#16114: strange "instance .. => .. => .. where ..." -------------------------------------+------------------------------------- Reporter: j.waldmann | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.6.3 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: | Unknown/Multiple Type of failure: None/Unknown | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Rev(s): Wiki Page: | -------------------------------------+------------------------------------- Comment (by RyanGlScott): Assuming that we don't wish to accept the original program, I think we can blame `rnClsInstDecl`. [https://gitlab.haskell.org/ghc/ghc/blob/master/compiler/rename/RnSource.hs#L... This] is the first part of that function: {{{#!hs rnClsInstDecl :: ClsInstDecl GhcPs -> RnM (ClsInstDecl GhcRn, FreeVars) rnClsInstDecl (ClsInstDecl { cid_poly_ty = inst_ty, cid_binds = mbinds , cid_sigs = uprags, cid_tyfam_insts = ats , cid_overlap_mode = oflag , cid_datafam_insts = adts }) = do { (inst_ty', inst_fvs) <- rnLHsInstType (text "an instance declaration") inst_ty ; let (ktv_names, _, head_ty') = splitLHsInstDeclTy inst_ty' ; let cls = case hsTyGetAppHead_maybe head_ty' of Nothing -> mkUnboundName (mkTcOccFS (fsLit "<class>")) Just (dL->L _ cls) -> cls -- rnLHsInstType has added an error message -- if hsTyGetAppHead_maybe fails }}} Since we're using `splitLHsInstDeclTy` to decompose `inst_ty`, if we feed it something sketchy like `Eq a => Eq a => Eq (T a)` as input, then `head_ty'` will be `Eq a => Eq (T a)` (which is not headed by a class, leading to havoc later on). Notice that that comment at the bottom assumes that `rnLHsInstType` throws an error message if `inst_ty` is malformed. But in commit 1c062b794bf71a329f65813ce7b72fe2bd3935f0, we have: {{{#!diff diff --git a/compiler/rename/RnTypes.hs b/compiler/rename/RnTypes.hs index 2305a04..b2dafb2 100644 (file) --- a/compiler/rename/RnTypes.hs +++ b/compiler/rename/RnTypes.hs @@ -326,20 +326,8 @@ rnImplicitBndrs bind_free_tvs doc rnLHsInstType :: SDoc -> LHsSigType GhcPs -> RnM (LHsSigType GhcRn, FreeVars) -- Rename the type in an instance or standalone deriving decl -- The 'doc_str' is "an instance declaration" or "a VECTORISE pragma" -rnLHsInstType doc_str inst_ty - | Just cls <- getLHsInstDeclClass_maybe inst_ty - , isTcOcc (rdrNameOcc (unLoc cls)) - -- The guards check that the instance type looks like - -- blah => C ty1 .. tyn - = do { let full_doc = doc_str <+> text "for" <+> quotes (ppr cls) - ; rnHsSigType (GenericCtx full_doc) inst_ty } - - | otherwise -- The instance is malformed, but we'd still like - -- to make progress rather than failing outright, so - -- we report more errors. So we rename it anyway. - = do { addErrAt (getLoc (hsSigType inst_ty)) $ - text "Malformed instance:" <+> ppr inst_ty - ; rnHsSigType (GenericCtx doc_str) inst_ty } +-- Do not try to decompose the inst_ty in case it is malformed +rnLHsInstType doc inst_ty = rnHsSigType (GenericCtx doc) inst_ty mk_implicit_bndrs :: [Name] -- implicitly bound -> a -- payload }}} Notice that `rnLHsInstType` no longer errors if given malformed input! So `rnClsInstDecl` charges on under the false pretense that `rnLHsInstType` succeeded. Perhaps the right thing to do would be to move the old validity check from `rnLHsInstType` to `rnClsInstDecl`? -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/16114#comment:2 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler