A good summary. After several days of thinking about this I can only come to two conclusions, neither of which I like:
a) Oracle is realizing that mobile is strategic and has decided this is how they will get in on the action. They think it is much more important than Java. They will force Google to license Java from them and that way gain not only royalties, but actual control of Android.
b) This is purely about damaging Android. There is an agreement we know nothing about yet between Apple and Oracle that if Oracle takes down Android something will happen in return from Apple. They don't care if they win or lose, they just want to smash adoption of Android. It could as easily be a move from MS but it is hard to think past the bitterness and sense of betrayal expressed by Jobs about Android, or the frequently reported "close friendship" between Ellison and Jobs.
I wish these were conspiracy theories but I can't account for Oracle's willingness to detonate what is still an incredibly valuable Java asset any other way.
> There is an agreement we know nothing about yet between Apple and Oracle that if Oracle takes down Android something will happen in return from Apple.
Like Apple gives a free iPod to everyone who purchases an Oracle database or something like that? Apple and Oracle are in such different markets it is hard to see how there is anything Apple can do for Oracle.
I have to think that if licensing fees were required for Android Google would re-tool 3.0 for Python or something. Mobile processors are getting fast enough that Python wouldn't be a problem
And if they didn't, I suspect device manufacturers would look to the truly free Linux platforms.
I think that retooling with Python would be such a huge undertaking that they could not do it. Not so much for their own software, but lets face there are ~100,000 Java apps out there for android now.
It would be interesting if they could do either of:
- Switch to use the GPL version of Java and be fully conformant with the java specs. This presumably would hurt performance unless they could do some miracle engineering to obviate the reasons Dalvik exists. Perhaps though, as phones get more powerful it's not so important.
- Keep Dalvik but work around whatever ip violations Oracle is successful in claiming. They probably believe they have already done this.
What is important for a phone platform is what's on top, not what's on bottom. Android can compete with iPhone because of the high level stuff Google provided--the Linux at the bottom is pretty much irrelevant.
Most Linux phones, other than Android and WebOS, run crappy top level stuff written by the device manufacturers. Rarely, a device manufacturer does come up with something that is not crap to put on top of the kernel, but even then they can't compete with iPhone because they lack the infrastructure, like app stores that can attract significant developer interest.
I agree that this is not primarily about money, given the uncertainty of an outcome. More than anything else though, this is Oracle proving they aren't going to play softball with infringement Sun's IP.
The whole episode reminds me of Nixon and Kissinger's plan to make the Soviets think he was nuts:
"Frustrated, Nixon decided to try something new: threaten the Soviet Union with a massive nuclear strike and make its leaders think he was crazy enough to go through with it. His hope was that the Soviets would be so frightened of events spinning out of control that they would strong-arm Hanoi, telling the North Vietnamese to start making concessions at the negotiating table or risk losing Soviet military support."
"Larry doesn't care about openness or community -- he holds them in contempt. What he cares about is winning. And this was his way to prove it."
That's a better way of putting it than anything I came up with. To give credit where credit is due, his approach has produced for the firm. And his disdain for communities like Linux (e.g. Oracle's Linux distribution) and OpenSolaris (e.g. the leaked memo) has not cost the firm much on the public markets thus far.
But the same was once true of Microsoft, and now they have felt compelled to attempt a difficult, long term rehab of their public image. Which makes one wonder whether a similar effort from Oracle is not an If, but rather a When.
Respectfully disagree. Reputation and image are important if yours is poor amongst a community that influences buying decisions.
Traditionally, this hasn't really been the case with Oracle, as their products are still adopted up the food chain from most developers who might be expected to have a poor image of Oracle.
The question is how long this remains the case, and what impact a poor image will have on hiring, partnership, investment in the Java ecosystem, and so on.
That's certainly going to be Google's strategy. The chances of Oracle receiving an injunction on this appear to be near zero. So Google will simply try to stretch this out for years. In the meantime Android will carry on business-as-usual. By the time the whole mess is sorted out, the whole mobile landscape will likely be much clearer and much more stable.
It's really a non-story from a practical strategic point of view for anyone involved (including application developers).
Im not so sure. there is a clear difference here. With SCO the majority opinion was that SCO are full of it and will loose. I don't think many people stopped using Linux as a result of the suit.
In contrast I can clearly envision a vendor deciding against building an Android based phone in fear of future royalty payments etc.
Some have already started calling this SCOracle. So yes, be prepared for a long protracted legal battle. The upside might be the end to software patents. The downside is years of legal maneuvering not to mention the death of Java.
As much as I would like to believe that this case could bring sufficient attention to the absurdity that is software patents that it would compel a remedy, I think it unlikely.
Frankly, this is likely to be regarded as business-as-usual outside of technology circles, and as long as large technology providers are extracting substantial revenue from their intellectual property in this fashion we can't expect much change.
Economics are, generally, the only real change agent.
I think they value Java quite a lot. The frightening part is that they must value something else quite a lot more. For my money, it has to be that they want a stake in the mobile market - they've decided they can't sit by and let the future of computing pass them by. This is their way of getting in on the action.
Total speculation but it seems very likely that Google has some database / data analysis / data processing related patents that Oracle would like to cross-license and rather than a starting with a soft discussion they throw a live grenade in the room and make the talks go quickly.
I wonder if Google made a mistake including any part of "Java" in Android. In retrospect, it seems like it would have been better to not only create their own VM but the whole rest of the stack as well.
It wouldn't have protected them, necessarily, from these patent claims but Oracle wouldn't even have bothered looking because there wouldn't have been any Java technology involved.
One of Oracle's selling points for their database was independence over hardware vendors. Java offers the same - although today's hardware is much less diverse.
Sun in fact did litigate over Java, as the article offhandedly mentions, though the motivation there was that MS different version was fragmenting the language, and undermining the compatibility of "write once, run anywhere".
I agree with the article that even a few billion in damages isn't worth it to Oracle. There must be some strategic advantage, quite possibly in some negotiations with Google that we haven't heard about (and perhaps will never hear about).
The point made about the mobile revolution is the right kind of path; another revolution in progress is the cloud. There is definite overlap between Google and Oracle in the cloud, with Google's technology expertise, and it being used by the enterprise - that's probably what Oracle hopes to gain.
I thought a while back that Google's Appengine could eventually compete with Oracle Database and Oracle Peoplesoft. As more people use Google Apps, Gmail and build custom apps on Appengine, they may start to compete with Oracle offerings. Maybe this is Oracle's way of slowing that down.
Not much. I've seen no steps taken against OpenJDK yet, and even if they did an OpenSolaris on that, you could always run it with the official JDK. It would be beyond crazy if the wouldn't make one available anymore.
What probably will happen is that some frustrated developers won't use anything associated with the JVM. I don't expect to see a lot of this, though. A lot of the people who use JVM languages already work in an environment where there's more than enough proprietary enterprise software. And I doubt that it matters a lot for complete newbies.
Personally, I was on the lookout for new languages to focus upon, after playing with a lot recently. And after this, Ocaml and Erlang moved to the top of the stack, past Scala and Clojure.
Let me recap the issue as I understand it so far: Oracle filed a lawsuit against Google regarding patents infringement for their use of Dalvik in Android. This is a big problem for Google and for every other alternative JVM (especially open-source) because even for a clean-room implementation you need to be licensed by Oracle.
But as far as I can see this is not so interesting to general application developers since, I think, the majority of them targets the official JVM (open or not) anyway. The same goes for other JVM based languages, since they can generally run well in the OpenJDK.
However the general consent that I see is that this is a bad move for the innovation of the JVM platform and a move that will likely reduce the trust of the industry in this technology.
Maybe new projects will be developed upon another and more open platform, but the vast majority of application-level projects (such as the myriad of enterprise webapps) won't be so interested.
What do you think?
Disclosure: I'm interested in alternative JVM languages because in my company (who is J[ava/VM] centric) we're evaluating technologies to rewrite one of our products from scratch. Aside from my personal interest in new languages obviously.
It's too soon to say what the general consent on that issue is. The next few days will be very interesting in that respect. I wonder what IBM will say. If I'm not mistaken, their current java products (they've got their own branch of the JDK) are covered by license agreements, probably made in the Sun era. So it's unlikely that they're in immediate legal danger. Still, they might hedge their bets differently for the long term. I wouldn't really count on it, though, considering that there's no decent alternative right now, and generally IBM takes a long while to wake up.
Most of the industry doesn't exactly mess with the JVM, so they don't really care whether it's free-as-in-beer, free-as-in-speech or something in between. I think that this includes academia (where Scala is from) and consulting agencies (where Clojure is from). There's no real hard-core GNU free software crowd behind them.
I hope that I'm wrong, but I think apart from a small flock of hackers, nobody will care too much. I do think that the efforts to port both Scala and Clojure to other platforms (CLR / LLVM) will increase a bit.
A lot also depends on what Google will do. If they throw down the gauntlet and put lots of effort and money into a migration towards a different language, then the whole IT market will look quite different. But I think that's too much money wasted, to there'll be some underhanded deals and Google and Oracle will become fast friends again. If Google doesn't want to loose face by settling de jure, Oracle might drop the case and there'll be a de facto settlement.
So to summarize: From my limited knowledge right now, I don't see big practical reasons to avoid JVM-based languages. People were quite content using Java, even before Sun made their Open Source initiative. We'll just regress to that state.
Personally, I was on the lookout for new languages to focus upon, after playing with a lot recently. And after this, Ocaml and Erlang moved to the top of the stack, past Scala and Clojure.
If your reaction is common this could be very damaging to Scala and Clojure. Both languages need a steady influx of new developers to sustain and build their current momentum. If enough potential new blood balks because of this that's definitely bad news.
The whole thing sickens me. I've been very happily back on the JVM with Scala lately.
Well, to be honest, after spending quite some time as a Java developer I regard the whole JVM stack as a mixed bag. Yes, you get a lot of libraries as a package deal, but most of them are rather bloated, and do require quite elaborate wrappers. And you might end up with something "good enough" and then you'll carry around the baggage for a long time.
And I think the Scala and Clojure communities will probably do fine without me. Never mind that for work I still prefer them to unadulterated Java, it's just that when I have the choice and the design choices are wholly mine, well…
If Scala and Clojure were my only options, I might have a look into that. But as far as I know, both version aren't production quality and I don't know anything about the whole infrastructure and as of yet, I don't have any projects at work or in private where interoperability with anything .NET related is an option.
> Apple and Microsoft will be thrilled by this development.
No kidding! Apple, especially. To the extent that Apple has suffered from developer frustration with it's App Store policies, this is a mitigating factor. Apple is fickle about approving apps, but who knows what will happen with Dalvik.
With the disclaimer that I'm the author of the linked piece, I'm not sure why the comment is silly. Dalvik is indeed at the heart of Oracle's claims, but many casual followers of its development are operating under the assumption that Google's cleanroom reimplementation shields it from patent claims.
An interpretation that is obviously false, as Dalvik was intended to circumvent copyright and - perhaps - trade secrets, not patents. But it's something you hear with some frequency, so spelling it out in clear terms seemed worthwhile.
I guess I can see where you're coming from vis-a-vis wanting to spell it out. But wouldn't it make more sense to emphasize that creating Dalvik is exactly what put Google in position to receive this suit? It feels like you are suggesting that there was some intent to avoid a patent suit when Google decided to create Dalvik rather than license the JVM but that it didn't work. When in reality, creating Dalvik is what brought about the suit in the first place. If Google wanted to safely avoid a suit, they would have just licensed the JVM like everyone else right?
You need to understand the difference between patents and copyrights.
Copyright infringement and trade secrets abuse (where exposure could have happened) can be avoided by clean-rooming original works. That's what it means to clean-room - re-implement without being directly affected by the original.
Patent infringement can not be avoided in this fashion.
Hence, Google had choices:
1. License JVM, and probably pay some fees not compatible with their "free and free" android plans.
2. Use JVM without licensing and get nailed for copyright infringement.
3. Clean-room their own VM and risk a patent fight. At the time Sun was known for not starting up fights (Shwartz let on this in a recent interview), so it looked like a good plan.
4. Go a completely different route. Still risk patent infringement.
Wait, I never suggested that software patents could be avoided by cleanroom implementations. And I don't see how anything I've said has anything to do with the difference between patent infringement and copyright infringement. But your laying out of Google's options is very useful.
My point is just that it's a little weird to say that Dalvik couldn't protect Google from a patent suit, because building it created the suit. It's a little like saying that playing Russian roulette can't protect you from getting shot.
Just because a suit is created doesn't mean it has merit.
What he's saying is that clean room reimplementations (Dalvik) side-step copyrights, but not patents. The fact that it was created to circumvent something, just not this particular thing, could lead to a lot of confusion for a lot of less savvy readers if this wasn't cleared up.
I never said the suit had merit, so I'm not sure what your first comment is referring to.
But your second point is valid, and I agree that some things needed clearing up. I just thought the way that particular comment was worded wasn't helping that much.
Going back to my Russian roulette analogy, of course playing Russian roulette won't protect you from getting shot. It's the fact that you're playing Russian roulette in the first place that is putting you at risk of being shot. If people are confused about that, then it's probably best to educate them about how playing Russian roulette can result in you getting shot.
Bringing the analogy back, I feel like it would have been more useful to just educate people about the fact that creating Dalvik is what directly led to this suit and for what reasons.
Let me restate the point differently, maybe it will make more sense that way.
While many people think that Dalvik was built to protect against all kinds of lawsuits, this is not true - it only protects against one kind - copyright/tradesecret lawsuit, and does not protect against patent lawsuit. This is exactly what article says where you quoted it. The reason why it needs to be pointed out is that, well, many people do not realize the difference.
Yes it does. If you followed the quoted sentence with one explaining that by creating Dalvik they specifically set themselves up for this suit I think it would be a little more clear. But I see exactly where you're coming from.
> Dalvik was intended to circumvent copyright and - perhaps - trade secrets, not patents.
I disagree. There is no reason that Google couldn't have built their own clean-room implementation of the JVM and called in Android. If they used no Sun code and no Sun trademarks, there were would be copyright or trademark infringement. Outside of some technical advantage, the only other reason to completely change the design of the VM would be to get around JVM related patents.
It's quite possible that Dalvik allows Google to avoid most of Sun's patents on Java technology. The patents included the suit would seem to be violated by most VM technologies, including Microsoft's .NET CLR.
Going further than that, given Google hired a number of staff that had worked on the Java implementation at Sun, and given that the head of Google was present at Sun as the Java patent-and-copyright trap was being constructed, it seems inconceivable to me that Dalvik would have been permitted to violate Sun's patents.
In a rational world and under a reasonable patent system, I might agree. But you know as well as I do, Simon, probably better, that under the current process it's difficult to guarantee that you're not violating anyone's patents. Particularly when you're reimplementing an existing system.
Is it possible that Oracle's patents don't read on Dalvik? Certainly. Did Google take care to minimize the risk of such? I'm sure they did.
But however careful the execution, the system at present is would be actively working against them.
a) Oracle is realizing that mobile is strategic and has decided this is how they will get in on the action. They think it is much more important than Java. They will force Google to license Java from them and that way gain not only royalties, but actual control of Android.
b) This is purely about damaging Android. There is an agreement we know nothing about yet between Apple and Oracle that if Oracle takes down Android something will happen in return from Apple. They don't care if they win or lose, they just want to smash adoption of Android. It could as easily be a move from MS but it is hard to think past the bitterness and sense of betrayal expressed by Jobs about Android, or the frequently reported "close friendship" between Ellison and Jobs.
I wish these were conspiracy theories but I can't account for Oracle's willingness to detonate what is still an incredibly valuable Java asset any other way.