* Posts by Andrew John Hughes 1

1 publicly visible post • joined 6 Sep 2009

Oracle should relax Sun's Java Community control grip

Andrew John Hughes 1

Poor article

'While Apache's has been able to build an implementation of Java Standard Edition under Project Harmony thanks to the opening of Java, Harmony cannot be certified because the TCKs contain proprietary code the open-source code cannot touch. Harmony, therefore, remains stuck in a limbo of having been built but being uncertified.'

That statement is factually inaccurate. Project Harmony did not occur 'thanks to the opening of Java'. First of all, Java is a trademark which is applied to certified implementations that have passed the TCK, a process the entire article talks about as being closed and proprietary still. I believe what you mean is Sun's JDK, and this certainly hasn't helped Harmony. If anything, it has hindered it. A very plausible argument can be made for the growing success of GNU Classpath and Apache Harmony being the trigger for Sun open-sourcing their implementation, giving them a means of both silencing their critics (which included RMS) over its proprietary nature and retaining control. Both Apache Harmony and GNU Classpath are clean room implementations which aren't based on Sun's code. Both are not eligible for the TCK under any rules Sun has put forward, either privately or through the OpenJDK TCK agreement. You can't say that either implementation has 'built an implementation of Java Standard Edition' until they have passed said TCK which is supposed to be the proof that these specifications have been met.

On the main content of the article, I and many other people working on OpenJDK, GNU Classpath and Apache Harmony would love to see both the JCP and the TCK become more open. I find it hard to trust a test of the specification which you can't look at or discuss without an NDA and which has only been passed on code derived from Sun's implementation (that's actually a term of the OpenJDK TCK license). The JCP results in specifications under terms which effectively prevent open source implementations, and we weren't able to use them to develop GNU Classpath.

Oracle could certainly do a lot to clear things up in their new position. Let's hope they do.