[GHC] #9220: type roles for unboxed arrays

[GHC]

#9220: type roles for unboxed arrays
-------------------------------------+-------------------------------------
              Reporter:  rwbarton    |            Owner:  goldfire
                  Type:  bug         |           Status:  new
              Priority:  normal      |        Milestone:  7.10.1
             Component:  libraries   |          Version:  7.8.1
  (other)                            |         Keywords:
            Resolution:              |     Architecture:  Unknown/Multiple
      Operating System:              |       Difficulty:  Unknown
  Unknown/Multiple                   |       Blocked By:
       Type of failure:              |  Related Tickets:
  None/Unknown                       |
             Test Case:              |
              Blocking:              |
Differential Revisions:              |
-------------------------------------+-------------------------------------

Comment (by rwbarton):

 Replying to [comment:6 goldfire]:
 > If I understand correctly, then, a representational role will allow
 users to get out-of-bounds errors, right? This would happen in the
 contrived scenario in comment:3.

 You won't actually get an out-of-bounds ''error'', you'll simply read
 outside the memory area allocated to the array (which could cause a
 segfault, or leak other data, or just return nonsense).

 Regarding `UArray` specifically, I believe it's true that all the
 different existing instances of `IArray UArray` are for non-
 representationally equal types (although under the hood many are actually
 represented identically, e.g. `data Word8 = W8# Word#`, `data Word16 =
 W16# Word#`; I believe these are still not inter-`coerce`-ible, though).
 And in order to write your own `IArray` instance you have to define
 functions with names like `unsafeAt`. So a representational role for
 `UArray` would be justifiable if ensuring that the in-array representation
 is identical is considered to be part of the contract when defining an
 `IArray` instance for a newtype. (One that would be satisfied
 automatically by GND.)

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9220#comment:8>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler