{- (c) The University of Glasgow 2006 (c) The GRASP/AQUA Project, Glasgow University, 1992-1998 Handles @deriving@ clauses on @data@ declarations. -} {-# LANGUAGE CPP #-} module TcDeriv ( tcDeriving, DerivInfo(..), mkDerivInfos ) where #include "HsVersions.h" import HsSyn import DynFlags import TcRnMonad import FamInst import TcDerivInfer import TcDerivUtils import TcValidity( allDistinctTyVars ) import TcClassDcl( tcATDefault, tcMkDeclCtxt ) import TcEnv import TcGenDeriv -- Deriv stuff import InstEnv import Inst import FamInstEnv import TcHsType import TcMType import RnNames( extendGlobalRdrEnvRn ) import RnBinds import RnEnv import RnSource ( addTcgDUs ) import Avail import Unify( tcUnifyTy ) import BasicTypes ( DerivStrategy(..) ) import Class import Type import ErrUtils import DataCon import Maybes import RdrName import Name import NameSet import TyCon import TcType import Var import VarEnv import VarSet import PrelNames import SrcLoc import Util import Outputable import FastString import Bag import Pair import FV (fvVarList, unionFV, mkFVs) import qualified GHC.LanguageExtensions as LangExt import Control.Monad import Data.List {- ************************************************************************ * * Overview * * ************************************************************************ Overall plan ~~~~~~~~~~~~ 1. Convert the decls (i.e. data/newtype deriving clauses, plus standalone deriving) to [EarlyDerivSpec] 2. Infer the missing contexts for the InferTheta's 3. Add the derived bindings, generating InstInfos -} data EarlyDerivSpec = InferTheta (DerivSpec [ThetaOrigin]) | GivenTheta (DerivSpec ThetaType) -- InferTheta ds => the context for the instance should be inferred -- In this case ds_theta is the list of all the sets of -- constraints needed, such as (Eq [a], Eq a), together with a -- suitable CtLoc to get good error messages. -- The inference process is to reduce this to a -- simpler form (e.g. Eq a) -- -- GivenTheta ds => the exact context for the instance is supplied -- by the programmer; it is ds_theta -- See Note [Inferring the instance context] in TcDerivInfer earlyDSLoc :: EarlyDerivSpec -> SrcSpan earlyDSLoc (InferTheta spec) = ds_loc spec earlyDSLoc (GivenTheta spec) = ds_loc spec splitEarlyDerivSpec :: [EarlyDerivSpec] -> ([DerivSpec [ThetaOrigin]], [DerivSpec ThetaType]) splitEarlyDerivSpec [] = ([],[]) splitEarlyDerivSpec (InferTheta spec : specs) = case splitEarlyDerivSpec specs of (is, gs) -> (spec : is, gs) splitEarlyDerivSpec (GivenTheta spec : specs) = case splitEarlyDerivSpec specs of (is, gs) -> (is, spec : gs) instance Outputable EarlyDerivSpec where ppr (InferTheta spec) = ppr spec <+> text "(Infer)" ppr (GivenTheta spec) = ppr spec <+> text "(Given)" {- Note [Data decl contexts] ~~~~~~~~~~~~~~~~~~~~~~~~~ Consider data (RealFloat a) => Complex a = !a :+ !a deriving( Read ) We will need an instance decl like: instance (Read a, RealFloat a) => Read (Complex a) where ... The RealFloat in the context is because the read method for Complex is bound to construct a Complex, and doing that requires that the argument type is in RealFloat. But this ain't true for Show, Eq, Ord, etc, since they don't construct a Complex; they only take them apart. Our approach: identify the offending classes, and add the data type context to the instance decl. The "offending classes" are Read, Enum? FURTHER NOTE ADDED March 2002. In fact, Haskell98 now requires that pattern matching against a constructor from a data type with a context gives rise to the constraints for that context -- or at least the thinned version. So now all classes are "offending". Note [Newtype deriving] ~~~~~~~~~~~~~~~~~~~~~~~ Consider this: class C a b instance C [a] Char newtype T = T Char deriving( C [a] ) Notice the free 'a' in the deriving. We have to fill this out to newtype T = T Char deriving( forall a. C [a] ) And then translate it to: instance C [a] Char => C [a] T where ... Note [Newtype deriving superclasses] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ (See also Trac #1220 for an interesting exchange on newtype deriving and superclasses.) The 'tys' here come from the partial application in the deriving clause. The last arg is the new instance type. We must pass the superclasses; the newtype might be an instance of them in a different way than the representation type E.g. newtype Foo a = Foo a deriving( Show, Num, Eq ) Then the Show instance is not done via Coercible; it shows Foo 3 as "Foo 3" The Num instance is derived via Coercible, but the Show superclass dictionary must the Show instance for Foo, *not* the Show dictionary gotten from the Num dictionary. So we must build a whole new dictionary not just use the Num one. The instance we want is something like: instance (Num a, Show (Foo a), Eq (Foo a)) => Num (Foo a) where (+) = ((+)@a) ...etc... There may be a coercion needed which we get from the tycon for the newtype when the dict is constructed in TcInstDcls.tcInstDecl2 Note [Unused constructors and deriving clauses] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ See Trac #3221. Consider data T = T1 | T2 deriving( Show ) Are T1 and T2 unused? Well, no: the deriving clause expands to mention both of them. So we gather defs/uses from deriving just like anything else. -} -- | Stuff needed to process a datatype's `deriving` clauses data DerivInfo = DerivInfo { di_rep_tc :: TyCon -- ^ The data tycon for normal datatypes, -- or the *representation* tycon for data families , di_clauses :: [LHsDerivingClause Name] , di_ctxt :: SDoc -- ^ error context } -- | Extract `deriving` clauses of proper data type (skips data families) mkDerivInfos :: [LTyClDecl Name] -> TcM [DerivInfo] mkDerivInfos decls = concatMapM (mk_deriv . unLoc) decls where mk_deriv decl@(DataDecl { tcdLName = L _ data_name , tcdDataDefn = HsDataDefn { dd_derivs = L _ clauses } }) = do { tycon <- tcLookupTyCon data_name ; return [DerivInfo { di_rep_tc = tycon, di_clauses = clauses , di_ctxt = tcMkDeclCtxt decl }] } mk_deriv _ = return [] {- ************************************************************************ * * \subsection[TcDeriv-driver]{Top-level function for \tr{derivings}} * * ************************************************************************ -} tcDeriving :: [DerivInfo] -- All `deriving` clauses -> [LDerivDecl Name] -- All stand-alone deriving declarations -> TcM (TcGblEnv, Bag (InstInfo Name), HsValBinds Name) tcDeriving deriv_infos deriv_decls = recoverM (do { g <- getGblEnv ; return (g, emptyBag, emptyValBindsOut)}) $ do { -- Fish the "deriving"-related information out of the TcEnv -- And make the necessary "equations". is_boot <- tcIsHsBootOrSig ; traceTc "tcDeriving" (ppr is_boot) ; early_specs <- makeDerivSpecs is_boot deriv_infos deriv_decls ; traceTc "tcDeriving 1" (ppr early_specs) ; let (infer_specs, given_specs) = splitEarlyDerivSpec early_specs ; insts1 <- mapM genInst given_specs ; insts2 <- mapM genInst infer_specs ; dflags <- getDynFlags ; let (_, deriv_stuff, maybe_fvs) = unzip3 (insts1 ++ insts2) ; loc <- getSrcSpanM ; let (binds, famInsts) = genAuxBinds dflags loc (unionManyBags deriv_stuff) ; let mk_inst_infos1 = map fstOf3 insts1 ; inst_infos1 <- apply_inst_infos mk_inst_infos1 given_specs -- We must put all the derived type family instances (from both -- infer_specs and given_specs) in the local instance environment -- before proceeding, or else simplifyInstanceContexts might -- get stuck if it has to reason about any of those family instances. -- See Note [Staging of tcDeriving] ; tcExtendLocalFamInstEnv (bagToList famInsts) $ -- NB: only call tcExtendLocalFamInstEnv once, as it performs -- validity checking for all of the family instances you give it. -- If the family instances have errors, calling it twice will result -- in duplicate error messages! do { -- the stand-alone derived instances (@inst_infos1@) are used when -- inferring the contexts for "deriving" clauses' instances -- (@infer_specs@) ; final_specs <- extendLocalInstEnv (map iSpec inst_infos1) $ simplifyInstanceContexts infer_specs ; let mk_inst_infos2 = map fstOf3 insts2 ; inst_infos2 <- apply_inst_infos mk_inst_infos2 final_specs ; let inst_infos = inst_infos1 ++ inst_infos2 ; (inst_info, rn_binds, rn_dus) <- renameDeriv is_boot inst_infos binds ; unless (isEmptyBag inst_info) $ liftIO (dumpIfSet_dyn dflags Opt_D_dump_deriv "Derived instances" (ddump_deriving inst_info rn_binds famInsts)) ; gbl_env <- tcExtendLocalInstEnv (map iSpec (bagToList inst_info)) getGblEnv ; let all_dus = rn_dus `plusDU` usesOnly (NameSet.mkFVs $ catMaybes maybe_fvs) ; return (addTcgDUs gbl_env all_dus, inst_info, rn_binds) } } where ddump_deriving :: Bag (InstInfo Name) -> HsValBinds Name -> Bag FamInst -- ^ Rep type family instances -> SDoc ddump_deriving inst_infos extra_binds repFamInsts = hang (text "Derived class instances:") 2 (vcat (map (\i -> pprInstInfoDetails i $$ text "") (bagToList inst_infos)) $$ ppr extra_binds) $$ hangP "Derived type family instances:" (vcat (map pprRepTy (bagToList repFamInsts))) hangP s x = text "" $$ hang (ptext (sLit s)) 2 x -- Apply the suspended computations given by genInst calls. -- See Note [Staging of tcDeriving] apply_inst_infos :: [ThetaType -> TcM (InstInfo RdrName)] -> [DerivSpec ThetaType] -> TcM [InstInfo RdrName] apply_inst_infos = zipWithM (\f ds -> f (ds_theta ds)) -- Prints the representable type family instance pprRepTy :: FamInst -> SDoc pprRepTy fi@(FamInst { fi_tys = lhs }) = text "type" <+> ppr (mkTyConApp (famInstTyCon fi) lhs) <+> equals <+> ppr rhs where rhs = famInstRHS fi renameDeriv :: Bool -> [InstInfo RdrName] -> Bag (LHsBind RdrName, LSig RdrName) -> TcM (Bag (InstInfo Name), HsValBinds Name, DefUses) renameDeriv is_boot inst_infos bagBinds | is_boot -- If we are compiling a hs-boot file, don't generate any derived bindings -- The inst-info bindings will all be empty, but it's easier to -- just use rn_inst_info to change the type appropriately = do { (rn_inst_infos, fvs) <- mapAndUnzipM rn_inst_info inst_infos ; return ( listToBag rn_inst_infos , emptyValBindsOut, usesOnly (plusFVs fvs)) } | otherwise = discardWarnings $ -- Discard warnings about unused bindings etc setXOptM LangExt.EmptyCase $ -- Derived decls (for empty types) can have -- case x of {} setXOptM LangExt.ScopedTypeVariables $ setXOptM LangExt.KindSignatures $ -- Derived decls (for newtype-deriving) can use ScopedTypeVariables & -- KindSignatures unsetXOptM LangExt.RebindableSyntax $ -- See Note [Avoid RebindableSyntax when deriving] do { -- Bring the extra deriving stuff into scope -- before renaming the instances themselves ; traceTc "rnd" (vcat (map (\i -> pprInstInfoDetails i $$ text "") inst_infos)) ; (aux_binds, aux_sigs) <- mapAndUnzipBagM return bagBinds ; let aux_val_binds = ValBindsIn aux_binds (bagToList aux_sigs) ; rn_aux_lhs <- rnTopBindsLHS emptyFsEnv aux_val_binds ; let bndrs = collectHsValBinders rn_aux_lhs ; envs <- extendGlobalRdrEnvRn (map avail bndrs) emptyFsEnv ; ; setEnvs envs $ do { (rn_aux, dus_aux) <- rnValBindsRHS (TopSigCtxt (mkNameSet bndrs)) rn_aux_lhs ; (rn_inst_infos, fvs_insts) <- mapAndUnzipM rn_inst_info inst_infos ; return (listToBag rn_inst_infos, rn_aux, dus_aux `plusDU` usesOnly (plusFVs fvs_insts)) } } where rn_inst_info :: InstInfo RdrName -> TcM (InstInfo Name, FreeVars) rn_inst_info inst_info@(InstInfo { iSpec = inst , iBinds = InstBindings { ib_binds = binds , ib_tyvars = tyvars , ib_pragmas = sigs , ib_extensions = exts -- Only for type-checking , ib_derived = sa } }) = ASSERT( null sigs ) bindLocalNamesFV tyvars $ do { (rn_binds,_, fvs) <- rnMethodBinds False (is_cls_nm inst) [] binds [] ; let binds' = InstBindings { ib_binds = rn_binds , ib_tyvars = tyvars , ib_pragmas = [] , ib_extensions = exts , ib_derived = sa } ; return (inst_info { iBinds = binds' }, fvs) } {- Note [Newtype deriving and unused constructors] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Consider this (see Trac #1954): module Bug(P) where newtype P a = MkP (IO a) deriving Monad If you compile with -Wunused-binds you do not expect the warning "Defined but not used: data constructor MkP". Yet the newtype deriving code does not explicitly mention MkP, but it should behave as if you had written instance Monad P where return x = MkP (return x) ...etc... So we want to signal a user of the data constructor 'MkP'. This is the reason behind the (Maybe Name) part of the return type of genInst. Note [Staging of tcDeriving] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Here's a tricky corner case for deriving (adapted from Trac #2721): class C a where type T a foo :: a -> T a instance C Int where type T Int = Int foo = id newtype N = N Int deriving C This will produce an instance something like this: instance C N where type T N = T Int foo = coerce (foo :: Int -> T Int) :: N -> T N We must be careful in order to typecheck this code. When determining the context for the instance (in simplifyInstanceContexts), we need to determine that T N and T Int have the same representation, but to do that, the T N instance must be in the local family instance environment. Otherwise, GHC would be unable to conclude that T Int is representationally equivalent to T Int, and simplifyInstanceContexts would get stuck. Previously, tcDeriving would defer adding any derived type family instances to the instance environment until the very end, which meant that simplifyInstanceContexts would get called without all the type family instances it needed in the environment in order to properly simplify instance like the C N instance above. To avoid this scenario, we carefully structure the order of events in tcDeriving. We first call genInst on the standalone derived instance specs and the instance specs obtained from deriving clauses. Note that the return type of genInst is a triple: TcM (ThetaType -> TcM (InstInfo RdrName), BagDerivStuff, Maybe Name) The type family instances are in the BagDerivStuff. The first field of the triple is a suspended computation which, given an instance context, produces the rest of the instance. The fact that it is suspended is important, because right now, we don't have ThetaTypes for the instances that use deriving clauses (only the standalone-derived ones). Now we can can collect the type family instances and extend the local instance environment. At this point, it is safe to run simplifyInstanceContexts on the deriving-clause instance specs, which gives us the ThetaTypes for the deriving-clause instances. Now we can feed all the ThetaTypes to the suspended computations and obtain our InstInfos, at which point tcDeriving is done. An alternative design would be to split up genInst so that the family instances are generated separately from the InstInfos. But this would require carving up a lot of the GHC deriving internals to accommodate the change. On the other hand, we can keep all of the InstInfo and type family instance logic together in genInst simply by converting genInst to continuation-returning style, so we opt for that route. Note [Why we don't pass rep_tc into deriveTyData] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Down in the bowels of mkEqnHelp, we need to convert the fam_tc back into the rep_tc by means of a lookup. And yet we have the rep_tc right here! Why look it up again? Answer: it's just easier this way. We drop some number of arguments from the end of the datatype definition in deriveTyData. The arguments are dropped from the fam_tc. This action may drop a *different* number of arguments passed to the rep_tc, depending on how many free variables, etc., the dropped patterns have. Also, this technique carries over the kind substitution from deriveTyData nicely. Note [Avoid RebindableSyntax when deriving] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ The RebindableSyntax extension interacts awkwardly with the derivation of any stock class whose methods require the use of string literals. The Show class is a simple example (see Trac #12688): {-# LANGUAGE RebindableSyntax, OverloadedStrings #-} newtype Text = Text String fromString :: String -> Text fromString = Text data Foo = Foo deriving Show This will generate code to the effect of: instance Show Foo where showsPrec _ Foo = showString "Foo" But because RebindableSyntax and OverloadedStrings are enabled, the "Foo" string literal is now of type Text, not String, which showString doesn't accept! This causes the generated Show instance to fail to typecheck. To avoid this kind of scenario, we simply turn off RebindableSyntax entirely in derived code. ************************************************************************ * * From HsSyn to DerivSpec * * ************************************************************************ @makeDerivSpecs@ fishes around to find the info about needed derived instances. -} makeDerivSpecs :: Bool -> [DerivInfo] -> [LDerivDecl Name] -> TcM [EarlyDerivSpec] makeDerivSpecs is_boot deriv_infos deriv_decls = do { eqns1 <- concatMapM (recoverM (return []) . deriveDerivInfo) deriv_infos ; eqns2 <- concatMapM (recoverM (return []) . deriveStandalone) deriv_decls ; let eqns = eqns1 ++ eqns2 ; if is_boot then -- No 'deriving' at all in hs-boot files do { unless (null eqns) (add_deriv_err (head eqns)) ; return [] } else return eqns } where add_deriv_err eqn = setSrcSpan (earlyDSLoc eqn) $ addErr (hang (text "Deriving not permitted in hs-boot file") 2 (text "Use an instance declaration instead")) ------------------------------------------------------------------ -- | Process a `deriving` clause deriveDerivInfo :: DerivInfo -> TcM [EarlyDerivSpec] deriveDerivInfo (DerivInfo { di_rep_tc = rep_tc, di_clauses = clauses , di_ctxt = err_ctxt }) = addErrCtxt err_ctxt $ concatMapM (deriveForClause . unLoc) clauses where tvs = tyConTyVars rep_tc (tc, tys) = case tyConFamInstSig_maybe rep_tc of -- data family: Just (fam_tc, pats, _) -> (fam_tc, pats) -- NB: deriveTyData wants the *user-specified* -- name. See Note [Why we don't pass rep_tc into deriveTyData] _ -> (rep_tc, mkTyVarTys tvs) -- datatype deriveForClause :: HsDerivingClause Name -> TcM [EarlyDerivSpec] deriveForClause (HsDerivingClause { deriv_clause_strategy = dcs , deriv_clause_tys = L _ preds }) = concatMapM (deriveTyData tvs tc tys (fmap unLoc dcs)) preds ------------------------------------------------------------------ deriveStandalone :: LDerivDecl Name -> TcM [EarlyDerivSpec] -- Standalone deriving declarations -- e.g. deriving instance Show a => Show (T a) -- Rather like tcLocalInstDecl deriveStandalone (L loc (DerivDecl deriv_ty deriv_strat' overlap_mode)) = setSrcSpan loc $ addErrCtxt (standaloneCtxt deriv_ty) $ do { traceTc "Standalone deriving decl for" (ppr deriv_ty) ; let deriv_strat = fmap unLoc deriv_strat' ; traceTc "Deriving strategy (standalone deriving)" $ vcat [ppr deriv_strat, ppr deriv_ty] ; (tvs, theta, cls, inst_tys) <- tcHsClsInstType TcType.InstDeclCtxt deriv_ty ; traceTc "Standalone deriving;" $ vcat [ text "tvs:" <+> ppr tvs , text "theta:" <+> ppr theta , text "cls:" <+> ppr cls , text "tys:" <+> ppr inst_tys ] -- C.f. TcInstDcls.tcLocalInstDecl1 ; checkTc (not (null inst_tys)) derivingNullaryErr ; let cls_tys = take (length inst_tys - 1) inst_tys inst_ty = last inst_tys ; traceTc "Standalone deriving:" $ vcat [ text "class:" <+> ppr cls , text "class types:" <+> ppr cls_tys , text "type:" <+> ppr inst_ty ] ; let bale_out msg = failWithTc (derivingThingErr False cls cls_tys inst_ty deriv_strat msg) ; case tcSplitTyConApp_maybe inst_ty of Just (tc, tc_args) | className cls == typeableClassName -> do warnUselessTypeable return [] | isUnboxedTupleTyCon tc -> bale_out $ unboxedTyConErr "tuple" | isUnboxedSumTyCon tc -> bale_out $ unboxedTyConErr "sum" | isAlgTyCon tc || isDataFamilyTyCon tc -- All other classes -> do { spec <- mkEqnHelp (fmap unLoc overlap_mode) tvs cls cls_tys tc tc_args (Just theta) deriv_strat ; return [spec] } _ -> -- Complain about functions, primitive types, etc, bale_out $ text "The last argument of the instance must be a data or newtype application" } warnUselessTypeable :: TcM () warnUselessTypeable = do { warn <- woptM Opt_WarnDerivingTypeable ; when warn $ addWarnTc (Reason Opt_WarnDerivingTypeable) $ text "Deriving" <+> quotes (ppr typeableClassName) <+> text "has no effect: all types now auto-derive Typeable" } ------------------------------------------------------------------ deriveTyData :: [TyVar] -> TyCon -> [Type] -- LHS of data or data instance -- Can be a data instance, hence [Type] args -> Maybe DerivStrategy -- The optional deriving strategy -> LHsSigType Name -- The deriving predicate -> TcM [EarlyDerivSpec] -- The deriving clause of a data or newtype declaration -- I.e. not standalone deriving deriveTyData tvs tc tc_args deriv_strat deriv_pred = setSrcSpan (getLoc (hsSigType deriv_pred)) $ -- Use loc of the 'deriving' item do { (deriv_tvs, cls, cls_tys, cls_arg_kinds) <- tcExtendTyVarEnv tvs $ tcHsDeriv deriv_pred -- Deriving preds may (now) mention -- the type variables for the type constructor, hence tcExtendTyVarenv -- The "deriv_pred" is a LHsType to take account of the fact that for -- newtype deriving we allow deriving (forall a. C [a]). -- Typeable is special, because Typeable :: forall k. k -> Constraint -- so the argument kind 'k' is not decomposable by splitKindFunTys -- as is the case for all other derivable type classes ; when (length cls_arg_kinds /= 1) $ failWithTc (nonUnaryErr deriv_pred) ; let [cls_arg_kind] = cls_arg_kinds ; if className cls == typeableClassName then do warnUselessTypeable return [] else do { -- Given data T a b c = ... deriving( C d ), -- we want to drop type variables from T so that (C d (T a)) is well-kinded let (arg_kinds, _) = splitFunTys cls_arg_kind n_args_to_drop = length arg_kinds n_args_to_keep = tyConArity tc - n_args_to_drop (tc_args_to_keep, args_to_drop) = splitAt n_args_to_keep tc_args inst_ty_kind = typeKind (mkTyConApp tc tc_args_to_keep) -- Match up the kinds, and apply the resulting kind substitution -- to the types. See Note [Unify kinds in deriving] -- We are assuming the tycon tyvars and the class tyvars are distinct mb_match = tcUnifyTy inst_ty_kind cls_arg_kind enough_args = n_args_to_keep >= 0 -- Check that the result really is well-kinded ; checkTc (enough_args && isJust mb_match) (derivingKindErr tc cls cls_tys cls_arg_kind enough_args) ; let Just kind_subst = mb_match ki_subst_range = getTCvSubstRangeFVs kind_subst all_tkvs = toposortTyVars $ fvVarList $ unionFV (tyCoFVsOfTypes tc_args_to_keep) (FV.mkFVs deriv_tvs) -- See Note [Unification of two kind variables in deriving] unmapped_tkvs = filter (\v -> v `notElemTCvSubst` kind_subst && not (v `elemVarSet` ki_subst_range)) all_tkvs (subst, _) = mapAccumL substTyVarBndr kind_subst unmapped_tkvs final_tc_args = substTys subst tc_args_to_keep final_cls_tys = substTys subst cls_tys tkvs = tyCoVarsOfTypesWellScoped $ final_cls_tys ++ final_tc_args ; traceTc "Deriving strategy (deriving clause)" $ vcat [ppr deriv_strat, ppr deriv_pred] ; traceTc "derivTyData1" (vcat [ pprTyVars tvs, ppr tc, ppr tc_args , ppr deriv_pred , pprTyVars (tyCoVarsOfTypesList tc_args) , ppr n_args_to_keep, ppr n_args_to_drop , ppr inst_ty_kind, ppr cls_arg_kind, ppr mb_match , ppr final_tc_args, ppr final_cls_tys ]) ; traceTc "derivTyData2" (vcat [ ppr tkvs ]) ; checkTc (allDistinctTyVars (mkVarSet tkvs) args_to_drop) -- (a, b, c) (derivingEtaErr cls final_cls_tys (mkTyConApp tc final_tc_args)) -- Check that -- (a) The args to drop are all type variables; eg reject: -- data instance T a Int = .... deriving( Monad ) -- (b) The args to drop are all *distinct* type variables; eg reject: -- class C (a :: * -> * -> *) where ... -- data instance T a a = ... deriving( C ) -- (c) The type class args, or remaining tycon args, -- do not mention any of the dropped type variables -- newtype T a s = ... deriving( ST s ) -- newtype instance K a a = ... deriving( Monad ) -- -- It is vital that the implementation of allDistinctTyVars -- expand any type synonyms. -- See Note [Eta-reducing type synonyms] ; spec <- mkEqnHelp Nothing tkvs cls final_cls_tys tc final_tc_args Nothing deriv_strat ; traceTc "derivTyData" (ppr spec) ; return [spec] } } {- Note [Unify kinds in deriving] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Consider (Trac #8534) data T a b = MkT a deriving( Functor ) -- where Functor :: (*->*) -> Constraint So T :: forall k. * -> k -> *. We want to get instance Functor (T * (a:*)) where ... Notice the '*' argument to T. Moreover, as well as instantiating T's kind arguments, we may need to instantiate C's kind args. Consider (Trac #8865): newtype T a b = MkT (Either a b) deriving( Category ) where Category :: forall k. (k -> k -> *) -> Constraint We need to generate the instance instance Category * (Either a) where ... Notice the '*' argument to Category. So we need to * drop arguments from (T a b) to match the number of arrows in the (last argument of the) class; * and then *unify* kind of the remaining type against the expected kind, to figure out how to instantiate C's and T's kind arguments. In the two examples, * we unify kind-of( T k (a:k) ) ~ kind-of( Functor ) i.e. (k -> *) ~ (* -> *) to find k:=*. yielding k:=* * we unify kind-of( Either ) ~ kind-of( Category ) i.e. (* -> * -> *) ~ (k -> k -> k) yielding k:=* Now we get a kind substitution. We then need to: 1. Remove the substituted-out kind variables from the quantified kind vars 2. Apply the substitution to the kinds of quantified *type* vars (and extend the substitution to reflect this change) 3. Apply that extended substitution to the non-dropped args (types and kinds) of the type and class Forgetting step (2) caused Trac #8893: data V a = V [a] deriving Functor data P (x::k->*) (a:k) = P (x a) deriving Functor data C (x::k->*) (a:k) = C (V (P x a)) deriving Functor When deriving Functor for P, we unify k to *, but we then want an instance $df :: forall (x:*->*). Functor x => Functor (P * (x:*->*)) and similarly for C. Notice the modified kind of x, both at binding and occurrence sites. This can lead to some surprising results when *visible* kind binder is unified (in contrast to the above examples, in which only non-visible kind binders were considered). Consider this example from Trac #11732: data T k (a :: k) = MkT deriving Functor Since unification yields k:=*, this results in a generated instance of: instance Functor (T *) where ... which looks odd at first glance, since one might expect the instance head to be of the form Functor (T k). Indeed, one could envision an alternative generated instance of: instance (k ~ *) => Functor (T k) where But this does not typecheck as the result of a -XTypeInType design decision: kind equalities are not allowed to be bound in types, only terms. But in essence, the two instance declarations are entirely equivalent, since even though (T k) matches any kind k, the only possibly value for k is *, since anything else is ill-typed. As a result, we can just as comfortably use (T *). Another way of thinking about is: deriving clauses often infer constraints. For example: data S a = S a deriving Eq infers an (Eq a) constraint in the derived instance. By analogy, when we are deriving Functor, we might infer an equality constraint (e.g., k ~ *). The only distinction is that GHC instantiates equality constraints directly during the deriving process. Another quirk of this design choice manifests when typeclasses have visible kind parameters. Consider this code (also from Trac #11732): class Cat k (cat :: k -> k -> *) where catId :: cat a a catComp :: cat b c -> cat a b -> cat a c instance Cat * (->) where catId = id catComp = (.) newtype Fun a b = Fun (a -> b) deriving (Cat k) Even though we requested an derived instance of the form (Cat k Fun), the kind unification will actually generate (Cat * Fun) (i.e., the same thing as if the user wrote deriving (Cat *)). Note [Unification of two kind variables in deriving] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ As a special case of the Note above, it is possible to derive an instance of a poly-kinded typeclass for a poly-kinded datatype. For example: class Category (cat :: k -> k -> *) where newtype T (c :: k -> k -> *) a b = MkT (c a b) deriving Category This case is suprisingly tricky. To see why, let's write out what instance GHC will attempt to derive (using -fprint-explicit-kinds syntax): instance Category k1 (T k2 c) where ... GHC will attempt to unify k1 and k2, which produces a substitution (kind_subst) that looks like [k2 :-> k1]. Importantly, we need to apply this substitution to the type variable binder for c, since its kind is (k2 -> k2 -> *). We used to accomplish this by doing the following: unmapped_tkvs = filter (`notElemTCvSubst` kind_subst) all_tkvs (subst, _) = mapAccumL substTyVarBndr kind_subst unmapped_tkvs Where all_tkvs contains all kind variables in the class and instance types (in this case, all_tkvs = [k1,k2]). But since kind_subst only has one mapping, this results in unmapped_tkvs being [k1], and as a consequence, k1 gets mapped to another kind variable in subst! That is, subst = [k2 :-> k1, k1 :-> k_new]. This is bad, because applying that substitution yields the following instance: instance Category k_new (T k1 c) where ... In other words, keeping k1 in unmapped_tvks taints the substitution, resulting in an ill-kinded instance (this caused Trac #11837). To prevent this, we need to filter out any variable from all_tkvs which either 1. Appears in the domain of kind_subst. notElemTCvSubst checks this. 2. Appears in the range of kind_subst. To do this, we compute the free variable set of the range of kind_subst with getTCvSubstRangeFVs, and check if a kind variable appears in that set. Note [Eta-reducing type synonyms] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ One can instantiate a type in a data family instance with a type synonym that mentions other type variables: type Const a b = a data family Fam (f :: * -> *) (a :: *) newtype instance Fam f (Const a f) = Fam (f a) deriving Functor With -XTypeInType, it is also possible to define kind synonyms, and they can mention other types in a datatype declaration. For example, type Const a b = a newtype T f (a :: Const * f) = T (f a) deriving Functor When deriving, we need to perform eta-reduction analysis to ensure that none of the eta-reduced type variables are mentioned elsewhere in the declaration. But we need to be careful, because if we don't expand through the Const type synonym, we will mistakenly believe that f is an eta-reduced type variable and fail to derive Functor, even though the code above is correct (see Trac #11416, where this was first noticed). For this reason, we expand the type synonyms in the eta-reduced types before doing any analysis. -} mkEqnHelp :: Maybe OverlapMode -> [TyVar] -> Class -> [Type] -> TyCon -> [Type] -> DerivContext -- Just => context supplied (standalone deriving) -- Nothing => context inferred (deriving on data decl) -> Maybe DerivStrategy -> TcRn EarlyDerivSpec -- Make the EarlyDerivSpec for an instance -- forall tvs. theta => cls (tys ++ [ty]) -- where the 'theta' is optional (that's the Maybe part) -- Assumes that this declaration is well-kinded mkEqnHelp overlap_mode tvs cls cls_tys tycon tc_args mtheta deriv_strat = do { -- Find the instance of a data family -- Note [Looking up family instances for deriving] fam_envs <- tcGetFamInstEnvs ; let (rep_tc, rep_tc_args, _co) = tcLookupDataFamInst fam_envs tycon tc_args -- If it's still a data family, the lookup failed; i.e no instance exists ; when (isDataFamilyTyCon rep_tc) (bale_out (text "No family instance for" <+> quotes (pprTypeApp tycon tc_args))) ; dflags <- getDynFlags ; if isDataTyCon rep_tc then mkDataTypeEqn dflags overlap_mode tvs cls cls_tys tycon tc_args rep_tc rep_tc_args mtheta deriv_strat else mkNewTypeEqn dflags overlap_mode tvs cls cls_tys tycon tc_args rep_tc rep_tc_args mtheta deriv_strat } where bale_out msg = failWithTc (derivingThingErr False cls cls_tys (mkTyConApp tycon tc_args) deriv_strat msg) {- Note [Looking up family instances for deriving] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ tcLookupFamInstExact is an auxiliary lookup wrapper which requires that looked-up family instances exist. If called with a vanilla tycon, the old type application is simply returned. If we have data instance F () = ... deriving Eq data instance F () = ... deriving Eq then tcLookupFamInstExact will be confused by the two matches; but that can't happen because tcInstDecls1 doesn't call tcDeriving if there are any overlaps. There are two other things that might go wrong with the lookup. First, we might see a standalone deriving clause deriving Eq (F ()) when there is no data instance F () in scope. Note that it's OK to have data instance F [a] = ... deriving Eq (F [(a,b)]) where the match is not exact; the same holds for ordinary data types with standalone deriving declarations. Note [Deriving, type families, and partial applications] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ When there are no type families, it's quite easy: newtype S a = MkS [a] -- :CoS :: S ~ [] -- Eta-reduced instance Eq [a] => Eq (S a) -- by coercion sym (Eq (:CoS a)) : Eq [a] ~ Eq (S a) instance Monad [] => Monad S -- by coercion sym (Monad :CoS) : Monad [] ~ Monad S When type familes are involved it's trickier: data family T a b newtype instance T Int a = MkT [a] deriving( Eq, Monad ) -- :RT is the representation type for (T Int a) -- :Co:RT :: :RT ~ [] -- Eta-reduced! -- :CoF:RT a :: T Int a ~ :RT a -- Also eta-reduced! instance Eq [a] => Eq (T Int a) -- easy by coercion -- d1 :: Eq [a] -- d2 :: Eq (T Int a) = d1 |> Eq (sym (:Co:RT a ; :coF:RT a)) instance Monad [] => Monad (T Int) -- only if we can eta reduce??? -- d1 :: Monad [] -- d2 :: Monad (T Int) = d1 |> Monad (sym (:Co:RT ; :coF:RT)) Note the need for the eta-reduced rule axioms. After all, we can write it out instance Monad [] => Monad (T Int) -- only if we can eta reduce??? return x = MkT [x] ... etc ... See Note [Eta reduction for data families] in FamInstEnv %************************************************************************ %* * Deriving data types * * ************************************************************************ -} mkDataTypeEqn :: DynFlags -> Maybe OverlapMode -> [TyVar] -- Universally quantified type variables in the instance -> Class -- Class for which we need to derive an instance -> [Type] -- Other parameters to the class except the last -> TyCon -- Type constructor for which the instance is requested -- (last parameter to the type class) -> [Type] -- Parameters to the type constructor -> TyCon -- rep of the above (for type families) -> [Type] -- rep of the above -> DerivContext -- Context of the instance, for standalone deriving -> Maybe DerivStrategy -- 'Just' if user requests a particular -- deriving strategy. -- Otherwise, 'Nothing'. -> TcRn EarlyDerivSpec -- Return 'Nothing' if error mkDataTypeEqn dflags overlap_mode tvs cls cls_tys tycon tc_args rep_tc rep_tc_args mtheta deriv_strat = case deriv_strat of Just StockStrategy -> mk_eqn_stock dflags mtheta cls cls_tys rep_tc go_for_it bale_out Just AnyclassStrategy -> mk_eqn_anyclass dflags go_for_it bale_out -- GeneralizedNewtypeDeriving makes no sense for non-newtypes Just NewtypeStrategy -> bale_out gndNonNewtypeErr -- Lacking a user-requested deriving strategy, we will try to pick -- between the stock or anyclass strategies Nothing -> mk_eqn_no_mechanism dflags tycon mtheta cls cls_tys rep_tc go_for_it bale_out where go_for_it = mk_data_eqn overlap_mode tvs cls cls_tys tycon tc_args rep_tc rep_tc_args mtheta (isJust deriv_strat) bale_out msg = failWithTc (derivingThingErr False cls cls_tys (mkTyConApp tycon tc_args) deriv_strat msg) mk_data_eqn :: Maybe OverlapMode -> [TyVar] -> Class -> [Type] -> TyCon -> [TcType] -> TyCon -> [TcType] -> DerivContext -> Bool -- True if an explicit deriving strategy keyword was -- provided -> DerivSpecMechanism -- How GHC should proceed attempting to -- derive this instance, determined in -- mkDataTypeEqn/mkNewTypeEqn -> TcM EarlyDerivSpec mk_data_eqn overlap_mode tvs cls cls_tys tycon tc_args rep_tc rep_tc_args mtheta strat_used mechanism = do doDerivInstErrorChecks1 cls cls_tys tycon tc_args rep_tc mtheta strat_used mechanism loc <- getSrcSpanM dfun_name <- newDFunName' cls tycon case mtheta of Nothing -> -- Infer context do { (inferred_constraints, tvs', inst_tys') <- inferConstraints tvs cls cls_tys inst_ty rep_tc rep_tc_args mechanism ; return $ InferTheta $ DS { ds_loc = loc , ds_name = dfun_name, ds_tvs = tvs' , ds_cls = cls, ds_tys = inst_tys' , ds_tc = rep_tc , ds_theta = inferred_constraints , ds_overlap = overlap_mode , ds_mechanism = mechanism } } Just theta -> do -- Specified context return $ GivenTheta $ DS { ds_loc = loc , ds_name = dfun_name, ds_tvs = tvs , ds_cls = cls, ds_tys = inst_tys , ds_tc = rep_tc , ds_theta = theta , ds_overlap = overlap_mode , ds_mechanism = mechanism } where inst_ty = mkTyConApp tycon tc_args inst_tys = cls_tys ++ [inst_ty] mk_eqn_stock :: DynFlags -> DerivContext -> Class -> [Type] -> TyCon -> (DerivSpecMechanism -> TcRn EarlyDerivSpec) -> (SDoc -> TcRn EarlyDerivSpec) -> TcRn EarlyDerivSpec mk_eqn_stock dflags mtheta cls cls_tys rep_tc go_for_it bale_out = case checkSideConditions dflags mtheta cls cls_tys rep_tc of CanDerive -> mk_eqn_stock' cls go_for_it DerivableClassError msg -> bale_out msg _ -> bale_out (nonStdErr cls) mk_eqn_stock' :: Class -> (DerivSpecMechanism -> TcRn EarlyDerivSpec) -> TcRn EarlyDerivSpec mk_eqn_stock' cls go_for_it = go_for_it $ case hasStockDeriving cls of Just gen_fn -> DerivSpecStock gen_fn Nothing -> pprPanic "mk_eqn_stock': Not a stock class!" (ppr cls) mk_eqn_anyclass :: DynFlags -> (DerivSpecMechanism -> TcRn EarlyDerivSpec) -> (SDoc -> TcRn EarlyDerivSpec) -> TcRn EarlyDerivSpec mk_eqn_anyclass dflags go_for_it bale_out = case canDeriveAnyClass dflags of IsValid -> go_for_it DerivSpecAnyClass NotValid msg -> bale_out msg mk_eqn_no_mechanism :: DynFlags -> TyCon -> DerivContext -> Class -> [Type] -> TyCon -> (DerivSpecMechanism -> TcRn EarlyDerivSpec) -> (SDoc -> TcRn EarlyDerivSpec) -> TcRn EarlyDerivSpec mk_eqn_no_mechanism dflags tc mtheta cls cls_tys rep_tc go_for_it bale_out = case checkSideConditions dflags mtheta cls cls_tys rep_tc of -- NB: pass the *representation* tycon to checkSideConditions NonDerivableClass msg -> bale_out (dac_error msg) DerivableClassError msg -> bale_out msg CanDerive -> mk_eqn_stock' cls go_for_it DerivableViaInstance -> go_for_it DerivSpecAnyClass where -- See Note [Deriving instances for classes themselves] dac_error msg | isClassTyCon rep_tc = quotes (ppr tc) <+> text "is a type class," <+> text "and can only have a derived instance" $+$ text "if DeriveAnyClass is enabled" | otherwise = nonStdErr cls $$ msg {- ************************************************************************ * * Deriving newtypes * * ************************************************************************ -} mkNewTypeEqn :: DynFlags -> Maybe OverlapMode -> [TyVar] -> Class -> [Type] -> TyCon -> [Type] -> TyCon -> [Type] -> DerivContext -> Maybe DerivStrategy -> TcRn EarlyDerivSpec mkNewTypeEqn dflags overlap_mode tvs cls cls_tys tycon tc_args rep_tycon rep_tc_args mtheta deriv_strat -- Want: instance (...) => cls (cls_tys ++ [tycon tc_args]) where ... = ASSERT( length cls_tys + 1 == classArity cls ) case deriv_strat of Just StockStrategy -> mk_eqn_stock dflags mtheta cls cls_tys rep_tycon go_for_it_other bale_out Just AnyclassStrategy -> mk_eqn_anyclass dflags go_for_it_other bale_out Just NewtypeStrategy -> -- Since the user explicitly asked for GeneralizedNewtypeDeriving, we -- don't need to perform all of the checks we normally would, such as -- if the class being derived is known to produce ill-roled coercions -- (e.g., Traversable), since we can just derive the instance and let -- it error if need be. -- See Note [Determining whether newtype-deriving is appropriate] if coercion_looks_sensible && newtype_deriving then go_for_it_gnd else bale_out (cant_derive_err $$ if newtype_deriving then empty else suggest_gnd) Nothing | might_derive_via_coercible && ((newtype_deriving && not deriveAnyClass) || std_class_via_coercible cls) -> go_for_it_gnd | otherwise -> case checkSideConditions dflags mtheta cls cls_tys rep_tycon of DerivableClassError msg -- There's a particular corner case where -- -- 1. -XGeneralizedNewtypeDeriving and -XDeriveAnyClass are both -- enabled at the same time -- 2. We're deriving a particular stock derivable class -- (such as Functor) -- -- and the previous cases won't catch it. This fixes the bug -- reported in Trac #10598. | might_derive_via_coercible && newtype_deriving -> go_for_it_gnd -- Otherwise, throw an error for a stock class | might_derive_via_coercible && not newtype_deriving -> bale_out (msg $$ suggest_gnd) | otherwise -> bale_out msg -- Must use newtype deriving or DeriveAnyClass NonDerivableClass _msg -- Too hard, even with newtype deriving | newtype_deriving -> bale_out cant_derive_err -- Try newtype deriving! -- Here we suggest GeneralizedNewtypeDeriving even in cases where -- it may not be applicable. See Trac #9600. | otherwise -> bale_out (non_std $$ suggest_gnd) -- DerivableViaInstance DerivableViaInstance -> do -- If both DeriveAnyClass and GeneralizedNewtypeDeriving are -- enabled, we take the diplomatic approach of defaulting to -- DeriveAnyClass, but emitting a warning about the choice. -- See Note [Deriving strategies] when (newtype_deriving && deriveAnyClass) $ addWarnTc NoReason $ sep [ text "Both DeriveAnyClass and" <+> text "GeneralizedNewtypeDeriving are enabled" , text "Defaulting to the DeriveAnyClass strategy" <+> text "for instantiating" <+> ppr cls ] go_for_it_other DerivSpecAnyClass -- CanDerive CanDerive -> mk_eqn_stock' cls go_for_it_other where newtype_deriving = xopt LangExt.GeneralizedNewtypeDeriving dflags deriveAnyClass = xopt LangExt.DeriveAnyClass dflags go_for_it_gnd = do traceTc "newtype deriving:" $ ppr tycon <+> ppr rep_tys <+> ppr all_thetas let mechanism = DerivSpecNewtype rep_inst_ty doDerivInstErrorChecks1 cls cls_tys tycon tc_args rep_tycon mtheta strat_used mechanism dfun_name <- newDFunName' cls tycon loc <- getSrcSpanM case mtheta of Just theta -> return $ GivenTheta $ DS { ds_loc = loc , ds_name = dfun_name, ds_tvs = tvs , ds_cls = cls, ds_tys = inst_tys , ds_tc = rep_tycon , ds_theta = theta , ds_overlap = overlap_mode , ds_mechanism = mechanism } Nothing -> return $ InferTheta $ DS { ds_loc = loc , ds_name = dfun_name, ds_tvs = tvs , ds_cls = cls, ds_tys = inst_tys , ds_tc = rep_tycon , ds_theta = all_thetas , ds_overlap = overlap_mode , ds_mechanism = mechanism } go_for_it_other = mk_data_eqn overlap_mode tvs cls cls_tys tycon tc_args rep_tycon rep_tc_args mtheta strat_used bale_out = bale_out' newtype_deriving bale_out' b = failWithTc . derivingThingErr b cls cls_tys inst_ty deriv_strat strat_used = isJust deriv_strat non_std = nonStdErr cls suggest_gnd = text "Try GeneralizedNewtypeDeriving for GHC's newtype-deriving extension" -- Here is the plan for newtype derivings. We see -- newtype T a1...an = MkT (t ak+1...an) deriving (.., C s1 .. sm, ...) -- where t is a type, -- ak+1...an is a suffix of a1..an, and are all tyvars -- ak+1...an do not occur free in t, nor in the s1..sm -- (C s1 ... sm) is a *partial applications* of class C -- with the last parameter missing -- (T a1 .. ak) matches the kind of C's last argument -- (and hence so does t) -- The latter kind-check has been done by deriveTyData already, -- and tc_args are already trimmed -- -- We generate the instance -- instance forall ({a1..ak} u fvs(s1..sm)). -- C s1 .. sm t => C s1 .. sm (T a1...ak) -- where T a1...ap is the partial application of -- the LHS of the correct kind and p >= k -- -- NB: the variables below are: -- tc_tvs = [a1, ..., an] -- tyvars_to_keep = [a1, ..., ak] -- rep_ty = t ak .. an -- deriv_tvs = fvs(s1..sm) \ tc_tvs -- tys = [s1, ..., sm] -- rep_fn' = t -- -- Running example: newtype T s a = MkT (ST s a) deriving( Monad ) -- We generate the instance -- instance Monad (ST s) => Monad (T s) where nt_eta_arity = newTyConEtadArity rep_tycon -- For newtype T a b = MkT (S a a b), the TyCon machinery already -- eta-reduces the representation type, so we know that -- T a ~ S a a -- That's convenient here, because we may have to apply -- it to fewer than its original complement of arguments -- Note [Newtype representation] -- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ -- Need newTyConRhs (*not* a recursive representation finder) -- to get the representation type. For example -- newtype B = MkB Int -- newtype A = MkA B deriving( Num ) -- We want the Num instance of B, *not* the Num instance of Int, -- when making the Num instance of A! rep_inst_ty = newTyConInstRhs rep_tycon rep_tc_args rep_tys = cls_tys ++ [rep_inst_ty] rep_pred = mkClassPred cls rep_tys rep_pred_o = mkPredOrigin DerivOrigin TypeLevel rep_pred -- rep_pred is the representation dictionary, from where -- we are gong to get all the methods for the newtype -- dictionary -- Next we figure out what superclass dictionaries to use -- See Note [Newtype deriving superclasses] above sc_preds :: [PredOrigin] cls_tyvars = classTyVars cls inst_ty = mkTyConApp tycon tc_args inst_tys = cls_tys ++ [inst_ty] sc_preds = map (mkPredOrigin DerivOrigin TypeLevel) $ substTheta (zipTvSubst cls_tyvars inst_tys) $ classSCTheta cls -- Next we collect constraints for the class methods -- If there are no methods, we don't need any constraints -- Otherwise we need (C rep_ty), for the representation methods, -- and constraints to coerce each individual method meth_preds :: [PredOrigin] meths = classMethods cls meth_preds | null meths = [] -- No methods => no constraints -- (Trac #12814) | otherwise = rep_pred_o : coercible_constraints coercible_constraints = [ mkPredOrigin (DerivOriginCoerce meth t1 t2) TypeLevel (mkReprPrimEqPred t1 t2) | meth <- meths , let (Pair t1 t2) = mkCoerceClassMethEqn cls tvs inst_tys rep_inst_ty meth ] all_thetas :: [ThetaOrigin] all_thetas = [mkThetaOriginFromPreds $ meth_preds ++ sc_preds] ------------------------------------------------------------------- -- Figuring out whether we can only do this newtype-deriving thing -- See Note [Determining whether newtype-deriving is appropriate] might_derive_via_coercible = not (non_coercible_class cls) && coercion_looks_sensible -- && not (isRecursiveTyCon tycon) -- Note [Recursive newtypes] coercion_looks_sensible = eta_ok -- Check (a) from Note [GND and associated type families] && ats_ok -- Check (b) from Note [GND and associated type families] && isNothing at_without_last_cls_tv -- Check that eta reduction is OK eta_ok = nt_eta_arity <= length rep_tc_args -- The newtype can be eta-reduced to match the number -- of type argument actually supplied -- newtype T a b = MkT (S [a] b) deriving( Monad ) -- Here the 'b' must be the same in the rep type (S [a] b) -- And the [a] must not mention 'b'. That's all handled -- by nt_eta_rity. (adf_tcs, atf_tcs) = partition isDataFamilyTyCon at_tcs ats_ok = null adf_tcs -- We cannot newtype-derive data family instances at_without_last_cls_tv = find (\tc -> last_cls_tv `notElem` tyConTyVars tc) atf_tcs at_tcs = classATs cls last_cls_tv = ASSERT( notNull cls_tyvars ) last cls_tyvars cant_derive_err = vcat [ ppUnless eta_ok eta_msg , ppUnless ats_ok ats_msg , maybe empty at_tv_msg at_without_last_cls_tv] eta_msg = text "cannot eta-reduce the representation type enough" ats_msg = text "the class has associated data types" at_tv_msg at_tc = hang (text "the associated type" <+> quotes (ppr at_tc) <+> text "is not parameterized over the last type variable") 2 (text "of the class" <+> quotes (ppr cls)) {- Note [Recursive newtypes] ~~~~~~~~~~~~~~~~~~~~~~~~~ Newtype deriving works fine, even if the newtype is recursive. e.g. newtype S1 = S1 [T1 ()] newtype T1 a = T1 (StateT S1 IO a ) deriving( Monad ) Remember, too, that type families are currently (conservatively) given a recursive flag, so this also allows newtype deriving to work for type famillies. We used to exclude recursive types, because we had a rather simple minded way of generating the instance decl: newtype A = MkA [A] instance Eq [A] => Eq A -- Makes typechecker loop! But now we require a simple context, so it's ok. Note [Determining whether newtype-deriving is appropriate] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ When we see newtype NT = MkNT Foo deriving C we have to decide how to perform the deriving. Do we do newtype deriving, or do we do normal deriving? In general, we prefer to do newtype deriving wherever possible. So, we try newtype deriving unless there's a glaring reason not to. "Glaring reasons not to" include trying to derive a class for which a coercion-based instance doesn't make sense. These classes are listed in the definition of non_coercible_class. They include Show (since it must show the name of the datatype) and Traversable (since a coercion-based Traversable instance is ill-roled). However, non_coercible_class is ignored if the user explicitly requests to derive an instance with GeneralizedNewtypeDeriving using the newtype deriving strategy. In such a scenario, GHC will unquestioningly try to derive the instance via coercions (even if the final generated code is ill-roled!). See Note [Deriving strategies]. Note that newtype deriving might fail, even after we commit to it. This is because the derived instance uses `coerce`, which must satisfy its `Coercible` constraint. This is different than other deriving scenarios, where we're sure that the resulting instance will type-check. Note [GND and associated type families] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ It's possible to use GeneralizedNewtypeDeriving (GND) to derive instances for classes with associated type families. A general recipe is: class C x y z where type T y z x op :: x -> [y] -> z newtype N a = MkN <rep-type> deriving( C ) =====> instance C x y <rep-type> => C x y (N a) where type T y (N a) x = T y <rep-type> x op = coerce (op :: x -> [y] -> <rep-type>) However, we must watch out for three things: (a) The class must not contain any data families. If it did, we'd have to generate a fresh data constructor name for the derived data family instance, and it's not clear how to do this. (b) Each associated type family's type variables must mention the last type variable of the class. As an example, you wouldn't be able to use GND to derive an instance of this class: class C a b where type T a But you would be able to derive an instance of this class: class C a b where type T b The difference is that in the latter T mentions the last parameter of C (i.e., it mentions b), but the former T does not. If you tried, e.g., newtype Foo x = Foo x deriving (C a) with the former definition of C, you'd end up with something like this: instance C a (Foo x) where type T a = T ??? This T family instance doesn't mention the newtype (or its representation type) at all, so we disallow such constructions with GND. (c) UndecidableInstances might need to be enabled. Here's a case where it is most definitely necessary: class C a where type T a newtype Loop = Loop MkLoop deriving C =====> instance C Loop where type T Loop = T Loop Obviously, T Loop would send the typechecker into a loop. Unfortunately, you might even need UndecidableInstances even in cases where the typechecker would be guaranteed to terminate. For example: instance C Int where type C Int = Int newtype MyInt = MyInt Int deriving C =====> instance C MyInt where type T MyInt = T Int GHC's termination checker isn't sophisticated enough to conclude that the definition of T MyInt terminates, so UndecidableInstances is required. ************************************************************************ * * \subsection[TcDeriv-normal-binds]{Bindings for the various classes} * * ************************************************************************ After all the trouble to figure out the required context for the derived instance declarations, all that's left is to chug along to produce them. They will then be shoved into @tcInstDecls2@, which will do all its usual business. There are lots of possibilities for code to generate. Here are various general remarks. PRINCIPLES: \begin{itemize} \item We want derived instances of @Eq@ and @Ord@ (both v common) to be ``you-couldn't-do-better-by-hand'' efficient. \item Deriving @Show@---also pretty common--- should also be reasonable good code. \item Deriving for the other classes isn't that common or that big a deal. \end{itemize} PRAGMATICS: \begin{itemize} \item Deriving @Ord@ is done mostly with the 1.3 @compare@ method. \item Deriving @Eq@ also uses @compare@, if we're deriving @Ord@, too. \item We {\em normally} generate code only for the non-defaulted methods; there are some exceptions for @Eq@ and (especially) @Ord@... \item Sometimes we use a @_con2tag_<tycon>@ function, which returns a data constructor's numeric (@Int#@) tag. These are generated by @gen_tag_n_con_binds@, and the heuristic for deciding if one of these is around is given by @hasCon2TagFun@. The examples under the different sections below will make this clearer. \item Much less often (really just for deriving @Ix@), we use a @_tag2con_<tycon>@ function. See the examples. \item We use the renamer!!! Reason: we're supposed to be producing @LHsBinds Name@ for the methods, but that means producing correctly-uniquified code on the fly. This is entirely possible (the @TcM@ monad has a @UniqueSupply@), but it is painful. So, instead, we produce @MonoBinds RdrName@ then heave 'em through the renamer. What a great hack! \end{itemize} -} -- Generate the InstInfo for the required instance paired with the -- *representation* tycon for that instance, -- plus any auxiliary bindings required -- -- Representation tycons differ from the tycon in the instance signature in -- case of instances for indexed families. -- genInst :: DerivSpec theta -> TcM (ThetaType -> TcM (InstInfo RdrName), BagDerivStuff, Maybe Name) -- We must use continuation-returning style here to get the order in which we -- typecheck family instances and derived instances right. -- See Note [Staging of tcDeriving] genInst spec@(DS { ds_tvs = tvs, ds_tc = rep_tycon , ds_mechanism = mechanism, ds_tys = tys , ds_cls = clas, ds_loc = loc }) = do (meth_binds, deriv_stuff) <- genDerivStuff mechanism loc clas rep_tycon tys tvs let mk_inst_info theta = do inst_spec <- newDerivClsInst theta spec doDerivInstErrorChecks2 clas inst_spec mechanism traceTc "newder" (ppr inst_spec) return $ InstInfo { iSpec = inst_spec , iBinds = InstBindings { ib_binds = meth_binds , ib_tyvars = map Var.varName tvs , ib_pragmas = [] , ib_extensions = extensions , ib_derived = True } } return (mk_inst_info, deriv_stuff, unusedConName) where unusedConName :: Maybe Name unusedConName | isDerivSpecNewtype mechanism -- See Note [Newtype deriving and unused constructors] = Just $ getName $ head $ tyConDataCons rep_tycon | otherwise = Nothing extensions :: [LangExt.Extension] extensions | isDerivSpecNewtype mechanism -- Both these flags are needed for higher-rank uses of coerce -- See Note [Newtype-deriving instances] in TcGenDeriv = [LangExt.ImpredicativeTypes, LangExt.RankNTypes] | otherwise = [] doDerivInstErrorChecks1 :: Class -> [Type] -> TyCon -> [Type] -> TyCon -> DerivContext -> Bool -> DerivSpecMechanism -> TcM () doDerivInstErrorChecks1 cls cls_tys tc tc_args rep_tc mtheta strat_used mechanism = do -- For standalone deriving (mtheta /= Nothing), -- check that all the data constructors are in scope... rdr_env <- getGlobalRdrEnv let data_con_names = map dataConName (tyConDataCons rep_tc) hidden_data_cons = not (isWiredInName (tyConName rep_tc)) && (isAbstractTyCon rep_tc || any not_in_scope data_con_names) not_in_scope dc = isNothing (lookupGRE_Name rdr_env dc) addUsedDataCons rdr_env rep_tc -- ...however, we don't perform this check if we're using DeriveAnyClass, -- since it doesn't generate any code that requires use of a data -- constructor. unless (anyclass_strategy || isNothing mtheta || not hidden_data_cons) $ bale_out $ derivingHiddenErr tc where anyclass_strategy = isDerivSpecAnyClass mechanism bale_out msg = failWithTc (derivingThingErrMechanism cls cls_tys (mkTyConApp tc tc_args) strat_used mechanism msg) doDerivInstErrorChecks2 :: Class -> ClsInst -> DerivSpecMechanism -> TcM () doDerivInstErrorChecks2 clas clas_inst mechanism = do { traceTc "doDerivInstErrorChecks2" (ppr clas_inst) ; dflags <- getDynFlags -- Check for Generic instances that are derived with an exotic -- deriving strategy like DAC -- See Note [Deriving strategies] ; when (exotic_mechanism && className clas `elem` genericClassNames) $ do { failIfTc (safeLanguageOn dflags) gen_inst_err ; when (safeInferOn dflags) (recordUnsafeInfer emptyBag) } } where exotic_mechanism = case mechanism of DerivSpecStock{} -> False _ -> True gen_inst_err = hang (text ("Generic instances can only be derived in " ++ "Safe Haskell using the stock strategy.") $+$ text "In the following instance:") 2 (pprInstanceHdr clas_inst) genDerivStuff :: DerivSpecMechanism -> SrcSpan -> Class -> TyCon -> [Type] -> [TyVar] -> TcM (LHsBinds RdrName, BagDerivStuff) genDerivStuff mechanism loc clas tycon inst_tys tyvars = case mechanism of -- See Note [Bindings for Generalised Newtype Deriving] DerivSpecNewtype rhs_ty -> gen_Newtype_binds loc clas tyvars inst_tys rhs_ty -- Try a stock deriver DerivSpecStock gen_fn -> gen_fn loc tycon inst_tys -- If there isn't a stock deriver, our last resort is -XDeriveAnyClass -- (since -XGeneralizedNewtypeDeriving fell through). DerivSpecAnyClass -> do let mini_env = mkVarEnv (classTyVars clas `zip` inst_tys) mini_subst = mkTvSubst (mkInScopeSet (mkVarSet tyvars)) mini_env dflags <- getDynFlags tyfam_insts <- -- canDeriveAnyClass should ensure that this code can't be reached -- unless -XDeriveAnyClass is enabled. ASSERT2( isValid (canDeriveAnyClass dflags) , ppr "genDerivStuff: bad derived class" <+> ppr clas ) mapM (tcATDefault False loc mini_subst emptyNameSet) (classATItems clas) return ( emptyBag -- No method bindings are needed... , listToBag (map DerivFamInst (concat tyfam_insts)) -- ...but we may need to generate binding for associated type -- family default instances. -- See Note [DeriveAnyClass and default family instances] ) {- Note [Bindings for Generalised Newtype Deriving] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Consider class Eq a => C a where f :: a -> a newtype N a = MkN [a] deriving( C ) instance Eq (N a) where ... The 'deriving C' clause generates, in effect instance (C [a], Eq a) => C (N a) where f = coerce (f :: [a] -> [a]) This generates a cast for each method, but allows the superclasse to be worked out in the usual way. In this case the superclass (Eq (N a)) will be solved by the explicit Eq (N a) instance. We do *not* create the superclasses by casting the superclass dictionaries for the representation type. See the paper "Safe zero-cost coercions for Haskell". Note [DeriveAnyClass and default family instances] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ When a class has a associated type family with a default instance, e.g.: class C a where type T a type T a = Char then there are a couple of scenarios in which a user would expect T a to default to Char. One is when an instance declaration for C is given without an implementation for T: instance C Int Another scenario in which this can occur is when the -XDeriveAnyClass extension is used: data Example = Example deriving (C, Generic) In the latter case, we must take care to check if C has any associated type families with default instances, because -XDeriveAnyClass will never provide an implementation for them. We "fill in" the default instances using the tcATDefault function from TcClsDcl (which is also used in TcInstDcls to handle the empty instance declaration case). Note [Deriving strategies] ~~~~~~~~~~~~~~~~~~~~~~~~~~ GHC has a notion of deriving strategies, which allow the user to explicitly request which approach to use when deriving an instance (enabled with the -XDerivingStrategies language extension). For more information, refer to the original Trac ticket (#10598) or the associated wiki page: https://ghc.haskell.org/trac/ghc/wiki/Commentary/Compiler/DerivingStrategies A deriving strategy can be specified in a deriving clause: newtype Foo = MkFoo Bar deriving newtype C Or in a standalone deriving declaration: deriving anyclass instance C Foo -XDerivingStrategies also allows the use of multiple deriving clauses per data declaration so that a user can derive some instance with one deriving strategy and other instances with another deriving strategy. For example: newtype Baz = Baz Quux deriving (Eq, Ord) deriving stock (Read, Show) deriving newtype (Num, Floating) deriving anyclass C Currently, the deriving strategies are: * stock: Have GHC implement a "standard" instance for a data type, if possible (e.g., Eq, Ord, Generic, Data, Functor, etc.) * anyclass: Use -XDeriveAnyClass * newtype: Use -XGeneralizedNewtypeDeriving If an explicit deriving strategy is not given, GHC has an algorithm it uses to determine which strategy it will actually use. The algorithm is quite long, so it lives in the Haskell wiki at https://ghc.haskell.org/trac/ghc/wiki/Commentary/Compiler/DerivingStrategies ("The deriving strategy resolution algorithm" section). Internally, GHC uses the DerivStrategy datatype to denote a user-requested deriving strategy, and it uses the DerivSpecMechanism datatype to denote what GHC will use to derive the instance after taking the above steps. In other words, GHC will always settle on a DerivSpecMechnism, even if the user did not ask for a particular DerivStrategy (using the algorithm linked to above). Note [Deriving instances for classes themselves] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Much of the code in TcDeriv assumes that deriving only works on data types. But this assumption doesn't hold true for DeriveAnyClass, since it's perfectly reasonable to do something like this: {-# LANGUAGE DeriveAnyClass #-} class C1 (a :: Constraint) where class C2 where deriving instance C1 C2 -- This is equivalent to `instance C1 C2` If DeriveAnyClass isn't enabled in the code above (i.e., it defaults to stock deriving), we throw a special error message indicating that DeriveAnyClass is the only way to go. We don't bother throwing this error if an explicit 'stock' or 'newtype' keyword is used, since both options have their own perfectly sensible error messages in the case of the above code (as C1 isn't a stock derivable class, and C2 isn't a newtype). ************************************************************************ * * \subsection[TcDeriv-taggery-Names]{What con2tag/tag2con functions are available?} * * ************************************************************************ -} nonUnaryErr :: LHsSigType Name -> SDoc nonUnaryErr ct = quotes (ppr ct) <+> text "is not a unary constraint, as expected by a deriving clause" nonStdErr :: Class -> SDoc nonStdErr cls = quotes (ppr cls) <+> text "is not a stock derivable class (Eq, Show, etc.)" gndNonNewtypeErr :: SDoc gndNonNewtypeErr = text "GeneralizedNewtypeDeriving cannot be used on non-newtypes" derivingNullaryErr :: MsgDoc derivingNullaryErr = text "Cannot derive instances for nullary classes" derivingKindErr :: TyCon -> Class -> [Type] -> Kind -> Bool -> MsgDoc derivingKindErr tc cls cls_tys cls_kind enough_args = sep [ hang (text "Cannot derive well-kinded instance of form" <+> quotes (pprClassPred cls cls_tys <+> parens (ppr tc <+> text "..."))) 2 gen1_suggestion , nest 2 (text "Class" <+> quotes (ppr cls) <+> text "expects an argument of kind" <+> quotes (pprKind cls_kind)) ] where gen1_suggestion | cls `hasKey` gen1ClassKey && enough_args = text "(Perhaps you intended to use PolyKinds)" | otherwise = Outputable.empty derivingEtaErr :: Class -> [Type] -> Type -> MsgDoc derivingEtaErr cls cls_tys inst_ty = sep [text "Cannot eta-reduce to an instance of form", nest 2 (text "instance (...) =>" <+> pprClassPred cls (cls_tys ++ [inst_ty]))] derivingThingErr :: Bool -> Class -> [Type] -> Type -> Maybe DerivStrategy -> MsgDoc -> MsgDoc derivingThingErr newtype_deriving clas tys ty deriv_strat why = derivingThingErr' newtype_deriving clas tys ty (isJust deriv_strat) (maybe empty ppr deriv_strat) why derivingThingErrMechanism :: Class -> [Type] -> Type -> Bool -- True if an explicit deriving strategy -- keyword was provided -> DerivSpecMechanism -> MsgDoc -> MsgDoc derivingThingErrMechanism clas tys ty strat_used mechanism why = derivingThingErr' (isDerivSpecNewtype mechanism) clas tys ty strat_used (ppr mechanism) why derivingThingErr' :: Bool -> Class -> [Type] -> Type -> Bool -> MsgDoc -> MsgDoc -> MsgDoc derivingThingErr' newtype_deriving clas tys ty strat_used strat_msg why = sep [(hang (text "Can't make a derived instance of") 2 (quotes (ppr pred) <+> via_mechanism) $$ nest 2 extra) <> colon, nest 2 why] where extra | not strat_used, newtype_deriving = text "(even with cunning GeneralizedNewtypeDeriving)" | otherwise = empty pred = mkClassPred clas (tys ++ [ty]) via_mechanism | strat_used = text "with the" <+> strat_msg <+> text "strategy" | otherwise = empty derivingHiddenErr :: TyCon -> SDoc derivingHiddenErr tc = hang (text "The data constructors of" <+> quotes (ppr tc) <+> ptext (sLit "are not all in scope")) 2 (text "so you cannot derive an instance for it") standaloneCtxt :: LHsSigType Name -> SDoc standaloneCtxt ty = hang (text "In the stand-alone deriving instance for") 2 (quotes (ppr ty)) unboxedTyConErr :: String -> MsgDoc unboxedTyConErr thing = text "The last argument of the instance cannot be an unboxed" <+> text thing