Repair gzip portability for frozen #332; conditional κ unchanged

[eumemic]

κ = 7.77268259042371640048112e-4

This is a portability repair for frozen #332, with the same conditional κ, scalar word, complete paid profile, supplier packages and exact assembly. The original head f9136e76568100034aa37337251f93d684496c13 remains unchanged.

The #332 Linux run passed all twenty preceding stages, including regeneration of all five complete source payloads, then failed finish.py:41 because it required identical compressed gzip container hashes. The source producer already documents that gzip bytes depend on the platform/zlib build and compares their complete expanded contents. Both local full replays of #332 had passed; that did not establish cross-platform compressed-byte identity.

The final admission now requires the exact five filenames, binds each actual stored and regenerated file to its own recorded container hash, and compares every expanded byte of both gzip files. The three plaintext files still require raw byte equality, and every expanded hash is checked. No JSON normalization or selected-field comparison is used. The complete stored package remains pinned by its manifest; all vendored sources, selectors, native mathematical checkers, profiles and arithmetic engines are byte-identical to #332.

A mandatory fresh control stage accepts valid changes in gzip representation and rejects altered full payloads, forged hashes and missing/extra entries. Independent controls also reject missing physical files, equal-cardinality key substitutions, JSON-equivalent formatting changes, stale hashes after recompression and invalid gzip CRCs. The workflow now uploads the producer-regeneration and control receipts for diagnosis.

Two new complete immutable replays actually exited 0 under Python 3.11 and 3.12, each passing all 22 stages. Their mathematical certificates match each other and #332 after excluding only top-level receipt hashes, whose referenced files include timings and paths. The independent five-file gate rechecked every actual container hash and complete expanded payload. The package contains 402 pinned files plus its manifest. Independent adversarial review passed four valid cases and rejected twenty invalid cases, including tests on the real regenerated gzip payloads. The complete stored package is checked against the committed Git blobs before publication. Linux CI for this new head is pending; the prior Linux failure is preserved as evidence.

Reproduce with the pinned dependencies, C++17 and Boost:

python -m pip install -r research/p10b-oriented-constructed42/requirements.txt
python -B research/p10b-oriented-constructed42/verify.py --output /tmp/p10b-portable-proof --cxx g++ --boost-include /usr/include
python -B research/p10b-portability-audit/controls.py --package research/p10b-oriented-constructed42 --regeneration /tmp/p10b-portable-proof/source-regeneration --output /tmp/p10b-binding-independent.json

The inherited all-size compiler, dirty lifting, chart, restored-row, covering, selector, routing, growing-prime, precision/recovery and analytic-transfer assumptions remain explicit in the unchanged construction documents. This repair makes no new unconditional theorem or Lean-build claim. The separate repository-wide joint-dual pin failure remains covered by #318.

Prepared and independently checked with substantial OpenAI Codex assistance. All original contributor credits and inherited AI disclosures are preserved.

[Mister-Meeseeks]

Thank you @eumemic. This contribution is now retained on main in the scoped 100-PR review and contributor credits.

Contribution: Portable producer binding for the oriented p10b word.

Review and limits: Compare all expanded gzip bytes and every plaintext byte, pin both containers, and reject forged hashes, omitted/extra files and altered payloads. The scientific construction and kappa are unchanged.

The complete original source, notices and relevant execution evidence are retained through the pinned source archive. The review distinguishes fresh local runs, source-bound historical Linux CI and prior exact-head review; it does not claim 100 fresh full mathematical audits. Closing this historical PR as reconciled. This preserves parallel and cumulative credit without replacing the selected frontier or promoting priced research targets to completed constructions.

Reviewed head: f04a11279fe1eedb54ac97c80d5a71ddc0275b13. Publication: 4040a0f46d67b41c36a472a7cd4af791fd4c36cc.