Skip to content
GitLab
Projects Groups Snippets
  • /
  • Help
    • Help
    • Support
    • Community forum
    • Submit feedback
  • Sign in / Register
  • GHC GHC
  • Project information
    • Project information
    • Activity
    • Labels
    • Members
  • Repository
    • Repository
    • Files
    • Commits
    • Branches
    • Tags
    • Contributors
    • Graph
    • Compare
    • Locked Files
  • Issues 5,398
    • Issues 5,398
    • List
    • Boards
    • Service Desk
    • Milestones
    • Iterations
  • Merge requests 590
    • Merge requests 590
  • CI/CD
    • CI/CD
    • Pipelines
    • Jobs
    • Schedules
    • Test Cases
  • Deployments
    • Deployments
    • Releases
  • Analytics
    • Analytics
    • Value stream
    • CI/CD
    • Code review
    • Insights
    • Issue
    • Repository
  • Wiki
    • Wiki
  • Snippets
    • Snippets
  • Activity
  • Graph
  • Create a new issue
  • Jobs
  • Commits
  • Issue Boards
Collapse sidebar
  • Glasgow Haskell CompilerGlasgow Haskell Compiler
  • GHCGHC
  • Issues
  • #10715
Closed
Open
Issue created Jul 31, 2015 by inaki@trac-inaki

Possible regression in Coercible a (X a) between 7.8 and 7.10

In upgrading to7.10, code of the form

{-# LANGUAGE FlexibleContexts #-}
import Data.Coerce (coerce, Coercible)

data X a

doCoerce :: Coercible a (X a) => a -> X a
doCoerce = coerce

fails to compile in 7.10.1 and 7.10.2 with the error

testCoerce.hs:6:13:
    Could not deduce (a ~ X a)
    from the context (Coercible a (X a))
      bound by the type signature for
                 doCoerce :: Coercible a (X a) => a -> X a
      at testCoerce.hs:6:13-41
      ‘a’ is a rigid type variable bound by
          the type signature for doCoerce :: Coercible a (X a) => a -> X a
          at testCoerce.hs:6:13
    Relevant role signatures: type role X phantom
    In the ambiguity check for the type signature for ‘doCoerce’:
      doCoerce :: forall a. Coercible a (X a) => a -> X a
    To defer the ambiguity check to use sites, enable AllowAmbiguousTypes
    In the type signature for ‘doCoerce’:
      doCoerce :: Coercible a (X a) => a -> X a

while it works in 7.8.4.

Surprisingly (to me at least), the code works in 7.10.1 and 7.10.2 if I change it to

{-# LANGUAGE FlexibleContexts #-}
import Data.Coerce (coerce, Coercible)

data X a

doCoerce :: Coercible a (X b) => a -> X a
doCoerce = coerce

while it fails to compile in 7.8.4 with the error

testCoerce.hs:6:13:
    Could not coerce from ‘a’ to ‘X b0’
      because ‘a’ and ‘X b0’ are different types.
      arising from the ambiguity check for ‘doCoerce’
    from the context (Coercible a (X b))
      bound by the type signature for
                 doCoerce :: Coercible a (X b) => a -> X a
      at testCoerce.hs:6:13-41
    The type variable ‘b0’ is ambiguous
    In the ambiguity check for:
      forall a b. Coercible a (X b) => a -> X a
    To defer the ambiguity check to use sites, enable AllowAmbiguousTypes
    In the type signature for ‘doCoerce’:
      doCoerce :: Coercible a (X b) => a -> X a

The coercion pattern may look a bit funny, but it is rather useful when one has newtypes of the form

newtype Y = Y (ForeignPtr Y)

which appear naturally when writing bindings to C libraries, and one wants to get access to the underlying ForeignPtr from Y (here X is ForeignPtr). The relevant Coercible instance here is Coercible Y (ForeignPtr Y), as above.

I would have expected the version with context "Coercible a (X a)" to be accepted, as 7.8.4 does, since it seems to be a specialization of the more general coerce, but maybe I am missing something?

Edited Mar 10, 2019 by inaki
To upload designs, you'll need to enable LFS and have an admin enable hashed storage. More information
Assignee
Assign to
Time tracking