Unexpected boxing in generated code

[Simon Peyton-Jones]

Very curious.  It does indeed look as though the strictness analyser is confused; but it should certainly not be confused by mutual recursion. I'll definitely look into it.  But don't hold your breath -- it's a very busy fortnight.

Simon

| -----Original Message-----
| From: glasgow-haskell-users-bounces at haskell.org [mailto:glasgow-haskell-users-bounces at haskell.org] On
| Behalf Of Neil Mitchell
| Sent: 09 July 2007 23:14
| To: glasgow-haskell-users at haskell.org
| Subject: Unexpected boxing in generated code
|
| Hi,
|
| I've got an inner loop that I think I can see is strict in the Int
| argument being passed around, but that GHC 6.6.1 isn't unboxing. In
| the following example both functions take a GHC.Base.Int, which I
| think should be an Int#.
|
| Rec {
| f60_rS5 :: GHC.Prim.State# GHC.Prim.RealWorld -> GHC.Base.Int -> GHC.Base.Int
| [GlobalId]
| [Arity 2
|  NoCafRefs
|  Str: DmdType LL]
| f60_rS5 =
|   \ (v1_aWH :: GHC.Prim.State# GHC.Prim.RealWorld) (v2_aWI :: GHC.Base.Int) ->
|     case $wccall_r2kv v1_aWH of wild_X2j { (# ds_d1V4, ds1_d1V3 #) ->
|     case ds1_d1V3 of wild1_X2L {
|       __DEFAULT -> f60_rS5 ds_d1V4 v2_aWI;
|       (-1) -> v2_aWI;
|       10 -> f561_r2kx ds_d1V4 v2_aWI
|     }
|     }
| f561_r2kx :: GHC.Prim.State# GHC.Prim.RealWorld -> GHC.Base.Int -> GHC.Base.Int
| [GlobalId]
| [Arity 2
|  NoCafRefs
|  Str: DmdType LL]
| f561_r2kx =
|   \ (v1_aWm :: GHC.Prim.State# GHC.Prim.RealWorld) (v2_aWn :: GHC.Base.Int) ->
|     case $wccall_r2kv v1_aWm of wild_X2j { (# ds_d1V4, ds1_d1V3 #) ->
|     case ds1_d1V3 of wild1_X2P {
|       __DEFAULT ->
|         case v2_aWn of wild2_a2du { GHC.Base.I# x_a2dw ->
|         case wild1_X2P of wild3_X35 {
|           __DEFAULT -> f60_rS5 ds_d1V4 (GHC.Base.I# (GHC.Prim.+# x_a2dw 1));
|           10 -> f561_r2kx ds_d1V4 (GHC.Base.I# (GHC.Prim.+# x_a2dw 1))
|         }
|         };
|       (-1) -> v2_aWn
|     }
|     }
| end Rec }
|
| This code comes from a line counting program, I have attached the
| entire source. My character counting program does infer the correct
| strictness, although that is based on a single self-recursive
| function. The largest obvious difference is that the strictness
| depends on the two functions which call each other - does this impeed
| GHC's strictness analysis?
|
| Thanks
|
| Neil