no need for bst-trav since there is an elements function. You can just map over the elements.
in the data declaration 'a' is a generic type, but Haskell will infer that the type must be an instance or class Ord (i.e. implement <,<=,>=,>,max,min,==). (If the type is not an instance of Ord it will tell you so at compile time)
Actually yes, in Haskell it is the same. Lazy evaluation means you could take the 4th element of a map over an infinite list. For example: "map (*2) [1..] !! 4"
I would argue that relying on overloaded comparison functions is a bad idea. Not all things have a 'natural ordering,' and even for those that do you might want to override it at some point.
I think reasonable people can differ on this. Passing comparators as arguments is an approach more typical of dynamically-typed languages (e.g. CL and Scheme 'sort'). In a statically-typed language, it is more common to use a comparison operator that specializes on type (e.g. Haskell Prelude 'sort'). If a set of objects don't have a 'natural ordering', they might not belong in a Binary Search Tree in the first place.
You have just seen a bunch of code that gets the job done, whereas there is still an argument about whether the lisp code should be passed a comparison function. If you are going to make this argument you should show some actual code, because it is obvious from looking at this example which way is better.