On Sun, Aug 23, 2009 at 5:21 PM, Jeremy Shaw
Also, you could define a forkM function like this:
forkM :: M () -> M ThreadId forkM (M r) = M $ mapReaderT forkIO r
which could be used like this:
test = runM "foo" $ do forkM $ loop callback liftIO $ print "here"
If we were to expand forkM in test, we would get something like:
test' = runM "foo" $ do env <- ask liftIO $ forkIO (runM env $ loop callback) liftIO $ print "here"
So, this does not change the 'rule' of forkIO having to come before a runM. It just wraps it up nicely.
Somehow I missed this e-mail when I replied last time. Now this is actually the solution to my problem. I didn't think of a second runM with the same environment. Brilliant (OTOH it probably highlights the fact that I am only a Monad novice). Many thanks Jeremy. You saved my day! Cheers, Levi