John D. Mitchell's Blog

IBM's open lettter to Sun: Open-Source Java

In response to Sun's Simon Phipps' rant about Eric Raymond's open letter to Sun to open-source Java, IBM Vice President Rod Smith has called for Sun to open-source Java in a letter to Sun Vice President Rob Gingell. Smith offers IBM's support to create an independent organization to control Java.

Comments

• Open source, closed spec?
  Someone will have to educate me on this issue. As a developer, particuarly a cross-platform developer, I really have one concern: don't let Java fracture.

  Yes, Java can be write-once, test everywhere, but that seems to be less painful than it used to be and it's still a heap better than write-everywhere. If Java becomes open source with 275 zillion implementations of the VM, what does that do to me, the developer? Java doesn't really become standard anywhere as there's no reasonable guarantee each of these implementations is fully compatible.

  As a customer, what's the advantage of having 275 zillion JVM's available? Do I, Joe Customer, really care?

  Is it possible to open Java, but only allow releases of JVM's that meet some Sun/IBM set of tests?

  Again, I'm largely ignorant here and I'm simply approaching it from my point of view. Java's already free for me, what does making it open source gain me?

  Posted by: ckessel on February 26, 2004 at 11:35 AM

• Vendor Lock In
  As a customer and developer, do you want to have to rely on a JVM from Sun?

  Or would you rather have an open source JVM that can be ported to multiple platforms consistantly that Sun doesn't have the resources for or doesn't care about?

  Forget the idea of 275 zillion JVMs. How about just ONE JVM that you don't have to rely on Sun for the success of your product and you can distribute freely.

  Posted by: dukefetish on February 26, 2004 at 12:59 PM

• Vendor Lock In
  >As a customer and developer, do you want to have to rely on a JVM from Sun?

  I believe HP, Apple, and IBM all supply the JVM themselves (porting Sun's implementation)? Yes, no? So, it seems to me there are already ways to create alternative implementations.

  I'm not sure what making Java open source adds to that. I suppose I don't really care if it's open source or not, what I care is that any implementation must by certified to meet a common specification before it can be unleashed on the world (e.g. no fracturing).

  I'm playing devil's advocate here to some extent to draw out some real information. I'm not really opposed, I just have yet to hear any real lists about the pros/cons of such a move for customers/developers.

  Posted by: ckessel on February 26, 2004 at 01:42 PM

• ust the Java J2ME,J2SE,J2EE Libraries
  posted before to Russ Miles's Sun and Open Sourcing, the debate continues...

  It would benefit the entire Java based industy, including the free software, open source and proprietary based vendors, to open license the core J2ME,J2SE,J2EE libaries and Java to bytecode compilers.

  Java's primary strength, the ability to write code which is constantly portable across many vendors platforms, would be greatly enhanced if all of vendors were using the same core libaries.

  To insure that the standard base core would not become polluted with incompatable forks, the source could be licensed with a clause requiring any incompatable changes or any additional classes or methords to be moved to and occupy only the vendors namespace. Another clause would require that the vendor version of Java bytecode compiler and any GUI IDE defaults to generating portable bytecode, without embedding any vendor specific references.

  Contributions to the core standard would be required to licensed under the same open source license. The existing JCP standard body could decide what becomes part of the Open Java Core.

  It should not be necessary to open source license Sun's JVMs. In the long run it could greatly benefit Sun to develop the JVM under a dual license as it doing with OpenOffice.org and selling StarOffice.
  

  Posted by: nzheretic on February 26, 2004 at 02:55 PM

• Vendor Lock In
  You are correct.

  But that's not always a good thing. You are just trading one evil for another.

  For example the Mozilla Mac developers are having a hard time finding the 'secret' API's to Apple's 1.4.2 JVM. That wouldn't be an issue with an open source JVM.

  I can't find the bugzilla reference about this right now.

  Posted by: dukefetish on February 26, 2004 at 03:23 PM

• Vendor Lock In
  Well, that's sort of my point I think. There aren't supposed to be secret API's. You make Java open, some companies will extend the API in secret non-compatible ways.
  

  Posted by: ckessel on February 26, 2004 at 03:43 PM

• Sun to meet with IBM?
  Eweek is claiming that an unnamed Sun spokesperson has told them that Sun representatives will meet with IBM folks to discuss open-sourcing Java.

  Posted by: johnm on February 26, 2004 at 03:50 PM

• Uncertified implementations can't compete
  "some companies will extend the API in secret non-compatible ways."
  And if they do that, will they still get Java certification? And if incompatible extended APIs don't get Java certification, will anyone use them instead of the 'real' Sun/IBM-led Open Source Java project?

  Posted by: philwebster on February 27, 2004 at 02:36 AM

• Ubiquity, dominance, standardisation.
  "I'm not sure what making Java open source adds to that. I suppose I don't really care if it's open source or not, what I care is that any implementation must by certified to meet a common specification before it can be unleashed on the world (e.g. no fracturing)."

  Any Open Source implementation would have to pass the TCK before using the 'Java' brand name. If it isn't compatible, it isn't Java - regardless of the license that the code is released under.

  The advantage to having Open Source JVM and API code is that it becomes a lot easier to extend Java's ubiquity onto exotic platforms. OpenBSD still doesn't have Java, FreeBSD only has 1.3, GNU/Hurd may have a JVM one day too - but these situations can improve more quickly only if Open Source Java implementations are available.

  Open implementations could end up on consumer devices such as games consoles too. Now *that's* ubiquity!

  And just to reiterate - no, Java cannot be forked.
  For every implementation, whether Open Source or not: If it doesn't pass the compatibility test, it isn't Java.
  

  Posted by: philwebster on February 27, 2004 at 02:50 AM

• Open source, closed spec?
  "As a customer, what's the advantage of having 275 zillion JVM's available? Do I, Joe Customer, really care?"

  The point is not to get 275 zillion VMs, but to have one standard, Open and portable VM on 275 zillion platforms, past, present and future. Games consoles, PDAs, countless OS platforms, obscure hardware.

  **Total Ubiquity**, and with it the benefits of the Java language for all. It's happening with mobile phones, and now with Open Source J2SE, it can happen everywhere else.

  And then 'Joe Customer' really WILL care.
  

  Posted by: philwebster on February 27, 2004 at 02:59 AM

• Efficiency
  "...it seems to me there are already ways to create alternative implementations."
  This is true, however it is a monumental waste of effort for all those talented programmers to have to 'reinvent the wheel' so to speak, time and time again. Wouldn't it be better if they were spending their time fixing bugs?

  With an Open Source / Free Software JVM and API, Java could extend its ubiquity. Who knows where that could lead...

  Posted by: philwebster on February 27, 2004 at 03:05 AM

• Vendor Lock In
  Now you've taken an exit from the topic of open source and are driving down the path of licenses.

  For example if code is licensed under the GPL you aren't allowed to go off and make secret changes.

  Posted by: dukefetish on February 27, 2004 at 09:24 AM

This work is licensed under a Creative Commons License.