{- (c) The University of Glasgow 2006 (c) The AQUA Project, Glasgow University, 1993-1998 TcRules: Typechecking transformation rules -} {-# LANGUAGE ViewPatterns #-} module TcRules ( tcRules ) where import HsSyn import TcRnMonad import TcSimplify import TcMType import TcType import TcHsType import TcExpr import TcEnv import TcEvidence import TcUnify( buildImplicationFor ) import Type import Id import Var ( EvVar ) import Name import BasicTypes ( RuleName ) import SrcLoc import Outputable import FastString import Bag import Data.List( partition ) {- Note [Typechecking rules] ~~~~~~~~~~~~~~~~~~~~~~~~~ We *infer* the typ of the LHS, and use that type to *check* the type of the RHS. That means that higher-rank rules work reasonably well. Here's an example (test simplCore/should_compile/rule2.hs) produced by Roman: foo :: (forall m. m a -> m b) -> m a -> m b foo f = ... bar :: (forall m. m a -> m a) -> m a -> m a bar f = ... {-# RULES "foo/bar" foo = bar #-} He wanted the rule to typecheck. -} tcRules :: [LRuleDecls Name] -> TcM [LRuleDecls TcId] tcRules decls = mapM (wrapLocM tcRuleDecls) decls tcRuleDecls :: RuleDecls Name -> TcM (RuleDecls TcId) tcRuleDecls (HsRules src decls) = do { tc_decls <- mapM (wrapLocM tcRule) decls ; return (HsRules src tc_decls) } tcRule :: RuleDecl Name -> TcM (RuleDecl TcId) tcRule (HsRule name act hs_bndrs lhs fv_lhs rhs fv_rhs) = addErrCtxt (ruleCtxt $ snd $ unLoc name) $ do { traceTc "---- Rule ------" (pprFullRuleName name) -- Note [Typechecking rules] ; (vars, bndr_wanted) <- captureConstraints $ tcRuleBndrs hs_bndrs -- bndr_wanted constraints can include wildcard hole -- constraints, which we should not forget about. -- It may mention the skolem type variables bound by -- the RULE. c.f. Trac #10072 ; let (id_bndrs, tv_bndrs) = partition isId vars ; (lhs', lhs_wanted, rhs', rhs_wanted, rule_ty) <- tcExtendTyVarEnv tv_bndrs $ tcExtendIdEnv id_bndrs $ do { -- See Note [Solve order for RULES] ((lhs', rule_ty), lhs_wanted) <- captureConstraints (tcInferRho lhs) ; (rhs', rhs_wanted) <- captureConstraints $ tcMonoExpr rhs (mkCheckExpType rule_ty) ; return (lhs', lhs_wanted, rhs', rhs_wanted, rule_ty) } ; traceTc "tcRule 1" (vcat [ pprFullRuleName name , ppr lhs_wanted , ppr rhs_wanted ]) ; let all_lhs_wanted = bndr_wanted `andWC` lhs_wanted ; lhs_evs <- simplifyRule (snd $ unLoc name) all_lhs_wanted rhs_wanted -- Now figure out what to quantify over -- c.f. TcSimplify.simplifyInfer -- We quantify over any tyvars free in *either* the rule -- *or* the bound variables. The latter is important. Consider -- ss (x,(y,z)) = (x,z) -- RULE: forall v. fst (ss v) = fst v -- The type of the rhs of the rule is just a, but v::(a,(b,c)) -- -- We also need to get the completely-uconstrained tyvars of -- the LHS, lest they otherwise get defaulted to Any; but we do that -- during zonking (see TcHsSyn.zonkRule) ; let tpl_ids = lhs_evs ++ id_bndrs ; forall_tkvs <- zonkTcTypesAndSplitDepVars $ rule_ty : map idType tpl_ids ; gbls <- tcGetGlobalTyCoVars -- Even though top level, there might be top-level -- monomorphic bindings from the MR; test tc111 ; qtkvs <- quantifyZonkedTyVars gbls forall_tkvs ; traceTc "tcRule" (vcat [ pprFullRuleName name , ppr forall_tkvs , ppr qtkvs , ppr rule_ty , vcat [ ppr id <+> dcolon <+> ppr (idType id) | id <- tpl_ids ] ]) -- Simplify the RHS constraints ; let skol_info = RuleSkol (snd $ unLoc name) ; (rhs_implic, rhs_binds) <- buildImplicationFor topTcLevel skol_info qtkvs lhs_evs rhs_wanted -- For the LHS constraints we must solve the remaining constraints -- (a) so that we report insoluble ones -- (b) so that we bind any soluble ones ; (lhs_implic, lhs_binds) <- buildImplicationFor topTcLevel skol_info qtkvs lhs_evs (all_lhs_wanted { wc_simple = emptyBag }) -- simplifyRule consumed all simple -- constraints ; emitImplications (lhs_implic `unionBags` rhs_implic) ; return (HsRule name act (map (noLoc . RuleBndr . noLoc) (qtkvs ++ tpl_ids)) (mkHsDictLet lhs_binds lhs') fv_lhs (mkHsDictLet rhs_binds rhs') fv_rhs) } tcRuleBndrs :: [LRuleBndr Name] -> TcM [Var] tcRuleBndrs [] = return [] tcRuleBndrs (L _ (RuleBndr (L _ name)) : rule_bndrs) = do { ty <- newOpenFlexiTyVarTy ; vars <- tcRuleBndrs rule_bndrs ; return (mkLocalId name ty : vars) } tcRuleBndrs (L _ (RuleBndrSig (L _ name) rn_ty) : rule_bndrs) -- e.g x :: a->a -- The tyvar 'a' is brought into scope first, just as if you'd written -- a::*, x :: a->a = do { let ctxt = RuleSigCtxt name ; (id_ty, tvs, _) <- tcHsPatSigType ctxt rn_ty ; let id = mkLocalIdOrCoVar name id_ty -- See Note [Pattern signature binders] in TcHsType -- The type variables scope over subsequent bindings; yuk ; vars <- tcExtendTyVarEnv tvs $ tcRuleBndrs rule_bndrs ; return (tvs ++ id : vars) } ruleCtxt :: FastString -> SDoc ruleCtxt name = text "When checking the transformation rule" <+> doubleQuotes (ftext name) {- ********************************************************************************* * * Constraint simplification for rules * * *********************************************************************************** Note [Simplifying RULE constraints] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Example. Consider the following left-hand side of a rule f (x == y) (y > z) = ... If we typecheck this expression we get constraints d1 :: Ord a, d2 :: Eq a We do NOT want to "simplify" to the LHS forall x::a, y::a, z::a, d1::Ord a. f ((==) (eqFromOrd d1) x y) ((>) d1 y z) = ... Instead we want forall x::a, y::a, z::a, d1::Ord a, d2::Eq a. f ((==) d2 x y) ((>) d1 y z) = ... Here is another example: fromIntegral :: (Integral a, Num b) => a -> b {-# RULES "foo" fromIntegral = id :: Int -> Int #-} In the rule, a=b=Int, and Num Int is a superclass of Integral Int. But we *dont* want to get forall dIntegralInt. fromIntegral Int Int dIntegralInt (scsel dIntegralInt) = id Int because the scsel will mess up RULE matching. Instead we want forall dIntegralInt, dNumInt. fromIntegral Int Int dIntegralInt dNumInt = id Int Even if we have g (x == y) (y == z) = .. where the two dictionaries are *identical*, we do NOT WANT forall x::a, y::a, z::a, d1::Eq a f ((==) d1 x y) ((>) d1 y z) = ... because that will only match if the dict args are (visibly) equal. Instead we want to quantify over the dictionaries separately. In short, simplifyRuleLhs must *only* squash equalities, leaving all dicts unchanged, with absolutely no sharing. Also note that we can't solve the LHS constraints in isolation: Example foo :: Ord a => a -> a foo_spec :: Int -> Int {-# RULE "foo" foo = foo_spec #-} Here, it's the RHS that fixes the type variable HOWEVER, under a nested implication things are different Consider f :: (forall a. Eq a => a->a) -> Bool -> ... {-# RULES "foo" forall (v::forall b. Eq b => b->b). f b True = ... #-} Here we *must* solve the wanted (Eq a) from the given (Eq a) resulting from skolemising the agument type of g. So we revert to SimplCheck when going under an implication. ------------------------ So the plan is this ----------------------- * Step 0: typecheck the LHS and RHS to get constraints from each * Step 1: Simplify the LHS and RHS constraints all together in one bag We do this to discover all unification equalities * Step 2: Zonk the ORIGINAL (unsimplified) lhs constraints, to take advantage of those unifications, and partition them into the ones we will quantify over, and the others See Note [RULE quantification over equalities] * Step 3: Decide on the type variables to quantify over * Step 4: Simplify the LHS and RHS constraints separately, using the quantified constraints as givens Note [Solve order for RULES] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~ In step 1 above, we need to be a bit careful about solve order. Consider f :: Int -> T Int type instance T Int = Bool RULE f 3 = True From the RULE we get lhs-constraints: T Int ~ alpha rhs-constraints: Bool ~ alpha where 'alpha' is the type that connects the two. If we glom them all together, and solve the RHS constraint first, we might solve with alpha := Bool. But then we'd end up with a RULE like RULE: f 3 |> (co :: T Int ~ Booo) = True which is terrible. We want RULE: f 3 = True |> (sym co :: Bool ~ T Int) So we are careful to solve the LHS constraints first, and *then* the RHS constraints. Actually much of this is done by the on-the-fly constraint solving, so the same order must be observed in tcRule. Note [RULE quantification over equalities] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Deciding which equalities to quantify over is tricky: * We do not want to quantify over insoluble equalities (Int ~ Bool) (a) because we prefer to report a LHS type error (b) because if such things end up in 'givens' we get a bogus "inaccessible code" error * But we do want to quantify over things like (a ~ F b), where F is a type function. The difficulty is that it's hard to tell what is insoluble! So we see whether the simplification step yielded any type errors, and if so refrain from quantifying over *any* equalities. Note [Quantifying over coercion holes] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Equality constraints from the LHS will emit coercion hole Wanteds. These don't have a name, so we can't quantify over them directly. Instead, because we really do want to quantify here, invent a new EvVar for the coercion, fill the hole with the invented EvVar, and then quantify over the EvVar. Not too tricky -- just some impedence matching, really. Note [Simplify *derived* constraints] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ At this stage, we're simplifying constraints only for insolubility and for unification. Note that all the evidence is quickly discarded. We make this explicit by working over derived constraints, for which there is no evidence. Using derived constraints also prevents solved equalities from being written to coercion holes. If we don't do this, then RHS coercion-hole constraints get filled in, only to get filled in *again* when solving the implications emitted from tcRule. That's terrible, so we avoid the problem by using derived constraints. -} simplifyRule :: RuleName -> WantedConstraints -- Constraints from LHS -> WantedConstraints -- Constraints from RHS -> TcM [EvVar] -- LHS evidence variables, -- See Note [Simplifying RULE constraints] in TcRule -- NB: This consumes all simple constraints on the LHS, but not -- any LHS implication constraints. simplifyRule name lhs_wanted rhs_wanted = do { -- We allow ourselves to unify environment -- variables: runTcS runs with topTcLevel ; tc_lvl <- getTcLevel ; insoluble <- runTcSDeriveds $ do { -- First solve the LHS and *then* solve the RHS -- See Note [Solve order for RULES] -- See Note [Simplify *derived* constraints] lhs_resid <- solveWanteds $ toDerivedWC lhs_wanted ; rhs_resid <- solveWanteds $ toDerivedWC rhs_wanted ; return ( insolubleWC tc_lvl lhs_resid || insolubleWC tc_lvl rhs_resid ) } ; zonked_lhs_simples <- zonkSimples (wc_simple lhs_wanted) ; ev_ids <- mapMaybeM (quantify_ct insoluble) $ bagToList zonked_lhs_simples ; traceTc "simplifyRule" $ vcat [ text "LHS of rule" <+> doubleQuotes (ftext name) , text "lhs_wantd" <+> ppr lhs_wanted , text "rhs_wantd" <+> ppr rhs_wanted , text "zonked_lhs_simples" <+> ppr zonked_lhs_simples , text "ev_ids" <+> ppr ev_ids ] ; return ev_ids } where quantify_ct insol -- Note [RULE quantification over equalities] | insol = quantify_insol | otherwise = quantify_normal quantify_insol ct | isEqPred (ctPred ct) = return Nothing | otherwise = return $ Just $ ctEvId $ ctEvidence ct quantify_normal (ctEvidence -> CtWanted { ctev_dest = dest , ctev_pred = pred }) = case dest of -- See Note [Quantifying over coercion holes] HoleDest hole | EqPred NomEq t1 t2 <- classifyPredType pred , t1 `tcEqType` t2 -> do { -- These are trivial. Don't quantify. But do fill in -- the hole. ; fillCoercionHole hole (mkTcNomReflCo t1) ; return Nothing } | otherwise -> do { ev_id <- newEvVar pred ; fillCoercionHole hole (mkTcCoVarCo ev_id) ; return (Just ev_id) } EvVarDest evar -> return (Just evar) quantify_normal ct = pprPanic "simplifyRule.quantify_normal" (ppr ct)