[GHC] #15696: Derived Ord instance for enumerations with more than 8 elements seems to be incorrect

[GHC]

#15696: Derived Ord instance for enumerations with more than 8 elements seems to be
incorrect
-------------------------------------+-------------------------------------
        Reporter:  mrkkrp            |                Owner:  (none)
            Type:  bug               |               Status:  patch
        Priority:  highest           |            Milestone:  8.6.2
       Component:  Compiler          |              Version:  8.6.1
      Resolution:                    |             Keywords:
Operating System:  Unknown/Multiple  |         Architecture:
 Type of failure:  Incorrect result  |  Unknown/Multiple
  at runtime                         |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:                    |  Differential Rev(s):  Phab:D5196,
       Wiki Page:                    |  Phab:D5201
-------------------------------------+-------------------------------------

Comment (by osa1):

 Simon, what do you think about marking `dataToTag#` as can't fail? I
 thought it
 makes sense, but when I do that I get this lint error: (add type
 annotation on scrutinee)

 {{{
 <no location info>: warning:
     In the expression: tagToEnum#
                          @ Bool
                          (case (ds :: Instr) of lwild {
                             __DEFAULT -> 0#;
                             C_ALabel_1 ipv -> 1#
                           })
     This argument does not satisfy the let/app invariant:
       case ds of lwild {
         __DEFAULT -> 0#;
         C_ALabel_1 ipv -> 1#
       }
 }}}

 Note that the expression does not actually have `dataToTag#`. Apparently
 making
 `dataToTag#` can't fail enables some transformations, which leads to this.

 (Type of `ds` is a lifted sum type named `Instr`)

 If I don't make it "can't fail", I get this expression instead

 {{{
               case GHC.Prim.dataToTag# @ Instr ds of b# { __DEFAULT ->
               GHC.Prim.tagToEnum#
                 @ Bool
                 (case b# of {
                    __DEFAULT -> 0#;
                    0# -> 1#
                  })
               };
 }}}

 The version with "can't fail" is better because we eliminate a redundant
 `dataToTag#` call, but apparently the resulting expression is not "OK for
 speculation".

 I think this expression is not OK for speculation because the scrutinee is
 lifted, but I'm not sure. I also don't know why lifted scrutinee is a
 problem
 for speculation..

-- 
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/15696#comment:49>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler