2 % (c) The University of Glasgow 2006
3 % (c) The GRASP/AQUA Project, Glasgow University, 1992-1998
10 tcSimplifyInfer, tcSimplifyInferCheck,
11 tcSimplifyCheck, tcSimplifyRestricted,
12 tcSimplifyRuleLhs, tcSimplifyIPs,
13 tcSimplifySuperClasses,
14 tcSimplifyTop, tcSimplifyInteractive,
15 tcSimplifyBracket, tcSimplifyCheckPat,
17 tcSimplifyDeriv, tcSimplifyDefault,
25 #include "HsVersions.h"
27 import {-# SOURCE #-} TcUnify( unifyType )
31 import TcHsSyn ( hsLPatType )
39 import DsUtils -- Big-tuple functions
69 %************************************************************************
73 %************************************************************************
75 --------------------------------------
76 Notes on functional dependencies (a bug)
77 --------------------------------------
84 instance D a b => C a b -- Undecidable
85 -- (Not sure if it's crucial to this eg)
86 f :: C a b => a -> Bool
89 g :: C a b => a -> Bool
92 Here f typechecks, but g does not!! Reason: before doing improvement,
93 we reduce the (C a b1) constraint from the call of f to (D a b1).
95 Here is a more complicated example:
98 > class Foo a b | a->b
100 > class Bar a b | a->b
104 > instance Bar Obj Obj
106 > instance (Bar a b) => Foo a b
108 > foo:: (Foo a b) => a -> String
111 > runFoo:: (forall a b. (Foo a b) => a -> w) -> w
117 Could not deduce (Bar a b) from the context (Foo a b)
118 arising from use of `foo' at <interactive>:1
120 Add (Bar a b) to the expected type of an expression
121 In the first argument of `runFoo', namely `foo'
122 In the definition of `it': it = runFoo foo
124 Why all of the sudden does GHC need the constraint Bar a b? The
125 function foo didn't ask for that...
128 The trouble is that to type (runFoo foo), GHC has to solve the problem:
130 Given constraint Foo a b
131 Solve constraint Foo a b'
133 Notice that b and b' aren't the same. To solve this, just do
134 improvement and then they are the same. But GHC currently does
139 That is usually fine, but it isn't here, because it sees that Foo a b is
140 not the same as Foo a b', and so instead applies the instance decl for
141 instance Bar a b => Foo a b. And that's where the Bar constraint comes
144 The Right Thing is to improve whenever the constraint set changes at
145 all. Not hard in principle, but it'll take a bit of fiddling to do.
147 Note [Choosing which variables to quantify]
148 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
149 Suppose we are about to do a generalisation step. We have in our hand
152 T the type of the RHS
153 C the constraints from that RHS
155 The game is to figure out
157 Q the set of type variables over which to quantify
158 Ct the constraints we will *not* quantify over
159 Cq the constraints we will quantify over
161 So we're going to infer the type
165 and float the constraints Ct further outwards.
167 Here are the things that *must* be true:
169 (A) Q intersect fv(G) = EMPTY limits how big Q can be
170 (B) Q superset fv(Cq union T) \ oclose(fv(G),C) limits how small Q can be
172 (A) says we can't quantify over a variable that's free in the environment.
173 (B) says we must quantify over all the truly free variables in T, else
174 we won't get a sufficiently general type.
176 We do not *need* to quantify over any variable that is fixed by the
177 free vars of the environment G.
179 BETWEEN THESE TWO BOUNDS, ANY Q WILL DO!
181 Example: class H x y | x->y where ...
183 fv(G) = {a} C = {H a b, H c d}
186 (A) Q intersect {a} is empty
187 (B) Q superset {a,b,c,d} \ oclose({a}, C) = {a,b,c,d} \ {a,b} = {c,d}
189 So Q can be {c,d}, {b,c,d}
191 In particular, it's perfectly OK to quantify over more type variables
192 than strictly necessary; there is no need to quantify over 'b', since
193 it is determined by 'a' which is free in the envt, but it's perfectly
194 OK to do so. However we must not quantify over 'a' itself.
196 Other things being equal, however, we'd like to quantify over as few
197 variables as possible: smaller types, fewer type applications, more
198 constraints can get into Ct instead of Cq. Here's a good way to
201 Q = grow( fv(T), C ) \ oclose( fv(G), C )
203 That is, quantify over all variable that that MIGHT be fixed by the
204 call site (which influences T), but which aren't DEFINITELY fixed by
205 G. This choice definitely quantifies over enough type variables,
206 albeit perhaps too many.
208 Why grow( fv(T), C ) rather than fv(T)? Consider
210 class H x y | x->y where ...
215 If we used fv(T) = {c} we'd get the type
217 forall c. H c d => c -> b
219 And then if the fn was called at several different c's, each of
220 which fixed d differently, we'd get a unification error, because
221 d isn't quantified. Solution: quantify d. So we must quantify
222 everything that might be influenced by c.
224 Why not oclose( fv(T), C )? Because we might not be able to see
225 all the functional dependencies yet:
227 class H x y | x->y where ...
228 instance H x y => Eq (T x y) where ...
233 Now oclose(fv(T),C) = {c}, because the functional dependency isn't
234 apparent yet, and that's wrong. We must really quantify over d too.
236 There really isn't any point in quantifying over any more than
237 grow( fv(T), C ), because the call sites can't possibly influence
238 any other type variables.
242 -------------------------------------
244 -------------------------------------
246 It's very hard to be certain when a type is ambiguous. Consider
250 instance H x y => K (x,y)
252 Is this type ambiguous?
253 forall a b. (K (a,b), Eq b) => a -> a
255 Looks like it! But if we simplify (K (a,b)) we get (H a b) and
256 now we see that a fixes b. So we can't tell about ambiguity for sure
257 without doing a full simplification. And even that isn't possible if
258 the context has some free vars that may get unified. Urgle!
260 Here's another example: is this ambiguous?
261 forall a b. Eq (T b) => a -> a
262 Not if there's an insance decl (with no context)
263 instance Eq (T b) where ...
265 You may say of this example that we should use the instance decl right
266 away, but you can't always do that:
268 class J a b where ...
269 instance J Int b where ...
271 f :: forall a b. J a b => a -> a
273 (Notice: no functional dependency in J's class decl.)
274 Here f's type is perfectly fine, provided f is only called at Int.
275 It's premature to complain when meeting f's signature, or even
276 when inferring a type for f.
280 However, we don't *need* to report ambiguity right away. It'll always
281 show up at the call site.... and eventually at main, which needs special
282 treatment. Nevertheless, reporting ambiguity promptly is an excellent thing.
284 So here's the plan. We WARN about probable ambiguity if
286 fv(Cq) is not a subset of oclose(fv(T) union fv(G), C)
288 (all tested before quantification).
289 That is, all the type variables in Cq must be fixed by the the variables
290 in the environment, or by the variables in the type.
292 Notice that we union before calling oclose. Here's an example:
294 class J a b c | a b -> c
298 forall b c. (J a b c) => b -> b
300 Only if we union {a} from G with {b} from T before using oclose,
301 do we see that c is fixed.
303 It's a bit vague exactly which C we should use for this oclose call. If we
304 don't fix enough variables we might complain when we shouldn't (see
305 the above nasty example). Nothing will be perfect. That's why we can
306 only issue a warning.
309 Can we ever be *certain* about ambiguity? Yes: if there's a constraint
311 c in C such that fv(c) intersect (fv(G) union fv(T)) = EMPTY
313 then c is a "bubble"; there's no way it can ever improve, and it's
314 certainly ambiguous. UNLESS it is a constant (sigh). And what about
319 instance H x y => K (x,y)
321 Is this type ambiguous?
322 forall a b. (K (a,b), Eq b) => a -> a
324 Urk. The (Eq b) looks "definitely ambiguous" but it isn't. What we are after
325 is a "bubble" that's a set of constraints
327 Cq = Ca union Cq' st fv(Ca) intersect (fv(Cq') union fv(T) union fv(G)) = EMPTY
329 Hence another idea. To decide Q start with fv(T) and grow it
330 by transitive closure in Cq (no functional dependencies involved).
331 Now partition Cq using Q, leaving the definitely-ambiguous and probably-ok.
332 The definitely-ambiguous can then float out, and get smashed at top level
333 (which squashes out the constants, like Eq (T a) above)
336 --------------------------------------
337 Notes on principal types
338 --------------------------------------
343 f x = let g y = op (y::Int) in True
345 Here the principal type of f is (forall a. a->a)
346 but we'll produce the non-principal type
347 f :: forall a. C Int => a -> a
350 --------------------------------------
351 The need for forall's in constraints
352 --------------------------------------
354 [Exchange on Haskell Cafe 5/6 Dec 2000]
356 class C t where op :: t -> Bool
357 instance C [t] where op x = True
359 p y = (let f :: c -> Bool; f x = op (y >> return x) in f, y ++ [])
360 q y = (y ++ [], let f :: c -> Bool; f x = op (y >> return x) in f)
362 The definitions of p and q differ only in the order of the components in
363 the pair on their right-hand sides. And yet:
365 ghc and "Typing Haskell in Haskell" reject p, but accept q;
366 Hugs rejects q, but accepts p;
367 hbc rejects both p and q;
368 nhc98 ... (Malcolm, can you fill in the blank for us!).
370 The type signature for f forces context reduction to take place, and
371 the results of this depend on whether or not the type of y is known,
372 which in turn depends on which component of the pair the type checker
375 Solution: if y::m a, float out the constraints
376 Monad m, forall c. C (m c)
377 When m is later unified with [], we can solve both constraints.
380 --------------------------------------
381 Notes on implicit parameters
382 --------------------------------------
384 Note [Inheriting implicit parameters]
385 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
390 where f is *not* a top-level binding.
391 From the RHS of f we'll get the constraint (?y::Int).
392 There are two types we might infer for f:
396 (so we get ?y from the context of f's definition), or
398 f :: (?y::Int) => Int -> Int
400 At first you might think the first was better, becuase then
401 ?y behaves like a free variable of the definition, rather than
402 having to be passed at each call site. But of course, the WHOLE
403 IDEA is that ?y should be passed at each call site (that's what
404 dynamic binding means) so we'd better infer the second.
406 BOTTOM LINE: when *inferring types* you *must* quantify
407 over implicit parameters. See the predicate isFreeWhenInferring.
410 Note [Implicit parameters and ambiguity]
411 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
412 Only a *class* predicate can give rise to ambiguity
413 An *implicit parameter* cannot. For example:
414 foo :: (?x :: [a]) => Int
416 is fine. The call site will suppply a particular 'x'
418 Furthermore, the type variables fixed by an implicit parameter
419 propagate to the others. E.g.
420 foo :: (Show a, ?x::[a]) => Int
422 The type of foo looks ambiguous. But it isn't, because at a call site
424 let ?x = 5::Int in foo
425 and all is well. In effect, implicit parameters are, well, parameters,
426 so we can take their type variables into account as part of the
427 "tau-tvs" stuff. This is done in the function 'FunDeps.grow'.
430 Question 2: type signatures
431 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
432 BUT WATCH OUT: When you supply a type signature, we can't force you
433 to quantify over implicit parameters. For example:
437 This is perfectly reasonable. We do not want to insist on
439 (?x + 1) :: (?x::Int => Int)
441 That would be silly. Here, the definition site *is* the occurrence site,
442 so the above strictures don't apply. Hence the difference between
443 tcSimplifyCheck (which *does* allow implicit paramters to be inherited)
444 and tcSimplifyCheckBind (which does not).
446 What about when you supply a type signature for a binding?
447 Is it legal to give the following explicit, user type
448 signature to f, thus:
453 At first sight this seems reasonable, but it has the nasty property
454 that adding a type signature changes the dynamic semantics.
457 (let f x = (x::Int) + ?y
458 in (f 3, f 3 with ?y=5)) with ?y = 6
464 in (f 3, f 3 with ?y=5)) with ?y = 6
468 Indeed, simply inlining f (at the Haskell source level) would change the
471 Nevertheless, as Launchbury says (email Oct 01) we can't really give the
472 semantics for a Haskell program without knowing its typing, so if you
473 change the typing you may change the semantics.
475 To make things consistent in all cases where we are *checking* against
476 a supplied signature (as opposed to inferring a type), we adopt the
479 a signature does not need to quantify over implicit params.
481 [This represents a (rather marginal) change of policy since GHC 5.02,
482 which *required* an explicit signature to quantify over all implicit
483 params for the reasons mentioned above.]
485 But that raises a new question. Consider
487 Given (signature) ?x::Int
488 Wanted (inferred) ?x::Int, ?y::Bool
490 Clearly we want to discharge the ?x and float the ?y out. But
491 what is the criterion that distinguishes them? Clearly it isn't
492 what free type variables they have. The Right Thing seems to be
493 to float a constraint that
494 neither mentions any of the quantified type variables
495 nor any of the quantified implicit parameters
497 See the predicate isFreeWhenChecking.
500 Question 3: monomorphism
501 ~~~~~~~~~~~~~~~~~~~~~~~~
502 There's a nasty corner case when the monomorphism restriction bites:
506 The argument above suggests that we *must* generalise
507 over the ?y parameter, to get
508 z :: (?y::Int) => Int,
509 but the monomorphism restriction says that we *must not*, giving
511 Why does the momomorphism restriction say this? Because if you have
513 let z = x + ?y in z+z
515 you might not expect the addition to be done twice --- but it will if
516 we follow the argument of Question 2 and generalise over ?y.
519 Question 4: top level
520 ~~~~~~~~~~~~~~~~~~~~~
521 At the top level, monomorhism makes no sense at all.
524 main = let ?x = 5 in print foo
528 woggle :: (?x :: Int) => Int -> Int
531 We definitely don't want (foo :: Int) with a top-level implicit parameter
532 (?x::Int) becuase there is no way to bind it.
537 (A) Always generalise over implicit parameters
538 Bindings that fall under the monomorphism restriction can't
542 * Inlining remains valid
543 * No unexpected loss of sharing
544 * But simple bindings like
546 will be rejected, unless you add an explicit type signature
547 (to avoid the monomorphism restriction)
548 z :: (?y::Int) => Int
550 This seems unacceptable
552 (B) Monomorphism restriction "wins"
553 Bindings that fall under the monomorphism restriction can't
555 Always generalise over implicit parameters *except* for bindings
556 that fall under the monomorphism restriction
559 * Inlining isn't valid in general
560 * No unexpected loss of sharing
561 * Simple bindings like
563 accepted (get value of ?y from binding site)
565 (C) Always generalise over implicit parameters
566 Bindings that fall under the monomorphism restriction can't
567 be generalised, EXCEPT for implicit parameters
569 * Inlining remains valid
570 * Unexpected loss of sharing (from the extra generalisation)
571 * Simple bindings like
573 accepted (get value of ?y from occurrence sites)
578 None of these choices seems very satisfactory. But at least we should
579 decide which we want to do.
581 It's really not clear what is the Right Thing To Do. If you see
585 would you expect the value of ?y to be got from the *occurrence sites*
586 of 'z', or from the valuue of ?y at the *definition* of 'z'? In the
587 case of function definitions, the answer is clearly the former, but
588 less so in the case of non-fucntion definitions. On the other hand,
589 if we say that we get the value of ?y from the definition site of 'z',
590 then inlining 'z' might change the semantics of the program.
592 Choice (C) really says "the monomorphism restriction doesn't apply
593 to implicit parameters". Which is fine, but remember that every
594 innocent binding 'x = ...' that mentions an implicit parameter in
595 the RHS becomes a *function* of that parameter, called at each
596 use of 'x'. Now, the chances are that there are no intervening 'with'
597 clauses that bind ?y, so a decent compiler should common up all
598 those function calls. So I think I strongly favour (C). Indeed,
599 one could make a similar argument for abolishing the monomorphism
600 restriction altogether.
602 BOTTOM LINE: we choose (B) at present. See tcSimplifyRestricted
606 %************************************************************************
608 \subsection{tcSimplifyInfer}
610 %************************************************************************
612 tcSimplify is called when we *inferring* a type. Here's the overall game plan:
614 1. Compute Q = grow( fvs(T), C )
616 2. Partition C based on Q into Ct and Cq. Notice that ambiguous
617 predicates will end up in Ct; we deal with them at the top level
619 3. Try improvement, using functional dependencies
621 4. If Step 3 did any unification, repeat from step 1
622 (Unification can change the result of 'grow'.)
624 Note: we don't reduce dictionaries in step 2. For example, if we have
625 Eq (a,b), we don't simplify to (Eq a, Eq b). So Q won't be different
626 after step 2. However note that we may therefore quantify over more
627 type variables than we absolutely have to.
629 For the guts, we need a loop, that alternates context reduction and
630 improvement with unification. E.g. Suppose we have
632 class C x y | x->y where ...
634 and tcSimplify is called with:
636 Then improvement unifies a with b, giving
639 If we need to unify anything, we rattle round the whole thing all over
646 -> TcTyVarSet -- fv(T); type vars
648 -> TcM ([TcTyVar], -- Tyvars to quantify (zonked and quantified)
649 [Inst], -- Dict Ids that must be bound here (zonked)
650 TcDictBinds) -- Bindings
651 -- Any free (escaping) Insts are tossed into the environment
656 tcSimplifyInfer doc tau_tvs wanted
657 = do { tau_tvs1 <- zonkTcTyVarsAndFV (varSetElems tau_tvs)
658 ; wanted' <- mapM zonkInst wanted -- Zonk before deciding quantified tyvars
659 ; gbl_tvs <- tcGetGlobalTyVars
660 ; let preds1 = fdPredsOfInsts wanted'
661 gbl_tvs1 = oclose preds1 gbl_tvs
662 qtvs = grow preds1 tau_tvs1 `minusVarSet` gbl_tvs1
663 -- See Note [Choosing which variables to quantify]
665 -- To maximise sharing, remove from consideration any
666 -- constraints that don't mention qtvs at all
667 ; let (free, bound) = partition (isFreeWhenInferring qtvs) wanted'
670 -- To make types simple, reduce as much as possible
671 ; traceTc (text "infer" <+> (ppr preds1 $$ ppr (grow preds1 tau_tvs1) $$ ppr gbl_tvs $$
672 ppr gbl_tvs1 $$ ppr free $$ ppr bound))
673 ; (irreds1, binds1) <- tryHardCheckLoop doc bound
675 -- Note [Inference and implication constraints]
676 ; let want_dict d = tyVarsOfInst d `intersectsVarSet` qtvs
677 ; (irreds2, binds2) <- approximateImplications doc want_dict irreds1
679 -- Now work out all over again which type variables to quantify,
680 -- exactly in the same way as before, but starting from irreds2. Why?
681 -- a) By now improvment may have taken place, and we must *not*
682 -- quantify over any variable free in the environment
683 -- tc137 (function h inside g) is an example
685 -- b) Do not quantify over constraints that *now* do not
686 -- mention quantified type variables, because they are
687 -- simply ambiguous (or might be bound further out). Example:
688 -- f :: Eq b => a -> (a, b)
690 -- From the RHS of g we get the MethodInst f77 :: alpha -> (alpha, beta)
691 -- We decide to quantify over 'alpha' alone, but free1 does not include f77
692 -- because f77 mentions 'alpha'. Then reducing leaves only the (ambiguous)
693 -- constraint (Eq beta), which we dump back into the free set
694 -- See test tcfail181
696 -- c) irreds may contain type variables not previously mentioned,
697 -- e.g. instance D a x => Foo [a]
699 -- Then after simplifying we'll get (D a x), and x is fresh
700 -- We must quantify over x else it'll be totally unbound
701 ; tau_tvs2 <- zonkTcTyVarsAndFV (varSetElems tau_tvs1)
702 ; gbl_tvs2 <- zonkTcTyVarsAndFV (varSetElems gbl_tvs1)
703 -- Note that we start from gbl_tvs1
704 -- We use tcGetGlobalTyVars, then oclose wrt preds2, because
705 -- we've already put some of the original preds1 into frees
706 -- E.g. wanteds = C a b (where a->b)
709 -- Then b is fixed by gbl_tvs, so (C a b) will be in free, and
710 -- irreds2 will be empty. But we don't want to generalise over b!
711 ; let preds2 = fdPredsOfInsts irreds2 -- irreds2 is zonked
712 qtvs = grow preds2 tau_tvs2 `minusVarSet` oclose preds2 gbl_tvs2
713 ; let (free, irreds3) = partition (isFreeWhenInferring qtvs) irreds2
716 -- Turn the quantified meta-type variables into real type variables
717 ; qtvs2 <- zonkQuantifiedTyVars (varSetElems qtvs)
719 -- We can't abstract over any remaining unsolved
720 -- implications so instead just float them outwards. Ugh.
721 ; let (q_dicts0, implics) = partition isAbstractableInst irreds3
722 ; loc <- getInstLoc (ImplicOrigin doc)
723 ; implic_bind <- bindIrreds loc qtvs2 q_dicts0 implics
725 -- Prepare equality instances for quantification
726 ; let (q_eqs0,q_dicts) = partition isEqInst q_dicts0
727 ; q_eqs <- mapM finalizeEqInst q_eqs0
729 ; return (qtvs2, q_eqs ++ q_dicts, binds1 `unionBags` binds2 `unionBags` implic_bind) }
730 -- NB: when we are done, we might have some bindings, but
731 -- the final qtvs might be empty. See Note [NO TYVARS] below.
733 approximateImplications :: SDoc -> (Inst -> Bool) -> [Inst] -> TcM ([Inst], TcDictBinds)
734 -- Note [Inference and implication constraints]
735 -- Given a bunch of Dict and ImplicInsts, try to approximate the implications by
736 -- - fetching any dicts inside them that are free
737 -- - using those dicts as cruder constraints, to solve the implications
738 -- - returning the extra ones too
740 approximateImplications doc want_dict irreds
742 = return (irreds, emptyBag)
744 = do { extra_dicts' <- mapM cloneDict extra_dicts
745 ; tryHardCheckLoop doc (extra_dicts' ++ irreds) }
746 -- By adding extra_dicts', we make them
747 -- available to solve the implication constraints
749 extra_dicts = get_dicts (filter isImplicInst irreds)
751 get_dicts :: [Inst] -> [Inst] -- Returns only Dicts
752 -- Find the wanted constraints in implication constraints that satisfy
753 -- want_dict, and are not bound by forall's in the constraint itself
754 get_dicts ds = concatMap get_dict ds
756 get_dict d@(Dict {}) | want_dict d = [d]
758 get_dict (ImplicInst {tci_tyvars = tvs, tci_wanted = wanteds})
759 = [ d | let tv_set = mkVarSet tvs
760 , d <- get_dicts wanteds
761 , not (tyVarsOfInst d `intersectsVarSet` tv_set)]
762 get_dict i@(EqInst {}) | want_dict i = [i]
764 get_dict other = pprPanic "approximateImplications" (ppr other)
767 Note [Inference and implication constraints]
768 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
769 Suppose we have a wanted implication constraint (perhaps arising from
770 a nested pattern match) like
772 and we are now trying to quantify over 'a' when inferring the type for
773 a function. In principle it's possible that there might be an instance
774 instance (C a, E a) => D [a]
775 so the context (E a) would suffice. The Right Thing is to abstract over
776 the implication constraint, but we don't do that (a) because it'll be
777 surprising to programmers and (b) because we don't have the machinery to deal
778 with 'given' implications.
780 So our best approximation is to make (D [a]) part of the inferred
781 context, so we can use that to discharge the implication. Hence
782 the strange function get_dicts in approximateImplications.
784 The common cases are more clear-cut, when we have things like
786 Here, abstracting over (C b) is not an approximation at all -- but see
787 Note [Freeness and implications].
789 See Trac #1430 and test tc228.
793 -----------------------------------------------------------
794 -- tcSimplifyInferCheck is used when we know the constraints we are to simplify
795 -- against, but we don't know the type variables over which we are going to quantify.
796 -- This happens when we have a type signature for a mutually recursive group
799 -> TcTyVarSet -- fv(T)
802 -> TcM ([TyVar], -- Fully zonked, and quantified
803 TcDictBinds) -- Bindings
805 tcSimplifyInferCheck loc tau_tvs givens wanteds
806 = do { traceTc (text "tcSimplifyInferCheck <-" <+> ppr wanteds)
807 ; (irreds, binds) <- gentleCheckLoop loc givens wanteds
809 -- Figure out which type variables to quantify over
810 -- You might think it should just be the signature tyvars,
811 -- but in bizarre cases you can get extra ones
812 -- f :: forall a. Num a => a -> a
813 -- f x = fst (g (x, head [])) + 1
815 -- Here we infer g :: forall a b. a -> b -> (b,a)
816 -- We don't want g to be monomorphic in b just because
817 -- f isn't quantified over b.
818 ; let all_tvs = varSetElems (tau_tvs `unionVarSet` tyVarsOfInsts givens)
819 ; all_tvs <- zonkTcTyVarsAndFV all_tvs
820 ; gbl_tvs <- tcGetGlobalTyVars
821 ; let qtvs = varSetElems (all_tvs `minusVarSet` gbl_tvs)
822 -- We could close gbl_tvs, but its not necessary for
823 -- soundness, and it'll only affect which tyvars, not which
824 -- dictionaries, we quantify over
826 ; qtvs' <- zonkQuantifiedTyVars qtvs
828 -- Now we are back to normal (c.f. tcSimplCheck)
829 ; implic_bind <- bindIrreds loc qtvs' givens irreds
831 ; traceTc (text "tcSimplifyInferCheck ->" <+> ppr (implic_bind))
832 ; return (qtvs', binds `unionBags` implic_bind) }
835 Note [Squashing methods]
836 ~~~~~~~~~~~~~~~~~~~~~~~~~
837 Be careful if you want to float methods more:
838 truncate :: forall a. RealFrac a => forall b. Integral b => a -> b
839 From an application (truncate f i) we get
842 If we have also have a second occurrence of truncate, we get
845 When simplifying with i,f free, we might still notice that
846 t1=t3; but alas, the binding for t2 (which mentions t1)
847 may continue to float out!
852 class Y a b | a -> b where
855 instance Y [[a]] a where
858 k :: X a -> X a -> X a
860 g :: Num a => [X a] -> [X a]
863 h ys = ys ++ map (k (y [[0]])) xs
865 The excitement comes when simplifying the bindings for h. Initially
866 try to simplify {y @ [[t1]] t2, 0 @ t1}, with initial qtvs = {t2}.
867 From this we get t1~t2, but also various bindings. We can't forget
868 the bindings (because of [LOOP]), but in fact t1 is what g is
871 The net effect of [NO TYVARS]
874 isFreeWhenInferring :: TyVarSet -> Inst -> Bool
875 isFreeWhenInferring qtvs inst
876 = isFreeWrtTyVars qtvs inst -- Constrains no quantified vars
877 && isInheritableInst inst -- and no implicit parameter involved
878 -- see Note [Inheriting implicit parameters]
880 {- No longer used (with implication constraints)
881 isFreeWhenChecking :: TyVarSet -- Quantified tyvars
882 -> NameSet -- Quantified implicit parameters
884 isFreeWhenChecking qtvs ips inst
885 = isFreeWrtTyVars qtvs inst
886 && isFreeWrtIPs ips inst
889 isFreeWrtTyVars :: VarSet -> Inst -> Bool
890 isFreeWrtTyVars qtvs inst = tyVarsOfInst inst `disjointVarSet` qtvs
891 isFreeWrtIPs :: NameSet -> Inst -> Bool
892 isFreeWrtIPs ips inst = not (any (`elemNameSet` ips) (ipNamesOfInst inst))
896 %************************************************************************
898 \subsection{tcSimplifyCheck}
900 %************************************************************************
902 @tcSimplifyCheck@ is used when we know exactly the set of variables
903 we are going to quantify over. For example, a class or instance declaration.
906 -----------------------------------------------------------
907 -- tcSimplifyCheck is used when checking expression type signatures,
908 -- class decls, instance decls etc.
909 tcSimplifyCheck :: InstLoc
910 -> [TcTyVar] -- Quantify over these
913 -> TcM TcDictBinds -- Bindings
914 tcSimplifyCheck loc qtvs givens wanteds
915 = ASSERT( all isTcTyVar qtvs && all isSkolemTyVar qtvs )
916 do { traceTc (text "tcSimplifyCheck")
917 ; (irreds, binds) <- gentleCheckLoop loc givens wanteds
918 ; implic_bind <- bindIrreds loc qtvs givens irreds
919 ; return (binds `unionBags` implic_bind) }
921 -----------------------------------------------------------
922 -- tcSimplifyCheckPat is used for existential pattern match
923 tcSimplifyCheckPat :: InstLoc
924 -> [TcTyVar] -- Quantify over these
927 -> TcM TcDictBinds -- Bindings
928 tcSimplifyCheckPat loc qtvs givens wanteds
929 = ASSERT( all isTcTyVar qtvs && all isSkolemTyVar qtvs )
930 do { traceTc (text "tcSimplifyCheckPat")
931 ; (irreds, binds) <- gentleCheckLoop loc givens wanteds
932 ; implic_bind <- bindIrredsR loc qtvs givens irreds
933 ; return (binds `unionBags` implic_bind) }
935 -----------------------------------------------------------
936 bindIrreds :: InstLoc -> [TcTyVar]
939 bindIrreds loc qtvs givens irreds
940 = bindIrredsR loc qtvs givens irreds
942 bindIrredsR :: InstLoc -> [TcTyVar] -> [Inst] -> [Inst] -> TcM TcDictBinds
943 -- Make a binding that binds 'irreds', by generating an implication
944 -- constraint for them, *and* throwing the constraint into the LIE
945 bindIrredsR loc qtvs givens irreds
949 = do { let givens' = filter isAbstractableInst givens
950 -- The givens can (redundantly) include methods
951 -- We want to retain both EqInsts and Dicts
952 -- There should be no implicadtion constraints
953 -- See Note [Pruning the givens in an implication constraint]
955 -- If there are no 'givens', then it's safe to
956 -- partition the 'wanteds' by their qtvs, thereby trimming irreds
957 -- See Note [Freeness and implications]
958 ; irreds' <- if null givens'
960 { let qtv_set = mkVarSet qtvs
961 (frees, real_irreds) = partition (isFreeWrtTyVars qtv_set) irreds
963 ; return real_irreds }
966 ; (implics, bind) <- makeImplicationBind loc qtvs givens' irreds'
967 -- This call does the real work
968 -- If irreds' is empty, it does something sensible
973 makeImplicationBind :: InstLoc -> [TcTyVar]
975 -> TcM ([Inst], TcDictBinds)
976 -- Make a binding that binds 'irreds', by generating an implication
977 -- constraint for them.
979 -- The binding looks like
980 -- (ir1, .., irn) = f qtvs givens
981 -- where f is (evidence for) the new implication constraint
982 -- f :: forall qtvs. givens => (ir1, .., irn)
983 -- qtvs includes coercion variables
985 -- This binding must line up the 'rhs' in reduceImplication
986 makeImplicationBind loc all_tvs
987 givens -- Guaranteed all Dicts or EqInsts
989 | null irreds -- If there are no irreds, we are done
990 = return ([], emptyBag)
991 | otherwise -- Otherwise we must generate a binding
992 = do { uniq <- newUnique
993 ; span <- getSrcSpanM
994 ; let (eq_givens, dict_givens) = partition isEqInst givens
996 -- extract equality binders
997 eq_cotvs = map eqInstType eq_givens
999 -- make the implication constraint instance
1000 name = mkInternalName uniq (mkVarOcc "ic") span
1001 implic_inst = ImplicInst { tci_name = name,
1002 tci_tyvars = all_tvs,
1003 tci_given = eq_givens ++ dict_givens,
1004 -- same order as binders
1005 tci_wanted = irreds,
1008 -- create binders for the irreducible dictionaries
1009 dict_irreds = filter (not . isEqInst) irreds
1010 dict_irred_ids = map instToId dict_irreds
1011 lpat = mkBigLHsPatTup (map (L span . VarPat) dict_irred_ids)
1013 -- create the binding
1014 rhs = L span (mkHsWrap co (HsVar (instToId implic_inst)))
1015 co = mkWpApps (map instToId dict_givens)
1016 <.> mkWpTyApps eq_cotvs
1017 <.> mkWpTyApps (mkTyVarTys all_tvs)
1018 bind | [dict_irred_id] <- dict_irred_ids
1019 = VarBind dict_irred_id rhs
1021 = PatBind { pat_lhs = lpat
1022 , pat_rhs = unguardedGRHSs rhs
1023 , pat_rhs_ty = hsLPatType lpat
1024 , bind_fvs = placeHolderNames
1027 ; traceTc $ text "makeImplicationBind" <+> ppr implic_inst
1028 ; return ([implic_inst], unitBag (L span bind))
1031 -----------------------------------------------------------
1032 tryHardCheckLoop :: SDoc
1034 -> TcM ([Inst], TcDictBinds)
1036 tryHardCheckLoop doc wanteds
1037 = do { (irreds,binds) <- checkLoop (mkInferRedEnv doc try_me) wanteds
1038 ; return (irreds,binds)
1042 -- Here's the try-hard bit
1044 -----------------------------------------------------------
1045 gentleCheckLoop :: InstLoc
1048 -> TcM ([Inst], TcDictBinds)
1050 gentleCheckLoop inst_loc givens wanteds
1051 = do { (irreds,binds) <- checkLoop env wanteds
1052 ; return (irreds,binds)
1055 env = mkRedEnv (pprInstLoc inst_loc) try_me givens
1057 try_me inst | isMethodOrLit inst = ReduceMe
1059 -- When checking against a given signature
1060 -- we MUST be very gentle: Note [Check gently]
1062 gentleInferLoop :: SDoc -> [Inst]
1063 -> TcM ([Inst], TcDictBinds)
1064 gentleInferLoop doc wanteds
1065 = do { (irreds, binds) <- checkLoop env wanteds
1066 ; return (irreds, binds) }
1068 env = mkInferRedEnv doc try_me
1069 try_me inst | isMethodOrLit inst = ReduceMe
1074 ~~~~~~~~~~~~~~~~~~~~
1075 We have to very careful about not simplifying too vigorously
1080 f :: Show b => T b -> b
1081 f (MkT x) = show [x]
1083 Inside the pattern match, which binds (a:*, x:a), we know that
1085 Hence we have a dictionary for Show [a] available; and indeed we
1086 need it. We are going to build an implication contraint
1087 forall a. (b~[a]) => Show [a]
1088 Later, we will solve this constraint using the knowledge (Show b)
1090 But we MUST NOT reduce (Show [a]) to (Show a), else the whole
1091 thing becomes insoluble. So we simplify gently (get rid of literals
1092 and methods only, plus common up equal things), deferring the real
1093 work until top level, when we solve the implication constraint
1094 with tryHardCheckLooop.
1098 -----------------------------------------------------------
1101 -> TcM ([Inst], TcDictBinds)
1102 -- Precondition: givens are completely rigid
1103 -- Postcondition: returned Insts are zonked
1105 checkLoop env wanteds
1107 where go env wanteds
1108 = do { -- We do need to zonk the givens; cf Note [Zonking RedEnv]
1109 ; env' <- zonkRedEnv env
1110 ; wanteds' <- zonkInsts wanteds
1112 ; (improved, binds, irreds) <- reduceContext env' wanteds'
1114 ; if null irreds || not improved then
1115 return (irreds, binds)
1118 -- If improvement did some unification, we go round again.
1119 -- We start again with irreds, not wanteds
1120 -- Using an instance decl might have introduced a fresh type
1121 -- variable which might have been unified, so we'd get an
1122 -- infinite loop if we started again with wanteds!
1124 { (irreds1, binds1) <- go env' irreds
1125 ; return (irreds1, binds `unionBags` binds1) } }
1128 Note [Zonking RedEnv]
1129 ~~~~~~~~~~~~~~~~~~~~~
1130 It might appear as if the givens in RedEnv are always rigid, but that is not
1131 necessarily the case for programs involving higher-rank types that have class
1132 contexts constraining the higher-rank variables. An example from tc237 in the
1135 class Modular s a | s -> a
1137 wim :: forall a w. Integral a
1138 => a -> (forall s. Modular s a => M s w) -> w
1139 wim i k = error "urk"
1141 test5 :: (Modular s a, Integral a) => M s a
1144 test4 = wim 4 test4'
1146 Notice how the variable 'a' of (Modular s a) in the rank-2 type of wim is
1147 quantified further outside. When type checking test4, we have to check
1148 whether the signature of test5 is an instance of
1150 (forall s. Modular s a => M s w)
1152 Consequently, we will get (Modular s t_a), where t_a is a TauTv into the
1155 Given the FD of Modular in this example, class improvement will instantiate
1156 t_a to 'a', where 'a' is the skolem from test5's signatures (due to the
1157 Modular s a predicate in that signature). If we don't zonk (Modular s t_a) in
1158 the givens, we will get into a loop as improveOne uses the unification engine
1159 Unify.tcUnifyTys, which doesn't know about mutable type variables.
1164 class If b t e r | b t e -> r
1167 class Lte a b c | a b -> c where lte :: a -> b -> c
1169 instance (Lte a b l,If l b a c) => Max a b c
1171 Wanted: Max Z (S x) y
1173 Then we'll reduce using the Max instance to:
1174 (Lte Z (S x) l, If l (S x) Z y)
1175 and improve by binding l->T, after which we can do some reduction
1176 on both the Lte and If constraints. What we *can't* do is start again
1177 with (Max Z (S x) y)!
1181 %************************************************************************
1183 tcSimplifySuperClasses
1185 %************************************************************************
1187 Note [SUPERCLASS-LOOP 1]
1188 ~~~~~~~~~~~~~~~~~~~~~~~~
1189 We have to be very, very careful when generating superclasses, lest we
1190 accidentally build a loop. Here's an example:
1194 class S a => C a where { opc :: a -> a }
1195 class S b => D b where { opd :: b -> b }
1197 instance C Int where
1200 instance D Int where
1203 From (instance C Int) we get the constraint set {ds1:S Int, dd:D Int}
1204 Simplifying, we may well get:
1205 $dfCInt = :C ds1 (opd dd)
1208 Notice that we spot that we can extract ds1 from dd.
1210 Alas! Alack! We can do the same for (instance D Int):
1212 $dfDInt = :D ds2 (opc dc)
1216 And now we've defined the superclass in terms of itself.
1217 Two more nasty cases are in
1222 - Satisfy the superclass context *all by itself*
1223 (tcSimplifySuperClasses)
1224 - And do so completely; i.e. no left-over constraints
1225 to mix with the constraints arising from method declarations
1228 Note [Recursive instances and superclases]
1229 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1230 Consider this code, which arises in the context of "Scrap Your
1231 Boilerplate with Class".
1235 instance Sat (ctx Char) => Data ctx Char
1236 instance (Sat (ctx [a]), Data ctx a) => Data ctx [a]
1238 class Data Maybe a => Foo a
1240 instance Foo t => Sat (Maybe t)
1242 instance Data Maybe a => Foo a
1243 instance Foo a => Foo [a]
1246 In the instance for Foo [a], when generating evidence for the superclasses
1247 (ie in tcSimplifySuperClasses) we need a superclass (Data Maybe [a]).
1248 Using the instance for Data, we therefore need
1249 (Sat (Maybe [a], Data Maybe a)
1250 But we are given (Foo a), and hence its superclass (Data Maybe a).
1251 So that leaves (Sat (Maybe [a])). Using the instance for Sat means
1252 we need (Foo [a]). And that is the very dictionary we are bulding
1253 an instance for! So we must put that in the "givens". So in this
1255 Given: Foo a, Foo [a]
1256 Watend: Data Maybe [a]
1258 BUT we must *not not not* put the *superclasses* of (Foo [a]) in
1259 the givens, which is what 'addGiven' would normally do. Why? Because
1260 (Data Maybe [a]) is the superclass, so we'd "satisfy" the wanted
1261 by selecting a superclass from Foo [a], which simply makes a loop.
1263 On the other hand we *must* put the superclasses of (Foo a) in
1264 the givens, as you can see from the derivation described above.
1266 Conclusion: in the very special case of tcSimplifySuperClasses
1267 we have one 'given' (namely the "this" dictionary) whose superclasses
1268 must not be added to 'givens' by addGiven. That is the *whole* reason
1269 for the red_given_scs field in RedEnv, and the function argument to
1273 tcSimplifySuperClasses
1275 -> Inst -- The dict whose superclasses
1276 -- are being figured out
1280 tcSimplifySuperClasses loc this givens sc_wanteds
1281 = do { traceTc (text "tcSimplifySuperClasses")
1282 ; (irreds,binds1) <- checkLoop env sc_wanteds
1283 ; let (tidy_env, tidy_irreds) = tidyInsts irreds
1284 ; reportNoInstances tidy_env (Just (loc, givens)) tidy_irreds
1287 env = RedEnv { red_doc = pprInstLoc loc,
1288 red_try_me = try_me,
1289 red_givens = this:givens,
1290 red_given_scs = add_scs,
1292 red_improve = False } -- No unification vars
1293 add_scs g | g==this = NoSCs
1294 | otherwise = AddSCs
1296 try_me _ = ReduceMe -- Try hard, so we completely solve the superclass
1297 -- constraints right here. See Note [SUPERCLASS-LOOP 1]
1301 %************************************************************************
1303 \subsection{tcSimplifyRestricted}
1305 %************************************************************************
1307 tcSimplifyRestricted infers which type variables to quantify for a
1308 group of restricted bindings. This isn't trivial.
1311 We want to quantify over a to get id :: forall a. a->a
1314 We do not want to quantify over a, because there's an Eq a
1315 constraint, so we get eq :: a->a->Bool (notice no forall)
1318 RHS has type 'tau', whose free tyvars are tau_tvs
1319 RHS has constraints 'wanteds'
1322 Quantify over (tau_tvs \ ftvs(wanteds))
1323 This is bad. The constraints may contain (Monad (ST s))
1324 where we have instance Monad (ST s) where...
1325 so there's no need to be monomorphic in s!
1327 Also the constraint might be a method constraint,
1328 whose type mentions a perfectly innocent tyvar:
1329 op :: Num a => a -> b -> a
1330 Here, b is unconstrained. A good example would be
1332 We want to infer the polymorphic type
1333 foo :: forall b. b -> b
1336 Plan B (cunning, used for a long time up to and including GHC 6.2)
1337 Step 1: Simplify the constraints as much as possible (to deal
1338 with Plan A's problem). Then set
1339 qtvs = tau_tvs \ ftvs( simplify( wanteds ) )
1341 Step 2: Now simplify again, treating the constraint as 'free' if
1342 it does not mention qtvs, and trying to reduce it otherwise.
1343 The reasons for this is to maximise sharing.
1345 This fails for a very subtle reason. Suppose that in the Step 2
1346 a constraint (Foo (Succ Zero) (Succ Zero) b) gets thrown upstairs as 'free'.
1347 In the Step 1 this constraint might have been simplified, perhaps to
1348 (Foo Zero Zero b), AND THEN THAT MIGHT BE IMPROVED, to bind 'b' to 'T'.
1349 This won't happen in Step 2... but that in turn might prevent some other
1350 constraint (Baz [a] b) being simplified (e.g. via instance Baz [a] T where {..})
1351 and that in turn breaks the invariant that no constraints are quantified over.
1353 Test typecheck/should_compile/tc177 (which failed in GHC 6.2) demonstrates
1358 Step 1: Simplify the constraints as much as possible (to deal
1359 with Plan A's problem). Then set
1360 qtvs = tau_tvs \ ftvs( simplify( wanteds ) )
1361 Return the bindings from Step 1.
1364 A note about Plan C (arising from "bug" reported by George Russel March 2004)
1367 instance (HasBinary ty IO) => HasCodedValue ty
1369 foo :: HasCodedValue a => String -> IO a
1371 doDecodeIO :: HasCodedValue a => () -> () -> IO a
1372 doDecodeIO codedValue view
1373 = let { act = foo "foo" } in act
1375 You might think this should work becuase the call to foo gives rise to a constraint
1376 (HasCodedValue t), which can be satisfied by the type sig for doDecodeIO. But the
1377 restricted binding act = ... calls tcSimplifyRestricted, and PlanC simplifies the
1378 constraint using the (rather bogus) instance declaration, and now we are stuffed.
1380 I claim this is not really a bug -- but it bit Sergey as well as George. So here's
1384 Plan D (a variant of plan B)
1385 Step 1: Simplify the constraints as much as possible (to deal
1386 with Plan A's problem), BUT DO NO IMPROVEMENT. Then set
1387 qtvs = tau_tvs \ ftvs( simplify( wanteds ) )
1389 Step 2: Now simplify again, treating the constraint as 'free' if
1390 it does not mention qtvs, and trying to reduce it otherwise.
1392 The point here is that it's generally OK to have too few qtvs; that is,
1393 to make the thing more monomorphic than it could be. We don't want to
1394 do that in the common cases, but in wierd cases it's ok: the programmer
1395 can always add a signature.
1397 Too few qtvs => too many wanteds, which is what happens if you do less
1402 tcSimplifyRestricted -- Used for restricted binding groups
1403 -- i.e. ones subject to the monomorphism restriction
1406 -> [Name] -- Things bound in this group
1407 -> TcTyVarSet -- Free in the type of the RHSs
1408 -> [Inst] -- Free in the RHSs
1409 -> TcM ([TyVar], -- Tyvars to quantify (zonked and quantified)
1410 TcDictBinds) -- Bindings
1411 -- tcSimpifyRestricted returns no constraints to
1412 -- quantify over; by definition there are none.
1413 -- They are all thrown back in the LIE
1415 tcSimplifyRestricted doc top_lvl bndrs tau_tvs wanteds
1416 -- Zonk everything in sight
1417 = do { traceTc (text "tcSimplifyRestricted")
1418 ; wanteds_z <- zonkInsts wanteds
1420 -- 'ReduceMe': Reduce as far as we can. Don't stop at
1421 -- dicts; the idea is to get rid of as many type
1422 -- variables as possible, and we don't want to stop
1423 -- at (say) Monad (ST s), because that reduces
1424 -- immediately, with no constraint on s.
1426 -- BUT do no improvement! See Plan D above
1427 -- HOWEVER, some unification may take place, if we instantiate
1428 -- a method Inst with an equality constraint
1429 ; let env = mkNoImproveRedEnv doc (\_ -> ReduceMe)
1430 ; (_imp, _binds, constrained_dicts) <- reduceContext env wanteds_z
1432 -- Next, figure out the tyvars we will quantify over
1433 ; tau_tvs' <- zonkTcTyVarsAndFV (varSetElems tau_tvs)
1434 ; gbl_tvs' <- tcGetGlobalTyVars
1435 ; constrained_dicts' <- zonkInsts constrained_dicts
1437 ; let qtvs1 = tau_tvs' `minusVarSet` oclose (fdPredsOfInsts constrained_dicts) gbl_tvs'
1438 -- As in tcSimplifyInfer
1440 -- Do not quantify over constrained type variables:
1441 -- this is the monomorphism restriction
1442 constrained_tvs' = tyVarsOfInsts constrained_dicts'
1443 qtvs = qtvs1 `minusVarSet` constrained_tvs'
1444 pp_bndrs = pprWithCommas (quotes . ppr) bndrs
1447 ; warn_mono <- doptM Opt_WarnMonomorphism
1448 ; warnTc (warn_mono && (constrained_tvs' `intersectsVarSet` qtvs1))
1449 (vcat[ ptext (sLit "the Monomorphism Restriction applies to the binding")
1450 <> plural bndrs <+> ptext (sLit "for") <+> pp_bndrs,
1451 ptext (sLit "Consider giving a type signature for") <+> pp_bndrs])
1453 ; traceTc (text "tcSimplifyRestricted" <+> vcat [
1454 pprInsts wanteds, pprInsts constrained_dicts',
1456 ppr constrained_tvs', ppr tau_tvs', ppr qtvs ])
1458 -- Zonk wanteds again! The first call to reduceContext may have
1459 -- instantiated some variables.
1460 -- FIXME: If red_improve would work, we could propagate that into
1461 -- the equality solver, too, to prevent instantating any
1463 ; wanteds_zz <- zonkInsts wanteds_z
1465 -- The first step may have squashed more methods than
1466 -- necessary, so try again, this time more gently, knowing the exact
1467 -- set of type variables to quantify over.
1469 -- We quantify only over constraints that are captured by qtvs;
1470 -- these will just be a subset of non-dicts. This in contrast
1471 -- to normal inference (using isFreeWhenInferring) in which we quantify over
1472 -- all *non-inheritable* constraints too. This implements choice
1473 -- (B) under "implicit parameter and monomorphism" above.
1475 -- Remember that we may need to do *some* simplification, to
1476 -- (for example) squash {Monad (ST s)} into {}. It's not enough
1477 -- just to float all constraints
1479 -- At top level, we *do* squash methods becuase we want to
1480 -- expose implicit parameters to the test that follows
1481 ; let is_nested_group = isNotTopLevel top_lvl
1482 try_me inst | isFreeWrtTyVars qtvs inst,
1483 (is_nested_group || isDict inst) = Stop
1484 | otherwise = ReduceMe
1485 env = mkNoImproveRedEnv doc try_me
1486 ; (_imp, binds, irreds) <- reduceContext env wanteds_zz
1488 -- See "Notes on implicit parameters, Question 4: top level"
1489 ; ASSERT( all (isFreeWrtTyVars qtvs) irreds ) -- None should be captured
1490 if is_nested_group then
1492 else do { let (bad_ips, non_ips) = partition isIPDict irreds
1493 ; addTopIPErrs bndrs bad_ips
1494 ; extendLIEs non_ips }
1496 ; qtvs' <- zonkQuantifiedTyVars (varSetElems qtvs)
1497 ; return (qtvs', binds) }
1501 %************************************************************************
1505 %************************************************************************
1507 On the LHS of transformation rules we only simplify methods and constants,
1508 getting dictionaries. We want to keep all of them unsimplified, to serve
1509 as the available stuff for the RHS of the rule.
1511 Example. Consider the following left-hand side of a rule
1513 f (x == y) (y > z) = ...
1515 If we typecheck this expression we get constraints
1517 d1 :: Ord a, d2 :: Eq a
1519 We do NOT want to "simplify" to the LHS
1521 forall x::a, y::a, z::a, d1::Ord a.
1522 f ((==) (eqFromOrd d1) x y) ((>) d1 y z) = ...
1526 forall x::a, y::a, z::a, d1::Ord a, d2::Eq a.
1527 f ((==) d2 x y) ((>) d1 y z) = ...
1529 Here is another example:
1531 fromIntegral :: (Integral a, Num b) => a -> b
1532 {-# RULES "foo" fromIntegral = id :: Int -> Int #-}
1534 In the rule, a=b=Int, and Num Int is a superclass of Integral Int. But
1535 we *dont* want to get
1537 forall dIntegralInt.
1538 fromIntegral Int Int dIntegralInt (scsel dIntegralInt) = id Int
1540 because the scsel will mess up RULE matching. Instead we want
1542 forall dIntegralInt, dNumInt.
1543 fromIntegral Int Int dIntegralInt dNumInt = id Int
1547 g (x == y) (y == z) = ..
1549 where the two dictionaries are *identical*, we do NOT WANT
1551 forall x::a, y::a, z::a, d1::Eq a
1552 f ((==) d1 x y) ((>) d1 y z) = ...
1554 because that will only match if the dict args are (visibly) equal.
1555 Instead we want to quantify over the dictionaries separately.
1557 In short, tcSimplifyRuleLhs must *only* squash LitInst and MethInts, leaving
1558 all dicts unchanged, with absolutely no sharing. It's simpler to do this
1559 from scratch, rather than further parameterise simpleReduceLoop etc.
1560 Simpler, maybe, but alas not simple (see Trac #2494)
1562 * Type errors may give rise to an (unsatisfiable) equality constraint
1564 * Applications of a higher-rank function on the LHS may give
1565 rise to an implication constraint, esp if there are unsatisfiable
1566 equality constraints inside.
1569 tcSimplifyRuleLhs :: [Inst] -> TcM ([Inst], TcDictBinds)
1570 tcSimplifyRuleLhs wanteds
1571 = do { wanteds' <- zonkInsts wanteds
1572 ; (irreds, binds) <- go [] emptyBag wanteds'
1573 ; let (dicts, bad_irreds) = partition isDict irreds
1574 ; traceTc (text "tcSimplifyrulelhs" <+> pprInsts bad_irreds)
1575 ; addNoInstanceErrs (nub bad_irreds)
1576 -- The nub removes duplicates, which has
1577 -- not happened otherwise (see notes above)
1578 ; return (dicts, binds) }
1580 go :: [Inst] -> TcDictBinds -> [Inst] -> TcM ([Inst], TcDictBinds)
1582 = return (irreds, binds)
1583 go irreds binds (w:ws)
1585 = go (w:irreds) binds ws
1586 | isImplicInst w -- Have a go at reducing the implication
1587 = do { (binds1, irreds1) <- reduceImplication red_env w
1588 ; let (bad_irreds, ok_irreds) = partition isImplicInst irreds1
1589 ; go (bad_irreds ++ irreds)
1590 (binds `unionBags` binds1)
1593 = do { w' <- zonkInst w -- So that (3::Int) does not generate a call
1594 -- to fromInteger; this looks fragile to me
1595 ; lookup_result <- lookupSimpleInst w'
1596 ; case lookup_result of
1597 NoInstance -> go (w:irreds) binds ws
1598 GenInst ws' rhs -> go irreds binds' (ws' ++ ws)
1600 binds' = addInstToDictBind binds w rhs
1603 -- Sigh: we need to reduce inside implications
1604 red_env = mkInferRedEnv doc try_me
1605 doc = ptext (sLit "Implication constraint in RULE lhs")
1606 try_me inst | isMethodOrLit inst = ReduceMe
1607 | otherwise = Stop -- Be gentle
1610 tcSimplifyBracket is used when simplifying the constraints arising from
1611 a Template Haskell bracket [| ... |]. We want to check that there aren't
1612 any constraints that can't be satisfied (e.g. Show Foo, where Foo has no
1613 Show instance), but we aren't otherwise interested in the results.
1614 Nor do we care about ambiguous dictionaries etc. We will type check
1615 this bracket again at its usage site.
1618 tcSimplifyBracket :: [Inst] -> TcM ()
1619 tcSimplifyBracket wanteds
1620 = do { tryHardCheckLoop doc wanteds
1623 doc = text "tcSimplifyBracket"
1627 %************************************************************************
1629 \subsection{Filtering at a dynamic binding}
1631 %************************************************************************
1636 we must discharge all the ?x constraints from B. We also do an improvement
1637 step; if we have ?x::t1 and ?x::t2 we must unify t1, t2.
1639 Actually, the constraints from B might improve the types in ?x. For example
1641 f :: (?x::Int) => Char -> Char
1644 then the constraint (?x::Int) arising from the call to f will
1645 force the binding for ?x to be of type Int.
1648 tcSimplifyIPs :: [Inst] -- The implicit parameters bound here
1651 -- We need a loop so that we do improvement, and then
1652 -- (next time round) generate a binding to connect the two
1654 -- Here the two ?x's have different types, and improvement
1655 -- makes them the same.
1657 tcSimplifyIPs given_ips wanteds
1658 = do { wanteds' <- zonkInsts wanteds
1659 ; given_ips' <- zonkInsts given_ips
1660 -- Unusually for checking, we *must* zonk the given_ips
1662 ; let env = mkRedEnv doc try_me given_ips'
1663 ; (improved, binds, irreds) <- reduceContext env wanteds'
1665 ; if null irreds || not improved then
1666 ASSERT( all is_free irreds )
1667 do { extendLIEs irreds
1670 -- If improvement did some unification, we go round again.
1671 -- We start again with irreds, not wanteds
1672 -- Using an instance decl might have introduced a fresh type
1673 -- variable which might have been unified, so we'd get an
1674 -- infinite loop if we started again with wanteds!
1676 { binds1 <- tcSimplifyIPs given_ips' irreds
1677 ; return $ binds `unionBags` binds1
1680 doc = text "tcSimplifyIPs" <+> ppr given_ips
1681 ip_set = mkNameSet (ipNamesOfInsts given_ips)
1682 is_free inst = isFreeWrtIPs ip_set inst
1684 -- Simplify any methods that mention the implicit parameter
1685 try_me inst | is_free inst = Stop
1686 | otherwise = ReduceMe
1690 %************************************************************************
1692 \subsection[binds-for-local-funs]{@bindInstsOfLocalFuns@}
1694 %************************************************************************
1696 When doing a binding group, we may have @Insts@ of local functions.
1697 For example, we might have...
1699 let f x = x + 1 -- orig local function (overloaded)
1700 f.1 = f Int -- two instances of f
1705 The point is: we must drop the bindings for @f.1@ and @f.2@ here,
1706 where @f@ is in scope; those @Insts@ must certainly not be passed
1707 upwards towards the top-level. If the @Insts@ were binding-ified up
1708 there, they would have unresolvable references to @f@.
1710 We pass in an @init_lie@ of @Insts@ and a list of locally-bound @Ids@.
1711 For each method @Inst@ in the @init_lie@ that mentions one of the
1712 @Ids@, we create a binding. We return the remaining @Insts@ (in an
1713 @LIE@), as well as the @HsBinds@ generated.
1716 bindInstsOfLocalFuns :: [Inst] -> [TcId] -> TcM TcDictBinds
1717 -- Simlifies only MethodInsts, and generate only bindings of form
1719 -- We're careful not to even generate bindings of the form
1721 -- You'd think that'd be fine, but it interacts with what is
1722 -- arguably a bug in Match.tidyEqnInfo (see notes there)
1724 bindInstsOfLocalFuns wanteds local_ids
1725 | null overloaded_ids = do
1728 return emptyLHsBinds
1731 = do { (irreds, binds) <- gentleInferLoop doc for_me
1732 ; extendLIEs not_for_me
1736 doc = text "bindInsts" <+> ppr local_ids
1737 overloaded_ids = filter is_overloaded local_ids
1738 is_overloaded id = isOverloadedTy (idType id)
1739 (for_me, not_for_me) = partition (isMethodFor overloaded_set) wanteds
1741 overloaded_set = mkVarSet overloaded_ids -- There can occasionally be a lot of them
1742 -- so it's worth building a set, so that
1743 -- lookup (in isMethodFor) is faster
1747 %************************************************************************
1749 \subsection{Data types for the reduction mechanism}
1751 %************************************************************************
1753 The main control over context reduction is here
1757 = RedEnv { red_doc :: SDoc -- The context
1758 , red_try_me :: Inst -> WhatToDo
1759 , red_improve :: Bool -- True <=> do improvement
1760 , red_givens :: [Inst] -- All guaranteed rigid
1761 -- Always dicts & equalities
1762 -- but see Note [Rigidity]
1764 , red_given_scs :: Inst -> WantSCs -- See Note [Recursive instances and superclases]
1766 , red_stack :: (Int, [Inst]) -- Recursion stack (for err msg)
1767 -- See Note [RedStack]
1771 -- The red_givens are rigid so far as cmpInst is concerned.
1772 -- There is one case where they are not totally rigid, namely in tcSimplifyIPs
1773 -- let ?x = e in ...
1774 -- Here, the given is (?x::a), where 'a' is not necy a rigid type
1775 -- But that doesn't affect the comparison, which is based only on mame.
1778 -- The red_stack pair (n,insts) pair is just used for error reporting.
1779 -- 'n' is always the depth of the stack.
1780 -- The 'insts' is the stack of Insts being reduced: to produce X
1781 -- I had to produce Y, to produce Y I had to produce Z, and so on.
1784 mkRedEnv :: SDoc -> (Inst -> WhatToDo) -> [Inst] -> RedEnv
1785 mkRedEnv doc try_me givens
1786 = RedEnv { red_doc = doc, red_try_me = try_me,
1787 red_givens = givens,
1788 red_given_scs = const AddSCs,
1790 red_improve = True }
1792 mkInferRedEnv :: SDoc -> (Inst -> WhatToDo) -> RedEnv
1794 mkInferRedEnv doc try_me
1795 = RedEnv { red_doc = doc, red_try_me = try_me,
1797 red_given_scs = const AddSCs,
1799 red_improve = True }
1801 mkNoImproveRedEnv :: SDoc -> (Inst -> WhatToDo) -> RedEnv
1802 -- Do not do improvement; no givens
1803 mkNoImproveRedEnv doc try_me
1804 = RedEnv { red_doc = doc, red_try_me = try_me,
1806 red_given_scs = const AddSCs,
1808 red_improve = True }
1811 = ReduceMe -- Try to reduce this
1812 -- If there's no instance, add the inst to the
1813 -- irreductible ones, but don't produce an error
1814 -- message of any kind.
1815 -- It might be quite legitimate such as (Eq a)!
1817 | Stop -- Return as irreducible unless it can
1818 -- be reduced to a constant in one step
1819 -- Do not add superclasses; see
1821 data WantSCs = NoSCs | AddSCs -- Tells whether we should add the superclasses
1822 -- of a predicate when adding it to the avails
1823 -- The reason for this flag is entirely the super-class loop problem
1824 -- Note [SUPER-CLASS LOOP 1]
1826 zonkRedEnv :: RedEnv -> TcM RedEnv
1828 = do { givens' <- mapM zonkInst (red_givens env)
1829 ; return $ env {red_givens = givens'}
1834 %************************************************************************
1836 \subsection[reduce]{@reduce@}
1838 %************************************************************************
1840 Note [Ancestor Equalities]
1841 ~~~~~~~~~~~~~~~~~~~~~~~~~~
1842 During context reduction, we add to the wanted equalities also those
1843 equalities that (transitively) occur in superclass contexts of wanted
1844 class constraints. Consider the following code
1846 class a ~ Int => C a
1849 If (C a) is wanted, we want to add (a ~ Int), which will be discharged by
1850 substituting Int for a. Hence, we ultimately want (C Int), which we
1851 discharge with the explicit instance.
1854 reduceContext :: RedEnv
1856 -> TcM (ImprovementDone,
1857 TcDictBinds, -- Dictionary bindings
1858 [Inst]) -- Irreducible
1860 reduceContext env wanteds0
1861 = do { traceTc (text "reduceContext" <+> (vcat [
1862 text "----------------------",
1864 text "given" <+> ppr (red_givens env),
1865 text "wanted" <+> ppr wanteds0,
1866 text "----------------------"
1869 -- We want to add as wanted equalities those that (transitively)
1870 -- occur in superclass contexts of wanted class constraints.
1871 -- See Note [Ancestor Equalities]
1872 ; ancestor_eqs <- ancestorEqualities wanteds0
1873 ; traceTc $ text "reduceContext: ancestor eqs" <+> ppr ancestor_eqs
1875 -- Normalise and solve all equality constraints as far as possible
1876 -- and normalise all dictionary constraints wrt to the reduced
1877 -- equalities. The returned wanted constraints include the
1878 -- irreducible wanted equalities.
1879 ; let wanteds = wanteds0 ++ ancestor_eqs
1880 givens = red_givens env
1884 eq_improved) <- tcReduceEqs givens wanteds
1885 ; traceTc $ text "reduceContext: tcReduceEqs result" <+> vcat
1886 [ppr givens', ppr wanteds', ppr normalise_binds]
1888 -- Build the Avail mapping from "given_dicts"
1889 ; (init_state, _) <- getLIE $ do
1890 { init_state <- foldlM (addGiven (red_given_scs env))
1895 -- Solve the *wanted* *dictionary* constraints (not implications)
1896 -- This may expose some further equational constraints in the course
1897 -- of improvement due to functional dependencies if any of the
1898 -- involved unifications gets deferred.
1899 ; let (wanted_implics, wanted_dicts) = partition isImplicInst wanteds'
1900 ; (avails, extra_eqs) <- getLIE (reduceList env wanted_dicts init_state)
1901 -- The getLIE is reqd because reduceList does improvement
1902 -- (via extendAvails) which may in turn do unification
1905 dict_irreds) <- extractResults avails wanted_dicts
1906 ; traceTc $ text "reduceContext: extractResults" <+> vcat
1907 [ppr avails, ppr wanted_dicts, ppr dict_binds]
1909 -- Solve the wanted *implications*. In doing so, we can provide
1910 -- as "given" all the dicts that were originally given,
1911 -- *or* for which we now have bindings,
1912 -- *or* which are now irreds
1913 -- NB: Equality irreds need to be converted, as the recursive
1914 -- invocation of the solver will still treat them as wanteds
1916 ; let implic_env = env { red_givens
1917 = givens ++ bound_dicts ++
1918 map wantedToLocalEqInst dict_irreds }
1919 ; (implic_binds_s, implic_irreds_s)
1920 <- mapAndUnzipM (reduceImplication implic_env) wanted_implics
1921 ; let implic_binds = unionManyBags implic_binds_s
1922 implic_irreds = concat implic_irreds_s
1924 -- Collect all irreducible instances, and determine whether we should
1925 -- go round again. We do so in either of two cases:
1926 -- (1) If dictionary reduction or equality solving led to
1927 -- improvement (i.e., instantiated type variables).
1928 -- (2) If we reduced dictionaries (i.e., got dictionary bindings),
1929 -- they may have exposed further opportunities to normalise
1930 -- family applications. See Note [Dictionary Improvement]
1932 -- NB: We do *not* go around for new extra_eqs. Morally, we should,
1933 -- but we can't without risking non-termination (see #2688). By
1934 -- not going around, we miss some legal programs mixing FDs and
1935 -- TFs, but we never claimed to support such programs in the
1936 -- current implementation anyway.
1938 ; let all_irreds = dict_irreds ++ implic_irreds ++ extra_eqs
1939 avails_improved = availsImproved avails
1940 improvedFlexible = avails_improved || eq_improved
1941 reduced_dicts = not (isEmptyBag dict_binds)
1942 improved = improvedFlexible || reduced_dicts
1944 improvedHint = (if avails_improved then " [AVAILS]" else "") ++
1945 (if eq_improved then " [EQ]" else "")
1947 ; traceTc (text "reduceContext end" <+> (vcat [
1948 text "----------------------",
1950 text "given" <+> ppr givens,
1951 text "wanted" <+> ppr wanteds0,
1953 text "avails" <+> pprAvails avails,
1954 text "improved =" <+> ppr improved <+> text improvedHint,
1955 text "(all) irreds = " <+> ppr all_irreds,
1956 text "dict-binds = " <+> ppr dict_binds,
1957 text "implic-binds = " <+> ppr implic_binds,
1958 text "----------------------"
1962 normalise_binds `unionBags` dict_binds
1963 `unionBags` implic_binds,
1967 tcImproveOne :: Avails -> Inst -> TcM ImprovementDone
1968 tcImproveOne avails inst
1969 | not (isDict inst) = return False
1971 = do { inst_envs <- tcGetInstEnvs
1972 ; let eqns = improveOne (classInstances inst_envs)
1973 (dictPred inst, pprInstArising inst)
1974 [ (dictPred p, pprInstArising p)
1975 | p <- availsInsts avails, isDict p ]
1976 -- Avails has all the superclasses etc (good)
1977 -- It also has all the intermediates of the deduction (good)
1978 -- It does not have duplicates (good)
1979 -- NB that (?x::t1) and (?x::t2) will be held separately in
1980 -- avails so that improve will see them separate
1981 ; traceTc (text "improveOne" <+> ppr inst)
1984 unifyEqns :: [(Equation, (PredType, SDoc), (PredType, SDoc))]
1985 -> TcM ImprovementDone
1986 unifyEqns [] = return False
1988 = do { traceTc (ptext (sLit "Improve:") <+> vcat (map pprEquationDoc eqns))
1989 ; improved <- mapM unify eqns
1990 ; return $ or improved
1993 unify ((qtvs, pairs), what1, what2)
1994 = addErrCtxtM (mkEqnMsg what1 what2) $
1995 do { let freeTyVars = unionVarSets (map tvs_pr pairs)
1997 ; (_, _, tenv) <- tcInstTyVars (varSetElems qtvs)
1998 ; mapM_ (unif_pr tenv) pairs
1999 ; anyM isFilledMetaTyVar $ varSetElems freeTyVars
2002 unif_pr tenv (ty1, ty2) = unifyType (substTy tenv ty1) (substTy tenv ty2)
2004 tvs_pr (ty1, ty2) = tyVarsOfType ty1 `unionVarSet` tyVarsOfType ty2
2006 pprEquationDoc :: (Equation, (PredType, SDoc), (PredType, SDoc)) -> SDoc
2007 pprEquationDoc (eqn, (p1, _), (p2, _))
2008 = vcat [pprEquation eqn, nest 2 (ppr p1), nest 2 (ppr p2)]
2010 mkEqnMsg :: (TcPredType, SDoc) -> (TcPredType, SDoc) -> TidyEnv
2011 -> TcM (TidyEnv, SDoc)
2012 mkEqnMsg (pred1,from1) (pred2,from2) tidy_env
2013 = do { pred1' <- zonkTcPredType pred1
2014 ; pred2' <- zonkTcPredType pred2
2015 ; let { pred1'' = tidyPred tidy_env pred1'
2016 ; pred2'' = tidyPred tidy_env pred2' }
2017 ; let msg = vcat [ptext (sLit "When using functional dependencies to combine"),
2018 nest 2 (sep [ppr pred1'' <> comma, nest 2 from1]),
2019 nest 2 (sep [ppr pred2'' <> comma, nest 2 from2])]
2020 ; return (tidy_env, msg) }
2023 Note [Dictionary Improvement]
2024 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2025 In reduceContext, we first reduce equalities and then class constraints.
2026 However, the letter may expose further opportunities for the former. Hence,
2027 we need to go around again if dictionary reduction produced any dictionary
2028 bindings. The following example demonstrated the point:
2030 data EX _x _y (p :: * -> *)
2035 class Base (Def p) => Prop p where
2039 instance Prop () where
2042 instance (Base (Def (p ANY))) => Base (EX _x _y p)
2043 instance (Prop (p ANY)) => Prop (EX _x _y p) where
2044 type Def (EX _x _y p) = EX _x _y p
2047 instance Prop (FOO x) where
2048 type Def (FOO x) = ()
2051 instance Prop BAR where
2052 type Def BAR = EX () () FOO
2054 During checking the last instance declaration, we need to check the superclass
2055 cosntraint Base (Def BAR), which family normalisation reduced to
2056 Base (EX () () FOO). Chasing the instance for Base (EX _x _y p), gives us
2057 Base (Def (FOO ANY)), which again requires family normalisation of Def to
2058 Base () before we can finish.
2061 The main context-reduction function is @reduce@. Here's its game plan.
2064 reduceList :: RedEnv -> [Inst] -> Avails -> TcM Avails
2065 reduceList env@(RedEnv {red_stack = (n,stk)}) wanteds state
2066 = do { traceTc (text "reduceList " <+> (ppr wanteds $$ ppr state))
2068 ; when (debugIsOn && (n > 8)) $ do
2069 debugDumpTcRn (hang (ptext (sLit "Interesting! Context reduction stack depth") <+> int n)
2070 2 (ifPprDebug (nest 2 (pprStack stk))))
2071 ; if n >= ctxtStkDepth dopts then
2072 failWithTc (reduceDepthErr n stk)
2076 go [] state = return state
2077 go (w:ws) state = do { state' <- reduce (env {red_stack = (n+1, w:stk)}) w state
2080 -- Base case: we're done!
2081 reduce :: RedEnv -> Inst -> Avails -> TcM Avails
2082 reduce env wanted avails
2084 -- We don't reduce equalities here (and they must not end up as irreds
2089 -- It's the same as an existing inst, or a superclass thereof
2090 | Just _ <- findAvail avails wanted
2091 = do { traceTc (text "reduce: found " <+> ppr wanted)
2096 = do { traceTc (text "reduce" <+> ppr wanted $$ ppr avails)
2097 ; case red_try_me env wanted of {
2098 Stop -> try_simple (addIrred NoSCs);
2099 -- See Note [No superclasses for Stop]
2101 ReduceMe -> do -- It should be reduced
2102 { (avails, lookup_result) <- reduceInst env avails wanted
2103 ; case lookup_result of
2104 NoInstance -> addIrred AddSCs avails wanted
2105 -- Add it and its superclasses
2107 GenInst [] rhs -> addWanted AddSCs avails wanted rhs []
2109 GenInst wanteds' rhs
2110 -> do { avails1 <- addIrred NoSCs avails wanted
2111 ; avails2 <- reduceList env wanteds' avails1
2112 ; addWanted AddSCs avails2 wanted rhs wanteds' } }
2113 -- Temporarily do addIrred *before* the reduceList,
2114 -- which has the effect of adding the thing we are trying
2115 -- to prove to the database before trying to prove the things it
2116 -- needs. See note [RECURSIVE DICTIONARIES]
2117 -- NB: we must not do an addWanted before, because that adds the
2118 -- superclasses too, and that can lead to a spurious loop; see
2119 -- the examples in [SUPERCLASS-LOOP]
2120 -- So we do an addIrred before, and then overwrite it afterwards with addWanted
2123 -- First, see if the inst can be reduced to a constant in one step
2124 -- Works well for literals (1::Int) and constant dictionaries (d::Num Int)
2125 -- Don't bother for implication constraints, which take real work
2126 try_simple do_this_otherwise
2127 = do { res <- lookupSimpleInst wanted
2129 GenInst [] rhs -> addWanted AddSCs avails wanted rhs []
2130 _ -> do_this_otherwise avails wanted }
2134 Note [RECURSIVE DICTIONARIES]
2135 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2137 data D r = ZeroD | SuccD (r (D r));
2139 instance (Eq (r (D r))) => Eq (D r) where
2140 ZeroD == ZeroD = True
2141 (SuccD a) == (SuccD b) = a == b
2144 equalDC :: D [] -> D [] -> Bool;
2147 We need to prove (Eq (D [])). Here's how we go:
2151 by instance decl, holds if
2155 by instance decl of Eq, holds if
2157 where d2 = dfEqList d3
2160 But now we can "tie the knot" to give
2166 and it'll even run! The trick is to put the thing we are trying to prove
2167 (in this case Eq (D []) into the database before trying to prove its
2168 contributing clauses.
2170 Note [SUPERCLASS-LOOP 2]
2171 ~~~~~~~~~~~~~~~~~~~~~~~~
2172 We need to be careful when adding "the constaint we are trying to prove".
2173 Suppose we are *given* d1:Ord a, and want to deduce (d2:C [a]) where
2175 class Ord a => C a where
2176 instance Ord [a] => C [a] where ...
2178 Then we'll use the instance decl to deduce C [a] from Ord [a], and then add the
2179 superclasses of C [a] to avails. But we must not overwrite the binding
2180 for Ord [a] (which is obtained from Ord a) with a superclass selection or we'll just
2183 Here's another variant, immortalised in tcrun020
2184 class Monad m => C1 m
2185 class C1 m => C2 m x
2186 instance C2 Maybe Bool
2187 For the instance decl we need to build (C1 Maybe), and it's no good if
2188 we run around and add (C2 Maybe Bool) and its superclasses to the avails
2189 before we search for C1 Maybe.
2191 Here's another example
2192 class Eq b => Foo a b
2193 instance Eq a => Foo [a] a
2197 we'll first deduce that it holds (via the instance decl). We must not
2198 then overwrite the Eq t constraint with a superclass selection!
2200 At first I had a gross hack, whereby I simply did not add superclass constraints
2201 in addWanted, though I did for addGiven and addIrred. This was sub-optimal,
2202 becuase it lost legitimate superclass sharing, and it still didn't do the job:
2203 I found a very obscure program (now tcrun021) in which improvement meant the
2204 simplifier got two bites a the cherry... so something seemed to be an Stop
2205 first time, but reducible next time.
2207 Now we implement the Right Solution, which is to check for loops directly
2208 when adding superclasses. It's a bit like the occurs check in unification.
2212 %************************************************************************
2214 Reducing a single constraint
2216 %************************************************************************
2219 ---------------------------------------------
2220 reduceInst :: RedEnv -> Avails -> Inst -> TcM (Avails, LookupInstResult)
2221 reduceInst _ avails other_inst
2222 = do { result <- lookupSimpleInst other_inst
2223 ; return (avails, result) }
2226 Note [Equational Constraints in Implication Constraints]
2227 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2229 An implication constraint is of the form
2231 where Given and Wanted may contain both equational and dictionary
2232 constraints. The delay and reduction of these two kinds of constraints
2235 -) In the generated code, wanted Dictionary constraints are wrapped up in an
2236 implication constraint that is created at the code site where the wanted
2237 dictionaries can be reduced via a let-binding. This let-bound implication
2238 constraint is deconstructed at the use-site of the wanted dictionaries.
2240 -) While the reduction of equational constraints is also delayed, the delay
2241 is not manifest in the generated code. The required evidence is generated
2242 in the code directly at the use-site. There is no let-binding and deconstruction
2243 necessary. The main disadvantage is that we cannot exploit sharing as the
2244 same evidence may be generated at multiple use-sites. However, this disadvantage
2245 is limited because it only concerns coercions which are erased.
2247 The different treatment is motivated by the different in representation. Dictionary
2248 constraints require manifest runtime dictionaries, while equations require coercions
2252 ---------------------------------------------
2253 reduceImplication :: RedEnv
2255 -> TcM (TcDictBinds, [Inst])
2258 Suppose we are simplifying the constraint
2259 forall bs. extras => wanted
2260 in the context of an overall simplification problem with givens 'givens'.
2263 * The 'givens' need not mention any of the quantified type variables
2264 e.g. forall {}. Eq a => Eq [a]
2265 forall {}. C Int => D (Tree Int)
2267 This happens when you have something like
2269 T1 :: Eq a => a -> T a
2272 f x = ...(case x of { T1 v -> v==v })...
2275 -- ToDo: should we instantiate tvs? I think it's not necessary
2277 -- Note on coercion variables:
2279 -- The extra given coercion variables are bound at two different
2282 -- -) in the creation context of the implication constraint
2283 -- the solved equational constraints use these binders
2285 -- -) at the solving site of the implication constraint
2286 -- the solved dictionaries use these binders;
2287 -- these binders are generated by reduceImplication
2289 -- Note [Binders for equalities]
2290 -- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2291 -- To reuse the binders of local/given equalities in the binders of
2292 -- implication constraints, it is crucial that these given equalities
2293 -- always have the form
2295 -- where cotv is a simple coercion type variable (and not a more
2296 -- complex coercion term). We require that the extra_givens always
2297 -- have this form and exploit the special form when generating binders.
2298 reduceImplication env
2299 orig_implic@(ImplicInst { tci_name = name, tci_loc = inst_loc,
2301 tci_given = extra_givens, tci_wanted = wanteds
2303 = do { -- Solve the sub-problem
2304 ; let try_me _ = ReduceMe -- Note [Freeness and implications]
2305 env' = env { red_givens = extra_givens ++ red_givens env
2306 , red_doc = sep [ptext (sLit "reduceImplication for")
2308 nest 2 (parens $ ptext (sLit "within")
2310 , red_try_me = try_me }
2312 ; traceTc (text "reduceImplication" <+> vcat
2313 [ ppr (red_givens env), ppr extra_givens,
2315 ; (irreds, binds) <- checkLoop env' wanteds
2317 ; traceTc (text "reduceImplication result" <+> vcat
2318 [ppr irreds, ppr binds])
2320 ; -- extract superclass binds
2321 -- (sc_binds,_) <- extractResults avails []
2322 -- ; traceTc (text "reduceImplication sc_binds" <+> vcat
2323 -- [ppr sc_binds, ppr avails])
2326 -- SLPJ Sept 07: what if improvement happened inside the checkLoop?
2327 -- Then we must iterate the outer loop too!
2329 ; didntSolveWantedEqs <- allM wantedEqInstIsUnsolved wanteds
2330 -- we solve wanted eqs by side effect!
2332 -- Progress is no longer measered by the number of bindings
2333 -- If there are any irreds, but no bindings and no solved
2334 -- equalities, we back off and do nothing
2335 ; let backOff = isEmptyLHsBinds binds && -- no new bindings
2336 (not $ null irreds) && -- but still some irreds
2337 didntSolveWantedEqs -- no instantiated cotv
2339 ; if backOff then -- No progress
2340 return (emptyBag, [orig_implic])
2342 { (simpler_implic_insts, bind)
2343 <- makeImplicationBind inst_loc tvs extra_givens irreds
2344 -- This binding is useless if the recursive simplification
2345 -- made no progress; but currently we don't try to optimise that
2346 -- case. After all, we only try hard to reduce at top level, or
2347 -- when inferring types.
2349 ; let -- extract Id binders for dicts and CoTyVar binders for eqs;
2350 -- see Note [Binders for equalities]
2351 (extra_eq_givens, extra_dict_givens) = partition isEqInst
2353 eq_cotvs = map instToVar extra_eq_givens
2354 dict_ids = map instToId extra_dict_givens
2356 -- Note [Always inline implication constraints]
2357 wrap_inline | null dict_ids = idHsWrapper
2358 | otherwise = WpInline
2361 <.> mkWpTyLams eq_cotvs
2362 <.> mkWpLams dict_ids
2363 <.> WpLet (binds `unionBags` bind)
2364 rhs = mkLHsWrap co payload
2365 loc = instLocSpan inst_loc
2366 -- wanted equalities are solved by updating their
2367 -- cotv; we don't generate bindings for them
2368 dict_bndrs = map (L loc . HsVar . instToId)
2369 . filter (not . isEqInst)
2371 payload = mkBigLHsTup dict_bndrs
2374 ; traceTc (vcat [text "reduceImplication" <+> ppr name,
2375 ppr simpler_implic_insts,
2376 text "->" <+> ppr rhs])
2377 ; return (unitBag (L loc (VarBind (instToId orig_implic) rhs)),
2378 simpler_implic_insts)
2381 reduceImplication _ i = pprPanic "reduceImplication" (ppr i)
2384 Note [Always inline implication constraints]
2385 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2386 Suppose an implication constraint floats out of an INLINE function.
2387 Then although the implication has a single call site, it won't be
2388 inlined. And that is bad because it means that even if there is really
2389 *no* overloading (type signatures specify the exact types) there will
2390 still be dictionary passing in the resulting code. To avert this,
2391 we mark the implication constraints themselves as INLINE, at least when
2392 there is no loss of sharing as a result.
2394 Note [Freeness and implications]
2395 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2396 It's hard to say when an implication constraint can be floated out. Consider
2397 forall {} Eq a => Foo [a]
2398 The (Foo [a]) doesn't mention any of the quantified variables, but it
2399 still might be partially satisfied by the (Eq a).
2401 There is a useful special case when it *is* easy to partition the
2402 constraints, namely when there are no 'givens'. Consider
2403 forall {a}. () => Bar b
2404 There are no 'givens', and so there is no reason to capture (Bar b).
2405 We can let it float out. But if there is even one constraint we
2406 must be much more careful:
2407 forall {a}. C a b => Bar (m b)
2408 because (C a b) might have a superclass (D b), from which we might
2409 deduce (Bar [b]) when m later gets instantiated to []. Ha!
2411 Here is an even more exotic example
2413 Now consider the constraint
2414 forall b. D Int b => C Int
2415 We can satisfy the (C Int) from the superclass of D, so we don't want
2416 to float the (C Int) out, even though it mentions no type variable in
2419 One more example: the constraint
2421 instance (C a, E c) => E (a,c)
2423 constraint: forall b. D Int b => E (Int,c)
2425 You might think that the (D Int b) can't possibly contribute
2426 to solving (E (Int,c)), since the latter mentions 'c'. But
2427 in fact it can, because solving the (E (Int,c)) constraint needs
2430 and the (C Int) can be satisfied from the superclass of (D Int b).
2431 So we must still not float (E (Int,c)) out.
2433 To think about: special cases for unary type classes?
2435 Note [Pruning the givens in an implication constraint]
2436 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2437 Suppose we are about to form the implication constraint
2438 forall tvs. Eq a => Ord b
2439 The (Eq a) cannot contribute to the (Ord b), because it has no access to
2440 the type variable 'b'. So we could filter out the (Eq a) from the givens.
2441 But BE CAREFUL of the examples above in [Freeness and implications].
2443 Doing so would be a bit tidier, but all the implication constraints get
2444 simplified away by the optimiser, so it's no great win. So I don't take
2445 advantage of that at the moment.
2447 If you do, BE CAREFUL of wobbly type variables.
2450 %************************************************************************
2452 Avails and AvailHow: the pool of evidence
2454 %************************************************************************
2458 data Avails = Avails !ImprovementDone !AvailEnv
2460 type ImprovementDone = Bool -- True <=> some unification has happened
2461 -- so some Irreds might now be reducible
2462 -- keys that are now
2464 type AvailEnv = FiniteMap Inst AvailHow
2466 = IsIrred -- Used for irreducible dictionaries,
2467 -- which are going to be lambda bound
2469 | Given Inst -- Used for dictionaries for which we have a binding
2470 -- e.g. those "given" in a signature
2472 | Rhs -- Used when there is a RHS
2473 (LHsExpr TcId) -- The RHS
2474 [Inst] -- Insts free in the RHS; we need these too
2476 instance Outputable Avails where
2479 pprAvails :: Avails -> SDoc
2480 pprAvails (Avails imp avails)
2481 = vcat [ ptext (sLit "Avails") <> (if imp then ptext (sLit "[improved]") else empty)
2483 vcat [ sep [ppr inst, nest 2 (equals <+> ppr avail)]
2484 | (inst,avail) <- fmToList avails ]]
2486 instance Outputable AvailHow where
2489 -------------------------
2490 pprAvail :: AvailHow -> SDoc
2491 pprAvail IsIrred = text "Irred"
2492 pprAvail (Given x) = text "Given" <+> ppr x
2493 pprAvail (Rhs rhs bs) = sep [text "Rhs" <+> ppr bs,
2496 -------------------------
2497 extendAvailEnv :: AvailEnv -> Inst -> AvailHow -> AvailEnv
2498 extendAvailEnv env inst avail = addToFM env inst avail
2500 findAvailEnv :: AvailEnv -> Inst -> Maybe AvailHow
2501 findAvailEnv env wanted = lookupFM env wanted
2502 -- NB 1: the Ord instance of Inst compares by the class/type info
2503 -- *not* by unique. So
2504 -- d1::C Int == d2::C Int
2506 emptyAvails :: Avails
2507 emptyAvails = Avails False emptyFM
2509 findAvail :: Avails -> Inst -> Maybe AvailHow
2510 findAvail (Avails _ avails) wanted = findAvailEnv avails wanted
2512 elemAvails :: Inst -> Avails -> Bool
2513 elemAvails wanted (Avails _ avails) = wanted `elemFM` avails
2515 extendAvails :: Avails -> Inst -> AvailHow -> TcM Avails
2517 extendAvails avails@(Avails imp env) inst avail
2518 = do { imp1 <- tcImproveOne avails inst -- Do any improvement
2519 ; return (Avails (imp || imp1) (extendAvailEnv env inst avail)) }
2521 availsInsts :: Avails -> [Inst]
2522 availsInsts (Avails _ avails) = keysFM avails
2524 availsImproved :: Avails -> ImprovementDone
2525 availsImproved (Avails imp _) = imp
2528 Extracting the bindings from a bunch of Avails.
2529 The bindings do *not* come back sorted in dependency order.
2530 We assume that they'll be wrapped in a big Rec, so that the
2531 dependency analyser can sort them out later
2534 type DoneEnv = FiniteMap Inst [Id]
2535 -- Tracks which things we have evidence for
2537 extractResults :: Avails
2539 -> TcM (TcDictBinds, -- Bindings
2540 [Inst], -- The insts bound by the bindings
2541 [Inst]) -- Irreducible ones
2542 -- Note [Reducing implication constraints]
2544 extractResults (Avails _ avails) wanteds
2545 = go emptyBag [] [] emptyFM wanteds
2547 go :: TcDictBinds -- Bindings for dicts
2548 -> [Inst] -- Bound by the bindings
2550 -> DoneEnv -- Has an entry for each inst in the above three sets
2552 -> TcM (TcDictBinds, [Inst], [Inst])
2553 go binds bound_dicts irreds _ []
2554 = return (binds, bound_dicts, irreds)
2556 go binds bound_dicts irreds done (w:ws)
2558 = go binds bound_dicts (w:irreds) done' ws
2560 | Just done_ids@(done_id : rest_done_ids) <- lookupFM done w
2561 = if w_id `elem` done_ids then
2562 go binds bound_dicts irreds done ws
2564 go (add_bind (nlHsVar done_id)) bound_dicts irreds
2565 (addToFM done w (done_id : w_id : rest_done_ids)) ws
2567 | otherwise -- Not yet done
2568 = case findAvailEnv avails w of
2569 Nothing -> pprTrace "Urk: extractResults" (ppr w) $
2570 go binds bound_dicts irreds done ws
2572 Just IsIrred -> go binds bound_dicts (w:irreds) done' ws
2574 Just (Rhs rhs ws') -> go (add_bind rhs) (w:bound_dicts) irreds done' (ws' ++ ws)
2576 Just (Given g) -> go binds' bound_dicts irreds (addToFM done w [g_id]) ws
2579 binds' | w_id == g_id = binds
2580 | otherwise = add_bind (nlHsVar g_id)
2583 done' = addToFM done w [w_id]
2584 add_bind rhs = addInstToDictBind binds w rhs
2588 Note [No superclasses for Stop]
2589 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2590 When we decide not to reduce an Inst -- the 'WhatToDo' --- we still
2591 add it to avails, so that any other equal Insts will be commoned up
2592 right here. However, we do *not* add superclasses. If we have
2595 but a is not bound here, then we *don't* want to derive dn from df
2596 here lest we lose sharing.
2599 addWanted :: WantSCs -> Avails -> Inst -> LHsExpr TcId -> [Inst] -> TcM Avails
2600 addWanted want_scs avails wanted rhs_expr wanteds
2601 = addAvailAndSCs want_scs avails wanted avail
2603 avail = Rhs rhs_expr wanteds
2605 addGiven :: (Inst -> WantSCs) -> Avails -> Inst -> TcM Avails
2606 addGiven want_scs avails given = addAvailAndSCs (want_scs given) avails given (Given given)
2607 -- Conditionally add superclasses for 'givens'
2608 -- See Note [Recursive instances and superclases]
2610 -- No ASSERT( not (given `elemAvails` avails) ) because in an instance
2611 -- decl for Ord t we can add both Ord t and Eq t as 'givens',
2612 -- so the assert isn't true
2616 addIrred :: WantSCs -> Avails -> Inst -> TcM Avails
2617 addIrred want_scs avails irred = ASSERT2( not (irred `elemAvails` avails), ppr irred $$ ppr avails )
2618 addAvailAndSCs want_scs avails irred IsIrred
2620 addAvailAndSCs :: WantSCs -> Avails -> Inst -> AvailHow -> TcM Avails
2621 addAvailAndSCs want_scs avails inst avail
2622 | not (isClassDict inst) = extendAvails avails inst avail
2623 | NoSCs <- want_scs = extendAvails avails inst avail
2624 | otherwise = do { traceTc (text "addAvailAndSCs" <+> vcat [ppr inst, ppr deps])
2625 ; avails' <- extendAvails avails inst avail
2626 ; addSCs is_loop avails' inst }
2628 is_loop pred = any (`tcEqType` mkPredTy pred) dep_tys
2629 -- Note: this compares by *type*, not by Unique
2630 deps = findAllDeps (unitVarSet (instToVar inst)) avail
2631 dep_tys = map idType (varSetElems deps)
2633 findAllDeps :: IdSet -> AvailHow -> IdSet
2634 -- Find all the Insts that this one depends on
2635 -- See Note [SUPERCLASS-LOOP 2]
2636 -- Watch out, though. Since the avails may contain loops
2637 -- (see Note [RECURSIVE DICTIONARIES]), so we need to track the ones we've seen so far
2638 findAllDeps so_far (Rhs _ kids) = foldl find_all so_far kids
2639 findAllDeps so_far _ = so_far
2641 find_all :: IdSet -> Inst -> IdSet
2643 | isEqInst kid = so_far
2644 | kid_id `elemVarSet` so_far = so_far
2645 | Just avail <- findAvail avails kid = findAllDeps so_far' avail
2646 | otherwise = so_far'
2648 so_far' = extendVarSet so_far kid_id -- Add the new kid to so_far
2649 kid_id = instToId kid
2651 addSCs :: (TcPredType -> Bool) -> Avails -> Inst -> TcM Avails
2652 -- Add all the superclasses of the Inst to Avails
2653 -- The first param says "don't do this because the original thing
2654 -- depends on this one, so you'd build a loop"
2655 -- Invariant: the Inst is already in Avails.
2657 addSCs is_loop avails dict
2658 = ASSERT( isDict dict )
2659 do { sc_dicts <- newDictBndrs (instLoc dict) sc_theta'
2660 ; foldlM add_sc avails (zipEqual "add_scs" sc_dicts sc_sels) }
2662 (clas, tys) = getDictClassTys dict
2663 (tyvars, sc_theta, sc_sels, _) = classBigSig clas
2664 sc_theta' = filter (not . isEqPred) $
2665 substTheta (zipTopTvSubst tyvars tys) sc_theta
2667 add_sc avails (sc_dict, sc_sel)
2668 | is_loop (dictPred sc_dict) = return avails -- See Note [SUPERCLASS-LOOP 2]
2669 | is_given sc_dict = return avails
2670 | otherwise = do { avails' <- extendAvails avails sc_dict (Rhs sc_sel_rhs [dict])
2671 ; addSCs is_loop avails' sc_dict }
2673 sc_sel_rhs = L (instSpan dict) (HsWrap co_fn (HsVar sc_sel))
2674 co_fn = WpApp (instToVar dict) <.> mkWpTyApps tys
2676 is_given :: Inst -> Bool
2677 is_given sc_dict = case findAvail avails sc_dict of
2678 Just (Given _) -> True -- Given is cheaper than superclass selection
2681 -- From the a set of insts obtain all equalities that (transitively) occur in
2682 -- superclass contexts of class constraints (aka the ancestor equalities).
2684 ancestorEqualities :: [Inst] -> TcM [Inst]
2686 = mapM mkWantedEqInst -- turn only equality predicates..
2687 . filter isEqPred -- ..into wanted equality insts
2689 . addAEsToBag emptyBag -- collect the superclass constraints..
2690 . map dictPred -- ..of all predicates in a bag
2691 . filter isClassDict
2693 addAEsToBag :: Bag PredType -> [PredType] -> Bag PredType
2694 addAEsToBag bag [] = bag
2695 addAEsToBag bag (pred:preds)
2696 | pred `elemBag` bag = addAEsToBag bag preds
2697 | isEqPred pred = addAEsToBag bagWithPred preds
2698 | isClassPred pred = addAEsToBag bagWithPred predsWithSCs
2699 | otherwise = addAEsToBag bag preds
2701 bagWithPred = bag `snocBag` pred
2702 predsWithSCs = preds ++ substTheta (zipTopTvSubst tyvars tys) sc_theta
2704 (tyvars, sc_theta, _, _) = classBigSig clas
2705 (clas, tys) = getClassPredTys pred
2709 %************************************************************************
2711 \section{tcSimplifyTop: defaulting}
2713 %************************************************************************
2716 @tcSimplifyTop@ is called once per module to simplify all the constant
2717 and ambiguous Insts.
2719 We need to be careful of one case. Suppose we have
2721 instance Num a => Num (Foo a b) where ...
2723 and @tcSimplifyTop@ is given a constraint (Num (Foo x y)). Then it'll simplify
2724 to (Num x), and default x to Int. But what about y??
2726 It's OK: the final zonking stage should zap y to (), which is fine.
2730 tcSimplifyTop, tcSimplifyInteractive :: [Inst] -> TcM TcDictBinds
2731 tcSimplifyTop wanteds
2732 = tc_simplify_top doc False wanteds
2734 doc = text "tcSimplifyTop"
2736 tcSimplifyInteractive wanteds
2737 = tc_simplify_top doc True wanteds
2739 doc = text "tcSimplifyInteractive"
2741 -- The TcLclEnv should be valid here, solely to improve
2742 -- error message generation for the monomorphism restriction
2743 tc_simplify_top :: SDoc -> Bool -> [Inst] -> TcM (Bag (LHsBind TcId))
2744 tc_simplify_top doc interactive wanteds
2745 = do { dflags <- getDOpts
2746 ; wanteds <- zonkInsts wanteds
2747 ; mapM_ zonkTopTyVar (varSetElems (tyVarsOfInsts wanteds))
2749 ; traceTc (text "tc_simplify_top 0: " <+> ppr wanteds)
2750 ; (irreds1, binds1) <- tryHardCheckLoop doc1 wanteds
2751 -- ; (irreds1, binds1) <- gentleInferLoop doc1 wanteds
2752 ; traceTc (text "tc_simplify_top 1: " <+> ppr irreds1)
2753 ; (irreds2, binds2) <- approximateImplications doc2 (\_ -> True) irreds1
2754 ; traceTc (text "tc_simplify_top 2: " <+> ppr irreds2)
2756 -- Use the defaulting rules to do extra unification
2757 -- NB: irreds2 are already zonked
2758 ; (irreds3, binds3) <- disambiguate doc3 interactive dflags irreds2
2760 -- Deal with implicit parameters
2761 ; let (bad_ips, non_ips) = partition isIPDict irreds3
2762 (ambigs, others) = partition isTyVarDict non_ips
2764 ; topIPErrs bad_ips -- Can arise from f :: Int -> Int
2766 ; addNoInstanceErrs others
2767 ; addTopAmbigErrs ambigs
2769 ; return (binds1 `unionBags` binds2 `unionBags` binds3) }
2771 doc1 = doc <+> ptext (sLit "(first round)")
2772 doc2 = doc <+> ptext (sLit "(approximate)")
2773 doc3 = doc <+> ptext (sLit "(disambiguate)")
2776 If a dictionary constrains a type variable which is
2777 * not mentioned in the environment
2778 * and not mentioned in the type of the expression
2779 then it is ambiguous. No further information will arise to instantiate
2780 the type variable; nor will it be generalised and turned into an extra
2781 parameter to a function.
2783 It is an error for this to occur, except that Haskell provided for
2784 certain rules to be applied in the special case of numeric types.
2786 * at least one of its classes is a numeric class, and
2787 * all of its classes are numeric or standard
2788 then the type variable can be defaulted to the first type in the
2789 default-type list which is an instance of all the offending classes.
2791 So here is the function which does the work. It takes the ambiguous
2792 dictionaries and either resolves them (producing bindings) or
2793 complains. It works by splitting the dictionary list by type
2794 variable, and using @disambigOne@ to do the real business.
2796 @disambigOne@ assumes that its arguments dictionaries constrain all
2797 the same type variable.
2799 ADR Comment 20/6/94: I've changed the @CReturnable@ case to default to
2800 @()@ instead of @Int@. I reckon this is the Right Thing to do since
2801 the most common use of defaulting is code like:
2803 _ccall_ foo `seqPrimIO` bar
2805 Since we're not using the result of @foo@, the result if (presumably)
2809 disambiguate :: SDoc -> Bool -> DynFlags -> [Inst] -> TcM ([Inst], TcDictBinds)
2810 -- Just does unification to fix the default types
2811 -- The Insts are assumed to be pre-zonked
2812 disambiguate doc interactive dflags insts
2814 = return (insts, emptyBag)
2816 | null defaultable_groups
2817 = do { traceTc (text "disambigutate, no defaultable groups" <+> vcat [ppr unaries, ppr insts, ppr bad_tvs, ppr defaultable_groups])
2818 ; return (insts, emptyBag) }
2821 = do { -- Figure out what default types to use
2822 default_tys <- getDefaultTys extended_defaulting ovl_strings
2824 ; traceTc (text "disambiguate1" <+> vcat [ppr insts, ppr unaries, ppr bad_tvs, ppr defaultable_groups])
2825 ; mapM_ (disambigGroup default_tys) defaultable_groups
2827 -- disambigGroup does unification, hence try again
2828 ; tryHardCheckLoop doc insts }
2831 extended_defaulting = interactive || dopt Opt_ExtendedDefaultRules dflags
2832 ovl_strings = dopt Opt_OverloadedStrings dflags
2834 unaries :: [(Inst, Class, TcTyVar)] -- (C tv) constraints
2835 bad_tvs :: TcTyVarSet -- Tyvars mentioned by *other* constraints
2836 (unaries, bad_tvs_s) = partitionWith find_unary insts
2837 bad_tvs = unionVarSets bad_tvs_s
2839 -- Finds unary type-class constraints
2840 find_unary d@(Dict {tci_pred = ClassP cls [ty]})
2841 | Just tv <- tcGetTyVar_maybe ty = Left (d,cls,tv)
2842 find_unary inst = Right (tyVarsOfInst inst)
2844 -- Group by type variable
2845 defaultable_groups :: [[(Inst,Class,TcTyVar)]]
2846 defaultable_groups = filter defaultable_group (equivClasses cmp_tv unaries)
2847 cmp_tv (_,_,tv1) (_,_,tv2) = tv1 `compare` tv2
2849 defaultable_group :: [(Inst,Class,TcTyVar)] -> Bool
2850 defaultable_group ds@((_,_,tv):_)
2851 = isTyConableTyVar tv -- Note [Avoiding spurious errors]
2852 && not (tv `elemVarSet` bad_tvs)
2853 && defaultable_classes [c | (_,c,_) <- ds]
2854 defaultable_group [] = panic "defaultable_group"
2856 defaultable_classes clss
2857 | extended_defaulting = any isInteractiveClass clss
2858 | otherwise = all is_std_class clss && (any is_num_class clss)
2860 -- In interactive mode, or with -XExtendedDefaultRules,
2861 -- we default Show a to Show () to avoid graututious errors on "show []"
2862 isInteractiveClass cls
2863 = is_num_class cls || (classKey cls `elem` [showClassKey, eqClassKey, ordClassKey])
2865 is_num_class cls = isNumericClass cls || (ovl_strings && (cls `hasKey` isStringClassKey))
2866 -- is_num_class adds IsString to the standard numeric classes,
2867 -- when -foverloaded-strings is enabled
2869 is_std_class cls = isStandardClass cls || (ovl_strings && (cls `hasKey` isStringClassKey))
2870 -- Similarly is_std_class
2872 -----------------------
2873 disambigGroup :: [Type] -- The default types
2874 -> [(Inst,Class,TcTyVar)] -- All standard classes of form (C a)
2875 -> TcM () -- Just does unification, to fix the default types
2877 disambigGroup default_tys dicts
2878 = try_default default_tys
2880 (_,_,tyvar) = ASSERT(not (null dicts)) head dicts -- Should be non-empty
2881 classes = [c | (_,c,_) <- dicts]
2883 try_default [] = return ()
2884 try_default (default_ty : default_tys)
2885 = tryTcLIE_ (try_default default_tys) $
2886 do { tcSimplifyDefault [mkClassPred clas [default_ty] | clas <- classes]
2887 -- This may fail; then the tryTcLIE_ kicks in
2888 -- Failure here is caused by there being no type in the
2889 -- default list which can satisfy all the ambiguous classes.
2890 -- For example, if Real a is reqd, but the only type in the
2891 -- default list is Int.
2893 -- After this we can't fail
2894 ; warnDefault dicts default_ty
2895 ; unifyType default_ty (mkTyVarTy tyvar)
2896 ; return () -- TOMDO: do something with the coercion
2900 -----------------------
2901 getDefaultTys :: Bool -> Bool -> TcM [Type]
2902 getDefaultTys extended_deflts ovl_strings
2903 = do { mb_defaults <- getDeclaredDefaultTys
2904 ; case mb_defaults of {
2905 Just tys -> return tys ; -- User-supplied defaults
2908 -- No use-supplied default
2909 -- Use [Integer, Double], plus modifications
2910 { integer_ty <- tcMetaTy integerTyConName
2911 ; checkWiredInTyCon doubleTyCon
2912 ; string_ty <- tcMetaTy stringTyConName
2913 ; return (opt_deflt extended_deflts unitTy
2914 -- Note [Default unitTy]
2916 [integer_ty,doubleTy]
2918 opt_deflt ovl_strings string_ty) } } }
2920 opt_deflt True ty = [ty]
2921 opt_deflt False _ = []
2924 Note [Default unitTy]
2925 ~~~~~~~~~~~~~~~~~~~~~
2926 In interative mode (or with -XExtendedDefaultRules) we add () as the first type we
2927 try when defaulting. This has very little real impact, except in the following case.
2929 Text.Printf.printf "hello"
2930 This has type (forall a. IO a); it prints "hello", and returns 'undefined'. We don't
2931 want the GHCi repl loop to try to print that 'undefined'. The neatest thing is to
2932 default the 'a' to (), rather than to Integer (which is what would otherwise happen;
2933 and then GHCi doesn't attempt to print the (). So in interactive mode, we add
2934 () to the list of defaulting types. See Trac #1200.
2936 Note [Avoiding spurious errors]
2937 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2938 When doing the unification for defaulting, we check for skolem
2939 type variables, and simply don't default them. For example:
2940 f = (*) -- Monomorphic
2941 g :: Num a => a -> a
2943 Here, we get a complaint when checking the type signature for g,
2944 that g isn't polymorphic enough; but then we get another one when
2945 dealing with the (Num a) context arising from f's definition;
2946 we try to unify a with Int (to default it), but find that it's
2947 already been unified with the rigid variable from g's type sig
2950 %************************************************************************
2952 \subsection[simple]{@Simple@ versions}
2954 %************************************************************************
2956 Much simpler versions when there are no bindings to make!
2958 @tcSimplifyThetas@ simplifies class-type constraints formed by
2959 @deriving@ declarations and when specialising instances. We are
2960 only interested in the simplified bunch of class/type constraints.
2962 It simplifies to constraints of the form (C a b c) where
2963 a,b,c are type variables. This is required for the context of
2964 instance declarations.
2967 tcSimplifyDeriv :: InstOrigin
2969 -> ThetaType -- Wanted
2970 -> TcM ThetaType -- Needed
2971 -- Given instance (wanted) => C inst_ty
2972 -- Simplify 'wanted' as much as possible
2974 tcSimplifyDeriv orig tyvars theta
2975 = do { (tvs, _, tenv) <- tcInstTyVars tyvars
2976 -- The main loop may do unification, and that may crash if
2977 -- it doesn't see a TcTyVar, so we have to instantiate. Sigh
2978 -- ToDo: what if two of them do get unified?
2979 ; wanteds <- newDictBndrsO orig (substTheta tenv theta)
2980 ; (irreds, _) <- tryHardCheckLoop doc wanteds
2982 ; let (tv_dicts, others) = partition ok irreds
2983 ; addNoInstanceErrs others
2984 -- See Note [Exotic derived instance contexts] in TcMType
2986 ; let rev_env = zipTopTvSubst tvs (mkTyVarTys tyvars)
2987 simpl_theta = substTheta rev_env (map dictPred tv_dicts)
2988 -- This reverse-mapping is a pain, but the result
2989 -- should mention the original TyVars not TcTyVars
2991 ; return simpl_theta }
2993 doc = ptext (sLit "deriving classes for a data type")
2995 ok dict | isDict dict = validDerivPred (dictPred dict)
3000 @tcSimplifyDefault@ just checks class-type constraints, essentially;
3001 used with \tr{default} declarations. We are only interested in
3002 whether it worked or not.
3005 tcSimplifyDefault :: ThetaType -- Wanted; has no type variables in it
3008 tcSimplifyDefault theta = do
3009 wanteds <- newDictBndrsO DefaultOrigin theta
3010 (irreds, _) <- tryHardCheckLoop doc wanteds
3011 addNoInstanceErrs irreds
3015 traceTc (ptext (sLit "tcSimplifyDefault failing")) >> failM
3017 doc = ptext (sLit "default declaration")
3020 @tcSimplifyStagedExpr@ performs a simplification but does so at a new
3021 stage. This is used when typechecking annotations and splices.
3025 tcSimplifyStagedExpr :: ThStage -> TcM a -> TcM (a, TcDictBinds)
3026 -- Type check an expression that runs at a top level stage as if
3027 -- it were going to be spliced and then simplify it
3028 tcSimplifyStagedExpr stage tc_action
3029 = setStage stage $ do {
3030 -- Typecheck the expression
3031 (thing', lie) <- getLIE tc_action
3033 -- Solve the constraints
3034 ; const_binds <- tcSimplifyTop lie
3036 ; return (thing', const_binds) }
3041 %************************************************************************
3043 \section{Errors and contexts}
3045 %************************************************************************
3047 ToDo: for these error messages, should we note the location as coming
3048 from the insts, or just whatever seems to be around in the monad just
3052 groupErrs :: ([Inst] -> TcM ()) -- Deal with one group
3053 -> [Inst] -- The offending Insts
3055 -- Group together insts with the same origin
3056 -- We want to report them together in error messages
3060 groupErrs report_err (inst:insts)
3061 = do { do_one (inst:friends)
3062 ; groupErrs report_err others }
3064 -- (It may seem a bit crude to compare the error messages,
3065 -- but it makes sure that we combine just what the user sees,
3066 -- and it avoids need equality on InstLocs.)
3067 (friends, others) = partition is_friend insts
3068 loc_msg = showSDoc (pprInstLoc (instLoc inst))
3069 is_friend friend = showSDoc (pprInstLoc (instLoc friend)) == loc_msg
3070 do_one insts = addInstCtxt (instLoc (head insts)) (report_err insts)
3071 -- Add location and context information derived from the Insts
3073 -- Add the "arising from..." part to a message about bunch of dicts
3074 addInstLoc :: [Inst] -> Message -> Message
3075 addInstLoc insts msg = msg $$ nest 2 (pprInstArising (head insts))
3077 addTopIPErrs :: [Name] -> [Inst] -> TcM ()
3080 addTopIPErrs bndrs ips
3081 = do { dflags <- getDOpts
3082 ; addErrTcM (tidy_env, mk_msg dflags tidy_ips) }
3084 (tidy_env, tidy_ips) = tidyInsts ips
3086 = vcat [sep [ptext (sLit "Implicit parameters escape from"),
3087 nest 2 (ptext (sLit "the monomorphic top-level binding")
3088 <> plural bndrs <+> ptext (sLit "of")
3089 <+> pprBinders bndrs <> colon)],
3090 nest 2 (vcat (map ppr_ip ips)),
3091 monomorphism_fix dflags]
3092 ppr_ip ip = pprPred (dictPred ip) <+> pprInstArising ip
3094 topIPErrs :: [Inst] -> TcM ()
3096 = groupErrs report tidy_dicts
3098 (tidy_env, tidy_dicts) = tidyInsts dicts
3099 report dicts = addErrTcM (tidy_env, mk_msg dicts)
3100 mk_msg dicts = addInstLoc dicts (ptext (sLit "Unbound implicit parameter") <>
3101 plural tidy_dicts <+> pprDictsTheta tidy_dicts)
3103 addNoInstanceErrs :: [Inst] -- Wanted (can include implications)
3105 addNoInstanceErrs insts
3106 = do { let (tidy_env, tidy_insts) = tidyInsts insts
3107 ; reportNoInstances tidy_env Nothing tidy_insts }
3111 -> Maybe (InstLoc, [Inst]) -- Context
3112 -- Nothing => top level
3113 -- Just (d,g) => d describes the construct
3115 -> [Inst] -- What is wanted (can include implications)
3118 reportNoInstances tidy_env mb_what insts
3119 = groupErrs (report_no_instances tidy_env mb_what) insts
3121 report_no_instances :: TidyEnv -> Maybe (InstLoc, [Inst]) -> [Inst] -> TcM ()
3122 report_no_instances tidy_env mb_what insts
3123 = do { inst_envs <- tcGetInstEnvs
3124 ; let (implics, insts1) = partition isImplicInst insts
3125 (insts2, overlaps) = partitionWith (check_overlap inst_envs) insts1
3126 (eqInsts, insts3) = partition isEqInst insts2
3127 ; traceTc (text "reportNoInstances" <+> vcat
3128 [ppr insts, ppr implics, ppr insts1, ppr insts2])
3129 ; mapM_ complain_implic implics
3130 ; mapM_ (\doc -> addErrTcM (tidy_env, doc)) overlaps
3131 ; groupErrs complain_no_inst insts3
3132 ; mapM_ (addErrTcM . mk_eq_err) eqInsts
3135 complain_no_inst insts = addErrTcM (tidy_env, mk_no_inst_err insts)
3137 complain_implic inst -- Recurse!
3138 = reportNoInstances tidy_env
3139 (Just (tci_loc inst, tci_given inst))
3142 check_overlap :: (InstEnv,InstEnv) -> Inst -> Either Inst SDoc
3143 -- Right msg => overlap message
3144 -- Left inst => no instance
3145 check_overlap inst_envs wanted
3146 | not (isClassDict wanted) = Left wanted
3148 = case lookupInstEnv inst_envs clas tys of
3149 ([], _) -> Left wanted -- No match
3150 -- The case of exactly one match and no unifiers means a
3151 -- successful lookup. That can't happen here, because dicts
3152 -- only end up here if they didn't match in Inst.lookupInst
3154 | debugIsOn -> pprPanic "reportNoInstance" (ppr wanted)
3155 res -> Right (mk_overlap_msg wanted res)
3157 (clas,tys) = getDictClassTys wanted
3159 mk_overlap_msg dict (matches, unifiers)
3160 = ASSERT( not (null matches) )
3161 vcat [ addInstLoc [dict] ((ptext (sLit "Overlapping instances for")
3162 <+> pprPred (dictPred dict))),
3163 sep [ptext (sLit "Matching instances") <> colon,
3164 nest 2 (vcat [pprInstances ispecs, pprInstances unifiers])],
3165 if not (isSingleton matches)
3166 then -- Two or more matches
3168 else -- One match, plus some unifiers
3169 ASSERT( not (null unifiers) )
3170 parens (vcat [ptext (sLit "The choice depends on the instantiation of") <+>
3171 quotes (pprWithCommas ppr (varSetElems (tyVarsOfInst dict))),
3172 ptext (sLit "To pick the first instance above, use -XIncoherentInstances"),
3173 ptext (sLit "when compiling the other instance declarations")])]
3175 ispecs = [ispec | (ispec, _) <- matches]
3177 mk_eq_err :: Inst -> (TidyEnv, SDoc)
3178 mk_eq_err inst = misMatchMsg tidy_env (eqInstTys inst)
3180 mk_no_inst_err insts
3181 | null insts = empty
3183 | Just (loc, givens) <- mb_what, -- Nested (type signatures, instance decls)
3184 not (isEmptyVarSet (tyVarsOfInsts insts))
3185 = vcat [ addInstLoc insts $
3186 sep [ ptext (sLit "Could not deduce") <+> pprDictsTheta insts
3187 , nest 2 $ ptext (sLit "from the context") <+> pprDictsTheta givens]
3188 , show_fixes (fix1 loc : fixes2) ]
3190 | otherwise -- Top level
3191 = vcat [ addInstLoc insts $
3192 ptext (sLit "No instance") <> plural insts
3193 <+> ptext (sLit "for") <+> pprDictsTheta insts
3194 , show_fixes fixes2 ]
3197 fix1 loc = sep [ ptext (sLit "add") <+> pprDictsTheta insts
3198 <+> ptext (sLit "to the context of"),
3199 nest 2 (ppr (instLocOrigin loc)) ]
3200 -- I'm not sure it helps to add the location
3201 -- nest 2 (ptext (sLit "at") <+> ppr (instLocSpan loc)) ]
3203 fixes2 | null instance_dicts = []
3204 | otherwise = [sep [ptext (sLit "add an instance declaration for"),
3205 pprDictsTheta instance_dicts]]
3206 instance_dicts = [d | d <- insts, isClassDict d, not (isTyVarDict d)]
3207 -- Insts for which it is worth suggesting an adding an instance declaration
3208 -- Exclude implicit parameters, and tyvar dicts
3210 show_fixes :: [SDoc] -> SDoc
3211 show_fixes [] = empty
3212 show_fixes (f:fs) = sep [ptext (sLit "Possible fix:"),
3213 nest 2 (vcat (f : map (ptext (sLit "or") <+>) fs))]
3215 addTopAmbigErrs :: [Inst] -> TcRn ()
3216 addTopAmbigErrs dicts
3217 -- Divide into groups that share a common set of ambiguous tyvars
3218 = ifErrsM (return ()) $ -- Only report ambiguity if no other errors happened
3219 -- See Note [Avoiding spurious errors]
3220 mapM_ report (equivClasses cmp [(d, tvs_of d) | d <- tidy_dicts])
3222 (tidy_env, tidy_dicts) = tidyInsts dicts
3224 tvs_of :: Inst -> [TcTyVar]
3225 tvs_of d = varSetElems (tyVarsOfInst d)
3226 cmp (_,tvs1) (_,tvs2) = tvs1 `compare` tvs2
3228 report :: [(Inst,[TcTyVar])] -> TcM ()
3229 report pairs@((inst,tvs) : _) = do -- The pairs share a common set of ambiguous tyvars
3230 (tidy_env, mono_msg) <- mkMonomorphismMsg tidy_env tvs
3231 setSrcSpan (instSpan inst) $
3232 -- the location of the first one will do for the err message
3233 addErrTcM (tidy_env, msg $$ mono_msg)
3235 dicts = map fst pairs
3236 msg = sep [text "Ambiguous type variable" <> plural tvs <+>
3237 pprQuotedList tvs <+> in_msg,
3238 nest 2 (pprDictsInFull dicts)]
3239 in_msg = text "in the constraint" <> plural dicts <> colon
3240 report [] = panic "addTopAmbigErrs"
3243 mkMonomorphismMsg :: TidyEnv -> [TcTyVar] -> TcM (TidyEnv, Message)
3244 -- There's an error with these Insts; if they have free type variables
3245 -- it's probably caused by the monomorphism restriction.
3246 -- Try to identify the offending variable
3247 -- ASSUMPTION: the Insts are fully zonked
3248 mkMonomorphismMsg tidy_env inst_tvs
3249 = do { dflags <- getDOpts
3250 ; (tidy_env, docs) <- findGlobals (mkVarSet inst_tvs) tidy_env
3251 ; return (tidy_env, mk_msg dflags docs) }
3253 mk_msg _ _ | any isRuntimeUnk inst_tvs
3254 = vcat [ptext (sLit "Cannot resolve unknown runtime types:") <+>
3255 (pprWithCommas ppr inst_tvs),
3256 ptext (sLit "Use :print or :force to determine these types")]
3257 mk_msg _ [] = ptext (sLit "Probable fix: add a type signature that fixes these type variable(s)")
3258 -- This happens in things like
3259 -- f x = show (read "foo")
3260 -- where monomorphism doesn't play any role
3262 = vcat [ptext (sLit "Possible cause: the monomorphism restriction applied to the following:"),
3264 monomorphism_fix dflags]
3266 monomorphism_fix :: DynFlags -> SDoc
3267 monomorphism_fix dflags
3268 = ptext (sLit "Probable fix:") <+> vcat
3269 [ptext (sLit "give these definition(s) an explicit type signature"),
3270 if dopt Opt_MonomorphismRestriction dflags
3271 then ptext (sLit "or use -XNoMonomorphismRestriction")
3272 else empty] -- Only suggest adding "-XNoMonomorphismRestriction"
3273 -- if it is not already set!
3275 warnDefault :: [(Inst, Class, Var)] -> Type -> TcM ()
3276 warnDefault ups default_ty = do
3277 warn_flag <- doptM Opt_WarnTypeDefaults
3278 addInstCtxt (instLoc (head (dicts))) (warnTc warn_flag warn_msg)
3280 dicts = [d | (d,_,_) <- ups]
3283 (_, tidy_dicts) = tidyInsts dicts
3284 warn_msg = vcat [ptext (sLit "Defaulting the following constraint(s) to type") <+>
3285 quotes (ppr default_ty),
3286 pprDictsInFull tidy_dicts]
3288 reduceDepthErr :: Int -> [Inst] -> SDoc
3289 reduceDepthErr n stack
3290 = vcat [ptext (sLit "Context reduction stack overflow; size =") <+> int n,
3291 ptext (sLit "Use -fcontext-stack=N to increase stack size to N"),
3292 nest 4 (pprStack stack)]
3294 pprStack :: [Inst] -> SDoc
3295 pprStack stack = vcat (map pprInstInFull stack)