Jamie Willis

I meant more are there any existing instances that _couldn't_ be commutative, I guess?

And also any `select` that executes both branches too (with `select = selectA`)

Ah, fair. Yes I was thinking about Applicatives where the order of effects matters More precisely then, it doesn't work for `select = selectA` when `()` doesn't "commute", i.e. where...

If you think it's worthwhile, I could probably dedicate some "tube time" to thinking about laws that might involve `select` and commutative `branch`'s interactions?

yeah, the break is annoying. If only Haskell had a rewrite tool for stuff like this (Scala has `scalafix`, which can be used for migrating to new breaking versions of...

Fair enough! In which case, I would advocate for the instance to either be removed, or instead placed behind a newtype _itself_. Library authors should, for sure, not create an...

Backwards incompatible changes by definition break code: I wouldn't be saying do it immediately, but if you're making a different breaking change that already necessitates changes in user code, then...

Most parser combinator libraries support some form of "chain" combinators that abstract away the repeated application of a parser to operators. They handle the recursion properly, and apply the operators...

Yeah, they are voluntarily left out, because I consider it bad form to keep them in the repo when they are essentially "precompiled" files. I've forgotten the exact commands I...

I don't know what would cause that: I would just recommend disabling the nandlang benchmarks in the cabal file.