Hi Günther, from the semantical point of view, you can replace
let x = e' in e by (\x -> e) e' Both should evaluate to the same thing.
However, from the typing point of view, it makes quite a difference. It is an integral part of the Hindley-Milner type inference algorithm, which is used by many functional languages. Consider the following two expressions:
f = (\x -> x x) (\y -> y) g = let x = \y -> y in x x
The function "f" is not typable in the Hindley-Milner type system, while "g" is is (and its type is "a -> a"). The reason is that in the first case (f), the typing system tries to assign a single type to "x", which is impossible (one would have to unify types "a" and "a -> b"). In the second case (g), the typing system assigns "x" a polymorphic type schema forall a.a -> a which can be instantiated to both "a -> a" (the second "x") and "(b -> b) -> (b -> b)" (the first "x"), and then applied together. I'd say that "let ... in ..." is a core feature of the language that gives us parametric polymorphism. If you are interested in the details, I'd suggest you to read the original paper [damas1982principal]. Best regards, Petr @conference{damas1982principal, title={{Principal type-schemes for functional programs}}, author={Damas, L. and Milner, R.}, booktitle={Proceedings of the 9th ACM SIGPLAN-SIGACT symposium on Principles of programming languages}, pages={207--212}, year={1982}, organization={ACM}, url = {http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.122.3604&rep=rep1&type=pdf} } On Tue, Nov 02, 2010 at 02:34:29AM +0100, Günther Schmidt wrote:
Hi all,
is there something special about "let"? I don't mean only its use in haskell, but in the general context of programming languages.
I've been given a few hints over time when I asked question concerning DSLs but regretfully didn't follow them up.
Günther
_______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe