Java Trap, 2010 Edition

http://journal.paul.querna.org/articles/2010/10/12/java-trap-2010-edition/

by pquerna • 16 years ago

177 96 16 years ago

journal.paul.querna.org

16s• 16 years ago

I agree with the article and use truly free dev tools whenever possible, but for some there is no choice because of the platform they have chosen to target. If you decided to do Android apps, you'll probably use Java for 90% (or more) of the work.

mcherm• 16 years ago

...and that's exactly the point. Oracle is suing Google, complaining that they are violating Oracle patents by running Dalvik on Android. Perhaps they'll win, and every app written in Java for Android will become useless.[1]

[1] - Note: not a realistic possibility. Someone would pay a billion dollar ransom instead. But the fact that the threat is out there, regardless how how credible it is, shows the problem.[2]

[2] - Note: the problem in this case is NOT due to non-free implementations: no one questions whether Dalvik is sufficiently free. The problem in this case is software patents. The only proper fix for that is to change the law. Please call your congresscritter.

bad_user7• 16 years ago

The other possible fix would be for a Java/Android related alliance to get formed by companies with relevant patent portfolios and sue the crap out of Oracle.

sanxiyn3• 16 years ago

This is not trolling. I honestly ask: why is TCK important? How does it prevent Apache Harmony from ever shipping a "real release"? After all, before OpenJDK, GCJ and lots of other free Java implementations released fine without TCK.

pilif4• 16 years ago

the problem is that Sun/Oracles patent license only covers implementations that passed TCK.

Conversely, by not passing TCK you don't get their patent license. Many of the patents in question are difficult (if even possible) to work around if you were to create your alternate implementation.

This means that using any non TCK implementation of Java makes you liable for patent lawsuits (as seen with the Google case).

sanxiyn3• 16 years ago

Ah, so anyone using GCJ in a product could be sued the same way Google was sued. I didn't know that.

The right conclusion to make seems to be that software patents are bad.

rbanffy10• 16 years ago

> the problem is that Sun/Oracles patent license only covers implementations that passed TCK.

Can they do that, being the OpenJDK GPL-licensed? IIRC, it shields the OpenJDK, as well as GPL-licensed derivatives, from patent lawsuits.

It doesn't protect Apache/MIT/BSD-licensed implementations, but who cares, if there is a GPL-licensed implementation available?

va_coder5• 16 years ago

I would also add the question: If I'm running Java on the server to deliver Web services, who cares? It's still open source, I have no plans to distribute MyPersonalVM that can run Java.

Google is the one that wants to keep it's advertisement machine going. It matters to them that over the next few years an open and mobile competitor is around. It doesn't matter to me.

hga9• 16 years ago

The author pointed out that Sun/Oracle's posture is not resulting in as healthy an ecosystem as other languages with multiple independent free implementations.

hugh4life4• 16 years ago

"After all, before OpenJDK, GCJ and lots of other free Java implementations released fine without TCK."

GCJ is old... and OpenJDK does have access to the TCK IIRC...

sanxiyn3• 16 years ago

GCJ is old, but is there any other implementation that can compile to native code? (Mainly good for simplifying distribution and better interfacing with native code instead of JNI.)

bad_user7• 16 years ago

> OpenJDK does have access to the TCK

Of course it does, access to the TCK was always at Sun's discretion to give.

OpenJDK doesn't matter, Java is not free to be implemented by open-source projects (even though JSR 270 explicitly says so) because the TCK licensing prohibits it.

It is also irrelevant if GCJ is old, patents don't have a "use it or you'll lose it" clause as trademarks do.

rbanffy10• 16 years ago

Didn't Stallman write an update to his original?

It's really hard to miss, right under the title, in big bold type:

http://www.gnu.org/philosophy/java-trap.html

bobds2• 16 years ago

I was browsing Wikipedia's list of Java implementations recently, so I was wondering in what way the Java Trap still exists, as now you can find dozens of both proprietary and open implementations.

http://en.wikipedia.org/wiki/List_of_Java_virtual_machines

kprobst• 16 years ago

The read from the Oracle lawsuit is that using any of those will get you sued.

I've read many arguments about how Dalvik is 'different' and some or all of these are 'safe' and so on, but the reality is that 'safe' is whatever Oracle decides it is.

nadam2• 16 years ago

What is not clear in the article is that languages and platforms are almost completely orthogonal in most cases. If we are speaking about the Java platform lock-in, languages are irrelevant, because I can use this platform from multiple languages. Even if I use Python, I can be locked into the Java platform because I call existing Java classes (using Jython). Also, for example whether Java the language is good or bad is irrelevant, because I can use Scala to compile to the Java platform, which is the best programming language I have ever seen (for my taste at least). I can also choose to target the .NET platform from Scala.

So Java platform lock-in can be a problem, I understand. But this seems to be totally orthogonal to the topic of from wich programming languages to which platforms you can compile. These different issues are somehow mixed in the article I think.

jacobolus2• 16 years ago

The article’s argument is this: If you write your code in Python targeting Jython and the JVM, and the Java platform becomes unusable, there’s still a lot of your code that can be easily ported to run in CPython or IronPython or PyPy or something. If you write your code in Java and the platform becomes unusable and Oracle decides to squish alternative Java platforms, then you’re stuffed.

nadam2• 16 years ago

Ok, so the problem is that Java the language supports only one target platform. This is not completely true, I don't know too much about about J#, but as far as I know using J# basically you can compile your program written with Java syntax to .NET.

borism• 16 years ago

I don't see how this could be true. What makes Java the Language so special that it can only be compiled into JVM bytecode and not something else?

tickle_me_elmo• 16 years ago

At the multinational I work at, we recently upgraded to Java 1.5. By the time we have to upgrade to Java 1.7, I expect this little storm-in-a-teacup to be well and truly forgotten.

hugh4life4• 16 years ago

I really wish more effort was put into D... I think that's really the best path to freedom vs Java and .Net... I wish Nokia would would adopt D because nobody wants to use C++ and Python is slow...

I just saw this on twitter...

http://twitter.com/casperbang/status/27134952930

"Java has an official open source reference implementation, but is not free. C# has only unofficial open source implementations, but is free!"

http://www2.research.att.com/~bs/bs_faq.html#Java

"Java isn't platform independent; it is a platform. Like Windows, it is a proprietary commercial platform. That is, you can write programs for Windows/Intel or Java/JVM, and in each case you are writing code for a platform owned by a single corporation and tweaked for the commercial benefit of that corporation."

Keyframe3• 16 years ago

Which D, 1 or 2? I have used D for several years and know a few people that also used D for a longer period of time. Consensus is that D is dead for all practical purposes (especially D2).

hugh4life4• 16 years ago

"Which D, 1 or 2?"

D2 of course... "Consensus is that D is dead for all practical purposes (especially D2)." BS..

There is a chicken and egg situation with D... if somebody big doesn't put some backing behind it then it will go nowhere...

MrRage• 16 years ago

Could you expand D being practically dead? or link to some sources on that? I haven't heard such strong statements about D before, and I'm curious to learn more.

wazoox2• 16 years ago

How a

wazoox2• 16 years ago

How a language with a proprietary reference implementation and library, namely D, could ever be a suitable solution to the Java trap? This is utter nonsense.

microtonal• 16 years ago

Most people will see it as C++ with a somewhat nicer syntax. That does not provide a lot of incentive to switch to D. You need more than marginal benefits to cause a huge shift.

Additionally, it doesn't help that D is torn. There used to be the Phobos/Tango split, now there is yet another split: D1/D2. The second is definitely a more serious competitor to C++, but there is no up-to-date fully open source compiler available. Digital Mars's D2 compiler has a proprietary backend, gdc (a GCC frontend) is outdated, and ldc (LLVM frontend) does not have usable D2 support.

Oh, and if you want a 64-bit D2 compiler, even proprietary, you are out in the woods (yes, work has started).

Even the average Scheme compiler is in a more usable state ;).

jakevoytko• 16 years ago

JNI minimizes the Java trap, since your entire program doesn't need to live in the JVM. Other functionality can be implemented in native code, and it will work with Java just fine. Oracle does not control native code, so they don't control what platform you target in the future. JNI is not a panacea - your code still runs on Oracle's JVM. But it provides an escape route, since pieces of your application can be ported away from the JVM without requiring an entire rewrite.

If you've reached this escape point, yes, you're wasting time by porting your application away from Java. But you knew Java is a closed platform owned by one company, and you took that risk.

tomjen3• 16 years ago

Even more importantly, you can start the JVM from C code -- indeed 'java' is nothing more than a simple wrapper around the necesarry DLL calls.

devmonk4• 16 years ago

I think the author is just taking advantage of Java's name having been drug through the mud by Oracle to push his point-of-view.

Many of Apache's projects are Java-based, and I don't seem them making any effort to port all of them over into a different language.

Though Oracle stunk up JavaOne this year, and the lawsuit against Google is making Oracle look like a supreme ass, companies like IBM, RedHat, SpringSource, and many, many others have a significant investment in Java.

Java has been and will continue to be close to the top of Paul's heap in terms of jobs for the foreseeable future:

http://www.indeed.com/jobtrends?q=java%2C+Python%2C+Javascri...

http://www.indeed.com/jobs?q=java

However, I would be the first to say that if you are looking to the future, explore dynamic and functional programming languages. If I had to pick one right now, I'd lean toward Ruby, not because it is the most well adopted (and for goodness sakes, it has had enough time to be), but because the language itself is great, and it has a growing and thriving community.

But Java is not dead yet, and it is far from a trap.

mbreese2• 16 years ago

I don't think that Apache would start porting existing projects, but do you think that there will be many new ones? Or how many existing projects will have any significant effort put behind them?

Imagine, if you will, that there were no Apache. What would the Java landscape look like today? If Java at Apache flounders, the platform as a whole will also diminish. Java has thrived for so long in a Microsoft dominated world thanks largely to third-party open source efforts. If a (the?) major contributor of open source code in Java goes away, which direction do you think Java will go? "Thriving community" is not how I see this turning out.

devmonk4• 16 years ago

"do you think that there will be many new ones?"

Yes, I think there will continue to be some new projects, and other projects will fall off. Apache isn't the end-all be-all of Java projects, however.

Several of the larger newer projects were under SpringSource's wing, and there continue to be new projects there. And there's Codehaus. And some Java projects under GitHub, Java.net, SourceForge, etc. Then there's Groovy, Grails, and related projects. JRuby (formerly at Sun, now at EngineYard) is Ruby, but it allows some reuse of other Java projects. Android (like the author states) is a subset of Java, so all of those Android app projects. Kindle's KDK is Java. Your cable box may even run on Java.

If you've been only looking at Apache for Java growth, you've been looking in the wrong place.

"Or how many existing projects will have any significant effort put behind them?"

The Java projects at Apache have had continued effort put behind them, but some of the people that would have put effort towards Apache projects have spread out into some of the places I mentioned.

"If Java at Apache flounders, the platform as a whole will also diminish."

That's BS. If Apache fell on its ass, Java would still survive. See all of the other projects I mentioned, plus all of the current Java jobs I mentioned.

Where "Java" itself isn't the new hot thing, Java is dominant, and things using/requiring Java are hot, for example- look at this Google trends of Java vs. Android to see what I mean:

http://www.google.com/trends?q=java%2C+android

rbranson• 16 years ago

I agree with the article, but I'm also starting to not see why Oracle even cares about Java anymore. There are enough fantastic free / open source products available that if you told an investor you were starting a company to build J2EE infrastructure, they'd probably laugh you out of the room. At this point the only reason to use a commercial J2EE platform is either lock-in (it'd be too much time to move to open source) or for some corporate reason (support, big name, impressive, etc). In what other ways can Oracle monetize the Java platform at this point?

doki_pen3• 16 years ago

Summary: "I don't agree that users deserve freedom, but if someone takes away my freedom, then its wrong."

I'd love to hear his reasoning for disagreeing with RMS's ideal of user freedom.

adbge• 16 years ago

A BSD license gives the user near absolute freedom to do what one wishes with a piece of source code -- they can sell it, modify and make proprietary, integrate with existing proprietary licensed projects, etc.

On the other hand, the GPL imposes heavy restrictions on what a user can do with a software's source code in order to preserve other users' freedoms.

That's the fundamental split in permissive licenses vs copyleft licenses. It has been argued to death and the author probably (correctly) assumed that most readers are already familiar with it.

doki_pen3• 16 years ago

Once they distribute it, they are not just users anymore. The people who receive it are the priority with GPL. The major difference with the GPL is that the user is king.

doki_pen3• 16 years ago

Why does anything need to be run the TCK? Why does anything need to use the name Java? Isn't this more of a people problem then a licence problem? Am I really to believe that if someone doesn't have access to Oracle's magical testing kit then it's impossible to make something that runs java source code? Isn't it more just a corporate perception problem? Who cares?

wrs• 16 years ago

Well, that's what Google thought...after they spend a few million dollars in attorney's fees, we'll see if they were right.

matrix• 16 years ago

Let's say we buy into the Java Trap Hypothesis; the question then is: what's next? What do you choose if you're a corporation with large projects and large teams of mostly non-rockstar developers. For the sake of the argument let's assume the JVM is out, and you can't use .Net (e.g. need Linux, or more flexible licensing, etc). I'm curious what people would choose, if given a clean slate for this kind of situation.

ciupicri2• 16 years ago

For Linux there is Mono for Linux and it has a pretty liberal license, a mix of MIT/X11, LGPL, GPL and MPL[1].

[1] http://mono-project.com/License

alextingle• 16 years ago

C? C++? (Or are you seriously suggesting that the average Java programmer is too feeble-minded to learn how malloc/free or new/delete work?)

Python is a great choice too, provided you don't mind about performance (which is true in the majority of applications).

jrockway• 16 years ago

Haskell. Faster than java, easier than java, safer than java. And it even has a centralized repository of libraries.

flogic• 16 years ago

Ok. From my limited understanding of the FAQ. Sun agreed to release the TCKs without "field of use" restrictions when they entered the community process? But they failed to meet that obligation with the JavaSE TCK? Isn't there some legal structure to force Sun/Oracle to meet that obligation?

hga9• 16 years ago

With the JavaSE TCK for the Apache Harmony project.

There may be a legal structure to address that, but for anyone who depends on the good will of Sun, now Oracle, which probably includes IBM, it certainly wouldn't have been practical.

JVerstry• 16 years ago

De facto, if Oracle/Sun refuses to TCK any independent implementation of Java SE, while tolerating their implementation under open source license, then there is ground for anti-trust action.

The argument being that 3rd parties have no means to validate their implementation. Hence, there is no practical way to implement Java SE independently.

Considering the dominant position of Oracle/Sun, we can't exclude the possibility that the European Commission will start a new anti-trust investigation against Oracle/Sun.

itissid• 16 years ago

Was there not news yesterday that IBM and oracle are going to work together on Open JDK?

kaffiene• 16 years ago

The article is FUD. Java is too important to too many people to not be around any time soon. Also, there's a GPL'd release and IBM's Java.

Java Trap, 2010 Edition

Related Stories