Kill non-deterministic foldUFM in TrieMap and TcAppMap
[ghc.git] / compiler / typecheck / TcSMonad.hs
1 {-# LANGUAGE CPP, TypeFamilies #-}
2
3 -- Type definitions for the constraint solver
4 module TcSMonad (
5
6 -- The work list
7 WorkList(..), isEmptyWorkList, emptyWorkList,
8 extendWorkListNonEq, extendWorkListCt, extendWorkListDerived,
9 extendWorkListCts, extendWorkListEq, extendWorkListFunEq,
10 appendWorkList,
11 selectNextWorkItem,
12 workListSize, workListWantedCount,
13 getWorkList, updWorkListTcS,
14
15 -- The TcS monad
16 TcS, runTcS, runTcSDeriveds, runTcSWithEvBinds,
17 failTcS, warnTcS, addErrTcS,
18 runTcSEqualities,
19 nestTcS, nestImplicTcS,
20
21 runTcPluginTcS, addUsedDataCons, deferTcSForAllEq,
22
23 -- Tracing etc
24 panicTcS, traceTcS,
25 traceFireTcS, bumpStepCountTcS, csTraceTcS,
26 wrapErrTcS, wrapWarnTcS,
27
28 -- Evidence creation and transformation
29 MaybeNew(..), freshGoals, isFresh, getEvTerm,
30
31 newTcEvBinds,
32 newWantedEq,
33 newWanted, newWantedEvVar, newWantedEvVarNC, newDerivedNC,
34 newBoundEvVarId,
35 unifyTyVar, unflattenFmv, reportUnifications,
36 setEvBind, setWantedEq, setEqIfWanted,
37 setWantedEvTerm, setWantedEvBind, setEvBindIfWanted,
38 newEvVar, newGivenEvVar, newGivenEvVars,
39 emitNewDerived, emitNewDeriveds, emitNewDerivedEq,
40 checkReductionDepth,
41
42 getInstEnvs, getFamInstEnvs, -- Getting the environments
43 getTopEnv, getGblEnv, getLclEnv,
44 getTcEvBinds, getTcEvBindsFromVar, getTcLevel,
45 getTcEvBindsMap,
46 tcLookupClass,
47
48 -- Inerts
49 InertSet(..), InertCans(..),
50 updInertTcS, updInertCans, updInertDicts, updInertIrreds,
51 getNoGivenEqs, setInertCans,
52 getInertEqs, getInertCans, getInertModel, getInertGivens,
53 emptyInert, getTcSInerts, setTcSInerts, takeGivenInsolubles,
54 matchableGivens, prohibitedSuperClassSolve,
55 getUnsolvedInerts,
56 removeInertCts, getPendingScDicts,
57 addInertCan, addInertEq, insertFunEq,
58 emitInsoluble, emitWorkNC,
59
60 -- The Model
61 InertModel, kickOutAfterUnification,
62
63 -- Inert Safe Haskell safe-overlap failures
64 addInertSafehask, insertSafeOverlapFailureTcS, updInertSafehask,
65 getSafeOverlapFailures,
66
67 -- Inert CDictCans
68 lookupInertDict, findDictsByClass, addDict, addDictsByClass,
69 delDict, partitionDicts, foldDicts, filterDicts,
70
71 -- Inert CTyEqCans
72 EqualCtList, findTyEqs, foldTyEqs, isInInertEqs,
73
74 -- Inert solved dictionaries
75 addSolvedDict, lookupSolvedDict,
76
77 -- Irreds
78 foldIrreds,
79
80 -- The flattening cache
81 lookupFlatCache, extendFlatCache, newFlattenSkolem, -- Flatten skolems
82
83 -- Inert CFunEqCans
84 updInertFunEqs, findFunEq, sizeFunEqMap, filterFunEqs,
85 findFunEqsByTyCon, partitionFunEqs, foldFunEqs,
86
87 instDFunType, -- Instantiation
88
89 -- MetaTyVars
90 newFlexiTcSTy, instFlexiTcS,
91 cloneMetaTyVar, demoteUnfilledFmv,
92
93 TcLevel, isTouchableMetaTyVarTcS,
94 isFilledMetaTyVar_maybe, isFilledMetaTyVar,
95 zonkTyCoVarsAndFV, zonkTcType, zonkTcTypes, zonkTcTyVar, zonkCo,
96 zonkTyCoVarsAndFVList,
97 zonkSimples, zonkWC,
98
99 -- References
100 newTcRef, readTcRef, updTcRef,
101
102 -- Misc
103 getDefaultInfo, getDynFlags, getGlobalRdrEnvTcS,
104 matchFam, matchFamTcM,
105 checkWellStagedDFun,
106 pprEq -- Smaller utils, re-exported from TcM
107 -- TODO (DV): these are only really used in the
108 -- instance matcher in TcSimplify. I am wondering
109 -- if the whole instance matcher simply belongs
110 -- here
111 ) where
112
113 #include "HsVersions.h"
114
115 import HscTypes
116
117 import qualified Inst as TcM
118 import InstEnv
119 import FamInst
120 import FamInstEnv
121
122 import qualified TcRnMonad as TcM
123 import qualified TcMType as TcM
124 import qualified TcEnv as TcM
125 ( checkWellStaged, topIdLvl, tcGetDefaultTys, tcLookupClass )
126 import Kind
127 import TcType
128 import DynFlags
129 import Type
130 import Coercion
131 import Unify
132
133 import TcEvidence
134 import Class
135 import TyCon
136 import TcErrors ( solverDepthErrorTcS )
137
138 import Name
139 import RdrName ( GlobalRdrEnv)
140 import qualified RnEnv as TcM
141 import Var
142 import VarEnv
143 import VarSet
144 import Outputable
145 import Bag
146 import UniqSupply
147 import Util
148 import TcRnTypes
149
150 import Unique
151 import UniqFM
152 import UniqDFM
153 import Maybes
154
155 import StaticFlags( opt_PprStyle_Debug )
156 import TrieMap
157 import Control.Monad
158 #if __GLASGOW_HASKELL__ > 710
159 import qualified Control.Monad.Fail as MonadFail
160 #endif
161 import MonadUtils
162 import Data.IORef
163 import Data.List ( foldl', partition )
164
165 #ifdef DEBUG
166 import Digraph
167 #endif
168
169 {-
170 ************************************************************************
171 * *
172 * Worklists *
173 * Canonical and non-canonical constraints that the simplifier has to *
174 * work on. Including their simplification depths. *
175 * *
176 * *
177 ************************************************************************
178
179 Note [WorkList priorities]
180 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
181 A WorkList contains canonical and non-canonical items (of all flavors).
182 Notice that each Ct now has a simplification depth. We may
183 consider using this depth for prioritization as well in the future.
184
185 As a simple form of priority queue, our worklist separates out
186 equalities (wl_eqs) from the rest of the canonical constraints,
187 so that it's easier to deal with them first, but the separation
188 is not strictly necessary. Notice that non-canonical constraints
189 are also parts of the worklist.
190
191 Note [Process derived items last]
192 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
193 We can often solve all goals without processing *any* derived constraints.
194 The derived constraints are just there to help us if we get stuck. So
195 we keep them in a separate list.
196
197 -}
198
199 -- See Note [WorkList priorities]
200 data WorkList
201 = WL { wl_eqs :: [Ct]
202 , wl_funeqs :: [Ct] -- LIFO stack of goals
203 , wl_rest :: [Ct]
204 , wl_deriv :: [CtEvidence] -- Implicitly non-canonical
205 -- See Note [Process derived items last]
206 , wl_implics :: Bag Implication -- See Note [Residual implications]
207 }
208
209 appendWorkList :: WorkList -> WorkList -> WorkList
210 appendWorkList
211 (WL { wl_eqs = eqs1, wl_funeqs = funeqs1, wl_rest = rest1
212 , wl_deriv = ders1, wl_implics = implics1 })
213 (WL { wl_eqs = eqs2, wl_funeqs = funeqs2, wl_rest = rest2
214 , wl_deriv = ders2, wl_implics = implics2 })
215 = WL { wl_eqs = eqs1 ++ eqs2
216 , wl_funeqs = funeqs1 ++ funeqs2
217 , wl_rest = rest1 ++ rest2
218 , wl_deriv = ders1 ++ ders2
219 , wl_implics = implics1 `unionBags` implics2 }
220
221 workListSize :: WorkList -> Int
222 workListSize (WL { wl_eqs = eqs, wl_funeqs = funeqs, wl_deriv = ders, wl_rest = rest })
223 = length eqs + length funeqs + length rest + length ders
224
225 workListWantedCount :: WorkList -> Int
226 workListWantedCount (WL { wl_eqs = eqs, wl_rest = rest })
227 = count isWantedCt eqs + count isWantedCt rest
228
229 extendWorkListEq :: Ct -> WorkList -> WorkList
230 extendWorkListEq ct wl = wl { wl_eqs = ct : wl_eqs wl }
231
232 extendWorkListEqs :: [Ct] -> WorkList -> WorkList
233 extendWorkListEqs cts wl = wl { wl_eqs = cts ++ wl_eqs wl }
234
235 extendWorkListFunEq :: Ct -> WorkList -> WorkList
236 extendWorkListFunEq ct wl = wl { wl_funeqs = ct : wl_funeqs wl }
237
238 extendWorkListNonEq :: Ct -> WorkList -> WorkList
239 -- Extension by non equality
240 extendWorkListNonEq ct wl = wl { wl_rest = ct : wl_rest wl }
241
242 extendWorkListDerived :: CtLoc -> CtEvidence -> WorkList -> WorkList
243 extendWorkListDerived loc ev wl
244 | isDroppableDerivedLoc loc = wl { wl_deriv = ev : wl_deriv wl }
245 | otherwise = extendWorkListEq (mkNonCanonical ev) wl
246
247 extendWorkListDeriveds :: CtLoc -> [CtEvidence] -> WorkList -> WorkList
248 extendWorkListDeriveds loc evs wl
249 | isDroppableDerivedLoc loc = wl { wl_deriv = evs ++ wl_deriv wl }
250 | otherwise = extendWorkListEqs (map mkNonCanonical evs) wl
251
252 extendWorkListImplic :: Implication -> WorkList -> WorkList
253 extendWorkListImplic implic wl = wl { wl_implics = implic `consBag` wl_implics wl }
254
255 extendWorkListCt :: Ct -> WorkList -> WorkList
256 -- Agnostic
257 extendWorkListCt ct wl
258 = case classifyPredType (ctPred ct) of
259 EqPred NomEq ty1 _
260 | Just (tc,_) <- tcSplitTyConApp_maybe ty1
261 , isTypeFamilyTyCon tc
262 -> extendWorkListFunEq ct wl
263 EqPred {}
264 -> extendWorkListEq ct wl
265
266 _ -> extendWorkListNonEq ct wl
267
268 extendWorkListCts :: [Ct] -> WorkList -> WorkList
269 -- Agnostic
270 extendWorkListCts cts wl = foldr extendWorkListCt wl cts
271
272 isEmptyWorkList :: WorkList -> Bool
273 isEmptyWorkList (WL { wl_eqs = eqs, wl_funeqs = funeqs
274 , wl_rest = rest, wl_deriv = ders, wl_implics = implics })
275 = null eqs && null rest && null funeqs && isEmptyBag implics && null ders
276
277 emptyWorkList :: WorkList
278 emptyWorkList = WL { wl_eqs = [], wl_rest = []
279 , wl_funeqs = [], wl_deriv = [], wl_implics = emptyBag }
280
281 selectWorkItem :: WorkList -> Maybe (Ct, WorkList)
282 selectWorkItem wl@(WL { wl_eqs = eqs, wl_funeqs = feqs
283 , wl_rest = rest })
284 | ct:cts <- eqs = Just (ct, wl { wl_eqs = cts })
285 | ct:fes <- feqs = Just (ct, wl { wl_funeqs = fes })
286 | ct:cts <- rest = Just (ct, wl { wl_rest = cts })
287 | otherwise = Nothing
288
289 getWorkList :: TcS WorkList
290 getWorkList = do { wl_var <- getTcSWorkListRef
291 ; wrapTcS (TcM.readTcRef wl_var) }
292
293 selectDerivedWorkItem :: WorkList -> Maybe (Ct, WorkList)
294 selectDerivedWorkItem wl@(WL { wl_deriv = ders })
295 | ev:evs <- ders = Just (mkNonCanonical ev, wl { wl_deriv = evs })
296 | otherwise = Nothing
297
298 selectNextWorkItem :: TcS (Maybe Ct)
299 selectNextWorkItem
300 = do { wl_var <- getTcSWorkListRef
301 ; wl <- wrapTcS (TcM.readTcRef wl_var)
302
303 ; let try :: Maybe (Ct,WorkList) -> TcS (Maybe Ct) -> TcS (Maybe Ct)
304 try mb_work do_this_if_fail
305 | Just (ct, new_wl) <- mb_work
306 = do { checkReductionDepth (ctLoc ct) (ctPred ct)
307 ; wrapTcS (TcM.writeTcRef wl_var new_wl)
308 ; return (Just ct) }
309 | otherwise
310 = do_this_if_fail
311
312 ; try (selectWorkItem wl) $
313
314 do { ics <- getInertCans
315 ; solve_deriveds <- keepSolvingDeriveds
316 ; if inert_count ics == 0 && not solve_deriveds
317 then return Nothing
318 else try (selectDerivedWorkItem wl) (return Nothing) } }
319
320 -- Pretty printing
321 instance Outputable WorkList where
322 ppr (WL { wl_eqs = eqs, wl_funeqs = feqs
323 , wl_rest = rest, wl_implics = implics, wl_deriv = ders })
324 = text "WL" <+> (braces $
325 vcat [ ppUnless (null eqs) $
326 text "Eqs =" <+> vcat (map ppr eqs)
327 , ppUnless (null feqs) $
328 text "Funeqs =" <+> vcat (map ppr feqs)
329 , ppUnless (null rest) $
330 text "Non-eqs =" <+> vcat (map ppr rest)
331 , ppUnless (null ders) $
332 text "Derived =" <+> vcat (map ppr ders)
333 , ppUnless (isEmptyBag implics) $
334 if opt_PprStyle_Debug -- Typically we only want the work list for this level
335 then text "Implics =" <+> vcat (map ppr (bagToList implics))
336 else text "(Implics omitted)"
337 ])
338
339
340 {- *********************************************************************
341 * *
342 InertSet: the inert set
343 * *
344 * *
345 ********************************************************************* -}
346
347 data InertSet
348 = IS { inert_cans :: InertCans
349 -- Canonical Given, Wanted, Derived (no Solved)
350 -- Sometimes called "the inert set"
351
352 , inert_flat_cache :: ExactFunEqMap (TcCoercion, TcType, CtFlavour)
353 -- See Note [Type family equations]
354 -- If F tys :-> (co, ty, ev),
355 -- then co :: F tys ~ ty
356 --
357 -- Just a hash-cons cache for use when flattening only
358 -- These include entirely un-processed goals, so don't use
359 -- them to solve a top-level goal, else you may end up solving
360 -- (w:F ty ~ a) by setting w:=w! We just use the flat-cache
361 -- when allocating a new flatten-skolem.
362 -- Not necessarily inert wrt top-level equations (or inert_cans)
363
364 -- NB: An ExactFunEqMap -- this doesn't match via loose types!
365
366 , inert_solved_dicts :: DictMap CtEvidence
367 -- Of form ev :: C t1 .. tn
368 -- See Note [Solved dictionaries]
369 -- and Note [Do not add superclasses of solved dictionaries]
370 }
371
372 instance Outputable InertSet where
373 ppr is = vcat [ ppr $ inert_cans is
374 , text "Solved dicts" <+> vcat (map ppr (bagToList (dictsToBag (inert_solved_dicts is)))) ]
375
376 emptyInert :: InertSet
377 emptyInert
378 = IS { inert_cans = IC { inert_count = 0
379 , inert_eqs = emptyVarEnv
380 , inert_dicts = emptyDicts
381 , inert_safehask = emptyDicts
382 , inert_funeqs = emptyFunEqs
383 , inert_irreds = emptyCts
384 , inert_insols = emptyCts
385 , inert_model = emptyVarEnv }
386 , inert_flat_cache = emptyExactFunEqs
387 , inert_solved_dicts = emptyDictMap }
388
389
390 {- Note [Solved dictionaries]
391 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
392 When we apply a top-level instance declararation, we add the "solved"
393 dictionary to the inert_solved_dicts. In general, we use it to avoid
394 creating a new EvVar when we have a new goal that we have solved in
395 the past.
396
397 But in particular, we can use it to create *recursive* dicationaries.
398 The simplest, degnerate case is
399 instance C [a] => C [a] where ...
400 If we have
401 [W] d1 :: C [x]
402 then we can apply the instance to get
403 d1 = $dfCList d
404 [W] d2 :: C [x]
405 Now 'd1' goes in inert_solved_dicts, and we can solve d2 directly from d1.
406 d1 = $dfCList d
407 d2 = d1
408
409 See Note [Example of recursive dictionaries]
410 Other notes about solved dictionaries
411
412 * See also Note [Do not add superclasses of solved dictionaries]
413
414 * The inert_solved_dicts field is not rewritten by equalities, so it may
415 get out of date.
416
417 Note [Do not add superclasses of solved dictionaries]
418 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
419 Every member of inert_solved_dicts is the result of applying a dictionary
420 function, NOT of applying superclass selection to anything.
421 Consider
422
423 class Ord a => C a where
424 instance Ord [a] => C [a] where ...
425
426 Suppose we are trying to solve
427 [G] d1 : Ord a
428 [W] d2 : C [a]
429
430 Then we'll use the instance decl to give
431
432 [G] d1 : Ord a Solved: d2 : C [a] = $dfCList d3
433 [W] d3 : Ord [a]
434
435 We must not add d4 : Ord [a] to the 'solved' set (by taking the
436 superclass of d2), otherwise we'll use it to solve d3, without ever
437 using d1, which would be a catastrophe.
438
439 Solution: when extending the solved dictionaries, do not add superclasses.
440 That's why each element of the inert_solved_dicts is the result of applying
441 a dictionary function.
442
443 Note [Example of recursive dictionaries]
444 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
445 --- Example 1
446
447 data D r = ZeroD | SuccD (r (D r));
448
449 instance (Eq (r (D r))) => Eq (D r) where
450 ZeroD == ZeroD = True
451 (SuccD a) == (SuccD b) = a == b
452 _ == _ = False;
453
454 equalDC :: D [] -> D [] -> Bool;
455 equalDC = (==);
456
457 We need to prove (Eq (D [])). Here's how we go:
458
459 [W] d1 : Eq (D [])
460 By instance decl of Eq (D r):
461 [W] d2 : Eq [D []] where d1 = dfEqD d2
462 By instance decl of Eq [a]:
463 [W] d3 : Eq (D []) where d2 = dfEqList d3
464 d1 = dfEqD d2
465 Now this wanted can interact with our "solved" d1 to get:
466 d3 = d1
467
468 -- Example 2:
469 This code arises in the context of "Scrap Your Boilerplate with Class"
470
471 class Sat a
472 class Data ctx a
473 instance Sat (ctx Char) => Data ctx Char -- dfunData1
474 instance (Sat (ctx [a]), Data ctx a) => Data ctx [a] -- dfunData2
475
476 class Data Maybe a => Foo a
477
478 instance Foo t => Sat (Maybe t) -- dfunSat
479
480 instance Data Maybe a => Foo a -- dfunFoo1
481 instance Foo a => Foo [a] -- dfunFoo2
482 instance Foo [Char] -- dfunFoo3
483
484 Consider generating the superclasses of the instance declaration
485 instance Foo a => Foo [a]
486
487 So our problem is this
488 [G] d0 : Foo t
489 [W] d1 : Data Maybe [t] -- Desired superclass
490
491 We may add the given in the inert set, along with its superclasses
492 Inert:
493 [G] d0 : Foo t
494 [G] d01 : Data Maybe t -- Superclass of d0
495 WorkList
496 [W] d1 : Data Maybe [t]
497
498 Solve d1 using instance dfunData2; d1 := dfunData2 d2 d3
499 Inert:
500 [G] d0 : Foo t
501 [G] d01 : Data Maybe t -- Superclass of d0
502 Solved:
503 d1 : Data Maybe [t]
504 WorkList:
505 [W] d2 : Sat (Maybe [t])
506 [W] d3 : Data Maybe t
507
508 Now, we may simplify d2 using dfunSat; d2 := dfunSat d4
509 Inert:
510 [G] d0 : Foo t
511 [G] d01 : Data Maybe t -- Superclass of d0
512 Solved:
513 d1 : Data Maybe [t]
514 d2 : Sat (Maybe [t])
515 WorkList:
516 [W] d3 : Data Maybe t
517 [W] d4 : Foo [t]
518
519 Now, we can just solve d3 from d01; d3 := d01
520 Inert
521 [G] d0 : Foo t
522 [G] d01 : Data Maybe t -- Superclass of d0
523 Solved:
524 d1 : Data Maybe [t]
525 d2 : Sat (Maybe [t])
526 WorkList
527 [W] d4 : Foo [t]
528
529 Now, solve d4 using dfunFoo2; d4 := dfunFoo2 d5
530 Inert
531 [G] d0 : Foo t
532 [G] d01 : Data Maybe t -- Superclass of d0
533 Solved:
534 d1 : Data Maybe [t]
535 d2 : Sat (Maybe [t])
536 d4 : Foo [t]
537 WorkList:
538 [W] d5 : Foo t
539
540 Now, d5 can be solved! d5 := d0
541
542 Result
543 d1 := dfunData2 d2 d3
544 d2 := dfunSat d4
545 d3 := d01
546 d4 := dfunFoo2 d5
547 d5 := d0
548 -}
549
550 {- *********************************************************************
551 * *
552 InertCans: the canonical inerts
553 * *
554 * *
555 ********************************************************************* -}
556
557 data InertCans -- See Note [Detailed InertCans Invariants] for more
558 = IC { inert_model :: InertModel
559 -- See Note [inert_model: the inert model]
560
561 , inert_eqs :: TyVarEnv EqualCtList
562 -- See Note [inert_eqs: the inert equalities]
563 -- All Given/Wanted CTyEqCans; index is the LHS tyvar
564
565 , inert_funeqs :: FunEqMap Ct
566 -- All CFunEqCans; index is the whole family head type.
567 -- All Nominal (that's an invarint of all CFunEqCans)
568 -- LHS is fully rewritten (modulo eqCanRewrite constraints)
569 -- wrt inert_eqs/inert_model
570 -- We can get Derived ones from e.g.
571 -- (a) flattening derived equalities
572 -- (b) emitDerivedShadows
573
574 , inert_dicts :: DictMap Ct
575 -- Dictionaries only
576 -- All fully rewritten (modulo flavour constraints)
577 -- wrt inert_eqs/inert_model
578
579 , inert_safehask :: DictMap Ct
580 -- Failed dictionary resolution due to Safe Haskell overlapping
581 -- instances restriction. We keep this seperate from inert_dicts
582 -- as it doesn't cause compilation failure, just safe inference
583 -- failure.
584 --
585 -- ^ See Note [Safe Haskell Overlapping Instances Implementation]
586 -- in TcSimplify
587
588 , inert_irreds :: Cts
589 -- Irreducible predicates
590
591 , inert_insols :: Cts
592 -- Frozen errors (as non-canonicals)
593
594 , inert_count :: Int
595 -- Number of Wanted goals in
596 -- inert_eqs, inert_dicts, inert_safehask, inert_irreds
597 -- Does not include insolubles
598 -- When non-zero, keep trying to solved
599 }
600
601 type InertModel = TyVarEnv Ct
602 -- If a -> ct, then ct is a
603 -- nominal, Derived, canonical CTyEqCan for [D] (a ~N rhs)
604 -- The index of the TyVarEnv is the 'a'
605 -- All saturated info for Given, Wanted, Derived is here
606
607
608 {- Note [Detailed InertCans Invariants]
609 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
610 The InertCans represents a collection of constraints with the following properties:
611
612 * All canonical
613
614 * No two dictionaries with the same head
615 * No two CIrreds with the same type
616
617 * Family equations inert wrt top-level family axioms
618
619 * Dictionaries have no matching top-level instance
620
621 * Given family or dictionary constraints don't mention touchable
622 unification variables
623
624 * Non-CTyEqCan constraints are fully rewritten with respect
625 to the CTyEqCan equalities (modulo canRewrite of course;
626 eg a wanted cannot rewrite a given)
627
628 * CTyEqCan equalities: see Note [Applying the inert substitution]
629 in TcFlatten
630
631 Note [Type family equations]
632 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
633 Type-family equations, of form (ev : F tys ~ ty), live in three places
634
635 * The work-list, of course
636
637 * The inert_flat_cache. This is used when flattening, to get maximal
638 sharing. It contains lots of things that are still in the work-list.
639 E.g Suppose we have (w1: F (G a) ~ Int), and (w2: H (G a) ~ Int) in the
640 work list. Then we flatten w1, dumping (w3: G a ~ f1) in the work
641 list. Now if we flatten w2 before we get to w3, we still want to
642 share that (G a).
643
644 Because it contains work-list things, DO NOT use the flat cache to solve
645 a top-level goal. Eg in the above example we don't want to solve w3
646 using w3 itself!
647
648 * The inert_funeqs are un-solved but fully processed and in the InertCans.
649
650 Note [inert_model: the inert model]
651 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
652 * Part of the inert set is the “model”
653
654 * The “Model” is an non-idempotent but no-occurs-check
655 substitution, reflecting *all* *Nominal* equalities (a ~N ty)
656 that are not immediately soluble by unification.
657
658 * All the constraints in the model are Derived CTyEqCans
659 That is if (a -> ty) is in the model, then
660 we have an inert constraint [D] a ~N ty.
661
662 * There are two sources of constraints in the model:
663
664 - Derived constraints arising from functional dependencies, or
665 decomposing injective arguments of type functions, and
666 suchlike.
667
668 - A Derived "shadow copy" for every Given or Wanted (a ~N ty) in
669 inert_eqs.
670
671 * The model is not subject to "kicking-out". Reason: we make a Derived
672 shadow copy of any Given/Wanted (a ~ ty), and that Derived copy will
673 be fully rewritten by the model before it is added
674
675 * The principal reason for maintaining the model is to generate
676 equalities that tell us how to unify a variable: that is, what
677 Mark Jones calls "improvement". The same idea is sometimes also
678 called "saturation"; find all the equalities that must hold in
679 any solution.
680
681 * Domain of the model = skolems + untouchables.
682 A touchable unification variable wouuld have been unified first.
683
684 * The inert_eqs are all Given/Wanted. The Derived ones are in the
685 inert_model only.
686
687 * However inert_dicts, inert_funeqs, inert_irreds
688 may well contain derived costraints.
689
690 Note [inert_eqs: the inert equalities]
691 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
692
693 Definition [Can-rewrite relation]
694 A "can-rewrite" relation between flavours, written f1 >= f2, is a
695 binary relation with the following properties
696
697 (R1) >= is transitive
698 (R2) If f1 >= f, and f2 >= f,
699 then either f1 >= f2 or f2 >= f1
700
701 Lemma. If f1 >= f then f1 >= f1
702 Proof. By property (R2), with f1=f2
703
704 Definition [Generalised substitution]
705 A "generalised substitution" S is a set of triples (a -f-> t), where
706 a is a type variable
707 t is a type
708 f is a flavour
709 such that
710 (WF1) if (a -f1-> t1) in S
711 (a -f2-> t2) in S
712 then neither (f1 >= f2) nor (f2 >= f1) hold
713 (WF2) if (a -f-> t) is in S, then t /= a
714
715 Definition [Applying a generalised substitution]
716 If S is a generalised substitution
717 S(f,a) = t, if (a -fs-> t) in S, and fs >= f
718 = a, otherwise
719 Application extends naturally to types S(f,t), modulo roles.
720 See Note [Flavours with roles].
721
722 Theorem: S(f,a) is well defined as a function.
723 Proof: Suppose (a -f1-> t1) and (a -f2-> t2) are both in S,
724 and f1 >= f and f2 >= f
725 Then by (R2) f1 >= f2 or f2 >= f1, which contradicts (WF1)
726
727 Notation: repeated application.
728 S^0(f,t) = t
729 S^(n+1)(f,t) = S(f, S^n(t))
730
731 Definition: inert generalised substitution
732 A generalised substitution S is "inert" iff
733
734 (IG1) there is an n such that
735 for every f,t, S^n(f,t) = S^(n+1)(f,t)
736
737 By (IG1) we define S*(f,t) to be the result of exahaustively
738 applying S(f,_) to t.
739
740 ----------------------------------------------------------------
741 Our main invariant:
742 the inert CTyEqCans should be an inert generalised substitution
743 ----------------------------------------------------------------
744
745 Note that inertness is not the same as idempotence. To apply S to a
746 type, you may have to apply it recursive. But inertness does
747 guarantee that this recursive use will terminate.
748
749 Note [Extending the inert equalities]
750 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
751 Theorem [Stability under extension]
752 This is the main theorem!
753 Suppose we have a "work item"
754 a -fw-> t
755 and an inert generalised substitution S,
756 such that
757 (T1) S(fw,a) = a -- LHS of work-item is a fixpoint of S(fw,_)
758 (T2) S(fw,t) = t -- RHS of work-item is a fixpoint of S(fw,_)
759 (T3) a not in t -- No occurs check in the work item
760
761 (K1) for every (a -fs-> s) in S, then not (fw >= fs)
762 Reason: the work item is fully rewritten by S, hence not (fs >= fw)
763 but if (fw >= fs) then the work item could rewrite
764 the inert item
765
766 (K2) for every (b -fs-> s) in S, where b /= a, then
767 (K2a) not (fs >= fs)
768 or (K2b) fs >= fw
769 or (K2c) not (fw >= fs)
770 or (K2d) a not in s
771
772 (K3) See Note [K3: completeness of solving]
773 If (b -fs-> s) is in S with (fw >= fs), then
774 (K3a) If the role of fs is nominal: s /= a
775 (K3b) If the role of fs is representational: EITHER
776 a not in s, OR
777 the path from the top of s to a includes at least one non-newtype
778
779 then the extended substition T = S+(a -fw-> t)
780 is an inert generalised substitution.
781
782 Conditions (T1-T3) are established by the canonicaliser
783 Conditions (K1-K3) are established by TcSMonad.kickOutRewriteable
784
785 The idea is that
786 * (T1-2) are guaranteed by exhaustively rewriting the work-item
787 with S(fw,_).
788
789 * T3 is guaranteed by a simple occurs-check on the work item.
790 This is done during canonicalisation, in canEqTyVar;
791 (invariant: a CTyEqCan never has an occurs check).
792
793 * (K1-3) are the "kick-out" criteria. (As stated, they are really the
794 "keep" criteria.) If the current inert S contains a triple that does
795 not satisfy (K1-3), then we remove it from S by "kicking it out",
796 and re-processing it.
797
798 * Note that kicking out is a Bad Thing, because it means we have to
799 re-process a constraint. The less we kick out, the better.
800 TODO: Make sure that kicking out really *is* a Bad Thing. We've assumed
801 this but haven't done the empirical study to check.
802
803 * Assume we have G>=G, G>=W and that's all. Then, when performing
804 a unification we add a new given a -G-> ty. But doing so does NOT require
805 us to kick out an inert wanted that mentions a, because of (K2a). This
806 is a common case, hence good not to kick out.
807
808 * Lemma (L2): if not (fw >= fw), then K1-K3 all hold.
809 Proof: using Definition [Can-rewrite relation], fw can't rewrite anything
810 and so K1-K3 hold. Intuitively, since fw can't rewrite anything,
811 adding it cannot cause any loops
812 This is a common case, because Wanteds cannot rewrite Wanteds.
813
814 * Lemma (L1): The conditions of the Main Theorem imply that there is no
815 (a -fs-> t) in S, s.t. (fs >= fw).
816 Proof. Suppose the contrary (fs >= fw). Then because of (T1),
817 S(fw,a)=a. But since fs>=fw, S(fw,a) = s, hence s=a. But now we
818 have (a -fs-> a) in S, which contradicts (WF2).
819
820 * The extended substitution satisfies (WF1) and (WF2)
821 - (K1) plus (L1) guarantee that the extended substitution satisfies (WF1).
822 - (T3) guarantees (WF2).
823
824 * (K2) is about inertness. Intuitively, any infinite chain T^0(f,t),
825 T^1(f,t), T^2(f,T).... must pass through the new work item infnitely
826 often, since the substution without the work item is inert; and must
827 pass through at least one of the triples in S infnitely often.
828
829 - (K2a): if not(fs>=fs) then there is no f that fs can rewrite (fs>=f),
830 and hence this triple never plays a role in application S(f,a).
831 It is always safe to extend S with such a triple.
832
833 (NB: we could strengten K1) in this way too, but see K3.
834
835 - (K2b): If this holds then, by (T2), b is not in t. So applying the
836 work item does not genenerate any new opportunities for applying S
837
838 - (K2c): If this holds, we can't pass through this triple infinitely
839 often, because if we did then fs>=f, fw>=f, hence by (R2)
840 * either fw>=fs, contradicting K2c
841 * or fs>=fw; so by the agument in K2b we can't have a loop
842
843 - (K2d): if a not in s, we hae no further opportunity to apply the
844 work item, similar to (K2b)
845
846 NB: Dimitrios has a PDF that does this in more detail
847
848 Key lemma to make it watertight.
849 Under the conditions of the Main Theorem,
850 forall f st fw >= f, a is not in S^k(f,t), for any k
851
852 Also, consider roles more carefully. See Note [Flavours with roles]
853
854 Note [K3: completeness of solving]
855 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
856 (K3) is not necessary for the extended substitution
857 to be inert. In fact K1 could be made stronger by saying
858 ... then (not (fw >= fs) or not (fs >= fs))
859 But it's not enough for S to be inert; we also want completeness.
860 That is, we want to be able to solve all soluble wanted equalities.
861 Suppose we have
862
863 work-item b -G-> a
864 inert-item a -W-> b
865
866 Assuming (G >= W) but not (W >= W), this fulfills all the conditions,
867 so we could extend the inerts, thus:
868
869 inert-items b -G-> a
870 a -W-> b
871
872 But if we kicked-out the inert item, we'd get
873
874 work-item a -W-> b
875 inert-item b -G-> a
876
877 Then rewrite the work-item gives us (a -W-> a), which is soluble via Refl.
878 So we add one more clause to the kick-out criteria
879
880 Another way to understand (K3) is that we treat an inert item
881 a -f-> b
882 in the same way as
883 b -f-> a
884 So if we kick out one, we should kick out the other. The orientation
885 is somewhat accidental.
886
887 When considering roles, we also need the second clause (K3b). Consider
888
889 inert-item a -W/R-> b c
890 work-item c -G/N-> a
891
892 The work-item doesn't get rewritten by the inert, because (>=) doesn't hold.
893 We've satisfied conditions (T1)-(T3) and (K1) and (K2). If all we had were
894 condition (K3a), then we would keep the inert around and add the work item.
895 But then, consider if we hit the following:
896
897 work-item2 b -G/N-> Id
898
899 where
900
901 newtype Id x = Id x
902
903 For similar reasons, if we only had (K3a), we wouldn't kick the
904 representational inert out. And then, we'd miss solving the inert, which
905 now reduced to reflexivity. The solution here is to kick out representational
906 inerts whenever the tyvar of a work item is "exposed", where exposed means
907 not under some proper data-type constructor, like [] or Maybe. See
908 isTyVarExposed in TcType. This is encoded in (K3b).
909
910 Note [Stability of flattening]
911 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
912 The inert_eqs and inert_model, *considered separately* are each stable;
913 that is, substituting using them will terminate. Considered *together*
914 they are not. E.g.
915
916 Add: [G] a~[b] to inert set with model [D] b~[a]
917
918 We add [G] a~[b] to inert_eqs, and emit [D] a~[b]. At this point
919 the combination of inert_eqs and inert_model is not stable.
920
921 Then we canonicalise [D] a~[b] to [D] a~[[a]], and add that to
922 insolubles as an occurs check.
923
924 * When canonicalizing, the flattener respects flavours. In particular,
925 when flattening a type variable 'a':
926 * Derived: look up 'a' in the inert_model
927 * Given/Wanted: look up 'a' in the inert_eqs
928
929
930 Note [Flavours with roles]
931 ~~~~~~~~~~~~~~~~~~~~~~~~~~
932 The system described in Note [inert_eqs: the inert equalities]
933 discusses an abstract
934 set of flavours. In GHC, flavours have two components: the flavour proper,
935 taken from {Wanted, Derived, Given} and the equality relation (often called
936 role), taken from {NomEq, ReprEq}.
937 When substituting w.r.t. the inert set,
938 as described in Note [inert_eqs: the inert equalities],
939 we must be careful to respect all components of a flavour.
940 For example, if we have
941
942 inert set: a -G/R-> Int
943 b -G/R-> Bool
944
945 type role T nominal representational
946
947 and we wish to compute S(W/R, T a b), the correct answer is T a Bool, NOT
948 T Int Bool. The reason is that T's first parameter has a nominal role, and
949 thus rewriting a to Int in T a b is wrong. Indeed, this non-congruence of
950 substitution means that the proof in Note [The inert equalities] may need
951 to be revisited, but we don't think that the end conclusion is wrong.
952
953 Note [Examples of how the inert_model helps completeness]
954 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
955
956 ----------- Example 2 (indexed-types/should_fail/T4093a)
957 Ambiguity check for f: (Foo e ~ Maybe e) => Foo e
958
959 We get [G] Foo e ~ Maybe e
960 [W] Foo e ~ Foo ee -- ee is a unification variable
961 [W] Foo ee ~ Maybe ee
962
963 Flatten: [G] Foo e ~ fsk
964 [G] fsk ~ Maybe e -- (A)
965
966 [W] Foo ee ~ fmv
967 [W] fmv ~ fsk -- (B) From Foo e ~ Foo ee
968 [W] fmv ~ Maybe ee
969
970 --> rewrite (B) with (A)
971 [W] Foo ee ~ fmv
972 [W] fmv ~ Maybe e
973 [W] fmv ~ Maybe ee
974
975 But now awe appear to be stuck, since we don't rewrite Wanteds with
976 Wanteds. But inert_model to the rescue. In the model we first added
977 fmv -> Maybe e
978 Then when adding [W] fmv -> Maybe ee to the inert set, we noticed
979 that the model can rewrite the constraint, and so emit [D] fmv ~ Maybe ee.
980 That canonicalises to
981 [D] Maybe e ~ Maybe ee
982 and that soon yields ee := e, and all is well
983
984 ----------- Example 3 (typecheck/should_compile/Improvement.hs)
985 type instance F Int = Bool
986 instance (b~Int) => C Bool b
987
988 [W] w1 : C (F alpha) alpha, [W] w2 : F alpha ~ Bool
989
990 If we rewrote wanteds with wanteds, we could rewrite w1 to
991 C Bool alpha, use the instance to get alpha ~ Int, and solve
992 the whole thing.
993
994 And that is exactly what happens, in the *Derived* constraints.
995 In effect we get
996
997 [D] F alpha ~ fmv
998 [D] C fmv alpha
999 [D] fmv ~ Bool
1000
1001 and now we can rewrite (C fmv alpha) with (fmv ~ Bool), ane
1002 we are off to the races.
1003
1004 ----------- Example 4 (Trac #10009, a nasty example):
1005
1006 f :: (UnF (F b) ~ b) => F b -> ()
1007
1008 g :: forall a. (UnF (F a) ~ a) => a -> ()
1009 g _ = f (undefined :: F a)
1010
1011 For g we get [G] UnF (F a) ~ a
1012 [W] UnF (F beta) ~ beta
1013 [W] F a ~ F beta
1014 Flatten:
1015 [G] g1: F a ~ fsk1 fsk1 := F a
1016 [G] g2: UnF fsk1 ~ fsk2 fsk2 := UnF fsk1
1017 [G] g3: fsk2 ~ a
1018
1019 [W] w1: F beta ~ fmv1
1020 [W] w2: UnF fmv1 ~ fmv2
1021 [W] w3: beta ~ fmv2
1022 [W] w5: fmv1 ~ fsk1 -- From F a ~ F beta using flat-cache
1023 -- and re-orient to put meta-var on left
1024
1025 Unify beta := fmv2
1026 [W] w1: F fmv2 ~ fmv1
1027 [W] w2: UnF fmv1 ~ fmv2
1028 [W] w5: fmv1 ~ fsk1
1029
1030 In the model, we have the shadow Deriveds of w1 and w2
1031 (I name them for convenience even though they are anonymous)
1032 [D] d1: F fmv2 ~ fmv1d
1033 [D] d2: fmv1d ~ fmv1
1034 [D] d3: UnF fmv1 ~ fmv2d
1035 [D] d4: fmv2d ~ fmv2
1036
1037 Now we can rewrite d3 with w5, and match with g2, to get
1038 fmv2d := fsk2
1039 [D] d1: F fmv2 ~ fmv1d
1040 [D] d2: fmv1d ~ fmv1
1041 [D] d4: fmv2 ~ fsk2
1042
1043 Use g2 to rewrite fsk2 to a.
1044 [D] d1: F fmv2 ~ fmv1d
1045 [D] d2: fmv1d ~ fmv1
1046 [D] d4: fmv2 ~ a
1047
1048 Use d4 to rewrite d1, rewrite with g3,
1049 match with g1, to get
1050 fmv1d := fsk1
1051 [D] d2: fmv1 ~ fsk1
1052 [D] d4: fmv2 ~ a
1053
1054 At this point we are stuck so we unflatten this set:
1055 See Note [Orientation of equalities with fmvs] in TcFlatten
1056 [W] w1: F fmv2 ~ fmv1
1057 [W] w2: UnF fmv1 ~ fmv2
1058 [W] w5: fmv1 ~ fsk1
1059 [D] d4: fmv2 ~ a
1060
1061 Unflattening will discharge w1: fmv1 := F fmv2
1062 It can't discharge w2, so it is kept. But we can
1063 unify fmv2 := fsk2, and that is "progress". Result
1064 [W] w2: UnF (F a) ~ a
1065 [W] w5: F a ~ fsk1
1066
1067 And now both of these are easily proved in the next iteration. Phew!
1068 -}
1069
1070 instance Outputable InertCans where
1071 ppr (IC { inert_model = model, inert_eqs = eqs
1072 , inert_funeqs = funeqs, inert_dicts = dicts
1073 , inert_safehask = safehask, inert_irreds = irreds
1074 , inert_insols = insols, inert_count = count })
1075 = braces $ vcat
1076 [ ppUnless (isEmptyVarEnv eqs) $
1077 text "Equalities:"
1078 <+> pprCts (foldVarEnv (\eqs rest -> listToBag eqs `andCts` rest) emptyCts eqs)
1079 , ppUnless (isEmptyTcAppMap funeqs) $
1080 text "Type-function equalities =" <+> pprCts (funEqsToBag funeqs)
1081 , ppUnless (isEmptyTcAppMap dicts) $
1082 text "Dictionaries =" <+> pprCts (dictsToBag dicts)
1083 , ppUnless (isEmptyTcAppMap safehask) $
1084 text "Safe Haskell unsafe overlap =" <+> pprCts (dictsToBag safehask)
1085 , ppUnless (isEmptyCts irreds) $
1086 text "Irreds =" <+> pprCts irreds
1087 , ppUnless (isEmptyCts insols) $
1088 text "Insolubles =" <+> pprCts insols
1089 , ppUnless (isEmptyVarEnv model) $
1090 text "Model =" <+> pprCts (foldVarEnv consCts emptyCts model)
1091 , text "Unsolved goals =" <+> int count
1092 ]
1093
1094 {- *********************************************************************
1095 * *
1096 Adding an inert
1097 * *
1098 ************************************************************************
1099
1100 Note [Adding an inert canonical constraint the InertCans]
1101 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1102 * Adding any constraint c *other* than a CTyEqCan (TcSMonad.addInertCan):
1103
1104 * If c can be rewritten by model, emit the shadow constraint [D] c
1105 as NonCanonical. See Note [Emitting shadow constraints]
1106
1107 * Reason for non-canonical: a CFunEqCan has a unique fmv on the RHS,
1108 so we must not duplicate it.
1109
1110 * Adding a *nominal* CTyEqCan (a ~N ty) to the inert set (TcSMonad.addInertEq).
1111
1112 (A) Always (G/W/D) kick out constraints that can be rewritten
1113 (respecting flavours) by the new constraint. This is done
1114 by kickOutRewritable.
1115
1116 (B) Applies only to nominal equalities: a ~ ty. Four cases:
1117
1118 [Representational] [G/W/D] a ~R ty:
1119 Just add it to inert_eqs
1120
1121 [Derived Nominal] [D] a ~N ty:
1122 1. Add (a~ty) to the model
1123 NB: 'a' cannot be in fv(ty), because the constraint is canonical.
1124
1125 2. (DShadow) Do emitDerivedShadows
1126 For every inert G/W constraint c, st
1127 (a) (a~ty) can rewrite c (see Note [Emitting shadow constraints]),
1128 and
1129 (b) the model cannot rewrite c
1130 kick out a Derived *copy*, leaving the original unchanged.
1131 Reason for (b) if the model can rewrite c, then we have already
1132 generated a shadow copy
1133
1134 [Given/Wanted Nominal] [G/W] a ~N ty:
1135 1. Add it to inert_eqs
1136 2. Emit [D] a~ty
1137 Step (2) is needed to allow the current model to fully
1138 rewrite [D] a~ty before adding it using the [Derived Nominal]
1139 steps above.
1140
1141 We must do this even for Givens, because
1142 work-item [G] a ~ [b], model has [D] b ~ a.
1143 We need a shadow [D] a ~ [b] in the work-list
1144 When we process it, we'll rewrite to a ~ [a] and get an occurs check
1145
1146
1147 * Unifying a:=ty, is like adding [G] a~ty, but we can't make a [D]
1148 a~ty, as in step (1) of the [G/W] case above. So instead, do
1149 kickOutAfterUnification:
1150 - Kick out from the model any equality (b~ty2) that mentions 'a'
1151 (i.e. a=b or a in ty2). Example:
1152 [G] a ~ [b], model [D] b ~ [a]
1153
1154 Note [Emitting shadow constraints]
1155 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1156 * Given a new model element [D] a ~ ty, we want to emit shadow
1157 [D] constraints for any inert constraints 'c' that can be
1158 rewritten [D] a-> ty
1159
1160 * And similarly given a new Given/Wanted 'c', we want to emit a
1161 shadow 'c' if the model can rewrite [D] c
1162
1163 See modelCanRewrite.
1164
1165 NB the use of rewritableTyVars. You might wonder whether, given the new
1166 constraint [D] fmv ~ ty and the inert [W] F alpha ~ fmv, do we want to
1167 emit a shadow constraint [D] F alpha ~ fmv? No, we don't, because
1168 it'll literally be a duplicate (since we do not rewrite the RHS of a
1169 CFunEqCan) and hence immediately eliminated again. Insetad we simply
1170 want to *kick-out* the [W] F alpha ~ fmv, so that it is reconsidered
1171 from a fudep point of view. See Note [Kicking out CFunEqCan for
1172 fundeps]
1173
1174 Note [Kicking out CFunEqCan for fundeps]
1175 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1176 Consider:
1177 New: [D] fmv1 ~ fmv2
1178 Inert: [W] F alpha ~ fmv1
1179 [W] F beta ~ fmv2
1180
1181 The new (derived) equality certainly can't rewrite the inerts. But we
1182 *must* kick out the first one, to get:
1183
1184 New: [W] F alpha ~ fmv1
1185 Inert: [W] F beta ~ fmv2
1186 Model: [D] fmv1 ~ fmv2
1187
1188 and now improvement will discover [D] alpha ~ beta. This is important;
1189 eg in Trac #9587.
1190 -}
1191
1192 addInertEq :: Ct -> TcS ()
1193 -- This is a key function, because of the kick-out stuff
1194 -- Precondition: item /is/ canonical
1195 addInertEq ct@(CTyEqCan { cc_tyvar = tv })
1196 = do { traceTcS "addInertEq {" $
1197 text "Adding new inert equality:" <+> ppr ct
1198 ; ics <- getInertCans
1199
1200 ; let (kicked_out, ics1) = kickOutRewritable (ctFlavourRole ct) tv ics
1201 ; ics2 <- add_inert_eq ics1 ct
1202
1203 ; setInertCans ics2
1204
1205 ; unless (isEmptyWorkList kicked_out) $
1206 do { updWorkListTcS (appendWorkList kicked_out)
1207 ; csTraceTcS $
1208 hang (text "Kick out, tv =" <+> ppr tv)
1209 2 (vcat [ text "n-kicked =" <+> int (workListSize kicked_out)
1210 , ppr kicked_out ]) }
1211
1212 ; traceTcS "addInertEq }" $ empty }
1213 addInertEq ct = pprPanic "addInertEq" (ppr ct)
1214
1215 add_inert_eq :: InertCans -> Ct -> TcS InertCans
1216 add_inert_eq ics@(IC { inert_count = n
1217 , inert_eqs = old_eqs
1218 , inert_model = old_model })
1219 ct@(CTyEqCan { cc_ev = ev, cc_eq_rel = eq_rel, cc_tyvar = tv
1220 , cc_rhs = _rhs })
1221 | ReprEq <- eq_rel
1222 = return new_ics
1223
1224 | isDerived ev
1225 = do { emitDerivedShadows ics tv
1226 ; return (ics { inert_model = extendVarEnv old_model tv ct }) }
1227
1228 | otherwise -- Given/Wanted Nominal equality [W] tv ~N ty
1229 = do { emitNewDerived loc pred
1230 ; return new_ics }
1231 where
1232 loc = ctEvLoc ev
1233 pred = ctEvPred ev
1234 new_ics = ics { inert_eqs = addTyEq old_eqs tv ct
1235 , inert_count = bumpUnsolvedCount ev n }
1236
1237 add_inert_eq _ ct = pprPanic "addInertEq" (ppr ct)
1238
1239 emitDerivedShadows :: InertCans -> TcTyVar -> TcS ()
1240 emitDerivedShadows IC { inert_eqs = tv_eqs
1241 , inert_dicts = dicts
1242 , inert_safehask = safehask
1243 , inert_funeqs = funeqs
1244 , inert_irreds = irreds
1245 , inert_model = model } new_tv
1246 | null shadows
1247 = return ()
1248 | otherwise
1249 = do { traceTcS "Emit derived shadows:" $
1250 vcat [ text "tyvar =" <+> ppr new_tv
1251 , text "shadows =" <+> vcat (map ppr shadows) ]
1252 ; emitWork shadows }
1253 where
1254 shadows = foldDicts get_ct dicts $
1255 foldDicts get_ct safehask $
1256 foldFunEqs get_ct funeqs $
1257 foldIrreds get_ct irreds $
1258 foldTyEqs get_ct tv_eqs []
1259 -- Ignore insolubles
1260
1261 get_ct ct cts | want_shadow ct = mkShadowCt ct : cts
1262 | otherwise = cts
1263
1264 want_shadow ct
1265 = not (isDerivedCt ct) -- No need for a shadow of a Derived!
1266 && (new_tv `elemVarSet` rw_tvs) -- New tv can rewrite ct, yielding a
1267 -- different ct
1268 && not (modelCanRewrite model rw_tvs)-- We have not already created a
1269 -- shadow
1270 where
1271 rw_tvs = rewritableTyCoVars ct
1272
1273 mkShadowCt :: Ct -> Ct
1274 -- Produce a Derived shadow constraint from the input
1275 -- If it is a CFunEqCan, make it NonCanonical, to avoid
1276 -- duplicating the flatten-skolems
1277 -- Otherwise keep the canonical shape. This just saves work, but
1278 -- is sometimes important; see Note [Keep CDictCan shadows as CDictCan]
1279 mkShadowCt ct
1280 | CFunEqCan {} <- ct = CNonCanonical { cc_ev = derived_ev }
1281 | otherwise = ct { cc_ev = derived_ev }
1282 where
1283 ev = ctEvidence ct
1284 derived_ev = CtDerived { ctev_pred = ctEvPred ev
1285 , ctev_loc = ctEvLoc ev }
1286
1287 {- Note [Keep CDictCan shadows as CDictCan]
1288 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1289 Suppose we have
1290 class C a => D a b
1291 and [G] D a b, [G] C a in the inert set. Now we insert
1292 [D] b ~ c. We want to kick out a derived shadow for [D] D a b,
1293 so we can rewrite it with the new constraint, and perhaps get
1294 instance reduction or other consequences.
1295
1296 BUT we do not want to kick out a *non-canonical* (D a b). If we
1297 did, we would do this:
1298 - rewrite it to [D] D a c, with pend_sc = True
1299 - use expandSuperClasses to add C a
1300 - go round again, which solves C a from the givens
1301 This loop goes on for ever and triggers the simpl_loop limit.
1302
1303 Solution: kick out the CDictCan which will have pend_sc = False,
1304 because we've already added its superclasses. So we won't re-add
1305 them. If we forget the pend_sc flag, our cunning scheme for avoiding
1306 generating superclasses repeatedly will fail.
1307
1308 See Trac #11379 for a case of this.
1309 -}
1310
1311 modelCanRewrite :: InertModel -> TcTyCoVarSet -> Bool
1312 -- See Note [Emitting shadow constraints]
1313 -- True if there is any intersection between dom(model) and tvs
1314 modelCanRewrite model tvs = not (disjointUFM model tvs)
1315 -- The low-level use of disjointUFM might e surprising.
1316 -- InertModel = TyVarEnv Ct, and we want to see if its domain
1317 -- is disjoint from that of a TcTyCoVarSet. So we drop down
1318 -- to the underlying UniqFM. A bit yukky, but efficient.
1319
1320 rewritableTyCoVars :: Ct -> TcTyVarSet
1321 -- The tyvars of a Ct that can be rewritten
1322 rewritableTyCoVars (CFunEqCan { cc_tyargs = tys }) = tyCoVarsOfTypes tys
1323 rewritableTyCoVars ct = tyCoVarsOfType (ctPred ct)
1324
1325 --------------
1326 addInertCan :: Ct -> TcS () -- Constraints *other than* equalities
1327 addInertCan ct
1328 = do { traceTcS "insertInertCan {" $
1329 text "Trying to insert new inert item:" <+> ppr ct
1330
1331 ; ics <- getInertCans
1332 ; setInertCans (add_item ics ct)
1333
1334 -- Emit shadow derived if necessary
1335 -- See Note [Emitting shadow constraints]
1336 ; let rw_tvs = rewritableTyCoVars ct
1337 ; when (not (isDerivedCt ct) && modelCanRewrite (inert_model ics) rw_tvs)
1338 (emitWork [mkShadowCt ct])
1339
1340 ; traceTcS "addInertCan }" $ empty }
1341
1342 add_item :: InertCans -> Ct -> InertCans
1343 add_item ics item@(CFunEqCan { cc_fun = tc, cc_tyargs = tys })
1344 = ics { inert_funeqs = insertFunEq (inert_funeqs ics) tc tys item }
1345
1346 add_item ics item@(CIrredEvCan { cc_ev = ev })
1347 = ics { inert_irreds = inert_irreds ics `Bag.snocBag` item
1348 , inert_count = bumpUnsolvedCount ev (inert_count ics) }
1349 -- The 'False' is because the irreducible constraint might later instantiate
1350 -- to an equality.
1351 -- But since we try to simplify first, if there's a constraint function FC with
1352 -- type instance FC Int = Show
1353 -- we'll reduce a constraint (FC Int a) to Show a, and never add an inert irreducible
1354
1355 add_item ics item@(CDictCan { cc_ev = ev, cc_class = cls, cc_tyargs = tys })
1356 = ics { inert_dicts = addDict (inert_dicts ics) cls tys item
1357 , inert_count = bumpUnsolvedCount ev (inert_count ics) }
1358
1359 add_item _ item
1360 = pprPanic "upd_inert set: can't happen! Inserting " $
1361 ppr item -- CTyEqCan is dealt with by addInertEq
1362 -- Can't be CNonCanonical, CHoleCan,
1363 -- because they only land in inert_insols
1364
1365 bumpUnsolvedCount :: CtEvidence -> Int -> Int
1366 bumpUnsolvedCount ev n | isWanted ev = n+1
1367 | otherwise = n
1368
1369
1370 -----------------------------------------
1371 kickOutRewritable :: CtFlavourRole -- Flavour/role of the equality that
1372 -- is being added to the inert set
1373 -> TcTyVar -- The new equality is tv ~ ty
1374 -> InertCans
1375 -> (WorkList, InertCans)
1376 -- See Note [kickOutRewritable]
1377 kickOutRewritable new_fr new_tv ics@(IC { inert_eqs = tv_eqs
1378 , inert_dicts = dictmap
1379 , inert_safehask = safehask
1380 , inert_funeqs = funeqmap
1381 , inert_irreds = irreds
1382 , inert_insols = insols
1383 , inert_count = n
1384 , inert_model = model })
1385 | not (new_fr `eqCanRewriteFR` new_fr)
1386 = (emptyWorkList, ics)
1387 -- If new_fr can't rewrite itself, it can't rewrite
1388 -- anything else, so no need to kick out anything.
1389 -- (This is a common case: wanteds can't rewrite wanteds)
1390 -- Lemma (L2) in Note [Extending the inert equalities]
1391
1392 | otherwise
1393 = (kicked_out, inert_cans_in)
1394 where
1395 inert_cans_in = IC { inert_eqs = tv_eqs_in
1396 , inert_dicts = dicts_in
1397 , inert_safehask = safehask -- ??
1398 , inert_funeqs = feqs_in
1399 , inert_irreds = irs_in
1400 , inert_insols = insols_in
1401 , inert_count = n - workListWantedCount kicked_out
1402 , inert_model = model }
1403 -- Leave the model unchanged
1404
1405 kicked_out = WL { wl_eqs = tv_eqs_out
1406 , wl_funeqs = feqs_out
1407 , wl_deriv = []
1408 , wl_rest = bagToList (dicts_out `andCts` irs_out
1409 `andCts` insols_out)
1410 , wl_implics = emptyBag }
1411
1412 (tv_eqs_out, tv_eqs_in) = foldVarEnv kick_out_eqs ([], emptyVarEnv) tv_eqs
1413 (feqs_out, feqs_in) = partitionFunEqs kick_out_fe funeqmap
1414 (dicts_out, dicts_in) = partitionDicts kick_out_ct dictmap
1415 (irs_out, irs_in) = partitionBag kick_out_ct irreds
1416 (insols_out, insols_in) = partitionBag kick_out_ct insols
1417 -- Kick out even insolubles; see Note [Kick out insolubles]
1418
1419 fr_can_rewrite :: CtEvidence -> Bool
1420 fr_can_rewrite ev = new_fr `eqCanRewriteFR` (ctEvFlavourRole ev)
1421
1422 kick_out_ct :: Ct -> Bool
1423 -- Kick it out if the new CTyEqCan can rewrite the inert
1424 -- one. See Note [kickOutRewritable]
1425 kick_out_ct ct
1426 = fr_can_rewrite ev
1427 && new_tv `elemVarSet` tyCoVarsOfType (ctEvPred ev)
1428 where
1429 ev = ctEvidence ct
1430
1431 kick_out_fe :: Ct -> Bool
1432 kick_out_fe (CFunEqCan { cc_ev = ev, cc_tyargs = tys, cc_fsk = fsk })
1433 = new_tv == fsk -- If RHS is new_tvs, kick out /regardless of flavour/
1434 -- See Note [Kicking out CFunEqCan for fundeps]
1435 || (fr_can_rewrite ev
1436 && new_tv `elemVarSet` tyCoVarsOfTypes tys)
1437 kick_out_fe ct = pprPanic "kick_out_fe" (ppr ct)
1438
1439 kick_out_eqs :: EqualCtList -> ([Ct], TyVarEnv EqualCtList)
1440 -> ([Ct], TyVarEnv EqualCtList)
1441 kick_out_eqs eqs (acc_out, acc_in)
1442 = (eqs_out ++ acc_out, case eqs_in of
1443 [] -> acc_in
1444 (eq1:_) -> extendVarEnv acc_in (cc_tyvar eq1) eqs_in)
1445 where
1446 (eqs_in, eqs_out) = partition keep_eq eqs
1447
1448 -- Implements criteria K1-K3 in Note [Extending the inert equalities]
1449 keep_eq (CTyEqCan { cc_tyvar = tv, cc_rhs = rhs_ty, cc_ev = ev
1450 , cc_eq_rel = eq_rel })
1451 | tv == new_tv
1452 = not (fr_can_rewrite ev) -- (K1)
1453
1454 | otherwise
1455 = check_k2 && check_k3
1456 where
1457 fs = ctEvFlavourRole ev
1458 check_k2 = not (fs `eqCanRewriteFR` fs) -- (K2a)
1459 || (fs `eqCanRewriteFR` new_fr) -- (K2b)
1460 || not (new_fr `eqCanRewriteFR` fs) -- (K2c)
1461 || not (new_tv `elemVarSet` tyCoVarsOfType rhs_ty) -- (K2d)
1462
1463 check_k3
1464 | new_fr `eqCanRewriteFR` fs
1465 = case eq_rel of
1466 NomEq -> not (rhs_ty `eqType` mkTyVarTy new_tv)
1467 ReprEq -> not (isTyVarExposed new_tv rhs_ty)
1468
1469 | otherwise
1470 = True
1471
1472 keep_eq ct = pprPanic "keep_eq" (ppr ct)
1473
1474 kickOutAfterUnification :: TcTyVar -> TcS Int
1475 kickOutAfterUnification new_tv
1476 = do { ics <- getInertCans
1477 ; let (kicked_out1, ics1) = kickOutModel new_tv ics
1478 (kicked_out2, ics2) = kickOutRewritable (Given,NomEq)
1479 new_tv ics1
1480 -- Given because the tv := xi is given; NomEq because
1481 -- only nominal equalities are solved by unification
1482 kicked_out = appendWorkList kicked_out1 kicked_out2
1483 ; setInertCans ics2
1484 ; updWorkListTcS (appendWorkList kicked_out)
1485
1486 ; unless (isEmptyWorkList kicked_out) $
1487 csTraceTcS $
1488 hang (text "Kick out (unify), tv =" <+> ppr new_tv)
1489 2 (vcat [ text "n-kicked =" <+> int (workListSize kicked_out)
1490 , text "kicked_out =" <+> ppr kicked_out
1491 , text "Residual inerts =" <+> ppr ics2 ])
1492 ; return (workListSize kicked_out) }
1493
1494 kickOutModel :: TcTyVar -> InertCans -> (WorkList, InertCans)
1495 kickOutModel new_tv ics@(IC { inert_model = model, inert_eqs = eqs })
1496 = (foldVarEnv add emptyWorkList der_out, ics { inert_model = new_model })
1497 where
1498 (der_out, new_model) = partitionVarEnv kick_out_der model
1499
1500 kick_out_der :: Ct -> Bool
1501 kick_out_der (CTyEqCan { cc_tyvar = tv, cc_rhs = rhs })
1502 = new_tv == tv || new_tv `elemVarSet` tyCoVarsOfType rhs
1503 kick_out_der _ = False
1504
1505 add :: Ct -> WorkList -> WorkList
1506 -- Don't kick out a Derived if there is a Given or Wanted with
1507 -- the same predicate. The model is just a shadow copy, and the
1508 -- Given/Wanted will serve the purpose.
1509 add (CTyEqCan { cc_ev = ev, cc_tyvar = tv, cc_rhs = rhs }) wl
1510 | not (isInInertEqs eqs tv rhs) = extendWorkListDerived (ctEvLoc ev) ev wl
1511 add _ wl = wl
1512
1513
1514 {- Note [kickOutRewritable]
1515 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
1516 See also Note [inert_eqs: the inert equalities].
1517
1518 When we add a new inert equality (a ~N ty) to the inert set,
1519 we must kick out any inert items that could be rewritten by the
1520 new equality, to maintain the inert-set invariants.
1521
1522 - We want to kick out an existing inert constraint if
1523 a) the new constraint can rewrite the inert one
1524 b) 'a' is free in the inert constraint (so that it *will*)
1525 rewrite it if we kick it out.
1526
1527 For (b) we use tyCoVarsOfCt, which returns the type variables /and
1528 the kind variables/ that are directly visible in the type. Hence
1529 we will have exposed all the rewriting we care about to make the
1530 most precise kinds visible for matching classes etc. No need to
1531 kick out constraints that mention type variables whose kinds
1532 contain this variable!
1533
1534 - We do not need to kick anything out from the model; we only
1535 add [D] constraints to the model (in effect) and they are
1536 fully rewritten by the model, so (K2b) holds
1537
1538 - A Derived equality can kick out [D] constraints in inert_dicts,
1539 inert_irreds etc. Nothing in inert_eqs because there are no
1540 Derived constraints in inert_eqs (they are in the model)
1541
1542 - We don't kick out constraints from inert_solved_dicts, and
1543 inert_solved_funeqs optimistically. But when we lookup we have to
1544 take the substitution into account
1545
1546
1547 Note [Kick out insolubles]
1548 ~~~~~~~~~~~~~~~~~~~~~~~~~~
1549 Suppose we have an insoluble alpha ~ [alpha], which is insoluble
1550 because an occurs check. And then we unify alpha := [Int].
1551 Then we really want to rewrite the insoluble to [Int] ~ [[Int]].
1552 Now it can be decomposed. Otherwise we end up with a "Can't match
1553 [Int] ~ [[Int]]" which is true, but a bit confusing because the
1554 outer type constructors match.
1555 -}
1556
1557
1558
1559 --------------
1560 addInertSafehask :: InertCans -> Ct -> InertCans
1561 addInertSafehask ics item@(CDictCan { cc_class = cls, cc_tyargs = tys })
1562 = ics { inert_safehask = addDict (inert_dicts ics) cls tys item }
1563
1564 addInertSafehask _ item
1565 = pprPanic "addInertSafehask: can't happen! Inserting " $ ppr item
1566
1567 insertSafeOverlapFailureTcS :: Ct -> TcS ()
1568 -- See Note [Safe Haskell Overlapping Instances Implementation] in TcSimplify
1569 insertSafeOverlapFailureTcS item
1570 = updInertCans (\ics -> addInertSafehask ics item)
1571
1572 getSafeOverlapFailures :: TcS Cts
1573 -- See Note [Safe Haskell Overlapping Instances Implementation] in TcSimplify
1574 getSafeOverlapFailures
1575 = do { IC { inert_safehask = safehask } <- getInertCans
1576 ; return $ foldDicts consCts safehask emptyCts }
1577
1578 --------------
1579 addSolvedDict :: CtEvidence -> Class -> [Type] -> TcS ()
1580 -- Add a new item in the solved set of the monad
1581 -- See Note [Solved dictionaries]
1582 addSolvedDict item cls tys
1583 | isIPPred (ctEvPred item) -- Never cache "solved" implicit parameters (not sure why!)
1584 = return ()
1585 | otherwise
1586 = do { traceTcS "updSolvedSetTcs:" $ ppr item
1587 ; updInertTcS $ \ ics ->
1588 ics { inert_solved_dicts = addDict (inert_solved_dicts ics) cls tys item } }
1589
1590 {- *********************************************************************
1591 * *
1592 Other inert-set operations
1593 * *
1594 ********************************************************************* -}
1595
1596 updInertTcS :: (InertSet -> InertSet) -> TcS ()
1597 -- Modify the inert set with the supplied function
1598 updInertTcS upd_fn
1599 = do { is_var <- getTcSInertsRef
1600 ; wrapTcS (do { curr_inert <- TcM.readTcRef is_var
1601 ; TcM.writeTcRef is_var (upd_fn curr_inert) }) }
1602
1603 getInertCans :: TcS InertCans
1604 getInertCans = do { inerts <- getTcSInerts; return (inert_cans inerts) }
1605
1606 setInertCans :: InertCans -> TcS ()
1607 setInertCans ics = updInertTcS $ \ inerts -> inerts { inert_cans = ics }
1608
1609 takeGivenInsolubles :: TcS Cts
1610 -- See Note [The inert set after solving Givens]
1611 takeGivenInsolubles
1612 = updRetInertCans $ \ cans ->
1613 ( inert_insols cans
1614 , cans { inert_insols = emptyBag
1615 , inert_funeqs = filterFunEqs isGivenCt (inert_funeqs cans) } )
1616
1617 {- Note [The inert set after solving Givens]
1618 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1619 After solving the Givens we take two things out of the inert set
1620
1621 a) The insolubles; we return these to report inaccessible code
1622 We return these separately. We don't want to leave them in
1623 the inert set, lest we confuse them with insolubles arising from
1624 solving wanteds
1625
1626 b) Any Derived CFunEqCans. Derived CTyEqCans are in the
1627 inert_model and do no harm. In contrast, Derived CFunEqCans
1628 get mixed up with the Wanteds later and confuse the
1629 post-solve-wanted unflattening (Trac #10507).
1630 E.g. From [G] 1 <= m, [G] m <= n
1631 We get [D] 1 <= n, and we must remove it!
1632 Otherwise we unflatten it more then once, and assign
1633 to its fmv more than once...disaster.
1634 It's ok to remove them because they turned not not to
1635 yield an insoluble, and hence have now done their work.
1636 -}
1637
1638 updRetInertCans :: (InertCans -> (a, InertCans)) -> TcS a
1639 -- Modify the inert set with the supplied function
1640 updRetInertCans upd_fn
1641 = do { is_var <- getTcSInertsRef
1642 ; wrapTcS (do { inerts <- TcM.readTcRef is_var
1643 ; let (res, cans') = upd_fn (inert_cans inerts)
1644 ; TcM.writeTcRef is_var (inerts { inert_cans = cans' })
1645 ; return res }) }
1646
1647 updInertCans :: (InertCans -> InertCans) -> TcS ()
1648 -- Modify the inert set with the supplied function
1649 updInertCans upd_fn
1650 = updInertTcS $ \ inerts -> inerts { inert_cans = upd_fn (inert_cans inerts) }
1651
1652 updInertDicts :: (DictMap Ct -> DictMap Ct) -> TcS ()
1653 -- Modify the inert set with the supplied function
1654 updInertDicts upd_fn
1655 = updInertCans $ \ ics -> ics { inert_dicts = upd_fn (inert_dicts ics) }
1656
1657 updInertSafehask :: (DictMap Ct -> DictMap Ct) -> TcS ()
1658 -- Modify the inert set with the supplied function
1659 updInertSafehask upd_fn
1660 = updInertCans $ \ ics -> ics { inert_safehask = upd_fn (inert_safehask ics) }
1661
1662 updInertFunEqs :: (FunEqMap Ct -> FunEqMap Ct) -> TcS ()
1663 -- Modify the inert set with the supplied function
1664 updInertFunEqs upd_fn
1665 = updInertCans $ \ ics -> ics { inert_funeqs = upd_fn (inert_funeqs ics) }
1666
1667 updInertIrreds :: (Cts -> Cts) -> TcS ()
1668 -- Modify the inert set with the supplied function
1669 updInertIrreds upd_fn
1670 = updInertCans $ \ ics -> ics { inert_irreds = upd_fn (inert_irreds ics) }
1671
1672 getInertEqs :: TcS (TyVarEnv EqualCtList)
1673 getInertEqs = do { inert <- getInertCans; return (inert_eqs inert) }
1674
1675 getInertModel :: TcS InertModel
1676 getInertModel = do { inert <- getInertCans; return (inert_model inert) }
1677
1678 getInertGivens :: TcS [Ct]
1679 -- Returns the Given constraints in the inert set,
1680 -- with type functions *not* unflattened
1681 getInertGivens
1682 = do { inerts <- getInertCans
1683 ; let all_cts = foldDicts (:) (inert_dicts inerts)
1684 $ foldFunEqs (:) (inert_funeqs inerts)
1685 $ concat (varEnvElts (inert_eqs inerts))
1686 ; return (filter isGivenCt all_cts) }
1687
1688 getPendingScDicts :: TcS [Ct]
1689 -- Find all inert Given dictionaries whose cc_pend_sc flag is True
1690 -- Set the flag to False in the inert set, and return that Ct
1691 getPendingScDicts = updRetInertCans get_sc_dicts
1692 where
1693 get_sc_dicts ic@(IC { inert_dicts = dicts })
1694 = (sc_pend_dicts, ic')
1695 where
1696 ic' = ic { inert_dicts = foldr add dicts sc_pend_dicts }
1697
1698 sc_pend_dicts :: [Ct]
1699 sc_pend_dicts = foldDicts get_pending dicts []
1700
1701 get_pending :: Ct -> [Ct] -> [Ct] -- Get dicts with cc_pend_sc = True
1702 -- but flipping the flag
1703 get_pending dict dicts
1704 | Just dict' <- isPendingScDict dict = dict' : dicts
1705 | otherwise = dicts
1706
1707 add :: Ct -> DictMap Ct -> DictMap Ct
1708 add ct@(CDictCan { cc_class = cls, cc_tyargs = tys }) dicts
1709 = addDict dicts cls tys ct
1710 add ct _ = pprPanic "getPendingScDicts" (ppr ct)
1711
1712 getUnsolvedInerts :: TcS ( Bag Implication
1713 , Cts -- Tyvar eqs: a ~ ty
1714 , Cts -- Fun eqs: F a ~ ty
1715 , Cts -- Insoluble
1716 , Cts ) -- All others
1717 -- Return all the unsolved [Wanted] or [Derived] constraints
1718 --
1719 -- Post-condition: the returned simple constraints are all fully zonked
1720 -- (because they come from the inert set)
1721 -- the unsolved implics may not be
1722 getUnsolvedInerts
1723 = do { IC { inert_eqs = tv_eqs
1724 , inert_funeqs = fun_eqs
1725 , inert_irreds = irreds
1726 , inert_dicts = idicts
1727 , inert_insols = insols
1728 , inert_model = model } <- getInertCans
1729 ; keep_derived <- keepSolvingDeriveds
1730
1731 ; let der_tv_eqs = foldVarEnv (add_der_eq keep_derived tv_eqs)
1732 emptyCts model
1733 unsolved_tv_eqs = foldTyEqs add_if_unsolved tv_eqs der_tv_eqs
1734 unsolved_fun_eqs = foldFunEqs add_if_unsolved fun_eqs emptyCts
1735 unsolved_irreds = Bag.filterBag is_unsolved irreds
1736 unsolved_dicts = foldDicts add_if_unsolved idicts emptyCts
1737 others = unsolved_irreds `unionBags` unsolved_dicts
1738
1739 ; implics <- getWorkListImplics
1740
1741 ; traceTcS "getUnsolvedInerts" $
1742 vcat [ text " tv eqs =" <+> ppr unsolved_tv_eqs
1743 , text "fun eqs =" <+> ppr unsolved_fun_eqs
1744 , text "insols =" <+> ppr insols
1745 , text "others =" <+> ppr others
1746 , text "implics =" <+> ppr implics ]
1747
1748 ; return ( implics, unsolved_tv_eqs, unsolved_fun_eqs, insols, others) }
1749 -- Keep even the given insolubles
1750 -- so that we can report dead GADT pattern match branches
1751 where
1752 add_der_eq keep_derived tv_eqs ct cts
1753 -- See Note [Unsolved Derived equalities]
1754 | CTyEqCan { cc_tyvar = tv, cc_rhs = rhs } <- ct
1755 , isMetaTyVar tv || keep_derived
1756 , not (isInInertEqs tv_eqs tv rhs) = ct `consBag` cts
1757 | otherwise = cts
1758 add_if_unsolved :: Ct -> Cts -> Cts
1759 add_if_unsolved ct cts | is_unsolved ct = ct `consCts` cts
1760 | otherwise = cts
1761
1762 is_unsolved ct = not (isGivenCt ct) -- Wanted or Derived
1763
1764 isInInertEqs :: TyVarEnv EqualCtList -> TcTyVar -> TcType -> Bool
1765 -- True if (a ~N ty) is in the inert set, in either Given or Wanted
1766 isInInertEqs eqs tv rhs
1767 = case lookupVarEnv eqs tv of
1768 Nothing -> False
1769 Just cts -> any (same_pred rhs) cts
1770 where
1771 same_pred rhs ct
1772 | CTyEqCan { cc_rhs = rhs2, cc_eq_rel = eq_rel } <- ct
1773 , NomEq <- eq_rel
1774 , rhs `eqType` rhs2 = True
1775 | otherwise = False
1776
1777 getNoGivenEqs :: TcLevel -- TcLevel of this implication
1778 -> [TcTyVar] -- Skolems of this implication
1779 -> TcS Bool -- True <=> definitely no residual given equalities
1780 -- See Note [When does an implication have given equalities?]
1781 getNoGivenEqs tclvl skol_tvs
1782 = do { inerts@(IC { inert_eqs = ieqs, inert_irreds = iirreds, inert_funeqs = funeqs })
1783 <- getInertCans
1784 ; let local_fsks = foldFunEqs add_fsk funeqs emptyVarSet
1785
1786 has_given_eqs = foldrBag ((||) . ev_given_here . ctEvidence) False iirreds
1787 || foldVarEnv ((||) . eqs_given_here local_fsks) False ieqs
1788
1789 ; traceTcS "getNoGivenEqs" (vcat [ppr has_given_eqs, ppr inerts])
1790 ; return (not has_given_eqs) }
1791 where
1792 eqs_given_here :: VarSet -> EqualCtList -> Bool
1793 eqs_given_here local_fsks [CTyEqCan { cc_tyvar = tv, cc_ev = ev }]
1794 -- Givens are always a sigleton
1795 = not (skolem_bound_here local_fsks tv) && ev_given_here ev
1796 eqs_given_here _ _ = False
1797
1798 ev_given_here :: CtEvidence -> Bool
1799 -- True for a Given bound by the curent implication,
1800 -- i.e. the current level
1801 ev_given_here ev
1802 = isGiven ev
1803 && tclvl == ctLocLevel (ctEvLoc ev)
1804
1805 add_fsk :: Ct -> VarSet -> VarSet
1806 add_fsk ct fsks | CFunEqCan { cc_fsk = tv, cc_ev = ev } <- ct
1807 , isGiven ev = extendVarSet fsks tv
1808 | otherwise = fsks
1809
1810 skol_tv_set = mkVarSet skol_tvs
1811 skolem_bound_here local_fsks tv -- See Note [Let-bound skolems]
1812 = case tcTyVarDetails tv of
1813 SkolemTv {} -> tv `elemVarSet` skol_tv_set
1814 FlatSkol {} -> not (tv `elemVarSet` local_fsks)
1815 _ -> False
1816
1817 -- | Returns Given constraints that might,
1818 -- potentially, match the given pred. This is used when checking to see if a
1819 -- Given might overlap with an instance. See Note [Instance and Given overlap]
1820 -- in TcInteract.
1821 matchableGivens :: CtLoc -> PredType -> InertSet -> Cts
1822 matchableGivens loc_w pred (IS { inert_cans = inert_cans })
1823 = filterBag matchable_given all_relevant_givens
1824 where
1825 -- just look in class constraints and irreds. matchableGivens does get called
1826 -- for ~R constraints, but we don't need to look through equalities, because
1827 -- canonical equalities are used for rewriting. We'll only get caught by
1828 -- non-canonical -- that is, irreducible -- equalities.
1829 all_relevant_givens :: Cts
1830 all_relevant_givens
1831 | Just (clas, _) <- getClassPredTys_maybe pred
1832 = findDictsByClass (inert_dicts inert_cans) clas
1833 `unionBags` inert_irreds inert_cans
1834 | otherwise
1835 = inert_irreds inert_cans
1836
1837 matchable_given :: Ct -> Bool
1838 matchable_given ct
1839 | CtGiven { ctev_loc = loc_g } <- ctev
1840 , Just _ <- tcUnifyTys bind_meta_tv [ctEvPred ctev] [pred]
1841 , not (prohibitedSuperClassSolve loc_g loc_w)
1842 = True
1843
1844 | otherwise
1845 = False
1846 where
1847 ctev = cc_ev ct
1848
1849 bind_meta_tv :: TcTyVar -> BindFlag
1850 -- Any meta tyvar may be unified later, so we treat it as
1851 -- bindable when unifying with givens. That ensures that we
1852 -- conservatively assume that a meta tyvar might get unified with
1853 -- something that matches the 'given', until demonstrated
1854 -- otherwise.
1855 bind_meta_tv tv | isMetaTyVar tv = BindMe
1856 | otherwise = Skolem
1857
1858 prohibitedSuperClassSolve :: CtLoc -> CtLoc -> Bool
1859 -- See Note [Solving superclass constraints] in TcInstDcls
1860 prohibitedSuperClassSolve from_loc solve_loc
1861 | GivenOrigin (InstSC given_size) <- ctLocOrigin from_loc
1862 , ScOrigin wanted_size <- ctLocOrigin solve_loc
1863 = given_size >= wanted_size
1864 | otherwise
1865 = False
1866
1867 {- Note [Unsolved Derived equalities]
1868 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1869 In getUnsolvedInerts, we return a derived equality from the model
1870 for two possible reasons:
1871
1872 * Because it is a candidate for floating out of this implication.
1873 We only float equalities with a meta-tyvar on the left, so we only
1874 pull those out here.
1875
1876 * If we are only solving derived constraints (i.e. tcs_need_derived
1877 is true; see Note [Solving for Derived constraints]), then we
1878 those Derived constraints are effectively unsolved, and we need
1879 them!
1880
1881 Note [When does an implication have given equalities?]
1882 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1883 Consider an implication
1884 beta => alpha ~ Int
1885 where beta is a unification variable that has already been unified
1886 to () in an outer scope. Then we can float the (alpha ~ Int) out
1887 just fine. So when deciding whether the givens contain an equality,
1888 we should canonicalise first, rather than just looking at the original
1889 givens (Trac #8644).
1890
1891 So we simply look at the inert, canonical Givens and see if there are
1892 any equalities among them, the calculation of has_given_eqs. There
1893 are some wrinkles:
1894
1895 * We must know which ones are bound in *this* implication and which
1896 are bound further out. We can find that out from the TcLevel
1897 of the Given, which is itself recorded in the tcl_tclvl field
1898 of the TcLclEnv stored in the Given (ev_given_here).
1899
1900 What about interactions between inner and outer givens?
1901 - Outer given is rewritten by an inner given, then there must
1902 have been an inner given equality, hence the “given-eq” flag
1903 will be true anyway.
1904
1905 - Inner given rewritten by outer, retains its level (ie. The inner one)
1906
1907 * We must take account of *potential* equalities, like the one above:
1908 beta => ...blah...
1909 If we still don't know what beta is, we conservatively treat it as potentially
1910 becoming an equality. Hence including 'irreds' in the calculation or has_given_eqs.
1911
1912 * When flattening givens, we generate Given equalities like
1913 <F [a]> : F [a] ~ f,
1914 with Refl evidence, and we *don't* want those to count as an equality
1915 in the givens! After all, the entire flattening business is just an
1916 internal matter, and the evidence does not mention any of the 'givens'
1917 of this implication. So we do not treat inert_funeqs as a 'given equality'.
1918
1919 * See Note [Let-bound skolems] for another wrinkle
1920
1921 * We do *not* need to worry about representational equalities, because
1922 these do not affect the ability to float constraints.
1923
1924 Note [Let-bound skolems]
1925 ~~~~~~~~~~~~~~~~~~~~~~~~
1926 If * the inert set contains a canonical Given CTyEqCan (a ~ ty)
1927 and * 'a' is a skolem bound in this very implication, b
1928
1929 then:
1930 a) The Given is pretty much a let-binding, like
1931 f :: (a ~ b->c) => a -> a
1932 Here the equality constraint is like saying
1933 let a = b->c in ...
1934 It is not adding any new, local equality information,
1935 and hence can be ignored by has_given_eqs
1936
1937 b) 'a' will have been completely substituted out in the inert set,
1938 so we can safely discard it. Notably, it doesn't need to be
1939 returned as part of 'fsks'
1940
1941 For an example, see Trac #9211.
1942 -}
1943
1944 removeInertCts :: [Ct] -> InertCans -> InertCans
1945 -- ^ Remove inert constraints from the 'InertCans', for use when a
1946 -- typechecker plugin wishes to discard a given.
1947 removeInertCts cts icans = foldl' removeInertCt icans cts
1948
1949 removeInertCt :: InertCans -> Ct -> InertCans
1950 removeInertCt is ct =
1951 case ct of
1952
1953 CDictCan { cc_class = cl, cc_tyargs = tys } ->
1954 is { inert_dicts = delDict (inert_dicts is) cl tys }
1955
1956 CFunEqCan { cc_fun = tf, cc_tyargs = tys } ->
1957 is { inert_funeqs = delFunEq (inert_funeqs is) tf tys }
1958
1959 CTyEqCan { cc_tyvar = x, cc_rhs = ty } ->
1960 is { inert_eqs = delTyEq (inert_eqs is) x ty }
1961
1962 CIrredEvCan {} -> panic "removeInertCt: CIrredEvCan"
1963 CNonCanonical {} -> panic "removeInertCt: CNonCanonical"
1964 CHoleCan {} -> panic "removeInertCt: CHoleCan"
1965
1966
1967 lookupFlatCache :: TyCon -> [Type] -> TcS (Maybe (TcCoercion, TcType, CtFlavour))
1968 lookupFlatCache fam_tc tys
1969 = do { IS { inert_flat_cache = flat_cache
1970 , inert_cans = IC { inert_funeqs = inert_funeqs } } <- getTcSInerts
1971 ; return (firstJusts [lookup_inerts inert_funeqs,
1972 lookup_flats flat_cache]) }
1973 where
1974 lookup_inerts inert_funeqs
1975 | Just (CFunEqCan { cc_ev = ctev, cc_fsk = fsk, cc_tyargs = xis })
1976 <- findFunEq inert_funeqs fam_tc tys
1977 , tys `eqTypes` xis -- the lookup might find a near-match; see
1978 -- Note [Use loose types in inert set]
1979 = Just (ctEvCoercion ctev, mkTyVarTy fsk, ctEvFlavour ctev)
1980 | otherwise = Nothing
1981
1982 lookup_flats flat_cache = findExactFunEq flat_cache fam_tc tys
1983
1984
1985 lookupInInerts :: TcPredType -> TcS (Maybe CtEvidence)
1986 -- Is this exact predicate type cached in the solved or canonicals of the InertSet?
1987 lookupInInerts pty
1988 | ClassPred cls tys <- classifyPredType pty
1989 = do { inerts <- getTcSInerts
1990 ; return (lookupSolvedDict inerts cls tys `mplus`
1991 lookupInertDict (inert_cans inerts) cls tys) }
1992 | otherwise -- NB: No caching for equalities, IPs, holes, or errors
1993 = return Nothing
1994
1995 -- | Look up a dictionary inert. NB: the returned 'CtEvidence' might not
1996 -- match the input exactly. Note [Use loose types in inert set].
1997 lookupInertDict :: InertCans -> Class -> [Type] -> Maybe CtEvidence
1998 lookupInertDict (IC { inert_dicts = dicts }) cls tys
1999 = case findDict dicts cls tys of
2000 Just ct -> Just (ctEvidence ct)
2001 _ -> Nothing
2002
2003 -- | Look up a solved inert. NB: the returned 'CtEvidence' might not
2004 -- match the input exactly. See Note [Use loose types in inert set].
2005 lookupSolvedDict :: InertSet -> Class -> [Type] -> Maybe CtEvidence
2006 -- Returns just if exactly this predicate type exists in the solved.
2007 lookupSolvedDict (IS { inert_solved_dicts = solved }) cls tys
2008 = case findDict solved cls tys of
2009 Just ev -> Just ev
2010 _ -> Nothing
2011
2012 {- *********************************************************************
2013 * *
2014 Irreds
2015 * *
2016 ********************************************************************* -}
2017
2018 foldIrreds :: (Ct -> b -> b) -> Cts -> b -> b
2019 foldIrreds k irreds z = foldrBag k z irreds
2020
2021
2022 {- *********************************************************************
2023 * *
2024 Type equalities
2025 * *
2026 ********************************************************************* -}
2027
2028 type EqualCtList = [Ct]
2029
2030 {- Note [EqualCtList invariants]
2031 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2032 * All are equalities
2033 * All these equalities have the same LHS
2034 * The list is never empty
2035 * No element of the list can rewrite any other
2036
2037 From the fourth invariant it follows that the list is
2038 - A single Given, or
2039 - Any number of Wanteds and/or Deriveds
2040 -}
2041
2042 addTyEq :: TyVarEnv EqualCtList -> TcTyVar -> Ct -> TyVarEnv EqualCtList
2043 addTyEq old_list tv it = extendVarEnv_C (\old_eqs _new_eqs -> it : old_eqs)
2044 old_list tv [it]
2045
2046 foldTyEqs :: (Ct -> b -> b) -> TyVarEnv EqualCtList -> b -> b
2047 foldTyEqs k eqs z
2048 = foldVarEnv (\cts z -> foldr k z cts) z eqs
2049
2050 findTyEqs :: InertCans -> TyVar -> EqualCtList
2051 findTyEqs icans tv = lookupVarEnv (inert_eqs icans) tv `orElse` []
2052
2053 delTyEq :: TyVarEnv EqualCtList -> TcTyVar -> TcType -> TyVarEnv EqualCtList
2054 delTyEq m tv t = modifyVarEnv (filter (not . isThisOne)) m tv
2055 where isThisOne (CTyEqCan { cc_rhs = t1 }) = eqType t t1
2056 isThisOne _ = False
2057
2058 {- *********************************************************************
2059 * *
2060 TcAppMap
2061 * *
2062 ************************************************************************
2063
2064 Note [Use loose types in inert set]
2065 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2066 Say we know (Eq (a |> c1)) and we need (Eq (a |> c2)). One is clearly
2067 solvable from the other. So, we do lookup in the inert set using
2068 loose types, which omit the kind-check.
2069
2070 We must be careful when using the result of a lookup because it may
2071 not match the requsted info exactly!
2072
2073 -}
2074
2075 type TcAppMap a = UniqDFM (ListMap LooseTypeMap a)
2076 -- Indexed by tycon then the arg types, using "loose" matching, where
2077 -- we don't require kind equality. This allows, for example, (a |> co)
2078 -- to match (a).
2079 -- See Note [Use loose types in inert set]
2080 -- Used for types and classes; hence UniqDFM
2081 -- See Note [foldTM determinism] for why we use UniqDFM here
2082
2083 isEmptyTcAppMap :: TcAppMap a -> Bool
2084 isEmptyTcAppMap m = isNullUDFM m
2085
2086 emptyTcAppMap :: TcAppMap a
2087 emptyTcAppMap = emptyUDFM
2088
2089 findTcApp :: TcAppMap a -> Unique -> [Type] -> Maybe a
2090 findTcApp m u tys = do { tys_map <- lookupUDFM m u
2091 ; lookupTM tys tys_map }
2092
2093 delTcApp :: TcAppMap a -> Unique -> [Type] -> TcAppMap a
2094 delTcApp m cls tys = adjustUDFM (deleteTM tys) m cls
2095
2096 insertTcApp :: TcAppMap a -> Unique -> [Type] -> a -> TcAppMap a
2097 insertTcApp m cls tys ct = alterUDFM alter_tm m cls
2098 where
2099 alter_tm mb_tm = Just (insertTM tys ct (mb_tm `orElse` emptyTM))
2100
2101 -- mapTcApp :: (a->b) -> TcAppMap a -> TcAppMap b
2102 -- mapTcApp f = mapUDFM (mapTM f)
2103
2104 filterTcAppMap :: (Ct -> Bool) -> TcAppMap Ct -> TcAppMap Ct
2105 filterTcAppMap f m
2106 = mapUDFM do_tm m
2107 where
2108 do_tm tm = foldTM insert_mb tm emptyTM
2109 insert_mb ct tm
2110 | f ct = insertTM tys ct tm
2111 | otherwise = tm
2112 where
2113 tys = case ct of
2114 CFunEqCan { cc_tyargs = tys } -> tys
2115 CDictCan { cc_tyargs = tys } -> tys
2116 _ -> pprPanic "filterTcAppMap" (ppr ct)
2117
2118 tcAppMapToBag :: TcAppMap a -> Bag a
2119 tcAppMapToBag m = foldTcAppMap consBag m emptyBag
2120
2121 foldTcAppMap :: (a -> b -> b) -> TcAppMap a -> b -> b
2122 foldTcAppMap k m z = foldUDFM (foldTM k) z m
2123
2124
2125 {- *********************************************************************
2126 * *
2127 DictMap
2128 * *
2129 ********************************************************************* -}
2130
2131 type DictMap a = TcAppMap a
2132
2133 emptyDictMap :: DictMap a
2134 emptyDictMap = emptyTcAppMap
2135
2136 -- sizeDictMap :: DictMap a -> Int
2137 -- sizeDictMap m = foldDicts (\ _ x -> x+1) m 0
2138
2139 findDict :: DictMap a -> Class -> [Type] -> Maybe a
2140 findDict m cls tys = findTcApp m (getUnique cls) tys
2141
2142 findDictsByClass :: DictMap a -> Class -> Bag a
2143 findDictsByClass m cls
2144 | Just tm <- lookupUDFM m cls = foldTM consBag tm emptyBag
2145 | otherwise = emptyBag
2146
2147 delDict :: DictMap a -> Class -> [Type] -> DictMap a
2148 delDict m cls tys = delTcApp m (getUnique cls) tys
2149
2150 addDict :: DictMap a -> Class -> [Type] -> a -> DictMap a
2151 addDict m cls tys item = insertTcApp m (getUnique cls) tys item
2152
2153 addDictsByClass :: DictMap Ct -> Class -> Bag Ct -> DictMap Ct
2154 addDictsByClass m cls items
2155 = addToUDFM m cls (foldrBag add emptyTM items)
2156 where
2157 add ct@(CDictCan { cc_tyargs = tys }) tm = insertTM tys ct tm
2158 add ct _ = pprPanic "addDictsByClass" (ppr ct)
2159
2160 filterDicts :: (Ct -> Bool) -> DictMap Ct -> DictMap Ct
2161 filterDicts f m = filterTcAppMap f m
2162
2163 partitionDicts :: (Ct -> Bool) -> DictMap Ct -> (Bag Ct, DictMap Ct)
2164 partitionDicts f m = foldTcAppMap k m (emptyBag, emptyDicts)
2165 where
2166 k ct (yeses, noes) | f ct = (ct `consBag` yeses, noes)
2167 | otherwise = (yeses, add ct noes)
2168 add ct@(CDictCan { cc_class = cls, cc_tyargs = tys }) m
2169 = addDict m cls tys ct
2170 add ct _ = pprPanic "partitionDicts" (ppr ct)
2171
2172 dictsToBag :: DictMap a -> Bag a
2173 dictsToBag = tcAppMapToBag
2174
2175 foldDicts :: (a -> b -> b) -> DictMap a -> b -> b
2176 foldDicts = foldTcAppMap
2177
2178 emptyDicts :: DictMap a
2179 emptyDicts = emptyTcAppMap
2180
2181
2182 {- *********************************************************************
2183 * *
2184 FunEqMap
2185 * *
2186 ********************************************************************* -}
2187
2188 type FunEqMap a = TcAppMap a -- A map whose key is a (TyCon, [Type]) pair
2189
2190 emptyFunEqs :: TcAppMap a
2191 emptyFunEqs = emptyTcAppMap
2192
2193 sizeFunEqMap :: FunEqMap a -> Int
2194 sizeFunEqMap m = foldFunEqs (\ _ x -> x+1) m 0
2195
2196 findFunEq :: FunEqMap a -> TyCon -> [Type] -> Maybe a
2197 findFunEq m tc tys = findTcApp m (getUnique tc) tys
2198
2199 funEqsToBag :: FunEqMap a -> Bag a
2200 funEqsToBag m = foldTcAppMap consBag m emptyBag
2201
2202 findFunEqsByTyCon :: FunEqMap a -> TyCon -> [a]
2203 -- Get inert function equation constraints that have the given tycon
2204 -- in their head. Not that the constraints remain in the inert set.
2205 -- We use this to check for derived interactions with built-in type-function
2206 -- constructors.
2207 findFunEqsByTyCon m tc
2208 | Just tm <- lookupUDFM m tc = foldTM (:) tm []
2209 | otherwise = []
2210
2211 foldFunEqs :: (a -> b -> b) -> FunEqMap a -> b -> b
2212 foldFunEqs = foldTcAppMap
2213
2214 -- mapFunEqs :: (a -> b) -> FunEqMap a -> FunEqMap b
2215 -- mapFunEqs = mapTcApp
2216
2217 filterFunEqs :: (Ct -> Bool) -> FunEqMap Ct -> FunEqMap Ct
2218 filterFunEqs = filterTcAppMap
2219
2220 insertFunEq :: FunEqMap a -> TyCon -> [Type] -> a -> FunEqMap a
2221 insertFunEq m tc tys val = insertTcApp m (getUnique tc) tys val
2222
2223 -- insertFunEqCt :: FunEqMap Ct -> Ct -> FunEqMap Ct
2224 -- insertFunEqCt m ct@(CFunEqCan { cc_fun = tc, cc_tyargs = tys })
2225 -- = insertFunEq m tc tys ct
2226 -- insertFunEqCt _ ct = pprPanic "insertFunEqCt" (ppr ct)
2227
2228 partitionFunEqs :: (Ct -> Bool) -> FunEqMap Ct -> ([Ct], FunEqMap Ct)
2229 -- Optimise for the case where the predicate is false
2230 -- partitionFunEqs is called only from kick-out, and kick-out usually
2231 -- kicks out very few equalities, so we want to optimise for that case
2232 partitionFunEqs f m = (yeses, foldr del m yeses)
2233 where
2234 yeses = foldTcAppMap k m []
2235 k ct yeses | f ct = ct : yeses
2236 | otherwise = yeses
2237 del (CFunEqCan { cc_fun = tc, cc_tyargs = tys }) m
2238 = delFunEq m tc tys
2239 del ct _ = pprPanic "partitionFunEqs" (ppr ct)
2240
2241 delFunEq :: FunEqMap a -> TyCon -> [Type] -> FunEqMap a
2242 delFunEq m tc tys = delTcApp m (getUnique tc) tys
2243
2244 ------------------------------
2245 type ExactFunEqMap a = UniqFM (ListMap TypeMap a)
2246
2247 emptyExactFunEqs :: ExactFunEqMap a
2248 emptyExactFunEqs = emptyUFM
2249
2250 findExactFunEq :: ExactFunEqMap a -> TyCon -> [Type] -> Maybe a
2251 findExactFunEq m tc tys = do { tys_map <- lookupUFM m (getUnique tc)
2252 ; lookupTM tys tys_map }
2253
2254 insertExactFunEq :: ExactFunEqMap a -> TyCon -> [Type] -> a -> ExactFunEqMap a
2255 insertExactFunEq m tc tys val = alterUFM alter_tm m (getUnique tc)
2256 where alter_tm mb_tm = Just (insertTM tys val (mb_tm `orElse` emptyTM))
2257
2258 {-
2259 ************************************************************************
2260 * *
2261 * The TcS solver monad *
2262 * *
2263 ************************************************************************
2264
2265 Note [The TcS monad]
2266 ~~~~~~~~~~~~~~~~~~~~
2267 The TcS monad is a weak form of the main Tc monad
2268
2269 All you can do is
2270 * fail
2271 * allocate new variables
2272 * fill in evidence variables
2273
2274 Filling in a dictionary evidence variable means to create a binding
2275 for it, so TcS carries a mutable location where the binding can be
2276 added. This is initialised from the innermost implication constraint.
2277
2278 Note [Solving for Derived constraints]
2279 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2280 Sometimes we invoke the solver on a bunch of Derived constraints, not to
2281 generate any evidence, but just to cause unification side effects or to
2282 produce a simpler set of constraints. If that is what we are doing, we
2283 should do two things differently:
2284 a) Don't stop when you've solved all the Wanteds; instead keep going
2285 if there are any Deriveds in the work queue.
2286 b) In getInertUnsolved, include Derived ones
2287 -}
2288
2289 data TcSEnv
2290 = TcSEnv {
2291 tcs_ev_binds :: Maybe EvBindsVar,
2292 -- this could be Nothing if we can't deal with non-equality
2293 -- constraints, because, say, we're in a top-level type signature
2294
2295 tcs_unified :: IORef Int,
2296 -- The number of unification variables we have filled
2297 -- The important thing is whether it is non-zero
2298
2299 tcs_count :: IORef Int, -- Global step count
2300
2301 tcs_inerts :: IORef InertSet, -- Current inert set
2302
2303 -- The main work-list and the flattening worklist
2304 -- See Note [Work list priorities] and
2305 tcs_worklist :: IORef WorkList, -- Current worklist
2306
2307 tcs_used_tcvs :: IORef TyCoVarSet,
2308 -- these variables were used when filling holes. Don't discard!
2309 -- See also Note [Tracking redundant constraints] in TcSimplify
2310
2311 tcs_need_deriveds :: Bool
2312 -- Keep solving, even if all the unsolved constraints are Derived
2313 -- See Note [Solving for Derived constraints]
2314 }
2315
2316 ---------------
2317 newtype TcS a = TcS { unTcS :: TcSEnv -> TcM a }
2318
2319 instance Functor TcS where
2320 fmap f m = TcS $ fmap f . unTcS m
2321
2322 instance Applicative TcS where
2323 pure x = TcS (\_ -> return x)
2324 (<*>) = ap
2325
2326 instance Monad TcS where
2327 fail err = TcS (\_ -> fail err)
2328 m >>= k = TcS (\ebs -> unTcS m ebs >>= \r -> unTcS (k r) ebs)
2329
2330 #if __GLASGOW_HASKELL__ > 710
2331 instance MonadFail.MonadFail TcS where
2332 fail err = TcS (\_ -> fail err)
2333 #endif
2334
2335 instance MonadUnique TcS where
2336 getUniqueSupplyM = wrapTcS getUniqueSupplyM
2337
2338 -- Basic functionality
2339 -- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2340 wrapTcS :: TcM a -> TcS a
2341 -- Do not export wrapTcS, because it promotes an arbitrary TcM to TcS,
2342 -- and TcS is supposed to have limited functionality
2343 wrapTcS = TcS . const -- a TcM action will not use the TcEvBinds
2344
2345 wrapErrTcS :: TcM a -> TcS a
2346 -- The thing wrapped should just fail
2347 -- There's no static check; it's up to the user
2348 -- Having a variant for each error message is too painful
2349 wrapErrTcS = wrapTcS
2350
2351 wrapWarnTcS :: TcM a -> TcS a
2352 -- The thing wrapped should just add a warning, or no-op
2353 -- There's no static check; it's up to the user
2354 wrapWarnTcS = wrapTcS
2355
2356 failTcS, panicTcS :: SDoc -> TcS a
2357 warnTcS :: WarningFlag -> SDoc -> TcS ()
2358 addErrTcS :: SDoc -> TcS ()
2359 failTcS = wrapTcS . TcM.failWith
2360 warnTcS flag = wrapTcS . TcM.addWarn (Reason flag)
2361 addErrTcS = wrapTcS . TcM.addErr
2362 panicTcS doc = pprPanic "TcCanonical" doc
2363
2364 traceTcS :: String -> SDoc -> TcS ()
2365 traceTcS herald doc = wrapTcS (TcM.traceTc herald doc)
2366
2367 runTcPluginTcS :: TcPluginM a -> TcS a
2368 runTcPluginTcS m = wrapTcS . runTcPluginM m =<< getTcEvBinds
2369
2370 instance HasDynFlags TcS where
2371 getDynFlags = wrapTcS getDynFlags
2372
2373 getGlobalRdrEnvTcS :: TcS GlobalRdrEnv
2374 getGlobalRdrEnvTcS = wrapTcS TcM.getGlobalRdrEnv
2375
2376 bumpStepCountTcS :: TcS ()
2377 bumpStepCountTcS = TcS $ \env -> do { let ref = tcs_count env
2378 ; n <- TcM.readTcRef ref
2379 ; TcM.writeTcRef ref (n+1) }
2380
2381 -- | Mark variables as used filling a coercion hole
2382 useVars :: TyCoVarSet -> TcS ()
2383 useVars vars = TcS $ \env -> do { let ref = tcs_used_tcvs env
2384 ; TcM.updTcRef ref (`unionVarSet` vars) }
2385
2386 csTraceTcS :: SDoc -> TcS ()
2387 csTraceTcS doc
2388 = wrapTcS $ csTraceTcM 1 (return doc)
2389
2390 traceFireTcS :: CtEvidence -> SDoc -> TcS ()
2391 -- Dump a rule-firing trace
2392 traceFireTcS ev doc
2393 = TcS $ \env -> csTraceTcM 1 $
2394 do { n <- TcM.readTcRef (tcs_count env)
2395 ; tclvl <- TcM.getTcLevel
2396 ; return (hang (int n <> brackets (text "U:" <> ppr tclvl
2397 <> ppr (ctLocDepth (ctEvLoc ev)))
2398 <+> doc <> colon)
2399 4 (ppr ev)) }
2400
2401 csTraceTcM :: Int -> TcM SDoc -> TcM ()
2402 -- Constraint-solver tracing, -ddump-cs-trace
2403 csTraceTcM trace_level mk_doc
2404 = do { dflags <- getDynFlags
2405 ; when ( (dopt Opt_D_dump_cs_trace dflags || dopt Opt_D_dump_tc_trace dflags)
2406 && trace_level <= traceLevel dflags ) $
2407 do { msg <- mk_doc
2408 ; TcM.traceTcRn Opt_D_dump_cs_trace msg } }
2409
2410 runTcS :: TcS a -- What to run
2411 -> TcM (a, EvBindMap)
2412 runTcS tcs
2413 = do { ev_binds_var <- TcM.newTcEvBinds
2414 ; res <- runTcSWithEvBinds False (Just ev_binds_var) tcs
2415 ; ev_binds <- TcM.getTcEvBindsMap ev_binds_var
2416 ; return (res, ev_binds) }
2417
2418 -- | This variant of 'runTcS' will keep solving, even when only Deriveds
2419 -- are left around. It also doesn't return any evidence, as callers won't
2420 -- need it.
2421 runTcSDeriveds :: TcS a -> TcM a
2422 runTcSDeriveds tcs
2423 = do { ev_binds_var <- TcM.newTcEvBinds
2424 ; runTcSWithEvBinds True (Just ev_binds_var) tcs }
2425
2426 -- | This can deal only with equality constraints.
2427 runTcSEqualities :: TcS a -> TcM a
2428 runTcSEqualities = runTcSWithEvBinds False Nothing
2429
2430 runTcSWithEvBinds :: Bool -- ^ keep running even if only Deriveds are left?
2431 -> Maybe EvBindsVar
2432 -> TcS a
2433 -> TcM a
2434 runTcSWithEvBinds solve_deriveds ev_binds_var tcs
2435 = do { unified_var <- TcM.newTcRef 0
2436 ; step_count <- TcM.newTcRef 0
2437 ; inert_var <- TcM.newTcRef emptyInert
2438 ; wl_var <- TcM.newTcRef emptyWorkList
2439 ; used_var <- TcM.newTcRef emptyVarSet -- never read from, but see
2440 -- nestImplicTcS
2441
2442 ; let env = TcSEnv { tcs_ev_binds = ev_binds_var
2443 , tcs_unified = unified_var
2444 , tcs_count = step_count
2445 , tcs_inerts = inert_var
2446 , tcs_worklist = wl_var
2447 , tcs_used_tcvs = used_var
2448 , tcs_need_deriveds = solve_deriveds }
2449
2450 -- Run the computation
2451 ; res <- unTcS tcs env
2452
2453 ; count <- TcM.readTcRef step_count
2454 ; when (count > 0) $
2455 csTraceTcM 0 $ return (text "Constraint solver steps =" <+> int count)
2456
2457 #ifdef DEBUG
2458 ; whenIsJust ev_binds_var $ \ebv ->
2459 do { ev_binds <- TcM.getTcEvBinds ebv
2460 ; checkForCyclicBinds ev_binds }
2461 #endif
2462
2463 ; return res }
2464
2465 #ifdef DEBUG
2466 checkForCyclicBinds :: Bag EvBind -> TcM ()
2467 checkForCyclicBinds ev_binds
2468 | null cycles
2469 = return ()
2470 | null coercion_cycles
2471 = TcM.traceTc "Cycle in evidence binds" $ ppr cycles
2472 | otherwise
2473 = pprPanic "Cycle in coercion bindings" $ ppr coercion_cycles
2474 where
2475 cycles :: [[EvBind]]
2476 cycles = [c | CyclicSCC c <- stronglyConnCompFromEdgedVertices edges]
2477
2478 coercion_cycles = [c | c <- cycles, any is_co_bind c]
2479 is_co_bind (EvBind { eb_lhs = b }) = isEqPred (varType b)
2480
2481 edges :: [(EvBind, EvVar, [EvVar])]
2482 edges = [ (bind, bndr, varSetElems (evVarsOfTerm rhs))
2483 | bind@(EvBind { eb_lhs = bndr, eb_rhs = rhs}) <- bagToList ev_binds ]
2484 #endif
2485
2486 nestImplicTcS :: Maybe EvBindsVar -> TyCoVarSet -- bound in this implication
2487 -> TcLevel -> TcS a
2488 -> TcS (a, TyCoVarSet) -- also returns any vars used when filling
2489 -- coercion holes (for redundant-constraint
2490 -- tracking)
2491 nestImplicTcS m_ref bound_tcvs inner_tclvl (TcS thing_inside)
2492 = do { (res, used_tcvs) <-
2493 TcS $ \ TcSEnv { tcs_unified = unified_var
2494 , tcs_inerts = old_inert_var
2495 , tcs_count = count
2496 , tcs_need_deriveds = solve_deriveds
2497 } ->
2498 do { inerts <- TcM.readTcRef old_inert_var
2499 ; let nest_inert = inerts { inert_flat_cache = emptyExactFunEqs }
2500 -- See Note [Do not inherit the flat cache]
2501 ; new_inert_var <- TcM.newTcRef nest_inert
2502 ; new_wl_var <- TcM.newTcRef emptyWorkList
2503 ; new_used_var <- TcM.newTcRef emptyVarSet
2504 ; let nest_env = TcSEnv { tcs_ev_binds = m_ref
2505 , tcs_unified = unified_var
2506 , tcs_count = count
2507 , tcs_inerts = new_inert_var
2508 , tcs_worklist = new_wl_var
2509 , tcs_used_tcvs = new_used_var
2510 , tcs_need_deriveds = solve_deriveds }
2511 ; res <- TcM.setTcLevel inner_tclvl $
2512 thing_inside nest_env
2513
2514 #ifdef DEBUG
2515 -- Perform a check that the thing_inside did not cause cycles
2516 ; whenIsJust m_ref $ \ ref ->
2517 do { ev_binds <- TcM.getTcEvBinds ref
2518 ; checkForCyclicBinds ev_binds }
2519 #endif
2520 ; used_tcvs <- TcM.readTcRef new_used_var
2521 ; return (res, used_tcvs) }
2522
2523 ; local_ev_vars <- case m_ref of
2524 Nothing -> return emptyVarSet
2525 Just ref -> do { binds <- wrapTcS $ TcM.getTcEvBinds ref
2526 ; return $ mkVarSet $ map evBindVar $ bagToList binds }
2527 ; let all_locals = bound_tcvs `unionVarSet` local_ev_vars
2528 (inner_used_tcvs, outer_used_tcvs)
2529 = partitionVarSet (`elemVarSet` all_locals) used_tcvs
2530 ; useVars outer_used_tcvs
2531
2532 ; return (res, inner_used_tcvs) }
2533
2534 {- Note [Do not inherit the flat cache]
2535 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2536 We do not want to inherit the flat cache when processing nested
2537 implications. Consider
2538 a ~ F b, forall c. b~Int => blah
2539 If we have F b ~ fsk in the flat-cache, and we push that into the
2540 nested implication, we might miss that F b can be rewritten to F Int,
2541 and hence perhpas solve it. Moreover, the fsk from outside is
2542 flattened out after solving the outer level, but and we don't
2543 do that flattening recursively.
2544 -}
2545
2546 nestTcS :: TcS a -> TcS a
2547 -- Use the current untouchables, augmenting the current
2548 -- evidence bindings, and solved dictionaries
2549 -- But have no effect on the InertCans, or on the inert_flat_cache
2550 -- (the latter because the thing inside a nestTcS does unflattening)
2551 nestTcS (TcS thing_inside)
2552 = TcS $ \ env@(TcSEnv { tcs_inerts = inerts_var }) ->
2553 do { inerts <- TcM.readTcRef inerts_var
2554 ; new_inert_var <- TcM.newTcRef inerts
2555 ; new_wl_var <- TcM.newTcRef emptyWorkList
2556 ; let nest_env = env { tcs_inerts = new_inert_var
2557 , tcs_worklist = new_wl_var }
2558
2559 ; res <- thing_inside nest_env
2560
2561 ; new_inerts <- TcM.readTcRef new_inert_var
2562
2563 -- we want to propogate the safe haskell failures
2564 ; let old_ic = inert_cans inerts
2565 new_ic = inert_cans new_inerts
2566 nxt_ic = old_ic { inert_safehask = inert_safehask new_ic }
2567
2568 ; TcM.writeTcRef inerts_var -- See Note [Propagate the solved dictionaries]
2569 (inerts { inert_solved_dicts = inert_solved_dicts new_inerts
2570 , inert_cans = nxt_ic })
2571
2572 ; return res }
2573
2574 {-
2575 Note [Propagate the solved dictionaries]
2576 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2577 It's really quite important that nestTcS does not discard the solved
2578 dictionaries from the thing_inside.
2579 Consider
2580 Eq [a]
2581 forall b. empty => Eq [a]
2582 We solve the simple (Eq [a]), under nestTcS, and then turn our attention to
2583 the implications. It's definitely fine to use the solved dictionaries on
2584 the inner implications, and it can make a signficant performance difference
2585 if you do so.
2586 -}
2587
2588 -- Getters and setters of TcEnv fields
2589 -- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2590
2591 -- Getter of inerts and worklist
2592 getTcSInertsRef :: TcS (IORef InertSet)
2593 getTcSInertsRef = TcS (return . tcs_inerts)
2594
2595 getTcSWorkListRef :: TcS (IORef WorkList)
2596 getTcSWorkListRef = TcS (return . tcs_worklist)
2597
2598 getTcSInerts :: TcS InertSet
2599 getTcSInerts = getTcSInertsRef >>= wrapTcS . (TcM.readTcRef)
2600
2601 setTcSInerts :: InertSet -> TcS ()
2602 setTcSInerts ics = do { r <- getTcSInertsRef; wrapTcS (TcM.writeTcRef r ics) }
2603
2604 getWorkListImplics :: TcS (Bag Implication)
2605 getWorkListImplics
2606 = do { wl_var <- getTcSWorkListRef
2607 ; wl_curr <- wrapTcS (TcM.readTcRef wl_var)
2608 ; return (wl_implics wl_curr) }
2609
2610 updWorkListTcS :: (WorkList -> WorkList) -> TcS ()
2611 updWorkListTcS f
2612 = do { wl_var <- getTcSWorkListRef
2613 ; wl_curr <- wrapTcS (TcM.readTcRef wl_var)
2614 ; let new_work = f wl_curr
2615 ; wrapTcS (TcM.writeTcRef wl_var new_work) }
2616
2617 -- | Should we keep solving even only deriveds are left?
2618 keepSolvingDeriveds :: TcS Bool
2619 keepSolvingDeriveds = TcS (return . tcs_need_deriveds)
2620
2621 emitWorkNC :: [CtEvidence] -> TcS ()
2622 emitWorkNC evs
2623 | null evs
2624 = return ()
2625 | otherwise
2626 = emitWork (map mkNonCanonical evs)
2627
2628 emitWork :: [Ct] -> TcS ()
2629 emitWork cts
2630 = do { traceTcS "Emitting fresh work" (vcat (map ppr cts))
2631 ; updWorkListTcS (extendWorkListCts cts) }
2632
2633 emitInsoluble :: Ct -> TcS ()
2634 -- Emits a non-canonical constraint that will stand for a frozen error in the inerts.
2635 emitInsoluble ct
2636 = do { traceTcS "Emit insoluble" (ppr ct $$ pprCtLoc (ctLoc ct))
2637 ; updInertTcS add_insol }
2638 where
2639 this_pred = ctPred ct
2640 add_insol is@(IS { inert_cans = ics@(IC { inert_insols = old_insols }) })
2641 | already_there = is
2642 | otherwise = is { inert_cans = ics { inert_insols = old_insols `snocCts` ct } }
2643 where
2644 already_there = not (isWantedCt ct) && anyBag (tcEqType this_pred . ctPred) old_insols
2645 -- See Note [Do not add duplicate derived insolubles]
2646
2647 newTcRef :: a -> TcS (TcRef a)
2648 newTcRef x = wrapTcS (TcM.newTcRef x)
2649
2650 readTcRef :: TcRef a -> TcS a
2651 readTcRef ref = wrapTcS (TcM.readTcRef ref)
2652
2653 updTcRef :: TcRef a -> (a->a) -> TcS ()
2654 updTcRef ref upd_fn = wrapTcS (TcM.updTcRef ref upd_fn)
2655
2656 getTcEvBinds :: TcS (Maybe EvBindsVar)
2657 getTcEvBinds = TcS (return . tcs_ev_binds)
2658
2659 getTcEvBindsFromVar :: EvBindsVar -> TcS (Bag EvBind)
2660 getTcEvBindsFromVar = wrapTcS . TcM.getTcEvBinds
2661
2662 getTcLevel :: TcS TcLevel
2663 getTcLevel = wrapTcS TcM.getTcLevel
2664
2665 getTcEvBindsMap :: TcS EvBindMap
2666 getTcEvBindsMap
2667 = do { ev_binds <- getTcEvBinds
2668 ; case ev_binds of
2669 Just (EvBindsVar ev_ref _) -> wrapTcS $ TcM.readTcRef ev_ref
2670 Nothing -> return emptyEvBindMap }
2671
2672 unifyTyVar :: TcTyVar -> TcType -> TcS ()
2673 -- Unify a meta-tyvar with a type
2674 -- We keep track of how many unifications have happened in tcs_unified,
2675 --
2676 -- We should never unify the same variable twice!
2677 unifyTyVar tv ty
2678 = ASSERT2( isMetaTyVar tv, ppr tv )
2679 TcS $ \ env ->
2680 do { TcM.traceTc "unifyTyVar" (ppr tv <+> text ":=" <+> ppr ty)
2681 ; TcM.writeMetaTyVar tv ty
2682 ; TcM.updTcRef (tcs_unified env) (+1) }
2683
2684 unflattenFmv :: TcTyVar -> TcType -> TcS ()
2685 -- Fill a flatten-meta-var, simply by unifying it.
2686 -- This does NOT count as a unification in tcs_unified.
2687 unflattenFmv tv ty
2688 = ASSERT2( isMetaTyVar tv, ppr tv )
2689 TcS $ \ _ ->
2690 do { TcM.traceTc "unflattenFmv" (ppr tv <+> text ":=" <+> ppr ty)
2691 ; TcM.writeMetaTyVar tv ty }
2692
2693 reportUnifications :: TcS a -> TcS (Int, a)
2694 reportUnifications (TcS thing_inside)
2695 = TcS $ \ env ->
2696 do { inner_unified <- TcM.newTcRef 0
2697 ; res <- thing_inside (env { tcs_unified = inner_unified })
2698 ; n_unifs <- TcM.readTcRef inner_unified
2699 ; TcM.updTcRef (tcs_unified env) (+ n_unifs)
2700 ; return (n_unifs, res) }
2701
2702 getDefaultInfo :: TcS ([Type], (Bool, Bool))
2703 getDefaultInfo = wrapTcS TcM.tcGetDefaultTys
2704
2705 -- Just get some environments needed for instance looking up and matching
2706 -- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2707
2708 getInstEnvs :: TcS InstEnvs
2709 getInstEnvs = wrapTcS $ TcM.tcGetInstEnvs
2710
2711 getFamInstEnvs :: TcS (FamInstEnv, FamInstEnv)
2712 getFamInstEnvs = wrapTcS $ FamInst.tcGetFamInstEnvs
2713
2714 getTopEnv :: TcS HscEnv
2715 getTopEnv = wrapTcS $ TcM.getTopEnv
2716
2717 getGblEnv :: TcS TcGblEnv
2718 getGblEnv = wrapTcS $ TcM.getGblEnv
2719
2720 getLclEnv :: TcS TcLclEnv
2721 getLclEnv = wrapTcS $ TcM.getLclEnv
2722
2723 tcLookupClass :: Name -> TcS Class
2724 tcLookupClass c = wrapTcS $ TcM.tcLookupClass c
2725
2726 -- Setting names as used (used in the deriving of Coercible evidence)
2727 -- Too hackish to expose it to TcS? In that case somehow extract the used
2728 -- constructors from the result of solveInteract
2729 addUsedDataCons :: GlobalRdrEnv -> TyCon -> TcS ()
2730 addUsedDataCons rdr_env tycon = wrapTcS $ TcM.addUsedDataCons rdr_env tycon
2731
2732 -- Various smaller utilities [TODO, maybe will be absorbed in the instance matcher]
2733 -- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2734
2735 checkWellStagedDFun :: PredType -> DFunId -> CtLoc -> TcS ()
2736 checkWellStagedDFun pred dfun_id loc
2737 = wrapTcS $ TcM.setCtLocM loc $
2738 do { use_stage <- TcM.getStage
2739 ; TcM.checkWellStaged pp_thing bind_lvl (thLevel use_stage) }
2740 where
2741 pp_thing = text "instance for" <+> quotes (ppr pred)
2742 bind_lvl = TcM.topIdLvl dfun_id
2743
2744 pprEq :: TcType -> TcType -> SDoc
2745 pprEq ty1 ty2 = pprParendType ty1 <+> char '~' <+> pprParendType ty2
2746
2747 isTouchableMetaTyVarTcS :: TcTyVar -> TcS Bool
2748 isTouchableMetaTyVarTcS tv
2749 = do { tclvl <- getTcLevel
2750 ; return $ isTouchableMetaTyVar tclvl tv }
2751
2752 isFilledMetaTyVar_maybe :: TcTyVar -> TcS (Maybe Type)
2753 isFilledMetaTyVar_maybe tv
2754 = case tcTyVarDetails tv of
2755 MetaTv { mtv_ref = ref }
2756 -> do { cts <- wrapTcS (TcM.readTcRef ref)
2757 ; case cts of
2758 Indirect ty -> return (Just ty)
2759 Flexi -> return Nothing }
2760 _ -> return Nothing
2761
2762 isFilledMetaTyVar :: TcTyVar -> TcS Bool
2763 isFilledMetaTyVar tv = wrapTcS (TcM.isFilledMetaTyVar tv)
2764
2765 zonkTyCoVarsAndFV :: TcTyCoVarSet -> TcS TcTyCoVarSet
2766 zonkTyCoVarsAndFV tvs = wrapTcS (TcM.zonkTyCoVarsAndFV tvs)
2767
2768 zonkTyCoVarsAndFVList :: [TcTyCoVar] -> TcS [TcTyCoVar]
2769 zonkTyCoVarsAndFVList tvs = wrapTcS (TcM.zonkTyCoVarsAndFVList tvs)
2770
2771 zonkCo :: Coercion -> TcS Coercion
2772 zonkCo = wrapTcS . TcM.zonkCo
2773
2774 zonkTcType :: TcType -> TcS TcType
2775 zonkTcType ty = wrapTcS (TcM.zonkTcType ty)
2776
2777 zonkTcTypes :: [TcType] -> TcS [TcType]
2778 zonkTcTypes tys = wrapTcS (TcM.zonkTcTypes tys)
2779
2780 zonkTcTyVar :: TcTyVar -> TcS TcType
2781 zonkTcTyVar tv = wrapTcS (TcM.zonkTcTyVar tv)
2782
2783 zonkSimples :: Cts -> TcS Cts
2784 zonkSimples cts = wrapTcS (TcM.zonkSimples cts)
2785
2786 zonkWC :: WantedConstraints -> TcS WantedConstraints
2787 zonkWC wc = wrapTcS (TcM.zonkWC wc)
2788
2789 {-
2790 Note [Do not add duplicate derived insolubles]
2791 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2792 In general we *must* add an insoluble (Int ~ Bool) even if there is
2793 one such there already, because they may come from distinct call
2794 sites. Not only do we want an error message for each, but with
2795 -fdefer-type-errors we must generate evidence for each. But for
2796 *derived* insolubles, we only want to report each one once. Why?
2797
2798 (a) A constraint (C r s t) where r -> s, say, may generate the same fundep
2799 equality many times, as the original constraint is successively rewritten.
2800
2801 (b) Ditto the successive iterations of the main solver itself, as it traverses
2802 the constraint tree. See example below.
2803
2804 Also for *given* insolubles we may get repeated errors, as we
2805 repeatedly traverse the constraint tree. These are relatively rare
2806 anyway, so removing duplicates seems ok. (Alternatively we could take
2807 the SrcLoc into account.)
2808
2809 Note that the test does not need to be particularly efficient because
2810 it is only used if the program has a type error anyway.
2811
2812 Example of (b): assume a top-level class and instance declaration:
2813
2814 class D a b | a -> b
2815 instance D [a] [a]
2816
2817 Assume we have started with an implication:
2818
2819 forall c. Eq c => { wc_simple = D [c] c [W] }
2820
2821 which we have simplified to:
2822
2823 forall c. Eq c => { wc_simple = D [c] c [W]
2824 , wc_insols = (c ~ [c]) [D] }
2825
2826 For some reason, e.g. because we floated an equality somewhere else,
2827 we might try to re-solve this implication. If we do not do a
2828 dropDerivedWC, then we will end up trying to solve the following
2829 constraints the second time:
2830
2831 (D [c] c) [W]
2832 (c ~ [c]) [D]
2833
2834 which will result in two Deriveds to end up in the insoluble set:
2835
2836 wc_simple = D [c] c [W]
2837 wc_insols = (c ~ [c]) [D], (c ~ [c]) [D]
2838 -}
2839
2840 -- Flatten skolems
2841 -- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2842 newFlattenSkolem :: CtFlavour -> CtLoc
2843 -> TcType -- F xis
2844 -> TcS (CtEvidence, Coercion, TcTyVar) -- [W] x:: F xis ~ fsk
2845 newFlattenSkolem Given loc fam_ty
2846 = do { fsk <- newFsk fam_ty
2847 ; let co = mkNomReflCo fam_ty
2848 ; ev <- newGivenEvVar loc (mkPrimEqPred fam_ty (mkTyVarTy fsk),
2849 EvCoercion co)
2850 ; return (ev, co, fsk) }
2851
2852 newFlattenSkolem Wanted loc fam_ty
2853 = do { fmv <- newFmv fam_ty
2854 ; (ev, hole_co) <- newWantedEq loc Nominal fam_ty (mkTyVarTy fmv)
2855 ; return (ev, hole_co, fmv) }
2856
2857 newFlattenSkolem Derived loc fam_ty
2858 = do { fmv <- newFmv fam_ty
2859 ; ev <- newDerivedNC loc (mkPrimEqPred fam_ty (mkTyVarTy fmv))
2860 ; return (ev, pprPanic "newFlattenSkolem [D]" (ppr fam_ty), fmv) }
2861
2862 newFsk, newFmv :: TcType -> TcS TcTyVar
2863 newFsk fam_ty = wrapTcS (TcM.newFskTyVar fam_ty)
2864 newFmv fam_ty = wrapTcS (TcM.newFmvTyVar fam_ty)
2865
2866 extendFlatCache :: TyCon -> [Type] -> (TcCoercion, TcType, CtFlavour) -> TcS ()
2867 extendFlatCache tc xi_args stuff
2868 = do { dflags <- getDynFlags
2869 ; when (gopt Opt_FlatCache dflags) $
2870 updInertTcS $ \ is@(IS { inert_flat_cache = fc }) ->
2871 is { inert_flat_cache = insertExactFunEq fc tc xi_args stuff } }
2872
2873 -- Instantiations
2874 -- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2875
2876 instDFunType :: DFunId -> [DFunInstType] -> TcS ([TcType], TcThetaType)
2877 instDFunType dfun_id inst_tys
2878 = wrapTcS $ TcM.instDFunType dfun_id inst_tys
2879
2880 newFlexiTcSTy :: Kind -> TcS TcType
2881 newFlexiTcSTy knd = wrapTcS (TcM.newFlexiTyVarTy knd)
2882
2883 cloneMetaTyVar :: TcTyVar -> TcS TcTyVar
2884 cloneMetaTyVar tv = wrapTcS (TcM.cloneMetaTyVar tv)
2885
2886 demoteUnfilledFmv :: TcTyVar -> TcS ()
2887 -- If a flatten-meta-var is still un-filled,
2888 -- turn it into an ordinary meta-var
2889 demoteUnfilledFmv fmv
2890 = wrapTcS $ do { is_filled <- TcM.isFilledMetaTyVar fmv
2891 ; unless is_filled $
2892 do { tv_ty <- TcM.newFlexiTyVarTy (tyVarKind fmv)
2893 ; TcM.writeMetaTyVar fmv tv_ty } }
2894
2895 instFlexiTcS :: [TKVar] -> TcS (TCvSubst, [TcType])
2896 instFlexiTcS tvs = wrapTcS (mapAccumLM inst_one emptyTCvSubst tvs)
2897 where
2898 inst_one subst tv
2899 = do { ty' <- instFlexiTcSHelper (tyVarName tv)
2900 (substTyUnchecked subst (tyVarKind tv))
2901 ; return (extendTvSubst subst tv ty', ty') }
2902
2903 instFlexiTcSHelper :: Name -> Kind -> TcM TcType
2904 instFlexiTcSHelper tvname kind
2905 = do { uniq <- TcM.newUnique
2906 ; details <- TcM.newMetaDetails TauTv
2907 ; let name = setNameUnique tvname uniq
2908 ; return (mkTyVarTy (mkTcTyVar name kind details)) }
2909
2910
2911
2912 -- Creating and setting evidence variables and CtFlavors
2913 -- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2914
2915 data MaybeNew = Fresh CtEvidence | Cached EvTerm
2916
2917 isFresh :: MaybeNew -> Bool
2918 isFresh (Fresh {}) = True
2919 isFresh (Cached {}) = False
2920
2921 freshGoals :: [MaybeNew] -> [CtEvidence]
2922 freshGoals mns = [ ctev | Fresh ctev <- mns ]
2923
2924 getEvTerm :: MaybeNew -> EvTerm
2925 getEvTerm (Fresh ctev) = ctEvTerm ctev
2926 getEvTerm (Cached evt) = evt
2927
2928 setEvBind :: EvBind -> TcS ()
2929 setEvBind ev_bind
2930 = do { tc_evbinds <- getTcEvBinds
2931 ; case tc_evbinds of
2932 Just evb -> wrapTcS $ TcM.addTcEvBind evb ev_bind
2933 Nothing -> pprPanic "setEvBind" (ppr ev_bind) }
2934
2935 -- | Equalities only
2936 setWantedEq :: TcEvDest -> Coercion -> TcS ()
2937 setWantedEq (HoleDest hole) co
2938 = do { useVars (tyCoVarsOfCo co)
2939 ; wrapTcS $ TcM.fillCoercionHole hole co }
2940 setWantedEq (EvVarDest ev) _ = pprPanic "setWantedEq" (ppr ev)
2941
2942 -- | Equalities only
2943 setEqIfWanted :: CtEvidence -> Coercion -> TcS ()
2944 setEqIfWanted (CtWanted { ctev_dest = dest }) co = setWantedEq dest co
2945 setEqIfWanted _ _ = return ()
2946
2947 -- | Good for equalities and non-equalities
2948 setWantedEvTerm :: TcEvDest -> EvTerm -> TcS ()
2949 setWantedEvTerm (HoleDest hole) tm
2950 = do { let co = evTermCoercion tm
2951 ; useVars (tyCoVarsOfCo co)
2952 ; wrapTcS $ TcM.fillCoercionHole hole co }
2953 setWantedEvTerm (EvVarDest ev) tm = setWantedEvBind ev tm
2954
2955 setWantedEvBind :: EvVar -> EvTerm -> TcS ()
2956 setWantedEvBind ev_id tm = setEvBind (mkWantedEvBind ev_id tm)
2957
2958 setEvBindIfWanted :: CtEvidence -> EvTerm -> TcS ()
2959 setEvBindIfWanted ev tm
2960 = case ev of
2961 CtWanted { ctev_dest = dest }
2962 -> setWantedEvTerm dest tm
2963 _ -> return ()
2964
2965 newTcEvBinds :: TcS EvBindsVar
2966 newTcEvBinds = wrapTcS TcM.newTcEvBinds
2967
2968 newEvVar :: TcPredType -> TcS EvVar
2969 newEvVar pred = wrapTcS (TcM.newEvVar pred)
2970
2971 newGivenEvVar :: CtLoc -> (TcPredType, EvTerm) -> TcS CtEvidence
2972 -- Make a new variable of the given PredType,
2973 -- immediately bind it to the given term
2974 -- and return its CtEvidence
2975 -- See Note [Bind new Givens immediately] in TcRnTypes
2976 newGivenEvVar loc (pred, rhs)
2977 = do { new_ev <- newBoundEvVarId pred rhs
2978 ; return (CtGiven { ctev_pred = pred, ctev_evar = new_ev, ctev_loc = loc }) }
2979
2980 -- | Make a new 'Id' of the given type, bound (in the monad's EvBinds) to the
2981 -- given term
2982 newBoundEvVarId :: TcPredType -> EvTerm -> TcS EvVar
2983 newBoundEvVarId pred rhs
2984 = do { new_ev <- newEvVar pred
2985 ; setEvBind (mkGivenEvBind new_ev rhs)
2986 ; return new_ev }
2987
2988 newGivenEvVars :: CtLoc -> [(TcPredType, EvTerm)] -> TcS [CtEvidence]
2989 newGivenEvVars loc pts = mapM (newGivenEvVar loc) pts
2990
2991 -- | Make a new equality CtEvidence
2992 newWantedEq :: CtLoc -> Role -> TcType -> TcType -> TcS (CtEvidence, Coercion)
2993 newWantedEq loc role ty1 ty2
2994 = do { hole <- wrapTcS $ TcM.newCoercionHole
2995 ; traceTcS "Emitting new coercion hole" (ppr hole <+> dcolon <+> ppr pty)
2996 ; return ( CtWanted { ctev_pred = pty, ctev_dest = HoleDest hole
2997 , ctev_loc = loc}
2998 , mkHoleCo hole role ty1 ty2 ) }
2999 where
3000 pty = mkPrimEqPredRole role ty1 ty2
3001
3002 -- no equalities here. Use newWantedEqNC instead
3003 newWantedEvVarNC :: CtLoc -> TcPredType -> TcS CtEvidence
3004 -- Don't look up in the solved/inerts; we know it's not there
3005 newWantedEvVarNC loc pty
3006 = do { -- checkReductionDepth loc pty
3007 ; new_ev <- newEvVar pty
3008 ; traceTcS "Emitting new wanted" (ppr new_ev <+> dcolon <+> ppr pty $$
3009 pprCtLoc loc)
3010 ; return (CtWanted { ctev_pred = pty, ctev_dest = EvVarDest new_ev
3011 , ctev_loc = loc })}
3012
3013 newWantedEvVar :: CtLoc -> TcPredType -> TcS MaybeNew
3014 -- For anything except ClassPred, this is the same as newWantedEvVarNC
3015 newWantedEvVar loc pty
3016 = do { mb_ct <- lookupInInerts pty
3017 ; case mb_ct of
3018 Just ctev
3019 | not (isDerived ctev)
3020 -> do { traceTcS "newWantedEvVar/cache hit" $ ppr ctev
3021 ; return $ Cached (ctEvTerm ctev) }
3022 _ -> do { ctev <- newWantedEvVarNC loc pty
3023 ; return (Fresh ctev) } }
3024
3025 -- deals with both equalities and non equalities. Tries to look
3026 -- up non-equalities in the cache
3027 newWanted :: CtLoc -> PredType -> TcS MaybeNew
3028 newWanted loc pty
3029 | Just (role, ty1, ty2) <- getEqPredTys_maybe pty
3030 = Fresh . fst <$> newWantedEq loc role ty1 ty2
3031 | otherwise
3032 = newWantedEvVar loc pty
3033
3034 emitNewDerived :: CtLoc -> TcPredType -> TcS ()
3035 emitNewDerived loc pred
3036 = do { ev <- newDerivedNC loc pred
3037 ; traceTcS "Emitting new derived" (ppr ev)
3038 ; updWorkListTcS (extendWorkListDerived loc ev) }
3039
3040 emitNewDeriveds :: CtLoc -> [TcPredType] -> TcS ()
3041 emitNewDeriveds loc preds
3042 | null preds
3043 = return ()
3044 | otherwise
3045 = do { evs <- mapM (newDerivedNC loc) preds
3046 ; traceTcS "Emitting new deriveds" (ppr evs)
3047 ; updWorkListTcS (extendWorkListDeriveds loc evs) }
3048
3049 emitNewDerivedEq :: CtLoc -> Role -> TcType -> TcType -> TcS ()
3050 -- Create new equality Derived and put it in the work list
3051 -- There's no caching, no lookupInInerts
3052 emitNewDerivedEq loc role ty1 ty2
3053 = do { ev <- newDerivedNC loc (mkPrimEqPredRole role ty1 ty2)
3054 ; traceTcS "Emitting new derived equality" (ppr ev $$ pprCtLoc loc)
3055 ; updWorkListTcS (extendWorkListDerived loc ev) }
3056
3057 newDerivedNC :: CtLoc -> TcPredType -> TcS CtEvidence
3058 newDerivedNC loc pred
3059 = do { -- checkReductionDepth loc pred
3060 ; return (CtDerived { ctev_pred = pred, ctev_loc = loc }) }
3061
3062 -- --------- Check done in TcInteract.selectNewWorkItem???? ---------
3063 -- | Checks if the depth of the given location is too much. Fails if
3064 -- it's too big, with an appropriate error message.
3065 checkReductionDepth :: CtLoc -> TcType -- ^ type being reduced
3066 -> TcS ()
3067 checkReductionDepth loc ty
3068 = do { dflags <- getDynFlags
3069 ; when (subGoalDepthExceeded dflags (ctLocDepth loc)) $
3070 wrapErrTcS $
3071 solverDepthErrorTcS loc ty }
3072
3073 matchFam :: TyCon -> [Type] -> TcS (Maybe (Coercion, TcType))
3074 matchFam tycon args = wrapTcS $ matchFamTcM tycon args
3075
3076 matchFamTcM :: TyCon -> [Type] -> TcM (Maybe (Coercion, TcType))
3077 -- Given (F tys) return (ty, co), where co :: F tys ~ ty
3078 matchFamTcM tycon args
3079 = do { fam_envs <- FamInst.tcGetFamInstEnvs
3080 ; return $ reduceTyFamApp_maybe fam_envs Nominal tycon args }
3081
3082 {-
3083 Note [Residual implications]
3084 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3085 The wl_implics in the WorkList are the residual implication
3086 constraints that are generated while solving or canonicalising the
3087 current worklist. Specifically, when canonicalising
3088 (forall a. t1 ~ forall a. t2)
3089 from which we get the implication
3090 (forall a. t1 ~ t2)
3091 See TcSMonad.deferTcSForAllEq
3092 -}
3093
3094 -- Deferring forall equalities as implications
3095 -- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3096
3097 deferTcSForAllEq :: Role -- Nominal or Representational
3098 -> CtLoc -- Original wanted equality flavor
3099 -> [Coercion] -- among the kinds of the binders
3100 -> ([TyBinder],TcType) -- ForAll tvs1 body1
3101 -> ([TyBinder],TcType) -- ForAll tvs2 body2
3102 -> TcS Coercion
3103 deferTcSForAllEq role loc kind_cos (bndrs1,body1) (bndrs2,body2)
3104 = do { let tvs1' = zipWithEqual "deferTcSForAllEq"
3105 mkCastTy (mkTyVarTys tvs1) kind_cos
3106 body2' = substTyWithUnchecked tvs2 tvs1' body2
3107 ; (subst, skol_tvs) <- wrapTcS $ TcM.tcInstSkolTyVars tvs1
3108 ; let phi1 = Type.substTyUnchecked subst body1
3109 phi2 = Type.substTyUnchecked subst body2'
3110 skol_info = UnifyForAllSkol phi1
3111
3112 ; (ctev, hole_co) <- newWantedEq loc role phi1 phi2
3113 ; env <- getLclEnv
3114 ; let new_tclvl = pushTcLevel (tcl_tclvl env)
3115 wc = WC { wc_simple = singleCt (mkNonCanonical ctev)
3116 , wc_impl = emptyBag
3117 , wc_insol = emptyCts }
3118 imp = Implic { ic_tclvl = new_tclvl
3119 , ic_skols = skol_tvs
3120 , ic_no_eqs = True
3121 , ic_given = []
3122 , ic_wanted = wc
3123 , ic_status = IC_Unsolved
3124 , ic_binds = Nothing -- no place to put binds
3125 , ic_env = env
3126 , ic_info = skol_info }
3127 ; updWorkListTcS (extendWorkListImplic imp)
3128 ; let cobndrs = zip skol_tvs kind_cos
3129 ; return $ mkForAllCos cobndrs hole_co }
3130 where
3131 tvs1 = map (binderVar "deferTcSForAllEq") bndrs1
3132 tvs2 = map (binderVar "deferTcSForAllEq") bndrs2