vlovgr/ciris

Functor/Monad instances for ConfigResult

valenterry opened this issue · 4 comments

valenterry commented

Any reason not to have instances for functor/monad etc. for ConfigResult[F, ?] as long as these instances exist for F? Would be nice to be able to use mapN instead of loadConfig - or use for comprehensions if the context type allows it.

vlovgr commented

loadConfig is essentially mapN if we swap Either with Validated. The reason it exists is because the core module is dependency-free, meaning we can't rely on cats. What also might prevent this from working nicely is that often ConfigResult is not returned, but a different type extending ConfigResult, e.g. ConfigValue. There's no reason why we can't write these instances in the cats module though. I'd happily accept a pull request. :)

valenterry commented

👍 let me make a PR for that - I will do it for the (hardcoded) Either for now.
@vlovgr do you think it makes sense in general to use (or allow) for a Validated type like in cats? In the end, the Either[ConfigErrors, ?] essentially does look very similar already.

vlovgr commented

Thanks! Yes, there is a case for it if you want to use the mapN syntax in cats.

vlovgr commented

Ciris 1.x (preview on the 1.x branch) will be based on cats and cats-effect and will provide FlatMap for ConfigValue, allowing flatMap and mapN, as well as a NonEmptyParallel instance for parMapN (which matches the behaviour of loadConfig in 0.x).