Oracle CEO Larry Ellison: I don't know if Java is free | Hacker News Reader

Oracle CEO Larry Ellison: I don't know if Java is free

news.cnet.com

45 pointsdonretag57 comments

While I certainly prefer iOS to Android ... it's really unfortunate that such a situation has emerged. I think the Android platform has only done good for our planet. Developers have more jobs, more interesting apps are being created, more devices are being sold, there are more powerful and less expensive devices in the smartphone world that are becoming popular in emerging markets and third world countries (don't quote me, but a nearly free Android handset in Nigeria?) Mobile is a paradigm shift. People are more and more connected while away from their homes and desks. Payments are going mobile. Also, Android is the biggest competitor to Apple. I'm certainly an Apple guy in all aspects, but I do appreciate competition in the market to keep everyone working hard and innovating.

Sucks that Oracle will not consider any of that. That no terms could be reached out of simple goodheartedness. Oracle isn't losing money because of Android as far as I am concerned. There might be profit lost due to licensing, but it just feels wrong to me.

> Oracle isn't losing money because of Android as far as I am concerned. There might be profit lost due to licensing

From an Oracle shareholder's point of view (which I'm not), lost licensing fees == losing money.

Aren't they losing money every time someone downloads desktop Java SE for free then?

Their argument (which I disagree with) is that creating Java applications is far different than creating a JVM and class libraries. The latter must be licensed. So in their view, driving developer adoption of their language increases the potential for licensing the implementation specification.

But let's be honest, OpenJDK is gonna be all there is (sadly). I liked the direction Harmony went, and it was very strict about the legality of what it developed and used.

> OpenJDK is gonna be all there is (sadly)

Not sad at all. A hostile move by Oracle may be all that's needed to drive a lot more effort into improving it. And the best feature is that it's absolutely Oracle-proof.

No, it their product to distribute as they please. IBM and others have paid for theirs so the thinking is Google should pay too. In particular, the licensing on the phone / device edition has always been different.

By scaring everybody away from the Java ecosystem, I believe they'll lose a lot more money on the long run.

Sun tried desperately to monetize Java and failed. There's really not much money there, this case is their best hope to see some revenue.

Yes, This is exactly what everybody fails to notice.

Everybody please note that Oracle is in the business to make money. And there is nothing wrong with that. They can't just sit on the sidelines and develop java watching other use it to make money while they themselves don't.

From Oracle and their shareholders perspective money that they can't make from licensing fees is lost money.

No company can justify spending ridiculous money on projects that are going to give them nothing in return.

Well... They can always try that old business model - to make things people actually want.

To be fair, Java probably brings a lot of business to Oracle. There are a lot of Java applications running on Oracle app servers storing data on Oracle database servers and interfaced to Oracle applications and running on Sun^H^H^HOracle servers. It's just that HNizens live in a much more interesting ecosystem than the average corporate Java programmer who has to wear a tie to the office.

Also, Sun did spend ridiculous amounts of money on Java - mostly to level the field against Windows, as Java makes apps easily move from Windows to Unix and vice versa - and made easier to sell SPARC servers that don't run Windows but can run Java applications that also run on Windows. And do so with performance and stability no Windows box could touch. And most can't, even today.

Remember: Sun was a hardware company. A confused one, perhaps, but hardware was their core.

Oracle, OTOH, invested (or burned) money fully knowing what to expect in return. They can't complain Java does not bring a lot of direct revenue because it was never intended to. And since Java runs very well on x86 hardware running Linux (thanks in part to Oracle), they can't even claim they expected to sell Sun hardware. Unless they had been smoking stuff they shouldn't mention in court.

Well this is almost similar to IBM saying COBOL being strategically important to their Mainframe sales. Although Java does drive Oracles sales indirectly, Oracle using Java as a bait for their business reasons will do good for Oracle but bad for Java on the longer run.

At the maximum what this would mean is Oracle will evolve Java in a way most suitable to their business needs.

For an outsider COBOL might be dead, but in BigCorp's I've seen IBM having regular COBOL conferences and people do attend them. But ask any COBOL only programmer and he will tell you how limited audience his resume gets because no 'state of the art' work as known to the rest of the world happens in COBOL.

OTOH Java is already the language of the corporates and the darling of layers bureaucratic management layers.

That's definitely not the kind of language I want to be associated with. I don't want to be in the next set of COBOL programmers.

Sometimes I get very unpopular for this, but I can't stop myself from saying "Java is the COBOL of the 21st century".

Is Oracle really scaring their customers (read: enterprise)? I really don't think they care about people on this board or twitter. They earn a lot of money with their current model which involves payments from companies. From a company point of view, Oracle is a lot more stable than Sun ever was.

>> On a personal note, I really wish both Oracle and Google had handled this quite a bit differently. Google probably should have gone with Python when it looked like Java was going to cause trouble. Sun / Oracle really needed to have a decent smart (not feature) phone spec for Java / JVM with a small licensing fee. Java seems cursed with anything that needs an actual UI, something always goes wrong.

Enterprises move very slowly, but, if Oracle starts trying to aggressively monetize Java, future deployments may end up using alternative technologies.

In Google's defense, when they started Android Java belonged to Sun. Sun never was as malignant as Oracle.

In retrospect, they should have acquired Sun (maybe Larry and Sergei would have to pool their credit cards). That Sun ended up being part of Oracle is a larger tragedy than DEC ending up being acquired by Compaq.

Really? I would have thought that in retrospect it's clear that Google should not have bought Sun. If they lose in the courts, they're still rather unlikely to end up paying $7 billion in damages...

Really Google wouldn't have done anything with Sun except for the patents. No use for the hardware or services business, no use for Solaris, limited use for controlling Java, etc.

There is a huge value in preventing year-long problems and just moving on.

Google is in the business of making people use the web. More Solaris servers, more Java applications (I have some issues with that, but let's play along this line) can only be a good thing for them.

Google either "stole" from Oracle or it didn't, what you say doesn't matter much. To play your game, the world might benefit if someone took Google's money (entire $45 billion pile) and invested in vaccines or another cause to help the poor. But that's their money, we don't get to decide for them. The same principle should apply for Oracle or Apple.

> don't quote me, but a nearly free Android handset in Nigeria?

It's so horrible to say it, but someone has to. Not until Android 4.1.9.

So the message is pretty clear. No one should ever again build anything on top of Java.

The lawsuit is not about building on top of Java; it's about building JVMs. Sun said all along that if you build a JVM you need to be compatible and in some cases you need to pay.

And who in their right mind would want to build software on top of the JVM if what you say is true?

Everyone? People generally like JVMs to be compatible. Since you're not making a specific argument I don't really know how to respond.

What you're implying is that the JVM is a proprietary platform. I think most people have treated the JVM as an open platform all these years (or they trusted Sun to effectively treat it as such), and I bet many of them wouldn't have used it if they'd thought it was a proprietary platform where the only allowed implementations have to be approved by Oracle.

You seem to be implying that it's no big deal if it is proprietary. I disagree.

You could argue that enforcing a TCK before you can legally call your product a Java Virtual Machine helps ensure consistency (whether or not the JVM you write is open source). The real problem is the FOU restrictions in the terms for the TCK to be called a "Java" implementation.

I haven't seen this mentioned yet, but Sun in the past had sued Microsoft due to Microsoft's own JVM implementation, which they asserted was not fully compatible (see http://en.wikipedia.org/wiki/Microsoft_Java_Virtual_Machine). Microsoft eventually settled and removed its JVM. This case seems somewhat similar.

This case is not similar at all.

Sun's objection to Microsoft's JVM was not that they implemented it, but that they implemented something which was clearly not Java (due to incompatibilities/incompleteness/etc.), but which Microsoft called "Java" anyway. At the time, Sun made no claim whatsoever (that I'm aware of) that third parties could not legally implement JVMs.

Oracle is now making the claim that third parties cannot implement their own JVMs without paying Oracle for a license, on grounds that copyright to a language or API specification extends to the right to forbid implementations of those specifications.

Except when they released that JVM source under the GPL, which quite clearly disallows such extra restrictions. Except that of course Dalvik isn't derived from that source anyway.

Which is why the Oracle position is so strange and precarious: they're basically arguing that GPL users of the JDK source have a free license to do things (i.e. make incompatible versions) that people who didn't accept any license to any code ... don't. It's weird. What parts of that clear license to the public are severable from the source distribution?

>they're basically arguing that GPL users of the JDK source have a free license to do things that people who didn't accept any license to any code ... don't.

Well, that's what a license gives you: the right to use something in ways people without the license can't.

Dalvik isn't distributed under the GPL, so I don't think any rights derived from the GPL distribution of Java can be used.

But what if it were? What if they re-release Dalvik under the GPL and admit to Oracle, the court and the world that Dalvik is clearly a derivative of the GPL sources from Sun. And then the case gets thrown out, because of the "extra limitations" clause in the license.

See why this is so weird? Oracle is arguing something that if taken to its logical conclusion would invalidate the case. So they can't argue it too far: Dalvik's success (in the market sense) has to be derived from Oracle's IP, but not so much to as to be a derived work (in the copyright sense) of the Java code. That's an awfully fine line.

Oracle has certainly disgraced itself with this case, but as the article makes plain on the first page there is a clear legal basis for assuming that Java development is safe from Oracle's interference in almost all use cases.

Bryan Cantrill (formerly Sun, then worked for Oracle, now working elsewhere) talks about Oracle in his interesting talk here: http://smartos.org/2011/12/15/fork-yeah-the-rise-and-develop...

He compares Oracle to a lawnmower - if you stick your hand in a mower, it will cut your hand off.

Oracle is simple in the way that a mower is simple - a mower cuts what is put in front of the blade. Oracle wants money from whoever they can get it from.

What's the worse case scenario in this dispute? I'm sure Oracle just wants money, but could they legally prevent Google from using Java entirely? I can hardly imagine rebuilding the Android ecosystem on another language.

From my superficial understanding, the direct result of this lawsuit affects only Google and Oracle and is only over money. Android will keep using Java (perhaps with a fee of some sort). There may be repercussions from the decision in other ways, but if you're an Android dev I don't think you need to worry.

The worst case scenario is that the court finds, and it is upheld on appeal, that you can copyright the names of API calls, the number of arguments to those API calls, and the types associated with those arguments.

In that scenario Novell would own the copyrights to things like read(int fd, char *buf, int size); and could sue you if you used them without compensation to Novell.

Now that would be hugely disruptive and counter to a lot of existing case law in adjacent areas, perhaps the most directly applicable are Memorex vs IBM in the 70's with regard to making 'plug compatible' storage devices. Memorex made a compatible "DASD" (Direct Access Storage Devices, otherwise known as a hard disk) product and sold it to people who had IBM mainframes. IBM sued, Memorex counter sued, eventually Memorex prevailed and removed the threat of litigation over people whole made 'third party' add-ons.

I don't think this case will try to contravene that case law but you did ask for the 'worst case' :-)

Oracle doesn't want to kill Android, they just want a cut of Google's revenue. The worst case scenario is that Google has to write them a big check.

How is it that Google (et al) does not just fall back on the GPL-licensed OpenJDK? I understand that Oracle has a lot of money to burn on lawyers and other forms of harassment, but why does it seem that the common wisdom (not just this case, but the whole Java community seems generally scared of Oracle) is to completely ignore the generally accepted legal grants of this one specific GPL-licensed project?

Probably because not all their software is open source. Google would not like having to open source maps and gmail as gpl because the runtime was.

They have specifically addressed this. By having a GPL-based set of libraries to implement from, they would then be forcing this license to their app developers.

The GPL on OpenJDK does not infect apps.

I understand the classpath exception. But if you look at the opening slides showing evidence Oracle will be presenting [1] (whether taken out of context or not), it specifically states from a Google email:

  "… and as far as GPL-ing the VM, everything that is linked with the VM would get infected. The problem with GPL in embedded systems is that it’s viral, and there is no way (for example) OEMs or Carriers to differentiate by adding proprietary works. We are building a platform where the entire purpose is to let people differentiate on top of it."

So classpath exception aside (and the first OpenJDK release was after Android was started, but before the above quoted email was sent), they were still afraid of infecting their developers.

[1] - http://www.oracle.com/us/corporate/features/opening-slides-1...

Edit: fixed wording

As clever as they may be, they are not always right.

I think that's excessive paranoia on Google's part. It would be interesting to look at actual Android customizations and see if any of them would be affected by a GPLed JVM.

> any of them would be affected by a GPLed JVM

Do you mean I can make proprietary Java code GPL'ed by running it on OpenJDK? Wow! That would be great ;-)

Unless someone customized the JVM itself, nothing would have to be released under the GPL beyond the modified JVM itself.

E-mails like that were probably why Oracle assumed Google can't hire good lawyers.

Emails like that are more likely evidence that Google's Android partners can't hire good lawyers (not that we don't have other evidence of that, of course). Google would have to take their concerns seriously regardless of what their own lawyers said so it wouldn't be surprising if their partners' position wound up being Google's baseline.

OpenJDK would completely suck on a phone; the memory overhead and startup time are about 10x worse than Dalvik. And fixing the overhead requires eliminating JAR files, which breaks compatibility.

Vanilla OpenJDK is a resource pig, yes. But one could start off with the OpenJDK code base, drop JAR files and perform other wildly-breaking changes, and end up with a GPL-licensed Dalvik. Claiming current Dalvik as a derivative work of OpenJDK would be dubious, of course. But my point is that a path exists that should be able to morph the Java technology into any new GPL+classpath-licensed system regardless of Oracle's whining, yet it seems like this option is never considered.

Ellison doesn't seem to know a lot these days (and this isn't just snark): his moves to purchase then effectively lose influence over various large aspects of the Sun assets (Jenkins, Open Office, MySQL), his vacillation on whether NoSQL is a threat or not, and this ever churning battle with Google over Java...just shows a very unfocused leadership at the helm.

There is nothing more dangerous for a person like Ellison than being surrounded by yes-men. When you are effectively shielded from reality, it's really hard to make informed decisions.

I'd agree with this. Ellison definitely seemed to have more of a steady hand when he had Ray Lane as COO; a textbook pragmatist, Lane was the diametric opposite of Ellison, and the balance between the two worked really well.

Is there any other programming language with similar legal history? I would like to know.

We probably would have gotten a good preview with the Ashton-Tate vs Fox Software if Ashton-Tate hadn't totally bungled their copyright filings for dBase.

I wouldn't answer in his position either. If there's any internal record that he wasn't sure (and how could he be before the trial's over), and if he mentioned it to anyone who isn't a lawyer (no attorney-client privilege), then a definite answer either way invites a perjury charge.

I wonder if a ruling that favors Oracle will taint JVM-based languages such as Clojure?

This has nothing to do with Clojure, JRuby, or Jython at all to my knowledge.

To understand why: assume Clojure is just a software running on top of Oracle's JVM. Just like Apache Tomcat.

Yes, but the same chilling effect that it would have on adoption of JVM-based solutions could transitively pass on to whatever builds on its stack.

I wonder what would have happened if IBM won the bidding war for Sun instead of Oracle.

Both companies have reputations to preserve. Oracle is just preserving theirs.

Reply on news.ycombinator.com