Watt's Up Musings of an original geek

I went to a technical lecture tonight that got my juices flowing. It dealt with modelling programs which model other programs, ad infinitum (the subject was model drive architecture and the eclipse modeling framework). Another term that is used is meta-models, and meta-meta-models, etc. I did similar work over 10 years ago, but I never followed-up on it once the project ended. I tend to do that. I start something that interests me, but then I give up on it. Later I come to recognize that the field has caught up and then surpassed me. Back then I built a program that modeled programs and then I built a transformation system that would input one model and output another. So a programmer doesn’t write code, but instead write a transformer which indirectly write code. Naturally, one can build a lattice of transformers where transformers can be modelled and transformed too. All very recursive and self-referential. And it’s only in the last 3-4 years that people have been publishing and writing books on it. I tried to write a paper on what I did back then and when I submitted it to a small internal conference it was rejected, and so I gave up and I went on to other things. I just don’t have that single minded determination.