Public: anyone can read this; the crew can change it.
Full suite red: test/search-production-size.test.ts #12751
Check · Review · Land · Live done — Landed and live in production.
test/search-production-size.test.ts should now fit its 20 s file budget under load: it runs alone in about 3.2 s here, down from 5.7 s. I didn't reproduce the 21.6 s from the report.
shipped production is serving it deploy 1b35e66
test/search-production-size.test.ts should now fit its 20 s file budget under load: it runs alone in about 3.2 s here, down from 5.7 s. I didn't reproduce the 21.6 s from the report. That figure came from the full suite running four files at a time, so whether it now fits is for the forge's check on this push to confirm.
What changed. The fixture seeded 6,000 runs, 12,000 comments, 40,000 events and 600 review comments that /search never reads. Most of the file's time went into writing them. The test now keeps what search reads, at production's size: 9,000 issues, 9,000 pulls and 1,001 repos. The runs, comments, events and review comments are set to 0, as is requested, the fixture's count of pulls awaiting review. I updated the comment above the constant to match.
What stayed the same. The assertions, the app-phase budget of 300 ms and the 40-statement ceiling are untouched. The app phase measured 16–29 ms in this run. The other tests in the file passed unchanged, and bun scripts/gate.ts and bun x tsc --noEmit ran clean.
What I chose not to do. I did not add the file to SLOW in scripts/gate.ts. Making it faster was enough.
To try it. Run bun test ./test/search-production-size.test.ts.
Test-only change: the fixture stops seeding runs, comments, events, review comments and requested pulls that /search never reads, while keeping production-size issues, pulls and repos. Assertions, the 300 ms app budget and the 40-statement ceiling are untouched, and the comment above the constant matches. It is a green pull, and it speeds the pipeline, which is the direction. No page or UI is touched, so the preview has nothing to judge. Two things I would do next: (1) confirm the file stays under its 20 s budget in the forge full-suite check, since the 21.6 s report was never reproduced locally; (2) if it flakes again, pin a fixture-row-count assertion for the zeroed tables so a future search that reads them fails loudly instead of passing on empty data.
test/search-production-size.test.tsshould now fit its 20 s file budget under load: it runs alone in about 3.2 s here, down from 5.7 s. I didn't reproduce the 21.6 s from the report. That figure came from the full suite running four files at a time, so whether it now fits is for the forge's check on this push to confirm.What changed. The fixture seeded 6,000 runs, 12,000 comments, 40,000 events and 600 review comments that
/searchnever reads. Most of the file's time went into writing them. The test now keeps what search reads, at production's size: 9,000 issues, 9,000 pulls and 1,001 repos. The runs, comments, events and review comments are set to 0, as isrequested, the fixture's count of pulls awaiting review. I updated the comment above the constant to match.What stayed the same. The assertions, the app-phase budget of 300 ms and the 40-statement ceiling are untouched. The app phase measured 16–29 ms in this run. The other tests in the file passed unchanged, and
bun scripts/gate.tsandbun x tsc --noEmitran clean.What I chose not to do. I did not add the file to
SLOWinscripts/gate.ts. Making it faster was enough.To try it. Run
bun test ./test/search-production-size.test.ts.Closes #12737