> I consider the current binding system to be consistent. No, it's not. It is possible to get the same code compiled with all the same settings and get different behaviors. This is exactly what consistency is not.
What I would be looking for is that the non-binding attributes were the defaults, that MXML promised a deterministic order of binding, and that the present model be fixed to actually do what it looks-like-it-does-but-not-really. I.e. employ the mechanism that ensures that all bindings were assigned values provided in attributes, and that the later be made clearly distinct and alert users on potential overhead. Best. Oleg
