The Creative Generation

Imagine combining the stimulating English landscape of Cambridge, with a collection of the world's brightest IT code generation minds. Have them exchange thoughts in creative sessions and tutorial, and you'll find that innovation will take on a completely new dimension. Yes, I'm talking about the Code Generation 2008 conference, which was held in Cambridge June 25-27.

Raising the abstraction level
The common theme amongst the sessions resolved around raising the level of abstraction, either by modeling in UML or a (oh no, save me from the buzzword) DSL. Everybody seems to agree that raising the abstraction level is a good thing. What's not always clear however is, what to raise the abstraction level to. Keeping a constant level of modeling, without sneaking in too many details of your concrete platform, remains the challenge. And for people still looking for the ultimate ride on "round-tripping": well there's a reason the term ends with "tripping".

Rise of the model-driven tools
Model-driven architecture has been around since the 2000's. Well, the theory has been, and depending on specific architect's diligence and guidelines, it has also worked brilliantly in practice. In order to become mainstream however, a better tool support was required. It's just too easy to mix concerns or hack some quick changes into a system, when code generation is a part of your process. Looking at 2008, this tool support is finally coming. The pattern / template / marking distinction is more and more explicitly available in tools, led by open architecture ware. Even the Microsoft DSL tools manage to grasp the concept, although their template and pattern implementation will face the same challenges EMF/Java have seen over the last several years. It seems a standard pattern that every framework hurdle needs to be dealt with at least twice.

Personally I can say that this has been probably one of the most inspiring conferences I've been to. The unique combination of a beautiful surrounding with highly intellectual people, created a steady flow of ideas. The one thing I brought home is that my own views on MDA, which I've always considered somewhat straightforward and general, are not that mainstream after all. A lot of MDA's "traps" and pitfalls still catch developers. Let's broaden conferences like this, and spread the word further: domain-driven modeling is here to stay!