Passing -fdefer-type-errors to GHC causes confusing errors from LH about liquidhaskell-tutorial

[mgritter]

I ran into this problem during the tutorial too -- due to a missing pair of parentheses -- and removing -fdefer-type-errors quickly clarified what the problem was. Without this change, I got the extremely confusing message in part 2:

**** LIQUID: ERROR :1:1-1:1: Error
  elaborate solver elabBE 294 "lq_anf$##7205759403792806839##d3rx" {lq_tmp$x##1878 : (GHC.Types.Any (GHC.Prim.TYPE GHC.Types.LiftedRep)) | [(lq_tmp$x##1878 = lq_anf$##7205759403792806838##d3rw);
                                                                         (lq_tmp$x##1878 = (GHC.Num.fromInteger lq_anf$##7205759403792806838##d3rw));
                                                                         (lq_tmp$x##1878 = (GHC.Num.fromInteger lq_anf$##7205759403792806838##d3rw))]} failed on:
      ((lq_tmp$x##1878 == lq_anf$##7205759403792806838##d3rw
        && lq_tmp$x##1878 == GHC.Num.fromInteger lq_anf$##7205759403792806838##d3rw)
       && lq_tmp$x##1878 == GHC.Num.fromInteger lq_anf$##7205759403792806838##d3rw)
  with error
      The sort (GHC.Types.Any (GHC.Prim.TYPE GHC.Types.LiftedRep)) is not numeric
  because
Cannot unify (GHC.Types.Any (GHC.Prim.TYPE GHC.Types.LiftedRep)) with int in expression: lq_tmp$x##1878 == lq_anf$##7205759403792806838##d3rw 
  because
Elaborate fails on lq_tmp$x##1878 == lq_anf$##7205759403792806838##d3rw
  in environment
      GHC.Num.fromInteger := func(1 , [int; @(0)])

      lq_anf$##7205759403792806838##d3rw := int

      lq_tmp$x##1878 := (GHC.Types.Any (GHC.Prim.TYPE GHC.Types.LiftedRep)) 

[ranjitjhala]

Wow thanks for catching this Mark! I think we should totally remove that flag (tbh can’t recall why it was put in in the first place!)

[matthew-healy]

@ranjitjhala I opened #120 a while ago to fix this. As far as I can tell removing the flag doesn't have any impact on the tutorial whatsoever.