In an interview with ComputerWorld, Simon Phipps, the chief open-source officer at Sun Microsystems reaffirms Sun's commitment to OpenSource. In the interview, Simon throws cold water on critics by pointing at a mere portion of the support that Sun has offered the open source world.

if you look at the history, it's pretty hard to sustain that [Sun isn't doing enough with open-source] as a position.

The meat of the article surrounds Simon's efforts to fully open source Java. In the interview Simon points out that many problems need to be resolved before Java can be open sourced.

If I could snap my fingers and make it happen tomorrow, I would. It's not a simple endeavor.

To change the license to open source Java, Sun first must deal with ownership, legal, access, encumbrances and relationships with Java licensees. It took Sun a full five years to solve these issues with Solaris. However Simon predicts that it won't take anything near this amount of time to complete the task with Java. One of those that have spoken out against the open sourcing of Java is none other then James Gosling. In a conversation I had with James, he stated that he felt that Java was effectively open source. With Sun allowing an unprecedented access to the Java source code it is hard to argue that point with the "father of Java". However this apparent disagreement on the future of Java doesn’t necessarily put James and Simon at odds. James is terribly concerned about Java remaining "write once run anywhere". It is the ability to fork incompatible versions that has Mr. Gosling worried. Although Simon has spoken about the economic disincentive of forking in his keynote at TSSJS-Europe, he also is concerned that Java remains WORA. In another interview, Simon spoke about how Java must remain compatible and that no company should be able to use its might to change that. Though James feels that Sun should maintain stewardship of Java to ensure compatibility, Simon feels the best way is in proper licensing and governance. Clearly Simon Phipps is a man whose vision has earned him many accolades in the industry and if he is right on this question he will earn even more. The question to be answer is; do we need protection from incompatible forks and if so, is proper licensing and governance enough?

I agree with James gosling about not to open source java. we would end with with thousands of versions and increase incompatibitlity etc. the very basic principle of "wirte once and run everywhere" will be gone. We can already see the extreme of open source where in there are thousdands of half baked ideas on sourceforge etc where in people have posted any crap they have developed in their garage and created so much confusion. We see a new web framework almost every week posted on TSS. . We have got to have one place to go to for standard java library. One company has to hold the real stable version. And in any case I highly doubt that any big organization will move into an open source version of java instead of sun built/suppored version. the language is very flexible and open to use . It supports almost everything in OOP. what else do you want ? Build what ever you can with it - why do we need to open source the very base of it. The gains arent that much compared to the confusuion you will create in the community. and we have to draw a line .. otherwise lets contact intel and ask them to open uo their microprocessors instructions set. ;)

I agree with James gosling about not to open source java. we would end with with thousands of versions and increase incompatibitlity etc. the very basic principle of "wirte once and run everywhere" will be gone.

This is a silly argument... Look at Eclipse... How many forks are there? Look at Linux... There are forks but 99.9% of cases an i386 binary will execute on any of the forks. Also there are already forks of Java where the Write One and Run Everywhere is broken. With IBM's and BEA's VMs there is a 99.9% of compatibility... There are always few tricks here and there.

Look at Eclipse

I looked...

• 5 releases in 13 months
• no decent plugin manager
• really goofy plugins (XDoclet required for WST, c'mon)
• no backward compatibility for plugins
• no coherent Pull-Down & Popup Menu synchronisation
• SWT
• (list goes on...)

all good reasons to leave Java the way it is. I use Eclipse and IBM's Rational products daily (standard at the office), and I can say I really enjoy tweaking all day with my IDE instead of coding - makes me look like the hero when someone else runs into the same problem & I can solve it.

I can agree on the backward compatibility of plugins, that's surely very difficult to work with. But: * What is wrong is with a lot of releases? If you take out the milestones we are talking about a lot less releases. All of them where of really high quality. * The plugin manager is decent (version 3.0+), but I know that this is not the case with the older version YOU are using. I am also a ex-WSAD developer. * SWT... what can I say? This is why I like Eclipse. It's fast and it looks native and when developing with it it is just like coding win32 (easy to understand unlike the weird object model of Swing).

I agree with James gosling about not to open source java. we would end with with thousands of versions and increase incompatibitlity etc. the very basic principle of "wirte once and run everywhere" will be gone.

This is a silly argument... Look at Eclipse... How many forks are there?
Look at Linux... There are forks but 99.9% of cases an i386 binary will execute on any of the forks.

Also there are already forks of Java where the Write One and Run Everywhere is broken. With IBM's and BEA's VMs there is a 99.9% of compatibility... There are always few tricks here and there.

how many organizations have bought and customized their own version of linux ? All they do is go to red hat or some other big vendor to provide a "consistent, supportable" version of linux. you cant survive wihout one. Its not your garage running a small funky website. Same argument applies to eclipse - if you go to sourceforge or plugins.eclipse website you will find so many plugins - but a very few are developed and supported over a year's time, not all are version compatible. I think for a lot of folks out there open source is a fashion. But good that there are some smart folks like Gosling who are involved in making these big decisions and not just open source fanatics running amok.

how many organizations have bought and customized their own version of linux ?

Surprisingly, many do. I see it all the time at various customers of ours (including banks, financial services houses, big online retailers, etc.) Peace, Cameron Purdy Tangosol Coherence: Clustered Shared Memory for Java

how many organizations have bought and customized their own version of linux ?

Surprisingly, many do. I see it all the time at various customers of ours (including banks, financial services houses, big online retailers, etc.)

Peace,

Cameron Purdy
Tangosol Coherence: Clustered Shared Memory for Java

I wonder how they support their own customizations ? Depend on their own employees ?

I wonder how they support their own customizations ?
Depend on their own employees ?

I don't think that the customizations tend to be that "deep". Most of these big companies will adopt a standard hardware platform (e.g. only two or three different server configurations for an entire data center), and then they'll tailor and tweak a Linux build just for that specific hardware and the tasks that it has to do. The amount of C-level code changes are going to be minimal, but they still tend to get locked into a particular combination of versions of various pieces of software (kernel, gcc, etc.) that they have gotten comfortable with, and refuse (or have problems) moving off of. Peace, Cameron Purdy Tangosol Coherence: The Java Data Grid

but they still tend to get locked into a particular combination of versions of various pieces of software (kernel, gcc, etc.) that they have gotten comfortable with, and refuse (or have problems) moving off of.

Peace,

Cameron Purdy
Tangosol Coherence: The Java Data Grid

Strangely enough I have seen this done with Windows also and all in the name of security. regards, Kirk

I wonder how they support their own customizations ?
Depend on their own employees ?

I don't think that the customizations tend to be that "deep". Most of these big companies will adopt a standard hardware platform (e.g. only two or three different server configurations for an entire data center), and then they'll tailor and tweak a Linux build just for that specific hardware and the tasks that it has to do. The amount of C-level code changes are going to be minimal, but they still tend to get locked into a particular combination of versions of various pieces of software (kernel, gcc, etc.) that they have gotten comfortable with, and refuse (or have problems) moving off of.

Peace,

Cameron Purdy
Tangosol Coherence: The Java Data Grid

so they still mostly depend on vendors like red hat to help them for majority of linux support.

so they still mostly depend on vendors like red hat to help them for majority of linux support.

Yes, I think that is a fair statement. In fact, I would suggest that their customization of a Linux distribution tends to make them even more dependent on the distributor of that particular distribution, and explains how RedHat has been able to make a profitable business on top of a commodity product that is downloadable for free. Peace, Cameron Purdy Tangosol Coherence: Clustered Shared Memory for Java

I agree with James gosling about not to open source java. we would end with with thousands of versions and increase incompatibitlity etc. the very basic principle of "wirte once and run everywhere" will be gone.

I tend to agree with this as well. However, I have to ask how languages such as Perl and Python seem to exist without thousands of forks being made while staying open source? Plus, is open source really the "holy grail" that we as developers should be seeking after?

I tend to agree with this as well. However, I have to ask how languages such as Perl and Python seem to exist without thousands of forks being made while staying open source?

Simple. There's no commercial pressure to make a better Perl or Python. Perl and Python are popular, and in use in a gazillion projects. Some very large sites run Perl (e.g. Slashdot). No doubt they are in use within probably most every organization on the planet that isn't 100% Microsoft. But the number of organizations running mission critical high load applications on Perl or Python is very small. No doubt they exist, but the vast majority are running Java or Microsoft. IBM, BEA, and Apple all have competing JVMs. Look at Azul Systems dedicated server. I don't see them doing a Perl or Python version any time soon. And the simple reason is there's no money in it. And there's no money in it because where the languages are being used, while obviously important, are not in the typical performance or management sensitive areas that Java and .NET are being deployed. The closest they have come to a fork was Active System ActivePerl and ActivePython ports to Windows, which offered dev environments on top of the actual languages themselves, running on top of Windows. More important in the past when it was more difficult to port Perl and Python to Windows, less so today. So, you will get a fork when someone can add value to the system and they wish to leverage and exploit that value that they can add.

The use of open source software in a commercial project, which is my primary concern, is a tricky issue. One of the obvious main bonuses of open source software is that you can change the code yourself if it doesn't do what you want it to do. This can mean that you want to add a feature that's not yet implemented, but in my experience, more often means you can fix a bug that the developer missed. More often than not, the practical upshot of this is that the customer is not paying you for time you spend getting the framework of your application to work - every hour spent trying to get XA Transactions working in an open source message queue (for example) is an hour you're not getting paid for. Now, I'm anticipating a counter-argument here, that commercial software can also suffer from quality issues and that argument is 100% correct. The point is that when you license commercial software as part of your stack (DB, App Server, MQ, whatever) you can persue the vendor and the customer can be made to appreciate the process. You usually have some form of warranty and support procedure that provides recourse, legal if it gets that far, that can be used in mitigation of any risks involved in procuring a commercial technology. In an open source context, this kind of protection is often absent and any resultant risks cannot be ignored from a commercial standpoint. I'm not anti-open source, far from it. Some of my best friends are open source frameworks - I just think that the same commercial rigour should be applied when selecting an open source component as a commercial one. I don't fancy writing a business case for the use of an open source VM, when there are commercial alternatives offering professional support (OK, stop sniggering at the bug parade - you know what I mean!). I think that in many cases it's up to us to educate management and customers that open source software is not free, it just has a different cost profile to most commercial products. I just know I'm going to have nightmares about "incompatible forks" tonight. ;¬) -Justin.

Sun pushes for open sourcing Java because they don't have enough resources to respond to new requirements and changes fast enough. Also Apache-Harmony is coming and they have to worry about that. IMHO, Sun is trying to avoid what happened to Solaris. There was no other choice for them but open sourcing when Solaris lost the market to Linux. All in all, I do agree with Gosling and worry about compatibility issues!