
#12067: warn-unused-imports does not detect coersions -------------------------------------+------------------------------------- Reporter: crockeea | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.10.3 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: Type of failure: Incorrect | Unknown/Multiple warning at compile-time | Test Case: Blocked By: | Blocking: Related Tickets: #10347 | Differential Rev(s): Wiki Page: | -------------------------------------+------------------------------------- Description changed by crockeea: @@ -34,2 +34,2 @@ - nor is it needed to import instances. In particular, the two implied - suggestions + nor is it needed to (only) import instances. In particular, the two + implied suggestions @@ -37,2 +37,2 @@ - 1. remove the import altogether - 2. change the import to `import Control.Monad.Identity` + 1. remove the import in Main.hs altogether + 2. change the import in Main.hs to `import Control.Monad.Identity ()` New description: When using `coerce`, GHC can erroneously warn that imports are unnecessary. For example: Foo.hs {{{ module Foo (MT(..), M) where import Control.Monad.Identity newtype MT m b = MT (m b) type M b = MT Identity b }}} Main.hs {{{ import Control.Monad.Identity import Data.Coerce import Foo foo :: M [a] -> MT [] a foo = coerce }}} GHC (with `-fwarn-unused-imports`) warns {{{ Main.hs:1:1: Warning: The import of ‘Control.Monad.Identity’ is redundant except perhaps to import instances from ‘Control.Monad.Identity’ To import instances alone, use: import Control.Monad.Identity() }}} But this is incorrect: we need the constructor `Identity` to be in scope in order for the coercion to succeed. Thus the import is not redundant, nor is it needed to (only) import instances. In particular, the two implied suggestions 1. remove the import in Main.hs altogether 2. change the import in Main.hs to `import Control.Monad.Identity ()` both (correctly) result in compile errors. At a minimum, the warning message should be updated to include this possibility, but it would be even better if GHC could detect this and not throw a warning at all. -- -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/12067#comment:2 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler