f n =

  let u = [1..n] in g n

  where 

    g x = u

Why?  I understand that it's "just the way Haskell works", but I'm 

wondering about the rationale of the language designers when deciding

on this scoping behavior.