Putting a constraint on the MonadIO class actually makes your code
less general rather than more since it prevents your code from ever
being used in a purely functional setting. If you leave out this
constraint, you can actually turn your code into a pure function by
doing things like feeding it a wrapper monad that supplies a list of
strings as input and returns the list of written strings at the end.
The only reason why you would add a MonadIO constraint would be if you
wanted to do other things that require the IO monad (e.g., connecting
to a database, working with files, etc.) and so you wanted to require
that the IO monad be ultimately present somewhere in the type so that
you could use it.
Cheers,
Greg
On Oct 6, 2009, at 8:22 AM, Floptical Logic wrote:
class StreamMonad m where
fetchLine = m
sendLine = String -> m ()
instance StreamMonad IO where
fetchLine = getLine
sendLine = putLine
fetchLineFromStream = lift fetchLine
sendLineToStream = lift . sendLine
This approach makes more sense to me. The equivalent to printing
something to the screen in the Net (IRC) monad is privmsg and it is
easy to make that connection using type classes. I don't see how to
make this connection using MonadIO.
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe
