[rmcdaniel]

Every cold acceptance process has its own worker ID. The parent registers the setup worker, but the fresh child calls run_once without registering its new ID, so published native Server3.10 correctly refuses its first poll with HTTP409 worker_not_registered. Register that child before polling, preserving the same workflow definition, original workflow/run, four process boundaries and one-observation/history/result assertions.

This changes only acceptance-fixture setup. The bounded #270 diagnostic run is the actual predecessor failure; no logger/library/package change or new credential is required. Relates to #269 and durable-workflow/.github#227.

Validation in progress: ordinary pre-client controls/compile against the unchanged exact candidate package, exact-head CI and source review. Then bind the new test/build/parent to the reviewed manual envelope before one bounded installed repeat. All preceding NONPASS records remain unchanged; no tag/upload or logger PASS is claimed.

[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]

Exact source7b6a3514dd9f7c6454a1b753a5b5247cab064722 / test SHAe66d177c6a1ffc486623169d1505766447c18c3c4650a2023a60f8bb9f9d11f0. Ordinary pre-client controls PASS3, real SDK flows ignored2, using Cargo1.86/offline build against unchanged candidate4.17 archive6de774 and all118 byte-equal installed members. Source/vendor/manifest stayed RO; the control container is absent. The initial staging command returned before the test-file copy completed; Docker refused the nonexistent bind before creating any container. That setup NONPASS is retained, and the repeat began only after complete source/hash verification.

The actual predecessor is the reviewed e943 diagnostic run: the first fresh child's poll receives HTTP409 worker_not_registered because its phase-specific worker ID was never registered. The one new line registers that same worker before run_once, with the original definition, queue and all scenario assertions unchanged. No product replay/codec fix or corpus change is needed for this acceptance-fixture correction. No new credential or SDK release is selected for the registration operation itself.

New test/build/parent/controller identities are returned separately on #269 before another bounded runtime; this source/compile result does not qualify installed logging or publication.

[rmcdaniel]

All19 checks on exact7b6a3514dd9f7c6454a1b753a5b5247cab064722 are terminal (17 successes, two intended Server qualification skips), mergeState=CLEAN. Source remains frozen; the actual predecessor HTTP409 worker_not_registered, offline controls and data packet remain bound here. No runtime has executed the corrected registration fixture yet. Exact source/data counterreview still precedes the one bounded observation; no package/tag/upload claim.

[rmcdaniel]

Security exact source/data review of 7b6a3514dd9f7c6454a1b753a5b5247cab064722 / tree 7a7016f7d51a3e9354d40cff685a5bc239f686eb and the bound #269 packet: PASS for one bounded diagnostic observation under the existing #269 safeguards.

• The source delta from reviewed Retain bounded cold-worker failure diagnostics #270 8b006e1 is only the diagnosed fixture correction: retain the constructed phase worker, register that same instance, then call run_once(). Definition, queue, phase ID, four-process/log/history/result assertions and bounded child-stderr handling are unchanged. This directly addresses the retained real HTTP 409 worker_not_registered; it does not infer an auth or product-logger defect.
• Exact-head Register each cold logging worker before polling #271 now has 19 terminal checks (17 success, two intended skips).
• I independently rehashed the retained registration root: test e66d177c... equals the exact Git blob; receipt 4ac6427e..., manifest 8b8fcaa2..., lock f01b9aba..., build messages 28b558a6..., and 22,396,960-byte parent 217aaf91... all equal the public receipt.
• The public controller rehashes to 2e2c2e486683b94abaf00c6cdb0ce589ca905f71033e0098b2da9bc0b32098b5. Its complete diff from reviewed e9437c72... is exactly the receipt, binary and test pins; all runtime/network/source/limit/evidence/cleanup code is byte-identical.

This permits one new diagnostic observation only. Preserve the five preceding NONPASS outcomes and return the exact result/cleanup evidence for counter-review. No installed logging PASS, SDK/package change, #980 result, tag, upload or publication authority is inferred.

[rmcdaniel]

Registration fixture accepted on main 8e13af5, tree 7a7016f7d51a3e9354d40cff685a5bc239f686eb, exactly equal to reviewed7b6a3514. Remote topic branch is deleted and independently absent. All19 PR checks (17success/two intended skips) and all four actual-main workflows pass: CI, Public Boundary, CodeQL, SonarQube.

The exact corrected test then passed the real four-cold-worker candidate-package scenario, with original identities/history and one new-region log per process; full terminal record, independent terminal/source/base PASS. Earlier fixture/launcher outcomes remain NONPASS. This completes the fixture correction, not registry publication or other-SDK/#980 acceptance. The exact4.17 package and publication follow-through stay on #269.