[GHC] #12150: Compile time performance degradation on code that uses undefined/error with CallStacks

[GHC]

#12150: Compile time performance degradation on code that uses undefined/error with
CallStacks
-------------------------------------+-------------------------------------
        Reporter:  thomie            |                Owner:  (none)
            Type:  bug               |               Status:  new
        Priority:  high              |            Milestone:  8.2.2
       Component:  Compiler          |              Version:  8.0.1
      Resolution:                    |             Keywords:
Operating System:  Unknown/Multiple  |         Architecture:
 Type of failure:  Compile-time      |  Unknown/Multiple
  performance bug                    |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:  #10844            |  Differential Rev(s):  Phab:D3753
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by simonpj):

 > For some reason, we're not eliminating the ...

 I assume this is just in the ''first'' simplifier run, when `case-of-case`
 is off?  In the second simplifier run, the `case` gets eliminated,
 correct?

 I know why this is.  In `Simplify` we have:
 {{{
 simplExprF1 env (Case scrut bndr _ alts) cont
   | sm_case_case (getMode env)
   = simplExprF env scrut (Select { sc_dup = NoDup, sc_bndr = bndr
                                  , sc_alts = alts
                                  , sc_env = env, sc_cont = cont })
   | otherwise
   = do { (env', scrut') <- simplExprF (zapFloats env) scrut $
                            mkBoringStop (substTy env (idType bndr))
        ; let scrut'' = wrapJoinFloats (seJoinFloats env') scrut'
              env''   = env `addLetFloats` env'
        ; rebuildCase env'' scrut'' bndr alts cont }
 }}}
 That is, if `case-of-case` is off, we simplify the scrutinee in a trivial
 continuation, and then wrap the alternatives around it.  So disabling
 case-of-case has also disabled case-of-bottom.  I did this fairly
 recently, which accounts for the change.

 Let me think about a better way.

 Meanwhile you are working on that `botSig` stuff, making it work with
 functions too, right?

-- 
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/12150#comment:19>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler