[GHC] #9617: Try to replace `quot` and `rem` with `quotRem` when possible

19 Sep 2014

      #9617: Try to replace `quot` and `rem` with `quotRem` when possible
-------------------------------------+-------------------------------------
       Reporter:  dfeuer             |                   Owner:
           Type:  feature request    |                  Status:  new
       Priority:  normal             |               Milestone:
      Component:  Compiler           |                 Version:  7.9
       Keywords:                     |        Operating System:
   Architecture:  Unknown/Multiple   |  Unknown/Multiple
     Difficulty:  Unknown            |         Type of failure:
     Blocked By:                     |  None/Unknown
Related Tickets:                     |               Test Case:
                                     |                Blocking:
                                     |  Differential Revisions:
-------------------------------------+-------------------------------------
 If I define

 {{{#!hs
 myquot x y = fst (quotRem x y)
 myrem x y = snd (quotRem x y)

 bob :: Int -> Int -> Int
 bob x y = myquot x y + myrem x y
 }}}

 Then I get this beautiful thing with GHC 7.9 (I ''don't'' get anything
 beautiful in 7.8.3, so great work, compiler gurus!):

 {{{#!hs
 bob :: Int -> Int -> Int
 bob =
   \ (w_sWx :: Int) (w1_sWy :: Int) ->
     case w_sWx of _ { I# ww1_sWB ->
     case w1_sWy of _ { I# ww3_sWF ->
     case ww3_sWF of wild_aSo {
       __DEFAULT ->
         case quotRemInt# ww1_sWB wild_aSo of _ { (# ipv_aSr, ipv1_aSs #)
 ->
         I# (+# ipv_aSr ipv1_aSs)
         };
       (-1) ->
         case ww1_sWB of wild1_aSu {
           __DEFAULT ->
             case quotRemInt# wild1_aSu (-1) of _ { (# ipv_aSx, ipv1_aSy #)
 ->
             I# (+# ipv_aSx ipv1_aSy)
             };
           (-9223372036854775808) -> case overflowError of wild2_00 { }
         };
       0 -> case divZeroError of wild1_00 { }
     }
     }
     }
 }}}

 However, if I write

 {{{
 jones :: Int -> Int -> Int
 jones x y = quot x y + rem x y
 }}}

 I don't get anything nice.

 What I'm thinking (perhaps out of ignorance) is that we might be able to
 use the `myquot` and `myrem` definitions if it's possible to get some sort
 of dead code elimination to recognize when one or the other is not used,
 at which point it can replace the `quotRemInt#` with `quotInt#` or
 `remInt#`. Then we can quit the silly `quotRem` dance in user code and
 just write what we actually mean. Of course, exactly the same thing
 applies to `divMod`.

--
Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/9617
GHC http://www.haskell.org/ghc/
The Glasgow Haskell Compiler