On 12 Jun 2015 08:19, "Matt Oliveri" <[email protected]> wrote:
> Well if you're just trying to enforce that ASTs are of well-typed
> program fragments, this is not true. There's a certain point after
> which you can encode any language's typing rules in the host type
> system. This is analogous to how you only need predicate logic to
> axiomatize any deductive system.

I'm not sure I get this. Let's say my language has intersection types,
which require unification with expansion variables, and I am writing the
compiler in Haskell. How / why would I want to encode that into the Haskell
type system which does not.

Keean.
_______________________________________________
bitc-dev mailing list
[email protected]
http://www.coyotos.org/mailman/listinfo/bitc-dev

Reply via email to