Sun to Change Java License for Linux

daria42 writes "It looks like the days of downloading Java every time you re-install a Linux box may be at an end. Reports are trickling in that Sun plans to alter the Java license to make it easier to bundle the JRE with Linux. From the article: 'Sun has faced calls several times to open-source Java, which advocates say would foster innovative open-source development. The company has resisted formally open-sourcing all of the Java software, but it has dramatically changed the development process around Java and changed licenses to make it easier to see Java source code.'"

Comments

Finally!

[pryonic]

It's about time they woke up and smelled the coffee!

I'm sorry...

Re:Finally!

[moro_666]

the exploding one ?

Sweet...

[racebit]

Three cheers for sun *reaches for mug java*....o wait, my self heating mug exploded

Hard..

[ZoWnX]

Because downloading the JDK or the JRE after installing linux was hard? If it wasnt for this, I wouldnt be periodically using the latest version.

Re:Hard..

[CastrTroy]

I think it has more to do with having the distro do it for you. If we want Joe User to be able to use linux for their desktop needs, then we are going to have to make it as easy as possible for them to use. Of course the people in control of the distro are the ones making the decision. If they don't want to include it because of some ideological values, then that's their business. If they feel the people using the OS has can just install it themselves, then that's their business. But if they're trying to put out an easy to use desktop distro, then they'd probably be smart thinking twice about including it.

Re:Hard..

[/ASCII]

That's the fault of lazy packagers, not a problem with RPM itself. You can specify dependencies on a particular file (like /usr/bin/java) insesad of a package if you want to. And if that's not enough (e.g. if you want to allow people installing into /usr/local/ or /opt) you can write little dependency checking scripts at install-time. For example, this snippet makes sure a few X headers are present, regardless of if they are installed in /usr or /usr/X11:

%define xinclude %( if test -d /usr/X11R6/include; then echo /usr/X11R6/include; else echo /usr/include; fi )
Requires: %{xinclude}/X11/StringDefs.h, %{xinclude}/X11/Xlib.h

Re:Hard..

[Reverend528]

Screw Joe User! Bob Sysadmin is lazy too!

days of downloading Java

[JoelMeow]

days of downloading Java

That's what you get for having a slow connection.

Re: days of downloading Java

[Arandir]

I don't care how speedy your connection is, you can only click so fast on all those Sun license agreements.

slackware has jre in 10.2?

[Edzor]

not sure about pervious versions of slack, but 10.2 ISO has it, i dont see the problem way other distro dont include it?

jre-1_5

Re:slackware has jre in 10.2?

[Ash-Fox]

Mandriva Linux includes Sun's JRE5 too.

Re:slackware has jre in 10.2?

[moro_666]

i think debian/ubuntu keep away from it because it has an uncomfortable license that doesn't match with the rest of the system ;)

most debian/gnulinux software is either gpl/lgpl or bsd (or alike) licensed, can be distributed without any restrictions just about anywhere. the license of java which you are supposed to read and accept while downloading and installnig, differs a lot from the "free as in beer" or "even more free than beer" licenses mentioned above.

for the commercial distros - no idea, but possibly the same issue.

Re:slackware has jre in 10.2?

[zerblat]

The Binary Code License Agreement, under which Sun's Java implementations are licensed, only allow you to distribute their software if "[...]you do not distribute additional software intended to replace any component(s) of the Software[...]".

That means that you can't also distribute e.g. gcj or GNU Classpath. The license isn't exactly clear on whether it means that you can't distribute Sun's JRE together with gcj or whether distributing Sun's software means you can't distribute gcj at all -- ever. It's also not clear exactly what they mean by "software intended to replace any component(s) of the Software". In the worst case, that could apply to any software that performs the same function as some part of the JVM, the byte compiler, the class library etc. Does distributing Swing mean you can't distribute GTK?

We already have open source Java

[CustomDesigned]

Sun provides an open spec (actually multiple specs). There are open source implementations of the spec. Sun and others have proprietary implementations. There are applications where the open source implementations are superior (typically small memory and embedded). If you have tons of memory, Sun's Hotspot VM is very fast.

If there are areas where the specs need improvement to get closer to the "Write Once Run Anywhere" goal, by all means complain about those areas.

We want multiple competing implementations, both open and proprietary. That said, I could see Sun open sourcing the Java libraries - at least the Java parts. The SDK comes with Sun source for the publically visible parts of libraries. However, the licence precludes using that source in an open source VM. Instead, the GNU classpath project has to rewrite them from the spec.

Keeping the Sun VM proprietary but opensourcing the libraries seems like a good compromise between maximum interoperability and competition.

Re:We already have open source Java

[The Warlock]

The problem is that the open-source implementations of Java tend to be a version or two behind the official Java, which can be a problem for some Linux users (such as CS students like myself).

Re:We already have open source Java

[Anonymous Coward]

Keeping the Sun VM proprietary but opensourcing the libraries seems like a good compromise between maximum interoperability and competition.

Which is exactly why it'll never happen. The giant Java library ensures that, even with a 100% compatible JVM, most Java applications will only run on the Sun runtime. (Especially applications that use "sun.*" or "com.sun.*" packages in open defiance of Sun themselves saying not to do that. But that's beside the point.)

Sun learned from their experience with Microsoft. When Microsoft had their own JVM implementation, Microsoft added various extra libraries and functionality to their runtime that Sun was missing. Sun responded by sueing, and forcing Microsoft to remove their JVM from Windows. (And then Sun responded to that by sueing Microsoft for removing their JVM from Windows...)

By having a massive and hard to implement class library, Sun ensures that anyone else trying to create a compatible Java runtime will always be playing catch-up. They'll never be able to be 100% compatible with the "latest and greatest" Java runtime.

And that's just the way Sun wants it - they're able to keep control of Java that way. Open sourcing the libraries would cause them to lose that control. And that's why it'll never happen.

(It's worth noting that the source for all the classes that appear in java.* and javax.* are, indeed, available with the Sun JDK. However, the license prevents doing anything useful with them, and much of those classes rely on large amounts of code implemented in com.sun.* or sun.* packages, for which the source is NOT available.)

Re:We already have open source Java

[Tim C]

Microsoft. When Microsoft had their own JVM implementation, Microsoft added various extra libraries and functionality to their runtime that Sun was missing. Sun responded by sueing

MS added classes to the java.* package hierarchy, in contravention to the terms of their licence. That's why Sun sued. Had MS put their classes in a com.microsoft package hierarchy like you're supposed to, Sun wouldn't have cared (or had a leg to stand on).

The restriction was/is in the licence to prevent exactly what started to happen - people started using the classes, and thus were writing code that could only run on MS's VM, which is completely against the core Java ethos of "write once, run anywhere". (Ok, so in practice that's often easier said than done, but this was threatening to make it completely impossible)

For what it's worth, MS didn't have to stop shipping a VM with Windows; they just had to stop shipping their non-compliant VM. They were perfectly at liberty to remove the offending classes and continue developing a compliant VM. Instead they chose not to do so, shifting their efforts to .NET instead.

Especially applications that use "sun.*" or "com.sun.*" packages in open defiance of Sun themselves saying not to do that.

That's a really dumb thing to do if you care about cross-release compatibility. There's no guarantee whatsoever that classes that are present in one release will be present in the next.

Re:We already have open source Java

[eddeye]

which can be a problem for some Linux users (such as CS students like myself)

As a CS student you should be able to get the sun JRE/JDK going on your linux box. (I speak as a former CS student who just did it... (again)...)

As a former CS student *and* instructor, take my advice: run away from Java as fast as you can. I'm not saying it's a bad langage/environment or doesn't serve some audiences very well. But Java's like cigarettes, starting on them too early stunts your growth.

CS students need to learn as many different programming approaches and concepts as they can. Procedural languages (C et al), iterative (generators, Python/Ruby), functional (lisp), declarative (prolog), message passing, object oriented, generic programming, closures, static vs dynamic typing, etc. Breadth of exposure to different approaches is crucial to knowing what approach to take with real-world problems. This should be coupled with a depth of understanding of what the system does 'under the covers' at each level. It makes all the difference in the world when facing unexpected problems and differentiates a code monkey from an engineer.

Unfortunately Java covers only a couple of these areas and none of them particularly well. Standardizing classes on Java is one of the worst things a CS dept can do. If you're stuck in this boat, all I can suggest is play around with other languages every chance you get.

Sun's commitement?

[SWroclawski]

I remember hearing about two or three weeks ago that Sun said it was committed to "Open Sourcing all of its software, everything they make."- this is from LugRadio and a Sun representative.

Given this /very/ progressive stance, I don't see why they're stalling when it comes to Java.

If anything, this slows Java adoption.

Java was all the rage in the late 90s. Had they made it Free, I think it would have been a tour de force. Now we see competition from simpler technologies. We're learning that we don't need a J2EE infrastructure when a simple Model-View-Controller model with a database backend will do the job just as well, and so on.

Freeing Java would spread adoption, if nothing else than by including it in every distribution shortly thereafter.

This new license system isn't good enough, it'll just frustrate people.

Re:Sun's commitement?

[fossa]

(Luke Skywalker stands hands bound on the edge of a plank)

"Java! This is your last chance. Free yourself, or die."

(laughter ...but Java the Hutt will soon learn he's too sluggish as he is choked to death by Princess Ruby)

Limits?

[J.Y.Kelly]

Unfortunately the article is a bit light on details. It says that Sun are going to make the JRE easier to redistribute but that on it's own isn't enough for many distros. It would also have to be at least able to be repackaged (so it goes somewhere more friendly that the Sun supplied RPM) and preferably modified (to make it play nicer with the rest of the system) before it's really useful.

Also, it's a shame it seems they're only going to include the JRE. Nice and easy for linux users to run java programs. Shame they won't be able to write any...

Re:But what about Windows?

[Ash-Fox]

Don't worry, there will be .NET instead.

Java as electricity

[aphaenogaster]

Odd analogy, but I guess it kind of makes a little sense maybe... http://www.forbes.com/2006/05/04/sun-microsystems- schwartz-cz_ec_0504schwartz.html?partner=yahootix In shwartz's words...

Forbes:

You're trying to woo customers with free hardware. How do you make them paying customers? You haven't monetized Java proportional to what's out there.
JShwartz:

That's a misnomer. Largely an American misnomer. Nearing 1 billion Java handsets.

Forbes:

So what's your Java revenue?

JS:

Close to $13 billion.

F:

That's not money in Sun's pocket, though.

JS:

It's like asking a company that produces generators how much of their demand comes from people using electricity. It's 100 percent.

F:

But it's about how many customers are paying you for the privilege of using Java.

S:

And I'll point out that a billion handsets fuels an enormous market in the telecommunications industry. Java running on Sun's Java Enterprise system, whether it's at American Express or General Electric or Vodafone, is fueling Sun's overall revenue. Asking us how much money we make on Java is like asking Verizon Communications how much money they make on handsets. The fact is that they lose a fortune on handsets, but they make a fortune in subscribers.

F:

So are you going to convert Java users to subscription service for Sun?

S:

Partially, we're already doing that. American Express runs on the Java Enterprise system. That's per employee subscription for core middleware for Sun. My broader point is that Java ensures Sun has access to an open market. Java allows us to reach out to customers who don't run on Sun hardware and ensure we can serve them wherever they may be--whether it's on a Dell box or HP box or in an IBM customer base.

Again, it's hard to explain to people. Here's an analogy. With the advent of electricity, Thomas Edison tried to patent a lightbulb so that you would have to use his lightbulbs if you used his dynamo. That strategy obviously failed. And what emerged was the standard plug. Asking Sun the value of Java is like asking GE--which is, I think, the largest manufacturer of power turbines in the world--what the value of the standard plug is. It ensures they can serve a global marketplace. So if you asked them what's the value of the plug, how would they respond?

Here are some stats on Java: There are more than 1 billion Java cards in the marketplace, securing everything from set-top boxes to handsets. There are more than a billion Java handsets, all driving demand for network infrastructure. There are nearly 1,000 members of the Java community process, who collectively contribute to the standard called "Java." It is the default standard for set-top boxes in Brazil. So what will the infrastructure opportunity be in Brazil to serve 100 million Java-enabled set-top boxes? I promise you it will be enormous, and Sun will be among many participants that can serve that demand.

Programs, getcha programs here!

[Chris Pimlott]

Section 5.3 of the Debian Java FAQ sums up the present licensing issues that prevent Debian from including Sun Java.

I wish Java was more like CPAN

[johnnnyboy]

The more I hear calls that Java to be more open source the more I wish all these Java libraries worked like the way CPAN does.

CPAN is great and its what keeps Perl relevant and it works well for the Perl community. All these java libraries bundled with the JDK should be more modular with a lean core distro and then the rest can be organized and installed as modules.

And like everything CPAN all these modules will be peer reviewed by other Java developers in the open source and corporate worlds.

Ah, one can only dream.

Well I do declare! (as grandma said)

[davecb]

Now that they've released their newest chipset under the GPL, they're going around and looking to see how much they can loosen their other licenses.

Java, due to MS's efforts to subvert it, is probably the hardest to free up, but this is a good, workmanlike step in the right direction.

--dave

Re:Well I do declare! (as grandma said)

[davecb]

Because MS tried so very hard to embrace and extand Java, it's reasonable to expect that the Sun lawyers are going to be very reluctant to change anything, for fear of starting the war back up.

Sun engineers, on the other hand, arguably agree with your desire to move to free reference implemntation, and in the short term, to fewer restrictions on the JRE.

--dave

foster?

[dghcasp]

... advocates say [open-souring Java] would foster innovative open-source development.

Because there are so few innovative open source java projects right now? Heck, I can hardly keep track.

Leaving aside the politics of open source, and the "I can't play with your toys" argument, the main issue here seems to be the license incompatability that keeps Java from being bundled with the 267 different Linux distributions.

If people want to be innovative, how about working to unify the basic functionality of all those distributions, specifically one common, simple way that works on all distributions and architectures to install 3rd party packages, like, say, Java?

ObMetaDig: And besides, why do you care? Every time I see java on /., the whole thread seems to be "it's slow / no it isn't / GC sucks / no it doesn't / .NET rules / no it doesn't"

64-bit?

[escay]

well they could start with providing the mozilla-firefox java plugin for amd64 systems on linux...libjavaplugin.so, anyone?

I'd Be Happy

[Greyfox]

If they'd just fix some broke-ass things about the language. Seems like every time I run up against a limitation in the language, I find a bug open from 1998 complaining about the problem and either closed wontfix or "We'll fix that in a future release of java" and then they don't.

Three recent thorns in my side:

The Process object's destroy method sends a SIGINT or some such rot to the child process, which may or may not kill the child process. There's no way to send a SIGKILL, no way to get the PID of the process, no way to set the process group and no way to get or kill children of the child process.

There's no way to get OS-Specific permission settings on a File. For that reason if you try to archive some files in Java using an InputStream that takes Files, you'll lose the permissions settings on them and the files will restored with something both generic and useless like 644. They make a halfhearted attempt to address this in 1.5, but it's still useless.

It would appear that the only way to get disk space left on the volume is to open a file and start writing 1 byte at a time until you get an IO Exception.

It's deficiencies like this (And the ~50MB VM overhead) that make Java a poor choice for system programming tasks, but the robustness of the language design itself could be so easily changed to address these issues. The fact that it hasn't and that all of these issues have been around for over half a decade lead me to believe that Sun isn't really serious about the language and probably shouldn't be in charge of the standard, either.

Re:I'd Be Happy

[LarsWestergren]

There's no way to get OS-Specific permission settings on a File. For that reason if you try to archive some files in Java using an InputStream that takes Files, you'll lose the permissions settings on them and the files will restored with something both generic and useless like 644. They make a halfhearted attempt to address this in 1.5, but it's still useless.

It would appear that the only way to get disk space left on the volume is to open a file and start writing 1 byte at a time until you get an IO Exception.

These two are finally fixed in Mustang. I agree it has taken long though:

    Three new methods have been added to java.io.File class:

          getFreeSpace()
          getUsableSpace()
          getTotalSpace()
[...]

  Changing File Attributes:

  In Mustang the java.io.File API provides access to the file attributes for changing its readability, ability to write and ability to make it executable. Check out the following methods for playing around with file attributes:

  Changing readability: owner-only, owner or everybody
  Making it writable or read-only: owner-only, owner or everybody
  Making it executable or not executable: owner-only, owner or everybody