3.5 KiB
Independent archive readback — 2026-09-24
No discrepancy found in this bounded delivery check. Reviewed the packaging script as text; did not execute it, archived evaluators, Cargo, product binaries or source tests. No staging, commit or remote mutation was performed. This report is additional to the frozen 50-member archive, not an archive member.
Independently read evidence.zip: 456,290 bytes, SHA256 99ee3761a77c2f634d36d4cdd0c3ec6d305cbd2f3322cf37a2f25f21e6d9fbed. All 50 member paths are unique, including case-insensitive comparison, and safe relative paths without traversal, backslashes or drive separators. Exact member set, byte lengths and SHA256 match manifest.json; every member CRC and ZIP readback pass. Manifest raw SHA256: 47040569ad955304e9266f5ba006ed089f84fc116038e977c7c5b43d3c5cac62.
Compared all 48 non-diff members byte-for-byte against their original files in .local/analysis-ordinary-index-gates-20260924, .local/analysis-mixed-index-gates-20260924, and .local/analysis-mixed-index-seam-20260924, following the builder's explicit mappings. Independently regenerated both diff byte streams with git --no-replace-objects diff --no-ext-diff --no-textconv: 50a5918fd64514b49ebfdb8881f7f388c02135bd → 147b247398cdae4b0fbbf977e301fffcaa450108, and 147b247398cdae4b0fbbf977e301fffcaa450108 → ee500908e3d77182ddd54e12524f9b6c6eb43b60. Both exactly match archived patches.
Read the actual staged Git blobs using git show :path for all nine existing delivery files, including ZIP, manifest, README and .gitattributes: every staged byte equals its raw working-file byte. The five readable review/script/result mirrors also exactly equal corresponding archive members. The staged ORDINARY-RUNTIME-READBACK.json retains its original 154 CRLF sequences; its SHA256 is 2a9d234a22299dd97698c8f555fc8a1d3cc2b9a5d1f426ba8df7b02e54cfe03d. The directory's -text attributes preserve JSON, Python, ORDINARY markdown and ZIP bytes. No line-ending normalization discrepancy exists. README reviewed SHA256: 16cac24d329f4549288b58cdd67d447ecf3f1a60a3d2202fff764d63682cc53c.
Reconciled all eleven initial/final/baseline result records to their raw log and input-manifest hashes, captured launcher hashes, unchanged source/head and empty verification-failure lists. Parsed actual named outcomes and listing rows directly from logs and compared them in order with receipts:
| Recorded execution | Raw result |
|---|---|
| Initial147 focused | exit101; 9 passed, 4 failed, 216 filtered |
| Initial147 listing / strict Clippy | 229 names / exit0 |
| Finalee50090 focused | exit0; 13 passed, 0 failed, 216 filtered |
| Finalee50090 full library | exit101; 213 passed, 16 failed |
| Finalee50090 listing | exit0; 229 names |
| Finalee50090 strict Clippy | exit0 |
| Finalee50090 formatting | exit1 |
| Baseline93eb611 full / listing | exit101; 200 passed, 16 failed / 216 names |
| Baseline93eb611 formatting | exit1 |
Every one of the 216 baseline named outcomes remains unchanged in final full execution; precisely 13 additional names pass. All 16 final failure bodies retain Unsupported schema version: expected arcadia 3, 4, 5, 6 or 7. The entire 6,446-byte final formatting log is byte-identical to baseline, including eight diff blocks across five files. The README accurately retains these failures and the initial four fixture failures, distinguishes targeted/Clippy success from failed full-library/format gates, and excludes this source from the Windows prerelease. No broader source correctness or runtime acceptance was reassessed here.