Ever since the Apache Harmony Java project was announced, I felt the future of Java lied within it. While there appears to have been a change in attitude of Sun, in the last three years, it’s debatable to associate the change to Harmony or the seemly aggressive and responsive CEO (Jonathan schwartz) . Quite a lot has happened, with Redhat’s icetea project passing Java’s JCK recently, there is no doubt a hive of activities within opensource Java. My interest rest is in the future of Harmony Java. Stefan Krause recently posted a performance benchmark of different Java implementations and c, and the results were kinda interesting. I constantly check the performance page of Harmony and it appears the page hasn’t been updated in the last six months. Openjdk, from which icetea is derived, got about 94% of its code from Sun’s donation and it has taken about 18 months to have a release. Harmony which is a clean room implementation, with a few code donations has been able to achieve over 97% completion of its Java implementation in about 3 three years. This is certainly impressive. With the recent announcement of the icetea project, the Linux community is most likely to rally around that implementation, with derivatives as the default Java on most Linux distros.
Now back to the benchmark. It was surprising that not much difference was seen between IBM’s Java implementation and Harmony. Commentaries about IBM’s JDK and Application server offerings around the blogosphere haven’t been too encouraging. But, hey IBM is still a great company. I simply wonder why they wouldn’t just open up their implementation to the prying eyes of the opensource community, that would drive innovation or better performance implementation. Some would argue that performance lies more in reliability/stability than in speed. But speed is certainly a concern. Even Sun’s openjdk is beginning to benefit from Harmony. David Dagastine has encouraged the harmony team to keep the optimizations coming. There are already efforts to port LLVM, which appears to have the very good performance, to openjdk/icetea. With no particular Big company backing Harmony, where does its future lie? Will Harmony remain a research project where the Big ones can feed on? Or are there chances corporations like Google might pick it up and deliver some incredible products with it. Yeah, I know about Android, but am looking at not just the mobile platform, but desktops, Servers and web in general. Or probably demi money gods like Microsoft? But I ask, how much revenue could such an investment generate or disruption in targeted competitor’s model and at what cost? Would such investments deliver most of the yearnings of Java enthusiasts? What is Harmony’s targeted market? I doubt if it’s Linux. I think Sun’s implementation would continue to dominate the Windows platform. Probably after seeing the first release of Harmony, some of these questions would be easier to answer.
I believe Nature has given few the ability to see the future, so I hope a few of these few would read this and share what the future holds. I pray long live Apache Harmony!
