\begin{code}
module TypeRep (
- Type(..), TyNote(..), SourceType(..), -- Representation visible to friends
+ TyThing(..),
+ Type(..), TyNote(..), -- Representation visible
+ PredType(..), -- to friends
- Kind, TauType, PredType, ThetaType, -- Synonyms
+ Kind, ThetaType, -- Synonyms
TyVarSubst,
superKind, superBoxity, -- KX and BX respectively
liftedTypeKind, unliftedTypeKind, openTypeKind, -- :: KX
mkArrowKind, mkArrowKinds, -- :: KX -> KX -> KX
- usageKindCon, -- :: KX
- usageTypeKind, -- :: KX
- usOnceTyCon, usManyTyCon, -- :: $
- usOnce, usMany, -- :: $
+ funTyCon,
- funTyCon
+ crudePprType -- Prints type representations for debugging
) where
#include "HsVersions.h"
--- friends:
-import Var ( TyVar )
-import VarEnv
-import VarSet
+import {-# SOURCE #-} DataCon( DataCon )
-import Name ( Name )
-import TyCon ( TyCon, KindCon, mkFunTyCon, mkKindCon, mkSuperKindCon )
-import Class ( Class )
+-- friends:
+import Var ( Id, TyVar, tyVarKind )
+import VarEnv ( TyVarEnv )
+import VarSet ( TyVarSet )
+import Name ( Name, mkWiredInName, mkInternalName )
+import OccName ( mkOccFS, mkKindOccFS, tcName )
+import BasicTypes ( IPName )
+import TyCon ( TyCon, KindCon, mkFunTyCon, mkKindCon, mkSuperKindCon, isNewTyCon )
+import Class ( Class )
-- others
-import PrelNames ( superKindName, superBoxityName, liftedConName,
- unliftedConName, typeConName, openKindConName,
- usageKindConName, usOnceTyConName, usManyTyConName,
- funTyConName
+import PrelNames ( gHC_PRIM, kindConKey, boxityConKey, liftedConKey,
+ unliftedConKey, typeConKey, anyBoxConKey,
+ funTyConKey
)
+import SrcLoc ( noSrcLoc )
+import Outputable
\end{code}
%************************************************************************
would give rise to infinite types, which in turn makes eqType diverge.
Similarly splitForAllTys and splitFunTys can get into a loop.
-Solution: for recursive newtypes use a coerce, and treat the newtype
-and its representation as distinct right through the compiler. That's
-what you get if you use recursive newtypes. (They are rare, so who
-cares if they are a tiny bit less efficient.)
+Solution:
+
+* Newtypes are always represented using NewTcApp, never as TyConApp.
+
+* For non-recursive newtypes, P, treat P just like a type synonym after
+ type-checking is done; i.e. it's opaque during type checking (functions
+ from TcType) but transparent afterwards (functions from Type).
+ "Treat P as a type synonym" means "all functions expand NewTcApps
+ on the fly".
+
+ Applications of the data constructor P simply vanish:
+ P x = x
+
-The TyCon still says "I'm a newtype", but we do not represent the
-newtype application as a SourceType; instead as a TyConApp.
+* For recursive newtypes Q, treat the Q and its representation as
+ distinct right through the compiler. Applications of the data consructor
+ use a coerce:
+ Q = \(x::Q->Q). coerce Q x
+ They are rare, so who cares if they are a tiny bit less efficient.
+
+The typechecker (TcTyDecls) identifies enough type construtors as 'recursive'
+to cut all loops. The other members of the loop may be marked 'non-recursive'.
%************************************************************************
\begin{code}
type SuperKind = Type
type Kind = Type
-type TauType = Type
type TyVarSubst = TyVarEnv Type
-- synonyms have their own constructors, below.
[Type] -- Might not be saturated.
+ | NewTcApp -- Application of a NewType TyCon. All newtype applications
+ TyCon -- show up like this until they are fed through newTypeRep,
+ -- which returns
+ -- * an ordinary TyConApp for non-saturated,
+ -- or recursive newtypes
+ --
+ -- * the representation type of the newtype for satuarted,
+ -- non-recursive ones
+ -- [But the result of a call to newTypeRep is always consumed
+ -- immediately; it never lives on in another type. So in any
+ -- type, newtypes are always represented with NewTcApp.]
+ [Type] -- Might not be saturated.
+
| FunTy -- Special case of TyConApp: TyConApp FunTyCon [t1,t2]
Type
Type
TyVar
Type
- | SourceTy -- A high level source type
- SourceType -- ...can be expanded to a representation type...
-
- | UsageTy -- A usage-annotated type
- Type -- - Annotation of kind $ (i.e., usage annotation)
- Type -- - Annotated type
+ | PredTy -- A high level source type
+ PredType -- ...can be expanded to a representation type...
| NoteTy -- A type with a note attached
TyNote
-- The type to which the note is attached is the expanded form.
\end{code}
-INVARIANT: UsageTys are optional, but may *only* appear immediately
-under a FunTy (either argument), or at top-level of a Type permitted
-to be annotated (such as the type of an Id). NoteTys are transparent
-for the purposes of this rule.
-
-------------------------------------
Source types
A type of the form
- SourceTy sty
-represents a value whose type is the Haskell source type sty.
+ PredTy p
+represents a value whose type is the Haskell predicate p,
+where a predicate is what occurs before the '=>' in a Haskell type.
It can be expanded into its representation, but:
* The type checker must treat it as opaque
* The rest of the compiler treats it as transparent
-There are two main uses
- a) Haskell predicates
- b) newtypes
-
Consider these examples:
f :: (Eq a) => a -> Int
g :: (?x :: Int -> Int) => a -> Int
Predicates are represented inside GHC by PredType:
\begin{code}
-data SourceType = ClassP Class [Type] -- Class predicate
- | IParam Name Type -- Implicit parameter
- | NType TyCon [Type] -- A *saturated*, *non-recursive* newtype application
- -- [See notes at top about newtypes]
+data PredType
+ = ClassP Class [Type] -- Class predicate
+ | IParam (IPName Name) Type -- Implicit parameter
-type PredType = SourceType -- A subtype for predicates
type ThetaType = [PredType]
\end{code}
| Type liftedness -- (Type *) is printed as just *
-- (Type #) is printed as just #
- | UsageKind -- Printed '$'; used for usage annotations
-
| OpenKind -- Can be lifted or unlifted
-- Printed '?'
type variable, one that may very well later be unified with a type.
For example, suppose f::a, and we see an application (f x). Then a
must be a function type, so we unify a with (b->c). But what kind
- are b and c? They can be lifted or unlifted types, so we give them
- kind '?'.
+ are b and c? They can be lifted or unlifted types, or indeed type schemes,
+ so we give them kind '?'.
When the type checker generalises over a bunch of type variables, it
makes any that still have kind '?' into kind '*'. So kind '?' is never
BX, the type of a boxity
\begin{code}
+superKindName = kindQual FSLIT("KX") kindConKey
+superBoxityName = kindQual FSLIT("BX") boxityConKey
+liftedConName = kindQual FSLIT("*") liftedConKey
+unliftedConName = kindQual FSLIT("#") unliftedConKey
+openKindConName = kindQual FSLIT("?") anyBoxConKey
+typeConName = kindQual FSLIT("Type") typeConKey
+
+kindQual str uq = mkInternalName uq (mkKindOccFS tcName str) noSrcLoc
+ -- Kinds are not z-encoded in interface file, hence mkKindOccFS
+ -- And they don't come from any particular module; indeed we always
+ -- want to print them unqualified. Hence the InternalName.
+\end{code}
+
+\begin{code}
superKind :: SuperKind -- KX, the type of all kinds
superKind = TyConApp (mkSuperKindCon superKindName) []
\begin{code}
liftedBoxity, unliftedBoxity :: Kind -- :: BX
-liftedBoxity = TyConApp (mkKindCon liftedConName superBoxity) []
+liftedBoxity = TyConApp liftedBoxityCon []
+unliftedBoxity = TyConApp unliftedBoxityCon []
-unliftedBoxity = TyConApp (mkKindCon unliftedConName superBoxity) []
+liftedBoxityCon = mkKindCon liftedConName superBoxity
+unliftedBoxityCon = mkKindCon unliftedConName superBoxity
\end{code}
------------------------------------------
-Define kinds: Type, Type *, Type #, OpenKind, and UsageKind
+Define kinds: Type, Type *, Type #, OpenKind
\begin{code}
typeCon :: KindCon -- :: BX -> KX
openKindCon = mkKindCon openKindConName superKind
openTypeKind = TyConApp openKindCon []
-
-usageKindCon = mkKindCon usageKindConName superKind
-usageTypeKind = TyConApp usageKindCon []
\end{code}
------------------------------------------
%************************************************************************
%* *
+ TyThing
+%* *
+%************************************************************************
+
+Despite the fact that DataCon has to be imported via a hi-boot route,
+this module seems the right place for TyThing, because it's needed for
+funTyCon and all the types in TysPrim.
+
+\begin{code}
+data TyThing = AnId Id
+ | ADataCon DataCon
+ | ATyCon TyCon
+ | AClass Class
+\end{code}
+
+
+%************************************************************************
+%* *
\subsection{Wired-in type constructors
%* *
%************************************************************************
\begin{code}
funTyCon = mkFunTyCon funTyConName (mkArrowKinds [liftedTypeKind, liftedTypeKind] liftedTypeKind)
+ -- You might think that (->) should have type (? -> ? -> *), and you'd be right
+ -- But if we do that we get kind errors when saying
+ -- instance Control.Arrow (->)
+ -- becuase the expected kind is (*->*->*). The trouble is that the
+ -- expected/actual stuff in the unifier does not go contra-variant, whereas
+ -- the kind sub-typing does. Sigh. It really only matters if you use (->) in
+ -- a prefix way, thus: (->) Int# Int#. And this is unusual.
+
+funTyConName = mkWiredInName gHC_PRIM
+ (mkOccFS tcName FSLIT("(->)"))
+ funTyConKey
+ Nothing -- No parent object
+ (ATyCon funTyCon) -- Relevant TyCon
\end{code}
-------------------------------------------
-Usage tycons @.@ and @!@
-The usage tycons are of kind usageTypeKind (`$'). The types contain
-no values, and are used purely for usage annotation.
-\begin{code}
-usOnceTyCon = mkKindCon usOnceTyConName usageTypeKind
-usOnce = TyConApp usOnceTyCon []
-
-usManyTyCon = mkKindCon usManyTyConName usageTypeKind
-usMany = TyConApp usManyTyCon []
-\end{code}
+%************************************************************************
+%* *
+ Crude printing
+ For debug purposes, we may want to print a type directly
+%* *
+%************************************************************************
+\begin{code}
+crudePprType :: Type -> SDoc
+crudePprType (TyVarTy tv) = ppr tv
+crudePprType (AppTy t1 t2) = crudePprType t1 <+> (parens (crudePprType t2))
+crudePprType (FunTy t1 t2) = crudePprType t1 <+> (parens (crudePprType t2))
+crudePprType (TyConApp tc tys) = ppr_tc_app (ppr tc <> pp_nt tc) tys
+crudePprType (NewTcApp tc tys) = ptext SLIT("<nt>") <+> ppr_tc_app (ppr tc <> pp_nt tc) tys
+crudePprType (ForAllTy tv ty) = sep [ptext SLIT("forall") <+>
+ parens (ppr tv <+> crudePprType (tyVarKind tv)) <> dot,
+ crudePprType ty]
+crudePprType (PredTy st) = braces (crudePprPredTy st)
+crudePprType (NoteTy (SynNote ty1) ty2) = crudePprType ty1
+crudePprType (NoteTy other ty) = crudePprType ty
+
+crudePprPredTy (ClassP cls tys) = ppr_tc_app (ppr cls) tys
+crudePprPredTy (IParam ip ty) = ppr ip <> dcolon <> crudePprType ty
+
+ppr_tc_app :: SDoc -> [Type] -> SDoc
+ppr_tc_app tc tys = tc <+> sep (map (parens . crudePprType) tys)
+
+pp_nt tc | isNewTyCon tc = ptext SLIT("(nt)")
+ | otherwise = empty
+\end{code}
\ No newline at end of file