If this was anybody else, I would tell them to explore alternative programming methodologies, styles, and languages. But not you gdp; you know what you're doing, so if you're hitting a dead end with interactive ML, I don't think many people can help you, you might be at a frontier here (at least I can't; I have dropped my last static language 8 years ago when I decided to have my cake and eat it too with the Lisp family.)