> I don't mind; honestly, it's the line number more than the error that I'm > interested in. It's definitely nice having the line numbers (and ununifiable types). I've already had my first instance of using MLton to find a type error almost instantly after giving up on an unhelpful SML/NJ type error.