No match in record selector Var.tcTyVarDetails

[morrow]

Summary:

ghci> :t fmap :: ((a -> b) -> b) -> (forall c. c -> a)
*** Exception: No match in record selector Var.tcTyVarDetails

I stumbled across this while in ghci with flags:

:set -fglasgow-exts -O2 -fno-print-bind-result
:set -XQuasiQuotes -XTemplateHaskell

Beginning with where I came across it and proceeding in reverse chrono order wrt ghc versions:

[m@ganon feio]$ ghc -V
The Glorious Glasgow Haskell Compilation System, version 6.10.0.20080927

ghci> :t fmap :: ((a -> b) -> b) -> (forall c. c -> a)
*** Exception: No match in record selector Var.tcTyVarDetails

[m@ganon feio]$ ghc -V
The Glorious Glasgow Haskell Compilation System, version 6.9.20080504

ghci> :t fmap :: ((a -> b) -> b) -> (forall c. c -> a)
*** Exception: No match in record selector Var.tcTyVarDetails

[m@ganon feio]$ /usr/bin/ghc-6.8.2 -V
The Glorious Glasgow Haskell Compilation System, version 6.8.2

ghci> :t fmap :: ((a -> b) -> b) -> (forall c. c -> a)

<interactive>:1:0:
    Couldn't match expected type `c'
           against inferred type `f (f b -> b)'
      `c' is a rigid type variable bound by
          the polymorphic type `forall c. c -> a' at <interactive>:1:0
    Probable cause: `fmap' is applied to too few arguments

<interactive>:1:0:
    Couldn't match expected type `a' against inferred type `f b'
      `a' is a rigid type variable bound by
          the polymorphic type
            `forall a b. ((a -> b) -> b) -> forall c. c -> a'
            at <interactive>:1:0
    Probable cause: `fmap' is applied to too few arguments

[m@ganon feio]$ /usr/bin/ghc-6.8.1 -V
The Glorious Glasgow Haskell Compilation System, version 6.8.1

ghci> :t fmap :: ((a -> b) -> b) -> (forall c. c -> a)

<interactive>:1:0:
    Couldn't match expected type `c'
           against inferred type `f (f b -> b)'
      `c' is a rigid type variable bound by
          the polymorphic type `forall c. c -> a' at <interactive>:1:0
    Probable cause: `fmap' is applied to too few arguments

Aside from filing a bug rpt, I'm curious in general about what change to whathaveyou, some way or another, had this as a side-effect.

[morrow]

Judging from the comment, this may have something to do with it? From TcUnify?:

unBox :: BoxyType -> TcM TcType
-- ...
unBox (TyVarTy tv)
  | isTcTyVar tv                                -- It's a boxy type variable
  , MetaTv BoxTv ref <- tcTyVarDetails tv       -- NB: non-TcTyVars are possible
  = do  { cts <- readMutVar ref                 --     under nested quantifiers
-- ...

[morrow]

I just came across ticket #1123, which seems possibly related.

[simonpj]

Thanks for the report. It's a small bug in the new code that deals with impredicativity. Fixed by

Wed Oct 22 15:51:38 GMT Daylight Time 2008  simonpj@microsoft.com
  * Fix Trac #2714 (a minor wibble)

Simon

[igloo]

Merged