[Haskell-cafe] "SameConstraints?" type constraints

[Emil Axelsson]

Den 2015-06-22 09:26, Emil Axelsson skrev:
> Den 2015-06-19 22:13, Leza Morais Lutonda skrev:
>> On 06/19/2015 08:44 AM, Sean Leather wrote:
>>> Note this won't work for a Num instance mentioned earlier because the
>>> existentially quantified d types in two SC values are not provably the
>>> same type. In other words, you can't write
>>>
>>> instance Num e => Show (S Num e) where
>>>     SC x1 y1 + SC x2 y2 = SC (x1 + x2) (y1 + y2)
>>>
>>> because x1 and x2 can have different types.
>> Another issue is: what if I want to constraint the type `e` to more
>> classes and make `d` to have the same constrains? I have to re-declare
>> the `S` data type like?:
>>
>> data S c1 c2 ... cN e where
>>
>>      SC :: (c1 d, c2 d, ..., cN d) -> d -> e -> S c1 c2 ... cN e
>>
>>
>> Does anyone ever needed such a feature?
>
> Yes, have a look at the paper "Deconstraing DSLs":
>
>    http://dl.acm.org/citation.cfm?id=2364571
>
> See also the constraint product (&&&) from the `constraints` package.

Or you might find this interface easier to work with:

 
https://github.com/emilaxelsson/imperative-edsl/blob/master/src/Data/TypePredicates.hs

It's based on type predicates (`* -> Constraint`), so you can write e.g. 
`(c1 /\ c2 /\ ...  /\ cN) d => ...`. It also implements "sub-classing" 
using a technique similar to Data Types à la Carte.

/ Emil