Dataloader equivalent (or alternative approach) about absinthe HOT 8 CLOSED

I agree about Sangria's approach in general, it seems to be the most mature GraphQL implementation right now. The original graphql-js was intended purely as a reference implementation and it's starting to show.

from absinthe.

Looks like Sangria has first-class concepts of Deferred resolvers (what i'm asking for) and Projections (what is already possible, but could be made simpler) which might be worth adopting: http://sangria-graphql.org/learn/#deferred-values-and-resolver

from absinthe.

@AndrewIngram The "how" of dataloader certainly isn't a direct port to Elixir/Erlang, but I'm confident we have better tools on hand for, eg, concurrency to apply to an implementation. Dataloader has been on our list of ecosystem tools to look into next, as I know that others in the GraphQL/JS space rely on it.

from absinthe.

I'll add that I'm very excited by the options pursued by Sangria, as I think those concepts are going to be more easily applicable to Elixir.

from absinthe.

👍

from absinthe.

Also see https://github.com/Shopify/graphql-batch for some similar concepts

from absinthe.

Tagging this as "roadmap" and closing it, but merely for temporary archival purposes. We know this and a number of other issues are big ticket items we need to address; we've got some of the foundational pieces going in now, and let's create more focused, actionable issues/PR discussion around this (eg, on deferred resolution, which I know @benwilson512 is very interested in as an approach).

from absinthe.

Just to close the loop here Absinthe now has a dataloader built-in: https://github.com/absinthe-graphql/dataloader

I believe it can also be considered feature-complete (for the majority of use-cases at least), although I'm not 100% sure on that.

from absinthe.