|
|
|
|
What do you think?
Posted:
May 9, 2005 5:38 AM
|
|
|
There is much excitement over this weekend's announcement of Harmony. What do you think? Is it important? If so, why? If not, why not?
|
|
|
|
|
|
|
Re: What do you think?
Posted:
May 9, 2005 8:00 AM
in response to: daniel
|
|
|
Personally I Am a long time user of apache projects, including Ant, Apache, and Tomcat. However, I doubt that I would switch from Sun's JVM. I am a fan of open source, but not a fanatic. I don't see what it solves. If they provide a JVM that is faster or in some way better than Sun's I would switch otherwise I don't really see what the point would be. I already know all the issues we have hit with Sun's JVM. I don't want to have to work around a new set of issues that may arise. For our product it is safer sticking to what we know.
|
|
|
|
|
|
|
|
|
Not just a JVM
Posted:
May 9, 2005 8:36 AM
in response to: haskovec
|
|
|
Note that this proposal is for an open-source implementation of J2SE, not just a JVM.
I'm worried by this prospect. In my organization we use both the Sun and BEA JRockit JVMs, and nasty little differences crop up when you least expect them.
I am concerned that multiple implementations of the J2SE libraries are bound to contain inconsistencies. The Java specifications are not perfect. The TCKs don't catch everything.
I understand the desire for Open Source Java, but this approach scares me.
|
|
|
|
|
|
|
|
|
Re: Not just a JVM
Posted:
May 10, 2005 12:26 AM
in response to: johnreynolds
|
|
|
> I'm worried by this prospect. In my organization we
> use both the Sun and BEA JRockit JVMs, and nasty
> little differences crop up when you least expect
> them.
>
And JRockit passes the TCK, this beast looks like it won't even try...
> I understand the desire for Open Source Java, but
> this approach scares me.
>
Personally I don't understand it, at least not from people who actually use and like Java.
Open source JVMs I can understand somewhat, but this goes way beyond that and wants to create an open source JLS and JVMS with no requirement to pass the TCK.
That's extremely worrying as it will inevitably lead to version hell where we'll be back to having to write our applications a hundred times for every possible version of the language out there (and support them all).
|
|
|
|
|
|
|
|
|
|
|
|
Re: Not just a JVM
Posted:
May 12, 2005 12:40 AM
in response to: marla_parker
|
|
|
Nothing about passing the TCK is mentioned in the project proposal...
In fact the proposal mentions key areas where the resulting JVM would (depending on how it's installed with its "modular" thinking) would fail the TCK.
|
|
|
|
|
|
|
|
|
Re: What do you think?
Posted:
May 9, 2005 8:26 AM
in response to: daniel
|
|
|
It is really a good step. But unless and until they get their first out really fast, it has the fate of being another Classpath.
I really hope that Sun would donate substantial amount of its codebase to Apache, so that they can get to it fast, and thus speed up the adoption of Java among the Open Source Fanatic Club (though I still doubt they would be willing to move out of their C++ trenchs).
If I may hazard a guess, I think it is Sun's way of experimenting with Open Source Java??
|
|
|
|
|
|
|
|
|
Open Source Fanatic Club?
Posted:
May 10, 2005 10:57 AM
in response to: vhi
|
|
|
> I really hope that Sun would donate substantial
> amount of its codebase to Apache, so that they can
> get to it fast, and thus speed up the adoption of
> Java among the Open Source Fanatic Club...
Would the Open Source Fanatic Club be satisfied with an Apache Licensed Java, or would they demand a GNU version? 
|
|
|
|
|
|
|
|
|
Re: What do you think?
Posted:
May 9, 2005 8:33 AM
in response to: daniel
|
|
|
I think it's step in the right direction. There's still a number of problems in the J2SE itself and hopefully this will spark new ways of implementing the write once, run anywhere concept. Have you seen how many methods require native calls? This seems like poor design or implementation at work.
My concern is this can lead to better cross-platform support (palm, linux, apple,...) then is presently availble.
|
|
|
|
|
|
|
|
|
Re: What do you think?
Posted:
May 9, 2005 8:48 AM
in response to: daniel
|
|
|
The world really needs an OSI-approved Open Source J2SE implementation. Frankly, for better or worse, the OS community is not going to truly embrace Java until it is 100% OS, and Java is not going to see its full potential until the OS community completely embraces it and takes it as far as it will go.
I completely understand the people worried about consistency across platforms. This is why I wish that it were Sun opening its implementation--I really think it would be in Sun's interest to do that--but we'll take what we can get. 
|
|
|
|
|
|
|
|
|
Re: What do you think?
Posted:
May 10, 2005 12:29 AM
in response to: afishionado
|
|
|
> The world really needs an OSI-approved Open Source
> J2SE implementation. Frankly, for better or worse,
> the OS community is not going to truly embrace Java
> until it is 100% OS, and Java is not going to see its
> full potential until the OS community completely
> embraces it and takes it as far as it will go.
>
The WORLD doesn't need it, only a few open source zealots need it and then only to destroy Java's main strengths (which are based on maintaining compatibility between deployment platforms).
> I completely understand the people worried about
> consistency across platforms. This is why I wish that
> it were Sun opening its implementation--I really
> think it would be in Sun's interest to do that--but
> we'll take what we can get.
Sun opening the JLS and JVMS and allowing them to be forked and changed at will by everyone without loosing the right to call it Java would be no different than this potential disaster.
It would in fact be worse as it would take less time for the problems to appear.
|
|
|
|
|
|
|
|
|
Re: What do you think?
Posted:
May 9, 2005 2:30 PM
in response to: daniel
|
|
|
Generally speaking competition is good.
Specifically speaking, usability is not Apache's strong point. Far from it actually.
I don't see that Harmony will make a difference in the grand scheme of things and I'm sure usability will be rather poor. I hope I'm proven wrong on both counts but there it is
|
|
|
|
|
|
|
|
|
Re: What do you think?
Posted:
May 10, 2005 1:34 AM
in response to: daniel
|
|
|
I think that this is bad.
1) The job is enormous. It will take years.
2) It will never be 100% compatible with Sun Java.
It would be better if all development was done on Sun Java. The only thing that might be good, is that Sun could be forced to open Java completely. But I think that the situation with Mustang is already quite good, the source code is there...
|
|
|
|
|
|
|
|
|
Re: What do you think?
Posted:
May 10, 2005 5:13 AM
in response to: fuerte2
|
|
|
With respect to the "It would be better if all development was done on Sun Java." line of thinking:
I use a Mac. It's the first Mac that I have ever owned, and I'm a convert. It's a really good desktop *nix, and a great development environment.
Sun does not release versions of J2SE for the Mac (they originally did not release versions for Linux). So, I use Apple's J2SE implementation. It works, but releases lag behind the 'official' releases from Sun (Tiger is a good example).
I don't understand the fear of an open source implementation. I think it would be great if we could have one open source implementation with distributions for all OSs. Get everybody to use the 'standard' implementation, rather than the implementation from one vendor.
If the OS vendor's contribute to Harmony, doing the grunge work to build distributions for their specific OSs, then we can get to a place where Java is truly standardized.
I think this is good news. Perhaps work needs to be done on the TCKs so that they _do_ catch everything, or a least as close to everything as we can get.
|
|
|
|
|
|
|
|
|
Re: What do you think?
Posted:
May 10, 2005 5:53 AM
in response to: aaston
|
|
|
I didn't know that Sun's Java does not work on Mac.
Still, perhaps Mac people could take part in Mustang development so that it is ported to Mac. Probably not at this point, but maybe later?
Opening Sun Java more would be much better than a whole new Java.
|
|
|
|
|
|
|
|
|
Re: What do you think?
Posted:
May 10, 2005 3:49 PM
in response to: fuerte2
|
|
|
What I would do for Sun to put out a Mac OSX JRE. You can write an application that runs on the Sun J2SE, and when deployed on OSX will crash, even though it is supposed to be a port. Then in order to make your application work on OSX you need to buy a Mac just to debug it on OSX. This is bunk. Of coarse it might be cheaper to outsource my development to another country and have them port it to Mac OSX. The sad thing in all of this is OSX is by far a superior GUI, and if I had the choice I would develop on that platform if support was available, or if it was feasible.
|
|
|
|
|
|
|
|
|
Re: What do you think?
Posted:
May 12, 2005 12:29 PM
in response to: aschiffman
|
|
|
> What I would do for Sun to put out a Mac OSX JRE.
Alternately, Apple could ask the community for help in porting the JRE. Even with a completely open J2SE, the work required to port it to the Mac is not going to miraculously disappear.
|
|
|
|
|
|
|
|
|
Re: What do you think?
Posted:
May 12, 2005 4:06 PM
in response to: johnreynolds
|
|
|
Why rewrite the class library when all you need is a JVM.
|
|
|
|
|
|
|
|
|
Re: What do you think?
Posted:
May 11, 2005 1:38 AM
in response to: aaston
|
|
|
> I don't understand the fear of an open source
> implementation. I think it would be great if we
> could have one open source implementation with
> distributions for all OSs. Get everybody to use the
> 'standard' implementation, rather than the
> implementation from one vendor.
>
there can be no "standard" implementation for all platforms.
You'd need hundreds of implementations, each created for a specific platform.
That's why Sun doesn't provide JVMs for all platforms...
The problem with Sun giving up the JLS and JVMS to the likes of the slashdot crowd (who are the ones screaming for Sun to open source Java even though they hate Java) is that it will destroy the platform and turn it into a quagmire of thousands of mutually incompatible versions all of which you'd have to code and test for.
> If the OS vendor's contribute to Harmony, doing the
> grunge work to build distributions for their specific
> OSs, then we can get to a place where Java is truly
> standardized.
>
Anything that passes the TCK already should run all Java code that's properly written.
But of course there seems to be no plan for Harmony to even try to pass the TCK, at least nothing of the kind was put down in the proposal (in fact there's several phrases in it that seem to make sure it will never pass the TCK).
So the tacid goal seems to be to deliberately create a Java implementation that is incompatible with the language standards.
|
|
|
|
|
|
|
|
|
Re: What do you think?
Posted:
May 10, 2005 2:32 AM
in response to: daniel
|
|
|
I think Harmony is a tremendous waste of time and talents. If there is something wrong with the J2SE reference implementation by Sun (and I wonder what this might be), why not fix this?
|
|
|
|
|
|
|
|
|
Waste of time and talents...
Posted:
May 10, 2005 7:43 AM
in response to: uhilger
|
|
|
> I think Harmony is a tremendous waste of time and
> talents.
I could not agree more.
This effort is not going to produce anything new, just another implementation with a different license. This is kind of like creating yet another J2EE app server (Geronimo, JOnAS, JBoss, etc.). Do we really need an implementation per license? If so, then we'll need a GNU J2SE to counter the Apache J2SE.
> If there is something wrong with the J2SE
> reference implementation by Sun (and I wonder what
> this might be), why not fix this?
Those who wish to improve the JDK can do so by joining the Mustang project.
(Check out the new JDK collaboration project: http://weblogs.java.net/blog/brinkley/archive/2005/05/jdk_community_l.html)
Other posters have referenced the lag in porting J2SE to various platforms. Open Source or not, the work necessary to port Java to different platforms remains the same. If Apple wants help in porting J2SE to the Mac faster, then they should open that process (I'm sure Sun would grant the proper license).
Message was edited by: johnreynolds
|
|
|
|
|
|
|
|
|
Re: Waste of time and talents...
Posted:
May 10, 2005 8:27 AM
in response to: johnreynolds
|
|
|
If I contribute to Mustang, doesn't Sun own my work? And without any license to protect access to my work? What if Sun decides it wants to change the way it distributes Java in the future? Working on Mustang may be fun, but it's still working for someone else for free.
Please correct me if I'm wrong.
|
|
|
|
|
|
|
|
|
Re: Waste of time and talents...
Posted:
May 10, 2005 9:04 AM
in response to: tjpalmer
|
|
|
It's a joint copyright.
|
|
|
|
|
|
|
|
|
Re: Waste of time and talents...
Posted:
May 10, 2005 9:20 AM
in response to: rabbe
|
|
|
Thanks for the info. That got me curious enough to look up more details. Here's a link to the agreement for other curious folks:
https://jdk.dev.java.net/Sun_Contributor_Agreement.pdf
It still seems fishy to improve a separate product I wouldn't have rights to (i.e., the rest of the JDK), but at least Sun doesn't claim sole ownership. So it's not worst case.
|
|
|
|
|
|
|
|
|
What do we really want?
Posted:
May 10, 2005 9:54 AM
in response to: tjpalmer
|
|
|
Interesting thought...
> It still seems fishy to improve a separate product I
> wouldn't have rights to (i.e., the rest of the JDK),
Not wishing to put words in your mouth, but this statement implies that we should only contribute improvements if we are granted rights to everything that has previously been developed.
Sun has spent a lot of money developing J2SE. Let's say that I improve one method in one class... Is it equitable for me to ask for rights to everything in return?
There's a whole spectrum from reporting bugs, to fixing bugs, to contributing new features. At some point I can well see the contributor feeling ownership in the whole... but at many levels it's just a win-win proposition. I want to use a better product, so I help the vendor improve it.
Assuming that you get rights to the JDK for contributing a fix: What rights do you want?
Do you want the right to use the JDK for free?
Do you want to distribute the JDK for free?
Do you want to use the JDK's code in a different product?
In my case, I just want to use the JDK for free, so I have no beef with Sun.
You might be interested in an article by Andy Tripp (http://www.javalobby.org/articles/fixing-the-jdk/) who actually did fix the JDK... He didn't get more rights, but he did get a free T-shirt
|
|
|
|
|
|
|
|
|
Re: What do we really want?
Posted:
May 10, 2005 12:13 PM
in response to: johnreynolds
|
|
|
I didn't say Sun is obligated to release their code under an open source license. I understand their concern. And also, thanks for clarifying that point.
I'm just saying that there are other people who do release their code in an open source fashion. In the long run, it's all a guess of which contribution can give a developer (or a company) the best bang for the buck (including whatever philanthropic world views people might or might not have).
|
|
|
|
|
|
|
|
|
Re: What do we really want?
Posted:
May 11, 2005 6:54 AM
in response to: atripp
|
|
|
> p.s. My first post on the java.net forums, and I use
> it to advertise that I'm a t-shirt whore. *sigh*
Andy, to me you are a saint. Keep up the good work.
|
|
|
|
|
|
|
|
|
Re: What do we really want?
Posted:
May 12, 2005 6:40 AM
in response to: atripp
|
|
|
Andy, Many FOSS developers are so paranoid about Sun licenses I'm surprised they'd want your contributions since you've seen Sun's source code
|
|
|
|
|
|
|
|
|
Re: What do you think?
Posted:
May 12, 2005 4:31 PM
in response to: uhilger
|
|
|
i think "competition" and "freedom of choice" is a great idea. but i hate the idea of java beginning to fork. if only the JLS and API specs were under one authority it would be perfect.
|
|
|
|
|
|
|
|
|
Re: What do you think?
Posted:
May 10, 2005 3:34 AM
in response to: daniel
|
|
|
I've only run into a small number of Open Source advocates who won't use Java because of its links to Sun. They often make me wonder: at what point does Open Source stop being a philosophy and become a religion?
No doubt this new incentive will placate them, but beyond pandering to a tiny number fanatics, I'm left wondering whether the positives outweigh the negatives(?)
|
|
|
|
|
|
|
|
|
Re: What do you think?
Posted:
May 11, 2005 1:32 AM
in response to: javakiddy
|
|
|
> I've only run into a small number of Open Source
> advocates who won't use Java because of its links to
> Sun. They often make me wonder: at what point does
> Open Source stop being a philosophy and become a
> religion?
>
Pretty quickly once you get into it too deep.
> No doubt this new incentive will placate them, but
> beyond pandering to a tiny number fanatics, I'm left
> wondering whether the positives outweigh the
> negatives(?)
It won't placate them one bit. They won't be happy until and unless the JLS and JVMS (which is what Java really is...) are open source and under their direct control to fork and fork and fork some more.
Under the current system they can already create an open source JVM (and even may call it Java if they can get it to pass the TCK) but they're not interested in that.
They're only interested in destroying the Java platform which is taking market share away from their beloved Perl (which ironically is also controlled by a limited group which is less open than the JCP...).
|
|
|
|
|
|
|
|
|
Re: What do you think?
Posted:
May 11, 2005 9:03 AM
in response to: jwenting
|
|
|
I keep pushing for OS Java in the hope that maybe we can channel some of that energy into constructive directions.
OS developers, in general, are more interested in have the "right" to fork than in actually forking. Sun's ownership of the Java trademark, along with the TCK, would, I hope, keep actual forking from being a real problem.
My dream would be to see those developers eventually latch onto Java the way they're latched onto Perl now. Maybe it's an impossible dream, but a guy can still dream.
|
|
|
|
|
|
|
|
|
Re: What do you think?
Posted:
May 11, 2005 9:21 PM
in response to: jwenting
|
|
|
Under the current system they can already create an open source JVM (and even may call it Java if they can get it to pass the TCK) but they're not interested in that.
They're only interested in destroying the Java platform which is taking market share away from their beloved Perl (which ironically is also controlled by a limited group which is less open than the JCP...).
And of course you have some evidence to backup this insane conspiracy theory of yours? I call bullshit, personally. Nobody is trying to destroy Java because it's taking market-share away from Perl. If you really believe that, you need to lay down the crack pipe for a while.
|
|
|
|
|
|
|
|
|
Re: What do you think?
Posted:
May 12, 2005 12:44 AM
in response to: sprhodes
|
|
|
Given that the people calling for Sun to give up ownership of the JLS, JVMS, and (implicitly) the Java name to the GPL fanatics (meaning themselves...) are people who are admitted Java haters like ESR there can be no other conclusion.
If they liked Java so much they want to prevent it from going under when Sun meets its demise they'd be using it now and maybe creating an OS JVM (which they're allowed to do now) of their own.
After all they don't care that other language specs they use (C, C++, etc.) aren't owned by them and are in fact controlled by groups that have no community input at all.
|
|
|
|
|
|
|
|
|
Re: What do you think?
Posted:
May 10, 2005 5:51 AM
in response to: daniel
|
|
|
The big, big advantage Java has over MS technologies is that it's open, so it avoids vendor lock-in. Isn't that the assertion we always make? (Ok, it's the assertion *I* always make.)
If it's so open, implementing an open-source clone should be feasible, right? And if the open-source clone is full of incompatibilities, I guess the spec wasn't that good. Sun has always called their stuff "reference implementations", meaning: "here's the spec and here's a sample implementation, in case there's any doubt about what the spec says".
So: do we honestly believe Sun's spec is incomplete or vague? I don't. I trust Sun to write good specs. (Except JSF. >:( ) So much for the "it'll be incompatible!" argument.
Sun is going down, slowly but surely. The world is moving to cheap, Intel/AMD-based hardware platforms. The software platform of choice is either something real cheap (Linux/*BSD) or something with a huge array of available software (MS flavor of the year). Sorry, Sun. So, when Sun gets bought, what happens to Java? If Oracle buys Sun: they proceed to jam it into their database, so now you have to be an Oracle customer to use Java (after all, Java was originally meant to be embedded, right?). If IBM buys Java; they have no interest in originating solutions any more, they're really only an integrator. IBM Global Services will integrate your MS and your UNIX environemts. IGS will help you get running on WebSphere or Tomcat. They'll do whatever you want, they won't come in w/their own solution. That's their schtick. If the customer wants a 100% Wintel solution, IBM will cheerfully provide it without making a strenuous argument for avoiding vendor lock-in. It's up to the customer to originate the push to avoid vendor lock-in, and how many customers do we know like that? Sure, a few really big ones, maybe, but those are the same guys who are shipping work overseas as fast as they can, and what's the overseas skill pool? Wintel.
So, when IBM or Oracle buy Sun, whither Sun's Java implementation? Wouldn't it be nice to have a viable open-source implementation at that point?
This project will take YEARS to reach fruition, so why don't we just wait and see how it turns out? (Or pitch in, even! [Not me, I have a full-time job {at the moment} and a family, but it's a big world, I'm sure somebody out there has both time and motivation.])
I can't believe all the fear and loathing this is creating.
|
|
|
|
|
|
|
|
|
Re: What do you think?
Posted:
May 10, 2005 9:09 AM
in response to: jlusk4
|
|
|
Open source does not ensure Java's existance as we know it. Sun does an excellent job of saying what can be considered a Java runtime and what cannot.
If Sun goes down as you suggest and no one maintains the TCK or enforces the Java trade marks then incompatible JVMs could become common. Java needs a steward. It's fine for an open source JVM to exist, but it in no way ensures Java's survival post Sun.
|
|
|
|
|
|
|
|
|
Re: What do you think?
Posted:
May 11, 2005 7:11 PM
in response to: rabbe
|
|
|
This is true if we're concerned about Java's future development in the absence of Sun.
However, simply adhering to the existing specs doesn't so much require a steward (and we'd have to adhere before moving forward, I would hope). Well, it *sort of* does, but I'm thinking the open-source "street" will serve as steward.
An open-source implementation might not be a sufficient condition to ensure Java's ongoing health, but I think it's a necessary condition.
|
|
|
|
|
|
|
|
|
Re: What do you think?
Posted:
May 10, 2005 9:09 AM
in response to: daniel
|
|
|
Sun Java is nice.
Having Different versions will make us to write the applications for every version.
Instead Its better to contribute Mustang.
Sun has to Concentrate on performance of JVM and J2SDK API's in its coming versions.
Sun has done right thing in making us to contribute to Mustang .
I thing its better to contribute to Mustang rather than making another JDK with different license.
|
|
|
|
|
|
|
|
|
Re: What do you think?
Posted:
May 11, 2005 9:26 PM
in response to: venu_juit
|
|
|
Sun Java is nice.
Sure is. Too bad it's not available on every platform I might want to run Java code on. And too bad I can't port Sun's existing code to another platform and distribute it.
Having Different versions will make us to write the applications for every version.
How will it? If the only JDK you are going to support is Sun's, then just document that fact, and refuse to provide support to anybody running anything else.
Instead Its better to contribute Mustang.
I disagree. The world needs a JDK that is under a true Open-Source license. We have a few already, but none is quite complete (not enough to pass the TCK). Maybe now we will get one that is.
Sun has to Concentrate on performance of JVM and J2SDK API's in its coming versions.
Sure. And competition fosters better development from both sides.
Sun has done right thing in making us to contribute to Mustang .
Make us? They haven't "made" anybody I know contribute to anything.
I thing its better to contribute to Mustang rather than making another JDK with different license.
|
|
|
|
|
|
|
|
|
Re: What do you think?
Posted:
Jan 22, 2006 1:05 PM
in response to: sprhodes
|
|
|
I wanted to write on this topic and I kept seeing the same things posted over and over basically, so this seemed like one I could add to and maybe try to shift some direction in this over all thread. The facts are simple: The group writing a JVM should not matter. If the JVM follows the standards (VM, language, APIs, JCP, and JSR) then it should work and if not it has a bug. The bug should be attended to and fixed.
If a JVM does not follow standards then software distributors should simply explain to their users they need to be using a JVM which follows standards. Then if there is a bug in the "standard" JVM the user is using then they need to deal with the situation and explain to the user what the problem is and how to work around it and that they need to talk to their JVM vendor.
If the application has an installation for it's different platforms then one possible solution is embedding a JVM within your application. This simply requires a less restrictive distribution model for the current JVMs....let distributors distribute only what they need for their application purposes, and if distributing the full distribution doesn't matter for one project...so be it.
> Sun Java is nice.
>
> Sure is. Too bad it's not available on every
> platform I might want to run Java code on. And too
> bad I can't port Sun's existing code to another
> platform and distribute it.
I completely agree here. I think what really needs to happen is people focus on the JCP and JSRs as the driving force for changes to APIs and the way they work. Then the JVM should not matter. I also think many of the APIs need to be broken out from the standard so that when distributing a JVM with your app (to avoid versioning issues) one should be able to package only what they need in their embedded JVM. Sun, Harmony, and everyone should do all this then it won't matter which JVM is being used as long as it's version compatible with your requirement.
I think we've all lost the big picture of Write Once Run Anywhere...the standards of the JVM, APIs, and the language. As long as the standards are adhered then who ever is writing the VM shouldn't matter one bit. It's real simple when speaking of a standards based JVM and distribution: If an application doesn't run on a VM it's for one of 3 reasons:
1) A bug in the VM: Report it and follow up and explain to your users there is an issue with that VM and they need to hammer on their vendor.
2) You used an incompatible non-standard API....lets not confuse this with a VM issue...it's your applications problem.
3) The VM is not version compatible with your applications requirement. So, you either package a VM with your application install: have a sep install for your supported platforms, or have an ability to find the VM you need when your application runs...this is why we find Web Start, launcher tools (JSmooth, JNI wrappers, etc), and all kinds of JAVA_HOME schemes. WORA doesn't have anything to do with installing your application on a given computer. We need a new standard for that actually which I think there is already a JSR for.
>
>
> Having Different versions will make us to write the
> applications for every version.
>
> How will it? If the only JDK you are going to support
> is Sun's, then just document that fact, and refuse to
> provide support to anybody running anything else.
Yes. The only way that would matter is if people do what currently happens...use classes which are documented to not be available in other vendors VMs. Libraries should not be bound to specific VM implementation classes unless they have some dynamic configurations hidden behind interfaces that allow it to be configured to work with another base framework. You see this all the time with libraries (even Apache ... see XML-RPC) which rely on specific things they should not ... (Sun JSSE seems to be a usual suspect for HTTPS).
Sometimes an API can be installed separately and helps get around these types of issues....currently what could change here is that all this stuff keeps getting added to the standard distribution. Extension APIs are fine for a public VM, but what happens when someone needs the API and needs it to run on another VM version or something....too much in the standard dist and too many restrictions on what can be redistributed separately seem to be a real issue which could be better addressed than worrying about who is writing what VM.
If it's a bug in a specific JVM then it's a different issue. You could have two Sun JVMs 1.x.x_y and 1.x.x_z and the z one have a bug that wasn't in the y version...so it's really the same as the 1.x.y_y version of one vendors JVM having a bug not in 1.x.x_y of the Sun JVM. That's just a fact of life, and the Java standard was never meant to lock people into using a specific vendors JVM...even Sun details this. Bug exist and need to be addressed, and that doesn't have anything to do with vendors besides they fix the bugs and should follow the standandars and it adds to a software distributors head aches...but again life...Java is WORA, so that has to be planned for, or ship an embedded JVM with your application. WORA doesn't mean you might not have to ship a VM with your application for a specific platform....
>
>
> Instead Its better to contribute Mustang.
>
>
> I disagree. The world needs a JDK that is under a
> true Open-Source license. We have a few already, but
> none is quite complete (not enough to pass the TCK).
> Maybe now we will get one that is.
Open source or closed source or what ever....it really shouldn't matter. I think we need different VMs to ensure the future, and I think they all need to be following the standards which should be laid down by the JCP and it's JSRs. If the JCP were the driving force behind the VMs and the APIs, as it should be, then the VM vendor won't matter as the VM, byte code instructions, the language itself, and the standard APIs should be standardized. Sure if they don't fix the bugs then the vendor really matters or their development just stinks...another issue.
Again, people are confusing a VM specific bug or API with Write Once Run Anywhere and following the standards....sum it up...I do this in my code and it's not supported in these other VMs so it's not standard...no...if it's a bug ask your the VM vendor to fix it...if the API is not a standard don't use it unless you have a replacement or a backup plan or expect to force the use of vendor specifics thus giving up some W.O.R.A and relying on a specific JVM.
>
>
> Sun has to Concentrate on performance of JVM and
> J2SDK API's in its coming versions.
>
> Sure. And competition fosters better development from
> both sides.
What does this has to do with any implementation being based on the standards and why it should or shouldn't exist? This really doesn't even belong in a sensible conservation as to the reason another JVM should be created. Nobody can stop anyone from creating an application (JVM) based on available standards...did anyone stop IBM...why would they?
I think it would be nice to have an open source community working on a JVM and as many ports as possible especially as the PDAs have moved up it would be nice to write apps for Palm, PocketPC, Windows, Linux, and Mac and get pretty close to WORA. The VM isn't the issue. It's who ever is writing it needs to follow the standards and who ever builds on an API needs to be sure they are using the standardized APIs and not some vendor specific API.
Focus on the standards and each company or project focus on their own performance and application (JVM)....seems simple. This way who cares who is making a new project....use it if you like it and if it has a bug and one of your users tells you about it...then do some testing with your application and tell the user they need to use this JVM until their vendor fixes their issue...customer expectations need to be managed better in todays software market as well as installation and packaging strategies.
>
>
> Sun has done right thing in making us to contribute
> to Mustang .
>
>
> Make us? They haven't "made" anybody I know
> contribute to anything.
> I thing its better to contribute to Mustang rather
> than making another JDK with different license.
Mustang is just a JVM. So really what we need to be talking about is not whether or not someone should work on this one or that one. If a JVM wants to survive it should follow standards and work and have it's bugs fixed and attended to...period.... it's the only reason for it to exist ... to best support running on multiple platforms and follow standards and pass the TCK. Can't over state this: The JCP should be the driving force behind the APIs. If someone joins the JCP and contributes to API and language standards and implement those same standards which would be implemented in other VMs of the same version requirement then how does Mustang, Harmony, YouComeUpWithAGoodName.....matter? Their JVM should just follow the standards.
I think this entire thread needs to be redirected to talking about standards and not a specific JVM or project. If the Sun JVM has a bug on Linux which isn't on Windows or the other way around what do you tell your customer then? You tell them the same thing you would tell them if the JVM they were using were not a Sun JVM and it had a bug that wasn't in a JVM your users were using on Windows....straight forward.
Yes this has been very repetitive, but it seems it needs to be. We need to keep in mind the initial good ideas of Java and the intent of it's designers and what those things mean and that they are all driving by standards and that needs to be the focus.
|
|
|
|
|
|
|
|
|
Re: What do you think?
Posted:
Feb 14, 2006 3:07 AM
in response to: wadechandler
|
|
|
The JCP, unfortunately, is not very useful in its current form for something that important, as it is fundamentally encumbered in NDAs, and has huge transparency problems, in particular wrt to J2SE and JVM specs.
This has all been known since JCP 2.6, and two years have passed without any changes to it or a movement to fix it. That does not make me very confident about the future relevance of the JCP.
cheers,
dalibor topic
|
|
|
|
|
|
|
|
|
Re: What do you think?
Posted:
Apr 13, 2006 3:48 PM
in response to: robilad
|
|
|
hey umm u sound like u grew up with some hp. dont worry everything is great. umm unfortunately i think i should tell u i work here and im 18.
|
|
|
|
|
|
|
|
|
Ignore the whinging it is a good idea
Posted:
May 11, 2005 3:08 AM
in response to: daniel
|
|
|
hi,
independently of whether /everyone/ wants different implementations of j2se it is going to happen - project Harmony is just as much about providing a framework for different parts of these implementations to fit into and play well together as it is about producing an entire j2se implementation
imho, it can only be a good thing if it produces results
asjf
|
|
|
|
|
|
|
|
|
Re: Ignore the whinging it is a good idea
Posted:
May 11, 2005 8:11 AM
in response to: asjf
|
|
|
It is a pretty idea. Everyone major in the GNU Classpath space is moving towards componentized runtimes, and there is a lot of cooperation going in between Kaffe and gcj, like actually sharing some code.
Working out good interfaces within the Classpath runtimes to be able to share state-of-the-art components more easily has been something that's been in the air for a while. But working together with runtime developers outside of the GNU Classpath runtime community could help do a better job at it, for example, and profit from a communally maintained, pluggable core VM framework (and associated component implementations) under a liberal license themselves.
To give people a more broader picture of things, where kaffe is slowly moving to is intergrating support for gcj-ed (core, rt.jar) libraries (gcj bindings), fast interpreter engines (jamvm, and Ertl's work), simple, quick jits (kaffe), more sophisticated jits (cacao, latte), in a mixed-engine mode environment (latte), with support for isolation (janosvm), single system image execution in clusters (jessica), and so on (J2ME, RTSJ, ...). The code is out there, but it would be easier to merge it into other existing runtimes if there actually were cross-VM component interfaces, for example.[1]
GNU Classpath has succeeded in creating those for VMs for the class library space, which explains the cambrian explosion of runtimes and projects around GNU Classpath. Two new projects (aot, and WinCE) just this week.
Getting such a lively, great, diverse family of runtimes to interoperate at the component level would let us more rapidly raise the state-of-the-art for many implementations, by letting them reuse other components from implementations with different strengths. That's a pretty good way, in my opinion, to encourage reuse, rather then NIH.
cheers,
dalibor topic
[1] I.e. if you really want to see a VM that lets you run your apps on large clusters seemlessly, with isolation, you'd be having a good start by merging kaffe, janosvm and jessica into one. I'd love to see someone volunteer to do it.
|
|
|
|
|
|
|
|
|
Re: Ignore the whinging it is a good idea
Posted:
May 11, 2005 8:12 AM
in response to: robilad
|
|
|
> GNU Classpath has succeeded in creating those for VMs
> for the class library space,
... those *interfaces* ...
cheers,
dalibor topic
|
|
|
|
|
|
|
|
|
Re: What do you think?
Posted:
May 11, 2005 12:46 PM
in response to: daniel
|
|
|
Well, I'm no defender of open source religious zealots, but overall an open source JDK seems to me a good thing. The open source community has *begged* Sun to open up Java for years, and Sun has consistently refused. So, the only option for those who want or need an open source solution is to create a separate, open source project.
I think this is a positive thing if for no other reason than to make sure Java continues to be available for all of us to use freely, before Sun goes bankrupt or gets bought by another company with a less enlightened attitude towards community participation. (All who think Sun will triumph over their evil enemies, kindly disclose the amount of Sun stock you own as a percentage of your retirement portfolio.)
|
|
|
|
|
|
|
|
|
Re: What do you think?
Posted:
May 12, 2005 2:32 PM
in response to: actionboy
|
|
|
If Sun went bankrupt(which seems highly unlikely) I am sure that J2SE will be donated to the open source community. Sun has always been a company with great ethics and integrity.. they will do the right thing. Look at what Netscape did went it out of business -> created mozilla.org.
|
|
|
|
|
|
|
|
|
Re: What do you think?
Posted:
May 13, 2005 12:16 AM
in response to: actionboy
|
|
|
> Well, I'm no defender of open source religious
> zealots, but overall an open source JDK seems to me a
> good thing. The open source community has *begged*
> Sun to open up Java for years, and Sun has
> consistently refused. So, the only option for those
> who want or need an open source solution is to create
> a separate, open source project.
>
and the fact that they can shows there's no reason for Sun to give up its intellectual property...
But the zealots don't want the code to the JDK (well, not just the code) but the keys to the language specs as well. This is something they don't demand from ISO for the C++ specs for example...
The moment Sun gives up control of the JLS/JVMS to the Java haters in the OS movement (who are the ones demanding Sun gives them the keys) is the moment we should all start looking for another line of employment.
|
|
|
|
|
|
|
|
|
Some reasons for an OpenSource java
Posted:
May 13, 2005 4:21 AM
in response to: daniel
|
|
|
Here are some people like jwenting, who think, that an OpenSource Java would not be a good thing.
They say, that with an OpenSource Java comes forks and so on.
But have a look at the current situation.
I think it is very hard to port Suns JVM to other Systems.
For example have a look at BeOS:
Ok, you can say that BeOS (and Zeta and Haiku) are dead Operating Systems, but there existing no Java for it.
Before Be Incorporated died, they were working together with Sun on a Java-port to BeOS.
They _worked_ on it, but nothing is happend. But there comes up an Java-implementaion for BeOS called BeKaffe. And who wonders, BeKaffe is OpenSource.
After Be Inc. died, the Java-port stops.
Later again people have begun to port it to BeOS. First beunited.org, then YellowTab and now they working together on it. But the old beginning of a port from Be Inc. is lost, because it is not allowed to public the intermediate steps.
And there existing the same problem: There is a little closed group, which shows time by time some screenshots of theire progress. But they are not allowed to public the steps between the porting. And so, there can not people - who are not in this closed group - helping them.
Do you see the problem, which the curreent Java-licenses have for making ports?
And what somebody say about "incompatible forks" (- if Java would be OpenSource -) is clearly wrong.
The GNU Classpath people, the GCJ people, the Kaffe people and all other want to create an OpenSource Java-implementation, which is compatible to Suns Java as possible.
But to be compatible to Suns Java, it would be helpful to use Suns JavaCompatiblityKit (JCK).
The last thing, which Sun does in this area was, to put the JCK under a "read only" license (https://jck.dev.java.net/).
So, you can read the Test-suide, but you are not allowed to compile or running it.
So, with this license, the JCK is for GNU Classpath and other projects unusable.
So until today Sun don't help OpenSource Java implementaions to be compatible to Suns Java.
But hope comes now from Apache, that they want to create an Java-implementation, where they then want to pass the JCK and so theire implementation can be called "Java".
And that is an other point some people don't mentioned.
Sun says, what program can be called "Java" and what program can not be called "Java". Sun have the trademark on this word.
So, if Java is OpenSource or not: Everybody knows, that only all things which have the name "Java" are really 100% compatible to Suns Java.
So I don't know, why some people are worry about, that with an OpenSource Java would be comes incomatible Java versions.
And to my example with BeOS:
In OpenSource-projects there would be published the intermediate steps. If Java would be OpenSource, then the steps are published, but they have then an other name.
And people then can using it. Not to program, only to run some Java-programs on a new platform, where no Java exists.
But why is it better, to have on a lot of platforms no Java, instead of work in progress ports, where the programs name says, that it isn't 100% compatible to Suns Java, but you already can some programs run on?
Greatings
theuserbl
|
|
|
|
|
|
|
|
|
More reasons
Posted:
May 13, 2005 5:16 AM
in response to: theuserbl
|
|
|
Oh, and I have forgotten the other often mentioned reasons to say:
- if Sun would be go bankrupt, then the future of Java is so long ended, until anybody other have rewritten it and continue with it
- if Sun would be bought up, then the future of Java is in the hand of an other firm, which can decide to change the license of newer Java versions in a negative way
- Sun can everytime decide to change the license, that newer Java versions are no longer cost-free or so
- if anybody contribute code to Java, then at the end Sun owns it. Contributing to Java is nothing for a community. It is only for Sun.
Greatings
theuserbl
|
|
|
|
|
|
|
|
|
Re: More reasons
Posted:
Sep 27, 2005 1:20 AM
in response to: theuserbl
|
|
|
> if anybody contribute code to Java, then at the end Sun
> owns it. Contributing to Java is nothing for a community.
> It is only for Sun.
Nothing for a community? How about a new feature / fixed bug that was added to JDK as the direct result of the contribution? Or are you expecting a paycheck for fixing the bug? Will you get one working on Harmony?
Your name is in the code, on the bugfix, just not in the copyright section. Is your ego really that big?
|
|
|
|
|
|
|
|
|
Re: More reasons
Posted:
Sep 27, 2005 6:41 AM
in response to: kirillcool
|
|
|
> Your name is in the code, on the bugfix, just not in
> the copyright section. Is your ego really that big?
I recently* put a full days work in creating a really cool looking front for my car radio. Afterwards I asked my neighbor to build the new radio into my car.
The next day I learned that he instead build it into his car and said I could listen to music in his car, no problem.
When I got into things like its my ownership, my blood sweat and tears he just said he will say to anyone who asks that its my work that made his car look so much nicer.
But, but, I said; Its mine and I want to do with it whatever I please!
To which the nasty neighbor answered;
Is your ego really that big?
*) recently is relative here, more like, in the possible future..
|
|
|
|
|
|
|
|
|
Re: More reasons
Posted:
Sep 27, 2005 7:01 AM
in response to: zander
|
|
|
> > Your name is in the code, on the bugfix, just not
> in
> > the copyright section. Is your ego really that
> big?
>
> I recently* put a full days work in creating a really
> cool looking front for my car radio. Afterwards I
> asked my neighbor to build the new radio into my
> car.
>
> The next day I learned that he instead build it into
> his car and said I could listen to music in his car,
> no problem.
> When I got into things like its my ownership, my
> blood sweat and tears he just said he will say to
> anyone who asks that its my work that made his car
> look so much nicer.
> But, but, I said; Its mine and I want to do with it
> whatever I please!
>
> To which the nasty neighbor answered;
> Is your ego really that big?
>
>
> *) recently is relative here, more like, in the
> possible future..
OK, so Harmony's yours? Why?
|
|
|
|
|
|
|
|
|
Re: More reasons
Posted:
Sep 27, 2005 8:17 AM
in response to: zander
|
|
|
I think it's more like this:
Once every year and a half you get brand new shining car from GMC, with free updates every few months. It's yours to ride, build on top of it, you just can't say it's your copyright. Uh, yes, the last thing - it's free. Doesn't cost you a single cent.
Then you decide that something is bothering you, and GMC agrees that you are fully qualified to make this fix and asks one thing in return - in order to makr it available to everybody, you will have to transfer copyrights on your work to GMC. You are absolutely free to decide that these terms are not good for you, and you know it in advance. GMC can't force you make these changes. It just kindly asks that you transfer the rights to it so it can provide the fix you made to everybody.
It's everybody's car, it's free, but it's not yours. You are most welcom to invent your own vehicle, decide on the terms of use for other individuals and do whatever you please with it. You can charge top dollar for it, you can parade it in front of everybody.
Back to GMC car - you can hardly expect GMC to make you the partner just because you provided an option to paint all door knobs black. However, the manual will clearly state your name.
|
|
|
|
|
|
|
|
|
Re: More reasons
Posted:
Sep 27, 2005 3:19 PM
in response to: kirillcool
|
|
|
> Back to GMC car - you can hardly expect GMC to make
> you the partner just because you provided an option
> to paint all door knobs black. However, the manual
> will clearly state your name.
No, the question is what can GMC do in the future? And the answer is they could charge me $100,000 for that car, including my feature, if they choose.
That is what open source gives you. The guarantee of a stable future. Right now Sun gives us a great deal. But nothing obligates them to continue to do that. And that is the point. It is not an accident that they have the option to pull the rug. It is fundamental to the value of Java for them.
Which is fine for them, and for anyone who is happy with those terms. Let's just recognize what the terms are.
|
|
|
|
|
|
|
|
|
Re: More reasons
Posted:
Sep 29, 2005 2:46 PM
in response to: yishai
|
|
|
> No, the question is what can GMC do in the future?
> And the answer is they could charge me $100,000 for
> that car, including my feature, if they choose.
More accurately; GMC is able to sell my work back to me, for any price that they want.
Anyone that expects a proprietary product that is free right now to stay free forever is guaranteed to be disappointed on a long enough timescale.
|
|
|
|
|
|
|
|
|
Re: More reasons
Posted:
Sep 27, 2005 2:31 PM
in response to: kirillcool
|
|
|
I wouldn't work on proprietary software without monetary compensation.
cheers,
dalibor topic
|
|
|
|
|
|
|
|
|
Re: Some reasons for an OpenSource java
Posted:
May 23, 2005 10:19 PM
in response to: theuserbl
|
|
|
Can't get u entirely....
the reason why Kaffe or Classpath are not certified JAVA
is that they didn't pass JCK. Wouldn't the same reason apply to Harmony as well? Is Sun disallowing others to pass the test or what?
|
|
|
|
|
java.net RSS
