58 lines
7.9 KiB
Markdown
58 lines
7.9 KiB
Markdown
# Independent rename-plan runtime reconciliation
|
||
|
||
2026-09-24. Final tested source **18073a6f2eabce83fc83997f094e331eec835a16**, baseline **ee500908e3d77182ddd54e12524f9b6c6eb43b60**. No material discrepancy found in final recorded gates or their qualified disposition: **17 new tests pass; strict Clippy passes; full library remains FAILED230pass/16fail; formatting remains FAILED**. Preserved upstream128KiB pipeline failures remain an unresolved limitation, not a fixed production issue.
|
||
|
||
## Read-only evaluator and source binding
|
||
|
||
Read the complete parent checker before executing it; earlier review's duplicate-list, summary-status/filtered-count, focused-status, raw-baseline and Git replacement-object concerns are addressed in this version. Executed only:
|
||
|
||
```text
|
||
python -B .local/analysis-mixed-index-seam-20260924/inspect_rename_runtime.py 18073a6f2eabce83fc83997f094e331eec835a16 18073a6 17
|
||
```
|
||
|
||
Assertions enabled; exit0. No Cargo, application binary, archived launcher or debugger was executed. Checker writes only a fresh readback JSON. Additional reviewer snippets read raw JSON/logs/ZIP and compare hashes; no source edits, commits or remote writes occurred.
|
||
|
||
| Reproducible artifact | Bytes | SHA256 |
|
||
| --- | ---: | --- |
|
||
| inspect_rename_runtime.py | 7468 | `68e428c8959ab8a98a06ac240fc379c734646e38cbb2ca4cdec47a3672ad3c02` |
|
||
| RENAME-PLAN-RUNTIME-READBACK.json | 2211 | `9b2d1631e8981ab01aa133cbc5cd41b407e9a40280f4f91921ce9c1f6500cc11` |
|
||
| Final test executable | 42,081,792 | `2c8bd7b9a6adfda3f74b74bae00453b1b1348cbed47f04c8e8c1de9152899416` |
|
||
|
||
Current exact executable at `.local/analysis-checked-save-target-20260913/debug/deps/oa_core-416e35542925bed6.exe` was rehashed against final focused/full/list records and their raw invocation paths. This is observed recorded-build binding, not authenticated compilation/reproducibility proof.
|
||
|
||
All **551** declared input paths from immutable final Git tree match raw disk lengths/bytes/SHA256 and each final gate input manifest. Git replacement objects were disabled. Current checkout is the exact source with no tracked changes. Shared final input-manifest SHA256 **e2c710230fa55bbefb4faeee73dab6b8104ddfb8f0291eef0eba15850799c2ff**. All five receipts report unchanged inputs, no verification failures and same final head; captured launcher SHA256 **843069930cd44278eef940930369b5a85a3b3412e1e50793d78d3d5738298fbe** matches before/after and actual captured bytes. Exact Cargo argv/cwd match intended commands; one compiler job, incremental0 and no lifecycle/RUST_MIN_STACK overrides are recorded. The declared input scope/environment subset is not claimed exhaustive toolchain or host attestation.
|
||
|
||
## Actual final results
|
||
|
||
| Gate suffix18073a6-1 | Exit | Raw outcome | Log SHA256 |
|
||
| --- | ---: | --- | --- |
|
||
| focused | 0 | 17pass,0fail,0ignore,229filtered | `193e7d2c82c65f2bb04d1d818835635152b702b5da56cae9c16fc6955f1cf091` |
|
||
| full | 101 | 230pass,16fail,0ignore,0filtered | `a3fef7c640ec9e59484d15b18a5fc8b9a0c25ca9bf01750b1da4431a101c4eec` |
|
||
| list | 0 | 246 unique names | `05e09d5280508a784dd97a894c92b8cdeb202c67592498d253e8f8c33b07e6ae` |
|
||
| clippy | 0 | core/all-target strict `-D warnings`; completed dev | `5f9c5607b71d64c343e6be3278aeba235ce95af6b9d28d8de7e2679560afa663` |
|
||
| fmt | 1 | unchanged inherited formatting diff | `c1c6b12861c8117ce6ffc6d6742ebf92618aafb10160ec6f99d370ca72ebab2d` |
|
||
|
||
Parsed every actual named result and compared receipt order, uniqueness, summary status/counts/filtered counts and expected exits. Listing set equals full-result set. The17 new names equal the exact focused set, all pass in both runs; all229 baseline named outcomes retain their prior result (213pass/16fail). All16 raw failure bodies compare exactly after CRLF handling and thread-PID-only normalization; no semantic diagnostics are removed. Final formatting raw bytes equal baseline, retaining eight diff blocks across five files.
|
||
|
||
Separately anchored baseline receipts, raw logs and input manifests for ordinary-full/fmt-ee50090-1 to the previously delivered50-member archive SHA256 **99ee3761a77c2f634d36d4cdd0c3ec6d305cbd2f3322cf37a2f25f21e6d9fbed**; every compared byte matches. Thus baseline was not silently revised to permit final failures. Exact final result-record hashes are captured in the readback JSON.
|
||
|
||
## Preserved failures and diagnostic scope
|
||
|
||
Read/rehashed original records and logs; validated captured ordinary/diagnostic launcher and input-manifest pins. Ordinary focused attempts9388fcf,70c20c3,f1c0956 each record exit3221225725 (`0xc00000fd`, STATUS_STACK_OVERFLOW), exactly four completed passes and no terminal test summary. They are not passing17-test runs. Log SHA256 respectively:
|
||
|
||
- 9388fcf: `92c0ef7acfaced00ec32a6f66203dd5b7e59fa3c10f844942c0f1e9291dc1a4d`.
|
||
- 70c20c3: `5ec1947d09e48c61c98d2396d176dd6acd6d8f72bbabab5ccca402908ce5faba`.
|
||
- f1c0956: `6e0542f463091d5b3924242c1aa4824dc9ab6dd076d22a018b09a9ada81d55d0`.
|
||
|
||
The separately recorded original9388-artifact invocation uses explicit `--skip` for only the small-stack case. Raw16pass/0fail/230filtered results equal precisely the final17-name set minus that case. Receipt artifact matches original focused42,074,624-byte executable SHA256 **bd927d4653a8915ee72d5df22096d29edb1853c12d929e20056647b05e565956** and records unchanged bytes; its read source-reviewed launcher pins/checks the artifact before running. Remaining16 log SHA256 **5ce011d7c1d09165363c44fb3dc248c3aaa8b916ce1a2bac0896a76d6814501c**. Historical binary bytes have since been replaced at the shared target path; this review rechecks their preserved receipts/linkage, not a currently retained original executable.
|
||
|
||
Diagnostic9e8e77e log **c4093678295bae3ef37ec125e707069187ca2d52dac6f09a00eaf3b28313a705** shows deep admission, original clone, successful planning/drop and desired clone completed before collision helper. Diagnostic8159a17 log **5adeb05bb91878a090297c7a0580609ed8d98d9fdb5eb73039214ae0d5b35e5d** adds `entered; before authored`, with no after-authored marker. Both record the same stack-overflow status. Those markers locate that helper attempt before its planning call.
|
||
|
||
All three debugger receipts identify the same prior f1c artifact SHA256 **b804768eccb350033fcfd01516d0c3253e43525718f42de7c75cda9a3d91c6fe**; their launcher bytes match recorded hashes. Debugger exit0 is **not** child-test success: run1 stops on stack exception, run2 shows a single test pass/245filtered under debugger, and run3 stops on stack exception then explicitly kills the process. Run3 log SHA256 **f10d5bcf20b0c4d64cb5e07eacfe7d19999a594f55a033e1e9df7b0ca79ffca0** contains a real backtrace through authored→mixed parse→Parser unit→typed declarations→binding→exact literal→Budget exact→Exact decimal→BigInt parse→Windows heap allocation. No rename-plan call appears in that stopped call path. Run2 log **4d707adebbd3c65b89aa558a99d86b463dadffcced6a9a1e98e50bdca1b77f1f** is retained as diagnostic variation, not ordinary acceptance. This supports an environment-sensitive upstream128KiB boundary, not a proved exact frame-size cause or repaired parser.
|
||
|
||
## Accepted bounded conclusion
|
||
|
||
Final18073a6 production is unchanged across these evaluator revisions. As independently reviewed in RENAME-PLAN-TEST-REVIEW.md, its planner-specific small-stack case performs all six deep128/wide129 × unchanged/collision/non-name cases with full original owner moved into a128KiB scoped worker, borrowed desired, exact outcomes and worker-local destruction. Admission prerequisites are prepared outside that worker. The final raw focused and full results both pass that named test. This establishes the accepted planner-scoped gate; **it does not establish that full parsing/admission reliably fits128KiB** or erase earlier aborts.
|
||
|
||
This is private rename planning with retained source authority and sizing, not accepted rewritten output, public source8, checked persistence, history adoption, GUI/API parity, a release build or full application completion. Full-library16 failures and formatting failure remain explicit. Parent owns packaging/publication and any next capability.
|