17 Oct
2009
17 Oct
'09
5:32 p.m.
Daniel Fischer wrote:
Am Samstag 17 Oktober 2009 22:24:08 schrieb Andrew Coppin:
Edward Z. Yang wrote:
Excerpts from Andrew Coppin's message of Sat Oct 17 15:21:28 -0400 2009:
Suppose we have
newtype Foo x instance Monad Foo runFoo :: Foo x -> IO x
Are you sure that's really the type you want for runFoo ? By the definition below, runFoo :: Foo x -> M -> IO x would be the natural type.
M is a type containing a bunch of IORefs. runFoo generates M, feeds M to Foo x, and then does magical stuff with M afterwards.