batchDelegateToSchema running resolvers twice

[TuurDutoit]

I couldn't find an answer to my question in the docs, so I'll try my luck here.

I'm using batchDelegateToSchema to resolve a field in my GraphQL API in a more optimized way. This works fine, except for the fact it runs my resolvers twice! I think I understand why it happens, but I'm at a loss on how to stop it from happening.

I made a minimal repro here. Given the following query:

{
  course(courseId: 1) {
    id
    title
    description
    pool {
      id
      title
      description
    }
  }
}

This is what happens:

1. GraphQL resolves Query.course
2. GraphQL resolves Course.pool. This uses batchDelegateToSchema, which essentially runs a sort of "subquery" against the same schema:
   3. GraphQL resolves Query.pools
   4. For each pool (just 1 in this case), GraphQL resolves Pool.id (this logs a message to the console)
   5. batchDelegateToSchema returns the Pool object (crucially: with all of its fields already being resolved!)
3. GraphQL resolves Pool.id again (logging a second message to the console)! It just received some object from the Course.pool resolver - it doesn't know this object has already been fully processed inside batchDelegateToSchema, so it runs the resolvers. GraphQL is "dumb" that way. Is there a way to make it "smart" and not have it rerun the same resolvers again?

Does anyone have experience dealing with this? It would probably be useful to add a warning about this in the docs!

[ardatan]

You use graphql-js execution here, so according to the regular resolution graphql-js will still call the outer resolvers.
You will see what I mean if you use batchDelegateToSchema against another schema.
Also that function is not intended to delegate queries in the same schema. The idea is to delegate queries to another schema or source.