ESE Day Two - Why Modelling Isn't Complex

Day One was always going to be a tough act to follow, but once again I had a great day at the conference. As well as meeting even more of the community to discuss the technologies that I use, as well some great ideas for future EclipseZone articles, I managed to fit in some quality talks.

Ed Merks talk was one that definitely catches your attention. As someone who's really interested in model driven development, and who has heard quite a few arguments against it, I was intrigued to see what Ed was going to speak about. One of the things that people say about model driven development is that it has a "steep learning curve" but what that really means is that you get to the more complex parts faster, rather than moving slowly at a lower complexity. Also, he pointed out that UML is not modelling!

I'm not a fan of the term "meta modelling", and it seems to be something that turns a lot of people off modelling. It seems Ed doesn't really like the term either, and he described that a meta-model is really just a model of a model. Another thing people seem to have in their heads is that "Generated Code Sucks", thinking what they write is more performant.

You can see some explanations of Ecore and EMF and lots more on Ed's slides here. It was a great talk, where we got to see the arguments people have against modelling, and why most of them are unfounded.

Still on EMF, Tom Schindl gave a great talk on writing Datacentric Application with RCP + EMF + Databinding.  It shows how easy it is to use RCP, EMF and CDO all together to create applications with good concurrency support out of the box. OSGi came in again - with some clever use of bundles to seperate functionality, you can use the OSGi console to stop and uninstall a bundle, make some code changes and reinstall it in your running application. This is a fantastic advantage of OSGi for debugging, which I hadn't considered previously.

Because I had to catch a flight, I wasn't able to get to the What's New In BIRT 2.3 talk, but I talked to Jason Weathersby, and he pointed me to the webinar which covers much the same content. From what I've seen of BIRT it's really flexible, and the report design is really usable.

I'd have really enjoyed another day or two hanging around all these leaders of the Eclipse community, but all good things have to come to an end. And I guess I only need to wait a few months before EclipseCon 2009!