The BoundedEnum thing is already something of a sunk cost, since there's already magic behavior of Enum when a type is also Bounded. (Go look at how Enum deriving is specified.)

On Fri, Jun 1, 2018 at 3:28 PM MarLinn <monkleyon@gmail.com> wrote:

Counterargument: overlapping instances

    instance (Bounded b, Enum b) => Enum (Either a b)
    instance (Bounded b) => Bounded (Either a b)
    instance (Applicative f, Bounded a) => Bounded (f a)
    instance (Bounded a, Enum a) => Enum (Either a b)
    instance (Bounded a) => Bounded (Either a b)
    instance (Bounded a, Enum a, Monoid b) => Enum (a, b)
    instance (Bounded b, Enum b, Monoid a) => Enum (a, b)
Also note that what you're talking about is a special type of objects, namely
    type BoundedEnum a = (Bounded a, Enum a) -- using ConstraintKinds
(I'm sure the mathematicians have a better name for this)

So IF someone where to add these somewhere, might I suggest also adding essentials like
    enumAll :: (Bounded a, Enum a) => [a]
    -- i.e. enumAll :: (BoundedEnum a) => [a]

Lastly, because it's its own type of objects, I'm sure there's a library out there doing just that. (Plus maybe other stuff like EnumMap's).

On 2018-06-01 20:32, Tom Ellis wrote:
True.  I think I would propose

    instance (Bounded a, Bounded b, Enum a, Enum b) => Enum (Either a b)
    instance (Bounded a, Bounded b) => Enum (Bounded a b)
    instance (Bounded a, Bounded b, Enum a, Enum b) => Enum (a, b)

On Fri, Jun 01, 2018 at 02:23:58PM -0400, Li-yao Xia wrote:
One issue is that (Int, Int) is too big to define toEnum/fromEnum.

On 06/01/2018 02:10 PM, Tom Ellis wrote:
I'm a bit surprised that whilst `Either` and `(,)` have instances for `Ord`

* `(,)` has no instance for `Enum`
* `Either` has no instance for `Enum` or `Bounded`

Is there a particular reason for that?  It might be tricky to implement

     toEnum :: Int -> a
     fromEnum :: a -> Int

but in the presence of `Bounded` that should be possible.

_______________________________________________
Haskell-Cafe mailing list
To (un)subscribe, modify options or view archives go to:
http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe
Only members subscribed via the mailman list are allowed to post.


--
brandon s allbery kf8nh                               sine nomine associates
allbery.b@gmail.com                                  ballbery@sinenomine.net
unix, openafs, kerberos, infrastructure, xmonad        http://sinenomine.net