[rmcdaniel]

The original #214 main run failed because test_wrong_h_refuses_under_python_optimization exceeded its five-second child-process timeout before producing the expected diagnostic. Its failed job is retained; no rerun was requested. The same source had passed PR CI. This evidence establishes the timeout, not a particular runner startup hook as its cause.

Run that source-only control with isolated Python startup and explicit optimization (-I -S -B -O), retaining the wrong-package refusal assertion. A synthetic sitecustomize.py countercontrol proves ordinary startup can intercept the child, while the isolated optimized child still reaches the exact package guard. Each inert child has a bounded 20-second test timeout. Runtime/parent/guest/cleanup ceilings are unchanged.

All 99 inner controls pass locally, including the new startup countercontrol. Diff whitespace, public boundary and changed-file secret/private-path checks pass. This changes one test file; every selected R 86f65e1 runtime member, manifest, outer command and package/fixture input is unchanged. Normal CI follows; no Docker, Cargo, lifecycle/native/consumer or publication operation was launched locally. Automatic docs paths do not include this test and release is manual-only.

Follows #214; refs #191 and #192. sdk-rust owns the CI correction and normal green integration. The original main NONPASS and specific effect/review gates remain explicit.

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[rmcdaniel]

Terminal actual-main receipt: #215 normally merged as 933f5a2164b61861eb63e80162e05722a9c3d9ad, tree 17904fe7c5ae207ce25fad8964867dd97a6c51a9, ordered parents [81e8ee7, df74098], independently read back. Original main CI 42098417172, Boundary, Sonar and all four CodeQL analyses are terminal successful: 16 successful checks/two intended Server skips; none pending/failed. The hosted run confirms 99 inner +47 outer controls and 715 unit +53 corpus tests on both Rust 1.86 and stable.

The preceding #214 main run 42096838917 remains its original NONPASS from the five-second optimization-child timeout. It was not rerun, canceled or relabeled. This one-test source correction passed fresh PR/main qualification. Every selected runtime member and the fixed real parent/guest/cleanup ceilings remain unchanged.

The separate unexecuted source/input packet is sealed at index SHA-256 d2c508014206b8d13ed8309bf35535475636bdcbaaccb24d8a2585814c1e0393. It binds inner R 86f65e1ca1b01baaa68f8d7256a7699dfda5f828, 13-runtime-member manifest 2b0cfb21bd1a051bf8222879a304dadb87e93a4873693eb6e6978671313ca715, outer a689a2fba28db34c888777c9ed5686afe5c615e2, clock arithmetic/domain/receipt contract, captured private-command source, staged guest argv and actual source qualification. The standalone public Git bundle is 2,795,280 bytes / SHA-256 ed43dd53a72d74bf1c464e99ad7bb27b7cf63940ad9fcfe729ea2b94a8fbbeb5, with exact R, published H and annotated 4.12.0 tag. Its independent bare-Git verification checked all five H fixture/official-consumer blobs without executing bundled application code. The index and all 13 data-packet members independently pass length/hash/type/link/mode readback. Native/final identities and actual effect selection remain unbound; this is not a runnable intent.

Both merged head branches are deleted locally/remotely, and each remote absence was verified with ls-remote --exit-code returning 2. Worktrees, predecessor branches and Git history are preserved; the local checkout is clean at accepted main. No local Docker/VM/store/consumer allocation occurred, and the sdk-rust container prefix is absent. Only ordinary source workflows followed; protected docs remains run 41959739724 at 46cf3651, with no new docs artifact or release.

Remaining actions/owners: sdk-rust owns the prepared successor and #191/#192's eventual installed proof. General owns return of actual changed authority for the specific policy-access blocker, current default-engine/foreign-workload owner handoff, and any separate exact lifecycle/native/installed selection. The selected host still lacks an admitted empty-engine/10-GiB-memory window. Actual launchd descriptor consumption, guest Git/Python/Docker/non-root/boot/device/daemon and clock/readiness/cleanup evidence remain unobserved prerequisites, with missing capability a refusal and no installer fallback. General has the private indexed handoff and has been asked for these concrete decisions. The frozen df617fb/bdbb4d4 packet is unchanged; no blocked response was retried/rerouted or claimed passed. Source integration does not close either issue or clear an operating/publication gate.