Premier League Predictor: Astro — Chapter 1, Exercise 3
====================================================

TASK
Set up the Astro project and the /api/health route yourself, confirm
it responds with a real server request, then explain — in your own
words — the real difference between a page that's server-rendered on
every request under output: 'server' and one that opts back into
static prerendering with export const prerender = true.

SOLUTION

Steps to complete this exercise:

1. Scaffold the project and add the Node adapter:

   npm create astro@latest -- pl-predictor-astro
   cd pl-predictor-astro
   npx astro add node
   npm install better-sqlite3

2. Set astro.config.mjs exactly as shown in the chapter:

   import { defineConfig } from 'astro/config';
   import node from '@astrojs/node';

   export default defineConfig({
     output: 'server',
     adapter: node({ mode: 'standalone' }),
   });

3. Create src/pages/api/health.ts exactly as shown in the chapter.

4. Start the dev server:

   npm run dev

5. Visit http://localhost:4321/api/health in a browser — it should
   return {"status":"ok"}.

Written explanation:

Under output: 'server', a page or API route with no prerender setting
is built fresh on every single incoming request — the server runs the
page's own code (or the endpoint's GET/POST function) each time a
visitor loads it, which is exactly what's needed for something that
genuinely changes request to request, like a league table that
reflects whatever the database currently holds, or an API route that
reads or writes real data.

A page marked export const prerender = true instead gets built once,
ahead of time, into a plain static HTML file during the build step —
the same way every page works in Astro's own default static mode. Once
built, that page is served as a fixed file with no server-side work
happening on each visit at all, which is faster and cheaper to serve,
but only correct for content that doesn't need to reflect anything
that changes between requests.

In short: output: 'server' switches the whole project's default from
"build once" to "run every time," and prerender = true is the escape
hatch that lets one specific page opt back into the faster "build
once" behavior when it doesn't actually need to be dynamic.

WHY THIS WORKS AS AN ANSWER
----------------------------
It walks through every real setup step in order, confirms the health
check works, and then gives a written explanation that correctly
identifies the real mechanism on both sides — request-time execution
vs. build-time generation — rather than just restating which setting
is "faster," which is the actual point Chapter 10 later builds on.