Grady Booch on verticalized tools and Domain Specific Languages
Grady correctly identifies why the fabled “San Francisco Project” failed all those years ago - it was over engineered and was not based upon real-world business processes. You can't simply make up patterns - they've got to be harvested from existing systems and implementations. If no one has ever used something how can it be a pattern? This is what IBM's researchers apparently forgot during the development of San Francisco.
This article is a little frusturating because it goes on to discuss how some people came over to Redmond from Rational to build something better and (even worse) wrongly implies that model-driven development wasn't happening prior to UML.
Just saw this: an interesting white paper explaining the productivity benefits one might gain from adopting a domain specific language instead of UML. Still confused? Keith does a great job of explaining UML and DSLs.
“UML is the Unified Modeling Language, not the Universal Modeling Language” - Fred Bershears