8 Mar
2016
8 Mar
'16
2 p.m.
On 03/08/2016 02:26 AM, Dan Burton wrote:
I would prefer that the Ord constraint be retained in the type signature, even if not used in the implementation. Sets and Maps conceptually do not have an ordering; the Ord constraint indicates in which order one is sequencing the values.
I don't quite understand the essence of your argument, but what you're proposing (annotating every function that deals with Sets or Maps with an Ord constraint) sounds like data type contexts[1], which was more or less universally admitted to be a bad idea. Could you explain in more detail why you want that? [1]: https://wiki.haskell.org/Data_declaration_with_constraint Roman