[bitc-dev] Objects in BitC

[Sandro Magi]

Jonathan S. Shapiro wrote:
> There isn't any deep theoretical issue. It's more that existential types
> seem to cross the threshold of user-manageable complexity unless they
> are very carefully constrained.
> 
> In a language with value types (i.e. explicit unboxing) there is also a
> problem that the compiler cannot determine the size of a field having
> existential type.

Polymorphic data types have existential type [1], so it would seem that
you already have a solution for this in place. Unless perhaps it's not
flexible enough?

Oleg shows some interesting encodings of existentials using type classes
on that page, so perhaps no language extension at all is required.

Sandro

http://okmij.org/ftp/Computation/Existentials.html#exist-via-forall