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

Full suite red: test/home-200.test.ts #12822

Merged opened this in answer to

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

Merged

test/home-200.test.ts should now finish inside the 20 s file budget. Here it runs in about 4.5 s, down from about 8 s on my first run and 5.7 s on a second run before the change. The failing 25.5 s was measured with four files running at once.

Notes

shipped production is serving it deploy c36cfce

opened this

test/home-200.test.ts should now finish inside the 20 s file budget. Here it runs in about 4.5 s, down from about 8 s on my first run and 5.7 s on a second run before the change. The failing 25.5 s was measured with four files running at once. I did not reproduce that, so I can't say the margin holds under that load.

What changed: the setup in bench/snappy.ts (snappyFixture) made its 200 repositories one at a time, each waiting on its own git init. It now makes them ten at a time. The per-repository work is the same, and repos keeps its order. The setup went from about 2.7 s to about 1.6 s.

What I left alone:

  • I did not add test/home-200.test.ts to SLOW in scripts/gate.ts. The test is faster rather than excused, and it still builds the same 200 repositories and 100 asks.
  • I did not shrink the fixture, so the test still covers the forge size it is named for.

Checks: bun x tsc --noEmit and bun scripts/gate.ts printed nothing. I ran every test file that uses snappyFixture (161 tests, 29 files): 160 pass and 1 fails. The failure is test/kpis-landings-scale.test.ts, which timed out at 5 s, and it fails the same way with my change stashed.

To try it: bun test ./test/home-200.test.ts

🤖 Generated with Claude Code

Closes #12791

commented

test/home-200.test.ts should now finish inside the 20 s file budget. Here it runs in about 4.5 s, down from about 8 s on my first run and 5.7 s on a second run before the change. The failing 25.5 s was measured with four files running at once. I did not reproduce that, so I can't say the margin holds under that load.

What changed: the setup in bench/snappy.ts (snappyFixture) made its 200 repositories one at a time, each waiting on its own git init. It now makes them ten at a time. The per-repository work is the same, and repos keeps its order. The setup went from about 2.7 s to about 1.6 s.

What I left alone:

  • I did not add test/home-200.test.ts to SLOW in scripts/gate.ts. The test is faster rather than excused, and it still builds the same 200 repositories and 100 asks.
  • I did not shrink the fixture, so the test still covers the forge size it is named for.

Checks: bun x tsc --noEmit and bun scripts/gate.ts printed nothing. I ran every test file that uses snappyFixture (161 tests, 29 files): 160 pass and 1 fails. The failure is test/kpis-landings-scale.test.ts, which timed out at 5 s, and it fails the same way with my change stashed.

To try it: bun test ./test/home-200.test.ts

🤖 Generated with Claude Code

Approved

Clean. Bench-only change: repositories are built ten at a time, repos keeps its order via indexed assignment, and per-repo work is unchanged. No security surface, no new dependency, no README change, and it correctly does not excuse the test in SLOW. The only side effect is that issue ids interleave across repos in the fixture. The fixture is internal, and the 161 tests that use it are reported passing, apart from the already-failing kpis-landings-scale timeout, so this is fine.

Sign in to join this conversation.

↑↓ navigate ↵ open ⌘↵ its app esc close

Keyboard shortcuts

Docs