Re: [GHC] #10528: compile time performance regression with OverloadedStrings and Text

24 Jul 2015

      #10528: compile time performance regression with OverloadedStrings and Text
-------------------------------------+-------------------------------------
        Reporter:  jakewheat         |                   Owner:  bgamari
            Type:  bug               |                  Status:  new
        Priority:  high              |               Milestone:  7.10.3
       Component:  Compiler          |                 Version:  7.10.2-rc2
      Resolution:                    |                Keywords:
Operating System:  Linux             |            Architecture:
 Type of failure:  Compile-time      |  Unknown/Multiple
  performance bug                    |               Test Case:
      Blocked By:                    |                Blocking:
 Related Tickets:                    |  Differential Revisions:
-------------------------------------+-------------------------------------

Comment (by bgamari):

 In the simplifier pass right before the two compilers diverse the code
 looks something like,
 {{{
 (c_d32a
    (Data.Text.pack
       (GHC.Base.build
          @ GHC.Types.Char
          (\ (@ b) ->
             GHC.CString.unpackFoldrCString#
               @ b "bpchar"#)))
    (c_d32a
       (Data.Text.pack
          (GHC.Base.build
             @ GHC.Types.Char
             (\ (@ b) ->
                GHC.CString.unpackFoldrCString#
                  @ b "bytea"#)))
 }}}

 Into this `GHC.Base.build` is inlined,
 {{{
 Considering inlining: build
   arg infos [ValueArg]
   interesting continuation RuleArgCtxt
   some_benefit True
   is exp: True
   is work-free: True
   guidance ALWAYS_IF(arity=1,unsat_ok=False,boring_ok=False)
   ANSWER = YES
 Inlining done: GHC.Base.build
     Inlined fn:  \ (@ a)
                    (g [Occ=Once!] :: forall b. (a -> b -> b) -> b -> b) ->
                    g @ [a] (GHC.Types.: @ a) (GHC.Types.[] @ a)
 }}}

 and then `pack`,
 {{{
 Considering inlining: pack
   arg infos [NonTrivArg]
   interesting continuation BoringCtxt
   some_benefit True
   is exp: True
   is work-free: True
   guidance ALWAYS_IF(arity=0,unsat_ok=False,boring_ok=False)
   ANSWER = YES
 Inlining done: Data.Text.pack
     Inlined fn:  \ (x [Occ=Once] :: GHC.Base.String) ->
                    Data.Text.Internal.Fusion.unstream
                      (Data.Text.Internal.Fusion.Common.map
                         Data.Text.Internal.safe
                         (Data.Text.Internal.Fusion.Common.streamList @
 GHC.Types.Char x))
 }}}

 which I believe is then supposed to cause this rule to fire,
 {{{
 {-# RULES "TEXT literal" forall a.
     unstream (S.map safe (S.streamList (GHC.unpackCString# a)))
       = unpackCString# a #-}
 }}}

 Yet this appears not to happen in 7.10.2.

--
Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/10528#comment:11
GHC http://www.haskell.org/ghc/
The Glasgow Haskell Compiler