[commit: ghc] : Experiment: allow common heap binders to clash with left/right heap binders (9df6981)
Max Bolingbroke <[email protected]>
| Newsgroups | gmane.comp.lang.haskell.cvs.ghc |
|---|---|
| Message-ID | <[email protected]> |
Repository : ssh://darcs.haskell.org//srv/darcs/ghc On branch : http://hackage.haskell.org/trac/ghc/changeset/9df69812193402b2fc4627cf52472d60cd9b2fc9 >--------------------------------------------------------------- commit 9df69812193402b2fc4627cf52472d60cd9b2fc9 Author: Max Bolingbroke <[email protected]> Date: Thu Apr 19 18:36:36 2012 +0100 Experiment: allow common heap binders to clash with left/right heap binders >--------------------------------------------------------------- compiler/supercompile/Supercompile/Drive/MSG.hs | 8 ++++++-- 1 files changed, 6 insertions(+), 2 deletions(-) diff --git a/compiler/supercompile/Supercompile/Drive/MSG.hs b/compiler/supercompile/Supercompile/Drive/MSG.hs index f4de232..01daec6 100644 --- a/compiler/supercompile/Supercompile/Drive/MSG.hs +++ b/compiler/supercompile/Supercompile/Drive/MSG.hs @@ -104,7 +104,7 @@ data Pending = PendingVar Var Var | PendingTerm AnnedTerm AnnedTerm data MSGState = MSGState { - msgInScopeSet :: InScopeSet, -- We have to ensure all new vars are distinct from *each other* AND *top level vars* + msgInScopeSet :: InScopeSet, -- We have to ensure all new vars introduced by MSG are distinct from each other msgKnownPendingVars :: VarEnv (VarEnv Var), -- INVARIANT: the "known" maps are inverse to the pending list, msgKnownPendingTypes :: TypeMap (TypeMap TyVar), -- except that PendingTerms are not recorded in a "known" map at all. msgKnownPendingCoercions :: CoercionMap (CoercionMap CoVar), -- We don't *want* them in one because we don't mant MSGing to increase work sharing! @@ -267,7 +267,11 @@ msgWithReason {- mm -} (deeds_l, Heap h_l ids_l, k_l, qa_l) (deeds_r, Heap h_r i -- This was the source of a very confusing bug :-( let ids = ids_l `unionInScope` ids_r init_rn2 = mkRnEnv2 ids - msg_s = MSGState ids emptyVarEnv emptyTM emptyTM [] + -- NB: I've started to use emptyInScopeSet here rather than ids. The effect this has is that variable names from ids_l + -- and ids_r will actually occur in the common heap. This is perfectly OK for our purposes because we never try to bind + -- names from the common and individual heaps at the same time. Doing this also helps debugging because we minimise + -- the creation of brand new names as far as possible. + msg_s = MSGState emptyInScopeSet emptyVarEnv emptyTM emptyTM [] -- TODO: test for multiple solutions? Attempt to choose best? firstSuccess [ do ((qa, (k_l, k, k_r)), (rn_l, (h_l, heap, h_r), rn_r)) <- mres return ((deeds_l, Heap h_l ids_l, rn_l, k_l), (heap, k, qa), (deeds_r, Heap h_r ids_r, rn_r, k_r))