Fetching Data

Each field exposed in the GraphQL schema has a corresponding resolver (aka data fetcher) associated with it. graphql-kotlin-schema-generator generates the GraphQL schema directly from the source code, automatically mapping all the fields either to use FunctionDataFetcher to resolve underlying functions or the PropertyDataFetcher to read a value from an underlying Kotlin property.

While all the fields in a GraphQL schema are resolved independently to produce a final result, whether a field is backed by a function or a property can have significant performance repercussions. For example, given the following schema:

type Query {
  product(id: ID!): Product
}

type Product {
  id: ID!
  name: String!
  reviews: [Review!]!
}

type Review {
  id: ID!
  text: String!
}

We can define Product as

data class Product(val id: ID, val name: String, reviews: List<Review>)

or

class Product(val id: ID, val name: String) {
  suspend fun reviews(): List<Reviews> {
     // logic to fetch reviews here
  }
}

If we expose the reviews field as a property it will always be populated regardless whether or not your client actually asks for it. On the other hand if reviews is backed by a function, it will only be called if your client asks for this data. In order to minimize the over-fetching of data from your underlying data sources we recommend to expose all your GraphQL fields that require some additional computations through functions.

Customizing Default Behavior

You can provide your own custom data fetchers to resolve functions and properties by providing an instance of KotlinDataFetcherFactoryProvider to your SchemaGeneratorConfig.

See our spring example app for an example of CustomDataFetcherFactoryProvider.