I am not sure how to write this but I’ve been a bit disappointed with something that affects Belgium as a whole lately. Everything started when I noticed that lots of advertising booth were in Dutch. I just thought it was a bit strange but it was confirmed the whole day. Many people were just speaking to us in Dutch in a first place just like this was perfectly normal. Come on, it’s a European conference, right? Don’t get me wrong, I think that I am really an open guy but I just feel too bad to give such view of Belgium.

That said, here are the sessions of today. Oh yes, I won’t talk about the keynotes and stuff because It’s always a chance for the sponsors to show how wonderful they are and bla bla bla. There’s good stuff in there and lots of commercial stuff so I won’t talk much except the JSE and JEE guideline document on which Stefan is working on which sounds interesting and which will be made available on http://www.bejug.org for free.

Pragmatic Clustering Guide

This really was the kind of sessions I was expecting. Mike is a brilliant guy, and his presentation is impressive end to end. He basically explained the kind of decisions and thoughts Atlassian had made to upgrade Confluence so that it’s clusterable. Very interesting materials and feedbacks, I just wish we had a couple of hours left so that he can show us code examples and more concrete implementations (I am a bit naive on this one, I admit it). Anyway, if you’re concerned by clustering an application, watch out parleys and make sure to follow this presentation. It really helps.

The Java SE Platform Past and Future

A very good wrap up of new features in JDK6, I really liked the kind remarks to Microsoft just like it was a kind of friendly established war. Quite funny. A good introduction to what’s the consequences of putting Java to open source with web sites such as openjdk

Location based J2EE application development made simple (fraud)

Well, i am biased on this one since I am working now for IONIC Software which is the world leader in GIS interoperability. I am saying it’s a fraud because the speaker was retained somewhere and we had this giant talk by some commercial guy which said at the beginning the presentation changed a bit. I really hope it was a bit more because it’s by far the worst presentation I have attended. It basically shown the business model of ESRI, the importance of GIS systems and a few technological buzzwords involved in their implementation. Totally useless. Too bad.

Java Performance Myths

I would place this one at the same level as Heinz and Kirk’s presentation. This one really deserved more than one hour and should have been placed in the University part of the conference. Brian explained in details five performance myths with very interesting details about the kind of optimizations the JVM could do for us behind the scene:

  • Allocation is slow
  • I’ll just write a benchmark
  • Synchronization is slow
  • Making method final makes it faster
  • Clever code is faster code

The most interesting part of the presentation was about dead code detection where he shown that the JVM could optimize/skip method call without side effect where the returned value was not used anywhere. Pretty much interesting.