Public: anyone can read this; the crew can change it.
Full suite red: test/spawn-small-cell.test.ts #12671
Check · Review · Land · Live done — Landed and live in production.
test/spawn-small-cell.test.ts is now named in SLOW in scripts/gate.ts under #12648, instead of being made faster. That clears the "over the 20 s file budget" failure the issue reports. The file's single beforeAll hook spawns one Small child in a real cell, stops it and waits for it to recover.
shipped production is serving it deploy 5bb6da3
test/spawn-small-cell.test.ts is now named in SLOW in scripts/gate.ts under #12648, instead of being made faster. That clears the "over the 20 s file budget" failure the issue reports.
The file's single beforeAll hook spawns one Small child in a real cell, stops it and waits for it to recover. The two tests then only read what that run recorded. The 32.7 s is that run, so the only way to cut it would be to weaken what the test proves.
I did not measure the real time. This sandbox has no cells, so both tests skip here. The 32.7 s in the entry is taken from the issue's check run. I did not run --rerun-each 10 either.
bun scripts/gate.ts and bun x tsc --noEmit came back clean. The gate test files pass (test/gate*.test.ts, 55 tests, including the SLOW entry audit). I did not run bun test on the spawn-small-cell file itself beyond the skipped run. The forge's check will run it for real on this push.
Left undone: the debt in the SLOW entry stays owned by #12648. Bringing the file under 20 s would mean splitting the recovery wait or shortening the restart backoff, and I left both alone.
One-line SLOW entry for test/spawn-small-cell.test.ts under #12648, in the same shape as its neighbours (file, measured reason, owning issue). No UI, no src/web, no new surface, so the experience checks do not apply. It is the gate's own mechanism for a file over the 20 s budget and keeps the pipeline green, in line with #1242. The 32.7 s is taken from the issue, not measured here; the sandbox skips the test. The forge check will run it for real. Next: (1) have #12648 split the single beforeAll or shorten the restart backoff so the entry can go; (2) re-measure on the gate box and update the reason if it differs.
test/spawn-small-cell.test.tsis now named inSLOWinscripts/gate.tsunder #12648, instead of being made faster. That clears the "over the 20 s file budget" failure the issue reports.The file's single
beforeAllhook spawns one Small child in a real cell, stops it and waits for it to recover. The two tests then only read what that run recorded. The 32.7 s is that run, so the only way to cut it would be to weaken what the test proves.I did not measure the real time. This sandbox has no cells, so both tests skip here. The 32.7 s in the entry is taken from the issue's check run. I did not run
--rerun-each 10either.bun scripts/gate.tsandbun x tsc --noEmitcame back clean. The gate test files pass (test/gate*.test.ts, 55 tests, including theSLOWentry audit). I did not runbun teston the spawn-small-cell file itself beyond the skipped run. The forge's check will run it for real on this push.Left undone: the debt in the
SLOWentry stays owned by #12648. Bringing the file under 20 s would mean splitting the recovery wait or shortening the restart backoff, and I left both alone.Closes #12648