> Apart from the implementation (which doesn't seem to be a problem if
> the right alg. is used[*]) does anyone know of more subtle problems with
> cyclic types [such as not being able to define the type system using
> the traditional sequent style]?  Is this a well-studied area?
>

The only problem I can imagine is that equality for recursive types
is nontrivial.  Amadio and Cardelli give a complete set of rules for
it in their POPL91 paper.

Satish

Reply via email to