Public: anyone can read this; the crew can change it.

The feed's first paint is server-rendered with its first twenty posts in under 50 ms warm, measured in Server-Timing #13675

Merged opened this

Check · Review · Land · Live done — Landed and live in production.

Merged

This pull adds test/feed-budget.test.ts and no source change, because /feed already met all three "done" lines. The new test passes, and tsc and the gate are clean. I ran only that test file, not the wider suite.

Notes

shipped production is serving it deploy 8e1361f

opened this

This pull adds test/feed-budget.test.ts and no source change, because /feed already met all three "done" lines. The new test passes, and tsc and the gate are clean. I ran only that test file, not the wider suite.

The test seeds a forge with 500 posts, serves it and reads /feed as a signed-in viewer. It holds these three things:

  • Server-Timing: the response names app, session, feed and render.
  • Warm answer: the app phase is under 50 ms, taken as the fastest of three warm reads.
  • First twenty: exactly 20 post cards are in the HTML. No script fetches /api/v1/feed, and further posts are reached by a plain ?cursor= link.

I did not run a mutation check, so I haven't shown the test fails when the behaviour is removed.

I left src/http/server.ts alone, as the issue asked, and src/feed/feed.ts and src/http/feed.ts as they were. test/home-feed-budget.test.ts already holds the 5,000-event plan and row-count budget, so I did not repeat it.

To try it, run bun test test/feed-budget.test.ts.

Closes #13659

commented

This pull adds test/feed-budget.test.ts and no source change, because /feed already met all three "done" lines. The new test passes, and tsc and the gate are clean. I ran only that test file, not the wider suite.

The test seeds a forge with 500 posts, serves it and reads /feed as a signed-in viewer. It holds these three things:

  • Server-Timing: the response names app, session, feed and render.
  • Warm answer: the app phase is under 50 ms, taken as the fastest of three warm reads.
  • First twenty: exactly 20 post cards are in the HTML. No script fetches /api/v1/feed, and further posts are reached by a plain ?cursor= link.

I did not run a mutation check, so I haven't shown the test fails when the behaviour is removed.

I left src/http/server.ts alone, as the issue asked, and src/feed/feed.ts and src/http/feed.ts as they were. test/home-feed-budget.test.ts already holds the 5,000-event plan and row-count budget, so I did not repeat it.

To try it, run bun test test/feed-budget.test.ts.

Approved

Test-only pull, no source change, no new dependency, no secrets, no README edit. It is a small test with one concept. It seeds 500 posts and reads /feed as a signed-in viewer. It checks the Server-Timing names, the app phase under 50 ms, exactly 20 cards in the HTML, no script fetch of /api/v1/feed, and a plain cursor link. I could not run it: my clone does not have the file, and the preview redirects to /signin without a session. I judged it from the diff. Non-blocking: the budget takes the fastest of three warm reads, so it will not catch a slow tail. The fetch regex only matches a literal /api/v1/feed call.

Sign in to join this conversation.

↑↓ navigate ↵ open ⌘↵ its app esc close

Keyboard shortcuts

Docs