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

/explore's rank stays warm on the live forge: a meter or heartbeat write no longer throws the ranking away #6219

shipped production is serving it opened this has it
opened this

Measured 2026-10-26 on the live forge, signed out: /explore reports rank;dur=30.7, 30.6 and 30.6 ms over repeated requests, and /api/v1/explore reports explore;dur=57.0. #6129 (59bcf621) made a warm rank under 10 ms in its test. Live, the rank is never warm. rankInputs (src/core/rank_inputs.ts) hashes max(started_at) and max(ended_at) of meter, and on a busy forge those move every few seconds.

Done when:

  • [ ] Meter-derived inputs move the hash at most once per the ranking's max age (or only when a ranked number would change), so a run starting or ending does not rebuild the ranking.
  • [ ] A test writes meter rows between two /explore requests and asserts the second rank phase is under 10 ms. The test fails on main today.
  • [ ] test/explore-speed.test.ts still passes: a fork, visibility, access, seat or team change still shows on the next request.

Files: src/core/rank_inputs.ts, src/actions/explore.ts, a test.

Sign in to join this conversation.

↑↓ navigate ↵ open ⌘↵ its app esc close

Keyboard shortcuts

Docs