech work with APIs. I guess I'm struggling to understand the difference between what, in the Google/Oracle case, constitutes:
- framework (you can't copyright the individual words in the dictionary nor the rules of grammar and syntax that surround them),
- content (the words from the dictionary in the order that I wrote them),
- the way the content is delivered (book, kindle, podcast) .
Anyone feel up to a short explanation, or is that what the court case is all about?
Sure, I'll give it a whack....
In order for programs to be inter-operable they have to have functions named the same that take the same parameters and return the same type of result. Take this example:
There is a library called Math and in the math library there is a collection of math related functions; add, subtract, multiply, divide.
A simple function to add two numbers together;
int add(int first, int second)
This mean the function name is "add" it takes two parameters that must be integers and it returns an integer.
To use it you would call "int X = add(2, 3)" and the variable X should contain the number 5.
For example Sun/Oracle could have written the function:
int add(int first, int second){
return first + second;
}
While Google could have written the function:
int add( int begin, int other){
int num_1 = begin;
int num_2 = other;
int answer = num_1 + num_2;
return answer;
}
What Google is accused of copying ISN'T any (well there was those 9 lines) actual SOURCE CODE (the stuff between the {} ). It was accused of copying the "int add(int, int)" and the Math.add(), Math.subtract(), ... arrangement. It is EXACTLY the things that Google is accused of copying that allow one program to be inter-operable with another. Up to this point is was always believed to be not protectable by copyright.
This was widely believed to have been decided in Baker v. Selden, a US Supreme Court case from 1879 ( https://www.law.cornell.edu/supremecourt/text/101/99 ) that defined the idea/expression dichotomy in copyright law. Unfortunately for all of us not named Oracle, Oracle threw a patent into the lawsuit and so it was appealed to the Court of Appeals for the Federal Circuit which has an absolutely horrible track record twisting the law to favor patenting everything under the sun. Recently it has had a number of cases overturned by the Supreme Court.
I hope this becomes another such case where the Supreme Court overturns the Appellate Court and rules that API's are unprotectable by copyright once and for all.
I hope that helps.
Or similar to a recipe. Oracle made a cake. Google also made a cake. Google know it has to be sweet, it has to be edible, but they decided to make it their way.
Now, Oracle has tried to say "we own the copyright on sweet edible cakes", because Google "copied our recipe". Now, some parts of a cake are copyrightable, such as a name "Kipling's Cakes", or an actual recipe *book*. But the recipe may not be, because it's too simple a functional action, or protected under free expression (or something else).
So here we have the question, is it reasonable for Google to put some flour and some sugar in a cake? Because someone, anyone, owning that "copyright" would be insane. So Oracle claiming ownership of APIs or names or maths in code is problematic. Even if they show it exactly the same.
Google went to the trouble to try and not copy the "book" but only make sure the end results were cakes, so people could easily cook with them. (Another example is a dictionary. Some parts can be owned, other parts are public domain) :)
After all, SCO also claimed copyright violations that centered around the header files of kernel functions.
So Google using the API defintions of Java to cleanly re-implement the bodies of internal Java functions ist to me ( programmer, not lawyer) exactly the same.
Google used the clean-room-developed Apache Harmony project for Android, which yes, used the same API call structure. They re-wrote all the content of those APIs to match up, but not using any of the actual Sun code. This is how reverse-engineering works (see BIOS, Linux, ... for examples).
If I took, say, one of the Harry Potter books, grabbed up the chapter titles BUT wrote new material under each, and put it out under a new title, would Rowling have any grounds for infringement? Doubtful.
Oracle is widely believed to have bought up Sun specifically to monetize Java (initially they'd only wanted the software side of the business and not SPARC; McNealy insisted on the whole bundle). Then Android went and took over the mobile world and Larry got peeved no one wanted to pony up for Java ME, so he unleashed the lawyers. Until Oracle threw in the patent issue, which sent it to the patent-friendly appeals court, they'd been losing consistently. Now Google is trying to hold onto that by begging the Supremes to ignore common engineering practice and make Google pay out.
Google did a clean room development of something against the specifications and API, which is almost like designing a new car from a dealership brochure.
In other words, Google did not really benefit from the engineering behind Java as it redeveloped what's behind the APIs from scratch, so I don't see why Oracle should get the big bucks.
I think that Judge Alsop (2012, Oracle vs. Google) had the keen sense to see where it was headed, and has prudently included findings of fact and other supporting cliff notes to insure that these Nine Seniles were able to have a carefully measured bullshit calibration routine to correctly call out both parties if they cross these lines.
So.. when IBM published Eclipse as a Java frontend, why didn't Oracle go after them? Oh, that's right - it was done as a nonprofit so there were no deep pockets to mine. Also, IBM had a mobile Java client but this somehow doesn't interest Oracle. Bottom line, Oracle is simply a shell company for legal actions against customers and competitors. If you think they're anything else, you haven't been paying attention.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
