Java Unbound

Tim Bray talks about some back-door strategies to pump more excitement into Sun's Java offerings. He lists some specific suggestions:

• Get serious about dynamic languages
• Be the first with the most on syndication technology
• Ship the plumbing to make it easy to build that kind of app and talk to that kind of app.

Bill de Hora chimes in with an even more compelling list:

• Get behind Jython.
• Tighten the J2EE/JCP feedback loop.
• Provide leadership for web frameworks.
• Get the SCSL out of the way.
• Support the Eclipse consortium.

I've been thinking about this subject a bit. Based on my feedback on the "101 reasons why Java is better" and the polls on this site, here's what I think really needs to be done:

1. Support the Eclipse consortium - You have to have blinders on not to see this. There's no way in hell that NetBeans, despite Gosling's stewardship, will ever catch up to the Eclipse. Sun should either port all NetBeans modules to Eclipse or acquire JetBeans and build of a more powerful foundation. The former option seems less risky than the latter.
2. Support alternative VMs - Let's face it, different contexts require different VMs. The most obvious reason why PHP lords over Java in the open source web application space is because it's cheap to host. It's cheap to host because its footprint is much lower than Java. The way around that are tools like MOD-GCJ. What locks people in to Java is not the Sun J2SDK, rather it is the Java language. Promote the language whether it be compiled (i.e. GCJ) or emulated on top of dot net (i.e. IKVM). See this proposal for more details.
3. Ship the Isolate API - It's high time that Java provide better manageability. What differentiates an enterprise app from a hobby project is reliability. JDK 1.5 has a host of new management functionalities, unfortunately it didn't incoporate the Isolate API proposal. You can't get reliability if you can't have modules crash gracefully. The Isolate API in additional opportunity to cleanup the Class loading mess.
4. Clean up the J2EE mess - Promote more simpler and powerful technology like Spring (i.e. lightweight containers) and Hibernate (i.e. O/R persistence ).
5. Build easier Java to OpenOffice Integration - You would think that Sun would have emphasized this more when they purchased Star Division. Unfortunately, despite the existence of Java APIs in OpenOffice, there has been practically no activity in this front. Where is the fledgling community of OpenOffice plugin developers? I feel that the developers at OpenOffice have a intrinsic disdain for Java and have made the absolute minimum attempt to incorporate it into their code. Well unfortunate for Sun, IBM is eating Sun's lunch by componentizing OpenOffice inside Lotus Workplace. The key for Sun to understand is that they have to make it easier for Java developers to build on top of the OpenOffice platform. Eclipse has been effective in promoting participation, but by contrast OpenOffice has failed miserably in this effort.
6. Build a meatier standard web application stack - Let's get real, to do anything worth while you have to go beyond Web frameworks, you have to add real functionality that people can re-use. Take the Eclipse platform as an example, sure there's a framework inside it, however, it is a real application that people have extended in all sort of direction. The Java world needs more beef on the stack for web applications that they can reuse out off. I want a content management, worflow, collaboration, portal, c2b, b2b engine all cohesively integrated that I can extend right of to build my app. Pick out a bunch of them that are already out there, consolidate them under an umbrella and provide a "architecture of participation". Lets face it, scripting is only as good as the foundation you are writing on top of.

Sun's current strategy is to make Java more accesible by building easier tools (i.e. Rave). It is too early to tell if this is going to work, however based on my own polling, it doesn't seem to be drumming up adequate interest. Easier tools is one strategy, however expanding the areas where different kinds of developers can participate in the platform should be given equal or greater emphasis.

In summary, Sun should eat their pride, support alternative platforms, support crash-only software, clean up their own mess, discipline its OpenOffice developers and provide more meat to the web development table. To indirectly expand revenues you expand the market by freely encouraging participation. Restricting usage and delivering half-baked usability are two sure fire ways to squelch enthusiasm.