Rob Rix
I’m not using `bound` (yet), but I’ve had some recent success adapting Hinze’s approach from _Efficient Generalized Folds_: https://github.com/robrix/path/blob/2da82c52cd17088fd6918ef3f37d3c13081dfd99/src/Path/Expr.hs#L32-L44
Would `b == 7` be the bit that fails, or would it be `b : Integer`? 😁
I’m pretty fond of `Reader`; there are cases where first-order definitions of effects (in which `local` is defined as a handler instead of an operation) give you _extremely_ counter-intuitive results....
Added to the milestone. Either we’ll add the old carrier back as well, or we’ll figure out a way to improve the ergonomics of the new one.
> We can fix that with a type alias, which I think we need. Great idea 👍 > I was hoping the new impredicative polymorphism would fix it, but it...
(Also, thank you very much for filing @mitchellwrosen! Your filed issues are good and you should feel good.)
Unfortunately, this approach is an order of magnitude slower. The intuitive explanation is that it’s equivalent to the old `ReaderC (Handler eff m)` implementation we had before, due to presence...
Okay, I’m… less hopeful now. (**Edit:** I haven’t been able to find a solution with better ergonomics that doesn’t sacrifice performance.) @mitchellwrosen, I know you were hoping to see this...
Thanks so much for the notes, @isovector! This might be a good one to wrap up in some ghc bug reports/enhancement requests, at least once we’ve got a good idea...
Removing from the 1.0 milestone.