[Clear1nsight]

Change and mathematical scope

This source archive connects the waiting-entry implementation to its actual fixed bootstrap, private loader, role constructor and scientific source recipes. It also corrects a reached cleanup defect: if timer cancellation and signal-handler restoration both fail, the caller now retains both caught exceptions and tracebacks in operation order, preserves an earlier working primary, and records uncertain retirement without retrying either native operation.

The package is under research/h25-connected-caller-assembly. Its readable source map binds 61 source roles, seven operating profiles and both complete scientific source recipes. It preserves the original scientific payloads. The assembly is a source review subject with pinned local paths, rather than a portable runtime installation. Genuine runtime roles, grants, current native/compiler subjects, financial provenance and guardian authority remain separate inputs.

Task and overlap

Original investigation: #6321. Immediate predecessor: #6862. Base: 7c49fe32c55de02c2c873655c804b16e4c3fef0a.

• Adapt Add parent waiting-entry source candidate and lifecycle controls #6862 at 47cf8c9af49d22686264db2762700215032eda90: preserve its waiting entry and prior timer-pop correction, then connect its actual caller bindings and repair the distinct final-retirement defect.
• Reuse/adapt Preserve compiler inputs with owned property-query observations #6830 at 39fcd67c991bd72918080b1d2c02feb3d679206d: retain the real bootstrap, loader, role-constructor and scientific source chain. Add readable snapshots of the eight external static source roles omitted from that package's source map.
• Retain the common-runner lineage from research(h25): expose reviewed primary baseline and runner ancestry #6837 at 92bd2213be787cc4451a3b61b00793dad26bebdd. Its primary predecessor-baseline comparison remains a separate integration task.
• Preserve inherited address-space limits in the primary baseline launcher #6847 at efb116491184bd16c5d1ada60d71efdefee61691 and the complete-outcome CLOSE implementation from Add reviewed complete-outcome CLOSE source extension #6762 at 3bcf99910a95248452e955d382c59bd23792a39b are review references for later baseline and retention integration.

Scope is refined to the connected static caller and this specific error-custody correction. It does not change a multiplication theorem, paid profile, selected frontier or claimed runtime speedup. Bounded all-author/all-state PR-body behavior searches and all 193 files of #6830 informed reuse; this is not a global novelty claim.

Validation

Ten bounded controls passed in 0.623 seconds under Python 3.12 with -S -s -P -B, a clean environment, a 20-second outer timeout, 256 MiB address-space limit and fixed file limits. The command was python3.12 -S -s -P -B -m unittest discover -s <local-assembly>/validation -p 'test_*.py' -v inside that guard. It requires the pinned local source inputs; the exported .txt snapshots do not constitute an executable hosted replay.

The controls cover the actual strict source reader, all selected source/profile/control pins, both complete scientific recipes and their predecessor inverse, the actual private-loader module lifecycle, the pure bootstrap argument constructor, and the exact changed timer/finalizer bodies with the real registry and explicitly inert signal/native-verification ports. They reproduce the frozen predecessor's lost-secondary defect; check both clean and existing-primary failures, raw exception/traceback retention, operation order, uncertainty, successful retirement and one call per native operation.

The first versioned attempt passed the timer controls but correctly rejected stale root-relative loader paths. Its failed evidence is retained, and the successor carries forward the predecessor's relative-root repair. No assertion or loader pin check was weakened. All 88 exported file bodies and Git blob identities were read back at c009a7514790e865b664ab8cd4688d37787d7eaa.

The bounded Pro review supports the inspected bootstrap, loader, source recipe, qualifier, entry, relay, receiver and earlier timer-pop connections. It does not comprehensively audit the full native guardian/view, compiler/property-observer theorem, all historical controls, scientific dependency closure or retention paths. A complete exact-commit Pro follow-up accepted the specific timer-retirement correction and its inspected integration, under the existing bounded caught-error and registry contract. That bounded AI source review does not expand the remaining dependency coverage. No financial owner, actual native signal, guardian, compiler, property observer or scientific workload was run. No complete comparison or gain is claimed. Physical and outside costs remain OPEN_NOT_ZERO.

Credit

Tekalign, with substantial OpenAI Codex and ChatGPT Pro assistance. Predecessor and parallel sources retain their original author, Apache-2.0 and source-specific notices. The two publication commits preserve the reviewed failed source subject before the correction.

Open next questions

Complete the remaining independent native/compiler/scientific source qualification before any actual execution. Then adopt the correctly reviewed primary-baseline and complete-outcome CLOSE owners through the genuine caller, source/role bindings and independently validated receiver. A complete baseline/candidate equality result, current resource fit, original runtime issuance, complete cost accounting and successful retention/CLOSE are still required; this archive grants none of them. Start from the public source map and the exact unresolved owner contracts rather than another detached helper design.

[Clear1nsight]

Delivery verified at exact head c009a7514790e865b664ab8cd4688d37787d7eaa: this PR is open and Ready for review, with two commits, 88 files and maintainer modification disabled. All exported file bodies and Git blob identities match the prepared source archive.

Hosted workflow 42417222488, attempt 1, completed successfully for that head. Onboarding helper tests, changed-package selection and the planning job passed; replay was skipped. These are archival workflow checks, not native or scientific qualification.

Ten local guarded source controls passed. A completed Pro follow-up accepted the specific ordered timer-retirement correction and its inspected caller integration, retaining the preceding review's native/compiler/scientific and retention coverage limits. Failed source/control evidence is preserved. No actual financial owner, native signal, guardian/compiler/property observer or scientific workload ran, and no full comparison or performance improvement is claimed.

Canonical handoff on the original investigation; short research-hub pointer. Remaining independent native/compiler/scientific qualification, genuine runtime issuance, baseline/equality/resource fit and successful retention/CLOSE are explicit next dependencies. The source round is frozen; publication grants no new execution or budget, and outside/physical costs remain OPEN_NOT_ZERO.