No, a spec is not code. It's possible to describe simple behavior that's nevertheless difficult to implement. Consider, say,

  fn sin(x: f16) -> f16

There are only 64k different f16s. Easy enough to test them all. A given sin() is either correct or it's not.

Yet sin() here can have a large number of different implementations. The spec alone under-determines the actual code.