Point-Free Style : Haskell doesn't like it ?

[Ahn Ki-yung]

I wan't to know why Haskell doesn't like pooint-free style.

For example, 0) and 1) are Unresolved top-level overloading,
but 2) is accepted by the type system.

0) show' = show
1) show' = (\x->show x)
2) show x = show x

What are the differences between 0), 1) and 2) ?
They must have the same type.

Point-free style is consice and more functional, so I like it.
I think quite many of functional programmers would prefer.
It is especially annoying when using compoisitoins to define
new function.

I should write like this in order to pass the type system
f x y = (g . h . k) x y
instead of writing just compositions like below.
f = g . h . k