Re: [Haskell-cafe] ANN: data-fix-cse -- Common subexpression elimination for EDSLs

Do you think the approach can be extended for non-regular (nested) algebraic types (where the recursive data type is sometimes at a different type instance)? For instance, it's very handy to use GADTs to capture embedded language types in host language (Haskell) types, which leads to non-regularity.

I'm not sure I understand the case you are talking about. Can you write a simple example of the types like this? Cheers, Anton