Yesterday, I discussed with some students who are implementing an SML plug-in for eclipse, whether types are necessary for a lambda-lifting refactoring for SML.

So today’s quiz is simply: Are types necessary for lambda-lifting refactoring in SML? Why/Why not?

Remember, the refactoring works on valid SML programs, and after the transformation the program should still be valid.

Example:

fun foo x =
    let fun bar y = (x,y)
    in bar x end

is transformed/refactored to

fun bar x y = (x,y)
fun foo x = bar x x

I’ll give the answer tomorrow.