Question

GHC compiles it without a hitch, but it fails miserably at runtime:

many_a x =
    let
        a = 2
    in
        let 
            a = 2*a
        in
            x*a

Intuitively, this shouldn’t work. But yet GHC accepted it.

Answer 1

Yes, that’s valid Haskell code. The thing is that the a in the second let...in expression is actually a new a; it’s perfectly permissible to shadow existing variables by defining a new variable with the same name. However, it doesn’t affect the value of the outer a. It’s generally considered bad style, however, due to the potential for confusion and mistakes; if you pass -Wall to GHC, it’ll issue a warning if you do this.

It “fails” at runtime because you’ve defined a to be 2*a, which results in an infinite loop at runtime. This is due to laziness; basically, (*) evaluates both its arguments before multiplying them. Evaluating 2 works fine, of course, but evaluating a causes it to go through the same process all over again. The same thing that allows infinite lists (like ones = 1:ones) makes this code be an infinite loop.