Java inches closer to open source

(continued from previous page)

To be sure, Sun is not the only company adjusting to open source: Giving away products away for free to developers has become commonplace. And the growing maturity of open-source software, such as Red Hat's JBoss and MySQL's database products, has driven down prices, analysts said.

Green said that the results of Sun's approach are promising. For example, Sun's Solaris operating system has been "rejuvenated" since Sun began an open-source project around it, whereas before it had been written off as a casualty of competition from Linux, Green said.

The number of subscribers to Sun's Java server suite, called Java Enterprise System, has risen every quarter over the past two years and topped one million subscribers earlier this year.

Still, Green acknowledged that it isn't always easy to see how those initiatives will drive revenue at the company.

"I know for many in the industry, the business model that enables Sun to generate revenue the context of this new software ecosystem is complex. You have to squint at it sideways to understand it," Green said. "On the other hand, companies with far fewer resources in terms of support and maintenance have figured out how to monetize this new software ecosystem."

Jonathan Schwartz, who replaced Scott McNealy as Sun CEO last month, has been a driving force behind the company's software moves. Schwartz is scheduled to speak at JavaOne on Tuesday.

Meeting grassroots efforts halfway
Even as Sun tries open source to better capitalize on its invention, the company--and other Java software providers--are facing a growing list of alternatives to Java, particularly for writing Web applications.

Java is widely used for industrial-strength corporate applications, but simpler, faster scripting languages are gaining interest among developers.

Sun's approach is to make scripting, or "dynamic," languages work well with Java. On Thursday, it said it will offer support to the jMaki project, which aims to create tools for writing AJAX-style Web applications with existing Java front-end development tools.

Still, the company continues to have a Java-centric view of the world, said Steve Mills, head of IBM's software group, which derives billions of dollars in revenue from its Java-based software.

Mills said a lot of "the action in software development" is focused on scripting and scripting "frameworks," such as Ruby on Rails, which make developers more productive. Many customers would rather use scripting tools to create a Web front-end for existing server Java programs, he said.

"There's always still work to do back in the engine room--it's not like everyone lost interest in Java and practical ways to extend it," Mills said. "But it's not the answer to every problem, and frankly this is where we've had debates with Sun. Let's not try to make Java what it's not."

All Java tools and middleware vendors face development software juggernaut Microsoft, which has a long track record of making harder programming tasks easier.

"Java has been less embraced?(in) a large swath of commercial applications in ERP (enterprise resource planning) and supply chain and such--apps that involve serious transactions and a real need to scale. Many of those apps, and the ISVs (independent software vendors) and developers that create them, are going to .Net and Microsoft," said Jonathan Eunice, an analyst at Illuminata.

The improvements in the Java Platform Enterprise Edition 5 standard, which was recently finalized, are specifically aimed at developer productivity, particularly for server-side Java applications. But many of the most popular productivity tools for developers spring from open-source groups, rather than the vendor-led Java Community Process standards body.

"What's happening is that the entire industry is having a backlash against complexity," said Burton Group's Monson-Haefel. "So JavaOne is not about the standards coming out of Java Community Process--it's about the open-source community and the de facto standards."

Joe Keller, vice president of marketing for service-oriented architecture and integration platforms at Sun, said that the Java Community Process is now able to capture many of the enhancements that spring from grassroots open-source efforts and standardize them.

"There are innovations happening in the community and the standard-setting process. It continues to happen over and over--it's a natural process of evolution," he said.

[pwawright]

Some corrections

Hi, I wanted to suggest a couple of corrections to your story. First, you write "the event won't see developers get their hands on the source code for the actual Java programming". This is not true--the source code for the development kit (the JDK) has shipped along with every Java release since it was introduced 10 years ago. And as of last year, the Peabody project on java.net allows you to see all the rest of it--including the full source code for the virtual machine--Sun's implementation, Hotspot. The issue is not releasing the source code, but the license under which it is released. Until recently, any changes you made to the JDK Java source code could not be legally redistributed; this was relaxed last year so it is legal to release it within, say, a company, as long as you assume the risk of making the changes. What some FOSS advocates want is the right to take the JDK source code (from Sun), change or "fix" it, and then release those changes to the world. That's what you can't do right now.

Second, you write, "simpler, faster scripting languages are gaining interest among developers". However, "faster" in this case means "lower cost of development" or "faster time to market". Whether the scripting language actually performs faster depends on many factors. Perl will probably be faster for string operations as it can use a highly tuned C library to that end. Whether it's faster in general, or where it performs better, is a matter of benchmarking. The issue is not that Java is slow (though it is probably slower for some purposes) but that development in Java follows a traditional design-code-compile-test cycle which just takes longer, partly because of the extensive type-checking during the compile cycle, and partly because a compile needs to complete before the program can be run. Scripting languages which support a full interpreter can be run without an intermediate compile step, which some developers prefer, as they get feedback about what is wrong sooner.

Cheers
Patrick

[pwawright]

Some corrections

Hi, I wanted to suggest a couple of corrections to your story. First, you write "the event won't see developers get their hands on the source code for the actual Java programming". This is not true--the source code for the development kit (the JDK) has shipped along with every Java release since it was introduced 10 years ago. And as of last year, the Peabody project on java.net allows you to see all the rest of it--including the full source code for the virtual machine--Sun's implementation, Hotspot. The issue is not releasing the source code, but the license under which it is released. Until recently, any changes you made to the JDK Java source code could not be legally redistributed; this was relaxed last year so it is legal to release it within, say, a company, as long as you assume the risk of making the changes. What some FOSS advocates want is the right to take the JDK source code (from Sun), change or "fix" it, and then release those changes to the world. That's what you can't do right now.

Second, you write, "simpler, faster scripting languages are gaining interest among developers". However, "faster" in this case means "lower cost of development" or "faster time to market". Whether the scripting language actually performs faster depends on many factors. Perl will probably be faster for string operations as it can use a highly tuned C library to that end. Whether it's faster in general, or where it performs better, is a matter of benchmarking. The issue is not that Java is slow (though it is probably slower for some purposes) but that development in Java follows a traditional design-code-compile-test cycle which just takes longer, partly because of the extensive type-checking during the compile cycle, and partly because a compile needs to complete before the program can be run. Scripting languages which support a full interpreter can be run without an intermediate compile step, which some developers prefer, as they get feedback about what is wrong sooner.

Cheers
Patrick

[Captain_Spock]

With All Of This Talk....

... about "Giving products away for free to developers has become commonplace...." I am very curious at this time as to why MICROSOFT, IBM and SUN MICROSYSTEMS are holding on to the OS/2 Source-Codes... when it is known that a few years ago IBM and SUN MICROSYSTEMS were engaged in a collaborative project for the "DESKTOP". Just what are NICROSOFT, IBM and SUN MICROSYSTEMS afraid of in moving the "JAVA OS" into the Open-Source Community when these companies are not actively supporting or have no interest in the OS/2 Platforms! Why not just give this "Java" OS away for free...!

[Captain_Spock]

With All Of This Talk....

... about "Giving products away for free to developers has become commonplace...." I am very curious at this time as to why MICROSOFT, IBM and SUN MICROSYSTEMS are holding on to the OS/2 Source-Codes... when it is known that a few years ago IBM and SUN MICROSYSTEMS were engaged in a collaborative project for the "DESKTOP". Just what are NICROSOFT, IBM and SUN MICROSYSTEMS afraid of in moving the "JAVA OS" into the Open-Source Community when these companies are not actively supporting or have no interest in the OS/2 Platforms! Why not just give this "Java" OS away for free...!