跳转至

Historical version

This page records design, operation, or validation facts from an earlier release. It is not current V7.17.3 guidance. Start from the documentation hub for current instructions.

V7.12.0 Audit and Handoff

Scope: reliability candidate work, executed-test evidence, governance routing, quick status diagnostics, LIGHT guidance, and the matching bilingual release material.

  • F01 has 50 targeted regressions and 16 passing stress repetitions. F03 retains complete-history validation, reuses one identity read per execution, and has 20 samples at each of 100, 1,000, and 10,000 scale; it does not claim improved asymptotic complexity.
  • F02 completed three-scenario evaluation. The candidate shortcut is not delivered and the original LIGHT rule is restored. The simple tool median changed from 6 to 5 (-16.7%, below the suggested 20% target), duration did not improve, and resume baseline passed 5/5 while candidate passed 4/5. This does not attribute the result to the rule or support general, stable, or causal efficiency claims.
  • F04 has 50 passing targeted tests. In 20 paired same-account concurrent native samples from historical payload 948c, full/quick medians were 2924.6905/110.248 ms and p95 values were 3521.541/115.587 ms, a local 96.23% reduction. The current final payload was not remeasured. Quick status is not complete validation or authorization and this does not generalize to other environments.
  • F05 has 52 passing cases. The original 16 independent host tasks had 13 PASS and three FAILED results, retained as history. Combining the fixed profile's unchanged SHA-256-bound evidence bytes for 13 original cases with three targeted follow-up tasks yields HOST_FINAL_REPORT coverage of 16 observed actual independent tasks, 16 PASS and none missing. It does not claim a single-cohort pass rate or an independent router trace, and applies only to this profile. F06 completed 19 capability-group reprojections and nine focused tests after complete local execution.
  • The security review round two passed. Delivery review round two only found stale release documentation, now corrected. Both are logical-readonly reviews, not system-isolated reviews.
  • Complete local package validation executed 617 package tests (616 PASS, one platform-conditional skip) and 231 PASS runtime tests against the earlier uncommitted candidate based on 6641ed24626925b3445a88014621ed530f8c2ee8; this is not final-source CI.
  • Advance the CLI anchor to 0.156.0 under the unchanged current-plus-ten policy. The account layout now includes sibling seal_worker.py; a missing entry fails before process creation. Installed append/seal, missing/tamper detection and removal regressions pass. This delivery-path delta received scoped INLINE inspection, not a new independent review.
  • Current installed payload 4d16bf233697500fd2f4087f356037dbe61aa93af61670b289b44cd0b1a0abc7 matches across source, marketplace and cache (245 files). Installed, enabled, HOST_COMPATIBLE and the backup were read back. A fresh parent/child task emitted all five real event types, with five records at SEALED_CURRENT.
  • This host's elevated sandbox reports a reparse-point error while pinning a directory. Acceptance processes use the documented temporary unelevated fallback with a read-only policy and unchanged global configuration. This does not claim that elevated setup is repaired. Final-commit CI and public-release acceptance remain separate.
  • Two bilingual ZIP candidate builds were reproducible, but documentation changes require rebuilding them. Commit, push, tag, CI, rebuilt artifact, GitHub Release, and additional installation/effectiveness evidence require separate readback.

  • New 0.156.0 host acceptance retains 13 original passing independent tasks and the host execution error that stopped that cohort. A targeted follow-up covers the interrupted case and two unstarted cases. Verified raw metadata, projections, report hashes and actual task IDs produce 13+3 combined coverage of the unchanged 16-case profile, all passing with no missing cases. Execution uses temporary unelevated plus read-only policy. Evidence remains HOST_FINAL_REPORT, not one-shot success or an independent router trace.