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,
25 bindInstsOfLocalFuns, bindIrreds,
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' <- mappM 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 <- mappM 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_dictsin 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 -> [CoVar] -> Refinement
925 -> [TcTyVar] -- Quantify over these
928 -> TcM TcDictBinds -- Bindings
929 tcSimplifyCheckPat loc co_vars reft qtvs givens wanteds
930 = ASSERT( all isTcTyVar qtvs && all isSkolemTyVar qtvs )
931 do { traceTc (text "tcSimplifyCheckPat")
932 ; (irreds, binds) <- gentleCheckLoop loc givens wanteds
933 ; implic_bind <- bindIrredsR loc qtvs co_vars reft
935 ; return (binds `unionBags` implic_bind) }
937 -----------------------------------------------------------
938 bindIrreds :: InstLoc -> [TcTyVar]
941 bindIrreds loc qtvs givens irreds
942 = bindIrredsR loc qtvs [] emptyRefinement givens irreds
944 bindIrredsR :: InstLoc -> [TcTyVar] -> [CoVar]
945 -> Refinement -> [Inst] -> [Inst]
947 -- Make a binding that binds 'irreds', by generating an implication
948 -- constraint for them, *and* throwing the constraint into the LIE
949 bindIrredsR loc qtvs co_vars reft givens irreds
953 = do { let givens' = filter isDict givens
954 -- The givens can include methods
955 -- See Note [Pruning the givens in an implication constraint]
957 -- If there are no 'givens' *and* the refinement is empty
958 -- (the refinement is like more givens), then it's safe to
959 -- partition the 'wanteds' by their qtvs, thereby trimming irreds
960 -- See Note [Freeness and implications]
961 ; irreds' <- if null givens' && isEmptyRefinement reft
963 { let qtv_set = mkVarSet qtvs
964 (frees, real_irreds) = partition (isFreeWrtTyVars qtv_set) irreds
966 ; return real_irreds }
969 ; let all_tvs = qtvs ++ co_vars -- Abstract over all these
970 ; (implics, bind) <- makeImplicationBind loc all_tvs reft givens' irreds'
971 -- This call does the real work
972 -- If irreds' is empty, it does something sensible
977 makeImplicationBind :: InstLoc -> [TcTyVar] -> Refinement
979 -> TcM ([Inst], TcDictBinds)
980 -- Make a binding that binds 'irreds', by generating an implication
981 -- constraint for them, *and* throwing the constraint into the LIE
982 -- The binding looks like
983 -- (ir1, .., irn) = f qtvs givens
984 -- where f is (evidence for) the new implication constraint
985 -- f :: forall qtvs. {reft} givens => (ir1, .., irn)
986 -- qtvs includes coercion variables
988 -- This binding must line up the 'rhs' in reduceImplication
989 makeImplicationBind loc all_tvs reft
990 givens -- Guaranteed all Dicts (TOMDO: true?)
992 | null irreds -- If there are no irreds, we are done
993 = return ([], emptyBag)
994 | otherwise -- Otherwise we must generate a binding
995 = do { uniq <- newUnique
996 ; span <- getSrcSpanM
997 ; let (eq_givens, dict_givens) = partition isEqInst givens
998 eq_tyvar_cos = map TyVarTy $ uniqSetToList $ tyVarsOfTypes $ map eqInstType eq_givens
999 ; let name = mkInternalName uniq (mkVarOcc "ic") span
1000 implic_inst = ImplicInst { tci_name = name, tci_reft = reft,
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) <.> mkWpTyApps eq_tyvar_cos <.> mkWpTyApps (mkTyVarTys all_tvs)
1012 bind | [dict_irred_id] <- dict_irred_ids = VarBind dict_irred_id rhs
1013 | otherwise = PatBind { pat_lhs = L span pat,
1014 pat_rhs = unguardedGRHSs rhs,
1015 pat_rhs_ty = tup_ty,
1016 bind_fvs = placeHolderNames }
1017 ; -- pprTrace "Make implic inst" (ppr (implic_inst,irreds,dict_irreds,tup_ty)) $
1018 return ([implic_inst], unitBag (L span bind)) }
1020 -----------------------------------------------------------
1021 tryHardCheckLoop :: SDoc
1023 -> TcM ([Inst], TcDictBinds)
1025 tryHardCheckLoop doc wanteds
1026 = do { (irreds,binds,_) <- checkLoop (mkRedEnv doc try_me []) wanteds
1027 ; return (irreds,binds)
1030 try_me inst = ReduceMe AddSCs
1031 -- Here's the try-hard bit
1033 -----------------------------------------------------------
1034 gentleCheckLoop :: InstLoc
1037 -> TcM ([Inst], TcDictBinds)
1039 gentleCheckLoop inst_loc givens wanteds
1040 = do { (irreds,binds,_) <- checkLoop env wanteds
1041 ; return (irreds,binds)
1044 env = mkRedEnv (pprInstLoc inst_loc) try_me givens
1046 try_me inst | isMethodOrLit inst = ReduceMe AddSCs
1048 -- When checking against a given signature
1049 -- we MUST be very gentle: Note [Check gently]
1053 ~~~~~~~~~~~~~~~~~~~~
1054 We have to very careful about not simplifying too vigorously
1059 f :: Show b => T b -> b
1060 f (MkT x) = show [x]
1062 Inside the pattern match, which binds (a:*, x:a), we know that
1064 Hence we have a dictionary for Show [a] available; and indeed we
1065 need it. We are going to build an implication contraint
1066 forall a. (b~[a]) => Show [a]
1067 Later, we will solve this constraint using the knowledg e(Show b)
1069 But we MUST NOT reduce (Show [a]) to (Show a), else the whole
1070 thing becomes insoluble. So we simplify gently (get rid of literals
1071 and methods only, plus common up equal things), deferring the real
1072 work until top level, when we solve the implication constraint
1073 with tryHardCheckLooop.
1077 -----------------------------------------------------------
1080 -> TcM ([Inst], TcDictBinds,
1081 [Inst]) -- needed givens
1082 -- Precondition: givens are completely rigid
1083 -- Postcondition: returned Insts are zonked
1085 checkLoop env wanteds
1087 where go env wanteds needed_givens
1088 = do { -- We do need to zonk the givens; cf Note [Zonking RedEnv]
1089 ; env' <- zonkRedEnv env
1090 ; wanteds' <- zonkInsts wanteds
1092 ; (improved, binds, irreds, more_needed_givens) <- reduceContext env' wanteds'
1094 ; let all_needed_givens = needed_givens ++ more_needed_givens
1096 ; if not improved then
1097 return (irreds, binds, all_needed_givens)
1100 -- If improvement did some unification, we go round again.
1101 -- We start again with irreds, not wanteds
1102 -- Using an instance decl might have introduced a fresh type variable
1103 -- which might have been unified, so we'd get an infinite loop
1104 -- if we started again with wanteds! See Note [LOOP]
1105 { (irreds1, binds1, all_needed_givens1) <- go env' irreds all_needed_givens
1106 ; return (irreds1, binds `unionBags` binds1, all_needed_givens1) } }
1109 Note [Zonking RedEnv]
1110 ~~~~~~~~~~~~~~~~~~~~~
1111 It might appear as if the givens in RedEnv are always rigid, but that is not
1112 necessarily the case for programs involving higher-rank types that have class
1113 contexts constraining the higher-rank variables. An example from tc237 in the
1116 class Modular s a | s -> a
1118 wim :: forall a w. Integral a
1119 => a -> (forall s. Modular s a => M s w) -> w
1120 wim i k = error "urk"
1122 test5 :: (Modular s a, Integral a) => M s a
1125 test4 = wim 4 test4'
1127 Notice how the variable 'a' of (Modular s a) in the rank-2 type of wim is
1128 quantified further outside. When type checking test4, we have to check
1129 whether the signature of test5 is an instance of
1131 (forall s. Modular s a => M s w)
1133 Consequently, we will get (Modular s t_a), where t_a is a TauTv into the
1136 Given the FD of Modular in this example, class improvement will instantiate
1137 t_a to 'a', where 'a' is the skolem from test5's signatures (due to the
1138 Modular s a predicate in that signature). If we don't zonk (Modular s t_a) in
1139 the givens, we will get into a loop as improveOne uses the unification engine
1140 TcGadt.tcUnifyTys, which doesn't know about mutable type variables.
1145 class If b t e r | b t e -> r
1148 class Lte a b c | a b -> c where lte :: a -> b -> c
1150 instance (Lte a b l,If l b a c) => Max a b c
1152 Wanted: Max Z (S x) y
1154 Then we'll reduce using the Max instance to:
1155 (Lte Z (S x) l, If l (S x) Z y)
1156 and improve by binding l->T, after which we can do some reduction
1157 on both the Lte and If constraints. What we *can't* do is start again
1158 with (Max Z (S x) y)!
1162 %************************************************************************
1164 tcSimplifySuperClasses
1166 %************************************************************************
1168 Note [SUPERCLASS-LOOP 1]
1169 ~~~~~~~~~~~~~~~~~~~~~~~~
1170 We have to be very, very careful when generating superclasses, lest we
1171 accidentally build a loop. Here's an example:
1175 class S a => C a where { opc :: a -> a }
1176 class S b => D b where { opd :: b -> b }
1178 instance C Int where
1181 instance D Int where
1184 From (instance C Int) we get the constraint set {ds1:S Int, dd:D Int}
1185 Simplifying, we may well get:
1186 $dfCInt = :C ds1 (opd dd)
1189 Notice that we spot that we can extract ds1 from dd.
1191 Alas! Alack! We can do the same for (instance D Int):
1193 $dfDInt = :D ds2 (opc dc)
1197 And now we've defined the superclass in terms of itself.
1199 Solution: never generate a superclass selectors at all when
1200 satisfying the superclass context of an instance declaration.
1202 Two more nasty cases are in
1207 tcSimplifySuperClasses
1212 tcSimplifySuperClasses loc givens sc_wanteds
1213 = do { traceTc (text "tcSimplifySuperClasses")
1214 ; (irreds,binds1,_) <- checkLoop env sc_wanteds
1215 ; let (tidy_env, tidy_irreds) = tidyInsts irreds
1216 ; reportNoInstances tidy_env (Just (loc, givens)) tidy_irreds
1219 env = mkRedEnv (pprInstLoc loc) try_me givens
1220 try_me inst = ReduceMe NoSCs
1221 -- Like tryHardCheckLoop, but with NoSCs
1225 %************************************************************************
1227 \subsection{tcSimplifyRestricted}
1229 %************************************************************************
1231 tcSimplifyRestricted infers which type variables to quantify for a
1232 group of restricted bindings. This isn't trivial.
1235 We want to quantify over a to get id :: forall a. a->a
1238 We do not want to quantify over a, because there's an Eq a
1239 constraint, so we get eq :: a->a->Bool (notice no forall)
1242 RHS has type 'tau', whose free tyvars are tau_tvs
1243 RHS has constraints 'wanteds'
1246 Quantify over (tau_tvs \ ftvs(wanteds))
1247 This is bad. The constraints may contain (Monad (ST s))
1248 where we have instance Monad (ST s) where...
1249 so there's no need to be monomorphic in s!
1251 Also the constraint might be a method constraint,
1252 whose type mentions a perfectly innocent tyvar:
1253 op :: Num a => a -> b -> a
1254 Here, b is unconstrained. A good example would be
1256 We want to infer the polymorphic type
1257 foo :: forall b. b -> b
1260 Plan B (cunning, used for a long time up to and including GHC 6.2)
1261 Step 1: Simplify the constraints as much as possible (to deal
1262 with Plan A's problem). Then set
1263 qtvs = tau_tvs \ ftvs( simplify( wanteds ) )
1265 Step 2: Now simplify again, treating the constraint as 'free' if
1266 it does not mention qtvs, and trying to reduce it otherwise.
1267 The reasons for this is to maximise sharing.
1269 This fails for a very subtle reason. Suppose that in the Step 2
1270 a constraint (Foo (Succ Zero) (Succ Zero) b) gets thrown upstairs as 'free'.
1271 In the Step 1 this constraint might have been simplified, perhaps to
1272 (Foo Zero Zero b), AND THEN THAT MIGHT BE IMPROVED, to bind 'b' to 'T'.
1273 This won't happen in Step 2... but that in turn might prevent some other
1274 constraint (Baz [a] b) being simplified (e.g. via instance Baz [a] T where {..})
1275 and that in turn breaks the invariant that no constraints are quantified over.
1277 Test typecheck/should_compile/tc177 (which failed in GHC 6.2) demonstrates
1282 Step 1: Simplify the constraints as much as possible (to deal
1283 with Plan A's problem). Then set
1284 qtvs = tau_tvs \ ftvs( simplify( wanteds ) )
1285 Return the bindings from Step 1.
1288 A note about Plan C (arising from "bug" reported by George Russel March 2004)
1291 instance (HasBinary ty IO) => HasCodedValue ty
1293 foo :: HasCodedValue a => String -> IO a
1295 doDecodeIO :: HasCodedValue a => () -> () -> IO a
1296 doDecodeIO codedValue view
1297 = let { act = foo "foo" } in act
1299 You might think this should work becuase the call to foo gives rise to a constraint
1300 (HasCodedValue t), which can be satisfied by the type sig for doDecodeIO. But the
1301 restricted binding act = ... calls tcSimplifyRestricted, and PlanC simplifies the
1302 constraint using the (rather bogus) instance declaration, and now we are stuffed.
1304 I claim this is not really a bug -- but it bit Sergey as well as George. So here's
1308 Plan D (a variant of plan B)
1309 Step 1: Simplify the constraints as much as possible (to deal
1310 with Plan A's problem), BUT DO NO IMPROVEMENT. Then set
1311 qtvs = tau_tvs \ ftvs( simplify( wanteds ) )
1313 Step 2: Now simplify again, treating the constraint as 'free' if
1314 it does not mention qtvs, and trying to reduce it otherwise.
1316 The point here is that it's generally OK to have too few qtvs; that is,
1317 to make the thing more monomorphic than it could be. We don't want to
1318 do that in the common cases, but in wierd cases it's ok: the programmer
1319 can always add a signature.
1321 Too few qtvs => too many wanteds, which is what happens if you do less
1326 tcSimplifyRestricted -- Used for restricted binding groups
1327 -- i.e. ones subject to the monomorphism restriction
1330 -> [Name] -- Things bound in this group
1331 -> TcTyVarSet -- Free in the type of the RHSs
1332 -> [Inst] -- Free in the RHSs
1333 -> TcM ([TyVar], -- Tyvars to quantify (zonked and quantified)
1334 TcDictBinds) -- Bindings
1335 -- tcSimpifyRestricted returns no constraints to
1336 -- quantify over; by definition there are none.
1337 -- They are all thrown back in the LIE
1339 tcSimplifyRestricted doc top_lvl bndrs tau_tvs wanteds
1340 -- Zonk everything in sight
1341 = do { traceTc (text "tcSimplifyRestricted")
1342 ; wanteds' <- zonkInsts wanteds
1344 -- 'ReduceMe': Reduce as far as we can. Don't stop at
1345 -- dicts; the idea is to get rid of as many type
1346 -- variables as possible, and we don't want to stop
1347 -- at (say) Monad (ST s), because that reduces
1348 -- immediately, with no constraint on s.
1350 -- BUT do no improvement! See Plan D above
1351 -- HOWEVER, some unification may take place, if we instantiate
1352 -- a method Inst with an equality constraint
1353 ; let env = mkNoImproveRedEnv doc (\i -> ReduceMe AddSCs)
1354 ; (_imp, _binds, constrained_dicts, _) <- reduceContext env wanteds'
1356 -- Next, figure out the tyvars we will quantify over
1357 ; tau_tvs' <- zonkTcTyVarsAndFV (varSetElems tau_tvs)
1358 ; gbl_tvs' <- tcGetGlobalTyVars
1359 ; constrained_dicts' <- zonkInsts constrained_dicts
1361 ; let qtvs1 = tau_tvs' `minusVarSet` oclose (fdPredsOfInsts constrained_dicts) gbl_tvs'
1362 -- As in tcSimplifyInfer
1364 -- Do not quantify over constrained type variables:
1365 -- this is the monomorphism restriction
1366 constrained_tvs' = tyVarsOfInsts constrained_dicts'
1367 qtvs = qtvs1 `minusVarSet` constrained_tvs'
1368 pp_bndrs = pprWithCommas (quotes . ppr) bndrs
1371 ; warn_mono <- doptM Opt_WarnMonomorphism
1372 ; warnTc (warn_mono && (constrained_tvs' `intersectsVarSet` qtvs1))
1373 (vcat[ ptext SLIT("the Monomorphism Restriction applies to the binding")
1374 <> plural bndrs <+> ptext SLIT("for") <+> pp_bndrs,
1375 ptext SLIT("Consider giving a type signature for") <+> pp_bndrs])
1377 ; traceTc (text "tcSimplifyRestricted" <+> vcat [
1378 pprInsts wanteds, pprInsts constrained_dicts',
1380 ppr constrained_tvs', ppr tau_tvs', ppr qtvs ])
1382 -- The first step may have squashed more methods than
1383 -- necessary, so try again, this time more gently, knowing the exact
1384 -- set of type variables to quantify over.
1386 -- We quantify only over constraints that are captured by qtvs;
1387 -- these will just be a subset of non-dicts. This in contrast
1388 -- to normal inference (using isFreeWhenInferring) in which we quantify over
1389 -- all *non-inheritable* constraints too. This implements choice
1390 -- (B) under "implicit parameter and monomorphism" above.
1392 -- Remember that we may need to do *some* simplification, to
1393 -- (for example) squash {Monad (ST s)} into {}. It's not enough
1394 -- just to float all constraints
1396 -- At top level, we *do* squash methods becuase we want to
1397 -- expose implicit parameters to the test that follows
1398 ; let is_nested_group = isNotTopLevel top_lvl
1399 try_me inst | isFreeWrtTyVars qtvs inst,
1400 (is_nested_group || isDict inst) = Stop
1401 | otherwise = ReduceMe AddSCs
1402 env = mkNoImproveRedEnv doc try_me
1403 ; (_imp, binds, irreds, _) <- reduceContext env wanteds'
1405 -- See "Notes on implicit parameters, Question 4: top level"
1406 ; ASSERT( all (isFreeWrtTyVars qtvs) irreds ) -- None should be captured
1407 if is_nested_group then
1409 else do { let (bad_ips, non_ips) = partition isIPDict irreds
1410 ; addTopIPErrs bndrs bad_ips
1411 ; extendLIEs non_ips }
1413 ; qtvs' <- zonkQuantifiedTyVars (varSetElems qtvs)
1414 ; return (qtvs', binds) }
1418 %************************************************************************
1422 %************************************************************************
1424 On the LHS of transformation rules we only simplify methods and constants,
1425 getting dictionaries. We want to keep all of them unsimplified, to serve
1426 as the available stuff for the RHS of the rule.
1428 Example. Consider the following left-hand side of a rule
1430 f (x == y) (y > z) = ...
1432 If we typecheck this expression we get constraints
1434 d1 :: Ord a, d2 :: Eq a
1436 We do NOT want to "simplify" to the LHS
1438 forall x::a, y::a, z::a, d1::Ord a.
1439 f ((==) (eqFromOrd d1) x y) ((>) d1 y z) = ...
1443 forall x::a, y::a, z::a, d1::Ord a, d2::Eq a.
1444 f ((==) d2 x y) ((>) d1 y z) = ...
1446 Here is another example:
1448 fromIntegral :: (Integral a, Num b) => a -> b
1449 {-# RULES "foo" fromIntegral = id :: Int -> Int #-}
1451 In the rule, a=b=Int, and Num Int is a superclass of Integral Int. But
1452 we *dont* want to get
1454 forall dIntegralInt.
1455 fromIntegral Int Int dIntegralInt (scsel dIntegralInt) = id Int
1457 because the scsel will mess up RULE matching. Instead we want
1459 forall dIntegralInt, dNumInt.
1460 fromIntegral Int Int dIntegralInt dNumInt = id Int
1464 g (x == y) (y == z) = ..
1466 where the two dictionaries are *identical*, we do NOT WANT
1468 forall x::a, y::a, z::a, d1::Eq a
1469 f ((==) d1 x y) ((>) d1 y z) = ...
1471 because that will only match if the dict args are (visibly) equal.
1472 Instead we want to quantify over the dictionaries separately.
1474 In short, tcSimplifyRuleLhs must *only* squash LitInst and MethInts, leaving
1475 all dicts unchanged, with absolutely no sharing. It's simpler to do this
1476 from scratch, rather than further parameterise simpleReduceLoop etc
1479 tcSimplifyRuleLhs :: [Inst] -> TcM ([Inst], TcDictBinds)
1480 tcSimplifyRuleLhs wanteds
1481 = go [] emptyBag wanteds
1484 = return (dicts, binds)
1485 go dicts binds (w:ws)
1487 = go (w:dicts) binds ws
1489 = do { w' <- zonkInst w -- So that (3::Int) does not generate a call
1490 -- to fromInteger; this looks fragile to me
1491 ; lookup_result <- lookupSimpleInst w'
1492 ; case lookup_result of
1494 go dicts (addInstToDictBind binds w rhs) (ws' ++ ws)
1495 NoInstance -> pprPanic "tcSimplifyRuleLhs" (ppr w)
1499 tcSimplifyBracket is used when simplifying the constraints arising from
1500 a Template Haskell bracket [| ... |]. We want to check that there aren't
1501 any constraints that can't be satisfied (e.g. Show Foo, where Foo has no
1502 Show instance), but we aren't otherwise interested in the results.
1503 Nor do we care about ambiguous dictionaries etc. We will type check
1504 this bracket again at its usage site.
1507 tcSimplifyBracket :: [Inst] -> TcM ()
1508 tcSimplifyBracket wanteds
1509 = do { tryHardCheckLoop doc wanteds
1512 doc = text "tcSimplifyBracket"
1516 %************************************************************************
1518 \subsection{Filtering at a dynamic binding}
1520 %************************************************************************
1525 we must discharge all the ?x constraints from B. We also do an improvement
1526 step; if we have ?x::t1 and ?x::t2 we must unify t1, t2.
1528 Actually, the constraints from B might improve the types in ?x. For example
1530 f :: (?x::Int) => Char -> Char
1533 then the constraint (?x::Int) arising from the call to f will
1534 force the binding for ?x to be of type Int.
1537 tcSimplifyIPs :: [Inst] -- The implicit parameters bound here
1540 -- We need a loop so that we do improvement, and then
1541 -- (next time round) generate a binding to connect the two
1543 -- Here the two ?x's have different types, and improvement
1544 -- makes them the same.
1546 tcSimplifyIPs given_ips wanteds
1547 = do { wanteds' <- zonkInsts wanteds
1548 ; given_ips' <- zonkInsts given_ips
1549 -- Unusually for checking, we *must* zonk the given_ips
1551 ; let env = mkRedEnv doc try_me given_ips'
1552 ; (improved, binds, irreds, _) <- reduceContext env wanteds'
1554 ; if not improved then
1555 ASSERT( all is_free irreds )
1556 do { extendLIEs irreds
1559 tcSimplifyIPs given_ips wanteds }
1561 doc = text "tcSimplifyIPs" <+> ppr given_ips
1562 ip_set = mkNameSet (ipNamesOfInsts given_ips)
1563 is_free inst = isFreeWrtIPs ip_set inst
1565 -- Simplify any methods that mention the implicit parameter
1566 try_me inst | is_free inst = Stop
1567 | otherwise = ReduceMe NoSCs
1571 %************************************************************************
1573 \subsection[binds-for-local-funs]{@bindInstsOfLocalFuns@}
1575 %************************************************************************
1577 When doing a binding group, we may have @Insts@ of local functions.
1578 For example, we might have...
1580 let f x = x + 1 -- orig local function (overloaded)
1581 f.1 = f Int -- two instances of f
1586 The point is: we must drop the bindings for @f.1@ and @f.2@ here,
1587 where @f@ is in scope; those @Insts@ must certainly not be passed
1588 upwards towards the top-level. If the @Insts@ were binding-ified up
1589 there, they would have unresolvable references to @f@.
1591 We pass in an @init_lie@ of @Insts@ and a list of locally-bound @Ids@.
1592 For each method @Inst@ in the @init_lie@ that mentions one of the
1593 @Ids@, we create a binding. We return the remaining @Insts@ (in an
1594 @LIE@), as well as the @HsBinds@ generated.
1597 bindInstsOfLocalFuns :: [Inst] -> [TcId] -> TcM TcDictBinds
1598 -- Simlifies only MethodInsts, and generate only bindings of form
1600 -- We're careful not to even generate bindings of the form
1602 -- You'd think that'd be fine, but it interacts with what is
1603 -- arguably a bug in Match.tidyEqnInfo (see notes there)
1605 bindInstsOfLocalFuns wanteds local_ids
1606 | null overloaded_ids
1608 = extendLIEs wanteds `thenM_`
1609 returnM emptyLHsBinds
1612 = do { (irreds, binds,_) <- checkLoop env for_me
1613 ; extendLIEs not_for_me
1617 env = mkRedEnv doc try_me []
1618 doc = text "bindInsts" <+> ppr local_ids
1619 overloaded_ids = filter is_overloaded local_ids
1620 is_overloaded id = isOverloadedTy (idType id)
1621 (for_me, not_for_me) = partition (isMethodFor overloaded_set) wanteds
1623 overloaded_set = mkVarSet overloaded_ids -- There can occasionally be a lot of them
1624 -- so it's worth building a set, so that
1625 -- lookup (in isMethodFor) is faster
1626 try_me inst | isMethod inst = ReduceMe NoSCs
1631 %************************************************************************
1633 \subsection{Data types for the reduction mechanism}
1635 %************************************************************************
1637 The main control over context reduction is here
1641 = RedEnv { red_doc :: SDoc -- The context
1642 , red_try_me :: Inst -> WhatToDo
1643 , red_improve :: Bool -- True <=> do improvement
1644 , red_givens :: [Inst] -- All guaranteed rigid
1646 -- but see Note [Rigidity]
1647 , red_stack :: (Int, [Inst]) -- Recursion stack (for err msg)
1648 -- See Note [RedStack]
1652 -- The red_givens are rigid so far as cmpInst is concerned.
1653 -- There is one case where they are not totally rigid, namely in tcSimplifyIPs
1654 -- let ?x = e in ...
1655 -- Here, the given is (?x::a), where 'a' is not necy a rigid type
1656 -- But that doesn't affect the comparison, which is based only on mame.
1659 -- The red_stack pair (n,insts) pair is just used for error reporting.
1660 -- 'n' is always the depth of the stack.
1661 -- The 'insts' is the stack of Insts being reduced: to produce X
1662 -- I had to produce Y, to produce Y I had to produce Z, and so on.
1665 mkRedEnv :: SDoc -> (Inst -> WhatToDo) -> [Inst] -> RedEnv
1666 mkRedEnv doc try_me givens
1667 = RedEnv { red_doc = doc, red_try_me = try_me,
1668 red_givens = givens, red_stack = (0,[]),
1669 red_improve = True }
1671 mkNoImproveRedEnv :: SDoc -> (Inst -> WhatToDo) -> RedEnv
1672 -- Do not do improvement; no givens
1673 mkNoImproveRedEnv doc try_me
1674 = RedEnv { red_doc = doc, red_try_me = try_me,
1675 red_givens = [], red_stack = (0,[]),
1676 red_improve = True }
1679 = ReduceMe WantSCs -- Try to reduce this
1680 -- If there's no instance, add the inst to the
1681 -- irreductible ones, but don't produce an error
1682 -- message of any kind.
1683 -- It might be quite legitimate such as (Eq a)!
1685 | Stop -- Return as irreducible unless it can
1686 -- be reduced to a constant in one step
1687 -- Do not add superclasses; see
1689 data WantSCs = NoSCs | AddSCs -- Tells whether we should add the superclasses
1690 -- of a predicate when adding it to the avails
1691 -- The reason for this flag is entirely the super-class loop problem
1692 -- Note [SUPER-CLASS LOOP 1]
1694 zonkRedEnv :: RedEnv -> TcM RedEnv
1696 = do { givens' <- mappM zonkInst (red_givens env)
1697 ; return $ env {red_givens = givens'}
1702 %************************************************************************
1704 \subsection[reduce]{@reduce@}
1706 %************************************************************************
1708 Note [Ancestor Equalities]
1709 ~~~~~~~~~~~~~~~~~~~~~~~~~~
1710 During context reduction, we add to the wanted equalities also those
1711 equalities that (transitively) occur in superclass contexts of wanted
1712 class constraints. Consider the following code
1714 class a ~ Int => C a
1717 If (C a) is wanted, we want to add (a ~ Int), which will be discharged by
1718 substituting Int for a. Hence, we ultimately want (C Int), which we
1719 discharge with the explicit instance.
1722 reduceContext :: RedEnv
1724 -> TcM (ImprovementDone,
1725 TcDictBinds, -- Dictionary bindings
1726 [Inst], -- Irreducible
1727 [Inst]) -- Needed givens
1729 reduceContext env wanteds
1730 = do { traceTc (text "reduceContext" <+> (vcat [
1731 text "----------------------",
1733 text "given" <+> ppr (red_givens env),
1734 text "wanted" <+> ppr wanteds,
1735 text "----------------------"
1738 ; let givens = red_givens env
1739 (given_eqs0, given_dicts0) = partition isEqInst givens
1740 (wanted_eqs0, wanted_dicts) = partition isEqInst wanteds
1742 -- We want to add as wanted equalities those that (transitively)
1743 -- occur in superclass contexts of wanted class constraints.
1744 -- See Note [Ancestor Equalities]
1745 ; ancestor_eqs <- ancestorEqualities wanted_dicts
1746 ; let wanted_eqs = wanted_eqs0 ++ ancestor_eqs
1747 ; traceTc $ text "reduceContext: ancestor eqs" <+> ppr ancestor_eqs
1749 -- 1. Normalise the *given* *equality* constraints
1750 ; (given_eqs, eliminate_skolems) <- normaliseGivenEqs given_eqs0
1752 -- 2. Normalise the *given* *dictionary* constraints
1753 -- wrt. the toplevel and given equations
1754 ; (given_dicts, given_binds) <- normaliseGivenDicts given_eqs
1757 -- 3. Solve the *wanted* *equation* constraints
1758 ; eq_irreds0 <- solveWantedEqs given_eqs wanted_eqs
1760 -- 4. Normalise the *wanted* equality constraints with respect to
1762 ; eq_irreds <- normaliseWantedEqs eq_irreds0
1764 -- 5. Build the Avail mapping from "given_dicts"
1765 ; init_state <- foldlM addGiven emptyAvails given_dicts
1767 -- 6. Solve the *wanted* *dictionary* constraints
1768 -- This may expose some further equational constraints...
1769 ; wanted_dicts' <- zonkInsts wanted_dicts
1770 ; avails <- reduceList env wanted_dicts' init_state
1771 ; (binds, irreds0, needed_givens) <- extractResults avails wanted_dicts'
1772 ; traceTc $ text "reduceContext extractresults" <+> vcat
1773 [ppr avails,ppr wanted_dicts',ppr binds,ppr needed_givens]
1775 -- 7. Normalise the *wanted* *dictionary* constraints
1776 -- wrt. the toplevel and given equations
1777 ; (irreds1,normalise_binds1) <- normaliseWantedDicts given_eqs irreds0
1779 -- 8. Substitute the wanted *equations* in the wanted *dictionaries*
1780 ; (irreds,normalise_binds2) <- substEqInDictInsts eq_irreds irreds1
1782 -- 9. eliminate the artificial skolem constants introduced in 1.
1785 -- If there was some FD improvement,
1786 -- or new wanted equations have been exposed,
1787 -- we should have another go at solving.
1788 ; let improved = availsImproved avails
1789 || (not $ isEmptyBag normalise_binds1)
1790 || (not $ isEmptyBag normalise_binds2)
1791 || (any isEqInst irreds)
1793 ; traceTc (text "reduceContext end" <+> (vcat [
1794 text "----------------------",
1796 text "given" <+> ppr (red_givens env),
1797 text "wanted" <+> ppr wanteds,
1799 text "avails" <+> pprAvails avails,
1800 text "improved =" <+> ppr improved,
1801 text "irreds = " <+> ppr irreds,
1802 text "binds = " <+> ppr binds,
1803 text "needed givens = " <+> ppr needed_givens,
1804 text "----------------------"
1808 given_binds `unionBags` normalise_binds1
1809 `unionBags` normalise_binds2
1811 irreds ++ eq_irreds,
1815 tcImproveOne :: Avails -> Inst -> TcM ImprovementDone
1816 tcImproveOne avails inst
1817 | not (isDict inst) = return False
1819 = do { inst_envs <- tcGetInstEnvs
1820 ; let eqns = improveOne (classInstances inst_envs)
1821 (dictPred inst, pprInstArising inst)
1822 [ (dictPred p, pprInstArising p)
1823 | p <- availsInsts avails, isDict p ]
1824 -- Avails has all the superclasses etc (good)
1825 -- It also has all the intermediates of the deduction (good)
1826 -- It does not have duplicates (good)
1827 -- NB that (?x::t1) and (?x::t2) will be held separately in avails
1828 -- so that improve will see them separate
1829 ; traceTc (text "improveOne" <+> ppr inst)
1832 unifyEqns :: [(Equation,(PredType,SDoc),(PredType,SDoc))]
1833 -> TcM ImprovementDone
1834 unifyEqns [] = return False
1836 = do { traceTc (ptext SLIT("Improve:") <+> vcat (map pprEquationDoc eqns))
1840 unify ((qtvs, pairs), what1, what2)
1841 = addErrCtxtM (mkEqnMsg what1 what2) $
1842 tcInstTyVars (varSetElems qtvs) `thenM` \ (_, _, tenv) ->
1843 mapM_ (unif_pr tenv) pairs
1844 unif_pr tenv (ty1,ty2) = unifyType (substTy tenv ty1) (substTy tenv ty2)
1846 pprEquationDoc (eqn, (p1,w1), (p2,w2)) = vcat [pprEquation eqn, nest 2 (ppr p1), nest 2 (ppr p2)]
1848 mkEqnMsg (pred1,from1) (pred2,from2) tidy_env
1849 = do { pred1' <- zonkTcPredType pred1; pred2' <- zonkTcPredType pred2
1850 ; let { pred1'' = tidyPred tidy_env pred1'; pred2'' = tidyPred tidy_env pred2' }
1851 ; let msg = vcat [ptext SLIT("When using functional dependencies to combine"),
1852 nest 2 (sep [ppr pred1'' <> comma, nest 2 from1]),
1853 nest 2 (sep [ppr pred2'' <> comma, nest 2 from2])]
1854 ; return (tidy_env, msg) }
1857 The main context-reduction function is @reduce@. Here's its game plan.
1860 reduceList :: RedEnv -> [Inst] -> Avails -> TcM Avails
1861 reduceList env@(RedEnv {red_stack = (n,stk)}) wanteds state
1862 = do { dopts <- getDOpts
1865 dumpTcRn (hang (ptext SLIT("Interesting! Context reduction stack depth") <+> int n)
1866 2 (ifPprDebug (nest 2 (pprStack stk))))
1869 ; if n >= ctxtStkDepth dopts then
1870 failWithTc (reduceDepthErr n stk)
1874 go [] state = return state
1875 go (w:ws) state = do { traceTc (text "reduceList " <+> (ppr (w:ws) $$ ppr state))
1876 ; state' <- reduce (env {red_stack = (n+1, w:stk)}) w state
1879 -- Base case: we're done!
1880 reduce env wanted avails
1881 -- It's the same as an existing inst, or a superclass thereof
1882 | Just avail <- findAvail avails wanted
1883 = do { traceTc (text "reduce: found " <+> ppr wanted)
1888 = do { traceTc (text "reduce" <+> ppr avails <+> ppr wanted)
1889 ; case red_try_me env wanted of {
1890 Stop -> try_simple (addIrred NoSCs);
1891 -- See Note [No superclasses for Stop]
1893 ReduceMe want_scs -> do -- It should be reduced
1894 { (avails, lookup_result) <- reduceInst env avails wanted
1895 ; case lookup_result of
1896 NoInstance -> addIrred want_scs avails wanted
1897 -- Add it and its superclasses
1899 GenInst [] rhs -> addWanted want_scs avails wanted rhs []
1901 GenInst wanteds' rhs
1902 -> do { avails1 <- addIrred NoSCs avails wanted
1903 ; avails2 <- reduceList env wanteds' avails1
1904 ; addWanted want_scs avails2 wanted rhs wanteds' } }
1905 -- Temporarily do addIrred *before* the reduceList,
1906 -- which has the effect of adding the thing we are trying
1907 -- to prove to the database before trying to prove the things it
1908 -- needs. See note [RECURSIVE DICTIONARIES]
1909 -- NB: we must not do an addWanted before, because that adds the
1910 -- superclasses too, and that can lead to a spurious loop; see
1911 -- the examples in [SUPERCLASS-LOOP]
1912 -- So we do an addIrred before, and then overwrite it afterwards with addWanted
1915 -- First, see if the inst can be reduced to a constant in one step
1916 -- Works well for literals (1::Int) and constant dictionaries (d::Num Int)
1917 -- Don't bother for implication constraints, which take real work
1918 try_simple do_this_otherwise
1919 = do { res <- lookupSimpleInst wanted
1921 GenInst [] rhs -> addWanted AddSCs avails wanted rhs []
1922 other -> do_this_otherwise avails wanted }
1926 Note [SUPERCLASS-LOOP 2]
1927 ~~~~~~~~~~~~~~~~~~~~~~~~
1928 But the above isn't enough. Suppose we are *given* d1:Ord a,
1929 and want to deduce (d2:C [a]) where
1931 class Ord a => C a where
1932 instance Ord [a] => C [a] where ...
1934 Then we'll use the instance decl to deduce C [a] from Ord [a], and then add the
1935 superclasses of C [a] to avails. But we must not overwrite the binding
1936 for Ord [a] (which is obtained from Ord a) with a superclass selection or we'll just
1939 Here's another variant, immortalised in tcrun020
1940 class Monad m => C1 m
1941 class C1 m => C2 m x
1942 instance C2 Maybe Bool
1943 For the instance decl we need to build (C1 Maybe), and it's no good if
1944 we run around and add (C2 Maybe Bool) and its superclasses to the avails
1945 before we search for C1 Maybe.
1947 Here's another example
1948 class Eq b => Foo a b
1949 instance Eq a => Foo [a] a
1953 we'll first deduce that it holds (via the instance decl). We must not
1954 then overwrite the Eq t constraint with a superclass selection!
1956 At first I had a gross hack, whereby I simply did not add superclass constraints
1957 in addWanted, though I did for addGiven and addIrred. This was sub-optimal,
1958 becuase it lost legitimate superclass sharing, and it still didn't do the job:
1959 I found a very obscure program (now tcrun021) in which improvement meant the
1960 simplifier got two bites a the cherry... so something seemed to be an Stop
1961 first time, but reducible next time.
1963 Now we implement the Right Solution, which is to check for loops directly
1964 when adding superclasses. It's a bit like the occurs check in unification.
1967 Note [RECURSIVE DICTIONARIES]
1968 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1970 data D r = ZeroD | SuccD (r (D r));
1972 instance (Eq (r (D r))) => Eq (D r) where
1973 ZeroD == ZeroD = True
1974 (SuccD a) == (SuccD b) = a == b
1977 equalDC :: D [] -> D [] -> Bool;
1980 We need to prove (Eq (D [])). Here's how we go:
1984 by instance decl, holds if
1988 by instance decl of Eq, holds if
1990 where d2 = dfEqList d3
1993 But now we can "tie the knot" to give
1999 and it'll even run! The trick is to put the thing we are trying to prove
2000 (in this case Eq (D []) into the database before trying to prove its
2001 contributing clauses.
2004 %************************************************************************
2006 Reducing a single constraint
2008 %************************************************************************
2011 ---------------------------------------------
2012 reduceInst :: RedEnv -> Avails -> Inst -> TcM (Avails, LookupInstResult)
2013 reduceInst env avails (ImplicInst { tci_tyvars = tvs, tci_reft = reft, tci_loc = loc,
2014 tci_given = extra_givens, tci_wanted = wanteds })
2015 = reduceImplication env avails reft tvs extra_givens wanteds loc
2017 reduceInst env avails other_inst
2018 = do { result <- lookupSimpleInst other_inst
2019 ; return (avails, result) }
2022 Note [Equational Constraints in Implication Constraints]
2023 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2025 An equational constraint is of the form
2027 where Given and Wanted may contain both equational and dictionary
2028 constraints. The delay and reduction of these two kinds of constraints
2031 -) In the generated code, wanted Dictionary constraints are wrapped up in an
2032 implication constraint that is created at the code site where the wanted
2033 dictionaries can be reduced via a let-binding. This let-bound implication
2034 constraint is deconstructed at the use-site of the wanted dictionaries.
2036 -) While the reduction of equational constraints is also delayed, the delay
2037 is not manifest in the generated code. The required evidence is generated
2038 in the code directly at the use-site. There is no let-binding and deconstruction
2039 necessary. The main disadvantage is that we cannot exploit sharing as the
2040 same evidence may be generated at multiple use-sites. However, this disadvantage
2041 is limited because it only concerns coercions which are erased.
2043 The different treatment is motivated by the different in representation. Dictionary
2044 constraints require manifest runtime dictionaries, while equations require coercions
2048 ---------------------------------------------
2049 reduceImplication :: RedEnv
2051 -> Refinement -- May refine the givens; often empty
2052 -> [TcTyVar] -- Quantified type variables; all skolems
2053 -> [Inst] -- Extra givens; all rigid
2056 -> TcM (Avails, LookupInstResult)
2059 Suppose we are simplifying the constraint
2060 forall bs. extras => wanted
2061 in the context of an overall simplification problem with givens 'givens',
2062 and refinment 'reft'.
2065 * The refinement is often empty
2067 * The 'extra givens' need not mention any of the quantified type variables
2068 e.g. forall {}. Eq a => Eq [a]
2069 forall {}. C Int => D (Tree Int)
2071 This happens when you have something like
2073 T1 :: Eq a => a -> T a
2076 f x = ...(case x of { T1 v -> v==v })...
2079 -- ToDo: should we instantiate tvs? I think it's not necessary
2081 -- Note on coercion variables:
2083 -- The extra given coercion variables are bound at two different sites:
2084 -- -) in the creation context of the implication constraint
2085 -- the solved equational constraints use these binders
2087 -- -) at the solving site of the implication constraint
2088 -- the solved dictionaries use these binders
2089 -- these binders are generated by reduceImplication
2091 reduceImplication env orig_avails reft tvs extra_givens wanteds inst_loc
2092 = do { -- Add refined givens, and the extra givens
2094 (refined_red_givens,refined_avails)
2095 <- if isEmptyRefinement reft then return (red_givens env,orig_avails)
2096 else foldlM (addRefinedGiven reft) ([],orig_avails) (red_givens env)
2098 -- Solve the sub-problem
2099 ; let try_me inst = ReduceMe AddSCs -- Note [Freeness and implications]
2100 env' = env { red_givens = refined_red_givens ++ extra_givens ++ availsInsts orig_avails
2101 , red_try_me = try_me }
2103 ; traceTc (text "reduceImplication" <+> vcat
2105 ppr (red_givens env), ppr extra_givens,
2106 ppr reft, ppr wanteds])
2107 ; (irreds,binds,needed_givens0) <- checkLoop env' wanteds
2108 ; let needed_givens1 = [ng | ng <- needed_givens0, notElem ng extra_givens]
2110 -- Note [Reducing implication constraints]
2111 -- Tom -- update note, put somewhere!
2113 ; traceTc (text "reduceImplication result" <+> vcat
2114 [ppr irreds, ppr binds, ppr needed_givens1])
2115 -- ; avails <- reduceList env' wanteds avails
2117 -- -- Extract the binding
2118 -- ; (binds, irreds) <- extractResults avails wanteds
2119 ; (refinement_binds,needed_givens) <- extractLocalResults refined_avails needed_givens1
2120 ; traceTc (text "reduceImplication local results" <+> vcat
2121 [ppr refinement_binds, ppr needed_givens])
2123 ; -- extract superclass binds
2124 -- (sc_binds,_) <- extractResults avails []
2125 -- ; traceTc (text "reduceImplication sc_binds" <+> vcat
2126 -- [ppr sc_binds, ppr avails])
2129 -- We always discard the extra avails we've generated;
2130 -- but we remember if we have done any (global) improvement
2131 -- ; let ret_avails = avails
2132 ; let ret_avails = orig_avails
2133 -- ; let ret_avails = updateImprovement orig_avails avails
2135 ; traceTc (text "reduceImplication condition" <+> ppr ((isEmptyLHsBinds binds) || (null irreds)))
2137 -- Porgress is no longer measered by the number of bindings
2138 -- ; if isEmptyLHsBinds binds then -- No progress
2139 ; if (isEmptyLHsBinds binds) && (not $ null irreds) then
2140 return (ret_avails, NoInstance)
2143 ; (implic_insts, bind) <- makeImplicationBind inst_loc tvs reft extra_givens irreds
2144 -- This binding is useless if the recursive simplification
2145 -- made no progress; but currently we don't try to optimise that
2146 -- case. After all, we only try hard to reduce at top level, or
2147 -- when inferring types.
2149 ; let dict_wanteds = filter (not . isEqInst) wanteds
2150 (extra_eq_givens, extra_dict_givens) = partition isEqInst extra_givens
2151 dict_ids = map instToId extra_dict_givens
2152 -- TOMDO: given equational constraints bug!
2153 -- we need a different evidence for given
2154 -- equations depending on whether we solve
2155 -- dictionary constraints or equational constraints
2156 eq_tyvars = uniqSetToList $ tyVarsOfTypes $ map eqInstType extra_eq_givens
2157 -- dict_ids = map instToId extra_givens
2158 co = mkWpTyLams tvs <.> mkWpTyLams eq_tyvars <.> mkWpLams dict_ids <.> WpLet (binds `unionBags` refinement_binds `unionBags` bind)
2159 rhs = mkHsWrap co payload
2160 loc = instLocSpan inst_loc
2161 payload | [dict_wanted] <- dict_wanteds = HsVar (instToId dict_wanted)
2162 | otherwise = ExplicitTuple (map (L loc . HsVar . instToId) dict_wanteds) Boxed
2165 ; traceTc (text "reduceImplication ->" <+> vcat
2168 -- If there are any irreds, we back off and return NoInstance
2169 ; return (ret_avails, GenInst (implic_insts ++ needed_givens) (L loc rhs))
2174 Note [Reducing implication constraints]
2175 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2176 Suppose we are trying to simplify
2178 ic: (forall b. C a b => (W [a] b, D c b)) )
2180 instance (C a b, Ord a) => W [a] b
2181 When solving the implication constraint, we'll start with
2183 in the Avails. Then we add (C a b -> Given) and solve. Extracting
2184 the results gives us a binding for the (W [a] b), with an Irred of
2185 (Ord a, D c b). Now, the (Ord a) comes from "outside" the implication,
2186 but the (D d b) is from "inside". So we want to generate a GenInst
2191 ic' :: forall b. C a b => D c b]
2192 (/\b \(dc:C a b). (df a b dc do, ic' b dc))
2194 The first arg of GenInst gives the free dictionary variables of the
2195 second argument -- the "needed givens". And that list in turn is
2196 vital because it's used to determine what other dicts must be solved.
2197 This very list ends up in the second field of the Rhs, and drives
2200 The need for this field is why we have to return "needed givens"
2201 from extractResults, reduceContext, checkLoop, and so on.
2203 NB: the "needed givens" in a GenInst or Rhs, may contain two dicts
2204 with the same type but different Ids, e.g. [d12 :: Eq a, d81 :: Eq a]
2205 That says we must generate a binding for both d12 and d81.
2207 The "inside" and "outside" distinction is what's going on with 'inner' and
2208 'outer' in reduceImplication
2211 Note [Freeness and implications]
2212 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2213 It's hard to say when an implication constraint can be floated out. Consider
2214 forall {} Eq a => Foo [a]
2215 The (Foo [a]) doesn't mention any of the quantified variables, but it
2216 still might be partially satisfied by the (Eq a).
2218 There is a useful special case when it *is* easy to partition the
2219 constraints, namely when there are no 'givens'. Consider
2220 forall {a}. () => Bar b
2221 There are no 'givens', and so there is no reason to capture (Bar b).
2222 We can let it float out. But if there is even one constraint we
2223 must be much more careful:
2224 forall {a}. C a b => Bar (m b)
2225 because (C a b) might have a superclass (D b), from which we might
2226 deduce (Bar [b]) when m later gets instantiated to []. Ha!
2228 Here is an even more exotic example
2230 Now consider the constraint
2231 forall b. D Int b => C Int
2232 We can satisfy the (C Int) from the superclass of D, so we don't want
2233 to float the (C Int) out, even though it mentions no type variable in
2236 Note [Pruning the givens in an implication constraint]
2237 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2238 Suppose we are about to form the implication constraint
2239 forall tvs. Eq a => Ord b
2240 The (Eq a) cannot contribute to the (Ord b), because it has no access to
2241 the type variable 'b'. So we could filter out the (Eq a) from the givens.
2243 Doing so would be a bit tidier, but all the implication constraints get
2244 simplified away by the optimiser, so it's no great win. So I don't take
2245 advantage of that at the moment.
2247 If you do, BE CAREFUL of wobbly type variables.
2250 %************************************************************************
2252 Avails and AvailHow: the pool of evidence
2254 %************************************************************************
2258 data Avails = Avails !ImprovementDone !AvailEnv
2260 type ImprovementDone = Bool -- True <=> some unification has happened
2261 -- so some Irreds might now be reducible
2262 -- keys that are now
2264 type AvailEnv = FiniteMap Inst AvailHow
2266 = IsIrred -- Used for irreducible dictionaries,
2267 -- which are going to be lambda bound
2269 | Given TcId -- Used for dictionaries for which we have a binding
2270 -- e.g. those "given" in a signature
2272 | Rhs -- Used when there is a RHS
2273 (LHsExpr TcId) -- The RHS
2274 [Inst] -- Insts free in the RHS; we need these too
2276 instance Outputable Avails where
2279 pprAvails (Avails imp avails)
2280 = vcat [ ptext SLIT("Avails") <> (if imp then ptext SLIT("[improved]") else empty)
2281 , nest 2 (vcat [sep [ppr inst, nest 2 (equals <+> ppr avail)]
2282 | (inst,avail) <- fmToList avails ])]
2284 instance Outputable AvailHow where
2287 -------------------------
2288 pprAvail :: AvailHow -> SDoc
2289 pprAvail IsIrred = text "Irred"
2290 pprAvail (Given x) = text "Given" <+> ppr x
2291 pprAvail (Rhs rhs bs) = text "Rhs" <+> sep [ppr rhs, braces (ppr bs)]
2293 -------------------------
2294 extendAvailEnv :: AvailEnv -> Inst -> AvailHow -> AvailEnv
2295 extendAvailEnv env inst avail = addToFM env inst avail
2297 findAvailEnv :: AvailEnv -> Inst -> Maybe AvailHow
2298 findAvailEnv env wanted = lookupFM env wanted
2299 -- NB 1: the Ord instance of Inst compares by the class/type info
2300 -- *not* by unique. So
2301 -- d1::C Int == d2::C Int
2303 emptyAvails :: Avails
2304 emptyAvails = Avails False emptyFM
2306 findAvail :: Avails -> Inst -> Maybe AvailHow
2307 findAvail (Avails _ avails) wanted = findAvailEnv avails wanted
2309 elemAvails :: Inst -> Avails -> Bool
2310 elemAvails wanted (Avails _ avails) = wanted `elemFM` avails
2312 extendAvails :: Avails -> Inst -> AvailHow -> TcM Avails
2314 extendAvails avails@(Avails imp env) inst avail
2315 = do { imp1 <- tcImproveOne avails inst -- Do any improvement
2316 ; return (Avails (imp || imp1) (extendAvailEnv env inst avail)) }
2318 availsInsts :: Avails -> [Inst]
2319 availsInsts (Avails _ avails) = keysFM avails
2321 availsImproved (Avails imp _) = imp
2323 updateImprovement :: Avails -> Avails -> Avails
2324 -- (updateImprovement a1 a2) sets a1's improvement flag from a2
2325 updateImprovement (Avails _ avails1) (Avails imp2 _) = Avails imp2 avails1
2328 Extracting the bindings from a bunch of Avails.
2329 The bindings do *not* come back sorted in dependency order.
2330 We assume that they'll be wrapped in a big Rec, so that the
2331 dependency analyser can sort them out later
2334 extractResults :: Avails
2336 -> TcM ( TcDictBinds, -- Bindings
2337 [Inst], -- Irreducible ones
2338 [Inst]) -- Needed givens, i.e. ones used in the bindings
2339 -- Postcondition: needed-givens = free vars( binds ) \ irreds
2340 -- Note [Reducing implication constraints]
2342 extractResults (Avails _ avails) wanteds
2343 = go avails emptyBag [] [] wanteds
2345 go :: AvailEnv -> TcDictBinds -> [Inst] -> [Inst] -> [Inst]
2346 -> TcM (TcDictBinds, [Inst], [Inst])
2347 go avails binds irreds givens []
2348 = returnM (binds, irreds, givens)
2350 go avails binds irreds givens (w:ws)
2351 = case findAvailEnv avails w of
2352 Nothing -> pprTrace "Urk: extractResults" (ppr w) $
2353 go avails binds irreds givens ws
2356 | id == w_id -> go avails binds irreds (w:givens) ws
2358 go avails (addInstToDictBind binds w (nlHsVar id)) irreds
2359 (update_id w id:givens) ws
2360 -- The sought Id can be one of the givens, via a superclass chain
2361 -- and then we definitely don't want to generate an x=x binding!
2363 Just IsIrred -> go (add_given avails w) binds (w:irreds) givens ws
2364 -- The add_given handles the case where we want (Ord a, Eq a), and we
2365 -- don't want to emit *two* Irreds for Ord a, one via the superclass chain
2366 -- This showed up in a dupliated Ord constraint in the error message for
2369 Just (Rhs rhs ws') -> go (add_given avails w) new_binds irreds givens (ws' ++ ws)
2371 new_binds = addInstToDictBind binds w rhs
2374 update_id m@(Method{}) id = m {tci_id = id}
2375 update_id w id = w {tci_name = idName id}
2377 add_given avails w = extendAvailEnv avails w (Given (instToId w))
2379 extractLocalResults :: Avails
2381 -> TcM ( TcDictBinds, -- Bindings
2382 [Inst]) -- Needed givens, i.e. ones used in the bindings
2384 extractLocalResults (Avails _ avails) wanteds
2385 = go avails emptyBag [] wanteds
2387 go :: AvailEnv -> TcDictBinds -> [Inst] -> [Inst]
2388 -> TcM (TcDictBinds, [Inst])
2389 go avails binds givens []
2390 = returnM (binds, givens)
2392 go avails binds givens (w:ws)
2393 = case findAvailEnv avails w of
2394 Nothing -> -- pprTrace "Urk: extractLocalResults" (ppr w) $
2395 go avails binds givens ws
2398 go avails binds givens ws
2401 | id == w_id -> go avails binds (w:givens) ws
2402 | otherwise -> go avails binds (w{tci_name=idName id}:givens) ws
2403 -- The sought Id can be one of the givens, via a superclass chain
2404 -- and then we definitely don't want to generate an x=x binding!
2406 Just (Rhs rhs ws') -> go (add_given avails w) new_binds givens (ws' ++ ws)
2408 new_binds = addInstToDictBind binds w rhs
2412 add_given avails w = extendAvailEnv avails w (Given (instToId w))
2416 Note [No superclasses for Stop]
2417 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2418 When we decide not to reduce an Inst -- the 'WhatToDo' --- we still
2419 add it to avails, so that any other equal Insts will be commoned up
2420 right here. However, we do *not* add superclasses. If we have
2423 but a is not bound here, then we *don't* want to derive dn from df
2424 here lest we lose sharing.
2427 addWanted :: WantSCs -> Avails -> Inst -> LHsExpr TcId -> [Inst] -> TcM Avails
2428 addWanted want_scs avails wanted rhs_expr wanteds
2429 = addAvailAndSCs want_scs avails wanted avail
2431 avail = Rhs rhs_expr wanteds
2433 addGiven :: Avails -> Inst -> TcM Avails
2434 addGiven avails given = addAvailAndSCs AddSCs avails given (Given (instToId given))
2435 -- Always add superclasses for 'givens'
2437 -- No ASSERT( not (given `elemAvails` avails) ) because in an instance
2438 -- decl for Ord t we can add both Ord t and Eq t as 'givens',
2439 -- so the assert isn't true
2441 addRefinedGiven :: Refinement -> ([Inst], Avails) -> Inst -> TcM ([Inst], Avails)
2442 addRefinedGiven reft (refined_givens, avails) given
2443 | isDict given -- We sometimes have 'given' methods, but they
2444 -- are always optional, so we can drop them
2445 , let pred = dictPred given
2446 , isRefineablePred pred -- See Note [ImplicInst rigidity]
2447 , Just (co, pred) <- refinePred reft pred
2448 = do { new_given <- newDictBndr (instLoc given) pred
2449 ; let rhs = L (instSpan given) $
2450 HsWrap (WpCo co) (HsVar (instToId given))
2451 ; avails <- addAvailAndSCs AddSCs avails new_given (Rhs rhs [given])
2452 ; return (new_given:refined_givens, avails) }
2453 -- ToDo: the superclasses of the original given all exist in Avails
2454 -- so we could really just cast them, but it's more awkward to do,
2455 -- and hopefully the optimiser will spot the duplicated work
2457 = return (refined_givens, avails)
2460 Note [ImplicInst rigidity]
2461 ~~~~~~~~~~~~~~~~~~~~~~~~~~
2463 C :: forall ab. (Eq a, Ord b) => b -> T a
2465 ...(case x of C v -> <body>)...
2467 From the case (where x::T ty) we'll get an implication constraint
2468 forall b. (Eq ty, Ord b) => <body-constraints>
2469 Now suppose <body-constraints> itself has an implication constraint
2471 forall c. <reft> => <payload>
2472 Then, we can certainly apply the refinement <reft> to the Ord b, becuase it is
2473 existential, but we probably should not apply it to the (Eq ty) because it may
2474 be wobbly. Hence the isRigidInst
2476 @Insts@ are ordered by their class/type info, rather than by their
2477 unique. This allows the context-reduction mechanism to use standard finite
2478 maps to do their stuff. It's horrible that this code is here, rather
2479 than with the Avails handling stuff in TcSimplify
2482 addIrred :: WantSCs -> Avails -> Inst -> TcM Avails
2483 addIrred want_scs avails irred = ASSERT2( not (irred `elemAvails` avails), ppr irred $$ ppr avails )
2484 addAvailAndSCs want_scs avails irred IsIrred
2486 addAvailAndSCs :: WantSCs -> Avails -> Inst -> AvailHow -> TcM Avails
2487 addAvailAndSCs want_scs avails inst avail
2488 | not (isClassDict inst) = extendAvails avails inst avail
2489 | NoSCs <- want_scs = extendAvails avails inst avail
2490 | otherwise = do { traceTc (text "addAvailAndSCs" <+> vcat [ppr inst, ppr deps])
2491 ; avails' <- extendAvails avails inst avail
2492 ; addSCs is_loop avails' inst }
2494 is_loop pred = any (`tcEqType` mkPredTy pred) dep_tys
2495 -- Note: this compares by *type*, not by Unique
2496 deps = findAllDeps (unitVarSet (instToVar inst)) avail
2497 dep_tys = map idType (varSetElems deps)
2499 findAllDeps :: IdSet -> AvailHow -> IdSet
2500 -- Find all the Insts that this one depends on
2501 -- See Note [SUPERCLASS-LOOP 2]
2502 -- Watch out, though. Since the avails may contain loops
2503 -- (see Note [RECURSIVE DICTIONARIES]), so we need to track the ones we've seen so far
2504 findAllDeps so_far (Rhs _ kids) = foldl find_all so_far kids
2505 findAllDeps so_far other = so_far
2507 find_all :: IdSet -> Inst -> IdSet
2509 | isEqInst kid = so_far
2510 | kid_id `elemVarSet` so_far = so_far
2511 | Just avail <- findAvail avails kid = findAllDeps so_far' avail
2512 | otherwise = so_far'
2514 so_far' = extendVarSet so_far kid_id -- Add the new kid to so_far
2515 kid_id = instToId kid
2517 addSCs :: (TcPredType -> Bool) -> Avails -> Inst -> TcM Avails
2518 -- Add all the superclasses of the Inst to Avails
2519 -- The first param says "dont do this because the original thing
2520 -- depends on this one, so you'd build a loop"
2521 -- Invariant: the Inst is already in Avails.
2523 addSCs is_loop avails dict
2524 = ASSERT( isDict dict )
2525 do { sc_dicts <- newDictBndrs (instLoc dict) sc_theta'
2526 ; foldlM add_sc avails (zipEqual "add_scs" sc_dicts sc_sels) }
2528 (clas, tys) = getDictClassTys dict
2529 (tyvars, sc_theta, sc_sels, _) = classBigSig clas
2530 sc_theta' = filter (not . isEqPred) $
2531 substTheta (zipTopTvSubst tyvars tys) sc_theta
2533 add_sc avails (sc_dict, sc_sel)
2534 | is_loop (dictPred sc_dict) = return avails -- See Note [SUPERCLASS-LOOP 2]
2535 | is_given sc_dict = return avails
2536 | otherwise = do { avails' <- extendAvails avails sc_dict (Rhs sc_sel_rhs [dict])
2537 ; addSCs is_loop avails' sc_dict }
2539 sc_sel_rhs = L (instSpan dict) (HsWrap co_fn (HsVar sc_sel))
2540 co_fn = WpApp (instToVar dict) <.> mkWpTyApps tys
2542 is_given :: Inst -> Bool
2543 is_given sc_dict = case findAvail avails sc_dict of
2544 Just (Given _) -> True -- Given is cheaper than superclass selection
2547 -- From the a set of insts obtain all equalities that (transitively) occur in
2548 -- superclass contexts of class constraints (aka the ancestor equalities).
2550 ancestorEqualities :: [Inst] -> TcM [Inst]
2552 = mapM mkWantedEqInst -- turn only equality predicates..
2553 . filter isEqPred -- ..into wanted equality insts
2555 . addAEsToBag emptyBag -- collect the superclass constraints..
2556 . map dictPred -- ..of all predicates in a bag
2557 . filter isClassDict
2559 addAEsToBag :: Bag PredType -> [PredType] -> Bag PredType
2560 addAEsToBag bag [] = bag
2561 addAEsToBag bag (pred:preds)
2562 | pred `elemBag` bag = addAEsToBag bag preds
2563 | isEqPred pred = addAEsToBag bagWithPred preds
2564 | isClassPred pred = addAEsToBag bagWithPred predsWithSCs
2565 | otherwise = addAEsToBag bag preds
2567 bagWithPred = bag `snocBag` pred
2568 predsWithSCs = preds ++ substTheta (zipTopTvSubst tyvars tys) sc_theta
2570 (tyvars, sc_theta, _, _) = classBigSig clas
2571 (clas, tys) = getClassPredTys pred
2575 %************************************************************************
2577 \section{tcSimplifyTop: defaulting}
2579 %************************************************************************
2582 @tcSimplifyTop@ is called once per module to simplify all the constant
2583 and ambiguous Insts.
2585 We need to be careful of one case. Suppose we have
2587 instance Num a => Num (Foo a b) where ...
2589 and @tcSimplifyTop@ is given a constraint (Num (Foo x y)). Then it'll simplify
2590 to (Num x), and default x to Int. But what about y??
2592 It's OK: the final zonking stage should zap y to (), which is fine.
2596 tcSimplifyTop, tcSimplifyInteractive :: [Inst] -> TcM TcDictBinds
2597 tcSimplifyTop wanteds
2598 = tc_simplify_top doc False wanteds
2600 doc = text "tcSimplifyTop"
2602 tcSimplifyInteractive wanteds
2603 = tc_simplify_top doc True wanteds
2605 doc = text "tcSimplifyInteractive"
2607 -- The TcLclEnv should be valid here, solely to improve
2608 -- error message generation for the monomorphism restriction
2609 tc_simplify_top doc interactive wanteds
2610 = do { dflags <- getDOpts
2611 ; wanteds <- zonkInsts wanteds
2612 ; mapM_ zonkTopTyVar (varSetElems (tyVarsOfInsts wanteds))
2614 ; traceTc (text "tc_simplify_top 0: " <+> ppr wanteds)
2615 ; (irreds1, binds1) <- tryHardCheckLoop doc1 wanteds
2616 ; traceTc (text "tc_simplify_top 1: " <+> ppr irreds1)
2617 ; (irreds2, binds2) <- approximateImplications doc2 (\d -> True) irreds1
2618 ; traceTc (text "tc_simplify_top 2: " <+> ppr irreds2)
2620 -- Use the defaulting rules to do extra unification
2621 -- NB: irreds2 are already zonked
2622 ; (irreds3, binds3) <- disambiguate doc3 interactive dflags irreds2
2624 -- Deal with implicit parameters
2625 ; let (bad_ips, non_ips) = partition isIPDict irreds3
2626 (ambigs, others) = partition isTyVarDict non_ips
2628 ; topIPErrs bad_ips -- Can arise from f :: Int -> Int
2630 ; addNoInstanceErrs others
2631 ; addTopAmbigErrs ambigs
2633 ; return (binds1 `unionBags` binds2 `unionBags` binds3) }
2635 doc1 = doc <+> ptext SLIT("(first round)")
2636 doc2 = doc <+> ptext SLIT("(approximate)")
2637 doc3 = doc <+> ptext SLIT("(disambiguate)")
2640 If a dictionary constrains a type variable which is
2641 * not mentioned in the environment
2642 * and not mentioned in the type of the expression
2643 then it is ambiguous. No further information will arise to instantiate
2644 the type variable; nor will it be generalised and turned into an extra
2645 parameter to a function.
2647 It is an error for this to occur, except that Haskell provided for
2648 certain rules to be applied in the special case of numeric types.
2650 * at least one of its classes is a numeric class, and
2651 * all of its classes are numeric or standard
2652 then the type variable can be defaulted to the first type in the
2653 default-type list which is an instance of all the offending classes.
2655 So here is the function which does the work. It takes the ambiguous
2656 dictionaries and either resolves them (producing bindings) or
2657 complains. It works by splitting the dictionary list by type
2658 variable, and using @disambigOne@ to do the real business.
2660 @disambigOne@ assumes that its arguments dictionaries constrain all
2661 the same type variable.
2663 ADR Comment 20/6/94: I've changed the @CReturnable@ case to default to
2664 @()@ instead of @Int@. I reckon this is the Right Thing to do since
2665 the most common use of defaulting is code like:
2667 _ccall_ foo `seqPrimIO` bar
2669 Since we're not using the result of @foo@, the result if (presumably)
2673 disambiguate :: SDoc -> Bool -> DynFlags -> [Inst] -> TcM ([Inst], TcDictBinds)
2674 -- Just does unification to fix the default types
2675 -- The Insts are assumed to be pre-zonked
2676 disambiguate doc interactive dflags insts
2678 = return (insts, emptyBag)
2680 | null defaultable_groups
2681 = do { traceTc (text "disambigutate, no defaultable groups" <+> vcat [ppr unaries, ppr insts, ppr bad_tvs, ppr defaultable_groups])
2682 ; return (insts, emptyBag) }
2685 = do { -- Figure out what default types to use
2686 default_tys <- getDefaultTys extended_defaulting ovl_strings
2688 ; traceTc (text "disambiguate1" <+> vcat [ppr insts, ppr unaries, ppr bad_tvs, ppr defaultable_groups])
2689 ; mapM_ (disambigGroup default_tys) defaultable_groups
2691 -- disambigGroup does unification, hence try again
2692 ; tryHardCheckLoop doc insts }
2695 extended_defaulting = interactive || dopt Opt_ExtendedDefaultRules dflags
2696 ovl_strings = dopt Opt_OverloadedStrings dflags
2698 unaries :: [(Inst, Class, TcTyVar)] -- (C tv) constraints
2699 bad_tvs :: TcTyVarSet -- Tyvars mentioned by *other* constraints
2700 (unaries, bad_tvs_s) = partitionWith find_unary insts
2701 bad_tvs = unionVarSets bad_tvs_s
2703 -- Finds unary type-class constraints
2704 find_unary d@(Dict {tci_pred = ClassP cls [ty]})
2705 | Just tv <- tcGetTyVar_maybe ty = Left (d,cls,tv)
2706 find_unary inst = Right (tyVarsOfInst inst)
2708 -- Group by type variable
2709 defaultable_groups :: [[(Inst,Class,TcTyVar)]]
2710 defaultable_groups = filter defaultable_group (equivClasses cmp_tv unaries)
2711 cmp_tv (_,_,tv1) (_,_,tv2) = tv1 `compare` tv2
2713 defaultable_group :: [(Inst,Class,TcTyVar)] -> Bool
2714 defaultable_group ds@((_,_,tv):_)
2715 = isTyConableTyVar tv -- Note [Avoiding spurious errors]
2716 && not (tv `elemVarSet` bad_tvs)
2717 && defaultable_classes [c | (_,c,_) <- ds]
2718 defaultable_group [] = panic "defaultable_group"
2720 defaultable_classes clss
2721 | extended_defaulting = any isInteractiveClass clss
2722 | otherwise = all is_std_class clss && (any is_num_class clss)
2724 -- In interactive mode, or with -fextended-default-rules,
2725 -- we default Show a to Show () to avoid graututious errors on "show []"
2726 isInteractiveClass cls
2727 = is_num_class cls || (classKey cls `elem` [showClassKey, eqClassKey, ordClassKey])
2729 is_num_class cls = isNumericClass cls || (ovl_strings && (cls `hasKey` isStringClassKey))
2730 -- is_num_class adds IsString to the standard numeric classes,
2731 -- when -foverloaded-strings is enabled
2733 is_std_class cls = isStandardClass cls || (ovl_strings && (cls `hasKey` isStringClassKey))
2734 -- Similarly is_std_class
2736 -----------------------
2737 disambigGroup :: [Type] -- The default types
2738 -> [(Inst,Class,TcTyVar)] -- All standard classes of form (C a)
2739 -> TcM () -- Just does unification, to fix the default types
2741 disambigGroup default_tys dicts
2742 = try_default default_tys
2744 (_,_,tyvar) = ASSERT(not (null dicts)) head dicts -- Should be non-empty
2745 classes = [c | (_,c,_) <- dicts]
2747 try_default [] = return ()
2748 try_default (default_ty : default_tys)
2749 = tryTcLIE_ (try_default default_tys) $
2750 do { tcSimplifyDefault [mkClassPred clas [default_ty] | clas <- classes]
2751 -- This may fail; then the tryTcLIE_ kicks in
2752 -- Failure here is caused by there being no type in the
2753 -- default list which can satisfy all the ambiguous classes.
2754 -- For example, if Real a is reqd, but the only type in the
2755 -- default list is Int.
2757 -- After this we can't fail
2758 ; warnDefault dicts default_ty
2759 ; unifyType default_ty (mkTyVarTy tyvar)
2760 ; return () -- TOMDO: do something with the coercion
2764 -----------------------
2765 getDefaultTys :: Bool -> Bool -> TcM [Type]
2766 getDefaultTys extended_deflts ovl_strings
2767 = do { mb_defaults <- getDeclaredDefaultTys
2768 ; case mb_defaults of {
2769 Just tys -> return tys ; -- User-supplied defaults
2772 -- No use-supplied default
2773 -- Use [Integer, Double], plus modifications
2774 { integer_ty <- tcMetaTy integerTyConName
2775 ; checkWiredInTyCon doubleTyCon
2776 ; string_ty <- tcMetaTy stringTyConName
2777 ; return (opt_deflt extended_deflts unitTy
2778 -- Note [Default unitTy]
2780 [integer_ty,doubleTy]
2782 opt_deflt ovl_strings string_ty) } } }
2784 opt_deflt True ty = [ty]
2785 opt_deflt False ty = []
2788 Note [Default unitTy]
2789 ~~~~~~~~~~~~~~~~~~~~~
2790 In interative mode (or with -fextended-default-rules) we add () as the first type we
2791 try when defaulting. This has very little real impact, except in the following case.
2793 Text.Printf.printf "hello"
2794 This has type (forall a. IO a); it prints "hello", and returns 'undefined'. We don't
2795 want the GHCi repl loop to try to print that 'undefined'. The neatest thing is to
2796 default the 'a' to (), rather than to Integer (which is what would otherwise happen;
2797 and then GHCi doesn't attempt to print the (). So in interactive mode, we add
2798 () to the list of defaulting types. See Trac #1200.
2800 Note [Avoiding spurious errors]
2801 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2802 When doing the unification for defaulting, we check for skolem
2803 type variables, and simply don't default them. For example:
2804 f = (*) -- Monomorphic
2805 g :: Num a => a -> a
2807 Here, we get a complaint when checking the type signature for g,
2808 that g isn't polymorphic enough; but then we get another one when
2809 dealing with the (Num a) context arising from f's definition;
2810 we try to unify a with Int (to default it), but find that it's
2811 already been unified with the rigid variable from g's type sig
2814 %************************************************************************
2816 \subsection[simple]{@Simple@ versions}
2818 %************************************************************************
2820 Much simpler versions when there are no bindings to make!
2822 @tcSimplifyThetas@ simplifies class-type constraints formed by
2823 @deriving@ declarations and when specialising instances. We are
2824 only interested in the simplified bunch of class/type constraints.
2826 It simplifies to constraints of the form (C a b c) where
2827 a,b,c are type variables. This is required for the context of
2828 instance declarations.
2831 tcSimplifyDeriv :: InstOrigin
2833 -> ThetaType -- Wanted
2834 -> TcM ThetaType -- Needed
2835 -- Given instance (wanted) => C inst_ty
2836 -- Simplify 'wanted' as much as possible
2838 tcSimplifyDeriv orig tyvars theta
2839 = do { (tvs, _, tenv) <- tcInstTyVars tyvars
2840 -- The main loop may do unification, and that may crash if
2841 -- it doesn't see a TcTyVar, so we have to instantiate. Sigh
2842 -- ToDo: what if two of them do get unified?
2843 ; wanteds <- newDictBndrsO orig (substTheta tenv theta)
2844 ; (irreds, _) <- tryHardCheckLoop doc wanteds
2846 ; let (tv_dicts, others) = partition ok irreds
2847 ; addNoInstanceErrs others
2848 -- See Note [Exotic derived instance contexts] in TcMType
2850 ; let rev_env = zipTopTvSubst tvs (mkTyVarTys tyvars)
2851 simpl_theta = substTheta rev_env (map dictPred tv_dicts)
2852 -- This reverse-mapping is a pain, but the result
2853 -- should mention the original TyVars not TcTyVars
2855 ; return simpl_theta }
2857 doc = ptext SLIT("deriving classes for a data type")
2859 ok dict | isDict dict = validDerivPred (dictPred dict)
2864 @tcSimplifyDefault@ just checks class-type constraints, essentially;
2865 used with \tr{default} declarations. We are only interested in
2866 whether it worked or not.
2869 tcSimplifyDefault :: ThetaType -- Wanted; has no type variables in it
2872 tcSimplifyDefault theta
2873 = newDictBndrsO DefaultOrigin theta `thenM` \ wanteds ->
2874 tryHardCheckLoop doc wanteds `thenM` \ (irreds, _) ->
2875 addNoInstanceErrs irreds `thenM_`
2881 doc = ptext SLIT("default declaration")
2885 %************************************************************************
2887 \section{Errors and contexts}
2889 %************************************************************************
2891 ToDo: for these error messages, should we note the location as coming
2892 from the insts, or just whatever seems to be around in the monad just
2896 groupErrs :: ([Inst] -> TcM ()) -- Deal with one group
2897 -> [Inst] -- The offending Insts
2899 -- Group together insts with the same origin
2900 -- We want to report them together in error messages
2902 groupErrs report_err []
2904 groupErrs report_err (inst:insts)
2905 = do_one (inst:friends) `thenM_`
2906 groupErrs report_err others
2909 -- (It may seem a bit crude to compare the error messages,
2910 -- but it makes sure that we combine just what the user sees,
2911 -- and it avoids need equality on InstLocs.)
2912 (friends, others) = partition is_friend insts
2913 loc_msg = showSDoc (pprInstLoc (instLoc inst))
2914 is_friend friend = showSDoc (pprInstLoc (instLoc friend)) == loc_msg
2915 do_one insts = addInstCtxt (instLoc (head insts)) (report_err insts)
2916 -- Add location and context information derived from the Insts
2918 -- Add the "arising from..." part to a message about bunch of dicts
2919 addInstLoc :: [Inst] -> Message -> Message
2920 addInstLoc insts msg = msg $$ nest 2 (pprInstArising (head insts))
2922 addTopIPErrs :: [Name] -> [Inst] -> TcM ()
2923 addTopIPErrs bndrs []
2925 addTopIPErrs bndrs ips
2926 = do { dflags <- getDOpts
2927 ; addErrTcM (tidy_env, mk_msg dflags tidy_ips) }
2929 (tidy_env, tidy_ips) = tidyInsts ips
2931 = vcat [sep [ptext SLIT("Implicit parameters escape from"),
2932 nest 2 (ptext SLIT("the monomorphic top-level binding")
2933 <> plural bndrs <+> ptext SLIT("of")
2934 <+> pprBinders bndrs <> colon)],
2935 nest 2 (vcat (map ppr_ip ips)),
2936 monomorphism_fix dflags]
2937 ppr_ip ip = pprPred (dictPred ip) <+> pprInstArising ip
2939 topIPErrs :: [Inst] -> TcM ()
2941 = groupErrs report tidy_dicts
2943 (tidy_env, tidy_dicts) = tidyInsts dicts
2944 report dicts = addErrTcM (tidy_env, mk_msg dicts)
2945 mk_msg dicts = addInstLoc dicts (ptext SLIT("Unbound implicit parameter") <>
2946 plural tidy_dicts <+> pprDictsTheta tidy_dicts)
2948 addNoInstanceErrs :: [Inst] -- Wanted (can include implications)
2950 addNoInstanceErrs insts
2951 = do { let (tidy_env, tidy_insts) = tidyInsts insts
2952 ; reportNoInstances tidy_env Nothing tidy_insts }
2956 -> Maybe (InstLoc, [Inst]) -- Context
2957 -- Nothing => top level
2958 -- Just (d,g) => d describes the construct
2960 -> [Inst] -- What is wanted (can include implications)
2963 reportNoInstances tidy_env mb_what insts
2964 = groupErrs (report_no_instances tidy_env mb_what) insts
2966 report_no_instances tidy_env mb_what insts
2967 = do { inst_envs <- tcGetInstEnvs
2968 ; let (implics, insts1) = partition isImplicInst insts
2969 (insts2, overlaps) = partitionWith (check_overlap inst_envs) insts1
2970 (eqInsts, insts3) = partition isEqInst insts2
2971 ; traceTc (text "reportNoInstances" <+> vcat
2972 [ppr implics, ppr insts1, ppr insts2])
2973 ; mapM_ complain_implic implics
2974 ; mapM_ (\doc -> addErrTcM (tidy_env, doc)) overlaps
2975 ; groupErrs complain_no_inst insts3
2976 ; mapM_ eqInstMisMatch eqInsts
2979 complain_no_inst insts = addErrTcM (tidy_env, mk_no_inst_err insts)
2981 complain_implic inst -- Recurse!
2982 = reportNoInstances tidy_env
2983 (Just (tci_loc inst, tci_given inst))
2986 check_overlap :: (InstEnv,InstEnv) -> Inst -> Either Inst SDoc
2987 -- Right msg => overlap message
2988 -- Left inst => no instance
2989 check_overlap inst_envs wanted
2990 | not (isClassDict wanted) = Left wanted
2992 = case lookupInstEnv inst_envs clas tys of
2993 -- The case of exactly one match and no unifiers means a
2994 -- successful lookup. That can't happen here, because dicts
2995 -- only end up here if they didn't match in Inst.lookupInst
2997 ([m],[]) -> pprPanic "reportNoInstance" (ppr wanted)
2999 ([], _) -> Left wanted -- No match
3000 res -> Right (mk_overlap_msg wanted res)
3002 (clas,tys) = getDictClassTys wanted
3004 mk_overlap_msg dict (matches, unifiers)
3005 = ASSERT( not (null matches) )
3006 vcat [ addInstLoc [dict] ((ptext SLIT("Overlapping instances for")
3007 <+> pprPred (dictPred dict))),
3008 sep [ptext SLIT("Matching instances") <> colon,
3009 nest 2 (vcat [pprInstances ispecs, pprInstances unifiers])],
3010 if not (isSingleton matches)
3011 then -- Two or more matches
3013 else -- One match, plus some unifiers
3014 ASSERT( not (null unifiers) )
3015 parens (vcat [ptext SLIT("The choice depends on the instantiation of") <+>
3016 quotes (pprWithCommas ppr (varSetElems (tyVarsOfInst dict))),
3017 ptext SLIT("To pick the first instance above, use -fallow-incoherent-instances"),
3018 ptext SLIT("when compiling the other instance declarations")])]
3020 ispecs = [ispec | (ispec, _) <- matches]
3022 mk_no_inst_err insts
3023 | null insts = empty
3025 | Just (loc, givens) <- mb_what, -- Nested (type signatures, instance decls)
3026 not (isEmptyVarSet (tyVarsOfInsts insts))
3027 = vcat [ addInstLoc insts $
3028 sep [ ptext SLIT("Could not deduce") <+> pprDictsTheta insts
3029 , nest 2 $ ptext SLIT("from the context") <+> pprDictsTheta givens]
3030 , show_fixes (fix1 loc : fixes2) ]
3032 | otherwise -- Top level
3033 = vcat [ addInstLoc insts $
3034 ptext SLIT("No instance") <> plural insts
3035 <+> ptext SLIT("for") <+> pprDictsTheta insts
3036 , show_fixes fixes2 ]
3039 fix1 loc = sep [ ptext SLIT("add") <+> pprDictsTheta insts
3040 <+> ptext SLIT("to the context of"),
3041 nest 2 (ppr (instLocOrigin loc)) ]
3042 -- I'm not sure it helps to add the location
3043 -- nest 2 (ptext SLIT("at") <+> ppr (instLocSpan loc)) ]
3045 fixes2 | null instance_dicts = []
3046 | otherwise = [sep [ptext SLIT("add an instance declaration for"),
3047 pprDictsTheta instance_dicts]]
3048 instance_dicts = [d | d <- insts, isClassDict d, not (isTyVarDict d)]
3049 -- Insts for which it is worth suggesting an adding an instance declaration
3050 -- Exclude implicit parameters, and tyvar dicts
3052 show_fixes :: [SDoc] -> SDoc
3053 show_fixes [] = empty
3054 show_fixes (f:fs) = sep [ptext SLIT("Possible fix:"),
3055 nest 2 (vcat (f : map (ptext SLIT("or") <+>) fs))]
3057 addTopAmbigErrs dicts
3058 -- Divide into groups that share a common set of ambiguous tyvars
3059 = ifErrsM (return ()) $ -- Only report ambiguity if no other errors happened
3060 -- See Note [Avoiding spurious errors]
3061 mapM_ report (equivClasses cmp [(d, tvs_of d) | d <- tidy_dicts])
3063 (tidy_env, tidy_dicts) = tidyInsts dicts
3065 tvs_of :: Inst -> [TcTyVar]
3066 tvs_of d = varSetElems (tyVarsOfInst d)
3067 cmp (_,tvs1) (_,tvs2) = tvs1 `compare` tvs2
3069 report :: [(Inst,[TcTyVar])] -> TcM ()
3070 report pairs@((inst,tvs) : _) -- The pairs share a common set of ambiguous tyvars
3071 = mkMonomorphismMsg tidy_env tvs `thenM` \ (tidy_env, mono_msg) ->
3072 setSrcSpan (instSpan inst) $
3073 -- the location of the first one will do for the err message
3074 addErrTcM (tidy_env, msg $$ mono_msg)
3076 dicts = map fst pairs
3077 msg = sep [text "Ambiguous type variable" <> plural tvs <+>
3078 pprQuotedList tvs <+> in_msg,
3079 nest 2 (pprDictsInFull dicts)]
3080 in_msg = text "in the constraint" <> plural dicts <> colon
3081 report [] = panic "addTopAmbigErrs"
3084 mkMonomorphismMsg :: TidyEnv -> [TcTyVar] -> TcM (TidyEnv, Message)
3085 -- There's an error with these Insts; if they have free type variables
3086 -- it's probably caused by the monomorphism restriction.
3087 -- Try to identify the offending variable
3088 -- ASSUMPTION: the Insts are fully zonked
3089 mkMonomorphismMsg tidy_env inst_tvs
3090 = do { dflags <- getDOpts
3091 ; (tidy_env, docs) <- findGlobals (mkVarSet inst_tvs) tidy_env
3092 ; return (tidy_env, mk_msg dflags docs) }
3094 mk_msg _ _ | any isRuntimeUnk inst_tvs
3095 = vcat [ptext SLIT("Cannot resolve unknown runtime types:") <+>
3096 (pprWithCommas ppr inst_tvs),
3097 ptext SLIT("Use :print or :force to determine these types")]
3098 mk_msg _ [] = ptext SLIT("Probable fix: add a type signature that fixes these type variable(s)")
3099 -- This happens in things like
3100 -- f x = show (read "foo")
3101 -- where monomorphism doesn't play any role
3103 = vcat [ptext SLIT("Possible cause: the monomorphism restriction applied to the following:"),
3105 monomorphism_fix dflags]
3107 isRuntimeUnk :: TcTyVar -> Bool
3108 isRuntimeUnk x | SkolemTv RuntimeUnkSkol <- tcTyVarDetails x = True
3111 monomorphism_fix :: DynFlags -> SDoc
3112 monomorphism_fix dflags
3113 = ptext SLIT("Probable fix:") <+> vcat
3114 [ptext SLIT("give these definition(s) an explicit type signature"),
3115 if dopt Opt_MonomorphismRestriction dflags
3116 then ptext SLIT("or use -fno-monomorphism-restriction")
3117 else empty] -- Only suggest adding "-fno-monomorphism-restriction"
3118 -- if it is not already set!
3120 warnDefault ups default_ty
3121 = doptM Opt_WarnTypeDefaults `thenM` \ warn_flag ->
3122 addInstCtxt (instLoc (head (dicts))) (warnTc warn_flag warn_msg)
3124 dicts = [d | (d,_,_) <- ups]
3127 (_, tidy_dicts) = tidyInsts dicts
3128 warn_msg = vcat [ptext SLIT("Defaulting the following constraint(s) to type") <+>
3129 quotes (ppr default_ty),
3130 pprDictsInFull tidy_dicts]
3132 reduceDepthErr n stack
3133 = vcat [ptext SLIT("Context reduction stack overflow; size =") <+> int n,
3134 ptext SLIT("Use -fcontext-stack=N to increase stack size to N"),
3135 nest 4 (pprStack stack)]
3137 pprStack stack = vcat (map pprInstInFull stack)