2 % (c) The University of Glasgow 2006
3 % (c) The GRASP/AQUA Project, Glasgow University, 1992-1998
10 -- The above warning supression flag is a temporary kludge.
11 -- While working on this module you are encouraged to remove it and fix
12 -- any warnings in the module. See
13 -- http://hackage.haskell.org/trac/ghc/wiki/Commentary/CodingStyle#Warnings
17 tcSimplifyInfer, tcSimplifyInferCheck,
18 tcSimplifyCheck, tcSimplifyRestricted,
19 tcSimplifyRuleLhs, tcSimplifyIPs,
20 tcSimplifySuperClasses,
21 tcSimplifyTop, tcSimplifyInteractive,
22 tcSimplifyBracket, tcSimplifyCheckPat,
24 tcSimplifyDeriv, tcSimplifyDefault,
30 #include "HsVersions.h"
32 import {-# SOURCE #-} TcUnify( unifyType )
73 %************************************************************************
77 %************************************************************************
79 --------------------------------------
80 Notes on functional dependencies (a bug)
81 --------------------------------------
88 instance D a b => C a b -- Undecidable
89 -- (Not sure if it's crucial to this eg)
90 f :: C a b => a -> Bool
93 g :: C a b => a -> Bool
96 Here f typechecks, but g does not!! Reason: before doing improvement,
97 we reduce the (C a b1) constraint from the call of f to (D a b1).
99 Here is a more complicated example:
101 | > class Foo a b | a->b
103 | > class Bar a b | a->b
107 | > instance Bar Obj Obj
109 | > instance (Bar a b) => Foo a b
111 | > foo:: (Foo a b) => a -> String
114 | > runFoo:: (forall a b. (Foo a b) => a -> w) -> w
120 | Could not deduce (Bar a b) from the context (Foo a b)
121 | arising from use of `foo' at <interactive>:1
123 | Add (Bar a b) to the expected type of an expression
124 | In the first argument of `runFoo', namely `foo'
125 | In the definition of `it': it = runFoo foo
127 | Why all of the sudden does GHC need the constraint Bar a b? The
128 | function foo didn't ask for that...
130 The trouble is that to type (runFoo foo), GHC has to solve the problem:
132 Given constraint Foo a b
133 Solve constraint Foo a b'
135 Notice that b and b' aren't the same. To solve this, just do
136 improvement and then they are the same. But GHC currently does
141 That is usually fine, but it isn't here, because it sees that Foo a b is
142 not the same as Foo a b', and so instead applies the instance decl for
143 instance Bar a b => Foo a b. And that's where the Bar constraint comes
146 The Right Thing is to improve whenever the constraint set changes at
147 all. Not hard in principle, but it'll take a bit of fiddling to do.
149 Note [Choosing which variables to quantify]
150 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
151 Suppose we are about to do a generalisation step. We have in our hand
154 T the type of the RHS
155 C the constraints from that RHS
157 The game is to figure out
159 Q the set of type variables over which to quantify
160 Ct the constraints we will *not* quantify over
161 Cq the constraints we will quantify over
163 So we're going to infer the type
167 and float the constraints Ct further outwards.
169 Here are the things that *must* be true:
171 (A) Q intersect fv(G) = EMPTY limits how big Q can be
172 (B) Q superset fv(Cq union T) \ oclose(fv(G),C) limits how small Q can be
174 (A) says we can't quantify over a variable that's free in the environment.
175 (B) says we must quantify over all the truly free variables in T, else
176 we won't get a sufficiently general type.
178 We do not *need* to quantify over any variable that is fixed by the
179 free vars of the environment G.
181 BETWEEN THESE TWO BOUNDS, ANY Q WILL DO!
183 Example: class H x y | x->y where ...
185 fv(G) = {a} C = {H a b, H c d}
188 (A) Q intersect {a} is empty
189 (B) Q superset {a,b,c,d} \ oclose({a}, C) = {a,b,c,d} \ {a,b} = {c,d}
191 So Q can be {c,d}, {b,c,d}
193 In particular, it's perfectly OK to quantify over more type variables
194 than strictly necessary; there is no need to quantify over 'b', since
195 it is determined by 'a' which is free in the envt, but it's perfectly
196 OK to do so. However we must not quantify over 'a' itself.
198 Other things being equal, however, we'd like to quantify over as few
199 variables as possible: smaller types, fewer type applications, more
200 constraints can get into Ct instead of Cq. Here's a good way to
203 Q = grow( fv(T), C ) \ oclose( fv(G), C )
205 That is, quantify over all variable that that MIGHT be fixed by the
206 call site (which influences T), but which aren't DEFINITELY fixed by
207 G. This choice definitely quantifies over enough type variables,
208 albeit perhaps too many.
210 Why grow( fv(T), C ) rather than fv(T)? Consider
212 class H x y | x->y where ...
217 If we used fv(T) = {c} we'd get the type
219 forall c. H c d => c -> b
221 And then if the fn was called at several different c's, each of
222 which fixed d differently, we'd get a unification error, because
223 d isn't quantified. Solution: quantify d. So we must quantify
224 everything that might be influenced by c.
226 Why not oclose( fv(T), C )? Because we might not be able to see
227 all the functional dependencies yet:
229 class H x y | x->y where ...
230 instance H x y => Eq (T x y) where ...
235 Now oclose(fv(T),C) = {c}, because the functional dependency isn't
236 apparent yet, and that's wrong. We must really quantify over d too.
238 There really isn't any point in quantifying over any more than
239 grow( fv(T), C ), because the call sites can't possibly influence
240 any other type variables.
244 -------------------------------------
246 -------------------------------------
248 It's very hard to be certain when a type is ambiguous. Consider
252 instance H x y => K (x,y)
254 Is this type ambiguous?
255 forall a b. (K (a,b), Eq b) => a -> a
257 Looks like it! But if we simplify (K (a,b)) we get (H a b) and
258 now we see that a fixes b. So we can't tell about ambiguity for sure
259 without doing a full simplification. And even that isn't possible if
260 the context has some free vars that may get unified. Urgle!
262 Here's another example: is this ambiguous?
263 forall a b. Eq (T b) => a -> a
264 Not if there's an insance decl (with no context)
265 instance Eq (T b) where ...
267 You may say of this example that we should use the instance decl right
268 away, but you can't always do that:
270 class J a b where ...
271 instance J Int b where ...
273 f :: forall a b. J a b => a -> a
275 (Notice: no functional dependency in J's class decl.)
276 Here f's type is perfectly fine, provided f is only called at Int.
277 It's premature to complain when meeting f's signature, or even
278 when inferring a type for f.
282 However, we don't *need* to report ambiguity right away. It'll always
283 show up at the call site.... and eventually at main, which needs special
284 treatment. Nevertheless, reporting ambiguity promptly is an excellent thing.
286 So here's the plan. We WARN about probable ambiguity if
288 fv(Cq) is not a subset of oclose(fv(T) union fv(G), C)
290 (all tested before quantification).
291 That is, all the type variables in Cq must be fixed by the the variables
292 in the environment, or by the variables in the type.
294 Notice that we union before calling oclose. Here's an example:
296 class J a b c | a b -> c
300 forall b c. (J a b c) => b -> b
302 Only if we union {a} from G with {b} from T before using oclose,
303 do we see that c is fixed.
305 It's a bit vague exactly which C we should use for this oclose call. If we
306 don't fix enough variables we might complain when we shouldn't (see
307 the above nasty example). Nothing will be perfect. That's why we can
308 only issue a warning.
311 Can we ever be *certain* about ambiguity? Yes: if there's a constraint
313 c in C such that fv(c) intersect (fv(G) union fv(T)) = EMPTY
315 then c is a "bubble"; there's no way it can ever improve, and it's
316 certainly ambiguous. UNLESS it is a constant (sigh). And what about
321 instance H x y => K (x,y)
323 Is this type ambiguous?
324 forall a b. (K (a,b), Eq b) => a -> a
326 Urk. The (Eq b) looks "definitely ambiguous" but it isn't. What we are after
327 is a "bubble" that's a set of constraints
329 Cq = Ca union Cq' st fv(Ca) intersect (fv(Cq') union fv(T) union fv(G)) = EMPTY
331 Hence another idea. To decide Q start with fv(T) and grow it
332 by transitive closure in Cq (no functional dependencies involved).
333 Now partition Cq using Q, leaving the definitely-ambiguous and probably-ok.
334 The definitely-ambiguous can then float out, and get smashed at top level
335 (which squashes out the constants, like Eq (T a) above)
338 --------------------------------------
339 Notes on principal types
340 --------------------------------------
345 f x = let g y = op (y::Int) in True
347 Here the principal type of f is (forall a. a->a)
348 but we'll produce the non-principal type
349 f :: forall a. C Int => a -> a
352 --------------------------------------
353 The need for forall's in constraints
354 --------------------------------------
356 [Exchange on Haskell Cafe 5/6 Dec 2000]
358 class C t where op :: t -> Bool
359 instance C [t] where op x = True
361 p y = (let f :: c -> Bool; f x = op (y >> return x) in f, y ++ [])
362 q y = (y ++ [], let f :: c -> Bool; f x = op (y >> return x) in f)
364 The definitions of p and q differ only in the order of the components in
365 the pair on their right-hand sides. And yet:
367 ghc and "Typing Haskell in Haskell" reject p, but accept q;
368 Hugs rejects q, but accepts p;
369 hbc rejects both p and q;
370 nhc98 ... (Malcolm, can you fill in the blank for us!).
372 The type signature for f forces context reduction to take place, and
373 the results of this depend on whether or not the type of y is known,
374 which in turn depends on which component of the pair the type checker
377 Solution: if y::m a, float out the constraints
378 Monad m, forall c. C (m c)
379 When m is later unified with [], we can solve both constraints.
382 --------------------------------------
383 Notes on implicit parameters
384 --------------------------------------
386 Note [Inheriting implicit parameters]
387 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
392 where f is *not* a top-level binding.
393 From the RHS of f we'll get the constraint (?y::Int).
394 There are two types we might infer for f:
398 (so we get ?y from the context of f's definition), or
400 f :: (?y::Int) => Int -> Int
402 At first you might think the first was better, becuase then
403 ?y behaves like a free variable of the definition, rather than
404 having to be passed at each call site. But of course, the WHOLE
405 IDEA is that ?y should be passed at each call site (that's what
406 dynamic binding means) so we'd better infer the second.
408 BOTTOM LINE: when *inferring types* you *must* quantify
409 over implicit parameters. See the predicate isFreeWhenInferring.
412 Note [Implicit parameters and ambiguity]
413 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
414 Only a *class* predicate can give rise to ambiguity
415 An *implicit parameter* cannot. For example:
416 foo :: (?x :: [a]) => Int
418 is fine. The call site will suppply a particular 'x'
420 Furthermore, the type variables fixed by an implicit parameter
421 propagate to the others. E.g.
422 foo :: (Show a, ?x::[a]) => Int
424 The type of foo looks ambiguous. But it isn't, because at a call site
426 let ?x = 5::Int in foo
427 and all is well. In effect, implicit parameters are, well, parameters,
428 so we can take their type variables into account as part of the
429 "tau-tvs" stuff. This is done in the function 'FunDeps.grow'.
432 Question 2: type signatures
433 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
434 BUT WATCH OUT: When you supply a type signature, we can't force you
435 to quantify over implicit parameters. For example:
439 This is perfectly reasonable. We do not want to insist on
441 (?x + 1) :: (?x::Int => Int)
443 That would be silly. Here, the definition site *is* the occurrence site,
444 so the above strictures don't apply. Hence the difference between
445 tcSimplifyCheck (which *does* allow implicit paramters to be inherited)
446 and tcSimplifyCheckBind (which does not).
448 What about when you supply a type signature for a binding?
449 Is it legal to give the following explicit, user type
450 signature to f, thus:
455 At first sight this seems reasonable, but it has the nasty property
456 that adding a type signature changes the dynamic semantics.
459 (let f x = (x::Int) + ?y
460 in (f 3, f 3 with ?y=5)) with ?y = 6
466 in (f 3, f 3 with ?y=5)) with ?y = 6
470 Indeed, simply inlining f (at the Haskell source level) would change the
473 Nevertheless, as Launchbury says (email Oct 01) we can't really give the
474 semantics for a Haskell program without knowing its typing, so if you
475 change the typing you may change the semantics.
477 To make things consistent in all cases where we are *checking* against
478 a supplied signature (as opposed to inferring a type), we adopt the
481 a signature does not need to quantify over implicit params.
483 [This represents a (rather marginal) change of policy since GHC 5.02,
484 which *required* an explicit signature to quantify over all implicit
485 params for the reasons mentioned above.]
487 But that raises a new question. Consider
489 Given (signature) ?x::Int
490 Wanted (inferred) ?x::Int, ?y::Bool
492 Clearly we want to discharge the ?x and float the ?y out. But
493 what is the criterion that distinguishes them? Clearly it isn't
494 what free type variables they have. The Right Thing seems to be
495 to float a constraint that
496 neither mentions any of the quantified type variables
497 nor any of the quantified implicit parameters
499 See the predicate isFreeWhenChecking.
502 Question 3: monomorphism
503 ~~~~~~~~~~~~~~~~~~~~~~~~
504 There's a nasty corner case when the monomorphism restriction bites:
508 The argument above suggests that we *must* generalise
509 over the ?y parameter, to get
510 z :: (?y::Int) => Int,
511 but the monomorphism restriction says that we *must not*, giving
513 Why does the momomorphism restriction say this? Because if you have
515 let z = x + ?y in z+z
517 you might not expect the addition to be done twice --- but it will if
518 we follow the argument of Question 2 and generalise over ?y.
521 Question 4: top level
522 ~~~~~~~~~~~~~~~~~~~~~
523 At the top level, monomorhism makes no sense at all.
526 main = let ?x = 5 in print foo
530 woggle :: (?x :: Int) => Int -> Int
533 We definitely don't want (foo :: Int) with a top-level implicit parameter
534 (?x::Int) becuase there is no way to bind it.
539 (A) Always generalise over implicit parameters
540 Bindings that fall under the monomorphism restriction can't
544 * Inlining remains valid
545 * No unexpected loss of sharing
546 * But simple bindings like
548 will be rejected, unless you add an explicit type signature
549 (to avoid the monomorphism restriction)
550 z :: (?y::Int) => Int
552 This seems unacceptable
554 (B) Monomorphism restriction "wins"
555 Bindings that fall under the monomorphism restriction can't
557 Always generalise over implicit parameters *except* for bindings
558 that fall under the monomorphism restriction
561 * Inlining isn't valid in general
562 * No unexpected loss of sharing
563 * Simple bindings like
565 accepted (get value of ?y from binding site)
567 (C) Always generalise over implicit parameters
568 Bindings that fall under the monomorphism restriction can't
569 be generalised, EXCEPT for implicit parameters
571 * Inlining remains valid
572 * Unexpected loss of sharing (from the extra generalisation)
573 * Simple bindings like
575 accepted (get value of ?y from occurrence sites)
580 None of these choices seems very satisfactory. But at least we should
581 decide which we want to do.
583 It's really not clear what is the Right Thing To Do. If you see
587 would you expect the value of ?y to be got from the *occurrence sites*
588 of 'z', or from the valuue of ?y at the *definition* of 'z'? In the
589 case of function definitions, the answer is clearly the former, but
590 less so in the case of non-fucntion definitions. On the other hand,
591 if we say that we get the value of ?y from the definition site of 'z',
592 then inlining 'z' might change the semantics of the program.
594 Choice (C) really says "the monomorphism restriction doesn't apply
595 to implicit parameters". Which is fine, but remember that every
596 innocent binding 'x = ...' that mentions an implicit parameter in
597 the RHS becomes a *function* of that parameter, called at each
598 use of 'x'. Now, the chances are that there are no intervening 'with'
599 clauses that bind ?y, so a decent compiler should common up all
600 those function calls. So I think I strongly favour (C). Indeed,
601 one could make a similar argument for abolishing the monomorphism
602 restriction altogether.
604 BOTTOM LINE: we choose (B) at present. See tcSimplifyRestricted
608 %************************************************************************
610 \subsection{tcSimplifyInfer}
612 %************************************************************************
614 tcSimplify is called when we *inferring* a type. Here's the overall game plan:
616 1. Compute Q = grow( fvs(T), C )
618 2. Partition C based on Q into Ct and Cq. Notice that ambiguous
619 predicates will end up in Ct; we deal with them at the top level
621 3. Try improvement, using functional dependencies
623 4. If Step 3 did any unification, repeat from step 1
624 (Unification can change the result of 'grow'.)
626 Note: we don't reduce dictionaries in step 2. For example, if we have
627 Eq (a,b), we don't simplify to (Eq a, Eq b). So Q won't be different
628 after step 2. However note that we may therefore quantify over more
629 type variables than we absolutely have to.
631 For the guts, we need a loop, that alternates context reduction and
632 improvement with unification. E.g. Suppose we have
634 class C x y | x->y where ...
636 and tcSimplify is called with:
638 Then improvement unifies a with b, giving
641 If we need to unify anything, we rattle round the whole thing all over
648 -> TcTyVarSet -- fv(T); type vars
650 -> TcM ([TcTyVar], -- Tyvars to quantify (zonked and quantified)
651 [Inst], -- Dict Ids that must be bound here (zonked)
652 TcDictBinds) -- Bindings
653 -- Any free (escaping) Insts are tossed into the environment
658 tcSimplifyInfer doc tau_tvs wanted
659 = do { tau_tvs1 <- zonkTcTyVarsAndFV (varSetElems tau_tvs)
660 ; wanted' <- mapM zonkInst wanted -- Zonk before deciding quantified tyvars
661 ; gbl_tvs <- tcGetGlobalTyVars
662 ; let preds1 = fdPredsOfInsts wanted'
663 gbl_tvs1 = oclose preds1 gbl_tvs
664 qtvs = grow preds1 tau_tvs1 `minusVarSet` gbl_tvs1
665 -- See Note [Choosing which variables to quantify]
667 -- To maximise sharing, remove from consideration any
668 -- constraints that don't mention qtvs at all
669 ; let (free, bound) = partition (isFreeWhenInferring qtvs) wanted'
672 -- To make types simple, reduce as much as possible
673 ; traceTc (text "infer" <+> (ppr preds1 $$ ppr (grow preds1 tau_tvs1) $$ ppr gbl_tvs $$
674 ppr gbl_tvs1 $$ ppr free $$ ppr bound))
675 ; (irreds1, binds1) <- tryHardCheckLoop doc bound
677 -- Note [Inference and implication constraints]
678 ; let want_dict d = tyVarsOfInst d `intersectsVarSet` qtvs
679 ; (irreds2, binds2) <- approximateImplications doc want_dict irreds1
681 -- Now work out all over again which type variables to quantify,
682 -- exactly in the same way as before, but starting from irreds2. Why?
683 -- a) By now improvment may have taken place, and we must *not*
684 -- quantify over any variable free in the environment
685 -- tc137 (function h inside g) is an example
687 -- b) Do not quantify over constraints that *now* do not
688 -- mention quantified type variables, because they are
689 -- simply ambiguous (or might be bound further out). Example:
690 -- f :: Eq b => a -> (a, b)
692 -- From the RHS of g we get the MethodInst f77 :: alpha -> (alpha, beta)
693 -- We decide to quantify over 'alpha' alone, but free1 does not include f77
694 -- because f77 mentions 'alpha'. Then reducing leaves only the (ambiguous)
695 -- constraint (Eq beta), which we dump back into the free set
696 -- See test tcfail181
698 -- c) irreds may contain type variables not previously mentioned,
699 -- e.g. instance D a x => Foo [a]
701 -- Then after simplifying we'll get (D a x), and x is fresh
702 -- We must quantify over x else it'll be totally unbound
703 ; tau_tvs2 <- zonkTcTyVarsAndFV (varSetElems tau_tvs1)
704 ; gbl_tvs2 <- zonkTcTyVarsAndFV (varSetElems gbl_tvs1)
705 -- Note that we start from gbl_tvs1
706 -- We use tcGetGlobalTyVars, then oclose wrt preds2, because
707 -- we've already put some of the original preds1 into frees
708 -- E.g. wanteds = C a b (where a->b)
711 -- Then b is fixed by gbl_tvs, so (C a b) will be in free, and
712 -- irreds2 will be empty. But we don't want to generalise over b!
713 ; let preds2 = fdPredsOfInsts irreds2 -- irreds2 is zonked
714 qtvs = grow preds2 tau_tvs2 `minusVarSet` oclose preds2 gbl_tvs2
715 ; let (free, irreds3) = partition (isFreeWhenInferring qtvs) irreds2
718 -- Turn the quantified meta-type variables into real type variables
719 ; qtvs2 <- zonkQuantifiedTyVars (varSetElems qtvs)
721 -- We can't abstract over any remaining unsolved
722 -- implications so instead just float them outwards. Ugh.
723 ; let (q_dicts0, implics) = partition isAbstractableInst irreds3
724 ; loc <- getInstLoc (ImplicOrigin doc)
725 ; implic_bind <- bindIrreds loc qtvs2 q_dicts0 implics
727 -- Prepare equality instances for quantification
728 ; let (q_eqs0,q_dicts) = partition isEqInst q_dicts0
729 ; q_eqs <- mapM finalizeEqInst q_eqs0
731 ; return (qtvs2, q_eqs ++ q_dicts, binds1 `unionBags` binds2 `unionBags` implic_bind) }
732 -- NB: when we are done, we might have some bindings, but
733 -- the final qtvs might be empty. See Note [NO TYVARS] below.
735 approximateImplications :: SDoc -> (Inst -> Bool) -> [Inst] -> TcM ([Inst], TcDictBinds)
736 -- Note [Inference and implication constraints]
737 -- Given a bunch of Dict and ImplicInsts, try to approximate the implications by
738 -- - fetching any dicts inside them that are free
739 -- - using those dicts as cruder constraints, to solve the implications
740 -- - returning the extra ones too
742 approximateImplications doc want_dict irreds
744 = return (irreds, emptyBag)
746 = do { extra_dicts' <- mapM cloneDict extra_dicts
747 ; tryHardCheckLoop doc (extra_dicts' ++ irreds) }
748 -- By adding extra_dicts', we make them
749 -- available to solve the implication constraints
751 extra_dicts = get_dicts (filter isImplicInst irreds)
753 get_dicts :: [Inst] -> [Inst] -- Returns only Dicts
754 -- Find the wanted constraints in implication constraints that satisfy
755 -- want_dict, and are not bound by forall's in the constraint itself
756 get_dicts ds = concatMap get_dict ds
758 get_dict d@(Dict {}) | want_dict d = [d]
760 get_dict (ImplicInst {tci_tyvars = tvs, tci_wanted = wanteds})
761 = [ d | let tv_set = mkVarSet tvs
762 , d <- get_dicts wanteds
763 , not (tyVarsOfInst d `intersectsVarSet` tv_set)]
764 get_dict i@(EqInst {}) | want_dict i = [i]
766 get_dict other = pprPanic "approximateImplications" (ppr other)
769 Note [Inference and implication constraints]
770 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
771 Suppose we have a wanted implication constraint (perhaps arising from
772 a nested pattern match) like
774 and we are now trying to quantify over 'a' when inferring the type for
775 a function. In principle it's possible that there might be an instance
776 instance (C a, E a) => D [a]
777 so the context (E a) would suffice. The Right Thing is to abstract over
778 the implication constraint, but we don't do that (a) because it'll be
779 surprising to programmers and (b) because we don't have the machinery to deal
780 with 'given' implications.
782 So our best approximation is to make (D [a]) part of the inferred
783 context, so we can use that to discharge the implication. Hence
784 the strange function get_dicts in approximateImplications.
786 The common cases are more clear-cut, when we have things like
788 Here, abstracting over (C b) is not an approximation at all -- but see
789 Note [Freeness and implications].
791 See Trac #1430 and test tc228.
795 -----------------------------------------------------------
796 -- tcSimplifyInferCheck is used when we know the constraints we are to simplify
797 -- against, but we don't know the type variables over which we are going to quantify.
798 -- This happens when we have a type signature for a mutually recursive group
801 -> TcTyVarSet -- fv(T)
804 -> TcM ([TyVar], -- Fully zonked, and quantified
805 TcDictBinds) -- Bindings
807 tcSimplifyInferCheck loc tau_tvs givens wanteds
808 = do { traceTc (text "tcSimplifyInferCheck <-" <+> ppr wanteds)
809 ; (irreds, binds) <- gentleCheckLoop loc givens wanteds
811 -- Figure out which type variables to quantify over
812 -- You might think it should just be the signature tyvars,
813 -- but in bizarre cases you can get extra ones
814 -- f :: forall a. Num a => a -> a
815 -- f x = fst (g (x, head [])) + 1
817 -- Here we infer g :: forall a b. a -> b -> (b,a)
818 -- We don't want g to be monomorphic in b just because
819 -- f isn't quantified over b.
820 ; let all_tvs = varSetElems (tau_tvs `unionVarSet` tyVarsOfInsts givens)
821 ; all_tvs <- zonkTcTyVarsAndFV all_tvs
822 ; gbl_tvs <- tcGetGlobalTyVars
823 ; let qtvs = varSetElems (all_tvs `minusVarSet` gbl_tvs)
824 -- We could close gbl_tvs, but its not necessary for
825 -- soundness, and it'll only affect which tyvars, not which
826 -- dictionaries, we quantify over
828 ; qtvs' <- zonkQuantifiedTyVars qtvs
830 -- Now we are back to normal (c.f. tcSimplCheck)
831 ; implic_bind <- bindIrreds loc qtvs' givens irreds
833 ; traceTc (text "tcSimplifyInferCheck ->" <+> ppr (implic_bind))
834 ; return (qtvs', binds `unionBags` implic_bind) }
837 Note [Squashing methods]
838 ~~~~~~~~~~~~~~~~~~~~~~~~~
839 Be careful if you want to float methods more:
840 truncate :: forall a. RealFrac a => forall b. Integral b => a -> b
841 From an application (truncate f i) we get
844 If we have also have a second occurrence of truncate, we get
847 When simplifying with i,f free, we might still notice that
848 t1=t3; but alas, the binding for t2 (which mentions t1)
849 may continue to float out!
854 class Y a b | a -> b where
857 instance Y [[a]] a where
860 k :: X a -> X a -> X a
862 g :: Num a => [X a] -> [X a]
865 h ys = ys ++ map (k (y [[0]])) xs
867 The excitement comes when simplifying the bindings for h. Initially
868 try to simplify {y @ [[t1]] t2, 0 @ t1}, with initial qtvs = {t2}.
869 From this we get t1:=:t2, but also various bindings. We can't forget
870 the bindings (because of [LOOP]), but in fact t1 is what g is
873 The net effect of [NO TYVARS]
876 isFreeWhenInferring :: TyVarSet -> Inst -> Bool
877 isFreeWhenInferring qtvs inst
878 = isFreeWrtTyVars qtvs inst -- Constrains no quantified vars
879 && isInheritableInst inst -- and no implicit parameter involved
880 -- see Note [Inheriting implicit parameters]
882 {- No longer used (with implication constraints)
883 isFreeWhenChecking :: TyVarSet -- Quantified tyvars
884 -> NameSet -- Quantified implicit parameters
886 isFreeWhenChecking qtvs ips inst
887 = isFreeWrtTyVars qtvs inst
888 && isFreeWrtIPs ips inst
891 isFreeWrtTyVars qtvs inst = tyVarsOfInst inst `disjointVarSet` qtvs
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, *and* throwing the constraint into the LIE
978 -- The binding looks like
979 -- (ir1, .., irn) = f qtvs givens
980 -- where f is (evidence for) the new implication constraint
981 -- f :: forall qtvs. {reft} givens => (ir1, .., irn)
982 -- qtvs includes coercion variables
984 -- This binding must line up the 'rhs' in reduceImplication
985 makeImplicationBind loc all_tvs
986 givens -- Guaranteed all Dicts
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
995 eq_tyvar_cos = mkTyVarTys (varSetElems $ tyVarsOfTypes $ map eqInstType eq_givens)
996 -- Urgh! See line 2187 or thereabouts. I believe that all these
997 -- 'givens' must be a simple CoVar. This MUST be cleaned up.
999 ; let name = mkInternalName uniq (mkVarOcc "ic") span
1000 implic_inst = ImplicInst { tci_name = name,
1001 tci_tyvars = all_tvs,
1002 tci_given = (eq_givens ++ dict_givens),
1003 tci_wanted = irreds, tci_loc = loc }
1004 ; let -- only create binder for dict_irreds
1005 (eq_irreds, dict_irreds) = partition isEqInst irreds
1006 n_dict_irreds = length dict_irreds
1007 dict_irred_ids = map instToId dict_irreds
1008 tup_ty = mkTupleTy Boxed n_dict_irreds (map idType dict_irred_ids)
1009 pat = TuplePat (map nlVarPat dict_irred_ids) Boxed tup_ty
1010 rhs = L span (mkHsWrap co (HsVar (instToId implic_inst)))
1011 co = mkWpApps (map instToId dict_givens)
1012 <.> mkWpTyApps eq_tyvar_cos
1013 <.> mkWpTyApps (mkTyVarTys all_tvs)
1014 bind | [dict_irred_id] <- dict_irred_ids = VarBind dict_irred_id rhs
1015 | otherwise = PatBind { pat_lhs = L span pat,
1016 pat_rhs = unguardedGRHSs rhs,
1017 pat_rhs_ty = tup_ty,
1018 bind_fvs = placeHolderNames }
1019 ; traceTc $ text "makeImplicationBind" <+> ppr implic_inst
1020 ; return ([implic_inst], unitBag (L span bind))
1023 -----------------------------------------------------------
1024 tryHardCheckLoop :: SDoc
1026 -> TcM ([Inst], TcDictBinds)
1028 tryHardCheckLoop doc wanteds
1029 = do { (irreds,binds) <- checkLoop (mkRedEnv doc try_me []) wanteds
1030 ; return (irreds,binds)
1033 try_me inst = ReduceMe AddSCs
1034 -- Here's the try-hard bit
1036 -----------------------------------------------------------
1037 gentleCheckLoop :: InstLoc
1040 -> TcM ([Inst], TcDictBinds)
1042 gentleCheckLoop inst_loc givens wanteds
1043 = do { (irreds,binds) <- checkLoop env wanteds
1044 ; return (irreds,binds)
1047 env = mkRedEnv (pprInstLoc inst_loc) try_me givens
1049 try_me inst | isMethodOrLit inst = ReduceMe AddSCs
1051 -- When checking against a given signature
1052 -- we MUST be very gentle: Note [Check gently]
1054 gentleInferLoop :: SDoc -> [Inst]
1055 -> TcM ([Inst], TcDictBinds)
1056 gentleInferLoop doc wanteds
1057 = do { (irreds, binds) <- checkLoop env wanteds
1058 ; return (irreds, binds) }
1060 env = mkRedEnv doc try_me []
1061 try_me inst | isMethodOrLit inst = ReduceMe AddSCs
1066 ~~~~~~~~~~~~~~~~~~~~
1067 We have to very careful about not simplifying too vigorously
1072 f :: Show b => T b -> b
1073 f (MkT x) = show [x]
1075 Inside the pattern match, which binds (a:*, x:a), we know that
1077 Hence we have a dictionary for Show [a] available; and indeed we
1078 need it. We are going to build an implication contraint
1079 forall a. (b~[a]) => Show [a]
1080 Later, we will solve this constraint using the knowledge (Show b)
1082 But we MUST NOT reduce (Show [a]) to (Show a), else the whole
1083 thing becomes insoluble. So we simplify gently (get rid of literals
1084 and methods only, plus common up equal things), deferring the real
1085 work until top level, when we solve the implication constraint
1086 with tryHardCheckLooop.
1090 -----------------------------------------------------------
1093 -> TcM ([Inst], TcDictBinds)
1094 -- Precondition: givens are completely rigid
1095 -- Postcondition: returned Insts are zonked
1097 checkLoop env wanteds
1098 = go env wanteds (return ())
1099 where go env wanteds elim_skolems
1100 = do { -- We do need to zonk the givens; cf Note [Zonking RedEnv]
1101 ; env' <- zonkRedEnv env
1102 ; wanteds' <- zonkInsts wanteds
1104 ; (improved, binds, irreds, elim_more_skolems)
1105 <- reduceContext env' wanteds'
1106 ; let elim_skolems' = elim_skolems >> elim_more_skolems
1108 ; if not improved then
1109 elim_skolems' >> return (irreds, binds)
1112 -- If improvement did some unification, we go round again.
1113 -- We start again with irreds, not wanteds
1114 -- Using an instance decl might have introduced a fresh type
1115 -- variable which might have been unified, so we'd get an
1116 -- infinite loop if we started again with wanteds!
1118 { (irreds1, binds1) <- go env' irreds elim_skolems'
1119 ; return (irreds1, binds `unionBags` binds1) } }
1122 Note [Zonking RedEnv]
1123 ~~~~~~~~~~~~~~~~~~~~~
1124 It might appear as if the givens in RedEnv are always rigid, but that is not
1125 necessarily the case for programs involving higher-rank types that have class
1126 contexts constraining the higher-rank variables. An example from tc237 in the
1129 class Modular s a | s -> a
1131 wim :: forall a w. Integral a
1132 => a -> (forall s. Modular s a => M s w) -> w
1133 wim i k = error "urk"
1135 test5 :: (Modular s a, Integral a) => M s a
1138 test4 = wim 4 test4'
1140 Notice how the variable 'a' of (Modular s a) in the rank-2 type of wim is
1141 quantified further outside. When type checking test4, we have to check
1142 whether the signature of test5 is an instance of
1144 (forall s. Modular s a => M s w)
1146 Consequently, we will get (Modular s t_a), where t_a is a TauTv into the
1149 Given the FD of Modular in this example, class improvement will instantiate
1150 t_a to 'a', where 'a' is the skolem from test5's signatures (due to the
1151 Modular s a predicate in that signature). If we don't zonk (Modular s t_a) in
1152 the givens, we will get into a loop as improveOne uses the unification engine
1153 TcGadt.tcUnifyTys, which doesn't know about mutable type variables.
1158 class If b t e r | b t e -> r
1161 class Lte a b c | a b -> c where lte :: a -> b -> c
1163 instance (Lte a b l,If l b a c) => Max a b c
1165 Wanted: Max Z (S x) y
1167 Then we'll reduce using the Max instance to:
1168 (Lte Z (S x) l, If l (S x) Z y)
1169 and improve by binding l->T, after which we can do some reduction
1170 on both the Lte and If constraints. What we *can't* do is start again
1171 with (Max Z (S x) y)!
1175 %************************************************************************
1177 tcSimplifySuperClasses
1179 %************************************************************************
1181 Note [SUPERCLASS-LOOP 1]
1182 ~~~~~~~~~~~~~~~~~~~~~~~~
1183 We have to be very, very careful when generating superclasses, lest we
1184 accidentally build a loop. Here's an example:
1188 class S a => C a where { opc :: a -> a }
1189 class S b => D b where { opd :: b -> b }
1191 instance C Int where
1194 instance D Int where
1197 From (instance C Int) we get the constraint set {ds1:S Int, dd:D Int}
1198 Simplifying, we may well get:
1199 $dfCInt = :C ds1 (opd dd)
1202 Notice that we spot that we can extract ds1 from dd.
1204 Alas! Alack! We can do the same for (instance D Int):
1206 $dfDInt = :D ds2 (opc dc)
1210 And now we've defined the superclass in terms of itself.
1212 Solution: never generate a superclass selectors at all when
1213 satisfying the superclass context of an instance declaration.
1215 Two more nasty cases are in
1220 tcSimplifySuperClasses
1225 tcSimplifySuperClasses loc givens sc_wanteds
1226 = do { traceTc (text "tcSimplifySuperClasses")
1227 ; (irreds,binds1) <- checkLoop env sc_wanteds
1228 ; let (tidy_env, tidy_irreds) = tidyInsts irreds
1229 ; reportNoInstances tidy_env (Just (loc, givens)) tidy_irreds
1232 env = mkRedEnv (pprInstLoc loc) try_me givens
1233 try_me inst = ReduceMe NoSCs
1234 -- Like tryHardCheckLoop, but with NoSCs
1238 %************************************************************************
1240 \subsection{tcSimplifyRestricted}
1242 %************************************************************************
1244 tcSimplifyRestricted infers which type variables to quantify for a
1245 group of restricted bindings. This isn't trivial.
1248 We want to quantify over a to get id :: forall a. a->a
1251 We do not want to quantify over a, because there's an Eq a
1252 constraint, so we get eq :: a->a->Bool (notice no forall)
1255 RHS has type 'tau', whose free tyvars are tau_tvs
1256 RHS has constraints 'wanteds'
1259 Quantify over (tau_tvs \ ftvs(wanteds))
1260 This is bad. The constraints may contain (Monad (ST s))
1261 where we have instance Monad (ST s) where...
1262 so there's no need to be monomorphic in s!
1264 Also the constraint might be a method constraint,
1265 whose type mentions a perfectly innocent tyvar:
1266 op :: Num a => a -> b -> a
1267 Here, b is unconstrained. A good example would be
1269 We want to infer the polymorphic type
1270 foo :: forall b. b -> b
1273 Plan B (cunning, used for a long time up to and including GHC 6.2)
1274 Step 1: Simplify the constraints as much as possible (to deal
1275 with Plan A's problem). Then set
1276 qtvs = tau_tvs \ ftvs( simplify( wanteds ) )
1278 Step 2: Now simplify again, treating the constraint as 'free' if
1279 it does not mention qtvs, and trying to reduce it otherwise.
1280 The reasons for this is to maximise sharing.
1282 This fails for a very subtle reason. Suppose that in the Step 2
1283 a constraint (Foo (Succ Zero) (Succ Zero) b) gets thrown upstairs as 'free'.
1284 In the Step 1 this constraint might have been simplified, perhaps to
1285 (Foo Zero Zero b), AND THEN THAT MIGHT BE IMPROVED, to bind 'b' to 'T'.
1286 This won't happen in Step 2... but that in turn might prevent some other
1287 constraint (Baz [a] b) being simplified (e.g. via instance Baz [a] T where {..})
1288 and that in turn breaks the invariant that no constraints are quantified over.
1290 Test typecheck/should_compile/tc177 (which failed in GHC 6.2) demonstrates
1295 Step 1: Simplify the constraints as much as possible (to deal
1296 with Plan A's problem). Then set
1297 qtvs = tau_tvs \ ftvs( simplify( wanteds ) )
1298 Return the bindings from Step 1.
1301 A note about Plan C (arising from "bug" reported by George Russel March 2004)
1304 instance (HasBinary ty IO) => HasCodedValue ty
1306 foo :: HasCodedValue a => String -> IO a
1308 doDecodeIO :: HasCodedValue a => () -> () -> IO a
1309 doDecodeIO codedValue view
1310 = let { act = foo "foo" } in act
1312 You might think this should work becuase the call to foo gives rise to a constraint
1313 (HasCodedValue t), which can be satisfied by the type sig for doDecodeIO. But the
1314 restricted binding act = ... calls tcSimplifyRestricted, and PlanC simplifies the
1315 constraint using the (rather bogus) instance declaration, and now we are stuffed.
1317 I claim this is not really a bug -- but it bit Sergey as well as George. So here's
1321 Plan D (a variant of plan B)
1322 Step 1: Simplify the constraints as much as possible (to deal
1323 with Plan A's problem), BUT DO NO IMPROVEMENT. Then set
1324 qtvs = tau_tvs \ ftvs( simplify( wanteds ) )
1326 Step 2: Now simplify again, treating the constraint as 'free' if
1327 it does not mention qtvs, and trying to reduce it otherwise.
1329 The point here is that it's generally OK to have too few qtvs; that is,
1330 to make the thing more monomorphic than it could be. We don't want to
1331 do that in the common cases, but in wierd cases it's ok: the programmer
1332 can always add a signature.
1334 Too few qtvs => too many wanteds, which is what happens if you do less
1339 tcSimplifyRestricted -- Used for restricted binding groups
1340 -- i.e. ones subject to the monomorphism restriction
1343 -> [Name] -- Things bound in this group
1344 -> TcTyVarSet -- Free in the type of the RHSs
1345 -> [Inst] -- Free in the RHSs
1346 -> TcM ([TyVar], -- Tyvars to quantify (zonked and quantified)
1347 TcDictBinds) -- Bindings
1348 -- tcSimpifyRestricted returns no constraints to
1349 -- quantify over; by definition there are none.
1350 -- They are all thrown back in the LIE
1352 tcSimplifyRestricted doc top_lvl bndrs tau_tvs wanteds
1353 -- Zonk everything in sight
1354 = do { traceTc (text "tcSimplifyRestricted")
1355 ; wanteds' <- zonkInsts wanteds
1357 -- 'ReduceMe': Reduce as far as we can. Don't stop at
1358 -- dicts; the idea is to get rid of as many type
1359 -- variables as possible, and we don't want to stop
1360 -- at (say) Monad (ST s), because that reduces
1361 -- immediately, with no constraint on s.
1363 -- BUT do no improvement! See Plan D above
1364 -- HOWEVER, some unification may take place, if we instantiate
1365 -- a method Inst with an equality constraint
1366 ; let env = mkNoImproveRedEnv doc (\i -> ReduceMe AddSCs)
1367 ; (_imp, _binds, constrained_dicts, elim_skolems)
1368 <- reduceContext env wanteds'
1371 -- Next, figure out the tyvars we will quantify over
1372 ; tau_tvs' <- zonkTcTyVarsAndFV (varSetElems tau_tvs)
1373 ; gbl_tvs' <- tcGetGlobalTyVars
1374 ; constrained_dicts' <- zonkInsts constrained_dicts
1376 ; let qtvs1 = tau_tvs' `minusVarSet` oclose (fdPredsOfInsts constrained_dicts) gbl_tvs'
1377 -- As in tcSimplifyInfer
1379 -- Do not quantify over constrained type variables:
1380 -- this is the monomorphism restriction
1381 constrained_tvs' = tyVarsOfInsts constrained_dicts'
1382 qtvs = qtvs1 `minusVarSet` constrained_tvs'
1383 pp_bndrs = pprWithCommas (quotes . ppr) bndrs
1386 ; warn_mono <- doptM Opt_WarnMonomorphism
1387 ; warnTc (warn_mono && (constrained_tvs' `intersectsVarSet` qtvs1))
1388 (vcat[ ptext SLIT("the Monomorphism Restriction applies to the binding")
1389 <> plural bndrs <+> ptext SLIT("for") <+> pp_bndrs,
1390 ptext SLIT("Consider giving a type signature for") <+> pp_bndrs])
1392 ; traceTc (text "tcSimplifyRestricted" <+> vcat [
1393 pprInsts wanteds, pprInsts constrained_dicts',
1395 ppr constrained_tvs', ppr tau_tvs', ppr qtvs ])
1397 -- The first step may have squashed more methods than
1398 -- necessary, so try again, this time more gently, knowing the exact
1399 -- set of type variables to quantify over.
1401 -- We quantify only over constraints that are captured by qtvs;
1402 -- these will just be a subset of non-dicts. This in contrast
1403 -- to normal inference (using isFreeWhenInferring) in which we quantify over
1404 -- all *non-inheritable* constraints too. This implements choice
1405 -- (B) under "implicit parameter and monomorphism" above.
1407 -- Remember that we may need to do *some* simplification, to
1408 -- (for example) squash {Monad (ST s)} into {}. It's not enough
1409 -- just to float all constraints
1411 -- At top level, we *do* squash methods becuase we want to
1412 -- expose implicit parameters to the test that follows
1413 ; let is_nested_group = isNotTopLevel top_lvl
1414 try_me inst | isFreeWrtTyVars qtvs inst,
1415 (is_nested_group || isDict inst) = Stop
1416 | otherwise = ReduceMe AddSCs
1417 env = mkNoImproveRedEnv doc try_me
1418 ; (_imp, binds, irreds, elim_skolems) <- reduceContext env wanteds'
1421 -- See "Notes on implicit parameters, Question 4: top level"
1422 ; ASSERT( all (isFreeWrtTyVars qtvs) irreds ) -- None should be captured
1423 if is_nested_group then
1425 else do { let (bad_ips, non_ips) = partition isIPDict irreds
1426 ; addTopIPErrs bndrs bad_ips
1427 ; extendLIEs non_ips }
1429 ; qtvs' <- zonkQuantifiedTyVars (varSetElems qtvs)
1430 ; return (qtvs', binds) }
1434 %************************************************************************
1438 %************************************************************************
1440 On the LHS of transformation rules we only simplify methods and constants,
1441 getting dictionaries. We want to keep all of them unsimplified, to serve
1442 as the available stuff for the RHS of the rule.
1444 Example. Consider the following left-hand side of a rule
1446 f (x == y) (y > z) = ...
1448 If we typecheck this expression we get constraints
1450 d1 :: Ord a, d2 :: Eq a
1452 We do NOT want to "simplify" to the LHS
1454 forall x::a, y::a, z::a, d1::Ord a.
1455 f ((==) (eqFromOrd d1) x y) ((>) d1 y z) = ...
1459 forall x::a, y::a, z::a, d1::Ord a, d2::Eq a.
1460 f ((==) d2 x y) ((>) d1 y z) = ...
1462 Here is another example:
1464 fromIntegral :: (Integral a, Num b) => a -> b
1465 {-# RULES "foo" fromIntegral = id :: Int -> Int #-}
1467 In the rule, a=b=Int, and Num Int is a superclass of Integral Int. But
1468 we *dont* want to get
1470 forall dIntegralInt.
1471 fromIntegral Int Int dIntegralInt (scsel dIntegralInt) = id Int
1473 because the scsel will mess up RULE matching. Instead we want
1475 forall dIntegralInt, dNumInt.
1476 fromIntegral Int Int dIntegralInt dNumInt = id Int
1480 g (x == y) (y == z) = ..
1482 where the two dictionaries are *identical*, we do NOT WANT
1484 forall x::a, y::a, z::a, d1::Eq a
1485 f ((==) d1 x y) ((>) d1 y z) = ...
1487 because that will only match if the dict args are (visibly) equal.
1488 Instead we want to quantify over the dictionaries separately.
1490 In short, tcSimplifyRuleLhs must *only* squash LitInst and MethInts, leaving
1491 all dicts unchanged, with absolutely no sharing. It's simpler to do this
1492 from scratch, rather than further parameterise simpleReduceLoop etc
1495 tcSimplifyRuleLhs :: [Inst] -> TcM ([Inst], TcDictBinds)
1496 tcSimplifyRuleLhs wanteds
1497 = go [] emptyBag wanteds
1500 = return (dicts, binds)
1501 go dicts binds (w:ws)
1503 = go (w:dicts) binds ws
1505 = do { w' <- zonkInst w -- So that (3::Int) does not generate a call
1506 -- to fromInteger; this looks fragile to me
1507 ; lookup_result <- lookupSimpleInst w'
1508 ; case lookup_result of
1510 go dicts (addInstToDictBind binds w rhs) (ws' ++ ws)
1511 NoInstance -> pprPanic "tcSimplifyRuleLhs" (ppr w)
1515 tcSimplifyBracket is used when simplifying the constraints arising from
1516 a Template Haskell bracket [| ... |]. We want to check that there aren't
1517 any constraints that can't be satisfied (e.g. Show Foo, where Foo has no
1518 Show instance), but we aren't otherwise interested in the results.
1519 Nor do we care about ambiguous dictionaries etc. We will type check
1520 this bracket again at its usage site.
1523 tcSimplifyBracket :: [Inst] -> TcM ()
1524 tcSimplifyBracket wanteds
1525 = do { tryHardCheckLoop doc wanteds
1528 doc = text "tcSimplifyBracket"
1532 %************************************************************************
1534 \subsection{Filtering at a dynamic binding}
1536 %************************************************************************
1541 we must discharge all the ?x constraints from B. We also do an improvement
1542 step; if we have ?x::t1 and ?x::t2 we must unify t1, t2.
1544 Actually, the constraints from B might improve the types in ?x. For example
1546 f :: (?x::Int) => Char -> Char
1549 then the constraint (?x::Int) arising from the call to f will
1550 force the binding for ?x to be of type Int.
1553 tcSimplifyIPs :: [Inst] -- The implicit parameters bound here
1556 -- We need a loop so that we do improvement, and then
1557 -- (next time round) generate a binding to connect the two
1559 -- Here the two ?x's have different types, and improvement
1560 -- makes them the same.
1562 tcSimplifyIPs given_ips wanteds
1563 = do { wanteds' <- zonkInsts wanteds
1564 ; given_ips' <- zonkInsts given_ips
1565 -- Unusually for checking, we *must* zonk the given_ips
1567 ; let env = mkRedEnv doc try_me given_ips'
1568 ; (improved, binds, irreds, elim_skolems) <- reduceContext env wanteds'
1571 ; if not improved then
1572 ASSERT( all is_free irreds )
1573 do { extendLIEs irreds
1576 tcSimplifyIPs given_ips wanteds }
1578 doc = text "tcSimplifyIPs" <+> ppr given_ips
1579 ip_set = mkNameSet (ipNamesOfInsts given_ips)
1580 is_free inst = isFreeWrtIPs ip_set inst
1582 -- Simplify any methods that mention the implicit parameter
1583 try_me inst | is_free inst = Stop
1584 | otherwise = ReduceMe NoSCs
1588 %************************************************************************
1590 \subsection[binds-for-local-funs]{@bindInstsOfLocalFuns@}
1592 %************************************************************************
1594 When doing a binding group, we may have @Insts@ of local functions.
1595 For example, we might have...
1597 let f x = x + 1 -- orig local function (overloaded)
1598 f.1 = f Int -- two instances of f
1603 The point is: we must drop the bindings for @f.1@ and @f.2@ here,
1604 where @f@ is in scope; those @Insts@ must certainly not be passed
1605 upwards towards the top-level. If the @Insts@ were binding-ified up
1606 there, they would have unresolvable references to @f@.
1608 We pass in an @init_lie@ of @Insts@ and a list of locally-bound @Ids@.
1609 For each method @Inst@ in the @init_lie@ that mentions one of the
1610 @Ids@, we create a binding. We return the remaining @Insts@ (in an
1611 @LIE@), as well as the @HsBinds@ generated.
1614 bindInstsOfLocalFuns :: [Inst] -> [TcId] -> TcM TcDictBinds
1615 -- Simlifies only MethodInsts, and generate only bindings of form
1617 -- We're careful not to even generate bindings of the form
1619 -- You'd think that'd be fine, but it interacts with what is
1620 -- arguably a bug in Match.tidyEqnInfo (see notes there)
1622 bindInstsOfLocalFuns wanteds local_ids
1623 | null overloaded_ids = do
1626 return emptyLHsBinds
1629 = do { (irreds, binds) <- gentleInferLoop doc for_me
1630 ; extendLIEs not_for_me
1634 doc = text "bindInsts" <+> ppr local_ids
1635 overloaded_ids = filter is_overloaded local_ids
1636 is_overloaded id = isOverloadedTy (idType id)
1637 (for_me, not_for_me) = partition (isMethodFor overloaded_set) wanteds
1639 overloaded_set = mkVarSet overloaded_ids -- There can occasionally be a lot of them
1640 -- so it's worth building a set, so that
1641 -- lookup (in isMethodFor) is faster
1645 %************************************************************************
1647 \subsection{Data types for the reduction mechanism}
1649 %************************************************************************
1651 The main control over context reduction is here
1655 = RedEnv { red_doc :: SDoc -- The context
1656 , red_try_me :: Inst -> WhatToDo
1657 , red_improve :: Bool -- True <=> do improvement
1658 , red_givens :: [Inst] -- All guaranteed rigid
1660 -- but see Note [Rigidity]
1661 , red_stack :: (Int, [Inst]) -- Recursion stack (for err msg)
1662 -- See Note [RedStack]
1666 -- The red_givens are rigid so far as cmpInst is concerned.
1667 -- There is one case where they are not totally rigid, namely in tcSimplifyIPs
1668 -- let ?x = e in ...
1669 -- Here, the given is (?x::a), where 'a' is not necy a rigid type
1670 -- But that doesn't affect the comparison, which is based only on mame.
1673 -- The red_stack pair (n,insts) pair is just used for error reporting.
1674 -- 'n' is always the depth of the stack.
1675 -- The 'insts' is the stack of Insts being reduced: to produce X
1676 -- I had to produce Y, to produce Y I had to produce Z, and so on.
1679 mkRedEnv :: SDoc -> (Inst -> WhatToDo) -> [Inst] -> RedEnv
1680 mkRedEnv doc try_me givens
1681 = RedEnv { red_doc = doc, red_try_me = try_me,
1682 red_givens = givens,
1684 red_improve = True }
1686 mkNoImproveRedEnv :: SDoc -> (Inst -> WhatToDo) -> RedEnv
1687 -- Do not do improvement; no givens
1688 mkNoImproveRedEnv doc try_me
1689 = RedEnv { red_doc = doc, red_try_me = try_me,
1692 red_improve = True }
1695 = ReduceMe WantSCs -- Try to reduce this
1696 -- If there's no instance, add the inst to the
1697 -- irreductible ones, but don't produce an error
1698 -- message of any kind.
1699 -- It might be quite legitimate such as (Eq a)!
1701 | Stop -- Return as irreducible unless it can
1702 -- be reduced to a constant in one step
1703 -- Do not add superclasses; see
1705 data WantSCs = NoSCs | AddSCs -- Tells whether we should add the superclasses
1706 -- of a predicate when adding it to the avails
1707 -- The reason for this flag is entirely the super-class loop problem
1708 -- Note [SUPER-CLASS LOOP 1]
1710 zonkRedEnv :: RedEnv -> TcM RedEnv
1712 = do { givens' <- mapM zonkInst (red_givens env)
1713 ; return $ env {red_givens = givens'}
1718 %************************************************************************
1720 \subsection[reduce]{@reduce@}
1722 %************************************************************************
1724 Note [Ancestor Equalities]
1725 ~~~~~~~~~~~~~~~~~~~~~~~~~~
1726 During context reduction, we add to the wanted equalities also those
1727 equalities that (transitively) occur in superclass contexts of wanted
1728 class constraints. Consider the following code
1730 class a ~ Int => C a
1733 If (C a) is wanted, we want to add (a ~ Int), which will be discharged by
1734 substituting Int for a. Hence, we ultimately want (C Int), which we
1735 discharge with the explicit instance.
1738 reduceContext :: RedEnv
1740 -> TcM (ImprovementDone,
1741 TcDictBinds, -- Dictionary bindings
1742 [Inst], -- Irreducible
1743 TcM ()) -- Undo skolems from SkolemOccurs
1745 reduceContext env wanteds
1746 = do { traceTc (text "reduceContext" <+> (vcat [
1747 text "----------------------",
1749 text "given" <+> ppr (red_givens env),
1750 text "wanted" <+> ppr wanteds,
1751 text "----------------------"
1755 ; let givens = red_givens env
1756 (given_eqs0, given_dicts0) = partition isEqInst givens
1757 (wanted_eqs0, wanted_non_eqs) = partition isEqInst wanteds
1758 (wanted_implics0, wanted_dicts) = partition isImplicInst wanted_non_eqs
1760 -- We want to add as wanted equalities those that (transitively)
1761 -- occur in superclass contexts of wanted class constraints.
1762 -- See Note [Ancestor Equalities]
1763 ; ancestor_eqs <- ancestorEqualities wanted_dicts
1764 ; let wanted_eqs = wanted_eqs0 ++ ancestor_eqs
1765 ; traceTc $ text "reduceContext: ancestor eqs" <+> ppr ancestor_eqs
1767 -- 1. Normalise the *given* *equality* constraints
1768 ; (given_eqs, eliminate_skolems) <- normaliseGivenEqs given_eqs0
1770 -- 2. Normalise the *given* *dictionary* constraints
1771 -- wrt. the toplevel and given equations
1772 ; (given_dicts, given_binds) <- normaliseGivenDicts given_eqs
1775 -- 5. Build the Avail mapping from "given_dicts"
1776 ; (init_state, extra_givens) <- getLIE $ do
1777 { init_state <- foldlM addGiven emptyAvails given_dicts
1781 -- *** ToDo: what to do with the "extra_givens"? For the
1782 -- moment I'm simply discarding them, which is probably wrong
1784 -- 6. Solve the *wanted* *dictionary* constraints (not implications)
1785 -- This may expose some further equational constraints...
1786 ; (avails, extra_eqs) <- getLIE (reduceList env wanted_dicts init_state)
1787 ; (dict_binds, bound_dicts, dict_irreds)
1788 <- extractResults avails wanted_dicts
1789 ; traceTc $ text "reduceContext extractresults" <+> vcat
1790 [ppr avails, ppr wanted_dicts, ppr dict_binds]
1792 -- Solve the wanted *implications*. In doing so, we can provide
1793 -- as "given" all the dicts that were originally given,
1794 -- *or* for which we now have bindings,
1795 -- *or* which are now irreds
1796 ; let implic_env = env { red_givens = givens ++ bound_dicts
1798 ; (implic_binds_s, implic_irreds_s)
1799 <- mapAndUnzipM (reduceImplication implic_env) wanted_implics0
1800 ; let implic_binds = unionManyBags implic_binds_s
1801 implic_irreds = concat implic_irreds_s
1803 -- Normalise the wanted equality constraints
1804 ; eq_irreds <- normaliseWantedEqs given_eqs (wanted_eqs ++ extra_eqs)
1806 -- Normalise the wanted dictionaries
1807 ; let irreds = dict_irreds ++ implic_irreds
1808 eqs = eq_irreds ++ given_eqs
1809 ; (norm_irreds, normalise_binds) <- normaliseWantedDicts eqs irreds
1811 -- Figure out whether we should go round again. We do so in either
1813 -- (1) If any of the mutable tyvars in givens or irreds has been
1814 -- filled in by improvement, there is merit in going around
1815 -- again, because we may make further progress.
1816 -- (2) If we managed to normalise any dicts, there is merit in going
1817 -- around gain, because reduceList may be able to get further.
1819 -- ToDo: We may have exposed new
1820 -- equality constraints and should probably go round again
1821 -- then as well. But currently we are dropping them on the
1824 ; let all_irreds = norm_irreds ++ eq_irreds
1825 ; improvedMetaTy <- anyM isFilledMetaTyVar $ varSetElems $
1826 tyVarsOfInsts (givens ++ all_irreds)
1827 ; let improvedDicts = not $ isEmptyBag normalise_binds
1828 improved = improvedMetaTy || improvedDicts
1830 -- The old plan (fragile)
1831 -- improveed = availsImproved avails
1832 -- || (not $ isEmptyBag normalise_binds1)
1833 -- || (not $ isEmptyBag normalise_binds2)
1834 -- || (any isEqInst irreds)
1836 ; traceTc (text "reduceContext end" <+> (vcat [
1837 text "----------------------",
1839 text "given" <+> ppr givens,
1840 text "given_eqs" <+> ppr given_eqs,
1841 text "wanted" <+> ppr wanteds,
1842 text "wanted_dicts" <+> ppr wanted_dicts,
1844 text "avails" <+> pprAvails avails,
1845 text "improved =" <+> ppr improved,
1846 text "(all) irreds = " <+> ppr all_irreds,
1847 text "dict-binds = " <+> ppr dict_binds,
1848 text "implic-binds = " <+> ppr implic_binds,
1849 text "----------------------"
1853 given_binds `unionBags` normalise_binds
1854 `unionBags` dict_binds
1855 `unionBags` implic_binds,
1860 tcImproveOne :: Avails -> Inst -> TcM ImprovementDone
1861 tcImproveOne avails inst
1862 | not (isDict inst) = return False
1864 = do { inst_envs <- tcGetInstEnvs
1865 ; let eqns = improveOne (classInstances inst_envs)
1866 (dictPred inst, pprInstArising inst)
1867 [ (dictPred p, pprInstArising p)
1868 | p <- availsInsts avails, isDict p ]
1869 -- Avails has all the superclasses etc (good)
1870 -- It also has all the intermediates of the deduction (good)
1871 -- It does not have duplicates (good)
1872 -- NB that (?x::t1) and (?x::t2) will be held separately in avails
1873 -- so that improve will see them separate
1874 ; traceTc (text "improveOne" <+> ppr inst)
1877 unifyEqns :: [(Equation,(PredType,SDoc),(PredType,SDoc))]
1878 -> TcM ImprovementDone
1879 unifyEqns [] = return False
1881 = do { traceTc (ptext SLIT("Improve:") <+> vcat (map pprEquationDoc eqns))
1885 unify ((qtvs, pairs), what1, what2)
1886 = addErrCtxtM (mkEqnMsg what1 what2) $ do
1887 (_, _, tenv) <- tcInstTyVars (varSetElems qtvs)
1888 mapM_ (unif_pr tenv) pairs
1889 unif_pr tenv (ty1,ty2) = unifyType (substTy tenv ty1) (substTy tenv ty2)
1891 pprEquationDoc (eqn, (p1,w1), (p2,w2)) = vcat [pprEquation eqn, nest 2 (ppr p1), nest 2 (ppr p2)]
1893 mkEqnMsg (pred1,from1) (pred2,from2) tidy_env
1894 = do { pred1' <- zonkTcPredType pred1; pred2' <- zonkTcPredType pred2
1895 ; let { pred1'' = tidyPred tidy_env pred1'; pred2'' = tidyPred tidy_env pred2' }
1896 ; let msg = vcat [ptext SLIT("When using functional dependencies to combine"),
1897 nest 2 (sep [ppr pred1'' <> comma, nest 2 from1]),
1898 nest 2 (sep [ppr pred2'' <> comma, nest 2 from2])]
1899 ; return (tidy_env, msg) }
1902 The main context-reduction function is @reduce@. Here's its game plan.
1905 reduceList :: RedEnv -> [Inst] -> Avails -> TcM Avails
1906 reduceList env@(RedEnv {red_stack = (n,stk)}) wanteds state
1907 = do { traceTc (text "reduceList " <+> (ppr wanteds $$ ppr state))
1909 ; when (debugIsOn && (n > 8)) $ do
1910 dumpTcRn (hang (ptext SLIT("Interesting! Context reduction stack depth") <+> int n)
1911 2 (ifPprDebug (nest 2 (pprStack stk))))
1912 ; if n >= ctxtStkDepth dopts then
1913 failWithTc (reduceDepthErr n stk)
1917 go [] state = return state
1918 go (w:ws) state = do { state' <- reduce (env {red_stack = (n+1, w:stk)}) w state
1921 -- Base case: we're done!
1922 reduce env wanted avails
1923 -- It's the same as an existing inst, or a superclass thereof
1924 | Just avail <- findAvail avails wanted
1925 = do { traceTc (text "reduce: found " <+> ppr wanted)
1930 = do { traceTc (text "reduce" <+> ppr wanted $$ ppr avails)
1931 ; case red_try_me env wanted of {
1932 Stop -> try_simple (addIrred NoSCs);
1933 -- See Note [No superclasses for Stop]
1935 ReduceMe want_scs -> do -- It should be reduced
1936 { (avails, lookup_result) <- reduceInst env avails wanted
1937 ; case lookup_result of
1938 NoInstance -> addIrred want_scs avails wanted
1939 -- Add it and its superclasses
1941 GenInst [] rhs -> addWanted want_scs avails wanted rhs []
1943 GenInst wanteds' rhs
1944 -> do { avails1 <- addIrred NoSCs avails wanted
1945 ; avails2 <- reduceList env wanteds' avails1
1946 ; addWanted want_scs avails2 wanted rhs wanteds' } }
1947 -- Temporarily do addIrred *before* the reduceList,
1948 -- which has the effect of adding the thing we are trying
1949 -- to prove to the database before trying to prove the things it
1950 -- needs. See note [RECURSIVE DICTIONARIES]
1951 -- NB: we must not do an addWanted before, because that adds the
1952 -- superclasses too, and that can lead to a spurious loop; see
1953 -- the examples in [SUPERCLASS-LOOP]
1954 -- So we do an addIrred before, and then overwrite it afterwards with addWanted
1957 -- First, see if the inst can be reduced to a constant in one step
1958 -- Works well for literals (1::Int) and constant dictionaries (d::Num Int)
1959 -- Don't bother for implication constraints, which take real work
1960 try_simple do_this_otherwise
1961 = do { res <- lookupSimpleInst wanted
1963 GenInst [] rhs -> addWanted AddSCs avails wanted rhs []
1964 other -> do_this_otherwise avails wanted }
1968 Note [SUPERCLASS-LOOP 2]
1969 ~~~~~~~~~~~~~~~~~~~~~~~~
1970 But the above isn't enough. Suppose we are *given* d1:Ord a,
1971 and want to deduce (d2:C [a]) where
1973 class Ord a => C a where
1974 instance Ord [a] => C [a] where ...
1976 Then we'll use the instance decl to deduce C [a] from Ord [a], and then add the
1977 superclasses of C [a] to avails. But we must not overwrite the binding
1978 for Ord [a] (which is obtained from Ord a) with a superclass selection or we'll just
1981 Here's another variant, immortalised in tcrun020
1982 class Monad m => C1 m
1983 class C1 m => C2 m x
1984 instance C2 Maybe Bool
1985 For the instance decl we need to build (C1 Maybe), and it's no good if
1986 we run around and add (C2 Maybe Bool) and its superclasses to the avails
1987 before we search for C1 Maybe.
1989 Here's another example
1990 class Eq b => Foo a b
1991 instance Eq a => Foo [a] a
1995 we'll first deduce that it holds (via the instance decl). We must not
1996 then overwrite the Eq t constraint with a superclass selection!
1998 At first I had a gross hack, whereby I simply did not add superclass constraints
1999 in addWanted, though I did for addGiven and addIrred. This was sub-optimal,
2000 becuase it lost legitimate superclass sharing, and it still didn't do the job:
2001 I found a very obscure program (now tcrun021) in which improvement meant the
2002 simplifier got two bites a the cherry... so something seemed to be an Stop
2003 first time, but reducible next time.
2005 Now we implement the Right Solution, which is to check for loops directly
2006 when adding superclasses. It's a bit like the occurs check in unification.
2009 Note [RECURSIVE DICTIONARIES]
2010 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2012 data D r = ZeroD | SuccD (r (D r));
2014 instance (Eq (r (D r))) => Eq (D r) where
2015 ZeroD == ZeroD = True
2016 (SuccD a) == (SuccD b) = a == b
2019 equalDC :: D [] -> D [] -> Bool;
2022 We need to prove (Eq (D [])). Here's how we go:
2026 by instance decl, holds if
2030 by instance decl of Eq, holds if
2032 where d2 = dfEqList d3
2035 But now we can "tie the knot" to give
2041 and it'll even run! The trick is to put the thing we are trying to prove
2042 (in this case Eq (D []) into the database before trying to prove its
2043 contributing clauses.
2046 %************************************************************************
2048 Reducing a single constraint
2050 %************************************************************************
2053 ---------------------------------------------
2054 reduceInst :: RedEnv -> Avails -> Inst -> TcM (Avails, LookupInstResult)
2055 reduceInst env avails other_inst
2056 = do { result <- lookupSimpleInst other_inst
2057 ; return (avails, result) }
2060 Note [Equational Constraints in Implication Constraints]
2061 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2063 An implication constraint is of the form
2065 where Given and Wanted may contain both equational and dictionary
2066 constraints. The delay and reduction of these two kinds of constraints
2069 -) In the generated code, wanted Dictionary constraints are wrapped up in an
2070 implication constraint that is created at the code site where the wanted
2071 dictionaries can be reduced via a let-binding. This let-bound implication
2072 constraint is deconstructed at the use-site of the wanted dictionaries.
2074 -) While the reduction of equational constraints is also delayed, the delay
2075 is not manifest in the generated code. The required evidence is generated
2076 in the code directly at the use-site. There is no let-binding and deconstruction
2077 necessary. The main disadvantage is that we cannot exploit sharing as the
2078 same evidence may be generated at multiple use-sites. However, this disadvantage
2079 is limited because it only concerns coercions which are erased.
2081 The different treatment is motivated by the different in representation. Dictionary
2082 constraints require manifest runtime dictionaries, while equations require coercions
2086 ---------------------------------------------
2087 reduceImplication :: RedEnv
2089 -> TcM (TcDictBinds, [Inst])
2092 Suppose we are simplifying the constraint
2093 forall bs. extras => wanted
2094 in the context of an overall simplification problem with givens 'givens'.
2097 * The 'givens' need not mention any of the quantified type variables
2098 e.g. forall {}. Eq a => Eq [a]
2099 forall {}. C Int => D (Tree Int)
2101 This happens when you have something like
2103 T1 :: Eq a => a -> T a
2106 f x = ...(case x of { T1 v -> v==v })...
2109 -- ToDo: should we instantiate tvs? I think it's not necessary
2111 -- Note on coercion variables:
2113 -- The extra given coercion variables are bound at two different sites:
2114 -- -) in the creation context of the implication constraint
2115 -- the solved equational constraints use these binders
2117 -- -) at the solving site of the implication constraint
2118 -- the solved dictionaries use these binders
2119 -- these binders are generated by reduceImplication
2121 reduceImplication env
2122 orig_implic@(ImplicInst { tci_name = name, tci_loc = inst_loc,
2124 tci_given = extra_givens, tci_wanted = wanteds })
2125 = do { -- Solve the sub-problem
2126 ; let try_me inst = ReduceMe AddSCs -- Note [Freeness and implications]
2127 env' = env { red_givens = extra_givens ++ red_givens env
2128 , red_doc = sep [ptext SLIT("reduceImplication for")
2130 nest 2 (parens $ ptext SLIT("within")
2132 , red_try_me = try_me }
2134 ; traceTc (text "reduceImplication" <+> vcat
2135 [ ppr (red_givens env), ppr extra_givens,
2137 ; (irreds, binds) <- checkLoop env' wanteds
2138 ; let (extra_eq_givens, extra_dict_givens) = partition isEqInst extra_givens
2139 -- SLPJ Sept 07: I think this is bogus; currently
2140 -- there are no Eqinsts in extra_givens
2141 dict_ids = map instToId extra_dict_givens
2143 -- Note [Reducing implication constraints]
2144 -- Tom -- update note, put somewhere!
2146 ; traceTc (text "reduceImplication result" <+> vcat
2147 [ppr irreds, ppr binds])
2149 ; -- extract superclass binds
2150 -- (sc_binds,_) <- extractResults avails []
2151 -- ; traceTc (text "reduceImplication sc_binds" <+> vcat
2152 -- [ppr sc_binds, ppr avails])
2155 -- SLPJ Sept 07: what if improvement happened inside the checkLoop?
2156 -- Then we must iterate the outer loop too!
2158 ; traceTc (text "reduceImplication condition" <+> ppr ((isEmptyLHsBinds binds) || (null irreds)))
2160 -- Progress is no longer measered by the number of bindings
2161 ; if (isEmptyLHsBinds binds) && (not $ null irreds) then -- No progress
2162 -- If there are any irreds, we back off and do nothing
2163 return (emptyBag, [orig_implic])
2165 { (simpler_implic_insts, bind)
2166 <- makeImplicationBind inst_loc tvs extra_givens irreds
2167 -- This binding is useless if the recursive simplification
2168 -- made no progress; but currently we don't try to optimise that
2169 -- case. After all, we only try hard to reduce at top level, or
2170 -- when inferring types.
2172 ; let dict_wanteds = filter (not . isEqInst) wanteds
2173 -- TOMDO: given equational constraints bug!
2174 -- we need a different evidence for given
2175 -- equations depending on whether we solve
2176 -- dictionary constraints or equational constraints
2178 eq_tyvars = varSetElems $ tyVarsOfTypes $ map eqInstType extra_eq_givens
2179 -- SLPJ Sept07: this looks Utterly Wrong to me, but I think
2180 -- that current extra_givens has no EqInsts, so
2181 -- it makes no difference
2182 co = wrap_inline -- Note [Always inline implication constraints]
2184 <.> mkWpLams eq_tyvars
2185 <.> mkWpLams dict_ids
2186 <.> WpLet (binds `unionBags` bind)
2187 wrap_inline | null dict_ids = idHsWrapper
2188 | otherwise = WpInline
2189 rhs = mkHsWrap co payload
2190 loc = instLocSpan inst_loc
2191 payload | [dict_wanted] <- dict_wanteds = HsVar (instToId dict_wanted)
2192 | otherwise = ExplicitTuple (map (L loc . HsVar . instToId) dict_wanteds) Boxed
2195 ; traceTc (vcat [text "reduceImplication" <+> ppr name,
2196 ppr simpler_implic_insts,
2197 text "->" <+> ppr rhs])
2198 ; return (unitBag (L loc (VarBind (instToId orig_implic) (L loc rhs))),
2199 simpler_implic_insts)
2204 Note [Always inline implication constraints]
2205 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2206 Suppose an implication constraint floats out of an INLINE function.
2207 Then although the implication has a single call site, it won't be
2208 inlined. And that is bad because it means that even if there is really
2209 *no* overloading (type signatures specify the exact types) there will
2210 still be dictionary passing in the resulting code. To avert this,
2211 we mark the implication constraints themselves as INLINE, at least when
2212 there is no loss of sharing as a result.
2214 Note [Freeness and implications]
2215 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2216 It's hard to say when an implication constraint can be floated out. Consider
2217 forall {} Eq a => Foo [a]
2218 The (Foo [a]) doesn't mention any of the quantified variables, but it
2219 still might be partially satisfied by the (Eq a).
2221 There is a useful special case when it *is* easy to partition the
2222 constraints, namely when there are no 'givens'. Consider
2223 forall {a}. () => Bar b
2224 There are no 'givens', and so there is no reason to capture (Bar b).
2225 We can let it float out. But if there is even one constraint we
2226 must be much more careful:
2227 forall {a}. C a b => Bar (m b)
2228 because (C a b) might have a superclass (D b), from which we might
2229 deduce (Bar [b]) when m later gets instantiated to []. Ha!
2231 Here is an even more exotic example
2233 Now consider the constraint
2234 forall b. D Int b => C Int
2235 We can satisfy the (C Int) from the superclass of D, so we don't want
2236 to float the (C Int) out, even though it mentions no type variable in
2239 Note [Pruning the givens in an implication constraint]
2240 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2241 Suppose we are about to form the implication constraint
2242 forall tvs. Eq a => Ord b
2243 The (Eq a) cannot contribute to the (Ord b), because it has no access to
2244 the type variable 'b'. So we could filter out the (Eq a) from the givens.
2246 Doing so would be a bit tidier, but all the implication constraints get
2247 simplified away by the optimiser, so it's no great win. So I don't take
2248 advantage of that at the moment.
2250 If you do, BE CAREFUL of wobbly type variables.
2253 %************************************************************************
2255 Avails and AvailHow: the pool of evidence
2257 %************************************************************************
2261 data Avails = Avails !ImprovementDone !AvailEnv
2263 type ImprovementDone = Bool -- True <=> some unification has happened
2264 -- so some Irreds might now be reducible
2265 -- keys that are now
2267 type AvailEnv = FiniteMap Inst AvailHow
2269 = IsIrred -- Used for irreducible dictionaries,
2270 -- which are going to be lambda bound
2272 | Given Inst -- Used for dictionaries for which we have a binding
2273 -- e.g. those "given" in a signature
2275 | Rhs -- Used when there is a RHS
2276 (LHsExpr TcId) -- The RHS
2277 [Inst] -- Insts free in the RHS; we need these too
2279 instance Outputable Avails where
2282 pprAvails (Avails imp avails)
2283 = vcat [ ptext SLIT("Avails") <> (if imp then ptext SLIT("[improved]") else empty)
2285 vcat [ sep [ppr inst, nest 2 (equals <+> ppr avail)]
2286 | (inst,avail) <- fmToList avails ]]
2288 instance Outputable AvailHow where
2291 -------------------------
2292 pprAvail :: AvailHow -> SDoc
2293 pprAvail IsIrred = text "Irred"
2294 pprAvail (Given x) = text "Given" <+> ppr x
2295 pprAvail (Rhs rhs bs) = sep [text "Rhs" <+> ppr bs,
2298 -------------------------
2299 extendAvailEnv :: AvailEnv -> Inst -> AvailHow -> AvailEnv
2300 extendAvailEnv env inst avail = addToFM env inst avail
2302 findAvailEnv :: AvailEnv -> Inst -> Maybe AvailHow
2303 findAvailEnv env wanted = lookupFM env wanted
2304 -- NB 1: the Ord instance of Inst compares by the class/type info
2305 -- *not* by unique. So
2306 -- d1::C Int == d2::C Int
2308 emptyAvails :: Avails
2309 emptyAvails = Avails False emptyFM
2311 findAvail :: Avails -> Inst -> Maybe AvailHow
2312 findAvail (Avails _ avails) wanted = findAvailEnv avails wanted
2314 elemAvails :: Inst -> Avails -> Bool
2315 elemAvails wanted (Avails _ avails) = wanted `elemFM` avails
2317 extendAvails :: Avails -> Inst -> AvailHow -> TcM Avails
2319 extendAvails avails@(Avails imp env) inst avail
2320 = do { imp1 <- tcImproveOne avails inst -- Do any improvement
2321 ; return (Avails (imp || imp1) (extendAvailEnv env inst avail)) }
2323 availsInsts :: Avails -> [Inst]
2324 availsInsts (Avails _ avails) = keysFM avails
2326 availsImproved (Avails imp _) = imp
2328 updateImprovement :: Avails -> Avails -> Avails
2329 -- (updateImprovement a1 a2) sets a1's improvement flag from a2
2330 updateImprovement (Avails _ avails1) (Avails imp2 _) = Avails imp2 avails1
2333 Extracting the bindings from a bunch of Avails.
2334 The bindings do *not* come back sorted in dependency order.
2335 We assume that they'll be wrapped in a big Rec, so that the
2336 dependency analyser can sort them out later
2339 type DoneEnv = FiniteMap Inst [Id]
2340 -- Tracks which things we have evidence for
2342 extractResults :: Avails
2344 -> TcM (TcDictBinds, -- Bindings
2345 [Inst], -- The insts bound by the bindings
2346 [Inst]) -- Irreducible ones
2347 -- Note [Reducing implication constraints]
2349 extractResults (Avails _ avails) wanteds
2350 = go emptyBag [] [] emptyFM wanteds
2352 go :: TcDictBinds -- Bindings for dicts
2353 -> [Inst] -- Bound by the bindings
2355 -> DoneEnv -- Has an entry for each inst in the above three sets
2357 -> TcM (TcDictBinds, [Inst], [Inst])
2358 go binds bound_dicts irreds done []
2359 = return (binds, bound_dicts, irreds)
2361 go binds bound_dicts irreds done (w:ws)
2362 | Just done_ids@(done_id : rest_done_ids) <- lookupFM done w
2363 = if w_id `elem` done_ids then
2364 go binds bound_dicts irreds done ws
2366 go (add_bind (nlHsVar done_id)) bound_dicts irreds
2367 (addToFM done w (done_id : w_id : rest_done_ids)) ws
2369 | otherwise -- Not yet done
2370 = case findAvailEnv avails w of
2371 Nothing -> pprTrace "Urk: extractResults" (ppr w) $
2372 go binds bound_dicts irreds done ws
2374 Just IsIrred -> go binds bound_dicts (w:irreds) done' ws
2376 Just (Rhs rhs ws') -> go (add_bind rhs) (w:bound_dicts) irreds done' (ws' ++ ws)
2378 Just (Given g) -> go binds' bound_dicts irreds (addToFM done w [g_id]) ws
2381 binds' | w_id == g_id = binds
2382 | otherwise = add_bind (nlHsVar g_id)
2385 done' = addToFM done w [w_id]
2386 add_bind rhs = addInstToDictBind binds w rhs
2390 Note [No superclasses for Stop]
2391 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2392 When we decide not to reduce an Inst -- the 'WhatToDo' --- we still
2393 add it to avails, so that any other equal Insts will be commoned up
2394 right here. However, we do *not* add superclasses. If we have
2397 but a is not bound here, then we *don't* want to derive dn from df
2398 here lest we lose sharing.
2401 addWanted :: WantSCs -> Avails -> Inst -> LHsExpr TcId -> [Inst] -> TcM Avails
2402 addWanted want_scs avails wanted rhs_expr wanteds
2403 = addAvailAndSCs want_scs avails wanted avail
2405 avail = Rhs rhs_expr wanteds
2407 addGiven :: Avails -> Inst -> TcM Avails
2408 addGiven avails given = addAvailAndSCs AddSCs avails given (Given given)
2409 -- Always add superclasses for 'givens'
2411 -- No ASSERT( not (given `elemAvails` avails) ) because in an instance
2412 -- decl for Ord t we can add both Ord t and Eq t as 'givens',
2413 -- so the assert isn't true
2417 addIrred :: WantSCs -> Avails -> Inst -> TcM Avails
2418 addIrred want_scs avails irred = ASSERT2( not (irred `elemAvails` avails), ppr irred $$ ppr avails )
2419 addAvailAndSCs want_scs avails irred IsIrred
2421 addAvailAndSCs :: WantSCs -> Avails -> Inst -> AvailHow -> TcM Avails
2422 addAvailAndSCs want_scs avails inst avail
2423 | not (isClassDict inst) = extendAvails avails inst avail
2424 | NoSCs <- want_scs = extendAvails avails inst avail
2425 | otherwise = do { traceTc (text "addAvailAndSCs" <+> vcat [ppr inst, ppr deps])
2426 ; avails' <- extendAvails avails inst avail
2427 ; addSCs is_loop avails' inst }
2429 is_loop pred = any (`tcEqType` mkPredTy pred) dep_tys
2430 -- Note: this compares by *type*, not by Unique
2431 deps = findAllDeps (unitVarSet (instToVar inst)) avail
2432 dep_tys = map idType (varSetElems deps)
2434 findAllDeps :: IdSet -> AvailHow -> IdSet
2435 -- Find all the Insts that this one depends on
2436 -- See Note [SUPERCLASS-LOOP 2]
2437 -- Watch out, though. Since the avails may contain loops
2438 -- (see Note [RECURSIVE DICTIONARIES]), so we need to track the ones we've seen so far
2439 findAllDeps so_far (Rhs _ kids) = foldl find_all so_far kids
2440 findAllDeps so_far other = so_far
2442 find_all :: IdSet -> Inst -> IdSet
2444 | isEqInst kid = so_far
2445 | kid_id `elemVarSet` so_far = so_far
2446 | Just avail <- findAvail avails kid = findAllDeps so_far' avail
2447 | otherwise = so_far'
2449 so_far' = extendVarSet so_far kid_id -- Add the new kid to so_far
2450 kid_id = instToId kid
2452 addSCs :: (TcPredType -> Bool) -> Avails -> Inst -> TcM Avails
2453 -- Add all the superclasses of the Inst to Avails
2454 -- The first param says "don't do this because the original thing
2455 -- depends on this one, so you'd build a loop"
2456 -- Invariant: the Inst is already in Avails.
2458 addSCs is_loop avails dict
2459 = ASSERT( isDict dict )
2460 do { sc_dicts <- newDictBndrs (instLoc dict) sc_theta'
2461 ; foldlM add_sc avails (zipEqual "add_scs" sc_dicts sc_sels) }
2463 (clas, tys) = getDictClassTys dict
2464 (tyvars, sc_theta, sc_sels, _) = classBigSig clas
2465 sc_theta' = filter (not . isEqPred) $
2466 substTheta (zipTopTvSubst tyvars tys) sc_theta
2468 add_sc avails (sc_dict, sc_sel)
2469 | is_loop (dictPred sc_dict) = return avails -- See Note [SUPERCLASS-LOOP 2]
2470 | is_given sc_dict = return avails
2471 | otherwise = do { avails' <- extendAvails avails sc_dict (Rhs sc_sel_rhs [dict])
2472 ; addSCs is_loop avails' sc_dict }
2474 sc_sel_rhs = L (instSpan dict) (HsWrap co_fn (HsVar sc_sel))
2475 co_fn = WpApp (instToVar dict) <.> mkWpTyApps tys
2477 is_given :: Inst -> Bool
2478 is_given sc_dict = case findAvail avails sc_dict of
2479 Just (Given _) -> True -- Given is cheaper than superclass selection
2482 -- From the a set of insts obtain all equalities that (transitively) occur in
2483 -- superclass contexts of class constraints (aka the ancestor equalities).
2485 ancestorEqualities :: [Inst] -> TcM [Inst]
2487 = mapM mkWantedEqInst -- turn only equality predicates..
2488 . filter isEqPred -- ..into wanted equality insts
2490 . addAEsToBag emptyBag -- collect the superclass constraints..
2491 . map dictPred -- ..of all predicates in a bag
2492 . filter isClassDict
2494 addAEsToBag :: Bag PredType -> [PredType] -> Bag PredType
2495 addAEsToBag bag [] = bag
2496 addAEsToBag bag (pred:preds)
2497 | pred `elemBag` bag = addAEsToBag bag preds
2498 | isEqPred pred = addAEsToBag bagWithPred preds
2499 | isClassPred pred = addAEsToBag bagWithPred predsWithSCs
2500 | otherwise = addAEsToBag bag preds
2502 bagWithPred = bag `snocBag` pred
2503 predsWithSCs = preds ++ substTheta (zipTopTvSubst tyvars tys) sc_theta
2505 (tyvars, sc_theta, _, _) = classBigSig clas
2506 (clas, tys) = getClassPredTys pred
2510 %************************************************************************
2512 \section{tcSimplifyTop: defaulting}
2514 %************************************************************************
2517 @tcSimplifyTop@ is called once per module to simplify all the constant
2518 and ambiguous Insts.
2520 We need to be careful of one case. Suppose we have
2522 instance Num a => Num (Foo a b) where ...
2524 and @tcSimplifyTop@ is given a constraint (Num (Foo x y)). Then it'll simplify
2525 to (Num x), and default x to Int. But what about y??
2527 It's OK: the final zonking stage should zap y to (), which is fine.
2531 tcSimplifyTop, tcSimplifyInteractive :: [Inst] -> TcM TcDictBinds
2532 tcSimplifyTop wanteds
2533 = tc_simplify_top doc False wanteds
2535 doc = text "tcSimplifyTop"
2537 tcSimplifyInteractive wanteds
2538 = tc_simplify_top doc True wanteds
2540 doc = text "tcSimplifyInteractive"
2542 -- The TcLclEnv should be valid here, solely to improve
2543 -- error message generation for the monomorphism restriction
2544 tc_simplify_top doc interactive wanteds
2545 = do { dflags <- getDOpts
2546 ; wanteds <- zonkInsts wanteds
2547 ; mapM_ zonkTopTyVar (varSetElems (tyVarsOfInsts wanteds))
2549 ; traceTc (text "tc_simplify_top 0: " <+> ppr wanteds)
2550 ; (irreds1, binds1) <- tryHardCheckLoop doc1 wanteds
2551 -- ; (irreds1, binds1) <- gentleInferLoop doc1 wanteds
2552 ; traceTc (text "tc_simplify_top 1: " <+> ppr irreds1)
2553 ; (irreds2, binds2) <- approximateImplications doc2 (\d -> True) irreds1
2554 ; traceTc (text "tc_simplify_top 2: " <+> ppr irreds2)
2556 -- Use the defaulting rules to do extra unification
2557 -- NB: irreds2 are already zonked
2558 ; (irreds3, binds3) <- disambiguate doc3 interactive dflags irreds2
2560 -- Deal with implicit parameters
2561 ; let (bad_ips, non_ips) = partition isIPDict irreds3
2562 (ambigs, others) = partition isTyVarDict non_ips
2564 ; topIPErrs bad_ips -- Can arise from f :: Int -> Int
2566 ; addNoInstanceErrs others
2567 ; addTopAmbigErrs ambigs
2569 ; return (binds1 `unionBags` binds2 `unionBags` binds3) }
2571 doc1 = doc <+> ptext SLIT("(first round)")
2572 doc2 = doc <+> ptext SLIT("(approximate)")
2573 doc3 = doc <+> ptext SLIT("(disambiguate)")
2576 If a dictionary constrains a type variable which is
2577 * not mentioned in the environment
2578 * and not mentioned in the type of the expression
2579 then it is ambiguous. No further information will arise to instantiate
2580 the type variable; nor will it be generalised and turned into an extra
2581 parameter to a function.
2583 It is an error for this to occur, except that Haskell provided for
2584 certain rules to be applied in the special case of numeric types.
2586 * at least one of its classes is a numeric class, and
2587 * all of its classes are numeric or standard
2588 then the type variable can be defaulted to the first type in the
2589 default-type list which is an instance of all the offending classes.
2591 So here is the function which does the work. It takes the ambiguous
2592 dictionaries and either resolves them (producing bindings) or
2593 complains. It works by splitting the dictionary list by type
2594 variable, and using @disambigOne@ to do the real business.
2596 @disambigOne@ assumes that its arguments dictionaries constrain all
2597 the same type variable.
2599 ADR Comment 20/6/94: I've changed the @CReturnable@ case to default to
2600 @()@ instead of @Int@. I reckon this is the Right Thing to do since
2601 the most common use of defaulting is code like:
2603 _ccall_ foo `seqPrimIO` bar
2605 Since we're not using the result of @foo@, the result if (presumably)
2609 disambiguate :: SDoc -> Bool -> DynFlags -> [Inst] -> TcM ([Inst], TcDictBinds)
2610 -- Just does unification to fix the default types
2611 -- The Insts are assumed to be pre-zonked
2612 disambiguate doc interactive dflags insts
2614 = return (insts, emptyBag)
2616 | null defaultable_groups
2617 = do { traceTc (text "disambigutate, no defaultable groups" <+> vcat [ppr unaries, ppr insts, ppr bad_tvs, ppr defaultable_groups])
2618 ; return (insts, emptyBag) }
2621 = do { -- Figure out what default types to use
2622 default_tys <- getDefaultTys extended_defaulting ovl_strings
2624 ; traceTc (text "disambiguate1" <+> vcat [ppr insts, ppr unaries, ppr bad_tvs, ppr defaultable_groups])
2625 ; mapM_ (disambigGroup default_tys) defaultable_groups
2627 -- disambigGroup does unification, hence try again
2628 ; tryHardCheckLoop doc insts }
2631 extended_defaulting = interactive || dopt Opt_ExtendedDefaultRules dflags
2632 ovl_strings = dopt Opt_OverloadedStrings dflags
2634 unaries :: [(Inst, Class, TcTyVar)] -- (C tv) constraints
2635 bad_tvs :: TcTyVarSet -- Tyvars mentioned by *other* constraints
2636 (unaries, bad_tvs_s) = partitionWith find_unary insts
2637 bad_tvs = unionVarSets bad_tvs_s
2639 -- Finds unary type-class constraints
2640 find_unary d@(Dict {tci_pred = ClassP cls [ty]})
2641 | Just tv <- tcGetTyVar_maybe ty = Left (d,cls,tv)
2642 find_unary inst = Right (tyVarsOfInst inst)
2644 -- Group by type variable
2645 defaultable_groups :: [[(Inst,Class,TcTyVar)]]
2646 defaultable_groups = filter defaultable_group (equivClasses cmp_tv unaries)
2647 cmp_tv (_,_,tv1) (_,_,tv2) = tv1 `compare` tv2
2649 defaultable_group :: [(Inst,Class,TcTyVar)] -> Bool
2650 defaultable_group ds@((_,_,tv):_)
2651 = isTyConableTyVar tv -- Note [Avoiding spurious errors]
2652 && not (tv `elemVarSet` bad_tvs)
2653 && defaultable_classes [c | (_,c,_) <- ds]
2654 defaultable_group [] = panic "defaultable_group"
2656 defaultable_classes clss
2657 | extended_defaulting = any isInteractiveClass clss
2658 | otherwise = all is_std_class clss && (any is_num_class clss)
2660 -- In interactive mode, or with -fextended-default-rules,
2661 -- we default Show a to Show () to avoid graututious errors on "show []"
2662 isInteractiveClass cls
2663 = is_num_class cls || (classKey cls `elem` [showClassKey, eqClassKey, ordClassKey])
2665 is_num_class cls = isNumericClass cls || (ovl_strings && (cls `hasKey` isStringClassKey))
2666 -- is_num_class adds IsString to the standard numeric classes,
2667 -- when -foverloaded-strings is enabled
2669 is_std_class cls = isStandardClass cls || (ovl_strings && (cls `hasKey` isStringClassKey))
2670 -- Similarly is_std_class
2672 -----------------------
2673 disambigGroup :: [Type] -- The default types
2674 -> [(Inst,Class,TcTyVar)] -- All standard classes of form (C a)
2675 -> TcM () -- Just does unification, to fix the default types
2677 disambigGroup default_tys dicts
2678 = try_default default_tys
2680 (_,_,tyvar) = ASSERT(not (null dicts)) head dicts -- Should be non-empty
2681 classes = [c | (_,c,_) <- dicts]
2683 try_default [] = return ()
2684 try_default (default_ty : default_tys)
2685 = tryTcLIE_ (try_default default_tys) $
2686 do { tcSimplifyDefault [mkClassPred clas [default_ty] | clas <- classes]
2687 -- This may fail; then the tryTcLIE_ kicks in
2688 -- Failure here is caused by there being no type in the
2689 -- default list which can satisfy all the ambiguous classes.
2690 -- For example, if Real a is reqd, but the only type in the
2691 -- default list is Int.
2693 -- After this we can't fail
2694 ; warnDefault dicts default_ty
2695 ; unifyType default_ty (mkTyVarTy tyvar)
2696 ; return () -- TOMDO: do something with the coercion
2700 -----------------------
2701 getDefaultTys :: Bool -> Bool -> TcM [Type]
2702 getDefaultTys extended_deflts ovl_strings
2703 = do { mb_defaults <- getDeclaredDefaultTys
2704 ; case mb_defaults of {
2705 Just tys -> return tys ; -- User-supplied defaults
2708 -- No use-supplied default
2709 -- Use [Integer, Double], plus modifications
2710 { integer_ty <- tcMetaTy integerTyConName
2711 ; checkWiredInTyCon doubleTyCon
2712 ; string_ty <- tcMetaTy stringTyConName
2713 ; return (opt_deflt extended_deflts unitTy
2714 -- Note [Default unitTy]
2716 [integer_ty,doubleTy]
2718 opt_deflt ovl_strings string_ty) } } }
2720 opt_deflt True ty = [ty]
2721 opt_deflt False ty = []
2724 Note [Default unitTy]
2725 ~~~~~~~~~~~~~~~~~~~~~
2726 In interative mode (or with -fextended-default-rules) we add () as the first type we
2727 try when defaulting. This has very little real impact, except in the following case.
2729 Text.Printf.printf "hello"
2730 This has type (forall a. IO a); it prints "hello", and returns 'undefined'. We don't
2731 want the GHCi repl loop to try to print that 'undefined'. The neatest thing is to
2732 default the 'a' to (), rather than to Integer (which is what would otherwise happen;
2733 and then GHCi doesn't attempt to print the (). So in interactive mode, we add
2734 () to the list of defaulting types. See Trac #1200.
2736 Note [Avoiding spurious errors]
2737 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2738 When doing the unification for defaulting, we check for skolem
2739 type variables, and simply don't default them. For example:
2740 f = (*) -- Monomorphic
2741 g :: Num a => a -> a
2743 Here, we get a complaint when checking the type signature for g,
2744 that g isn't polymorphic enough; but then we get another one when
2745 dealing with the (Num a) context arising from f's definition;
2746 we try to unify a with Int (to default it), but find that it's
2747 already been unified with the rigid variable from g's type sig
2750 %************************************************************************
2752 \subsection[simple]{@Simple@ versions}
2754 %************************************************************************
2756 Much simpler versions when there are no bindings to make!
2758 @tcSimplifyThetas@ simplifies class-type constraints formed by
2759 @deriving@ declarations and when specialising instances. We are
2760 only interested in the simplified bunch of class/type constraints.
2762 It simplifies to constraints of the form (C a b c) where
2763 a,b,c are type variables. This is required for the context of
2764 instance declarations.
2767 tcSimplifyDeriv :: InstOrigin
2769 -> ThetaType -- Wanted
2770 -> TcM ThetaType -- Needed
2771 -- Given instance (wanted) => C inst_ty
2772 -- Simplify 'wanted' as much as possible
2774 tcSimplifyDeriv orig tyvars theta
2775 = do { (tvs, _, tenv) <- tcInstTyVars tyvars
2776 -- The main loop may do unification, and that may crash if
2777 -- it doesn't see a TcTyVar, so we have to instantiate. Sigh
2778 -- ToDo: what if two of them do get unified?
2779 ; wanteds <- newDictBndrsO orig (substTheta tenv theta)
2780 ; (irreds, _) <- tryHardCheckLoop doc wanteds
2782 ; let (tv_dicts, others) = partition ok irreds
2783 ; addNoInstanceErrs others
2784 -- See Note [Exotic derived instance contexts] in TcMType
2786 ; let rev_env = zipTopTvSubst tvs (mkTyVarTys tyvars)
2787 simpl_theta = substTheta rev_env (map dictPred tv_dicts)
2788 -- This reverse-mapping is a pain, but the result
2789 -- should mention the original TyVars not TcTyVars
2791 ; return simpl_theta }
2793 doc = ptext SLIT("deriving classes for a data type")
2795 ok dict | isDict dict = validDerivPred (dictPred dict)
2800 @tcSimplifyDefault@ just checks class-type constraints, essentially;
2801 used with \tr{default} declarations. We are only interested in
2802 whether it worked or not.
2805 tcSimplifyDefault :: ThetaType -- Wanted; has no type variables in it
2808 tcSimplifyDefault theta = do
2809 wanteds <- newDictBndrsO DefaultOrigin theta
2810 (irreds, _) <- tryHardCheckLoop doc wanteds
2811 addNoInstanceErrs irreds
2815 traceTc (ptext SLIT("tcSimplifyDefault failing")) >> failM
2817 doc = ptext SLIT("default declaration")
2821 %************************************************************************
2823 \section{Errors and contexts}
2825 %************************************************************************
2827 ToDo: for these error messages, should we note the location as coming
2828 from the insts, or just whatever seems to be around in the monad just
2832 groupErrs :: ([Inst] -> TcM ()) -- Deal with one group
2833 -> [Inst] -- The offending Insts
2835 -- Group together insts with the same origin
2836 -- We want to report them together in error messages
2838 groupErrs report_err []
2840 groupErrs report_err (inst:insts)
2841 = do { do_one (inst:friends)
2842 ; groupErrs report_err others }
2844 -- (It may seem a bit crude to compare the error messages,
2845 -- but it makes sure that we combine just what the user sees,
2846 -- and it avoids need equality on InstLocs.)
2847 (friends, others) = partition is_friend insts
2848 loc_msg = showSDoc (pprInstLoc (instLoc inst))
2849 is_friend friend = showSDoc (pprInstLoc (instLoc friend)) == loc_msg
2850 do_one insts = addInstCtxt (instLoc (head insts)) (report_err insts)
2851 -- Add location and context information derived from the Insts
2853 -- Add the "arising from..." part to a message about bunch of dicts
2854 addInstLoc :: [Inst] -> Message -> Message
2855 addInstLoc insts msg = msg $$ nest 2 (pprInstArising (head insts))
2857 addTopIPErrs :: [Name] -> [Inst] -> TcM ()
2858 addTopIPErrs bndrs []
2860 addTopIPErrs bndrs ips
2861 = do { dflags <- getDOpts
2862 ; addErrTcM (tidy_env, mk_msg dflags tidy_ips) }
2864 (tidy_env, tidy_ips) = tidyInsts ips
2866 = vcat [sep [ptext SLIT("Implicit parameters escape from"),
2867 nest 2 (ptext SLIT("the monomorphic top-level binding")
2868 <> plural bndrs <+> ptext SLIT("of")
2869 <+> pprBinders bndrs <> colon)],
2870 nest 2 (vcat (map ppr_ip ips)),
2871 monomorphism_fix dflags]
2872 ppr_ip ip = pprPred (dictPred ip) <+> pprInstArising ip
2874 topIPErrs :: [Inst] -> TcM ()
2876 = groupErrs report tidy_dicts
2878 (tidy_env, tidy_dicts) = tidyInsts dicts
2879 report dicts = addErrTcM (tidy_env, mk_msg dicts)
2880 mk_msg dicts = addInstLoc dicts (ptext SLIT("Unbound implicit parameter") <>
2881 plural tidy_dicts <+> pprDictsTheta tidy_dicts)
2883 addNoInstanceErrs :: [Inst] -- Wanted (can include implications)
2885 addNoInstanceErrs insts
2886 = do { let (tidy_env, tidy_insts) = tidyInsts insts
2887 ; reportNoInstances tidy_env Nothing tidy_insts }
2891 -> Maybe (InstLoc, [Inst]) -- Context
2892 -- Nothing => top level
2893 -- Just (d,g) => d describes the construct
2895 -> [Inst] -- What is wanted (can include implications)
2898 reportNoInstances tidy_env mb_what insts
2899 = groupErrs (report_no_instances tidy_env mb_what) insts
2901 report_no_instances tidy_env mb_what insts
2902 = do { inst_envs <- tcGetInstEnvs
2903 ; let (implics, insts1) = partition isImplicInst insts
2904 (insts2, overlaps) = partitionWith (check_overlap inst_envs) insts1
2905 (eqInsts, insts3) = partition isEqInst insts2
2906 ; traceTc (text "reportNoInstances" <+> vcat
2907 [ppr insts, ppr implics, ppr insts1, ppr insts2])
2908 ; mapM_ complain_implic implics
2909 ; mapM_ (\doc -> addErrTcM (tidy_env, doc)) overlaps
2910 ; groupErrs complain_no_inst insts3
2911 ; mapM_ (addErrTcM . mk_eq_err) eqInsts
2914 complain_no_inst insts = addErrTcM (tidy_env, mk_no_inst_err insts)
2916 complain_implic inst -- Recurse!
2917 = reportNoInstances tidy_env
2918 (Just (tci_loc inst, tci_given inst))
2921 check_overlap :: (InstEnv,InstEnv) -> Inst -> Either Inst SDoc
2922 -- Right msg => overlap message
2923 -- Left inst => no instance
2924 check_overlap inst_envs wanted
2925 | not (isClassDict wanted) = Left wanted
2927 = case lookupInstEnv inst_envs clas tys of
2928 ([], _) -> Left wanted -- No match
2929 -- The case of exactly one match and no unifiers means a
2930 -- successful lookup. That can't happen here, because dicts
2931 -- only end up here if they didn't match in Inst.lookupInst
2933 | debugIsOn -> pprPanic "reportNoInstance" (ppr wanted)
2934 res -> Right (mk_overlap_msg wanted res)
2936 (clas,tys) = getDictClassTys wanted
2938 mk_overlap_msg dict (matches, unifiers)
2939 = ASSERT( not (null matches) )
2940 vcat [ addInstLoc [dict] ((ptext SLIT("Overlapping instances for")
2941 <+> pprPred (dictPred dict))),
2942 sep [ptext SLIT("Matching instances") <> colon,
2943 nest 2 (vcat [pprInstances ispecs, pprInstances unifiers])],
2944 if not (isSingleton matches)
2945 then -- Two or more matches
2947 else -- One match, plus some unifiers
2948 ASSERT( not (null unifiers) )
2949 parens (vcat [ptext SLIT("The choice depends on the instantiation of") <+>
2950 quotes (pprWithCommas ppr (varSetElems (tyVarsOfInst dict))),
2951 ptext SLIT("To pick the first instance above, use -fallow-incoherent-instances"),
2952 ptext SLIT("when compiling the other instance declarations")])]
2954 ispecs = [ispec | (ispec, _) <- matches]
2956 mk_eq_err :: Inst -> (TidyEnv, SDoc)
2957 mk_eq_err inst = misMatchMsg tidy_env (eqInstTys inst)
2959 mk_no_inst_err insts
2960 | null insts = empty
2962 | Just (loc, givens) <- mb_what, -- Nested (type signatures, instance decls)
2963 not (isEmptyVarSet (tyVarsOfInsts insts))
2964 = vcat [ addInstLoc insts $
2965 sep [ ptext SLIT("Could not deduce") <+> pprDictsTheta insts
2966 , nest 2 $ ptext SLIT("from the context") <+> pprDictsTheta givens]
2967 , show_fixes (fix1 loc : fixes2) ]
2969 | otherwise -- Top level
2970 = vcat [ addInstLoc insts $
2971 ptext SLIT("No instance") <> plural insts
2972 <+> ptext SLIT("for") <+> pprDictsTheta insts
2973 , show_fixes fixes2 ]
2976 fix1 loc = sep [ ptext SLIT("add") <+> pprDictsTheta insts
2977 <+> ptext SLIT("to the context of"),
2978 nest 2 (ppr (instLocOrigin loc)) ]
2979 -- I'm not sure it helps to add the location
2980 -- nest 2 (ptext SLIT("at") <+> ppr (instLocSpan loc)) ]
2982 fixes2 | null instance_dicts = []
2983 | otherwise = [sep [ptext SLIT("add an instance declaration for"),
2984 pprDictsTheta instance_dicts]]
2985 instance_dicts = [d | d <- insts, isClassDict d, not (isTyVarDict d)]
2986 -- Insts for which it is worth suggesting an adding an instance declaration
2987 -- Exclude implicit parameters, and tyvar dicts
2989 show_fixes :: [SDoc] -> SDoc
2990 show_fixes [] = empty
2991 show_fixes (f:fs) = sep [ptext SLIT("Possible fix:"),
2992 nest 2 (vcat (f : map (ptext SLIT("or") <+>) fs))]
2994 addTopAmbigErrs dicts
2995 -- Divide into groups that share a common set of ambiguous tyvars
2996 = ifErrsM (return ()) $ -- Only report ambiguity if no other errors happened
2997 -- See Note [Avoiding spurious errors]
2998 mapM_ report (equivClasses cmp [(d, tvs_of d) | d <- tidy_dicts])
3000 (tidy_env, tidy_dicts) = tidyInsts dicts
3002 tvs_of :: Inst -> [TcTyVar]
3003 tvs_of d = varSetElems (tyVarsOfInst d)
3004 cmp (_,tvs1) (_,tvs2) = tvs1 `compare` tvs2
3006 report :: [(Inst,[TcTyVar])] -> TcM ()
3007 report pairs@((inst,tvs) : _) = do -- The pairs share a common set of ambiguous tyvars
3008 (tidy_env, mono_msg) <- mkMonomorphismMsg tidy_env tvs
3009 setSrcSpan (instSpan inst) $
3010 -- the location of the first one will do for the err message
3011 addErrTcM (tidy_env, msg $$ mono_msg)
3013 dicts = map fst pairs
3014 msg = sep [text "Ambiguous type variable" <> plural tvs <+>
3015 pprQuotedList tvs <+> in_msg,
3016 nest 2 (pprDictsInFull dicts)]
3017 in_msg = text "in the constraint" <> plural dicts <> colon
3018 report [] = panic "addTopAmbigErrs"
3021 mkMonomorphismMsg :: TidyEnv -> [TcTyVar] -> TcM (TidyEnv, Message)
3022 -- There's an error with these Insts; if they have free type variables
3023 -- it's probably caused by the monomorphism restriction.
3024 -- Try to identify the offending variable
3025 -- ASSUMPTION: the Insts are fully zonked
3026 mkMonomorphismMsg tidy_env inst_tvs
3027 = do { dflags <- getDOpts
3028 ; (tidy_env, docs) <- findGlobals (mkVarSet inst_tvs) tidy_env
3029 ; return (tidy_env, mk_msg dflags docs) }
3031 mk_msg _ _ | any isRuntimeUnk inst_tvs
3032 = vcat [ptext SLIT("Cannot resolve unknown runtime types:") <+>
3033 (pprWithCommas ppr inst_tvs),
3034 ptext SLIT("Use :print or :force to determine these types")]
3035 mk_msg _ [] = ptext SLIT("Probable fix: add a type signature that fixes these type variable(s)")
3036 -- This happens in things like
3037 -- f x = show (read "foo")
3038 -- where monomorphism doesn't play any role
3040 = vcat [ptext SLIT("Possible cause: the monomorphism restriction applied to the following:"),
3042 monomorphism_fix dflags]
3044 monomorphism_fix :: DynFlags -> SDoc
3045 monomorphism_fix dflags
3046 = ptext SLIT("Probable fix:") <+> vcat
3047 [ptext SLIT("give these definition(s) an explicit type signature"),
3048 if dopt Opt_MonomorphismRestriction dflags
3049 then ptext SLIT("or use -fno-monomorphism-restriction")
3050 else empty] -- Only suggest adding "-fno-monomorphism-restriction"
3051 -- if it is not already set!
3053 warnDefault ups default_ty = do
3054 warn_flag <- doptM Opt_WarnTypeDefaults
3055 addInstCtxt (instLoc (head (dicts))) (warnTc warn_flag warn_msg)
3057 dicts = [d | (d,_,_) <- ups]
3060 (_, tidy_dicts) = tidyInsts dicts
3061 warn_msg = vcat [ptext SLIT("Defaulting the following constraint(s) to type") <+>
3062 quotes (ppr default_ty),
3063 pprDictsInFull tidy_dicts]
3065 reduceDepthErr n stack
3066 = vcat [ptext SLIT("Context reduction stack overflow; size =") <+> int n,
3067 ptext SLIT("Use -fcontext-stack=N to increase stack size to N"),
3068 nest 4 (pprStack stack)]
3070 pprStack stack = vcat (map pprInstInFull stack)