Tony Morris

@djspiewak @nuttycom nope, it's not F which is universally quantified.

@puffnfresh @nuttycom But it is! I’m talking about LiftIO here. You’re claiming forall F . LiftIO[F] => (F ~> IO), which is impossible.

@djspiewak @nuttycom wow, I did not claim that. I would never.

@puffnfresh @nuttycom Because all LiftIO[F] must be able to define F[Unit] => IO[Unit], and all programs need it, it should be on the class.

@djspiewak @nuttycom not all programs need it.

@puffnfresh @nuttycom Please show me an example program *in Haskell* which uses an IO-based monad trans stack and doesn’t use this function.

@djspiewak @puffnfresh @nuttycom A [anything] program with an IO-based monad trans stack is rubbish. Yes I know they are out there.

@dibblego @puffnfresh @nuttycom It’s not always practical to do that. Always possible, surely, but I’m not sure about practical.