[project @ 2003-04-10 15:46:11 by simonpj]

A subtle type checker bug, which showed up in rebindable1 test

diff --git a/ghc/compiler/typecheck/Inst.lhs b/ghc/compiler/typecheck/Inst.lhs
index ce020f6..86522ad 100644 (file)
--- a/ghc/compiler/typecheck/Inst.lhs
+++ b/ghc/compiler/typecheck/Inst.lhs
@@ -353,7 +353,13 @@ newOverloadedLit :: InstOrigin
                 -> HsOverLit
                 -> TcType
                 -> TcM TcExpr
-newOverloadedLit orig lit@(HsIntegral i fi) expected_ty
+newOverloadedLit orig lit expected_ty
+  = zapToType expected_ty      `thenM_` 
+       -- The expected type might be a 'hole' type variable, 
+       -- in which case we must zap it to an ordinary type variable
+    new_over_lit orig lit expected_ty
+
+new_over_lit orig lit@(HsIntegral i fi) expected_ty
   | fi /= fromIntegerName      -- Do not generate a LitInst for rebindable
                                -- syntax.  Reason: tcSyntaxName does unification
                                -- which is very inconvenient in tcSimplify
@@ -366,7 +372,7 @@ newOverloadedLit orig lit@(HsIntegral i fi) expected_ty
   | otherwise
   = newLitInst orig lit expected_ty
 
-newOverloadedLit orig lit@(HsFractional r fr) expected_ty
+new_over_lit orig lit@(HsFractional r fr) expected_ty
   | fr /= fromRationalName     -- c.f. HsIntegral case
   = tcSyntaxName orig expected_ty fromRationalName fr  `thenM` \ (expr, _) ->
     mkRatLit r                                         `thenM` \ rat_lit ->
@@ -381,9 +387,6 @@ newOverloadedLit orig lit@(HsFractional r fr) expected_ty
 newLitInst orig lit expected_ty
   = getInstLoc orig            `thenM` \ loc ->
     newUnique                  `thenM` \ new_uniq ->
-    zapToType expected_ty      `thenM_` 
-       -- The expected type might be a 'hole' type variable, 
-       -- in which case we must zap it to an ordinary type variable
     let
        lit_inst = LitInst lit_id lit expected_ty loc
        lit_id   = mkSysLocal FSLIT("lit") new_uniq expected_ty