[Haskell-cafe] Curious data family bug

[Michael Snoyman]

Hey all,

While trying to get a commit pushed for Yesod[1], Alexander Dunlap
pointed out one of his programs didn't work with the new code. After
some investigation, I was able to reproduce the bug with the following
code snippet:

{-# LANGUAGE TypeFamilies #-}
data family Foo a
data Bar = Bar
data instance Foo Bar
    = Bar1 | Bar2 | Bar3 | Bar4 | Bar5 | Bar6 | Bar7 | Bar8 | Bar9
    deriving Eq

This produces:

    Couldn't match expected type `Main.R:FooBar'
           against inferred type `Foo Bar'
      NB: `Foo' is a type function
    In the first argument of `Main.$con2tag_R:FooBar', namely `a'
    In the expression: (Main.$con2tag_R:FooBar a)
    In the expression:
        case (Main.$con2tag_R:FooBar a) of {
          a#
            -> case (Main.$con2tag_R:FooBar b) of {
                 b# -> (a# GHC.Prim.==# b#) } }

The especially strange thing about this bug is that it only occurs
when there are more than 8 constructors; if I remove Bar9, everything
seems to work. Does anyone have experience with this occuring?

Michael

[1] http://docs.yesodweb.com/blog/please-break-yesod/