[GHC] #10532: Unnecessary constraints when using overlapping instances

[GHC]

#10532: Unnecessary constraints when using overlapping instances
-------------------------------------+-------------------------------------
        Reporter:  MikeIzbicki       |                   Owner:
            Type:  bug               |                  Status:  closed
        Priority:  normal            |               Milestone:
       Component:  Compiler (Type    |                 Version:  7.10.1
  checker)                           |                Keywords:
      Resolution:  invalid           |            Architecture:
Operating System:  Unknown/Multiple  |  Unknown/Multiple
 Type of failure:  None/Unknown      |               Test Case:
      Blocked By:                    |                Blocking:
 Related Tickets:                    |  Differential Revisions:
-------------------------------------+-------------------------------------
Changes (by simonpj):

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


Comment:

 Consider what happens, in your modified code with overlapping instances,
 when you call `f` thus: `f (4::Int)`.  What result do you expect?  You
 might say:

  1. The instance `Foo Int` clearly says that `foo` is the identity
 function on `Int`.  I'm giving it an `Int` so `f 4` should return `4`.

  2. Or you might say: when I compiled the code for `f`, GHC had to choose
 which instance to use, of the two available. Since the constraint is `Foo
 a`, and GHC knows nothing about `a` it must choose the `Foo a` instance.
 Result, `f 4` returns 5.

 This incoherence is what GHC is complaining about.  If it can't make a
 unique choice, it decides not to decide, and rejects the program instead.
 If it silently chose (2), which I think you intend, then if you replaced
 the call `(f 4)` with the result of inlining `f`, thus `(foo 4)`, the
 result of the program would change.

 This has been GHC's behaviour (and Hugs) ever since we introduced
 overlapping instances.  The flag `-XIncoherentInstances` switches off this
 behaviour, and does (2).

 Simon

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10532#comment:7>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler