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

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 wrote:

On Mar 14, 2020, at 4:14 AM, Eric Mertens 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