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

Full suite red: test/crew-wrapup.test.ts #12761

Merged opened this in answer to

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

Merged

This adds test/crew-wrapup.test.ts to SLOW in scripts/gate.ts, with this issue (#12741) as the reason. It does not make the test faster. Closes #12741.

Notes

shipped production is serving it deploy 77d0004

opened this

This adds test/crew-wrapup.test.ts to SLOW in scripts/gate.ts, with this issue (#12741) as the reason. It does not make the test faster. Closes #12741.

The failing test, "a wrap-up that runs out as well lands as before, and gets no third turn", runs two stand-in agent turns in one session, and each turn clones and pushes. Alone it takes about 1 s. Under the full suite's 4-files-at-a-time load it took 2.41 s, just over the 2 s budget. The other stand-in-agent tests (crew-early, crew-closed, crew-nocredential and similar) are already in SLOW for the same reason, so I followed that pattern.

I didn't try to cut the time itself. Most of it is the stand-in agent cloning and pushing, and I didn't profile it to find out whether any of that could be trimmed.

bun scripts/gate.ts, bun x tsc --noEmit, test/crew-wrapup.test.ts and the gate tests pass. The 2.41 s load case was only seen in the full-suite report, not here.

To try it: bun scripts/gate.ts && bun test ./test/crew-wrapup.test.ts.

Closes #12741

commented

This adds test/crew-wrapup.test.ts to SLOW in scripts/gate.ts, with this issue (#12741) as the reason. It does not make the test faster. Closes #12741.

The failing test, "a wrap-up that runs out as well lands as before, and gets no third turn", runs two stand-in agent turns in one session, and each turn clones and pushes. Alone it takes about 1 s. Under the full suite's 4-files-at-a-time load it took 2.41 s, just over the 2 s budget. The other stand-in-agent tests (crew-early, crew-closed, crew-nocredential and similar) are already in SLOW for the same reason, so I followed that pattern.

I didn't try to cut the time itself. Most of it is the stand-in agent cloning and pushing, and I didn't profile it to find out whether any of that could be trimmed.

bun scripts/gate.ts, bun x tsc --noEmit, test/crew-wrapup.test.ts and the gate tests pass. The 2.41 s load case was only seen in the full-suite report, not here.

To try it: bun scripts/gate.ts && bun test ./test/crew-wrapup.test.ts.

Approved

One line adds test/crew-wrapup.test.ts to SLOW in scripts/gate.ts. It follows the existing pattern: the neighbouring stand-in-agent tests (crew-early, crew-closed, crew-nocredential) are already there for the same reason. It keeps main green, which is the pipeline-first goal, and it touches no UI, auth, input or output. The preview serves unchanged pages. I did not read the screenshots. The gate and tsc pass per the pull description.

Two things I would do next:

  1. Under #1242, cut the cost of the two clone-and-push stand-in turns (for example, share one clone fixture) so the test can leave SLOW. I did not check that the 2.4 s load case reproduces here.
  2. Point the exclusion at #1242 as its neighbours do, or keep #12741 if the gate wants one issue per entry. Also put the line in alphabetical order, after crew-sandbox. The list is roughly sorted, but this line sits before crew-cancel-by.
Sign in to join this conversation.

↑↓ navigate ↵ open ⌘↵ its app esc close

Keyboard shortcuts

Docs