Koz Ross

@Fuuzetsu @ocramz Sorry to disappoint, but further discussions in my organization means that we're going to maintain our own copy. We've added a lot of modifications, and removed considerable functionality...

Aww, that's a bit disappointing. I actually prefer structured error types massively, and the API is also much more pleasant.

I think you might get some mileage out of [`QualifiedDo`](https://ghc.gitlab.haskell.org/ghc/doc/users_guide/exts/qualified_do.html)?

Does op-rigidity enable any capabilities or methods that regular rigidity wouldn't? Would it perhaps make sense to have `Selective`, which is rigid, but not op-rigid, and `OpSelective` which is both?

@snowleopard If as you say most instances are rigid, marking the exceptions alone, while adding something like "Unless stated otherwise, the instances of `Selective` defined in this package are rigid.`...

@Kleidukos Except that `before :: IO a -> SpecWith a -> Spec`. How do I get access to, and use for comparisons (for instance) the value of type `a` produced...

@Kleidukos Therein lies my precise complaint.

Not particularly - that example is tied up with a lot of unnecessary context, and self-constructing explanations from that is not exactly easy. The issue isn't my lack of understanding;...

Notably, [this already arose][1]. There's even an [_almost_ copy-pasteable solution][2] by a key contributor. [1]: https://github.com/hspec/hspec/issues/140 [2]: https://stackoverflow.com/a/30554467/2629787

I agree with @gbaz for the most part. However, I am a little confused about the 'memory footprint' being referred to. If we're using a standard relational schema to encode...