Your second option may or may not be dangerous, but it sure looks tricky. I wouldn't use 'maybe' that way. How about this? main = do val <- someIOFunc -- val is a Maybe String maybe (someAction >> someOtherActionEtc) (\theValue -> putStrLn ("the string in reverse is " ++ (reverse theValue)) val If you like point-free definitions: main = do val <- someIOFunc -- val is a Maybe String maybe (someAction >> someOtherActionEtc) (putStrLn . ("the string in reverse is " ++) . reverse) val The difference is that my use of 'maybe' involves giving it actions, and not just pure functions. On Tue, 2011-05-24 at 19:09 -0800, Christopher Howard wrote:
Since I started into Haskell, I frequently run into a scenario like this: some function returns a Maybe value. If it is a Nothing, I want to do one thing, but if it isn't, I want to do something with the value encapsulated by the Maybe. But before I can use that value (say, to print it) I've got to pull it out of the Maybe in a way that satisfies the compiler. So one option is to use a separate function with pattern matching, for example:
handleMaybeVal Nothing = do someAction someOtherActionEtc handleMaybeVal (Just a) = putStrLn ("The string in reverse is " ++ (reverse a))
Or I can use the maybe function with a conditional
main = do val <- someIOFunc -- val is a Maybe String if val == Nothing then do someAction someOtherActionEtc else putStrLn ("The string in reverse is " ++ ((reverse . unpack) val)) where unpack a = maybe (error "unreachable error") (\b -> b) a
The second option I like better, for some reason, but it is a bit dangerous, as I have to double check my code to make sure the default value passed to "maybe" cannot possibly be evaluated. (Alternatively, I could pass in a regular default value, like a blank string, though that is actually more dangerous in my mind because a bad value might be used silently.)
Anyway: how do you pros go about handling situations like these?