Archive for May, 2006

Sun Open-Sourcing Java? It’s Too Late, Baby.

Monday, May 1st, 2006

There’s rampant speculation about whether Sun will announce that it is going to open-source Java at the upcoming JavaOne conference. With Scott “I’ll open-source Java when IBM open-sources DB2″ McNealy stepping aside, and John “what about the shareholders” Loiacono now safely ensconsed at Adobe, maybe this time Jonathan Schwartz will follow through all the way and release the reference implementation for Java Enterprise Edition under some form of open-source license, as Sun has done with Solaris. And who knows? Maybe Sun will do the same for Java Standard Edition and Java Mobile Edition.

There’s just one problem: open-sourcing Java won’t change anything at this point.

I’ve been beating the open-source-Java drum for a very long time. Seven years ago (back when some of the plankholders of the JCP were bitching about Sun’s grip on the process, and Sun was arguing that it didn’t want the IP corrupted), I suggested that Sun spin off the JCP as a nonprofit and retain the rights to the Java trademark for revenue-generating purposes–as sort of a half-step toward open-sourcing.

(Ironically, Sun did eventually spin off a nonprofit out of the Java community–except it was GELC, and not the JCP.)

But now that Sun seems close to actually going all the way, it’s almost moot. Open-sourcing Sun’s Java intellectual property won’t dramatically expand the reach of Java into the developer community on the middleware side–there are already open source Java Enterprise implementations that have paved the way. And with an Apache-licensed, J2EE-certified server already available, why would anyone use a CDDL-licensed Java server instead for development (aside from it being the ‘official’ reference version)?

Besides, Java Enterprise Edition has already won the war for a substantial chunk of enterprise architecture, and it did so in great part because of the success of JBoss and Tomcat as development environments (and Eclipse, and all of the other open-source tools built around the Java standard).  On the other hand, Sun has been giving away its reference implementation for free
to developers for a while now. And despite a host of curiosity-inspired
downloads, Sun Java System Application Server Platform Edition hasn’t
exactly won Sun much more market share in the enterprise application
server space.

On the desktop, there’s not much room for celebration either.  Sure, an open-source Java Standard Edition could get bundled more tightly into Eclipse–if the licensing model allows it.  But open-sourcing Java probably won’t increase acceptance of Java Standard Edition on Linux, because Sun will most likely veer away from a GNU-compatible licensing scheme. And with Loiacono’s new home making major inroads into the mobile space with Flash as a developer plaform for phones (and fragmentation of the Java Mobile code base) don’t exactly make an open-source JME any more attractive to the mobile developer community.

So, Sun won’t be giving anything up really. Except its royalties off Java certification. But there’s still an opportunity to derive revenue–perhaps a better opportunity than there is with Java’s current licensing scheme–to get trademark licensing and support licensing revenue. Sun wouldn’t necessarily lose any more control over the destiny of enterprise Java than it has already, between the Java Community Process and JBoss and Geronimo already available as open-source.

In effect, Sun has waited until there was nothing left to really lose by open-sourcing Java. As a result, the potential impact that the move could have had, say, three or four years ago has been squandered. With nothing to lose, there is also very little besides PR to gain–other than possibly hoisting off the cost of maintenance of future versions of Java onto the Java developer community.