it doesn't mean that the language's closure semantics have suddenly changed. Yes, but what are the issues with fresh let-bindings per iteration? Presumably th

it doesn't mean that the language's closure semantics have suddenly changed. Yes, but what are the issues with fresh let-bindings per iteration? Presumably the alternative you mean is where the variables present in the local environment are copied on creation of a closure. No, that's bad -- it leaves you unable to show that objects with mutable state are reducible to closures.