Jason Dagit wrote:
On 11/23/06, Conor McBride <[EMAIL PROTECTED]> wrote:
*Grr> ifM (Just True) (Just 3) Nothing
Nothing
More care required!
Thank you. Now that you point this out I recall that I've made this
mistake in the past
You and me both. It's really insidious and can hide for weeks, looking
perfectly innocent and
doing all sorts of mad stuff.
Ross and I talk about this issue in our 'Applicative' paper. It's really
what distinguishes monads from applicative functors. Monads let you use
the value from one computation to choose which *computation* to run
next; applicative functors fix the structure of computations but allow
you to do what you like with the *values*.
So, again, I know it's not going to happen in the immediate future, but
I hope we will eventually adopt Ashley's Functor Hierarchy proposal,
then shift the liftM family to the Applicative layer, where they belong.
The fact that the above ifM can be typed with Applicative m, not just
Monad m, is the clue to why it is broken.
All the best
Conor
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe
