[GHC] #3806: Type checking errors in foreign declarations reported at wrong locations

[GHC]

#3806: Type checking errors in foreign declarations reported at wrong locations
---------------------------+------------------------------------------------
  Reporter:  asuffield     |          Owner:                  
      Type:  bug           |         Status:  closed          
  Priority:  normal        |      Milestone:                  
 Component:  Compiler      |        Version:  6.12.1          
Resolution:  invalid       |       Keywords:                  
Difficulty:                |             Os:  Unknown/Multiple
  Testcase:                |   Architecture:  Unknown/Multiple
   Failure:  None/Unknown  |  
---------------------------+------------------------------------------------
Changes (by simonpj):

  * status:  new => closed
  * difficulty:  =>
  * resolution:  => invalid

Comment:

 I see your point, but  your follow-up remark hits the nail on the head.
 It's fine, for example, to say this:
 {{{
 module M where

 import Foo( foo1 )

 newtype Blip = MkBlip { foo2 :: CInt }

 class Flop a where
   foo3 :: a -> IO ()
 instance Flop CInt where ...

 foreign export ccall foo1 :: CInt -> IO ()
 foreign export ccall foo2 :: Blip -> IO ()
 foreign export ccall foo3 :: CInt -> IO ()
 }}}
 All three are fine.  See Section 3.4 in
 http://www.cse.unsw.edu.au/~chak/haskell/ffi/ffi/ffise3.html#x6-100003.

 In short, a `foreign export` is not a type signature at all.  It's a
 ''use'' of `foo`.

 An alternative design would be:
  * `foreign export foo` really is a type signature for `foo`
  * A `foreign export foo` must be accompanied by a top-level definition of
 `foo` in the same module.
 On the whole I think that might be a better design.  It would be simpler,
 and I don't think the above cases are important in practice.  But that
 would mean changing the FFI spec.

 I'll close the GHC ticket as invalid, but there's a possible discussion to
 be had about what the FFI spec should say.

 Simon

-- 
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/3806#comment:2>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler