14:38 <repo> Blacksmoke16: https://p.jokke.space/m-9PYZs/

14:38 <repo> was easier than i expected

14:40 <repo> i also managed to make the field resolving concurrent in the graphql shard. It seems a bit like a hack though... I have to build individual json fragments for the fields and then add them to the current builder using JSON::Builder#raw...

14:42 <repo> i'll try another approach where i build a nested hash and then have the json serialization as a separate step

14:42 <repo> would seem cleaner to me

14:44 <FromGitter> <Blacksmoke16> would it makes sense for the data loader to like include `Iterator`

14:45 <FromGitter> <Blacksmoke16> or at least `Enumerable`

21:26 <repo> Blacksmoke16: how come?

21:26 <repo> i managed to resolve to hash and to_json in a separate step! \o/ much cleaner imo: https://github.com/graphql-crystal/graphql/pull/36

21:50 <FromGitter> <Blacksmoke16> so you could pass it to things that are typed like that. but after looking at the code again prob wouldnt make too much sense

21:51 <FromGitter> <Blacksmoke16> nice one!

21:54 <repo> yeah the dataloaders would be pretty short lived objects and live in a request context

21:55 <repo> the cache would otherwise make no sense

21:55 <repo> it should reduce database calls made during a single request

21:57 <repo> (which in a graphql request can be veeeery many)

22:03 <FromGitter> <Blacksmoke16> 👍 makes sense yea