April 2016 Summaries

GraphQL schema decorators are a proposal that aims to provide a standardized and modular way to add features to GraphQL schemas. They enable developers to add annotations, metadata, and behavior to different parts of their schema using reusable modules, making it easier to build and maintain complex GraphQL applications. The goal is to achieve flexibility, well-specified semantics, introspectability, self-containment, and portability across programming languages and implementations. Schema decorators share similarities with GraphQL directives but differ in application scope, implementation, and behavior. They can be used to add features such as authorization checks, argument validation, logging, error handling, and more to the schema, making it easier to build complex and maintainable GraphQL applications.

This tutorial covers building a GraphQL server using Apollo Server, with the goal of connecting to multiple backends such as SQL databases and MongoDB. The server is built using Node.js and npm packages like Sequelize for the SQL database connection and mongoose for the MongoDB connection. The resolvers are used to define how data should be retrieved from each backend, and caching is set up to improve query performance. The tutorial covers various aspects of building a GraphQL server, including setting up tracing, connecting to REST endpoints, and configuring caching. It also includes an example of using Apollo Engine for server-side caching. Overall, this tutorial provides a comprehensive guide to building a GraphQL server with multiple backends.

The Apollo GraphQL data stack is a technical preview release announced by Sashko Stubailo, aiming to enable experimentation with client and server-side components that interact between UI frameworks and backend services. The initial set of tools and integrations includes the Apollo Client, GraphQL Tools, and native integrations for React, Angular, Meteor, and Redux. Three example apps are available to demonstrate how the different parts might fit together, including a full-stack app with Material UI on top of Discourse, a GraphQL server example using SQL and MongoDB, and a bare-bones Meteor starter kit. The documentation site provides detailed information on the current set of packages, while the website introduces people to the project. The focus is on building a full-stack solution for GraphQL that's easy to get started with and scales as the app grows.

GraphQL-JS is a reference implementation of the GraphQL spec that also serves as a library for building servers, clients, and tooling for GraphQL. It exports several useful functions such as parse, which can be used to parse queries and schema notation, introspectionQuery, buildClientSchema, printSchema, validate, execute, and buildASTSchema. These functions provide various features for building and working with GraphQL schemas, including parsing and validating queries, printing schemas in a readable format, executing queries on the server, and creating executable schemas. Additionally, it has a little tool kit called graphql-tools that can be used to add resolve functions to a schema created with buildASTSchema.

Mocking a server with GraphQL tools has become significantly easier, thanks to the introduction of type systems. This allows for faster and more convenient mocking of backend APIs, making it possible to start developing frontend features without a working backend, run tests locally without connecting to a real backend, and even customize mock data to suit specific needs. The Apollo mocking tool provides a simple way to create mock servers with just one line of code, using the GraphQL schema as a reference point. With this tool, developers can quickly get started with mocking their backend APIs and fine-tune their mocks in tiny steps as needed.

Sashko Stubailo, a Meteor developer, introduced Apollo, the new GraphQL-based data stack for modern apps. The goal of Apollo is to decouple the client and server components, allowing developers to work with a unified data stack that can be adopted in any app, regardless of the platform or backend technology used. Meteor's existing data system was compared to GraphQL, highlighting the benefits of using GraphQL, such as reduced roundtrips, self-documenting APIs, and abstraction between services. The Apollo project aims to create a principled, unified, and scalable API for all client apps, agnostic to what is going on with the backend. The team has been working on the Apollo Server and Client, with the latter expected to be released soon.

The author of the text is working on creating a GraphQL server called Apollo Server, which aims to simplify building a GraphQL server by providing a clear structure and great tooling. The server consists of three main parts: a type system, stateless resolve functions, and loaders. A type system defines the types and their relationships in the schema, while resolve functions specify how to respond to queries for each field. Loaders separate data loading from resolve functions, allowing for better performance and maintainability. The author is also working on providing guides, documentation, error handling, logging, profiling, and performance monitoring features to enhance the developer experience.