[project @ 2001-10-15 15:03:48 by simonpj]

Add comments re eta expansion

diff --git a/ghc/compiler/coreSyn/CorePrep.lhs b/ghc/compiler/coreSyn/CorePrep.lhs
index e03683b..5a4b636 100644 (file)
--- a/ghc/compiler/coreSyn/CorePrep.lhs
+++ b/ghc/compiler/coreSyn/CorePrep.lhs
@@ -160,9 +160,11 @@ corePrepRecPairs env pairs
     bndrs = map fst pairs
 
 corePrepRhs :: CloneEnv -> (Id, CoreExpr) -> UniqSM CoreExpr
+       -- Used for top-level bindings, and local recursive bindings
+       -- c.f. mkLocalNonRec, which does the other case
+       -- No nonsense about floating.
+       -- Prepare the RHS and eta expand it. 
 corePrepRhs env (bndr, rhs)
--- Prepare the RHS and eta expand it. 
--- No nonsense about floating
   = corePrepAnExpr env rhs     `thenUs` \ rhs' ->
     getUniquesUs               `thenUs` \ us ->
     returnUs (etaExpand (exprArity rhs') us rhs' (idType bndr))
@@ -416,6 +418,9 @@ mkLocalNonRec bndr dem floats rhs
        -- because floating the case would make it evaluated too early
        --
        -- Finally, eta-expand the RHS, for the benefit of the code gen
+       -- This might not have happened already, because eta expansion
+       -- is done by the simplifier only when there at least one lambda already.
+       --
        -- NB: we could refrain when the RHS is trivial (which can happen
        --     for exported things.  This would reduce the amount of code
        --     generated (a little) and make things a little words for