On Sun, Jan 03, 2010 at 10:38:28PM +0100, Francesco Guerrieri wrote:
But doesn't ever happen that poor liftM be confused? and do poor haskell programmers never feel the need to explicitly state to which monad they "wish to lift" ?
If by "explicitly state to which monad" you mean specializing liftM or fmap, then the answer is no. It is not uncommon to see fmap (fmap function) being used to map two layers of structures. If by stating you mean giving a type signature, then yes, it is a good practice to give type signatures to all top-level definitions. The compiler never gets "confused". The only thing that may happen is having more than one possible type that fulfills your needs. The classical example is Num overloading: $ ghci GHCi, version 6.10.4: http://www.haskell.org/ghc/ :? for help Loading package ghc-prim ... linking ... done. Loading package integer ... linking ... done. Loading package base ... linking ... done. Prelude> :s -Wall Prelude> :m Data.Int Prelude Data.Int> let e :: Int; e = 31 Prelude Data.Int> 2^e <interactive>:1:0: Warning: Defaulting the following constraint(s) to type `Integer' `Num t' arising from a use of `^' at <interactive>:1:0-2 In the expression: 2 ^ e In the definition of `it': it = 2 ^ e <interactive>:1:0: Warning: Defaulting the following constraint(s) to type `Integer' `Num t' arising from a use of `^' at <interactive>:1:0-2 In the expression: 2 ^ e In the definition of `it': it = 2 ^ e 2147483648 Prelude Data.Int> 2^e :: Int32 -2147483648 Note that using just '2^e' in the GHCi prompt defaults to Integer. If I say I want Int32 as a type, then the result is completely different, in this case it overflows. Another case when you arrive at those oddities is when dealing with exceptions. The function you use to catch the exceptions need to explicitly say which type of exceptions it wants to catch. Failures to list the correct type will cause the exception to fall through and propagate. HTH, -- Felipe.