[Haskell-cafe] Re[2]: strict Haskell dialect

Brian Hulley brianh at metamilk.com
Sun Feb 5 13:11:06 EST 2006


Bulat Ziganshin wrote:
> Hello Brian,
>
> Saturday, February 04, 2006, 4:50:44 AM, you wrote:
>
>>>> One question is how to get some kind of "do" notation that would
>>>> work well in a strict setting.
>>>> The existing "do" notation makes use of lazyness in so far as the
>>>> second arg of  >> is only evaluated when needed. Perhaps a new
>>>> keyword such as "go" could be used to use >>= instead ie:
>
>> If strictness was the default (eg if the language were ML not
>> Haskell), then in
>
>>              putStr "hello" >> putStr (show 1)
>
>> both args to >>> would be evaluated before >> was called. Thus
>> putStr (show 1) would be evaluated before the combined monad is
>> actually run, which would be wasteful if we were using a monad with
>> a >> function that only runs the rhs conditionally on the result of
>> the lhs.
>> If Haskell were a strict language I think an equivalent for the do
>> notation would have to lift everything (except the first expression)
>> and use >>= instead of >>> .
>
> it seems that you misunderstand the monads (or may be i misunderstand
> :)
>
> each and every monadic operation is a function! type "IO a" is really
> "RealWorld -> (RealWorld,a)" and the same for any other monad. concept
> of the monad by itself means carrying "hidden" state from one monadic
> operation to the next. that allows to _order_ monadic operations plus
> this state used for zillions other things, including state, logs,
> fails and so on, so on

exp1 >> exp2 in a strict setting would force exp1 to be evaluated to a 
monad, exp2 to be evaluated to a monad, then these monads to be combined 
using >> into another monad, which at some later point would actually be 
run. But it is this eager evaluation of exp2 into the rhs monad that is the 
problem, because in the example above, (show 1) would be evaluated during 
the evaluation of (putStr "hello" >> putStr (show 1)) whereas in Haskell it 
would only be evaluated when the combined monad is actually run (because it 
is only at this point that Haskell actually creates the combined monad from 
the thunk).

Regards, Brian.



More information about the Haskell-Cafe mailing list