If a functor contains both a newtype and a type synonym referring to the newtype, then during instantiation we must apply the instantiation map to the types in the functor, so that the type synonym now refers to the instantiated newtype instead. Previously, we only applied type substitutions to the types in the functor, which would replace type variables with concrete types defined in the functor arguments, but would not replace any actual names.
Confusingly, the code referred to the instantiation map as "value substitutions" even though it can contain names in the type namespace. The "type substitutions" and "value substitutions" are better thought of as "TVar -> Type mappings" and "Name -> Name mappings". I have renamed them to better reflect this.
If a functor contains both a newtype and a type synonym referring to the newtype, then during instantiation we must apply the instantiation map to the types in the functor, so that the type synonym now refers to the instantiated newtype instead. Previously, we only applied type substitutions to the types in the functor, which would replace type variables with concrete types defined in the functor arguments, but would not replace any actual names.
Confusingly, the code referred to the instantiation map as "value substitutions" even though it can contain names in the type namespace. The "type substitutions" and "value substitutions" are better thought of as "
TVar -> Typemappings" and "Name -> Namemappings". I have renamed them to better reflect this.Fixes #1590.