Language.Fortran.Repr.Type.Scalar.Common

Documentation

type FKindTerm = Natural

Fortran kinds are represented by natural numbers. We use them on both type and term levels.

type FKindType = NaturalK

reifyKinded :: forall k (a :: k) n. (n ~ FKindOf a, KnownNat n) => Sing a -> FKindTerm

Reify a kind tag to its Natural equivalent.

class FKinded (a :: Type) where

Fortran types which use simple integer kinds.

Associated Types

type FKindOf (x :: a) :: FKindType

type FKindDefault :: a

Methods

printFKind :: a -> FKindTerm

This we get via the type family, but require singletons.

parseFKind :: FKindTerm -> Maybe a

This we *should* get via the type family, but again require singletons.

Instances

FKinded FTComplexWrapper

Defined in Language.Fortran.Repr.Type.Scalar.Complex

Associated Types

type FKindOf x :: FKindType

type FKindDefault :: a

Methods

printFKind :: FTComplexWrapper -> FKindTerm

parseFKind :: FKindTerm -> Maybe FTComplexWrapper

FKinded FTInt

Defined in Language.Fortran.Repr.Type.Scalar.Int

Associated Types

type FKindOf x :: FKindType

type FKindDefault :: a

Methods

printFKind :: FTInt -> FKindTerm

parseFKind :: FKindTerm -> Maybe FTInt

FKinded FTReal

Defined in Language.Fortran.Repr.Type.Scalar.Real

Associated Types

type FKindOf x :: FKindType

type FKindDefault :: a

Methods

printFKind :: FTReal -> FKindTerm

parseFKind :: FKindTerm -> Maybe FTReal

data SingCmp (l :: k) (r :: k)

Constructors

SingEq (l :~: r)

SingLt

SingGt

singCompare :: forall k (a :: k) (b :: k). SOrd k => Sing a -> Sing b -> SingCmp a b

Upgrade an SOrdering to include a proof of type equality for the equal case.

We have no choice but to fake the Refl with unsafeCoerce. But assuming SEQ is used correctly, it should be safe.