GraphQL

GraphQL

If there’s a GraphQL API that we want to use data from, we have a few options:

• If we want to include parts of the API’s schema in our schema:
  • If it supports federation, we should use that. For example, FaunaDB is working on support, and some third-party services we use might have a GraphQL API that supports federation. And if we have control over the API (e.g., if it’s one of our services), we can add support for federation.
  • We can use schema stitching if the API doesn’t support federation. But unless we want a significant part of the API’s schema, it may be easier to use one of the below methods instead.
• If we just want to use data from the API in our resolvers:
  • Use GraphQLDataSource from apollo-datasource-graphql to create a data source class. Similarly to RESTDataSource, we can define a willSendRequest method that adds an authorization header to all requests. But in our data fetching methods, instead of this.get('path'), we use this.query(QUERY_DOCUMENT).
  • Use graphql-request in our resolvers to fetch data from the data source (similar to our githubStars subscription where we fetch data from GitHub’s GraphQL API). While graphql-request is nice for extremely simple uses like githubStars, usually GraphQLDataSource is a better choice, as it’s a data source class.