Ryan Scott

This is indeed a bug. Currently, `makeRep0Inline` really only works if you're writing an instance of `Generic (T a_1 ... a_n)`, where `a_1 ... a_n` are all distinct type variables....

It would still be rather fragile. The reason that `makeRep0Inline` works today is because it gathers all of the type variables it encounters within `[t| ... |]` from left-to-right. It's...

Having taken a quick look at this, I'm not convinced that a Template Haskell-based solution would necessarily be much cleaner. It might even present its own unique set of maintenance...

I pushed a proof-of-concept implementation of the ideas in https://github.com/RyanGlScott/text-show/issues/33#issuecomment-1211919793 to the [`T33-alt`](https://github.com/RyanGlScott/text-show/tree/T33-alt) branch. Unfortunately, the benchmarks confirm my fears that this approach is not as efficient as the current,...

Thanks! I'm able to use the Linux bindist for HLint 3.3.4 now. > Either way, it should use the latest available for that OS, so I consider it a bug...

Eek, sorry for not noticing this until now. There has been independent movement on `Data.Constraint.Forall` in 978b4cd930efe1d28b44703d6190af8c59d44a82, which revamps the module to use `QuantifiedConstraints`. Does this supersede the work in...

I'm in favor of fixing this by adding a `1

It's certainly possible. This patch would suffice: ```diff diff --git a/src/Data/Constraint/Forall.hs b/src/Data/Constraint/Forall.hs index 3932f29..f628993 100644 --- a/src/Data/Constraint/Forall.hs +++ b/src/Data/Constraint/Forall.hs @@ -139,11 +139,11 @@ class Forall (R p t a) =>...

Although I'm unfamiliar with the ins and outs of `stack`, this looks related to the fact that old versions of `distributive` had a custom `Setup` script. [`distributive-0.6.2.1`](http://hackage.haskell.org/package/distributive-0.6.2.1) no longer has...

Do you have a standalone example of a program that doesn't simplify the way you'd expect because of this? I'm not opposed to adding an `INLINE` pragma to these functions,...