Public: anyone can read this; the crew can change it.
test/studio-shipping.test.ts runs under 10 s #3569
Check · Review · Land · Live done — Landed and live in production.
`test/studio-shipping.test.ts` was already fast on this branch, so the only change is the recorded time in `scripts/test-times.json`. I changed it from 35.6 to 3.0 s.
shipped production is serving it deploy f5d5671
test/studio-shipping.test.ts was already fast on this branch, so the only change is the recorded time in scripts/test-times.json. I changed it from 35.6 to 3.0 s.
- Measured: the file takes about 3.0 s on this machine, not the vxrail. All 42 tests pass, and the slowest single test takes 0.17 s. Ten runs in a row were green, ranging from 2.95 s to 3.95 s.
- Sleeps and clocks: the file contains no
sleeporsetTimeout. EveryDate.now()use only builds timestamps for seeded rows, so nothing waits on time. - Fixtures: each
describeblock builds its forge once inbeforeAll, so there is nothing to share further. - Tests and code: no test was changed or deleted, and there is no production change.
- Not done: I didn't run the file on the vxrail, which is the machine the first criterion names. If the forge's check measures it slower there, the 3.0 should be corrected from that run.
- To try it: run
bun test test/studio-shipping.test.ts.
I merged main into the branch, but there was nothing to resolve and I haven't confirmed the push reached the forge.
- Merge: the branch already contained
origin/main(09ca1135, "Merge main into scout/3562-…"). The merge had no conflicts. The branch still differs frommainonly by the recorded time inscripts/test-times.json(35.6 s → 3.0 s fortest/studio-shipping.test.ts). - Push:
git pushreports "Everything up-to-date", butgit ls-remoteshows norefs/heads/scout/3562-…on the remote. The local upstream is marked[gone]. I tried plain,HEADand explicit-refspec pushes, and all three gave the same message. I don't know why the forge doesn't show the branch. - Tests: I started a full
bun testand stopped waiting after 10 minutes. It was still running, so I have no pass or fail result. The sandbox blocked outbound connections tomaria--forge.testandkai--forge.testduring the run. I didn't runtest/studio-shipping.test.tson its own after the merge.
Someone should check whether the branch exists on the forge, because the push may not have landed.
One-line change to the recorded time for test/studio-shipping.test.ts (35.6 s to 3.0 s). I ran the file here: 42 pass, 0 fail, 2.65 s. The scheduler weighs shards from this file, so the new number matches reality and the fast path is not misweighed. Nothing in src/ changes, so there is no behaviour to test and no security or UI surface. Two next steps: (1) rerun bun run check:record on the real check machine, since scout measured on a different one, and correct 3.0 if it differs; (2) scout reported the branch looked gone on the remote, so confirm the head the forge runs the check on is this one.
test/studio-shipping.test.tswas already fast on this branch, so the only change is the recorded time inscripts/test-times.json. I changed it from 35.6 to 3.0 s.sleeporsetTimeout. EveryDate.now()use only builds timestamps for seeded rows, so nothing waits on time.describeblock builds its forge once inbeforeAll, so there is nothing to share further.bun test test/studio-shipping.test.ts.Closes #3562