#8184: Compiler panic in the presence of cyclic imports ---------------------------------------+----------------------------------- Reporter: parcs | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.7 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: Type of failure: Compile-time crash | Unknown/Multiple Test Case: | Difficulty: Unknown Blocking: 910 | Blocked By: | Related Tickets: ---------------------------------------+----------------------------------- Comment (by parcs): Digging deeper, the problem seems to be that in `WwLib.mkWWstr_one`, `deepSplitProductType_maybe y_type' returns `Nothing` when compiling `B`. But when compiling `C` (after re-typechecking the loop) it returns `Just y_datacon`. The immediate problem seem to lie in `WwLib.deepSplitProductType_maybe`: {{{ #!haskell deepSplitProductType_maybe :: Type -> Maybe (DataCon, [Type], [Type], Coercion) deepSplitProductType_maybe ty | let (ty1, co) = topNormaliseNewType ty `orElse` (ty, mkReflCo Representational ty) , Just (tc, tc_args) <- splitTyConApp_maybe ty1 , Just con <- isDataProductTyCon_maybe tc -- ^ This pattern fails for Y when compiling A.hs (because Y looks empty then) -- and succeeds when compiling C (because the import loop has been -- re-typechecked by then). This discrepancy causes `WwLib.mkWWstr_one` to -- return a different worker type depending on whether one is compiling -- a module inside the loop or outside the loop. = Just (con, tc_args, dataConInstArgTys con tc_args, co) deepSplitProductType_maybe _ = Nothing }}} -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/8184#comment:2 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler