Using -> as both type and value constructor
Studying the paper *A Simple Implementation for Priority Search Queues*, by Ralf Hinze, I came across the following syntax that I didn't understand and I couldn't use in GHCi 7.0.3 for defining a binding data type (page 3): Bindings are represented by the following data type: *data k -> p = k -> p* key :: (k -> p) -> k key (k -> p) = k prio :: (k -> p) -> p prio (k -> p) = p Note that we use '->' both as a type constructor and value constructor. [...] data PSQ k p The following page has these value constructors: *0 :: PSQ k p* *{.} :: (k -> p) -> PSQ k p* The paper says that's Haskell '98 code. I'll appreciate info on what kind of data declaration those are and any pointers to related documentation. Or maybe those are deprecated features. I'm stuck in this part. -armando www.cs.ox.ac.uk/people/ralf.hinze/talks/ICFP01.pdf
On Wednesday 24 August 2011, 20:24:14, Armando Blancas wrote:
Studying the paper *A Simple Implementation for Priority Search Queues*, by Ralf Hinze, I came across the following syntax that I didn't understand and I couldn't use in GHCi 7.0.3 for defining a binding data type (page 3):
Bindings are represented by the following data type: *data k -> p = k -> p* key :: (k -> p) -> k key (k -> p) = k prio :: (k -> p) -> p prio (k -> p) = p Note that we use '->' both as a type constructor and value constructor. [...] data PSQ k p
The following page has these value constructors:
*0 :: PSQ k p* *{.} :: (k -> p) -> PSQ k p*
The paper says that's Haskell '98 code.
I didn't see that claim in the linked slides, and it's not Haskell '98 (nor Haskell 2010). There's a translation of the principles to Haskell from page 17 on or so.
I'll appreciate info on what kind of data declaration those are and any pointers to related documentation. Or maybe those are deprecated features.
I don't think the notation from the beginning ever was legal Haskell, I think it was chosen to present the ideas separated from syntax.
I'm stuck in this part.
-armando
www.cs.ox.ac.uk/people/ralf.hinze/talks/ICFP01.pdf
I didn't see that claim in the linked slides, and it's not Haskell '98 (nor Haskell 2010).
I didn't realize it linked to the slides; I thought that pointed to the article. I just found another version of the paper, A Simple Implementation Technique for Priority Search Queues, by Hinze, which shows the binding type as a pair, with access functions for k and p. I'll continue to work my way through the material there. Thanks for you response.
On 8/24/11 5:03 PM, Armando Blancas wrote:
I didn't see that claim in the linked slides, and it's not Haskell '98 (nor Haskell 2010).
I didn't realize it linked to the slides; I thought that pointed to the article. I just found another version of the paper, A Simple Implementation Technique for Priority Search Queues, by Hinze, which shows the binding type as a pair, with access functions for k and p. I'll continue to work my way through the material there. Thanks for you response.
With -XTypeOperators you can define data a :-> b = a :-> b where (:->) is both a type constructor and a data constructor. Note the leading colon which is the only "capital" punctuation letter. Hinze may have just chosen to elide the colon for prettiness. -- Live well, ~wren
Thanks for the info; that's good to know. The ICFP '01 version uses pairs;
not sure when the other came out or where.
On Wed, Aug 24, 2011 at 4:46 PM, wren ng thornton
On 8/24/11 5:03 PM, Armando Blancas wrote:
I didn't see that claim in the linked slides, and it's not Haskell '98
(nor Haskell 2010).
I didn't realize it linked to the slides; I thought that pointed to the article. I just found another version of the paper, A Simple Implementation Technique for Priority Search Queues, by Hinze, which shows the binding type as a pair, with access functions for k and p. I'll continue to work my way through the material there. Thanks for you response.
With -XTypeOperators you can define
data a :-> b = a :-> b
where (:->) is both a type constructor and a data constructor. Note the leading colon which is the only "capital" punctuation letter. Hinze may have just chosen to elide the colon for prettiness.
-- Live well, ~wren
______________________________**_________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/**mailman/listinfo/haskell-cafehttp://www.haskell.org/mailman/listinfo/haskell-cafe
participants (3)
-
Armando Blancas -
Daniel Fischer -
wren ng thornton