Fetching Data

A custom server can help when an app needs custom request handling or authentication, and a standalone server may cost less for some workloads. But it gives up serverless API routes and automatic static optimization. Robinson estimates that about 95 percent of Next.js use cases can avoid a custom server.

API routes are files under `pages` that handle requests and responses, with helpers for reading cookies, query parameters, and request bodies and returning JSON. Dynamic routes such as `/api/users/[id]` can return a specific user or query a database. For client-side data that can load after the page, SWR fetches and revalidates responses, helping keep pages statically served.

GraphQL lets clients request specific fields, including nested data, with queries for reading and mutations for changing data. Hasura builds a GraphQL API over Postgres and generates CRUD mutations from table models. Robinson demonstrates creating tables, querying and filtering rows, and joining related deals and locations in the Day Drink app.

The Day Drink app uses Apollo Client to query deals by day and retrieve related users, active days, and locations. Apollo also supports mutations and a cache; after adding a deal, the app updates cached query results directly so the new deal appears without another database request. Robinson also shows fetching location and city data for display and possible map use.