[PaulHoule]

I am a volunteer working on the ISO 20022 standard and we use Ecore, an object modelling system that was designed to bootstrap UML, to represent financial messages.

I developed a toolset for working with Ecore that was adequate for what I needed to do. I took a crack at bootstrapping UML 2 but I found there were "chicken an egg" problems that I figured would take me a month to resolve, probably some important objects would have to be duplicated but then hidden from the upper layers.

(Resolving "chicken and egg" problems is something I do very well but I've been burned by spending a month working on something I felt uniquely inspired to do and then spending much more than a month trying to sell it and finding nobody gives a damn.)

UML has a fascinating vision which parallels the W3C's vision of the the semantic web, particularly "executable UML" ought to be possible, but it is held back by vendors who thrive on almost-functional standards that require proprietary additions to really work.