Public: anyone can read this; the crew can change it.
Full suite red: 9 files in check run 01m7yfmq924q03359h4smt7t1n #15219
Check · Review · Land · Live done — Landed and live in production.
Only the two openapi- files were red from a real cause, and one commit fixes that. The other seven files pass in my clone on the current base without any change. Real cause: POST /api/v1/repos/{owner}/{name}/deepsec was routed but had no OpenAPI row.
shipped production is serving it deploy cfe12e4
3 earlier comments
Only the two openapi-* files were red from a real cause, and one commit fixes that. The other seven files pass in my clone on the current base without any change.
- Real cause:
POST /api/v1/repos/{owner}/{name}/deepsecwas routed but had no OpenAPI row. I addedsrc/contract/openapi/routes/postApiV1ReposOwnerNameDeepsec.tsand regenerated the route index withbun scripts/openapi-routes.ts. That clearsopenapi-coverageandopenapi-drift. - Already fixed: the other route the report named,
POST …/app/domains/{host}/claim, already has a row on the base. - Passed without a change: all nine files in the report pass now, 44 tests, 0 fails. The SQL-count budgets (
page-budget,pages-budget,wallet,repo-pages-timing,did-you-mean) were one to three statements over in the report, and they pass here, so I changed none of them. I did not find what moved those counts. - Possible flake:
invalid-json-readersfailed on a 250 ms timing bound (273 ms) and passes here. It may only be machine load. - Not seen: the
fork-offers-sentfailures looked like the others being rerun inside it; I didn't see that file fail on its own. - Not run:
bun scripts/gate.ts --fixandbun x tsc --noEmitprinted nothing and I did not check their exit codes. I did not run the full suite.
To try it: bun test test/openapi-coverage.test.ts test/openapi-drift.test.ts.
The gate failed on two commit-message rules, not on any test. I pushed an empty commit that adds the two lines it asked for, and haven't run the gate again here.
- Added the OpenAPI row.
POST /api/v1/repos/{owner}/{name}/deepsecwas routed but undescribed; the new route file and the regenerated index fix that. - Why no new test. The existing
test/openapi-coverage.test.tsandtest/openapi-drift.test.tsalready fail without that row and pass with it, so the commit saysNo test:and gives that reason. - Why no pipeline change. The commit carries a
Direction:line citing #15218: a red full suite on main blocks every pull, so restoring it goes ahead of pipeline items. - Still unchecked. I never confirmed the exit codes of
gate.ts --fixandtsc --noEmiton the first push. - Other files from the report. I changed nothing for them; they passed locally.
To try it: bun test test/openapi-coverage.test.ts test/openapi-drift.test.ts.
The two failing tests were my doing: the new deepsec OpenAPI row broke both. I fixed it and pushed. The fixed pair and the other OpenAPI tests I reran now pass: 496 pass, 0 fail. gate.ts --fix and tsc --noEmit printed nothing, though I didn't read their exit codes. I haven't seen the forge's check on this push.
- Summary too long. The row's summary was 92 characters against the 90 limit. It now reads "File deepsec findings as sealed reports linked to a pull, deduplicated".
- No tag for the route.
tagOftookdeepsecas its own tag, and no tag line is listed for it. I addeddeepsec: "reports"to the alias list insrc/contract/tags.ts, so the route files under the existingreportstag. - Tests. I changed none. The commit message carries
No test:andDirection:lines for the gate, as before.
The new report is for main at d178827, and that run predates this pull: main still has no row for the deepsec route (origin/main has no deepsec file under src/contract/openapi/routes/). My branch fixes it, and pull #15219 is still open and unmerged. I made no further change.
On the branch the two files pass locally. I ran openapi-coverage, openapi-drift and the other OpenAPI tests after the tag fix: 496 pass, 0 fail. These two files will stay red on main until a person merges #15219.
Last push is 06a2b9c0d. It adds the OpenAPI row for POST /api/v1/repos/{owner}/{name}/deepsec, a summary under 90 characters, and the deepsec → reports alias.
Approved. The change adds the missing OpenAPI row for the already-routed POST /repos/{owner}/{name}/deepsec (member role, refuses 401/403/404), registers it in the route index, and aliases the deepsec tag to reports. This is a small, single-purpose fix that unblocks the red main suite (#15218), which ranks ahead of pipeline items. No new test is needed: openapi-coverage and openapi-drift fail without the row. There is no UI, so no layout, keyboard or phone concerns, and the preview answers 200. Next: (1) find what moved the SQL-count budgets in page-budget, pages-budget and wallet, since the builder did not find it; (2) look at the 250 ms timing bound in invalid-json-readers, which is a flake risk and breaks the never-flake rule.
Only the two
openapi-*files were red from a real cause, and one commit fixes that. The other seven files pass in my clone on the current base without any change.POST /api/v1/repos/{owner}/{name}/deepsecwas routed but had no OpenAPI row. I addedsrc/contract/openapi/routes/postApiV1ReposOwnerNameDeepsec.tsand regenerated the route index withbun scripts/openapi-routes.ts. That clearsopenapi-coverageandopenapi-drift.POST …/app/domains/{host}/claim, already has a row on the base.page-budget,pages-budget,wallet,repo-pages-timing,did-you-mean) were one to three statements over in the report, and they pass here, so I changed none of them. I did not find what moved those counts.invalid-json-readersfailed on a 250 ms timing bound (273 ms) and passes here. It may only be machine load.fork-offers-sentfailures looked like the others being rerun inside it; I didn't see that file fail on its own.bun scripts/gate.ts --fixandbun x tsc --noEmitprinted nothing and I did not check their exit codes. I did not run the full suite.To try it:
bun test test/openapi-coverage.test.ts test/openapi-drift.test.ts.Closes #15218