Really 2.5-3.5. What happened (at least what I saw), etc.
Last night was great. I attended the Java Communities in Action: Jini, JXTA, JCP, and Java.net. Good food and drink plus great technical cocktail chat. I had a demo which made it a little difficult to enjoy myself, but fortunately, few were interested in demos. Most people wanted to talk and I think I spent more time discussing mobile agents than I did our product. I hope my boss doesn't read this.
Afterwards, I attended a very interesting BoF by PsiNaptics on their very small device Jini work with Ford Research for in-vehicle services. Sadly I had to leave early to get ready for my BoF.
Next, I presented my BoF: Service Oriented Architectures - Separating Hype from Reality. I was astonished that so many people would come to a 10:30 pm session. For my part, I thought it went well. I'll post something here about the ideas later when I'm more than half conscious. JavaOne tends to wear one down a bit.
The Nokia General Session, although corny, showed that small device Java is slowly improving (Oh so slowly). Because of my past, this stuff warms my cockles. What the heck is a cockle anyway? To me the interesting thing was that peer-to-peer, e.g., phone-to-phone, was a key element of their roadmap. This is interesting for a few reasons. First, there are a ton of cool applications that can be built. But maybe more interesting are the business implications. Phone manufacturers want to differentiate and avoid diminishing margins so p2p makes sense to them. But operators want network billing so where's their motivation to buy p2p phones from manufacturers? Is this an end-around by the phone manufacturers? In the past, the operators had the say in what phones could do. Have the rules changed?
I saw an interesting presentation on Jini and Auto-ID. Basically, everything in the supply chain will be tracked using RFID tags but there are huge scalability problems. A team from Sun is using nearly every Java technology, including my friend Jini, to manage this. This looks like a very exciting and profitable space but it also appears everybody else has come to the same conclusion.
Michael Warres gave an excellent presentation on Jini Extensible Remote Invocation, a third release of RMI that is ultra configurable. He eloquently presented the details and then showed how to customize it through two interesting examples: Firewall traversal, and serialization caching. The former allows RMI through FWs, the later is a big performance optimization. These are available on Michael's Jini page.
Followed my friend Steve into a NetBeans presentation. This was more interesting than I expected. I've decided I must explore this IDE. Their upcoming 4.0 release will have some very cool features. One is refactoring which is a big time-saver. The other is performance and memory profiling using FluidJ a Sun project. The demo was very impressive and there's no doubt I could make good use of this tool.
There was a disappointing panel discussion on the merits of Aspect Oriented Programming. Summary: Some nice capabilities, some really dangerous/bad capabilities. Still needs work before hitting prime time. I didn't stay for most of it, but I did catch that FluidJ Java profiler does use AOP.
Had a great lunch in Chinatown at House of Nanking. Not the kind of place you'd walk into w/o a recommendation. Appearances are deceiving. The food was excellent and reasonably priced. It's at 9-something Kearny. Try it.