hub.darcs.net :: jcpetruzza -> hint -> issue

#7GHC Warnings propagate as exceptions from runInterpreter

vlstill

Hi, when compiling program which gets warning inside runInterpreter, I got

GHC returned a result but said: [GhcError {errMsg = "student30293.hs:6:24:Pattern match(es) are overlappedIn an equation for 216ack217:ack 0 n = ...ack m 0 = ...ack m n = ..."}]

It is very unfortunate that compilation errors are propagated as return value but warnings as exception.

• vlstillWed Sep 10 13:22:44 UTC 2014
  • description updated
• jcpetruzzaTue Feb 3 09:35:07 UTC 2015

  Could you provide a case to reproduce?

• meditansWed Oct 21 10:27:33 UTC 2015

  I'll try to provide a case for this error, just substitute the function testHint in your example file with this one:

  testHint :: Interpreter ()
  testHint = do
    setImportsQ [("Control.Lens", Nothing)]
    say =<< typeOf "(^.)"
    say =<< typeOf "mapOf"

• meditansWed Oct 21 13:28:58 UTC 2015

  It would suffice to modify this function in Hint.Base:

  mayFail :: MonadInterpreter m => m (Maybe a) -> m a
  mayFail action =
      do
          maybe_res <- action
          --
          es <- modifySessionRef ghcErrListRef (const [])
          --
          case (maybe_res, null es) of
              (Nothing,True)  -> throwError $ UnknownError "Got no error message"
              (Nothing,False) -> throwError $ WontCompile (reverse es)
              (Just a, True)  -> return a
              (Just _, False) -> fail $ "GHC returned a result but said: " ++ show es

  handling differently the last case. What problems do you expect if the last line is substituted with:

  (Just a, False)  -> return a    --  ?

  What other errors are possible?

• mvdanWed Mar 9 16:32:13 UTC 2016

  hint is now on GitHub, feel free to open this issue over there or open pull requests with patches:

  https://github.com/mvdan/hint