Re: Improving the instances of Data.Functor.{Product,Sum}

Big +1 to fixing these issues. https://github.com/phadej/some/pull/21 is a PR I'd like to see merged, but which cannot build because of base shunning flexible instances. We at Obsidian frequently use `Sum` and `Product` (or when it usually fails their `Data.Generic` equivalents, `:+:` and `:*:`) with GADTs that are *not* functors, and anything but the `Flexible*` instances  prevents that. (Yup, `QuantifiedConstraints` instead of *1 is also no good.) Do note that none of specific problems with `Flexible*` raised in this seem to apply in this case (no multi-param type class, no missing `~`, etc.), only the general wariness. As mentioned, the `Data.Generics` ones get it the way I want, and I do try to use them accordingly, but if we can be "right" there why can't we be right here? (And wouldn't it be nice of one was a newtype or alias of the other...) If we can all do that, secondarily I would actually like to *keep* the *1 classes, but give them `QuantifiedConstraints` super-classes:

(forall a. Eq (f a)) => Eq1 f

It doesn't look like this has been brought up yet, but the idea is that one implements a *1 class because of the extra power that the lift* methods provide----the functionality on the type parameter need not be canonical---not as a crude hack around the lack of `QuantifiedConstraints`. I would also advocate writing a `Lift1` for this reason: "shallowly quoting" a functor of syntax (i.e. `liftLift id`) is quite useful and very much in the spirit of the lisp quasi-quoting that our quoting and splicing is based one. Finally, while the change is breaking, no CPP should be needed and there is less room for user-error as the new quantified constraint super classes forces anything which used the old instances to also support the new instances---you can unthinkingly fix the the type errors and end up with the right instances for old and new base. John P.S. Maybe one might argue that the the new ret-conned use for the *1 classes is much more niche so they belong outside of base. I always like anything to do with breaking up base, so that's fine with me too. Just don't say they are altogether useless because of `QuantifiedConstraints`. On 3/15/20 4:43 PM, Carter Schonwald wrote:

as an additional point of data, https://github.com/ekmett/bytes/pull/49#issuecomment-580924670 and the indirectly linked ticket https://gitlab.haskell.org/ghc/ghc/issues/17767#note_251123 is an example of how not trivial a pretty simple use of quantified constraints / related machinery can be!

a good "guiding light" for core libraries in haskell perhaps should be "what choices jointly give the best type inference, composability, generality, performance, and usability" and when we hit a trade off for these parameters, make sure we understand those and have very well understood tradeoffs.

@daniel  for these sorts of instance renovation proposals, its probably best to show case "heres the type  and instance and uses" in each of the K different flavors and what gets better/easier/harder simpler.  I honestly dont do it as often as I should myself!

In this case, putting together a tiny repo with different module doing the various flavors and how they work might be best for grounding this. bonus points if its cabalized so that folks on different lagging rates of ghc versions can poke around :)

(one idea i've been thinking about is how to best make various phases of possibly nontrivial proposals easy to evaluate and compare for impact and benefit, but thats a discussion for another time)

as ever, be well all on this bizarre spring we're all experiencing

-Carter

On Sat, Mar 14, 2020 at 11:44 AM Oleg Grenrus mailto:oleg.grenrus@iki.fi> wrote:

QuantifiedConstraints are not buggy, but they are not _complete_ (I do mean that as "buggy" = not sound, properly expressive = complete).

There are at least three issues:

1. Instance definition need UndecidableInstances (in my opinion this is big deal) 2. Instances are not elegant (easy to write, but not elegant). 3. QuantifiedConstraints resists to be abstracted over

See https://gist.github.com/phadej/266d68cf5cc1229c3548b7965f4335f8 for the standalone code file.

For the reasons below I wouldn't recommend using QuantifiedConstraints. I very like them, but I'm not convinced the feature is ready for "prime time". To put into perspective, I think code classes of `singletons` are more ready to be included in `base` than changing instances of Data.Functor.Sum and .Product. I do use singletons in my code more than QuantifiedConstraints. :)

I'm very worried how this change will affect libraries like `free` and `recursion-schemes` and what builds on top of them. This is not only change to `base`, it strongly guides how downstream libraries should be written (or changed) as well.

On the other hand, I don't feel strongly about

    instance (Eq (f a), Eq (g a)) => Eq (Product f g a)

Yet, in the light of `free` it is "a step backwards". See https://hackage.haskell.org/package/free-5/changelog

---

Simple example with `newtype Fix f = Fix (f (Fix f))`

    class Eq1 f where         liftEq :: (a -> b -> Bool) -> f a -> f b -> Bool

    class Eq1 f => Ord1 f where         liftCompare :: (a -> b -> Ordering) -> f a -> f b -> Ordering

    instance Eq1 f => Eq (Fix f) where         (==) = eq where eq (Fix x) (Fix y) = liftEq eq x y

    instance Ord1 f => Ord (Fix f) where         compare = cmp where cmp (Fix x) (Fix y) = liftCompare cmp x y

works. It's boilerplate, but it's already written.

However, if we want to use QuantifiedConstraints, then we

1. need UndecidableInstances 2. and then the code turns out to be less elegant:

instance (forall x. Eq x => Eq (f x)) => Eq (Fix f) where     (==) = eq where eq (Fix x) (Fix y) = x == y

instance (forall x. Ord x => Ord (f x)) => Ord (Fix f) where     compare = cmp where cmp (Fix x) (Fix y) = compare x y

fails to compile with

    GHCi, version 8.8.3: https://www.haskell.org/ghc/ :? for help     [1 of 1] Compiling Main             ( Ord1.hs, interpreted )

    Ord1.hs:28:10: error:         • Could not deduce (Ord x)             arising from the superclasses of an instance declaration

we need to write Ord instance differently

   instance (forall x. Ord x => Ord (f x), forall x. Eq x => Eq (f x)) => Ord (Fix f) where        compare = cmp where cmp (Fix x) (Fix y) = compare x y

This _cannot_ be made to work:

    forall x. Ord x => Ord (f x)

doesn't entail

    forall x. Eq x => Eq (f x)

If you try to write defaultLiftEq using liftCompare

    defaultLiftEq :: Ord1 f => (a -> b -> Bool) -> f a -> f b -> Bool     defaultLiftEq eq x y = EQ == liftCompare _problem_ x y

then you will se a problem.

---

Third issue is that We cannot abstract over QuantifiedConstraints.

Take an example `Dict`. We can use `Dict` for various things.

    data Dict :: (k -> Constraint) -> k -> * where         Dict :: c a => Dict c a

It nicely uses PolyKinds extension so we can write:

    eqInt :: Dict Eq Int     eqInt = Dict

    eq1List :: Dict Eq1 []     eq1List = Dict

And we can use Dict to *manually* thread information

    entail :: Dict Ord1 f -> Dict Eq1 f     entail Dict = Dict

The selling pitch of QuantifiedConstraints, that we could get this for free. Above Ord (Fix) example however makes me suspicious. Let's try:

    eqQList :: Dict (forall x. Eq x => Eq (f x)) []     eqQList = undefined

But it doesn't work!

    Ord1.hs:53:28: error:         • Expected kind ‘(* -> *) -> Constraint’,             but ‘Eq (f x)’ has kind ‘*’         • In the first argument of ‘Dict’, namely             ‘(forall x. Eq x => Eq (f x))’           In the type signature:             eqQList :: Dict (forall x. Eq x => Eq (f x)) []        |     53 | eqQList :: Dict (forall x. Eq x => Eq (f x)) []        |

    Ord1.hs:53:36: error:         • Expected a type, but ‘Eq (f x)’ has kind ‘Constraint’         • In the first argument of ‘Dict’, namely             ‘(forall x. Eq x => Eq (f x))’           In the type signature:             eqQList :: Dict (forall x. Eq x => Eq (f x)) []        |     53 | eqQList :: Dict (forall x. Eq x => Eq (f x)) []        |                                    ^^^^^^^^

Then we remember that we have seen that, we **cannot define** type synonyms for quantified constraints

    type Eq1' f = forall x. Eq x => Eq (f x)

errors with

    Ord1.hs:56:33: error:         • Expected a type, but ‘Eq (f x)’ has kind ‘Constraint’

Luckily GHC-8.10.1 (which is not released at the moment of writing) will give us ability to say

    type Eq1' :: (* -> *) -> Constraint     type Eq1' f = forall x. Eq x => Eq (f x)

This is promising! Let's try to fix eqQList

    eqQList2 :: Dict Eq1' []     eqQList2 = undefined

But that doesn't work. Type-aliases have to be fully applied!

How in Haskell we fix issues when type-aliases (of classes) need to be partially evaluated? We defined

    class ... => Example a b c     instance ... => Example a b c

Third try

    eqQList3 :: Dict Eq1'' f     eqQList3 = Dict

The type signature is accepted, but the implementation is not

   Ord1.hs:67:12: error:        • Could not deduce (Eq (f x)) arising from a use of ‘Dict’

At this point I'm clueless.

---

Best regards, Oleg

P.S. If we would like to take QuantifiedConstraints somewhere into use, then IMO we should start with MonadTrans class. IIRC it's well motivated in the paper. But UndecidableInstances is very unfortunate.

On 14.3.2020 16.10, chessai . wrote:

...

I can second Richard's estimation of QuantifiedConstraints, I have used them a lot in my own code since they were in HEAD. I consider it a sufficiently stable feature to include in base or any library.

On Sat, Mar 14, 2020, 4:16 AM Richard Eisenberg mailto:rae@richarde.dev> wrote:

...

On Mar 14, 2020, at 4:14 AM, Eric Mertens mailto:emertens@gmail.com> wrote:

The last thing I'd heard about quantified constraints was that they were buggy and I've been avoiding relying on them. (I should probably review that assumptions at some point.)

Without expressing an opinion about chessai's proposal (which I have not really thought about): quantified constraints are in good shape and ready for prime time. They have limitations (e.g. you can't mention a type family to the right of the =>), but when they are valid, they work well. I'll never swear that a feature is bug-free, but I think it's reasonable to consider using quantified constraints in `base`.

Richard

_______________________________________________ Libraries mailing list Libraries@haskell.org mailto:Libraries@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/libraries

_______________________________________________ Libraries mailing list Libraries@haskell.org mailto:Libraries@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/libraries

_______________________________________________ Libraries mailing list Libraries@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/libraries