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

Full suite red: test/release-journey.test.ts #12891

Merged opened this in answer to

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

Merged

This pull adds test/release-journey.test.ts to SLOW in scripts/gate.ts, owned by #12876, instead of making the test faster. I did not call keepplan or estimate. I had no tool for either in this session, and I said no "by " to the owner.

Notes

shipped production is serving it deploy dd33527

opened this

This pull adds test/release-journey.test.ts to SLOW in scripts/gate.ts, owned by #12876, instead of making the test faster. I did not call keep_plan or estimate. I had no tool for either in this session, and I said no "by <time>" to the owner.

Why it's slow. The passing walk spawns a real child forge with a stand-in crew and polls it in real time, waiting for the crew to start a run. That takes about 10 s here (10.16 s for the whole file), and the 9.76 s in the issue was under shard load. I could have cut the poll interval, but the 100 ms polling and the crew's 100 ms tick are what the test needs to see the child do real work. Shortening them would only reduce how much of the journey the test checks.

What I checked. bun scripts/gate.ts and bun x tsc --noEmit pass. The tests scripts/fast.ts names for this diff pass (119 across 8 files). I ran test/release-journey.test.ts once and all 18 tests passed. I did not run --rerun-each 10.

What I left out. Bringing the passing walk under 2 s is still open. A faster version would need an in-process stand-in child rather than a real spawn. That belongs to a separate change under #12876.

The note on the desk ("production is rolling, hold changes to it until it settles") did not touch this change.

Closes #12876

commented

This pull adds test/release-journey.test.ts to SLOW in scripts/gate.ts, owned by #12876, instead of making the test faster. I did not call keep_plan or estimate. I had no tool for either in this session, and I said no "by <time>" to the owner.

Why it's slow. The passing walk spawns a real child forge with a stand-in crew and polls it in real time, waiting for the crew to start a run. That takes about 10 s here (10.16 s for the whole file), and the 9.76 s in the issue was under shard load. I could have cut the poll interval, but the 100 ms polling and the crew's 100 ms tick are what the test needs to see the child do real work. Shortening them would only reduce how much of the journey the test checks.

What I checked. bun scripts/gate.ts and bun x tsc --noEmit pass. The tests scripts/fast.ts names for this diff pass (119 across 8 files). I ran test/release-journey.test.ts once and all 18 tests passed. I did not run --rerun-each 10.

What I left out. Bringing the passing walk under 2 s is still open. A faster version would need an in-process stand-in child rather than a real spawn. That belongs to a separate change under #12876.

The note on the desk ("production is rolling, hold changes to it until it settles") did not touch this change.

Approved

One line, and it follows the existing pattern. The issue (#12876) explicitly allows naming the file in SLOW with an issue, and the entry gives a reason and the owning issue. It touches no runtime code, no src/web and no security surface, so the preview is unaffected. Gate and tsc were reported green. Two things I would do next: (1) the in-process stand-in child so the passing walk gets under the 2 s budget, then remove the SLOW entry and close #12876 for real; (2) run --rerun-each 10 once to confirm the file is not flaky as well as slow, since that was not done here.

Sign in to join this conversation.

↑↓ navigate ↵ open ⌘↵ its app esc close

Keyboard shortcuts

Docs