 Codata and Comonads in Haskell
started 8/13/2003; 5:30:09 AM - last post 8/19/2003; 4:20:41 AM
|
|
Ehud Lamm - Codata and Comonads in Haskell 
8/13/2003; 5:30:09 AM (reads: 2273, responses: 2)
|
|
| Codata and Comonads in Haskell |
|
Codata and Comonads in Haskell. Richard B. Kieburtz
This paper suggests that a useful interpretation of comonads in programming is that they account for effects originating in the context of a program fragment, effects to which the program fragment may react and which it may further propagate.
I cam across this paper browsing the Haskell Wiki, and it looks pretty interesting.
I am not sure I fully grasp the utility of comonads, but maybe someone more knowledgeable will care to comment.
Editors, we need to post some non-FP links quickly!
Posted to functional by Ehud Lamm on 8/13/03; 5:32:35 AM
|
|
|
|
Pseudonym - Re: Codata and Comonads in Haskell
8/17/2003; 6:07:12 PM (reads: 450, responses: 0)
|
|
|
Codata is a very powerful idea, and it's used (partly unknowingly) in a lot of APIs, both declarative and non-declarative. Take, for example, the Java Iterator object:
interface Iterator {
public boolean hasNext();
public Object next();
};
(It's been a while since I used any Java. Sorry if I got the syntax wrong.)
This is an implementation of the coalgebraic formulation of (possibly infinite) streams:
<A_inf, f: A_inf -> 1 + (A x A_inf)>
On comonads, though, note this thread from the Haskell mailing list. We still don't know if comonads break referential transparency without some kind of linear type constraint or not.
|
|
Dave Herman - Re: Codata and Comonads in Haskell
8/19/2003; 4:20:41 AM (reads: 418, responses: 0)
|
|
|
Am I going cross-eyed or is there a typo on page 5 in the definition of the State comonad? He defines co-ext as:
co-ext f = (λ(g, s) → (λs′ → f (g, s′)), s)
but I think it should be:
co-ext f = λ(g, s) → (λs′ → f (g, s′), s)
|
|
|
|