Jacques Garrigue wrote:
> The reason is mostly wrong :-)

That'll teach me to comment on type theory on this list :o)

> And neither polymorphic variants nor object require type anotations in
> ocaml; they just make it much more painful to understand error
> messages.

Though I'm confused by this - I thought that polymorphic methods in classes
(a part of the object system) do require type annotations and there are
cases with polymorphic variants where coercions (which I'd regard as a type
annotation?) must be explicitly written for a valid program to type. I
wasn't trying to say that all uses of them require type annotations, just
that there are occasions where you *have* to use them whereas for "core" ML
you never *have* to include a type annotation for *any* valid program - your
types just might be more general than you expect/want.

Or am I still barking up the wrong tree?


David