Skip to content

Audit selected JSON locale failures without changing frozen sources - #6165

Merged
Mister-Meeseeks merged 1 commit into
CrocSwap:mainfrom
josecarlosgonzalezroque-design:codex/selected-json-utf8-replay
Oct 29, 2026
Merged

Mister-Meeseeks merged 1 commit into
CrocSwap:mainfrom
josecarlosgonzalezroque-design:codex/selected-json-utf8-replay

Conversation

@josecarlosgonzalezroque-design

Copy link
Copy Markdown
Contributor

The selected-result checker uses the locale codec for its record, certificate and source-manifest JSON. Complete UTF-8 fixtures fail on native Windows CPython 3.12.10 (cp1252): U+211D raises UnicodeDecodeError, and a Unicode input filename becomes mojibake. This package reproduces the failure, verifies python -X utf8 on the unchanged checker, and tests a three-line explicit-encoding patch in private temporary copies.

The checker is itself pinned by frontier6030-source.json, so applying the patch directly would break the selected source closure. Only research/selected-json-utf8-replay/ is added. The patch is reviewable evidence, not a modification to the selected checker. The current selected metadata passes its normal baseline; the failures use complete synthetic fixtures. No new kappa or mathematical claim.

Evidence and reproduction

python3 -B scripts/research_package.py --package research/selected-json-utf8-replay
  • Eight test methods pass in ordinary and optimized-driver registered runs, including native cp1252 CLI execution without the portable codec mock. Checker children retain assertions.
  • All three JSON boundaries, exact Unicode path spelling, malformed UTF-8, source corruption and raw certificate-byte mismatch are covered.
  • The unchanged selected binding-only baseline and explicit UTF-8 checker/baseline invocations pass.
  • Exact committed-byte checking passes for seven package/input files. Official export integrity and declared-input coverage pass. Both Git-free exported replays pass with Windows read-only files, unchanged contents and empty PATH; corrupted source and review-patch controls reject.
  • Publication tree 378985908c2138ee3ef33abd956b3cc07fcb8e2e equals the tested local tree. Full source-history preflight is unavailable in the shallow checkout. POSIX execution and full scientific cold replay are not claimed; hosted CI is separate.

Current main, overlap and credit

Started from canonical main 721c03822c421b7f7fcbb2598993b947c4d88419, priority revision 8. Before publication on October 28, fetched canonical main 3f0152a32eeebac0df11946dbd50cabe5a7f7839. Guidance, catalogue, selected record and audited source are unchanged. The changed upstream package runner was additionally executed read-only at its logical repository path against this package; both registered commands passed. The package retains its original provenance pins.

The portable-replay card came from the verifier-engineering picker, nonce 269949e813c9ca9d693f6a23b2d0d0c2; TASK.json records the refined scope and stop rule. Handoffs, active Discussions and bounded all-state PR searches were checked. The final targeted updated-PR search returned no matches. Remote helper coverage was unavailable without gh; sparse local note coverage was partial. No exhaustive novelty claim.

This differs from #5973's invoice stdout repair. Credit Douglas Colkitt's selected-checker contract, Mihir Srivastava's baseline, and the codec precedents #2531, #1834 and #2554. Prepared for josecarlosgonzalezroque-design with substantial OpenAI Codex assistance, one agent, Apache-2.0.

Open next questions: choose a source-preserving integration policy for the explicit-read patch, and corroborate the same fixtures on native POSIX/another supported Windows interpreter. The five framework assumptions and selected frontier remain unchanged.

@Mister-Meeseeks
Mister-Meeseeks merged commit 8c6b69f into CrocSwap:main Oct 29, 2026
2 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants