Oracle v Google round-up: The show so far
What’s going on here? Who’s winning?
After years of waiting for the contenders to open fire, the Oracle-Google shooting match is now on, and the bullets are pretty expensive. The opening salvos have landed, so let's take stock.
It’s worth keeping at the back of your mind how strange it is to be here at all. Judge Alsup has spent three years trying to persuade both parties to settle privately – they’re haggling over the odd billion. But he hadn’t counted on the obstinacy of the Two Larrys. So to the case.
Sun wanted to pull off the hitherto impossible goal of keeping its Java platform open enough to achieve widespread industry adoption, but closed enough so it could retain control to avoid fragmentation. Opening your code is easy, and controlling your proprietary code is easy, but Sun wanted a unique middle way. So the procedures Sun wrapped around Java are pretty complex – and they have changed over the years. But Sun achieved what it had pretty much wanted: extensive industry use of Java, and absence of fragmentation.
That was, until Android.
In a nutshell, Google admits copying Sun Java code into Android. Google maintains that the code it copied didn’t require a licence. Google also argues that the Android platform, in which Java language code is written against Java classes (before being crunched into something that Google’s VM can run) doesn’t need a Java licence.
Oracle's case against Google is laid out in a 90-page slide presentation released at the start of the trial this week.
Sun launched Java in 1995 and Oracle gobbled up Sun in 2009. The hostilities began that year – but from new evidence we can see they were always simmering under surface. Android was founded in late 2003 and snapped up by Google two years later. That year, in 2005 [p38 of the slides] Android founder and chief Andy Rubin wrote to Larry Page:
Google wanted to open-source Android, but it knew Sun was uneasy about that. In October 2005, Rubin suggested to Larry Page that Google ought to pay Sun for a Java licence, and swallow the cost. According to Oracle’s narrative, negotiations were dragging on into 2006, and Google should have at that point contemplated a non-Java option – an avenue quashed by Google manager Brian Swetland in August 2006. By 2007, Android’s attitude could be summed up as: ‘Hey, what the hell’.
I don’t see how we can work together and not have it revert to arguments of control. I’m done with Sun (tail between my legs, you were right). They won’t be happy when we release our stuff, but now we have a huge alignment with industry, and they are just beginning.
In 2007, Rubin noted how cunning Sun had spun the law around Java. The awareness that Google was striking out into new and dangerous territory was evident in 2008: taking a license was “now water under the bridge” and excerpts from emails show engineers and management coyly avoiding the subject in discussions.
“Please don’t demonstrate to any Sun employees or lawyers,” Rubin warned an engineer in 2008, as he prepared to take Android on the road. (As an aside, here’s the engineer’s blog from his first fortnight on the job at Google: “As with most companies that rely on their IP, you learn very quickly that Google has a lot to lose from loose lips, and I don't want to see that happen.”).
The “What the hell” strategy is in Google’s DNA: there are echoes of the Google Book project and Google’s acquisition of YouTube. In each case it knew it was potentially in the wrong, and acquiring huge liabilities. But it pressed on.
Next page: Enter Schmidt
Google didn't need a license & others have taken one either.
So much wrong in so few words.....
There is _no_ license required to build/use Java the language, nor Java the JVM. You _do_ have to acquire a license to the test suite (TCK) if you want to call your result 'Java' or Java compatible.
It's in the TCK that Sun (now Oracle) hid such nastiness as the dreaded Field Of Use (FOU) restrictions that among other things prevents you from running what you've built in the mobile space.
The Apache Foundation never accepted the FOU for the Harmony project. They never took a license from Sun, didn't need to. Still doesn't need to.
HP doesn't have a license for their ChaiVM or their MicrochaiVM implementations.
Google wrote their own VM Dalvik. It's _not_ a JVM. It can't run Java programs. Google wanted a programming language that many programmers already knew to write code for Android. The apparently went with Java_the_language. To be compatible with Java_the_language they based the language on Java as documented by the freely available API's. They started with the Apache licensed Harmony project and adjusted it to create code that they felt was more appropriate to mobile development and compiled to code that would run under their Dalvik VM. Dalvik byecode won't run under a Java VM and Java bytecode won't run under the Dalvik VM. The only part of Java that android uses is Java_the_language.
Most people, I would have said all but obviously Oracle doesn't agree, agree that you can't copyright a language. Courts in the US have already ruled that having a copyright on a system does _not_ give you any rights on a subsequent implementation of that system.[Baker v. Selden, 101 US 99 (1879)]. The only parts that are identical are those _required_ to be identical for compatibility. (Yes, I realize that there were a couple of cases of literal copying, but that's a different issue that the one here. Google's already removed them and they may face a small fine for that mistake.)
Oracle's stance is that since they literally own the copyright to the book the defines the Java API's that you need a license to implement any system that uses them. Unfortunately the courts have never actually ruled on the applicability of Copyright to API's. Up until this point no one ever thought that they needed to. It makes as much sense as having to get a license from Webster to use the English language simply due to the fact that they hold the copyright to an English dictionary.
Is a shame that Oracle's sunk so far.....
Strange. Read the dissection at Groklaw and you get a totally different perspective.
The general view there is that this is largely about whether or not Google need a licence to use the JAVA library API, so this largely rests on whether or not an API can be covered by "copyright" as a creative work. Just because an ex-SUN engineer thinks that Google DO need to acquire a licence is entirely irrelevant: he is not a lawyer and probably doesn't know what he is talking about. The context of that statement is likely to be risk aversion and playing "fair" to his prior employer.
Most of the examples of supposed "copying" are debatable, although not necessarily without merit.
Re: Again : Mueller
No it doesn't make him corrupt but it puts a huge slant on anything he says - something that Andrew fails to point out in his article
"As Florian Mueller points out in his analysis, Schwartz’s comments cut little ice with the Judge" has a very different meaning if he's being paid by Oracle - "As an Oracle paid consultant points out in his analysis, Schwartz’s comments cut little ice with the Judge"
It's a big difference.