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 { -- Givens are skolems, so no need to zonk them
1089 wanteds' <- zonkInsts wanteds
1091 ; (improved, binds, irreds, more_needed_givens) <- reduceContext env wanteds'
1093 ; let all_needed_givens = needed_givens ++ more_needed_givens
1095 ; if not improved then
1096 return (irreds, binds, all_needed_givens)
1099 -- If improvement did some unification, we go round again.
1100 -- We start again with irreds, not wanteds
1101 -- Using an instance decl might have introduced a fresh type variable
1102 -- which might have been unified, so we'd get an infinite loop
1103 -- if we started again with wanteds! See Note [LOOP]
1104 { (irreds1, binds1, all_needed_givens1) <- go env irreds all_needed_givens
1105 ; return (irreds1, binds `unionBags` binds1, all_needed_givens1) } }
1110 class If b t e r | b t e -> r
1113 class Lte a b c | a b -> c where lte :: a -> b -> c
1115 instance (Lte a b l,If l b a c) => Max a b c
1117 Wanted: Max Z (S x) y
1119 Then we'll reduce using the Max instance to:
1120 (Lte Z (S x) l, If l (S x) Z y)
1121 and improve by binding l->T, after which we can do some reduction
1122 on both the Lte and If constraints. What we *can't* do is start again
1123 with (Max Z (S x) y)!
1127 %************************************************************************
1129 tcSimplifySuperClasses
1131 %************************************************************************
1133 Note [SUPERCLASS-LOOP 1]
1134 ~~~~~~~~~~~~~~~~~~~~~~~~
1135 We have to be very, very careful when generating superclasses, lest we
1136 accidentally build a loop. Here's an example:
1140 class S a => C a where { opc :: a -> a }
1141 class S b => D b where { opd :: b -> b }
1143 instance C Int where
1146 instance D Int where
1149 From (instance C Int) we get the constraint set {ds1:S Int, dd:D Int}
1150 Simplifying, we may well get:
1151 $dfCInt = :C ds1 (opd dd)
1154 Notice that we spot that we can extract ds1 from dd.
1156 Alas! Alack! We can do the same for (instance D Int):
1158 $dfDInt = :D ds2 (opc dc)
1162 And now we've defined the superclass in terms of itself.
1164 Solution: never generate a superclass selectors at all when
1165 satisfying the superclass context of an instance declaration.
1167 Two more nasty cases are in
1172 tcSimplifySuperClasses
1177 tcSimplifySuperClasses loc givens sc_wanteds
1178 = do { traceTc (text "tcSimplifySuperClasses")
1179 ; (irreds,binds1,_) <- checkLoop env sc_wanteds
1180 ; let (tidy_env, tidy_irreds) = tidyInsts irreds
1181 ; reportNoInstances tidy_env (Just (loc, givens)) tidy_irreds
1184 env = mkRedEnv (pprInstLoc loc) try_me givens
1185 try_me inst = ReduceMe NoSCs
1186 -- Like tryHardCheckLoop, but with NoSCs
1190 %************************************************************************
1192 \subsection{tcSimplifyRestricted}
1194 %************************************************************************
1196 tcSimplifyRestricted infers which type variables to quantify for a
1197 group of restricted bindings. This isn't trivial.
1200 We want to quantify over a to get id :: forall a. a->a
1203 We do not want to quantify over a, because there's an Eq a
1204 constraint, so we get eq :: a->a->Bool (notice no forall)
1207 RHS has type 'tau', whose free tyvars are tau_tvs
1208 RHS has constraints 'wanteds'
1211 Quantify over (tau_tvs \ ftvs(wanteds))
1212 This is bad. The constraints may contain (Monad (ST s))
1213 where we have instance Monad (ST s) where...
1214 so there's no need to be monomorphic in s!
1216 Also the constraint might be a method constraint,
1217 whose type mentions a perfectly innocent tyvar:
1218 op :: Num a => a -> b -> a
1219 Here, b is unconstrained. A good example would be
1221 We want to infer the polymorphic type
1222 foo :: forall b. b -> b
1225 Plan B (cunning, used for a long time up to and including GHC 6.2)
1226 Step 1: Simplify the constraints as much as possible (to deal
1227 with Plan A's problem). Then set
1228 qtvs = tau_tvs \ ftvs( simplify( wanteds ) )
1230 Step 2: Now simplify again, treating the constraint as 'free' if
1231 it does not mention qtvs, and trying to reduce it otherwise.
1232 The reasons for this is to maximise sharing.
1234 This fails for a very subtle reason. Suppose that in the Step 2
1235 a constraint (Foo (Succ Zero) (Succ Zero) b) gets thrown upstairs as 'free'.
1236 In the Step 1 this constraint might have been simplified, perhaps to
1237 (Foo Zero Zero b), AND THEN THAT MIGHT BE IMPROVED, to bind 'b' to 'T'.
1238 This won't happen in Step 2... but that in turn might prevent some other
1239 constraint (Baz [a] b) being simplified (e.g. via instance Baz [a] T where {..})
1240 and that in turn breaks the invariant that no constraints are quantified over.
1242 Test typecheck/should_compile/tc177 (which failed in GHC 6.2) demonstrates
1247 Step 1: Simplify the constraints as much as possible (to deal
1248 with Plan A's problem). Then set
1249 qtvs = tau_tvs \ ftvs( simplify( wanteds ) )
1250 Return the bindings from Step 1.
1253 A note about Plan C (arising from "bug" reported by George Russel March 2004)
1256 instance (HasBinary ty IO) => HasCodedValue ty
1258 foo :: HasCodedValue a => String -> IO a
1260 doDecodeIO :: HasCodedValue a => () -> () -> IO a
1261 doDecodeIO codedValue view
1262 = let { act = foo "foo" } in act
1264 You might think this should work becuase the call to foo gives rise to a constraint
1265 (HasCodedValue t), which can be satisfied by the type sig for doDecodeIO. But the
1266 restricted binding act = ... calls tcSimplifyRestricted, and PlanC simplifies the
1267 constraint using the (rather bogus) instance declaration, and now we are stuffed.
1269 I claim this is not really a bug -- but it bit Sergey as well as George. So here's
1273 Plan D (a variant of plan B)
1274 Step 1: Simplify the constraints as much as possible (to deal
1275 with Plan A's problem), BUT DO NO IMPROVEMENT. Then set
1276 qtvs = tau_tvs \ ftvs( simplify( wanteds ) )
1278 Step 2: Now simplify again, treating the constraint as 'free' if
1279 it does not mention qtvs, and trying to reduce it otherwise.
1281 The point here is that it's generally OK to have too few qtvs; that is,
1282 to make the thing more monomorphic than it could be. We don't want to
1283 do that in the common cases, but in wierd cases it's ok: the programmer
1284 can always add a signature.
1286 Too few qtvs => too many wanteds, which is what happens if you do less
1291 tcSimplifyRestricted -- Used for restricted binding groups
1292 -- i.e. ones subject to the monomorphism restriction
1295 -> [Name] -- Things bound in this group
1296 -> TcTyVarSet -- Free in the type of the RHSs
1297 -> [Inst] -- Free in the RHSs
1298 -> TcM ([TyVar], -- Tyvars to quantify (zonked and quantified)
1299 TcDictBinds) -- Bindings
1300 -- tcSimpifyRestricted returns no constraints to
1301 -- quantify over; by definition there are none.
1302 -- They are all thrown back in the LIE
1304 tcSimplifyRestricted doc top_lvl bndrs tau_tvs wanteds
1305 -- Zonk everything in sight
1306 = do { traceTc (text "tcSimplifyRestricted")
1307 ; wanteds' <- zonkInsts wanteds
1309 -- 'ReduceMe': Reduce as far as we can. Don't stop at
1310 -- dicts; the idea is to get rid of as many type
1311 -- variables as possible, and we don't want to stop
1312 -- at (say) Monad (ST s), because that reduces
1313 -- immediately, with no constraint on s.
1315 -- BUT do no improvement! See Plan D above
1316 -- HOWEVER, some unification may take place, if we instantiate
1317 -- a method Inst with an equality constraint
1318 ; let env = mkNoImproveRedEnv doc (\i -> ReduceMe AddSCs)
1319 ; (_imp, _binds, constrained_dicts, _) <- reduceContext env wanteds'
1321 -- Next, figure out the tyvars we will quantify over
1322 ; tau_tvs' <- zonkTcTyVarsAndFV (varSetElems tau_tvs)
1323 ; gbl_tvs' <- tcGetGlobalTyVars
1324 ; constrained_dicts' <- zonkInsts constrained_dicts
1326 ; let qtvs1 = tau_tvs' `minusVarSet` oclose (fdPredsOfInsts constrained_dicts) gbl_tvs'
1327 -- As in tcSimplifyInfer
1329 -- Do not quantify over constrained type variables:
1330 -- this is the monomorphism restriction
1331 constrained_tvs' = tyVarsOfInsts constrained_dicts'
1332 qtvs = qtvs1 `minusVarSet` constrained_tvs'
1333 pp_bndrs = pprWithCommas (quotes . ppr) bndrs
1336 ; warn_mono <- doptM Opt_WarnMonomorphism
1337 ; warnTc (warn_mono && (constrained_tvs' `intersectsVarSet` qtvs1))
1338 (vcat[ ptext SLIT("the Monomorphism Restriction applies to the binding")
1339 <> plural bndrs <+> ptext SLIT("for") <+> pp_bndrs,
1340 ptext SLIT("Consider giving a type signature for") <+> pp_bndrs])
1342 ; traceTc (text "tcSimplifyRestricted" <+> vcat [
1343 pprInsts wanteds, pprInsts constrained_dicts',
1345 ppr constrained_tvs', ppr tau_tvs', ppr qtvs ])
1347 -- The first step may have squashed more methods than
1348 -- necessary, so try again, this time more gently, knowing the exact
1349 -- set of type variables to quantify over.
1351 -- We quantify only over constraints that are captured by qtvs;
1352 -- these will just be a subset of non-dicts. This in contrast
1353 -- to normal inference (using isFreeWhenInferring) in which we quantify over
1354 -- all *non-inheritable* constraints too. This implements choice
1355 -- (B) under "implicit parameter and monomorphism" above.
1357 -- Remember that we may need to do *some* simplification, to
1358 -- (for example) squash {Monad (ST s)} into {}. It's not enough
1359 -- just to float all constraints
1361 -- At top level, we *do* squash methods becuase we want to
1362 -- expose implicit parameters to the test that follows
1363 ; let is_nested_group = isNotTopLevel top_lvl
1364 try_me inst | isFreeWrtTyVars qtvs inst,
1365 (is_nested_group || isDict inst) = Stop
1366 | otherwise = ReduceMe AddSCs
1367 env = mkNoImproveRedEnv doc try_me
1368 ; (_imp, binds, irreds, _) <- reduceContext env wanteds'
1370 -- See "Notes on implicit parameters, Question 4: top level"
1371 ; ASSERT( all (isFreeWrtTyVars qtvs) irreds ) -- None should be captured
1372 if is_nested_group then
1374 else do { let (bad_ips, non_ips) = partition isIPDict irreds
1375 ; addTopIPErrs bndrs bad_ips
1376 ; extendLIEs non_ips }
1378 ; qtvs' <- zonkQuantifiedTyVars (varSetElems qtvs)
1379 ; return (qtvs', binds) }
1383 %************************************************************************
1387 %************************************************************************
1389 On the LHS of transformation rules we only simplify methods and constants,
1390 getting dictionaries. We want to keep all of them unsimplified, to serve
1391 as the available stuff for the RHS of the rule.
1393 Example. Consider the following left-hand side of a rule
1395 f (x == y) (y > z) = ...
1397 If we typecheck this expression we get constraints
1399 d1 :: Ord a, d2 :: Eq a
1401 We do NOT want to "simplify" to the LHS
1403 forall x::a, y::a, z::a, d1::Ord a.
1404 f ((==) (eqFromOrd d1) x y) ((>) d1 y z) = ...
1408 forall x::a, y::a, z::a, d1::Ord a, d2::Eq a.
1409 f ((==) d2 x y) ((>) d1 y z) = ...
1411 Here is another example:
1413 fromIntegral :: (Integral a, Num b) => a -> b
1414 {-# RULES "foo" fromIntegral = id :: Int -> Int #-}
1416 In the rule, a=b=Int, and Num Int is a superclass of Integral Int. But
1417 we *dont* want to get
1419 forall dIntegralInt.
1420 fromIntegral Int Int dIntegralInt (scsel dIntegralInt) = id Int
1422 because the scsel will mess up RULE matching. Instead we want
1424 forall dIntegralInt, dNumInt.
1425 fromIntegral Int Int dIntegralInt dNumInt = id Int
1429 g (x == y) (y == z) = ..
1431 where the two dictionaries are *identical*, we do NOT WANT
1433 forall x::a, y::a, z::a, d1::Eq a
1434 f ((==) d1 x y) ((>) d1 y z) = ...
1436 because that will only match if the dict args are (visibly) equal.
1437 Instead we want to quantify over the dictionaries separately.
1439 In short, tcSimplifyRuleLhs must *only* squash LitInst and MethInts, leaving
1440 all dicts unchanged, with absolutely no sharing. It's simpler to do this
1441 from scratch, rather than further parameterise simpleReduceLoop etc
1444 tcSimplifyRuleLhs :: [Inst] -> TcM ([Inst], TcDictBinds)
1445 tcSimplifyRuleLhs wanteds
1446 = go [] emptyBag wanteds
1449 = return (dicts, binds)
1450 go dicts binds (w:ws)
1452 = go (w:dicts) binds ws
1454 = do { w' <- zonkInst w -- So that (3::Int) does not generate a call
1455 -- to fromInteger; this looks fragile to me
1456 ; lookup_result <- lookupSimpleInst w'
1457 ; case lookup_result of
1459 go dicts (addInstToDictBind binds w rhs) (ws' ++ ws)
1460 NoInstance -> pprPanic "tcSimplifyRuleLhs" (ppr w)
1464 tcSimplifyBracket is used when simplifying the constraints arising from
1465 a Template Haskell bracket [| ... |]. We want to check that there aren't
1466 any constraints that can't be satisfied (e.g. Show Foo, where Foo has no
1467 Show instance), but we aren't otherwise interested in the results.
1468 Nor do we care about ambiguous dictionaries etc. We will type check
1469 this bracket again at its usage site.
1472 tcSimplifyBracket :: [Inst] -> TcM ()
1473 tcSimplifyBracket wanteds
1474 = do { tryHardCheckLoop doc wanteds
1477 doc = text "tcSimplifyBracket"
1481 %************************************************************************
1483 \subsection{Filtering at a dynamic binding}
1485 %************************************************************************
1490 we must discharge all the ?x constraints from B. We also do an improvement
1491 step; if we have ?x::t1 and ?x::t2 we must unify t1, t2.
1493 Actually, the constraints from B might improve the types in ?x. For example
1495 f :: (?x::Int) => Char -> Char
1498 then the constraint (?x::Int) arising from the call to f will
1499 force the binding for ?x to be of type Int.
1502 tcSimplifyIPs :: [Inst] -- The implicit parameters bound here
1505 -- We need a loop so that we do improvement, and then
1506 -- (next time round) generate a binding to connect the two
1508 -- Here the two ?x's have different types, and improvement
1509 -- makes them the same.
1511 tcSimplifyIPs given_ips wanteds
1512 = do { wanteds' <- zonkInsts wanteds
1513 ; given_ips' <- zonkInsts given_ips
1514 -- Unusually for checking, we *must* zonk the given_ips
1516 ; let env = mkRedEnv doc try_me given_ips'
1517 ; (improved, binds, irreds, _) <- reduceContext env wanteds'
1519 ; if not improved then
1520 ASSERT( all is_free irreds )
1521 do { extendLIEs irreds
1524 tcSimplifyIPs given_ips wanteds }
1526 doc = text "tcSimplifyIPs" <+> ppr given_ips
1527 ip_set = mkNameSet (ipNamesOfInsts given_ips)
1528 is_free inst = isFreeWrtIPs ip_set inst
1530 -- Simplify any methods that mention the implicit parameter
1531 try_me inst | is_free inst = Stop
1532 | otherwise = ReduceMe NoSCs
1536 %************************************************************************
1538 \subsection[binds-for-local-funs]{@bindInstsOfLocalFuns@}
1540 %************************************************************************
1542 When doing a binding group, we may have @Insts@ of local functions.
1543 For example, we might have...
1545 let f x = x + 1 -- orig local function (overloaded)
1546 f.1 = f Int -- two instances of f
1551 The point is: we must drop the bindings for @f.1@ and @f.2@ here,
1552 where @f@ is in scope; those @Insts@ must certainly not be passed
1553 upwards towards the top-level. If the @Insts@ were binding-ified up
1554 there, they would have unresolvable references to @f@.
1556 We pass in an @init_lie@ of @Insts@ and a list of locally-bound @Ids@.
1557 For each method @Inst@ in the @init_lie@ that mentions one of the
1558 @Ids@, we create a binding. We return the remaining @Insts@ (in an
1559 @LIE@), as well as the @HsBinds@ generated.
1562 bindInstsOfLocalFuns :: [Inst] -> [TcId] -> TcM TcDictBinds
1563 -- Simlifies only MethodInsts, and generate only bindings of form
1565 -- We're careful not to even generate bindings of the form
1567 -- You'd think that'd be fine, but it interacts with what is
1568 -- arguably a bug in Match.tidyEqnInfo (see notes there)
1570 bindInstsOfLocalFuns wanteds local_ids
1571 | null overloaded_ids
1573 = extendLIEs wanteds `thenM_`
1574 returnM emptyLHsBinds
1577 = do { (irreds, binds,_) <- checkLoop env for_me
1578 ; extendLIEs not_for_me
1582 env = mkRedEnv doc try_me []
1583 doc = text "bindInsts" <+> ppr local_ids
1584 overloaded_ids = filter is_overloaded local_ids
1585 is_overloaded id = isOverloadedTy (idType id)
1586 (for_me, not_for_me) = partition (isMethodFor overloaded_set) wanteds
1588 overloaded_set = mkVarSet overloaded_ids -- There can occasionally be a lot of them
1589 -- so it's worth building a set, so that
1590 -- lookup (in isMethodFor) is faster
1591 try_me inst | isMethod inst = ReduceMe NoSCs
1596 %************************************************************************
1598 \subsection{Data types for the reduction mechanism}
1600 %************************************************************************
1602 The main control over context reduction is here
1606 = RedEnv { red_doc :: SDoc -- The context
1607 , red_try_me :: Inst -> WhatToDo
1608 , red_improve :: Bool -- True <=> do improvement
1609 , red_givens :: [Inst] -- All guaranteed rigid
1611 -- but see Note [Rigidity]
1612 , red_stack :: (Int, [Inst]) -- Recursion stack (for err msg)
1613 -- See Note [RedStack]
1617 -- The red_givens are rigid so far as cmpInst is concerned.
1618 -- There is one case where they are not totally rigid, namely in tcSimplifyIPs
1619 -- let ?x = e in ...
1620 -- Here, the given is (?x::a), where 'a' is not necy a rigid type
1621 -- But that doesn't affect the comparison, which is based only on mame.
1624 -- The red_stack pair (n,insts) pair is just used for error reporting.
1625 -- 'n' is always the depth of the stack.
1626 -- The 'insts' is the stack of Insts being reduced: to produce X
1627 -- I had to produce Y, to produce Y I had to produce Z, and so on.
1630 mkRedEnv :: SDoc -> (Inst -> WhatToDo) -> [Inst] -> RedEnv
1631 mkRedEnv doc try_me givens
1632 = RedEnv { red_doc = doc, red_try_me = try_me,
1633 red_givens = givens, red_stack = (0,[]),
1634 red_improve = True }
1636 mkNoImproveRedEnv :: SDoc -> (Inst -> WhatToDo) -> RedEnv
1637 -- Do not do improvement; no givens
1638 mkNoImproveRedEnv doc try_me
1639 = RedEnv { red_doc = doc, red_try_me = try_me,
1640 red_givens = [], red_stack = (0,[]),
1641 red_improve = True }
1644 = ReduceMe WantSCs -- Try to reduce this
1645 -- If there's no instance, add the inst to the
1646 -- irreductible ones, but don't produce an error
1647 -- message of any kind.
1648 -- It might be quite legitimate such as (Eq a)!
1650 | Stop -- Return as irreducible unless it can
1651 -- be reduced to a constant in one step
1652 -- Do not add superclasses; see
1654 data WantSCs = NoSCs | AddSCs -- Tells whether we should add the superclasses
1655 -- of a predicate when adding it to the avails
1656 -- The reason for this flag is entirely the super-class loop problem
1657 -- Note [SUPER-CLASS LOOP 1]
1661 %************************************************************************
1663 \subsection[reduce]{@reduce@}
1665 %************************************************************************
1667 Note [Ancestor Equalities]
1668 ~~~~~~~~~~~~~~~~~~~~~~~~~~
1669 During context reduction, we add to the wanted equalities also those
1670 equalities that (transitively) occur in superclass contexts of wanted
1671 class constraints. Consider the following code
1673 class a ~ Int => C a
1676 If (C a) is wanted, we want to add (a ~ Int), which will be discharged by
1677 substituting Int for a. Hence, we ultimately want (C Int), which we
1678 discharge with the explicit instance.
1681 reduceContext :: RedEnv
1683 -> TcM (ImprovementDone,
1684 TcDictBinds, -- Dictionary bindings
1685 [Inst], -- Irreducible
1686 [Inst]) -- Needed givens
1688 reduceContext env wanteds
1689 = do { traceTc (text "reduceContext" <+> (vcat [
1690 text "----------------------",
1692 text "given" <+> ppr (red_givens env),
1693 text "wanted" <+> ppr wanteds,
1694 text "----------------------"
1697 ; let givens = red_givens env
1698 (given_eqs0, given_dicts0) = partition isEqInst givens
1699 (wanted_eqs0, wanted_dicts) = partition isEqInst wanteds
1701 -- We want to add as wanted equalities those that (transitively)
1702 -- occur in superclass contexts of wanted class constraints.
1703 -- See Note [Ancestor Equalities]
1704 ; ancestor_eqs <- ancestorEqualities wanted_dicts
1705 ; let wanted_eqs = wanted_eqs0 ++ ancestor_eqs
1706 ; traceTc $ text "reduceContext: ancestor eqs" <+> ppr ancestor_eqs
1708 -- 1. Normalise the *given* *equality* constraints
1709 ; (given_eqs, eliminate_skolems) <- normaliseGivenEqs given_eqs0
1711 -- 2. Normalise the *given* *dictionary* constraints
1712 -- wrt. the toplevel and given equations
1713 ; (given_dicts, given_binds) <- normaliseGivenDicts given_eqs
1716 -- 3. Solve the *wanted* *equation* constraints
1717 ; eq_irreds0 <- solveWantedEqs given_eqs wanted_eqs
1719 -- 4. Normalise the *wanted* equality constraints with respect to
1721 ; eq_irreds <- normaliseWantedEqs eq_irreds0
1723 -- 5. Build the Avail mapping from "given_dicts"
1724 ; init_state <- foldlM addGiven emptyAvails given_dicts
1726 -- 6. Solve the *wanted* *dictionary* constraints
1727 -- This may expose some further equational constraints...
1728 ; wanted_dicts' <- zonkInsts wanted_dicts
1729 ; avails <- reduceList env wanted_dicts' init_state
1730 ; (binds, irreds0, needed_givens) <- extractResults avails wanted_dicts'
1731 ; traceTc $ text "reduceContext extractresults" <+> vcat
1732 [ppr avails,ppr wanted_dicts',ppr binds,ppr needed_givens]
1734 -- 7. Normalise the *wanted* *dictionary* constraints
1735 -- wrt. the toplevel and given equations
1736 ; (irreds1,normalise_binds1) <- normaliseWantedDicts given_eqs irreds0
1738 -- 8. Substitute the wanted *equations* in the wanted *dictionaries*
1739 ; (irreds,normalise_binds2) <- substEqInDictInsts eq_irreds irreds1
1741 -- 9. eliminate the artificial skolem constants introduced in 1.
1744 -- If there was some FD improvement,
1745 -- or new wanted equations have been exposed,
1746 -- we should have another go at solving.
1747 ; let improved = availsImproved avails
1748 || (not $ isEmptyBag normalise_binds1)
1749 || (not $ isEmptyBag normalise_binds2)
1750 || (any isEqInst irreds)
1752 ; traceTc (text "reduceContext end" <+> (vcat [
1753 text "----------------------",
1755 text "given" <+> ppr (red_givens env),
1756 text "wanted" <+> ppr wanteds,
1758 text "avails" <+> pprAvails avails,
1759 text "improved =" <+> ppr improved,
1760 text "irreds = " <+> ppr irreds,
1761 text "binds = " <+> ppr binds,
1762 text "needed givens = " <+> ppr needed_givens,
1763 text "----------------------"
1767 given_binds `unionBags` normalise_binds1
1768 `unionBags` normalise_binds2
1770 irreds ++ eq_irreds,
1774 tcImproveOne :: Avails -> Inst -> TcM ImprovementDone
1775 tcImproveOne avails inst
1776 | not (isDict inst) = return False
1778 = do { inst_envs <- tcGetInstEnvs
1779 ; let eqns = improveOne (classInstances inst_envs)
1780 (dictPred inst, pprInstArising inst)
1781 [ (dictPred p, pprInstArising p)
1782 | p <- availsInsts avails, isDict p ]
1783 -- Avails has all the superclasses etc (good)
1784 -- It also has all the intermediates of the deduction (good)
1785 -- It does not have duplicates (good)
1786 -- NB that (?x::t1) and (?x::t2) will be held separately in avails
1787 -- so that improve will see them separate
1788 ; traceTc (text "improveOne" <+> ppr inst)
1791 unifyEqns :: [(Equation,(PredType,SDoc),(PredType,SDoc))]
1792 -> TcM ImprovementDone
1793 unifyEqns [] = return False
1795 = do { traceTc (ptext SLIT("Improve:") <+> vcat (map pprEquationDoc eqns))
1799 unify ((qtvs, pairs), what1, what2)
1800 = addErrCtxtM (mkEqnMsg what1 what2) $
1801 tcInstTyVars (varSetElems qtvs) `thenM` \ (_, _, tenv) ->
1802 mapM_ (unif_pr tenv) pairs
1803 unif_pr tenv (ty1,ty2) = unifyType (substTy tenv ty1) (substTy tenv ty2)
1805 pprEquationDoc (eqn, (p1,w1), (p2,w2)) = vcat [pprEquation eqn, nest 2 (ppr p1), nest 2 (ppr p2)]
1807 mkEqnMsg (pred1,from1) (pred2,from2) tidy_env
1808 = do { pred1' <- zonkTcPredType pred1; pred2' <- zonkTcPredType pred2
1809 ; let { pred1'' = tidyPred tidy_env pred1'; pred2'' = tidyPred tidy_env pred2' }
1810 ; let msg = vcat [ptext SLIT("When using functional dependencies to combine"),
1811 nest 2 (sep [ppr pred1'' <> comma, nest 2 from1]),
1812 nest 2 (sep [ppr pred2'' <> comma, nest 2 from2])]
1813 ; return (tidy_env, msg) }
1816 The main context-reduction function is @reduce@. Here's its game plan.
1819 reduceList :: RedEnv -> [Inst] -> Avails -> TcM Avails
1820 reduceList env@(RedEnv {red_stack = (n,stk)}) wanteds state
1821 = do { dopts <- getDOpts
1824 dumpTcRn (hang (ptext SLIT("Interesting! Context reduction stack depth") <+> int n)
1825 2 (ifPprDebug (nest 2 (pprStack stk))))
1828 ; if n >= ctxtStkDepth dopts then
1829 failWithTc (reduceDepthErr n stk)
1833 go [] state = return state
1834 go (w:ws) state = do { traceTc (text "reduceList " <+> ppr (w:ws) <+> ppr state)
1835 ; state' <- reduce (env {red_stack = (n+1, w:stk)}) w state
1838 -- Base case: we're done!
1839 reduce env wanted avails
1840 -- It's the same as an existing inst, or a superclass thereof
1841 | Just avail <- findAvail avails wanted
1842 = do { traceTc (text "reduce: found " <+> ppr wanted)
1847 = do { traceTc (text "reduce" <+> ppr avails <+> ppr wanted)
1848 ; case red_try_me env wanted of {
1849 Stop -> try_simple (addIrred NoSCs);
1850 -- See Note [No superclasses for Stop]
1852 ReduceMe want_scs -> do -- It should be reduced
1853 { (avails, lookup_result) <- reduceInst env avails wanted
1854 ; case lookup_result of
1855 NoInstance -> addIrred want_scs avails wanted
1856 -- Add it and its superclasses
1858 GenInst [] rhs -> addWanted want_scs avails wanted rhs []
1860 GenInst wanteds' rhs
1861 -> do { avails1 <- addIrred NoSCs avails wanted
1862 ; avails2 <- reduceList env wanteds' avails1
1863 ; addWanted want_scs avails2 wanted rhs wanteds' } }
1864 -- Temporarily do addIrred *before* the reduceList,
1865 -- which has the effect of adding the thing we are trying
1866 -- to prove to the database before trying to prove the things it
1867 -- needs. See note [RECURSIVE DICTIONARIES]
1868 -- NB: we must not do an addWanted before, because that adds the
1869 -- superclasses too, and that can lead to a spurious loop; see
1870 -- the examples in [SUPERCLASS-LOOP]
1871 -- So we do an addIrred before, and then overwrite it afterwards with addWanted
1874 -- First, see if the inst can be reduced to a constant in one step
1875 -- Works well for literals (1::Int) and constant dictionaries (d::Num Int)
1876 -- Don't bother for implication constraints, which take real work
1877 try_simple do_this_otherwise
1878 = do { res <- lookupSimpleInst wanted
1880 GenInst [] rhs -> addWanted AddSCs avails wanted rhs []
1881 other -> do_this_otherwise avails wanted }
1885 Note [SUPERCLASS-LOOP 2]
1886 ~~~~~~~~~~~~~~~~~~~~~~~~
1887 But the above isn't enough. Suppose we are *given* d1:Ord a,
1888 and want to deduce (d2:C [a]) where
1890 class Ord a => C a where
1891 instance Ord [a] => C [a] where ...
1893 Then we'll use the instance decl to deduce C [a] from Ord [a], and then add the
1894 superclasses of C [a] to avails. But we must not overwrite the binding
1895 for Ord [a] (which is obtained from Ord a) with a superclass selection or we'll just
1898 Here's another variant, immortalised in tcrun020
1899 class Monad m => C1 m
1900 class C1 m => C2 m x
1901 instance C2 Maybe Bool
1902 For the instance decl we need to build (C1 Maybe), and it's no good if
1903 we run around and add (C2 Maybe Bool) and its superclasses to the avails
1904 before we search for C1 Maybe.
1906 Here's another example
1907 class Eq b => Foo a b
1908 instance Eq a => Foo [a] a
1912 we'll first deduce that it holds (via the instance decl). We must not
1913 then overwrite the Eq t constraint with a superclass selection!
1915 At first I had a gross hack, whereby I simply did not add superclass constraints
1916 in addWanted, though I did for addGiven and addIrred. This was sub-optimal,
1917 becuase it lost legitimate superclass sharing, and it still didn't do the job:
1918 I found a very obscure program (now tcrun021) in which improvement meant the
1919 simplifier got two bites a the cherry... so something seemed to be an Stop
1920 first time, but reducible next time.
1922 Now we implement the Right Solution, which is to check for loops directly
1923 when adding superclasses. It's a bit like the occurs check in unification.
1926 Note [RECURSIVE DICTIONARIES]
1927 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1929 data D r = ZeroD | SuccD (r (D r));
1931 instance (Eq (r (D r))) => Eq (D r) where
1932 ZeroD == ZeroD = True
1933 (SuccD a) == (SuccD b) = a == b
1936 equalDC :: D [] -> D [] -> Bool;
1939 We need to prove (Eq (D [])). Here's how we go:
1943 by instance decl, holds if
1947 by instance decl of Eq, holds if
1949 where d2 = dfEqList d3
1952 But now we can "tie the knot" to give
1958 and it'll even run! The trick is to put the thing we are trying to prove
1959 (in this case Eq (D []) into the database before trying to prove its
1960 contributing clauses.
1963 %************************************************************************
1965 Reducing a single constraint
1967 %************************************************************************
1970 ---------------------------------------------
1971 reduceInst :: RedEnv -> Avails -> Inst -> TcM (Avails, LookupInstResult)
1972 reduceInst env avails (ImplicInst { tci_tyvars = tvs, tci_reft = reft, tci_loc = loc,
1973 tci_given = extra_givens, tci_wanted = wanteds })
1974 = reduceImplication env avails reft tvs extra_givens wanteds loc
1976 reduceInst env avails other_inst
1977 = do { result <- lookupSimpleInst other_inst
1978 ; return (avails, result) }
1981 Note [Equational Constraints in Implication Constraints]
1982 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1984 An equational constraint is of the form
1986 where Given and Wanted may contain both equational and dictionary
1987 constraints. The delay and reduction of these two kinds of constraints
1990 -) In the generated code, wanted Dictionary constraints are wrapped up in an
1991 implication constraint that is created at the code site where the wanted
1992 dictionaries can be reduced via a let-binding. This let-bound implication
1993 constraint is deconstructed at the use-site of the wanted dictionaries.
1995 -) While the reduction of equational constraints is also delayed, the delay
1996 is not manifest in the generated code. The required evidence is generated
1997 in the code directly at the use-site. There is no let-binding and deconstruction
1998 necessary. The main disadvantage is that we cannot exploit sharing as the
1999 same evidence may be generated at multiple use-sites. However, this disadvantage
2000 is limited because it only concerns coercions which are erased.
2002 The different treatment is motivated by the different in representation. Dictionary
2003 constraints require manifest runtime dictionaries, while equations require coercions
2007 ---------------------------------------------
2008 reduceImplication :: RedEnv
2010 -> Refinement -- May refine the givens; often empty
2011 -> [TcTyVar] -- Quantified type variables; all skolems
2012 -> [Inst] -- Extra givens; all rigid
2015 -> TcM (Avails, LookupInstResult)
2018 Suppose we are simplifying the constraint
2019 forall bs. extras => wanted
2020 in the context of an overall simplification problem with givens 'givens',
2021 and refinment 'reft'.
2024 * The refinement is often empty
2026 * The 'extra givens' need not mention any of the quantified type variables
2027 e.g. forall {}. Eq a => Eq [a]
2028 forall {}. C Int => D (Tree Int)
2030 This happens when you have something like
2032 T1 :: Eq a => a -> T a
2035 f x = ...(case x of { T1 v -> v==v })...
2038 -- ToDo: should we instantiate tvs? I think it's not necessary
2040 -- Note on coercion variables:
2042 -- The extra given coercion variables are bound at two different sites:
2043 -- -) in the creation context of the implication constraint
2044 -- the solved equational constraints use these binders
2046 -- -) at the solving site of the implication constraint
2047 -- the solved dictionaries use these binders
2048 -- these binders are generated by reduceImplication
2050 reduceImplication env orig_avails reft tvs extra_givens wanteds inst_loc
2051 = do { -- Add refined givens, and the extra givens
2053 (refined_red_givens,refined_avails)
2054 <- if isEmptyRefinement reft then return (red_givens env,orig_avails)
2055 else foldlM (addRefinedGiven reft) ([],orig_avails) (red_givens env)
2057 -- Solve the sub-problem
2058 ; let try_me inst = ReduceMe AddSCs -- Note [Freeness and implications]
2059 env' = env { red_givens = refined_red_givens ++ extra_givens ++ availsInsts orig_avails
2060 , red_try_me = try_me }
2062 ; traceTc (text "reduceImplication" <+> vcat
2064 ppr (red_givens env), ppr extra_givens,
2065 ppr reft, ppr wanteds])
2066 ; (irreds,binds,needed_givens0) <- checkLoop env' wanteds
2067 ; let needed_givens1 = [ng | ng <- needed_givens0, notElem ng extra_givens]
2069 -- Note [Reducing implication constraints]
2070 -- Tom -- update note, put somewhere!
2072 ; traceTc (text "reduceImplication result" <+> vcat
2073 [ppr irreds, ppr binds, ppr needed_givens1])
2074 -- ; avails <- reduceList env' wanteds avails
2076 -- -- Extract the binding
2077 -- ; (binds, irreds) <- extractResults avails wanteds
2078 ; (refinement_binds,needed_givens) <- extractLocalResults refined_avails needed_givens1
2079 ; traceTc (text "reduceImplication local results" <+> vcat
2080 [ppr refinement_binds, ppr needed_givens])
2082 ; -- extract superclass binds
2083 -- (sc_binds,_) <- extractResults avails []
2084 -- ; traceTc (text "reduceImplication sc_binds" <+> vcat
2085 -- [ppr sc_binds, ppr avails])
2088 -- We always discard the extra avails we've generated;
2089 -- but we remember if we have done any (global) improvement
2090 -- ; let ret_avails = avails
2091 ; let ret_avails = orig_avails
2092 -- ; let ret_avails = updateImprovement orig_avails avails
2094 ; traceTc (text "reduceImplication condition" <+> ppr ((isEmptyLHsBinds binds) || (null irreds)))
2096 -- Porgress is no longer measered by the number of bindings
2097 -- ; if isEmptyLHsBinds binds then -- No progress
2098 ; if (isEmptyLHsBinds binds) && (not $ null irreds) then
2099 return (ret_avails, NoInstance)
2102 ; (implic_insts, bind) <- makeImplicationBind inst_loc tvs reft extra_givens irreds
2103 -- This binding is useless if the recursive simplification
2104 -- made no progress; but currently we don't try to optimise that
2105 -- case. After all, we only try hard to reduce at top level, or
2106 -- when inferring types.
2108 ; let dict_wanteds = filter (not . isEqInst) wanteds
2109 (extra_eq_givens, extra_dict_givens) = partition isEqInst extra_givens
2110 dict_ids = map instToId extra_dict_givens
2111 -- TOMDO: given equational constraints bug!
2112 -- we need a different evidence for given
2113 -- equations depending on whether we solve
2114 -- dictionary constraints or equational constraints
2115 eq_tyvars = uniqSetToList $ tyVarsOfTypes $ map eqInstType extra_eq_givens
2116 -- dict_ids = map instToId extra_givens
2117 co = mkWpTyLams tvs <.> mkWpTyLams eq_tyvars <.> mkWpLams dict_ids <.> WpLet (binds `unionBags` refinement_binds `unionBags` bind)
2118 rhs = mkHsWrap co payload
2119 loc = instLocSpan inst_loc
2120 payload | [dict_wanted] <- dict_wanteds = HsVar (instToId dict_wanted)
2121 | otherwise = ExplicitTuple (map (L loc . HsVar . instToId) dict_wanteds) Boxed
2124 ; traceTc (text "reduceImplication ->" <+> vcat
2127 -- If there are any irreds, we back off and return NoInstance
2128 ; return (ret_avails, GenInst (implic_insts ++ needed_givens) (L loc rhs))
2133 Note [Reducing implication constraints]
2134 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2135 Suppose we are trying to simplify
2136 (Ord a, forall b. C a b => (W [a] b, D c b))
2138 instance (C a b, Ord a) => W [a] b
2139 When solving the implication constraint, we'll start with
2141 in the Avails. Then we add (C a b -> Given) and solve. Extracting
2142 the results gives us a binding for the (W [a] b), with an Irred of
2143 (Ord a, D c b). Now, the (Ord a) comes from "outside" the implication,
2144 but the (D d b) is from "inside". So we want to generate a Rhs binding
2147 ic = /\b \dc:C a b). (df a b dc do, ic' b dc)
2150 ic' :: forall b. C a b => D c b
2152 The 'depending on' part of the Rhs is important, because it drives
2153 the extractResults code.
2155 The "inside" and "outside" distinction is what's going on with 'inner' and
2156 'outer' in reduceImplication
2159 Note [Freeness and implications]
2160 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2161 It's hard to say when an implication constraint can be floated out. Consider
2162 forall {} Eq a => Foo [a]
2163 The (Foo [a]) doesn't mention any of the quantified variables, but it
2164 still might be partially satisfied by the (Eq a).
2166 There is a useful special case when it *is* easy to partition the
2167 constraints, namely when there are no 'givens'. Consider
2168 forall {a}. () => Bar b
2169 There are no 'givens', and so there is no reason to capture (Bar b).
2170 We can let it float out. But if there is even one constraint we
2171 must be much more careful:
2172 forall {a}. C a b => Bar (m b)
2173 because (C a b) might have a superclass (D b), from which we might
2174 deduce (Bar [b]) when m later gets instantiated to []. Ha!
2176 Here is an even more exotic example
2178 Now consider the constraint
2179 forall b. D Int b => C Int
2180 We can satisfy the (C Int) from the superclass of D, so we don't want
2181 to float the (C Int) out, even though it mentions no type variable in
2184 Note [Pruning the givens in an implication constraint]
2185 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2186 Suppose we are about to form the implication constraint
2187 forall tvs. Eq a => Ord b
2188 The (Eq a) cannot contribute to the (Ord b), because it has no access to
2189 the type variable 'b'. So we could filter out the (Eq a) from the givens.
2191 Doing so would be a bit tidier, but all the implication constraints get
2192 simplified away by the optimiser, so it's no great win. So I don't take
2193 advantage of that at the moment.
2195 If you do, BE CAREFUL of wobbly type variables.
2198 %************************************************************************
2200 Avails and AvailHow: the pool of evidence
2202 %************************************************************************
2206 data Avails = Avails !ImprovementDone !AvailEnv
2208 type ImprovementDone = Bool -- True <=> some unification has happened
2209 -- so some Irreds might now be reducible
2210 -- keys that are now
2212 type AvailEnv = FiniteMap Inst AvailHow
2214 = IsIrred -- Used for irreducible dictionaries,
2215 -- which are going to be lambda bound
2217 | Given TcId -- Used for dictionaries for which we have a binding
2218 -- e.g. those "given" in a signature
2220 | Rhs -- Used when there is a RHS
2221 (LHsExpr TcId) -- The RHS
2222 [Inst] -- Insts free in the RHS; we need these too
2224 instance Outputable Avails where
2227 pprAvails (Avails imp avails)
2228 = vcat [ ptext SLIT("Avails") <> (if imp then ptext SLIT("[improved]") else empty)
2229 , nest 2 (vcat [sep [ppr inst, nest 2 (equals <+> ppr avail)]
2230 | (inst,avail) <- fmToList avails ])]
2232 instance Outputable AvailHow where
2235 -------------------------
2236 pprAvail :: AvailHow -> SDoc
2237 pprAvail IsIrred = text "Irred"
2238 pprAvail (Given x) = text "Given" <+> ppr x
2239 pprAvail (Rhs rhs bs) = text "Rhs" <+> sep [ppr rhs, braces (ppr bs)]
2241 -------------------------
2242 extendAvailEnv :: AvailEnv -> Inst -> AvailHow -> AvailEnv
2243 extendAvailEnv env inst avail = addToFM env inst avail
2245 findAvailEnv :: AvailEnv -> Inst -> Maybe AvailHow
2246 findAvailEnv env wanted = lookupFM env wanted
2247 -- NB 1: the Ord instance of Inst compares by the class/type info
2248 -- *not* by unique. So
2249 -- d1::C Int == d2::C Int
2251 emptyAvails :: Avails
2252 emptyAvails = Avails False emptyFM
2254 findAvail :: Avails -> Inst -> Maybe AvailHow
2255 findAvail (Avails _ avails) wanted = findAvailEnv avails wanted
2257 elemAvails :: Inst -> Avails -> Bool
2258 elemAvails wanted (Avails _ avails) = wanted `elemFM` avails
2260 extendAvails :: Avails -> Inst -> AvailHow -> TcM Avails
2262 extendAvails avails@(Avails imp env) inst avail
2263 = do { imp1 <- tcImproveOne avails inst -- Do any improvement
2264 ; return (Avails (imp || imp1) (extendAvailEnv env inst avail)) }
2266 availsInsts :: Avails -> [Inst]
2267 availsInsts (Avails _ avails) = keysFM avails
2269 availsImproved (Avails imp _) = imp
2271 updateImprovement :: Avails -> Avails -> Avails
2272 -- (updateImprovement a1 a2) sets a1's improvement flag from a2
2273 updateImprovement (Avails _ avails1) (Avails imp2 _) = Avails imp2 avails1
2276 Extracting the bindings from a bunch of Avails.
2277 The bindings do *not* come back sorted in dependency order.
2278 We assume that they'll be wrapped in a big Rec, so that the
2279 dependency analyser can sort them out later
2282 extractResults :: Avails
2284 -> TcM ( TcDictBinds, -- Bindings
2285 [Inst], -- Irreducible ones
2286 [Inst]) -- Needed givens, i.e. ones used in the bindings
2288 extractResults (Avails _ avails) wanteds
2289 = go avails emptyBag [] [] wanteds
2291 go :: AvailEnv -> TcDictBinds -> [Inst] -> [Inst] -> [Inst]
2292 -> TcM (TcDictBinds, [Inst], [Inst])
2293 go avails binds irreds givens []
2294 = returnM (binds, irreds, givens)
2296 go avails binds irreds givens (w:ws)
2297 = case findAvailEnv avails w of
2298 Nothing -> pprTrace "Urk: extractResults" (ppr w) $
2299 go avails binds irreds givens ws
2302 | id == w_id -> go avails binds irreds (w:givens) ws
2304 go avails (addInstToDictBind binds w (nlHsVar id)) irreds
2305 (update_id w id:givens) ws
2306 -- The sought Id can be one of the givens, via a superclass chain
2307 -- and then we definitely don't want to generate an x=x binding!
2309 Just IsIrred -> go (add_given avails w) binds (w:irreds) givens ws
2310 -- The add_given handles the case where we want (Ord a, Eq a), and we
2311 -- don't want to emit *two* Irreds for Ord a, one via the superclass chain
2312 -- This showed up in a dupliated Ord constraint in the error message for
2315 Just (Rhs rhs ws') -> go (add_given avails w) new_binds irreds givens (ws' ++ ws)
2317 new_binds = addInstToDictBind binds w rhs
2320 update_id m@(Method{}) id = m {tci_id = id}
2321 update_id w id = w {tci_name = idName id}
2323 add_given avails w = extendAvailEnv avails w (Given (instToId w))
2325 extractLocalResults :: Avails
2327 -> TcM ( TcDictBinds, -- Bindings
2328 [Inst]) -- Needed givens, i.e. ones used in the bindings
2330 extractLocalResults (Avails _ avails) wanteds
2331 = go avails emptyBag [] wanteds
2333 go :: AvailEnv -> TcDictBinds -> [Inst] -> [Inst]
2334 -> TcM (TcDictBinds, [Inst])
2335 go avails binds givens []
2336 = returnM (binds, givens)
2338 go avails binds givens (w:ws)
2339 = case findAvailEnv avails w of
2340 Nothing -> -- pprTrace "Urk: extractLocalResults" (ppr w) $
2341 go avails binds givens ws
2344 go avails binds givens ws
2347 | id == w_id -> go avails binds (w:givens) ws
2348 | otherwise -> go avails binds (w{tci_name=idName id}:givens) ws
2349 -- The sought Id can be one of the givens, via a superclass chain
2350 -- and then we definitely don't want to generate an x=x binding!
2352 Just (Rhs rhs ws') -> go (add_given avails w) new_binds givens (ws' ++ ws)
2354 new_binds = addInstToDictBind binds w rhs
2358 add_given avails w = extendAvailEnv avails w (Given (instToId w))
2362 Note [No superclasses for Stop]
2363 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2364 When we decide not to reduce an Inst -- the 'WhatToDo' --- we still
2365 add it to avails, so that any other equal Insts will be commoned up
2366 right here. However, we do *not* add superclasses. If we have
2369 but a is not bound here, then we *don't* want to derive dn from df
2370 here lest we lose sharing.
2373 addWanted :: WantSCs -> Avails -> Inst -> LHsExpr TcId -> [Inst] -> TcM Avails
2374 addWanted want_scs avails wanted rhs_expr wanteds
2375 = addAvailAndSCs want_scs avails wanted avail
2377 avail = Rhs rhs_expr wanteds
2379 addGiven :: Avails -> Inst -> TcM Avails
2380 addGiven avails given = addAvailAndSCs AddSCs avails given (Given (instToId given))
2381 -- Always add superclasses for 'givens'
2383 -- No ASSERT( not (given `elemAvails` avails) ) because in an instance
2384 -- decl for Ord t we can add both Ord t and Eq t as 'givens',
2385 -- so the assert isn't true
2387 addRefinedGiven :: Refinement -> ([Inst], Avails) -> Inst -> TcM ([Inst], Avails)
2388 addRefinedGiven reft (refined_givens, avails) given
2389 | isDict given -- We sometimes have 'given' methods, but they
2390 -- are always optional, so we can drop them
2391 , let pred = dictPred given
2392 , isRefineablePred pred -- See Note [ImplicInst rigidity]
2393 , Just (co, pred) <- refinePred reft pred
2394 = do { new_given <- newDictBndr (instLoc given) pred
2395 ; let rhs = L (instSpan given) $
2396 HsWrap (WpCo co) (HsVar (instToId given))
2397 ; avails <- addAvailAndSCs AddSCs avails new_given (Rhs rhs [given])
2398 ; return (new_given:refined_givens, avails) }
2399 -- ToDo: the superclasses of the original given all exist in Avails
2400 -- so we could really just cast them, but it's more awkward to do,
2401 -- and hopefully the optimiser will spot the duplicated work
2403 = return (refined_givens, avails)
2406 Note [ImplicInst rigidity]
2407 ~~~~~~~~~~~~~~~~~~~~~~~~~~
2409 C :: forall ab. (Eq a, Ord b) => b -> T a
2411 ...(case x of C v -> <body>)...
2413 From the case (where x::T ty) we'll get an implication constraint
2414 forall b. (Eq ty, Ord b) => <body-constraints>
2415 Now suppose <body-constraints> itself has an implication constraint
2417 forall c. <reft> => <payload>
2418 Then, we can certainly apply the refinement <reft> to the Ord b, becuase it is
2419 existential, but we probably should not apply it to the (Eq ty) because it may
2420 be wobbly. Hence the isRigidInst
2422 @Insts@ are ordered by their class/type info, rather than by their
2423 unique. This allows the context-reduction mechanism to use standard finite
2424 maps to do their stuff. It's horrible that this code is here, rather
2425 than with the Avails handling stuff in TcSimplify
2428 addIrred :: WantSCs -> Avails -> Inst -> TcM Avails
2429 addIrred want_scs avails irred = ASSERT2( not (irred `elemAvails` avails), ppr irred $$ ppr avails )
2430 addAvailAndSCs want_scs avails irred IsIrred
2432 addAvailAndSCs :: WantSCs -> Avails -> Inst -> AvailHow -> TcM Avails
2433 addAvailAndSCs want_scs avails inst avail
2434 | not (isClassDict inst) = extendAvails avails inst avail
2435 | NoSCs <- want_scs = extendAvails avails inst avail
2436 | otherwise = do { traceTc (text "addAvailAndSCs" <+> vcat [ppr inst, ppr deps])
2437 ; avails' <- extendAvails avails inst avail
2438 ; addSCs is_loop avails' inst }
2440 is_loop pred = any (`tcEqType` mkPredTy pred) dep_tys
2441 -- Note: this compares by *type*, not by Unique
2442 deps = findAllDeps (unitVarSet (instToVar inst)) avail
2443 dep_tys = map idType (varSetElems deps)
2445 findAllDeps :: IdSet -> AvailHow -> IdSet
2446 -- Find all the Insts that this one depends on
2447 -- See Note [SUPERCLASS-LOOP 2]
2448 -- Watch out, though. Since the avails may contain loops
2449 -- (see Note [RECURSIVE DICTIONARIES]), so we need to track the ones we've seen so far
2450 findAllDeps so_far (Rhs _ kids) = foldl find_all so_far kids
2451 findAllDeps so_far other = so_far
2453 find_all :: IdSet -> Inst -> IdSet
2455 | isEqInst kid = so_far
2456 | kid_id `elemVarSet` so_far = so_far
2457 | Just avail <- findAvail avails kid = findAllDeps so_far' avail
2458 | otherwise = so_far'
2460 so_far' = extendVarSet so_far kid_id -- Add the new kid to so_far
2461 kid_id = instToId kid
2463 addSCs :: (TcPredType -> Bool) -> Avails -> Inst -> TcM Avails
2464 -- Add all the superclasses of the Inst to Avails
2465 -- The first param says "dont do this because the original thing
2466 -- depends on this one, so you'd build a loop"
2467 -- Invariant: the Inst is already in Avails.
2469 addSCs is_loop avails dict
2470 = ASSERT( isDict dict )
2471 do { sc_dicts <- newDictBndrs (instLoc dict) sc_theta'
2472 ; foldlM add_sc avails (zipEqual "add_scs" sc_dicts sc_sels) }
2474 (clas, tys) = getDictClassTys dict
2475 (tyvars, sc_theta, sc_sels, _) = classBigSig clas
2476 sc_theta' = filter (not . isEqPred) $
2477 substTheta (zipTopTvSubst tyvars tys) sc_theta
2479 add_sc avails (sc_dict, sc_sel)
2480 | is_loop (dictPred sc_dict) = return avails -- See Note [SUPERCLASS-LOOP 2]
2481 | is_given sc_dict = return avails
2482 | otherwise = do { avails' <- extendAvails avails sc_dict (Rhs sc_sel_rhs [dict])
2483 ; addSCs is_loop avails' sc_dict }
2485 sc_sel_rhs = L (instSpan dict) (HsWrap co_fn (HsVar sc_sel))
2486 co_fn = WpApp (instToVar dict) <.> mkWpTyApps tys
2488 is_given :: Inst -> Bool
2489 is_given sc_dict = case findAvail avails sc_dict of
2490 Just (Given _) -> True -- Given is cheaper than superclass selection
2493 -- From the a set of insts obtain all equalities that (transitively) occur in
2494 -- superclass contexts of class constraints (aka the ancestor equalities).
2496 ancestorEqualities :: [Inst] -> TcM [Inst]
2498 = mapM mkWantedEqInst -- turn only equality predicates..
2499 . filter isEqPred -- ..into wanted equality insts
2501 . addAEsToBag emptyBag -- collect the superclass constraints..
2502 . map dictPred -- ..of all predicates in a bag
2503 . filter isClassDict
2505 addAEsToBag :: Bag PredType -> [PredType] -> Bag PredType
2506 addAEsToBag bag [] = bag
2507 addAEsToBag bag (pred:preds)
2508 | pred `elemBag` bag = addAEsToBag bag preds
2509 | isEqPred pred = addAEsToBag bagWithPred preds
2510 | isClassPred pred = addAEsToBag bagWithPred predsWithSCs
2511 | otherwise = addAEsToBag bag preds
2513 bagWithPred = bag `snocBag` pred
2514 predsWithSCs = preds ++ substTheta (zipTopTvSubst tyvars tys) sc_theta
2516 (tyvars, sc_theta, _, _) = classBigSig clas
2517 (clas, tys) = getClassPredTys pred
2521 %************************************************************************
2523 \section{tcSimplifyTop: defaulting}
2525 %************************************************************************
2528 @tcSimplifyTop@ is called once per module to simplify all the constant
2529 and ambiguous Insts.
2531 We need to be careful of one case. Suppose we have
2533 instance Num a => Num (Foo a b) where ...
2535 and @tcSimplifyTop@ is given a constraint (Num (Foo x y)). Then it'll simplify
2536 to (Num x), and default x to Int. But what about y??
2538 It's OK: the final zonking stage should zap y to (), which is fine.
2542 tcSimplifyTop, tcSimplifyInteractive :: [Inst] -> TcM TcDictBinds
2543 tcSimplifyTop wanteds
2544 = tc_simplify_top doc False wanteds
2546 doc = text "tcSimplifyTop"
2548 tcSimplifyInteractive wanteds
2549 = tc_simplify_top doc True wanteds
2551 doc = text "tcSimplifyInteractive"
2553 -- The TcLclEnv should be valid here, solely to improve
2554 -- error message generation for the monomorphism restriction
2555 tc_simplify_top doc interactive wanteds
2556 = do { dflags <- getDOpts
2557 ; wanteds <- zonkInsts wanteds
2558 ; mapM_ zonkTopTyVar (varSetElems (tyVarsOfInsts wanteds))
2560 ; traceTc (text "tc_simplify_top 0: " <+> ppr wanteds)
2561 ; (irreds1, binds1) <- tryHardCheckLoop doc1 wanteds
2562 ; traceTc (text "tc_simplify_top 1: " <+> ppr irreds1)
2563 ; (irreds2, binds2) <- approximateImplications doc2 (\d -> True) irreds1
2564 ; traceTc (text "tc_simplify_top 2: " <+> ppr irreds2)
2566 -- Use the defaulting rules to do extra unification
2567 -- NB: irreds2 are already zonked
2568 ; (irreds3, binds3) <- disambiguate doc3 interactive dflags irreds2
2570 -- Deal with implicit parameters
2571 ; let (bad_ips, non_ips) = partition isIPDict irreds3
2572 (ambigs, others) = partition isTyVarDict non_ips
2574 ; topIPErrs bad_ips -- Can arise from f :: Int -> Int
2576 ; addNoInstanceErrs others
2577 ; addTopAmbigErrs ambigs
2579 ; return (binds1 `unionBags` binds2 `unionBags` binds3) }
2581 doc1 = doc <+> ptext SLIT("(first round)")
2582 doc2 = doc <+> ptext SLIT("(approximate)")
2583 doc3 = doc <+> ptext SLIT("(disambiguate)")
2586 If a dictionary constrains a type variable which is
2587 * not mentioned in the environment
2588 * and not mentioned in the type of the expression
2589 then it is ambiguous. No further information will arise to instantiate
2590 the type variable; nor will it be generalised and turned into an extra
2591 parameter to a function.
2593 It is an error for this to occur, except that Haskell provided for
2594 certain rules to be applied in the special case of numeric types.
2596 * at least one of its classes is a numeric class, and
2597 * all of its classes are numeric or standard
2598 then the type variable can be defaulted to the first type in the
2599 default-type list which is an instance of all the offending classes.
2601 So here is the function which does the work. It takes the ambiguous
2602 dictionaries and either resolves them (producing bindings) or
2603 complains. It works by splitting the dictionary list by type
2604 variable, and using @disambigOne@ to do the real business.
2606 @disambigOne@ assumes that its arguments dictionaries constrain all
2607 the same type variable.
2609 ADR Comment 20/6/94: I've changed the @CReturnable@ case to default to
2610 @()@ instead of @Int@. I reckon this is the Right Thing to do since
2611 the most common use of defaulting is code like:
2613 _ccall_ foo `seqPrimIO` bar
2615 Since we're not using the result of @foo@, the result if (presumably)
2619 disambiguate :: SDoc -> Bool -> DynFlags -> [Inst] -> TcM ([Inst], TcDictBinds)
2620 -- Just does unification to fix the default types
2621 -- The Insts are assumed to be pre-zonked
2622 disambiguate doc interactive dflags insts
2624 = return (insts, emptyBag)
2626 | null defaultable_groups
2627 = do { traceTc (text "disambigutate, no defaultable groups" <+> vcat [ppr unaries, ppr insts, ppr bad_tvs, ppr defaultable_groups])
2628 ; return (insts, emptyBag) }
2631 = do { -- Figure out what default types to use
2632 default_tys <- getDefaultTys extended_defaulting ovl_strings
2634 ; traceTc (text "disambiguate1" <+> vcat [ppr insts, ppr unaries, ppr bad_tvs, ppr defaultable_groups])
2635 ; mapM_ (disambigGroup default_tys) defaultable_groups
2637 -- disambigGroup does unification, hence try again
2638 ; tryHardCheckLoop doc insts }
2641 extended_defaulting = interactive || dopt Opt_ExtendedDefaultRules dflags
2642 ovl_strings = dopt Opt_OverloadedStrings dflags
2644 unaries :: [(Inst, Class, TcTyVar)] -- (C tv) constraints
2645 bad_tvs :: TcTyVarSet -- Tyvars mentioned by *other* constraints
2646 (unaries, bad_tvs_s) = partitionWith find_unary insts
2647 bad_tvs = unionVarSets bad_tvs_s
2649 -- Finds unary type-class constraints
2650 find_unary d@(Dict {tci_pred = ClassP cls [ty]})
2651 | Just tv <- tcGetTyVar_maybe ty = Left (d,cls,tv)
2652 find_unary inst = Right (tyVarsOfInst inst)
2654 -- Group by type variable
2655 defaultable_groups :: [[(Inst,Class,TcTyVar)]]
2656 defaultable_groups = filter defaultable_group (equivClasses cmp_tv unaries)
2657 cmp_tv (_,_,tv1) (_,_,tv2) = tv1 `compare` tv2
2659 defaultable_group :: [(Inst,Class,TcTyVar)] -> Bool
2660 defaultable_group ds@((_,_,tv):_)
2661 = isTyConableTyVar tv -- Note [Avoiding spurious errors]
2662 && not (tv `elemVarSet` bad_tvs)
2663 && defaultable_classes [c | (_,c,_) <- ds]
2664 defaultable_group [] = panic "defaultable_group"
2666 defaultable_classes clss
2667 | extended_defaulting = any isInteractiveClass clss
2668 | otherwise = all is_std_class clss && (any is_num_class clss)
2670 -- In interactive mode, or with -fextended-default-rules,
2671 -- we default Show a to Show () to avoid graututious errors on "show []"
2672 isInteractiveClass cls
2673 = is_num_class cls || (classKey cls `elem` [showClassKey, eqClassKey, ordClassKey])
2675 is_num_class cls = isNumericClass cls || (ovl_strings && (cls `hasKey` isStringClassKey))
2676 -- is_num_class adds IsString to the standard numeric classes,
2677 -- when -foverloaded-strings is enabled
2679 is_std_class cls = isStandardClass cls || (ovl_strings && (cls `hasKey` isStringClassKey))
2680 -- Similarly is_std_class
2682 -----------------------
2683 disambigGroup :: [Type] -- The default types
2684 -> [(Inst,Class,TcTyVar)] -- All standard classes of form (C a)
2685 -> TcM () -- Just does unification, to fix the default types
2687 disambigGroup default_tys dicts
2688 = try_default default_tys
2690 (_,_,tyvar) = ASSERT(not (null dicts)) head dicts -- Should be non-empty
2691 classes = [c | (_,c,_) <- dicts]
2693 try_default [] = return ()
2694 try_default (default_ty : default_tys)
2695 = tryTcLIE_ (try_default default_tys) $
2696 do { tcSimplifyDefault [mkClassPred clas [default_ty] | clas <- classes]
2697 -- This may fail; then the tryTcLIE_ kicks in
2698 -- Failure here is caused by there being no type in the
2699 -- default list which can satisfy all the ambiguous classes.
2700 -- For example, if Real a is reqd, but the only type in the
2701 -- default list is Int.
2703 -- After this we can't fail
2704 ; warnDefault dicts default_ty
2705 ; unifyType default_ty (mkTyVarTy tyvar)
2706 ; return () -- TOMDO: do something with the coercion
2710 -----------------------
2711 getDefaultTys :: Bool -> Bool -> TcM [Type]
2712 getDefaultTys extended_deflts ovl_strings
2713 = do { mb_defaults <- getDeclaredDefaultTys
2714 ; case mb_defaults of {
2715 Just tys -> return tys ; -- User-supplied defaults
2718 -- No use-supplied default
2719 -- Use [Integer, Double], plus modifications
2720 { integer_ty <- tcMetaTy integerTyConName
2721 ; checkWiredInTyCon doubleTyCon
2722 ; string_ty <- tcMetaTy stringTyConName
2723 ; return (opt_deflt extended_deflts unitTy
2724 -- Note [Default unitTy]
2726 [integer_ty,doubleTy]
2728 opt_deflt ovl_strings string_ty) } } }
2730 opt_deflt True ty = [ty]
2731 opt_deflt False ty = []
2734 Note [Default unitTy]
2735 ~~~~~~~~~~~~~~~~~~~~~
2736 In interative mode (or with -fextended-default-rules) we add () as the first type we
2737 try when defaulting. This has very little real impact, except in the following case.
2739 Text.Printf.printf "hello"
2740 This has type (forall a. IO a); it prints "hello", and returns 'undefined'. We don't
2741 want the GHCi repl loop to try to print that 'undefined'. The neatest thing is to
2742 default the 'a' to (), rather than to Integer (which is what would otherwise happen;
2743 and then GHCi doesn't attempt to print the (). So in interactive mode, we add
2744 () to the list of defaulting types. See Trac #1200.
2746 Note [Avoiding spurious errors]
2747 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2748 When doing the unification for defaulting, we check for skolem
2749 type variables, and simply don't default them. For example:
2750 f = (*) -- Monomorphic
2751 g :: Num a => a -> a
2753 Here, we get a complaint when checking the type signature for g,
2754 that g isn't polymorphic enough; but then we get another one when
2755 dealing with the (Num a) context arising from f's definition;
2756 we try to unify a with Int (to default it), but find that it's
2757 already been unified with the rigid variable from g's type sig
2760 %************************************************************************
2762 \subsection[simple]{@Simple@ versions}
2764 %************************************************************************
2766 Much simpler versions when there are no bindings to make!
2768 @tcSimplifyThetas@ simplifies class-type constraints formed by
2769 @deriving@ declarations and when specialising instances. We are
2770 only interested in the simplified bunch of class/type constraints.
2772 It simplifies to constraints of the form (C a b c) where
2773 a,b,c are type variables. This is required for the context of
2774 instance declarations.
2777 tcSimplifyDeriv :: InstOrigin
2779 -> ThetaType -- Wanted
2780 -> TcM ThetaType -- Needed
2781 -- Given instance (wanted) => C inst_ty
2782 -- Simplify 'wanted' as much as possible
2784 tcSimplifyDeriv orig tyvars theta
2785 = do { (tvs, _, tenv) <- tcInstTyVars tyvars
2786 -- The main loop may do unification, and that may crash if
2787 -- it doesn't see a TcTyVar, so we have to instantiate. Sigh
2788 -- ToDo: what if two of them do get unified?
2789 ; wanteds <- newDictBndrsO orig (substTheta tenv theta)
2790 ; (irreds, _) <- tryHardCheckLoop doc wanteds
2792 ; let (tv_dicts, others) = partition ok irreds
2793 ; addNoInstanceErrs others
2794 -- See Note [Exotic derived instance contexts] in TcMType
2796 ; let rev_env = zipTopTvSubst tvs (mkTyVarTys tyvars)
2797 simpl_theta = substTheta rev_env (map dictPred tv_dicts)
2798 -- This reverse-mapping is a pain, but the result
2799 -- should mention the original TyVars not TcTyVars
2801 ; return simpl_theta }
2803 doc = ptext SLIT("deriving classes for a data type")
2805 ok dict | isDict dict = validDerivPred (dictPred dict)
2810 @tcSimplifyDefault@ just checks class-type constraints, essentially;
2811 used with \tr{default} declarations. We are only interested in
2812 whether it worked or not.
2815 tcSimplifyDefault :: ThetaType -- Wanted; has no type variables in it
2818 tcSimplifyDefault theta
2819 = newDictBndrsO DefaultOrigin theta `thenM` \ wanteds ->
2820 tryHardCheckLoop doc wanteds `thenM` \ (irreds, _) ->
2821 addNoInstanceErrs irreds `thenM_`
2827 doc = ptext SLIT("default declaration")
2831 %************************************************************************
2833 \section{Errors and contexts}
2835 %************************************************************************
2837 ToDo: for these error messages, should we note the location as coming
2838 from the insts, or just whatever seems to be around in the monad just
2842 groupErrs :: ([Inst] -> TcM ()) -- Deal with one group
2843 -> [Inst] -- The offending Insts
2845 -- Group together insts with the same origin
2846 -- We want to report them together in error messages
2848 groupErrs report_err []
2850 groupErrs report_err (inst:insts)
2851 = do_one (inst:friends) `thenM_`
2852 groupErrs report_err others
2855 -- (It may seem a bit crude to compare the error messages,
2856 -- but it makes sure that we combine just what the user sees,
2857 -- and it avoids need equality on InstLocs.)
2858 (friends, others) = partition is_friend insts
2859 loc_msg = showSDoc (pprInstLoc (instLoc inst))
2860 is_friend friend = showSDoc (pprInstLoc (instLoc friend)) == loc_msg
2861 do_one insts = addInstCtxt (instLoc (head insts)) (report_err insts)
2862 -- Add location and context information derived from the Insts
2864 -- Add the "arising from..." part to a message about bunch of dicts
2865 addInstLoc :: [Inst] -> Message -> Message
2866 addInstLoc insts msg = msg $$ nest 2 (pprInstArising (head insts))
2868 addTopIPErrs :: [Name] -> [Inst] -> TcM ()
2869 addTopIPErrs bndrs []
2871 addTopIPErrs bndrs ips
2872 = do { dflags <- getDOpts
2873 ; addErrTcM (tidy_env, mk_msg dflags tidy_ips) }
2875 (tidy_env, tidy_ips) = tidyInsts ips
2877 = vcat [sep [ptext SLIT("Implicit parameters escape from"),
2878 nest 2 (ptext SLIT("the monomorphic top-level binding")
2879 <> plural bndrs <+> ptext SLIT("of")
2880 <+> pprBinders bndrs <> colon)],
2881 nest 2 (vcat (map ppr_ip ips)),
2882 monomorphism_fix dflags]
2883 ppr_ip ip = pprPred (dictPred ip) <+> pprInstArising ip
2885 topIPErrs :: [Inst] -> TcM ()
2887 = groupErrs report tidy_dicts
2889 (tidy_env, tidy_dicts) = tidyInsts dicts
2890 report dicts = addErrTcM (tidy_env, mk_msg dicts)
2891 mk_msg dicts = addInstLoc dicts (ptext SLIT("Unbound implicit parameter") <>
2892 plural tidy_dicts <+> pprDictsTheta tidy_dicts)
2894 addNoInstanceErrs :: [Inst] -- Wanted (can include implications)
2896 addNoInstanceErrs insts
2897 = do { let (tidy_env, tidy_insts) = tidyInsts insts
2898 ; reportNoInstances tidy_env Nothing tidy_insts }
2902 -> Maybe (InstLoc, [Inst]) -- Context
2903 -- Nothing => top level
2904 -- Just (d,g) => d describes the construct
2906 -> [Inst] -- What is wanted (can include implications)
2909 reportNoInstances tidy_env mb_what insts
2910 = groupErrs (report_no_instances tidy_env mb_what) insts
2912 report_no_instances tidy_env mb_what insts
2913 = do { inst_envs <- tcGetInstEnvs
2914 ; let (implics, insts1) = partition isImplicInst insts
2915 (insts2, overlaps) = partitionWith (check_overlap inst_envs) insts1
2916 (eqInsts, insts3) = partition isEqInst insts2
2917 ; traceTc (text "reportNoInstances" <+> vcat
2918 [ppr implics, ppr insts1, ppr insts2])
2919 ; mapM_ complain_implic implics
2920 ; mapM_ (\doc -> addErrTcM (tidy_env, doc)) overlaps
2921 ; groupErrs complain_no_inst insts3
2922 ; mapM_ eqInstMisMatch eqInsts
2925 complain_no_inst insts = addErrTcM (tidy_env, mk_no_inst_err insts)
2927 complain_implic inst -- Recurse!
2928 = reportNoInstances tidy_env
2929 (Just (tci_loc inst, tci_given inst))
2932 check_overlap :: (InstEnv,InstEnv) -> Inst -> Either Inst SDoc
2933 -- Right msg => overlap message
2934 -- Left inst => no instance
2935 check_overlap inst_envs wanted
2936 | not (isClassDict wanted) = Left wanted
2938 = case lookupInstEnv inst_envs clas tys of
2939 -- The case of exactly one match and no unifiers means a
2940 -- successful lookup. That can't happen here, because dicts
2941 -- only end up here if they didn't match in Inst.lookupInst
2943 ([m],[]) -> pprPanic "reportNoInstance" (ppr wanted)
2945 ([], _) -> Left wanted -- No match
2946 res -> Right (mk_overlap_msg wanted res)
2948 (clas,tys) = getDictClassTys wanted
2950 mk_overlap_msg dict (matches, unifiers)
2951 = ASSERT( not (null matches) )
2952 vcat [ addInstLoc [dict] ((ptext SLIT("Overlapping instances for")
2953 <+> pprPred (dictPred dict))),
2954 sep [ptext SLIT("Matching instances") <> colon,
2955 nest 2 (vcat [pprInstances ispecs, pprInstances unifiers])],
2956 if not (isSingleton matches)
2957 then -- Two or more matches
2959 else -- One match, plus some unifiers
2960 ASSERT( not (null unifiers) )
2961 parens (vcat [ptext SLIT("The choice depends on the instantiation of") <+>
2962 quotes (pprWithCommas ppr (varSetElems (tyVarsOfInst dict))),
2963 ptext SLIT("To pick the first instance above, use -fallow-incoherent-instances"),
2964 ptext SLIT("when compiling the other instance declarations")])]
2966 ispecs = [ispec | (ispec, _) <- matches]
2968 mk_no_inst_err insts
2969 | null insts = empty
2971 | Just (loc, givens) <- mb_what, -- Nested (type signatures, instance decls)
2972 not (isEmptyVarSet (tyVarsOfInsts insts))
2973 = vcat [ addInstLoc insts $
2974 sep [ ptext SLIT("Could not deduce") <+> pprDictsTheta insts
2975 , nest 2 $ ptext SLIT("from the context") <+> pprDictsTheta givens]
2976 , show_fixes (fix1 loc : fixes2) ]
2978 | otherwise -- Top level
2979 = vcat [ addInstLoc insts $
2980 ptext SLIT("No instance") <> plural insts
2981 <+> ptext SLIT("for") <+> pprDictsTheta insts
2982 , show_fixes fixes2 ]
2985 fix1 loc = sep [ ptext SLIT("add") <+> pprDictsTheta insts
2986 <+> ptext SLIT("to the context of"),
2987 nest 2 (ppr (instLocOrigin loc)) ]
2988 -- I'm not sure it helps to add the location
2989 -- nest 2 (ptext SLIT("at") <+> ppr (instLocSpan loc)) ]
2991 fixes2 | null instance_dicts = []
2992 | otherwise = [sep [ptext SLIT("add an instance declaration for"),
2993 pprDictsTheta instance_dicts]]
2994 instance_dicts = [d | d <- insts, isClassDict d, not (isTyVarDict d)]
2995 -- Insts for which it is worth suggesting an adding an instance declaration
2996 -- Exclude implicit parameters, and tyvar dicts
2998 show_fixes :: [SDoc] -> SDoc
2999 show_fixes [] = empty
3000 show_fixes (f:fs) = sep [ptext SLIT("Possible fix:"),
3001 nest 2 (vcat (f : map (ptext SLIT("or") <+>) fs))]
3003 addTopAmbigErrs dicts
3004 -- Divide into groups that share a common set of ambiguous tyvars
3005 = ifErrsM (return ()) $ -- Only report ambiguity if no other errors happened
3006 -- See Note [Avoiding spurious errors]
3007 mapM_ report (equivClasses cmp [(d, tvs_of d) | d <- tidy_dicts])
3009 (tidy_env, tidy_dicts) = tidyInsts dicts
3011 tvs_of :: Inst -> [TcTyVar]
3012 tvs_of d = varSetElems (tyVarsOfInst d)
3013 cmp (_,tvs1) (_,tvs2) = tvs1 `compare` tvs2
3015 report :: [(Inst,[TcTyVar])] -> TcM ()
3016 report pairs@((inst,tvs) : _) -- The pairs share a common set of ambiguous tyvars
3017 = mkMonomorphismMsg tidy_env tvs `thenM` \ (tidy_env, mono_msg) ->
3018 setSrcSpan (instSpan inst) $
3019 -- the location of the first one will do for the err message
3020 addErrTcM (tidy_env, msg $$ mono_msg)
3022 dicts = map fst pairs
3023 msg = sep [text "Ambiguous type variable" <> plural tvs <+>
3024 pprQuotedList tvs <+> in_msg,
3025 nest 2 (pprDictsInFull dicts)]
3026 in_msg = text "in the constraint" <> plural dicts <> colon
3027 report [] = panic "addTopAmbigErrs"
3030 mkMonomorphismMsg :: TidyEnv -> [TcTyVar] -> TcM (TidyEnv, Message)
3031 -- There's an error with these Insts; if they have free type variables
3032 -- it's probably caused by the monomorphism restriction.
3033 -- Try to identify the offending variable
3034 -- ASSUMPTION: the Insts are fully zonked
3035 mkMonomorphismMsg tidy_env inst_tvs
3036 = do { dflags <- getDOpts
3037 ; (tidy_env, docs) <- findGlobals (mkVarSet inst_tvs) tidy_env
3038 ; return (tidy_env, mk_msg dflags docs) }
3040 mk_msg _ _ | any isRuntimeUnk inst_tvs
3041 = vcat [ptext SLIT("Cannot resolve unknown runtime types:") <+>
3042 (pprWithCommas ppr inst_tvs),
3043 ptext SLIT("Use :print or :force to determine these types")]
3044 mk_msg _ [] = ptext SLIT("Probable fix: add a type signature that fixes these type variable(s)")
3045 -- This happens in things like
3046 -- f x = show (read "foo")
3047 -- where monomorphism doesn't play any role
3049 = vcat [ptext SLIT("Possible cause: the monomorphism restriction applied to the following:"),
3051 monomorphism_fix dflags]
3053 isRuntimeUnk :: TcTyVar -> Bool
3054 isRuntimeUnk x | SkolemTv RuntimeUnkSkol <- tcTyVarDetails x = True
3057 monomorphism_fix :: DynFlags -> SDoc
3058 monomorphism_fix dflags
3059 = ptext SLIT("Probable fix:") <+> vcat
3060 [ptext SLIT("give these definition(s) an explicit type signature"),
3061 if dopt Opt_MonomorphismRestriction dflags
3062 then ptext SLIT("or use -fno-monomorphism-restriction")
3063 else empty] -- Only suggest adding "-fno-monomorphism-restriction"
3064 -- if it is not already set!
3066 warnDefault ups default_ty
3067 = doptM Opt_WarnTypeDefaults `thenM` \ warn_flag ->
3068 addInstCtxt (instLoc (head (dicts))) (warnTc warn_flag warn_msg)
3070 dicts = [d | (d,_,_) <- ups]
3073 (_, tidy_dicts) = tidyInsts dicts
3074 warn_msg = vcat [ptext SLIT("Defaulting the following constraint(s) to type") <+>
3075 quotes (ppr default_ty),
3076 pprDictsInFull tidy_dicts]
3078 reduceDepthErr n stack
3079 = vcat [ptext SLIT("Context reduction stack overflow; size =") <+> int n,
3080 ptext SLIT("Use -fcontext-stack=N to increase stack size to N"),
3081 nest 4 (pprStack stack)]
3083 pprStack stack = vcat (map pprInstInFull stack)