[GHC] #10875: Unexpected defaulting of partial type signatures and inconsistent behaviour when -fdefer-typed-holes is set.

[GHC]

#10875: Unexpected defaulting of partial type signatures and inconsistent behaviour
when -fdefer-typed-holes is set.
-------------------------------------+-------------------------------------
        Reporter:  holzensp          |                   Owner:
            Type:  bug               |                  Status:  new
        Priority:  normal            |               Milestone:
       Component:  Compiler          |                 Version:  7.10.2
      Resolution:                    |                Keywords:
                                     |  PartialTypeSignatures TypedHoles
Operating System:  MacOS X           |            Architecture:  x86_64
 Type of failure:  Incorrect         |  (amd64)
  warning at compile-time            |               Test Case:
      Blocked By:                    |                Blocking:
 Related Tickets:                    |  Differential Revisions:
-------------------------------------+-------------------------------------

Comment (by holzensp):

 If you're done with the refactoring and want someone to help test, by all
 means let me know.

 I see your point with regards to the defaulting rule from the language
 definition. In light of the typing rules, I'm unsure whether I would
 consider a typed hole to be like any other term. With a variable-typed
 term, like `undefined`, I agree that the typing rules should find
 ''whatever'' type satisfies. My interpretation of a typed hole, however,
 is that I'm asking GHC: "Please tell me what you expect me to put in
 here."

 After typing up my comment, though, I understand that `RealFrac b =>
 a_bound_by_the_type_of_foo -> b` actually is not the correct type, because
 `round` actually binds `b`. In other words, it's not just that `round`
 imposes a `RealFrac` constraint on `b`, but the `b` of the hole should be
 precisely the same `b` as that of `round`. Somehow it should be something
 like `RealFrac b_bound_by_the_type_of_round => a_bound_by_the_type_of_foo
 -> b_bound_by_the_type_of_round`.

 I admit it's complicated, but purely from a TypedHoles user perspective, I
 think defaulting is... less then intuitive (fully aware of the
 subjectivity of that term).

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10875#comment:4>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler