[GHC] #13426: compile-time memory-usage regression for DynFlags between GHC 8.0 and GHC 8.2

[GHC]

#13426: compile-time memory-usage regression for DynFlags between GHC 8.0 and GHC
8.2
-------------------------------------+-------------------------------------
        Reporter:  hvr               |                Owner:  (none)
            Type:  bug               |               Status:  new
        Priority:  high              |            Milestone:  8.2.1
       Component:  Compiler          |              Version:  8.1
      Resolution:                    |             Keywords:
Operating System:  Unknown/Multiple  |         Architecture:
 Type of failure:  Compile-time      |  Unknown/Multiple
  performance bug                    |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:                    |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by rwbarton):

 Replying to [comment:10 simonpj]:
 > I have an idea.  Consider `Simplify` line 1005
 > {{{
 > simplExprF1 env (Type ty)      cont = ASSERT( contIsRhsOrArg cont )
 >                                       rebuild env (Type (substTy env
 ty)) cont
 > }}}

 This wasn't enough unfortunately.

 Some more RTS/profiling hackery led me to... the next case of the same
 function:
 {{{
 simplExprF1 env (App fun arg) cont
   = simplExprF env fun $
     case arg of
       Type ty -> ApplyToTy  { sc_arg_ty  = substTy env ty
                             , sc_hole_ty = substTy env (exprType fun)
                             , sc_cont    = cont }
       _       -> ApplyToVal { sc_arg = arg, sc_env = env
                             , sc_dup = NoDup, sc_cont = cont }
 }}}
 Here `sc_arg_ty` and `sc_hole_ty` apparently need a similar treatment. And
 with that the space leak seems to be gone! I'll confirm with a clean
 rebuild.

 What I still don't understand is why this wasn't always a problem, since
 those lines have not changed in a long time. I guess there is some new
 code that manipulates `ApplyToTy`, so maybe the old version of that code
 did enough evaluation to avoid the space leak.

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/13426#comment:11>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler