[GHC] #10447: DeriveFoldable rejects instances with constraints in last argument of data type

[GHC]

#10447: DeriveFoldable rejects instances with constraints in last argument of data
type
-------------------------------------+-------------------------------------
        Reporter:  RyanGlScott       |                   Owner:
            Type:  bug               |                  Status:  new
        Priority:  normal            |               Milestone:
       Component:  Compiler          |                 Version:  7.10.1
      Resolution:                    |                Keywords:
Operating System:  Unknown/Multiple  |            Architecture:
 Type of failure:  GHC rejects       |  Unknown/Multiple
  valid program                      |               Test Case:
      Blocked By:                    |                Blocking:
 Related Tickets:  #8678             |  Differential Revisions:
-------------------------------------+-------------------------------------

Comment (by simonpj):

 OK, but can I ask: can you say precisely what `Foldable` code you would
 expect to get for these two data types?
 {{{
 data T1 a where
   MkT1 :: (a~Int) => a -> T1 a

 data T2 a where
   MkT2 :: Int -> T2 Int

 data T3 a where
   MkT3 :: (a~Int) => a -> T3 Int
 }}}
 All three are equivalent.  But I am guessing that you intend that they are
 '''not equivalent''' to the "deriving `Foldable`" algorithm?  If so, that
 had better be clearly stated in the user manual.

 (Regrettably, the user manual is silent on how `Foldable`, `Traversable`,
 and `Functor` are generated.  It'd be jolly good to have a wiki page that
 explained the deriving algorithm;
 [https://www.haskell.org/onlinereport/haskell2010/haskellch11.html#x18-18200011
 here's how it's done for Eq/Ord etc].)

 The documentation is probably the hard bit.  The change to lift the
 condition is pretty easy: look at `TcDeriv.cond_functorOK`.  (This same
 function actually handles `Traversable` and `Foldable`.  A little
 refactoring would be needed to distinguish the `Foldable` case, perhaps by
 passing the class name instead of a boolean to the function.)

 I'd be happy to review a patch.

 Simon

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10447#comment:5>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler