The difference with liftBase is that before lifting the base computation
#11125: Typechecker can't infer StM m Bool ~ Bool from StM m a ~ a -------------------------------------+------------------------------------- Reporter: nikomi | Owner: Type: bug | Status: closed Priority: normal | Milestone: Component: Compiler (Type | Version: 7.10.2 checker) | Resolution: invalid | Keywords: Operating System: Unknown/Multiple | Architecture: | Unknown/Multiple Type of failure: None/Unknown | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Rev(s): Wiki Page: | -------------------------------------+------------------------------------- Changes (by nomeata): * status: new => closed * resolution: => invalid Comment: More relevant types: {{{ liftBaseWith :: (RunInBase m b -> b a) -> m a type RunInBase m b = forall a. m a -> b (StM m a) }}} so the `runInST` function in your code changes the return type from `a` to `StM m a`. Note the docs: liftBaseWith captures the state of m. It then provides the base computation with a RunInBase function that allows running m computations in the base monad on the captured state. I’m confident that GHC is doing the right thing here, and it’s just `monad-control`’s API that is confusing here. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/11125#comment:5 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler