4.4 KiB
Independent entry-producer package readback
Pass for this finite evidence package; no unresolved material discrepancy. Reviewed source 29417130c71f70c4eaefe8ed4c879efba78eb72c and the current artifact before releasing the source freeze. This audit used file reads, independent data comparisons and read-only Git. It did not execute archived helpers, Cargo, a compiler, debugger or product/test executable, and did not access the network.
The immutable archive is exactly 434,046 bytes / 53 unique members, SHA256 19a87583a8dc4c22f35999044de5cac2c2c97b04fb457d2c3df3f6d2c054c865. CRC checks pass; every member name, byte count and SHA256 matches manifest.json, whose SHA256 is 074fbdf32ef7012a31d56ed723f04813edc5b4c5ba2b9fed260b6f2f7610600c. Names have no absolute paths or parent traversal. All 51 file-backed members equal their original local files; the other two members exactly equal immutable Git diffs from97a8703 to a629ae0 and2941713. All seven readable mirrors equal both archive and original review files.
The baseline full/list/fmt members also match the previously frozen text-producer archive, SHA256 252ea94e7c7f500a2afb3d0fa47c43a051f39d9c844328337289eea0c463becd. Independently checked all556 baseline and557 initial/final declared inputs against their respective immutable Git blobs. All557 final files also matched raw disk; HEAD was2941713 with no tracked changes. Focused/full/list bind the same current 42,898,944-byte executable, SHA256 adb0bd907f614c512bc3883dfe9a88ca0e4268638d85f4899ad07a5c1d3ec4e9, independently reread without execution. This is declared-input/artifact concordance, not exhaustive compiler or undeclared-dependency provenance.
Reparsed all nine gate receipts and raw logs, checking log/input/launcher hashes, source stability, exact recorded outcomes and listings. The five README commands equal the final argument arrays. Final results are six focused passes, 265 passed /16 failed,281 unique listed/runtime names, strict core all-target Clippy exit0 and format exit1. All275 prior outcomes and ten existing stack controls are preserved. All16 failure bodies match97 after replacing only the numeric thread identifier in Rust panic headers. Format output is byte-identical: eight blocks in five files, SHA256 c1c6b12861c8117ce6ffc6d6742ebf92618aafb10160ec6f99d370ca72ebab2d. Recorded jobs1/incremental0 and absent stack override agree with the stated environment. The captured launcher records unfiltered combined output and actual return codes; its source/input/helper checks support the stated bindings.
The initial a629ae0 E0277 compile refusal is preserved, with zero outcomes/artifacts. The exact a629ae0→2941713 diff changes only the quota test's unwrap_err extraction to explicit Err/Ok matching and its implementation note; production and assertions remain unchanged. The archived source/test reviews retain both precompile latch repairs and the separate-roomy-owner inventory evaluator correction. None is presented as an erased or successful initial run.
Read all six baseline/final prologue records and their capture helpers as source. Their source/artifact/helper identities and raw immediates agree with STACK-OBSERVATIONS.md: selected parser changes sum2,048 bytes; binding2520 plus prefix11720 stays14,240; continuation13416 is a separate alternative. Projector616→1032 and ordinary584→80+1000 are accurately distinguished from parser phases. Worker7368 and shim56 are not confused. These partial compiler reservations omit other stack costs and do not establish total headroom or a new128KiB strong-mode pipeline guarantee. Prior projector observations are correctly described as newly captured baseline evidence, not retroactively part of the older archive.
Reviewed README, current STATE/DECISIONS statements and readable reviews for scope; checked102 local Markdown links with no missing targets. Parent corrected the minor README/STATE thread-identifier terminology; archived historical checker labels remain unchanged. Final reviewed README SHA256 is 6a8f82a51b9ed4bee14172c1d17aaea4a2d3cb87a7b9bf1ebd4b9e7802ea99b6. The documents accurately preserve local-only publication status, failed inherited gates, finite retained-accounting exclusions and unfinished output acceptance/public editing. The reported remote storage failure and last remote commit are contextual prior observations, not independently revalidated by this offline audit.