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

Sentinel: shared:01m23vygyffx5hg2f5vm8r74sr:228tok7cq4 is past its budget: trevato/forge: "every registered /api/v1 route has an OpenAPI row or is listed as not yet described" failed on 3 pulls … #8152

Closed opened this
Closed without merging

I changed the test so a loaded check machine can't turn its file scan into a red verdict. I never saw it fail, so this is a likely cause, not a confirmed one. The test passed on this branch, and bun x tsc --noEmit is clean.

Notes

stopped closed without shipping deploy 1914e5b

opened this

I changed the test so a loaded check machine can't turn its file scan into a red verdict. I never saw it fail, so this is a likely cause, not a confirmed one.

The test passed on this branch, and bun x tsc --noEmit is clean. I didn't run scripts/gate.ts or the full suite. I couldn't check what the three failing pulls actually hit, because I didn't pull their logs. If they were simply pulls that registered a route without an OpenAPI row, the test was right to fail them and this change doesn't affect that.

In test/openapi-coverage.test.ts, the route scan now runs once and is shared by the two tests that need it. Each of the three tests has an explicit 30-second timeout in place of Bun's 5-second default.

I did not touch the OpenAPI routes or NOT_YET_DESCRIBED, which is already empty and still correct.

To try it: bun test test/openapi-coverage.test.ts.

🤖 Generated with Claude Code

Closes #8140

1 earlier comment · 1 event
commented

I changed the test so a loaded check machine can't turn its file scan into a red verdict. I never saw it fail, so this is a likely cause, not a confirmed one.

The test passed on this branch, and bun x tsc --noEmit is clean. I didn't run scripts/gate.ts or the full suite. I couldn't check what the three failing pulls actually hit, because I didn't pull their logs. If they were simply pulls that registered a route without an OpenAPI row, the test was right to fail them and this change doesn't affect that.

In test/openapi-coverage.test.ts, the route scan now runs once and is shared by the two tests that need it. Each of the three tests has an explicit 30-second timeout in place of Bun's 5-second default.

I did not touch the OpenAPI routes or NOT_YET_DESCRIBED, which is already empty and still correct.

To try it: bun test test/openapi-coverage.test.ts.

🤖 Generated with Claude Code

Changes requested — before the latest push

No security finding: this is a test-only change. Two things to fix, though.

  1. The cause is a guess. The author says they never saw the test fail and did not read the logs of the three failing pulls. A 30s timeout on a synchronous scan of a few source files hides a real failure instead of fixing it. It also works against #1242, which wants a time budget per test and per file. Read the three failing logs first. If the pulls simply had no OpenAPI row, the test was right and this pull should close with no change. If it was a real timeout, say how long the scan takes under load and fix that cause, for example by scanning once. Keep the default timeout, or set one far tighter than 30s.
  2. The memoisation moves the doc comment. The registered docblock now sits above let memo, and scan has no doc. Put the docblock back on scan. Also add a comment saying why the cache is safe: the scan reads source files that do not change during a run.

Also, no test fails without the change, so the rule that every behaviour needs a test that would fail without it is not met. I did not run the preview: the change touches no served behaviour.

commented

Declined: I couldn't find the cause of the failure, so I've reverted my change and the branch now matches main (fd794a7b) with no net diff. Run log: https://forge.open-platform.sh/trevato/forge/runs/01m6d9s260e88hybrayvtjaky4

Sign in to join this conversation.

↑↓ navigate ↵ open ⌘↵ its app esc close

Keyboard shortcuts

Docs