
I'm not sure I follow you? The compiler can't reorder the two updates or do them in parallel (IO is not a commutative monad). You might tell the compiler this explicitly, but then are you writing lower and lower level code, further removed from the functional paradigm. Edsko On 4 Nov 2009, at 15:27, David Leimbach wrote:
On Wed, Nov 4, 2009 at 7:11 AM, Edsko de Vries
wrote: On 4 Nov 2009, at 13:36, Alberto G. Corona wrote:
Artyom.
I know what uniqueness means. What I meant is that the context in which uniqueness is used, for imperative sequences:
(y, s')= proc1 s x (z, s'')= proc2 s' y .....
is essentially the same sequence as if we rewrite an state monad to make the state explicit. When the state is the "world" state, then it is similar to the IO monad.
Yes, as long as there is a single thing that is being updated there's little difference between the state monad and a unique type. But uniqueness typing is more general. For instance, a function which updates two arrays
f (arr1, arr2) = (update arr1 0 'x', update arr2 0 'y')
is easily written in functional style in Clean, whereas in Haskell we need to sequentialize the two updates:
f (arr1, arr2) = do writeArray arr1 0 'x' writeArray arr2 0 'y'
Those sequential updates can be run concurrently on both, just with different syntax though right?
You can find a more detailed comparison in my thesis (https://www.cs.tcd.ie/Edsko.de.Vries/pub/MakingUniquenessTypingLessUnique-sc... , Section 2.8.7).
-Edsko
_______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe