Challenge 1: Add a Second Loader to Context — Possible Solution
====================================================================

import express from "express";
import { expressMiddleware } from "@apollo/server/express4";

const app = express();
const server = new ApolloServer({ typeDefs, resolvers });
await server.start();

app.use(
  "/graphql",
  express.json(),
  expressMiddleware(server, {
    async context({ req }) {
      return {
        userLoader: createUserLoader(pool),
        postsByAuthorLoader: createPostsByAuthorLoader(pool), // NEW
      };
    },
  })
);

app.listen(4000);

// A matching batch function, following the same pattern as Challenge 1
// from Chapter 5 (batching a one-to-many relationship):
function createPostsByAuthorLoader(pool) {
  return new DataLoader(async (authorIds) => {
    const [rows] = await pool.query(
      "SELECT * FROM posts WHERE author_id IN (?)", [authorIds]
    );
    return authorIds.map((authorId) =>
      rows.filter((post) => post.author_id === authorId)
    );
  });
}

WHY THIS WORKS
--------------
- Both loaders are created INSIDE the same context callback, on every
  single request — postsByAuthorLoader isn't hoisted out any more than
  userLoader is, which is exactly the discipline the chapter's tip box
  insists on. Adding a second loader doesn't change WHERE loaders get
  created, only that there are now two of them built the same way.
- The context object simply grows to carry both loaders as separate
  named properties — any resolver needing user data reads
  context.userLoader, and any resolver needing a user's posts reads
  context.postsByAuthorLoader, each getting its own independent,
  per-request cache and batching behavior.
- createPostsByAuthorLoader follows the one-to-many batching shape
  (using .filter() to group multiple posts per author) rather than the
  one-to-one shape (using .find() for a single user per id) — the same
  distinction covered in Chapter 5's own comments-batching challenge,
  applied here to a different relationship (posts grouped by author
  instead of comments grouped by post).
- This scales naturally: a real production schema typically needs
  several DataLoaders (one per relationship that risks an N+1 pattern),
  and the context function is simply the one place all of them get
  constructed together, once per request.